Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
Overview
Comment: | Copy fts2 to fts3, renaming, and replacing references to fts2 with fts3, including capitalization variants. (CVS 4249) |
---|---|
Downloads: | Tarball | ZIP archive |
Timelines: | family | ancestors | descendants | both | trunk |
Files: | files | file ages | folders |
SHA1: |
216c91d2fc49792d9ff53596746f1162 |
User & Date: | shess 2007-08-20 17:37:03.000 |
Context
2007-08-20
| ||
17:37 | Add support for building with fts3. (CVS 4250) (check-in: 95ef963232 user: shess tags: trunk) | |
17:37 | Copy fts2 to fts3, renaming, and replacing references to fts2 with fts3, including capitalization variants. (CVS 4249) (check-in: 216c91d2fc user: shess tags: trunk) | |
16:07 | Call sqlite3OsSleep() with an argument in microseconds, not milliseconds. (CVS 4248) (check-in: 917790413c user: danielk1977 tags: trunk) | |
Changes
Added ext/fts3/README.tokenizers.
> > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 | 1. FTS3 Tokenizers When creating a new full-text table, FTS3 allows the user to select the text tokenizer implementation to be used when indexing text by specifying a "tokenizer" clause as part of the CREATE VIRTUAL TABLE statement: CREATE VIRTUAL TABLE <table-name> USING fts3( <columns ...> [, tokenizer <tokenizer-name> [<tokenizer-args>]] ); The built-in tokenizers (valid values to pass as <tokenizer name>) are "simple" and "porter". <tokenizer-args> should consist of zero or more white-space separated arguments to pass to the selected tokenizer implementation. The interpretation of the arguments, if any, depends on the individual tokenizer. 2. Custom Tokenizers FTS3 allows users to provide custom tokenizer implementations. The interface used to create a new tokenizer is defined and described in the fts3_tokenizer.h source file. Registering a new FTS3 tokenizer is similar to registering a new virtual table module with SQLite. The user passes a pointer to a structure containing pointers to various callback functions that make up the implementation of the new tokenizer type. For tokenizers, the structure (defined in fts3_tokenizer.h) is called "sqlite3_tokenizer_module". FTS3 does not expose a C-function that users call to register new tokenizer types with a database handle. Instead, the pointer must be encoded as an SQL blob value and passed to FTS3 through the SQL engine by evaluating a special scalar function, "fts3_tokenizer()". The fts3_tokenizer() function may be called with one or two arguments, as follows: SELECT fts3_tokenizer(<tokenizer-name>); SELECT fts3_tokenizer(<tokenizer-name>, <sqlite3_tokenizer_module ptr>); Where <tokenizer-name> is a string identifying the tokenizer and <sqlite3_tokenizer_module ptr> is a pointer to an sqlite3_tokenizer_module structure encoded as an SQL blob. If the second argument is present, it is registered as tokenizer <tokenizer-name> and a copy of it returned. If only one argument is passed, a pointer to the tokenizer implementation currently registered as <tokenizer-name> is returned, encoded as a blob. Or, if no such tokenizer exists, an SQL exception (error) is raised. SECURITY: If the fts3 extension is used in an environment where potentially malicious users may execute arbitrary SQL (i.e. gears), they should be prevented from invoking the fts3_tokenizer() function, possibly using the authorisation callback. See "Sample code" below for an example of calling the fts3_tokenizer() function from C code. 3. ICU Library Tokenizers If this extension is compiled with the SQLITE_ENABLE_ICU pre-processor symbol defined, then there exists a built-in tokenizer named "icu" implemented using the ICU library. The first argument passed to the xCreate() method (see fts3_tokenizer.h) of this tokenizer may be an ICU locale identifier. For example "tr_TR" for Turkish as used in Turkey, or "en_AU" for English as used in Australia. For example: "CREATE VIRTUAL TABLE thai_text USING fts3(text, tokenizer icu th_TH)" The ICU tokenizer implementation is very simple. It splits the input text according to the ICU rules for finding word boundaries and discards any tokens that consist entirely of white-space. This may be suitable for some applications in some locales, but not all. If more complex processing is required, for example to implement stemming or discard punctuation, this can be done by creating a tokenizer implementation that uses the ICU tokenizer as part of it's implementation. When using the ICU tokenizer this way, it is safe to overwrite the contents of the strings returned by the xNext() method (see fts3_tokenizer.h). 4. Sample code. The following two code samples illustrate the way C code should invoke the fts3_tokenizer() scalar function: int registerTokenizer( sqlite3 *db, char *zName, const sqlite3_tokenizer_module *p ){ int rc; sqlite3_stmt *pStmt; const char zSql[] = "SELECT fts3_tokenizer(?, ?)"; rc = sqlite3_prepare_v2(db, zSql, -1, &pStmt, 0); if( rc!=SQLITE_OK ){ return rc; } sqlite3_bind_text(pStmt, 1, zName, -1, SQLITE_STATIC); sqlite3_bind_blob(pStmt, 2, &p, sizeof(p), SQLITE_STATIC); sqlite3_step(pStmt); return sqlite3_finalize(pStmt); } int queryTokenizer( sqlite3 *db, char *zName, const sqlite3_tokenizer_module **pp ){ int rc; sqlite3_stmt *pStmt; const char zSql[] = "SELECT fts3_tokenizer(?)"; *pp = 0; rc = sqlite3_prepare_v2(db, zSql, -1, &pStmt, 0); if( rc!=SQLITE_OK ){ return rc; } sqlite3_bind_text(pStmt, 1, zName, -1, SQLITE_STATIC); if( SQLITE_ROW==sqlite3_step(pStmt) ){ if( sqlite3_column_type(pStmt, 0)==SQLITE_BLOB ){ memcpy(pp, sqlite3_column_blob(pStmt, 0), sizeof(*pp)); } } return sqlite3_finalize(pStmt); } |
Added ext/fts3/README.txt.
> > > > | 1 2 3 4 | This folder contains source code to the second full-text search extension for SQLite. While the API is the same, this version uses a substantially different storage schema from fts1, so tables will need to be rebuilt. |
Added ext/fts3/fts3.c.
> > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073 4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184 4185 4186 4187 4188 4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221 4222 4223 4224 4225 4226 4227 4228 4229 4230 4231 4232 4233 4234 4235 4236 4237 4238 4239 4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250 4251 4252 4253 4254 4255 4256 4257 4258 4259 4260 4261 4262 4263 4264 4265 4266 4267 4268 4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307 4308 4309 4310 4311 4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365 4366 4367 4368 4369 4370 4371 4372 4373 4374 4375 4376 4377 4378 4379 4380 4381 4382 4383 4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461 4462 4463 4464 4465 4466 4467 4468 4469 4470 4471 4472 4473 4474 4475 4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486 4487 4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512 4513 4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526 4527 4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542 4543 4544 4545 4546 4547 4548 4549 4550 4551 4552 4553 4554 4555 4556 4557 4558 4559 4560 4561 4562 4563 4564 4565 4566 4567 4568 4569 4570 4571 4572 4573 4574 4575 4576 4577 4578 4579 4580 4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604 4605 4606 4607 4608 4609 4610 4611 4612 4613 4614 4615 4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628 4629 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657 4658 4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807 4808 4809 4810 4811 4812 4813 4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828 4829 4830 4831 4832 4833 4834 4835 4836 4837 4838 4839 4840 4841 4842 4843 4844 4845 4846 4847 4848 4849 4850 4851 4852 4853 4854 4855 4856 4857 4858 4859 4860 4861 4862 4863 4864 4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875 4876 4877 4878 4879 4880 4881 4882 4883 4884 4885 4886 4887 4888 4889 4890 4891 4892 4893 4894 4895 4896 4897 4898 4899 4900 4901 4902 4903 4904 4905 4906 4907 4908 4909 4910 4911 4912 4913 4914 4915 4916 4917 4918 4919 4920 4921 4922 4923 4924 4925 4926 4927 4928 4929 4930 4931 4932 4933 4934 4935 4936 4937 4938 4939 4940 4941 4942 4943 4944 4945 4946 4947 4948 4949 4950 4951 4952 4953 4954 4955 4956 4957 4958 4959 4960 4961 4962 4963 4964 4965 4966 4967 4968 4969 4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982 4983 4984 4985 4986 4987 4988 4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065 5066 5067 5068 5069 5070 5071 5072 5073 5074 5075 5076 5077 5078 5079 5080 5081 5082 5083 5084 5085 5086 5087 5088 5089 5090 5091 5092 5093 5094 5095 5096 5097 5098 5099 5100 5101 5102 5103 5104 5105 5106 5107 5108 5109 5110 5111 5112 5113 5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124 5125 5126 5127 5128 5129 5130 5131 5132 5133 5134 5135 5136 5137 5138 5139 5140 5141 5142 5143 5144 5145 5146 5147 5148 5149 5150 5151 5152 5153 5154 5155 5156 5157 5158 5159 5160 5161 5162 5163 5164 5165 5166 5167 5168 5169 5170 5171 5172 5173 5174 5175 5176 5177 5178 5179 5180 5181 5182 5183 5184 5185 5186 5187 5188 5189 5190 5191 5192 5193 5194 5195 5196 5197 5198 5199 5200 5201 5202 5203 5204 5205 5206 5207 5208 5209 5210 5211 5212 5213 5214 5215 5216 5217 5218 5219 5220 5221 5222 5223 5224 5225 5226 5227 5228 5229 5230 5231 5232 5233 5234 5235 5236 5237 5238 5239 5240 5241 5242 5243 5244 5245 5246 5247 5248 5249 5250 5251 5252 5253 5254 5255 5256 5257 5258 5259 5260 5261 5262 5263 5264 5265 5266 5267 5268 5269 5270 5271 5272 5273 5274 5275 5276 5277 5278 5279 5280 5281 5282 5283 5284 5285 5286 5287 5288 5289 5290 5291 5292 5293 5294 5295 5296 5297 5298 5299 5300 5301 5302 5303 5304 5305 5306 5307 5308 5309 5310 5311 5312 5313 5314 5315 5316 5317 5318 5319 5320 5321 5322 5323 5324 5325 5326 5327 5328 5329 5330 5331 5332 5333 5334 5335 5336 5337 5338 5339 5340 5341 5342 5343 5344 5345 5346 5347 5348 5349 5350 5351 5352 5353 5354 5355 5356 5357 5358 5359 5360 5361 5362 5363 5364 5365 5366 5367 5368 5369 5370 5371 5372 5373 5374 5375 5376 5377 5378 5379 5380 5381 5382 5383 5384 5385 5386 5387 5388 5389 5390 5391 5392 5393 5394 5395 5396 5397 5398 5399 5400 5401 5402 5403 5404 5405 5406 5407 5408 5409 5410 5411 5412 5413 5414 5415 5416 5417 5418 5419 5420 5421 5422 5423 5424 5425 5426 5427 5428 5429 5430 5431 5432 5433 5434 5435 5436 5437 5438 5439 5440 5441 5442 5443 5444 5445 5446 5447 5448 5449 5450 5451 5452 5453 5454 5455 5456 5457 5458 5459 5460 5461 5462 5463 5464 5465 5466 5467 5468 5469 5470 5471 5472 5473 5474 5475 5476 5477 5478 5479 5480 5481 5482 5483 5484 5485 5486 5487 5488 5489 5490 5491 5492 5493 5494 5495 5496 5497 5498 5499 5500 5501 5502 5503 5504 5505 5506 5507 5508 5509 5510 5511 5512 5513 5514 5515 5516 5517 5518 5519 5520 5521 5522 5523 5524 5525 5526 5527 5528 5529 5530 5531 5532 5533 5534 5535 5536 5537 5538 5539 5540 5541 5542 5543 5544 5545 5546 5547 5548 5549 5550 5551 5552 5553 5554 5555 5556 5557 5558 5559 5560 5561 5562 5563 5564 5565 5566 5567 5568 5569 5570 5571 5572 5573 5574 5575 5576 5577 5578 5579 5580 5581 5582 5583 5584 5585 5586 5587 5588 5589 5590 5591 5592 5593 5594 5595 5596 5597 5598 5599 5600 5601 5602 5603 5604 5605 5606 5607 5608 5609 5610 5611 5612 5613 5614 5615 5616 5617 5618 5619 5620 5621 5622 5623 5624 5625 5626 5627 5628 5629 5630 5631 5632 5633 5634 5635 5636 5637 5638 5639 5640 5641 5642 5643 5644 5645 5646 5647 5648 5649 5650 5651 5652 5653 5654 5655 5656 5657 5658 5659 5660 5661 5662 5663 5664 5665 5666 5667 5668 5669 5670 5671 5672 5673 5674 5675 5676 5677 5678 5679 5680 5681 5682 5683 5684 5685 5686 5687 5688 5689 5690 5691 5692 5693 5694 5695 5696 5697 5698 5699 5700 5701 5702 5703 5704 5705 5706 5707 5708 5709 5710 5711 5712 5713 5714 5715 5716 5717 5718 5719 5720 5721 5722 5723 5724 5725 5726 5727 5728 5729 5730 5731 5732 5733 5734 5735 5736 5737 5738 5739 5740 5741 5742 5743 5744 5745 5746 5747 5748 5749 5750 5751 5752 5753 5754 5755 5756 5757 5758 5759 5760 5761 5762 5763 5764 5765 5766 5767 5768 5769 5770 5771 5772 5773 5774 5775 5776 5777 5778 5779 5780 5781 5782 5783 5784 5785 5786 5787 5788 5789 5790 5791 5792 5793 5794 5795 5796 5797 5798 5799 5800 5801 5802 5803 5804 5805 5806 5807 5808 5809 5810 5811 5812 5813 5814 5815 5816 5817 5818 5819 5820 5821 5822 5823 5824 5825 5826 5827 5828 5829 5830 5831 5832 5833 5834 5835 5836 5837 5838 5839 5840 5841 5842 5843 5844 5845 5846 5847 5848 5849 5850 5851 5852 5853 5854 5855 5856 5857 5858 5859 5860 5861 5862 5863 5864 5865 5866 5867 5868 5869 5870 5871 5872 5873 5874 5875 5876 5877 5878 5879 5880 5881 5882 5883 5884 5885 5886 5887 5888 5889 5890 5891 5892 5893 5894 5895 5896 5897 5898 5899 5900 5901 5902 5903 5904 5905 5906 5907 5908 5909 5910 5911 5912 5913 5914 5915 5916 5917 5918 | /* fts3 is not ready for use. So don't use it yet! Contact ** shess@google.com with questions. */ #ifndef SQLITE_FTS3_IS_NOT_READY #error fts3 is not ready for use. #endif /* ** 2006 Oct 10 ** ** The author disclaims copyright to this source code. In place of ** a legal notice, here is a blessing: ** ** May you do good and not evil. ** May you find forgiveness for yourself and forgive others. ** May you share freely, never taking more than you give. ** ****************************************************************************** ** ** This is an SQLite module implementing full-text search. */ /* ** The code in this file is only compiled if: ** ** * The FTS3 module is being built as an extension ** (in which case SQLITE_CORE is not defined), or ** ** * The FTS3 module is being built into the core of ** SQLite (in which case SQLITE_ENABLE_FTS3 is defined). */ /* TODO(shess) Consider exporting this comment to an HTML file or the ** wiki. */ /* The full-text index is stored in a series of b+tree (-like) ** structures called segments which map terms to doclists. The ** structures are like b+trees in layout, but are constructed from the ** bottom up in optimal fashion and are not updatable. Since trees ** are built from the bottom up, things will be described from the ** bottom up. ** ** **** Varints **** ** The basic unit of encoding is a variable-length integer called a ** varint. We encode variable-length integers in little-endian order ** using seven bits * per byte as follows: ** ** KEY: ** A = 0xxxxxxx 7 bits of data and one flag bit ** B = 1xxxxxxx 7 bits of data and one flag bit ** ** 7 bits - A ** 14 bits - BA ** 21 bits - BBA ** and so on. ** ** This is identical to how sqlite encodes varints (see util.c). ** ** **** Document lists **** ** A doclist (document list) holds a docid-sorted list of hits for a ** given term. Doclists hold docids, and can optionally associate ** token positions and offsets with docids. ** ** A DL_POSITIONS_OFFSETS doclist is stored like this: ** ** array { ** varint docid; ** array { (position list for column 0) ** varint position; (delta from previous position plus POS_BASE) ** varint startOffset; (delta from previous startOffset) ** varint endOffset; (delta from startOffset) ** } ** array { ** varint POS_COLUMN; (marks start of position list for new column) ** varint column; (index of new column) ** array { ** varint position; (delta from previous position plus POS_BASE) ** varint startOffset;(delta from previous startOffset) ** varint endOffset; (delta from startOffset) ** } ** } ** varint POS_END; (marks end of positions for this document. ** } ** ** Here, array { X } means zero or more occurrences of X, adjacent in ** memory. A "position" is an index of a token in the token stream ** generated by the tokenizer, while an "offset" is a byte offset, ** both based at 0. Note that POS_END and POS_COLUMN occur in the ** same logical place as the position element, and act as sentinals ** ending a position list array. ** ** A DL_POSITIONS doclist omits the startOffset and endOffset ** information. A DL_DOCIDS doclist omits both the position and ** offset information, becoming an array of varint-encoded docids. ** ** On-disk data is stored as type DL_DEFAULT, so we don't serialize ** the type. Due to how deletion is implemented in the segmentation ** system, on-disk doclists MUST store at least positions. ** ** **** Segment leaf nodes **** ** Segment leaf nodes store terms and doclists, ordered by term. Leaf ** nodes are written using LeafWriter, and read using LeafReader (to ** iterate through a single leaf node's data) and LeavesReader (to ** iterate through a segment's entire leaf layer). Leaf nodes have ** the format: ** ** varint iHeight; (height from leaf level, always 0) ** varint nTerm; (length of first term) ** char pTerm[nTerm]; (content of first term) ** varint nDoclist; (length of term's associated doclist) ** char pDoclist[nDoclist]; (content of doclist) ** array { ** (further terms are delta-encoded) ** varint nPrefix; (length of prefix shared with previous term) ** varint nSuffix; (length of unshared suffix) ** char pTermSuffix[nSuffix];(unshared suffix of next term) ** varint nDoclist; (length of term's associated doclist) ** char pDoclist[nDoclist]; (content of doclist) ** } ** ** Here, array { X } means zero or more occurrences of X, adjacent in ** memory. ** ** Leaf nodes are broken into blocks which are stored contiguously in ** the %_segments table in sorted order. This means that when the end ** of a node is reached, the next term is in the node with the next ** greater node id. ** ** New data is spilled to a new leaf node when the current node ** exceeds LEAF_MAX bytes (default 2048). New data which itself is ** larger than STANDALONE_MIN (default 1024) is placed in a standalone ** node (a leaf node with a single term and doclist). The goal of ** these settings is to pack together groups of small doclists while ** making it efficient to directly access large doclists. The ** assumption is that large doclists represent terms which are more ** likely to be query targets. ** ** TODO(shess) It may be useful for blocking decisions to be more ** dynamic. For instance, it may make more sense to have a 2.5k leaf ** node rather than splitting into 2k and .5k nodes. My intuition is ** that this might extend through 2x or 4x the pagesize. ** ** **** Segment interior nodes **** ** Segment interior nodes store blockids for subtree nodes and terms ** to describe what data is stored by the each subtree. Interior ** nodes are written using InteriorWriter, and read using ** InteriorReader. InteriorWriters are created as needed when ** SegmentWriter creates new leaf nodes, or when an interior node ** itself grows too big and must be split. The format of interior ** nodes: ** ** varint iHeight; (height from leaf level, always >0) ** varint iBlockid; (block id of node's leftmost subtree) ** optional { ** varint nTerm; (length of first term) ** char pTerm[nTerm]; (content of first term) ** array { ** (further terms are delta-encoded) ** varint nPrefix; (length of shared prefix with previous term) ** varint nSuffix; (length of unshared suffix) ** char pTermSuffix[nSuffix]; (unshared suffix of next term) ** } ** } ** ** Here, optional { X } means an optional element, while array { X } ** means zero or more occurrences of X, adjacent in memory. ** ** An interior node encodes n terms separating n+1 subtrees. The ** subtree blocks are contiguous, so only the first subtree's blockid ** is encoded. The subtree at iBlockid will contain all terms less ** than the first term encoded (or all terms if no term is encoded). ** Otherwise, for terms greater than or equal to pTerm[i] but less ** than pTerm[i+1], the subtree for that term will be rooted at ** iBlockid+i. Interior nodes only store enough term data to ** distinguish adjacent children (if the rightmost term of the left ** child is "something", and the leftmost term of the right child is ** "wicked", only "w" is stored). ** ** New data is spilled to a new interior node at the same height when ** the current node exceeds INTERIOR_MAX bytes (default 2048). ** INTERIOR_MIN_TERMS (default 7) keeps large terms from monopolizing ** interior nodes and making the tree too skinny. The interior nodes ** at a given height are naturally tracked by interior nodes at ** height+1, and so on. ** ** **** Segment directory **** ** The segment directory in table %_segdir stores meta-information for ** merging and deleting segments, and also the root node of the ** segment's tree. ** ** The root node is the top node of the segment's tree after encoding ** the entire segment, restricted to ROOT_MAX bytes (default 1024). ** This could be either a leaf node or an interior node. If the top ** node requires more than ROOT_MAX bytes, it is flushed to %_segments ** and a new root interior node is generated (which should always fit ** within ROOT_MAX because it only needs space for 2 varints, the ** height and the blockid of the previous root). ** ** The meta-information in the segment directory is: ** level - segment level (see below) ** idx - index within level ** - (level,idx uniquely identify a segment) ** start_block - first leaf node ** leaves_end_block - last leaf node ** end_block - last block (including interior nodes) ** root - contents of root node ** ** If the root node is a leaf node, then start_block, ** leaves_end_block, and end_block are all 0. ** ** **** Segment merging **** ** To amortize update costs, segments are groups into levels and ** merged in matches. Each increase in level represents exponentially ** more documents. ** ** New documents (actually, document updates) are tokenized and ** written individually (using LeafWriter) to a level 0 segment, with ** incrementing idx. When idx reaches MERGE_COUNT (default 16), all ** level 0 segments are merged into a single level 1 segment. Level 1 ** is populated like level 0, and eventually MERGE_COUNT level 1 ** segments are merged to a single level 2 segment (representing ** MERGE_COUNT^2 updates), and so on. ** ** A segment merge traverses all segments at a given level in ** parallel, performing a straightforward sorted merge. Since segment ** leaf nodes are written in to the %_segments table in order, this ** merge traverses the underlying sqlite disk structures efficiently. ** After the merge, all segment blocks from the merged level are ** deleted. ** ** MERGE_COUNT controls how often we merge segments. 16 seems to be ** somewhat of a sweet spot for insertion performance. 32 and 64 show ** very similar performance numbers to 16 on insertion, though they're ** a tiny bit slower (perhaps due to more overhead in merge-time ** sorting). 8 is about 20% slower than 16, 4 about 50% slower than ** 16, 2 about 66% slower than 16. ** ** At query time, high MERGE_COUNT increases the number of segments ** which need to be scanned and merged. For instance, with 100k docs ** inserted: ** ** MERGE_COUNT segments ** 16 25 ** 8 12 ** 4 10 ** 2 6 ** ** This appears to have only a moderate impact on queries for very ** frequent terms (which are somewhat dominated by segment merge ** costs), and infrequent and non-existent terms still seem to be fast ** even with many segments. ** ** TODO(shess) That said, it would be nice to have a better query-side ** argument for MERGE_COUNT of 16. Also, it's possible/likely that ** optimizations to things like doclist merging will swing the sweet ** spot around. ** ** ** **** Handling of deletions and updates **** ** Since we're using a segmented structure, with no docid-oriented ** index into the term index, we clearly cannot simply update the term ** index when a document is deleted or updated. For deletions, we ** write an empty doclist (varint(docid) varint(POS_END)), for updates ** we simply write the new doclist. Segment merges overwrite older ** data for a particular docid with newer data, so deletes or updates ** will eventually overtake the earlier data and knock it out. The ** query logic likewise merges doclists so that newer data knocks out ** older data. ** ** TODO(shess) Provide a VACUUM type operation to clear out all ** deletions and duplications. This would basically be a forced merge ** into a single segment. */ #if !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS3) #if defined(SQLITE_ENABLE_FTS3) && !defined(SQLITE_CORE) # define SQLITE_CORE 1 #endif #include <assert.h> #include <stdlib.h> #include <stdio.h> #include <string.h> #include <ctype.h> #include "fts3.h" #include "fts3_hash.h" #include "fts3_tokenizer.h" #include "sqlite3.h" #include "sqlite3ext.h" SQLITE_EXTENSION_INIT1 /* TODO(shess) MAN, this thing needs some refactoring. At minimum, it ** would be nice to order the file better, perhaps something along the ** lines of: ** ** - utility functions ** - table setup functions ** - table update functions ** - table query functions ** ** Put the query functions last because they're likely to reference ** typedefs or functions from the table update section. */ #if 0 # define TRACE(A) printf A; fflush(stdout) #else # define TRACE(A) #endif /* It is not safe to call isspace(), tolower(), or isalnum() on ** hi-bit-set characters. This is the same solution used in the ** tokenizer. */ /* TODO(shess) The snippet-generation code should be using the ** tokenizer-generated tokens rather than doing its own local ** tokenization. */ /* TODO(shess) Is __isascii() a portable version of (c&0x80)==0? */ static int safe_isspace(char c){ return (c&0x80)==0 ? isspace(c) : 0; } static int safe_tolower(char c){ return (c&0x80)==0 ? tolower(c) : c; } static int safe_isalnum(char c){ return (c&0x80)==0 ? isalnum(c) : 0; } typedef enum DocListType { DL_DOCIDS, /* docids only */ DL_POSITIONS, /* docids + positions */ DL_POSITIONS_OFFSETS /* docids + positions + offsets */ } DocListType; /* ** By default, only positions and not offsets are stored in the doclists. ** To change this so that offsets are stored too, compile with ** ** -DDL_DEFAULT=DL_POSITIONS_OFFSETS ** ** If DL_DEFAULT is set to DL_DOCIDS, your table can only be inserted ** into (no deletes or updates). */ #ifndef DL_DEFAULT # define DL_DEFAULT DL_POSITIONS #endif enum { POS_END = 0, /* end of this position list */ POS_COLUMN, /* followed by new column number */ POS_BASE }; /* MERGE_COUNT controls how often we merge segments (see comment at ** top of file). */ #define MERGE_COUNT 16 /* utility functions */ /* CLEAR() and SCRAMBLE() abstract memset() on a pointer to a single ** record to prevent errors of the form: ** ** my_function(SomeType *b){ ** memset(b, '\0', sizeof(b)); // sizeof(b)!=sizeof(*b) ** } */ /* TODO(shess) Obvious candidates for a header file. */ #define CLEAR(b) memset(b, '\0', sizeof(*(b))) #ifndef NDEBUG # define SCRAMBLE(b) memset(b, 0x55, sizeof(*(b))) #else # define SCRAMBLE(b) #endif /* We may need up to VARINT_MAX bytes to store an encoded 64-bit integer. */ #define VARINT_MAX 10 /* Write a 64-bit variable-length integer to memory starting at p[0]. * The length of data written will be between 1 and VARINT_MAX bytes. * The number of bytes written is returned. */ static int putVarint(char *p, sqlite_int64 v){ unsigned char *q = (unsigned char *) p; sqlite_uint64 vu = v; do{ *q++ = (unsigned char) ((vu & 0x7f) | 0x80); vu >>= 7; }while( vu!=0 ); q[-1] &= 0x7f; /* turn off high bit in final byte */ assert( q - (unsigned char *)p <= VARINT_MAX ); return (int) (q - (unsigned char *)p); } /* Read a 64-bit variable-length integer from memory starting at p[0]. * Return the number of bytes read, or 0 on error. * The value is stored in *v. */ static int getVarint(const char *p, sqlite_int64 *v){ const unsigned char *q = (const unsigned char *) p; sqlite_uint64 x = 0, y = 1; while( (*q & 0x80) == 0x80 ){ x += y * (*q++ & 0x7f); y <<= 7; if( q - (unsigned char *)p >= VARINT_MAX ){ /* bad data */ assert( 0 ); return 0; } } x += y * (*q++); *v = (sqlite_int64) x; return (int) (q - (unsigned char *)p); } static int getVarint32(const char *p, int *pi){ sqlite_int64 i; int ret = getVarint(p, &i); *pi = (int) i; assert( *pi==i ); return ret; } /*******************************************************************/ /* DataBuffer is used to collect data into a buffer in piecemeal ** fashion. It implements the usual distinction between amount of ** data currently stored (nData) and buffer capacity (nCapacity). ** ** dataBufferInit - create a buffer with given initial capacity. ** dataBufferReset - forget buffer's data, retaining capacity. ** dataBufferDestroy - free buffer's data. ** dataBufferExpand - expand capacity without adding data. ** dataBufferAppend - append data. ** dataBufferAppend2 - append two pieces of data at once. ** dataBufferReplace - replace buffer's data. */ typedef struct DataBuffer { char *pData; /* Pointer to malloc'ed buffer. */ int nCapacity; /* Size of pData buffer. */ int nData; /* End of data loaded into pData. */ } DataBuffer; static void dataBufferInit(DataBuffer *pBuffer, int nCapacity){ assert( nCapacity>=0 ); pBuffer->nData = 0; pBuffer->nCapacity = nCapacity; pBuffer->pData = nCapacity==0 ? NULL : malloc(nCapacity); } static void dataBufferReset(DataBuffer *pBuffer){ pBuffer->nData = 0; } static void dataBufferDestroy(DataBuffer *pBuffer){ if( pBuffer->pData!=NULL ) free(pBuffer->pData); SCRAMBLE(pBuffer); } static void dataBufferExpand(DataBuffer *pBuffer, int nAddCapacity){ assert( nAddCapacity>0 ); /* TODO(shess) Consider expanding more aggressively. Note that the ** underlying malloc implementation may take care of such things for ** us already. */ if( pBuffer->nData+nAddCapacity>pBuffer->nCapacity ){ pBuffer->nCapacity = pBuffer->nData+nAddCapacity; pBuffer->pData = realloc(pBuffer->pData, pBuffer->nCapacity); } } static void dataBufferAppend(DataBuffer *pBuffer, const char *pSource, int nSource){ assert( nSource>0 && pSource!=NULL ); dataBufferExpand(pBuffer, nSource); memcpy(pBuffer->pData+pBuffer->nData, pSource, nSource); pBuffer->nData += nSource; } static void dataBufferAppend2(DataBuffer *pBuffer, const char *pSource1, int nSource1, const char *pSource2, int nSource2){ assert( nSource1>0 && pSource1!=NULL ); assert( nSource2>0 && pSource2!=NULL ); dataBufferExpand(pBuffer, nSource1+nSource2); memcpy(pBuffer->pData+pBuffer->nData, pSource1, nSource1); memcpy(pBuffer->pData+pBuffer->nData+nSource1, pSource2, nSource2); pBuffer->nData += nSource1+nSource2; } static void dataBufferReplace(DataBuffer *pBuffer, const char *pSource, int nSource){ dataBufferReset(pBuffer); dataBufferAppend(pBuffer, pSource, nSource); } /* StringBuffer is a null-terminated version of DataBuffer. */ typedef struct StringBuffer { DataBuffer b; /* Includes null terminator. */ } StringBuffer; static void initStringBuffer(StringBuffer *sb){ dataBufferInit(&sb->b, 100); dataBufferReplace(&sb->b, "", 1); } static int stringBufferLength(StringBuffer *sb){ return sb->b.nData-1; } static char *stringBufferData(StringBuffer *sb){ return sb->b.pData; } static void stringBufferDestroy(StringBuffer *sb){ dataBufferDestroy(&sb->b); } static void nappend(StringBuffer *sb, const char *zFrom, int nFrom){ assert( sb->b.nData>0 ); if( nFrom>0 ){ sb->b.nData--; dataBufferAppend2(&sb->b, zFrom, nFrom, "", 1); } } static void append(StringBuffer *sb, const char *zFrom){ nappend(sb, zFrom, strlen(zFrom)); } /* Append a list of strings separated by commas. */ static void appendList(StringBuffer *sb, int nString, char **azString){ int i; for(i=0; i<nString; ++i){ if( i>0 ) append(sb, ", "); append(sb, azString[i]); } } static int endsInWhiteSpace(StringBuffer *p){ return stringBufferLength(p)>0 && safe_isspace(stringBufferData(p)[stringBufferLength(p)-1]); } /* If the StringBuffer ends in something other than white space, add a ** single space character to the end. */ static void appendWhiteSpace(StringBuffer *p){ if( stringBufferLength(p)==0 ) return; if( !endsInWhiteSpace(p) ) append(p, " "); } /* Remove white space from the end of the StringBuffer */ static void trimWhiteSpace(StringBuffer *p){ while( endsInWhiteSpace(p) ){ p->b.pData[--p->b.nData-1] = '\0'; } } /*******************************************************************/ /* DLReader is used to read document elements from a doclist. The ** current docid is cached, so dlrDocid() is fast. DLReader does not ** own the doclist buffer. ** ** dlrAtEnd - true if there's no more data to read. ** dlrDocid - docid of current document. ** dlrDocData - doclist data for current document (including docid). ** dlrDocDataBytes - length of same. ** dlrAllDataBytes - length of all remaining data. ** dlrPosData - position data for current document. ** dlrPosDataLen - length of pos data for current document (incl POS_END). ** dlrStep - step to current document. ** dlrInit - initial for doclist of given type against given data. ** dlrDestroy - clean up. ** ** Expected usage is something like: ** ** DLReader reader; ** dlrInit(&reader, pData, nData); ** while( !dlrAtEnd(&reader) ){ ** // calls to dlrDocid() and kin. ** dlrStep(&reader); ** } ** dlrDestroy(&reader); */ typedef struct DLReader { DocListType iType; const char *pData; int nData; sqlite_int64 iDocid; int nElement; } DLReader; static int dlrAtEnd(DLReader *pReader){ assert( pReader->nData>=0 ); return pReader->nData==0; } static sqlite_int64 dlrDocid(DLReader *pReader){ assert( !dlrAtEnd(pReader) ); return pReader->iDocid; } static const char *dlrDocData(DLReader *pReader){ assert( !dlrAtEnd(pReader) ); return pReader->pData; } static int dlrDocDataBytes(DLReader *pReader){ assert( !dlrAtEnd(pReader) ); return pReader->nElement; } static int dlrAllDataBytes(DLReader *pReader){ assert( !dlrAtEnd(pReader) ); return pReader->nData; } /* TODO(shess) Consider adding a field to track iDocid varint length ** to make these two functions faster. This might matter (a tiny bit) ** for queries. */ static const char *dlrPosData(DLReader *pReader){ sqlite_int64 iDummy; int n = getVarint(pReader->pData, &iDummy); assert( !dlrAtEnd(pReader) ); return pReader->pData+n; } static int dlrPosDataLen(DLReader *pReader){ sqlite_int64 iDummy; int n = getVarint(pReader->pData, &iDummy); assert( !dlrAtEnd(pReader) ); return pReader->nElement-n; } static void dlrStep(DLReader *pReader){ assert( !dlrAtEnd(pReader) ); /* Skip past current doclist element. */ assert( pReader->nElement<=pReader->nData ); pReader->pData += pReader->nElement; pReader->nData -= pReader->nElement; /* If there is more data, read the next doclist element. */ if( pReader->nData!=0 ){ sqlite_int64 iDocidDelta; int iDummy, n = getVarint(pReader->pData, &iDocidDelta); pReader->iDocid += iDocidDelta; if( pReader->iType>=DL_POSITIONS ){ assert( n<pReader->nData ); while( 1 ){ n += getVarint32(pReader->pData+n, &iDummy); assert( n<=pReader->nData ); if( iDummy==POS_END ) break; if( iDummy==POS_COLUMN ){ n += getVarint32(pReader->pData+n, &iDummy); assert( n<pReader->nData ); }else if( pReader->iType==DL_POSITIONS_OFFSETS ){ n += getVarint32(pReader->pData+n, &iDummy); n += getVarint32(pReader->pData+n, &iDummy); assert( n<pReader->nData ); } } } pReader->nElement = n; assert( pReader->nElement<=pReader->nData ); } } static void dlrInit(DLReader *pReader, DocListType iType, const char *pData, int nData){ assert( pData!=NULL && nData!=0 ); pReader->iType = iType; pReader->pData = pData; pReader->nData = nData; pReader->nElement = 0; pReader->iDocid = 0; /* Load the first element's data. There must be a first element. */ dlrStep(pReader); } static void dlrDestroy(DLReader *pReader){ SCRAMBLE(pReader); } #ifndef NDEBUG /* Verify that the doclist can be validly decoded. Also returns the ** last docid found because it's convenient in other assertions for ** DLWriter. */ static void docListValidate(DocListType iType, const char *pData, int nData, sqlite_int64 *pLastDocid){ sqlite_int64 iPrevDocid = 0; assert( nData>0 ); assert( pData!=0 ); assert( pData+nData>pData ); while( nData!=0 ){ sqlite_int64 iDocidDelta; int n = getVarint(pData, &iDocidDelta); iPrevDocid += iDocidDelta; if( iType>DL_DOCIDS ){ int iDummy; while( 1 ){ n += getVarint32(pData+n, &iDummy); if( iDummy==POS_END ) break; if( iDummy==POS_COLUMN ){ n += getVarint32(pData+n, &iDummy); }else if( iType>DL_POSITIONS ){ n += getVarint32(pData+n, &iDummy); n += getVarint32(pData+n, &iDummy); } assert( n<=nData ); } } assert( n<=nData ); pData += n; nData -= n; } if( pLastDocid ) *pLastDocid = iPrevDocid; } #define ASSERT_VALID_DOCLIST(i, p, n, o) docListValidate(i, p, n, o) #else #define ASSERT_VALID_DOCLIST(i, p, n, o) assert( 1 ) #endif /*******************************************************************/ /* DLWriter is used to write doclist data to a DataBuffer. DLWriter ** always appends to the buffer and does not own it. ** ** dlwInit - initialize to write a given type doclistto a buffer. ** dlwDestroy - clear the writer's memory. Does not free buffer. ** dlwAppend - append raw doclist data to buffer. ** dlwCopy - copy next doclist from reader to writer. ** dlwAdd - construct doclist element and append to buffer. ** Only apply dlwAdd() to DL_DOCIDS doclists (else use PLWriter). */ typedef struct DLWriter { DocListType iType; DataBuffer *b; sqlite_int64 iPrevDocid; #ifndef NDEBUG int has_iPrevDocid; #endif } DLWriter; static void dlwInit(DLWriter *pWriter, DocListType iType, DataBuffer *b){ pWriter->b = b; pWriter->iType = iType; pWriter->iPrevDocid = 0; #ifndef NDEBUG pWriter->has_iPrevDocid = 0; #endif } static void dlwDestroy(DLWriter *pWriter){ SCRAMBLE(pWriter); } /* iFirstDocid is the first docid in the doclist in pData. It is ** needed because pData may point within a larger doclist, in which ** case the first item would be delta-encoded. ** ** iLastDocid is the final docid in the doclist in pData. It is ** needed to create the new iPrevDocid for future delta-encoding. The ** code could decode the passed doclist to recreate iLastDocid, but ** the only current user (docListMerge) already has decoded this ** information. */ /* TODO(shess) This has become just a helper for docListMerge. ** Consider a refactor to make this cleaner. */ static void dlwAppend(DLWriter *pWriter, const char *pData, int nData, sqlite_int64 iFirstDocid, sqlite_int64 iLastDocid){ sqlite_int64 iDocid = 0; char c[VARINT_MAX]; int nFirstOld, nFirstNew; /* Old and new varint len of first docid. */ #ifndef NDEBUG sqlite_int64 iLastDocidDelta; #endif /* Recode the initial docid as delta from iPrevDocid. */ nFirstOld = getVarint(pData, &iDocid); assert( nFirstOld<nData || (nFirstOld==nData && pWriter->iType==DL_DOCIDS) ); nFirstNew = putVarint(c, iFirstDocid-pWriter->iPrevDocid); /* Verify that the incoming doclist is valid AND that it ends with ** the expected docid. This is essential because we'll trust this ** docid in future delta-encoding. */ ASSERT_VALID_DOCLIST(pWriter->iType, pData, nData, &iLastDocidDelta); assert( iLastDocid==iFirstDocid-iDocid+iLastDocidDelta ); /* Append recoded initial docid and everything else. Rest of docids ** should have been delta-encoded from previous initial docid. */ if( nFirstOld<nData ){ dataBufferAppend2(pWriter->b, c, nFirstNew, pData+nFirstOld, nData-nFirstOld); }else{ dataBufferAppend(pWriter->b, c, nFirstNew); } pWriter->iPrevDocid = iLastDocid; } static void dlwCopy(DLWriter *pWriter, DLReader *pReader){ dlwAppend(pWriter, dlrDocData(pReader), dlrDocDataBytes(pReader), dlrDocid(pReader), dlrDocid(pReader)); } static void dlwAdd(DLWriter *pWriter, sqlite_int64 iDocid){ char c[VARINT_MAX]; int n = putVarint(c, iDocid-pWriter->iPrevDocid); /* Docids must ascend. */ assert( !pWriter->has_iPrevDocid || iDocid>pWriter->iPrevDocid ); assert( pWriter->iType==DL_DOCIDS ); dataBufferAppend(pWriter->b, c, n); pWriter->iPrevDocid = iDocid; #ifndef NDEBUG pWriter->has_iPrevDocid = 1; #endif } /*******************************************************************/ /* PLReader is used to read data from a document's position list. As ** the caller steps through the list, data is cached so that varints ** only need to be decoded once. ** ** plrInit, plrDestroy - create/destroy a reader. ** plrColumn, plrPosition, plrStartOffset, plrEndOffset - accessors ** plrAtEnd - at end of stream, only call plrDestroy once true. ** plrStep - step to the next element. */ typedef struct PLReader { /* These refer to the next position's data. nData will reach 0 when ** reading the last position, so plrStep() signals EOF by setting ** pData to NULL. */ const char *pData; int nData; DocListType iType; int iColumn; /* the last column read */ int iPosition; /* the last position read */ int iStartOffset; /* the last start offset read */ int iEndOffset; /* the last end offset read */ } PLReader; static int plrAtEnd(PLReader *pReader){ return pReader->pData==NULL; } static int plrColumn(PLReader *pReader){ assert( !plrAtEnd(pReader) ); return pReader->iColumn; } static int plrPosition(PLReader *pReader){ assert( !plrAtEnd(pReader) ); return pReader->iPosition; } static int plrStartOffset(PLReader *pReader){ assert( !plrAtEnd(pReader) ); return pReader->iStartOffset; } static int plrEndOffset(PLReader *pReader){ assert( !plrAtEnd(pReader) ); return pReader->iEndOffset; } static void plrStep(PLReader *pReader){ int i, n; assert( !plrAtEnd(pReader) ); if( pReader->nData==0 ){ pReader->pData = NULL; return; } n = getVarint32(pReader->pData, &i); if( i==POS_COLUMN ){ n += getVarint32(pReader->pData+n, &pReader->iColumn); pReader->iPosition = 0; pReader->iStartOffset = 0; n += getVarint32(pReader->pData+n, &i); } /* Should never see adjacent column changes. */ assert( i!=POS_COLUMN ); if( i==POS_END ){ pReader->nData = 0; pReader->pData = NULL; return; } pReader->iPosition += i-POS_BASE; if( pReader->iType==DL_POSITIONS_OFFSETS ){ n += getVarint32(pReader->pData+n, &i); pReader->iStartOffset += i; n += getVarint32(pReader->pData+n, &i); pReader->iEndOffset = pReader->iStartOffset+i; } assert( n<=pReader->nData ); pReader->pData += n; pReader->nData -= n; } static void plrInit(PLReader *pReader, DLReader *pDLReader){ pReader->pData = dlrPosData(pDLReader); pReader->nData = dlrPosDataLen(pDLReader); pReader->iType = pDLReader->iType; pReader->iColumn = 0; pReader->iPosition = 0; pReader->iStartOffset = 0; pReader->iEndOffset = 0; plrStep(pReader); } static void plrDestroy(PLReader *pReader){ SCRAMBLE(pReader); } /*******************************************************************/ /* PLWriter is used in constructing a document's position list. As a ** convenience, if iType is DL_DOCIDS, PLWriter becomes a no-op. ** PLWriter writes to the associated DLWriter's buffer. ** ** plwInit - init for writing a document's poslist. ** plwDestroy - clear a writer. ** plwAdd - append position and offset information. ** plwCopy - copy next position's data from reader to writer. ** plwTerminate - add any necessary doclist terminator. ** ** Calling plwAdd() after plwTerminate() may result in a corrupt ** doclist. */ /* TODO(shess) Until we've written the second item, we can cache the ** first item's information. Then we'd have three states: ** ** - initialized with docid, no positions. ** - docid and one position. ** - docid and multiple positions. ** ** Only the last state needs to actually write to dlw->b, which would ** be an improvement in the DLCollector case. */ typedef struct PLWriter { DLWriter *dlw; int iColumn; /* the last column written */ int iPos; /* the last position written */ int iOffset; /* the last start offset written */ } PLWriter; /* TODO(shess) In the case where the parent is reading these values ** from a PLReader, we could optimize to a copy if that PLReader has ** the same type as pWriter. */ static void plwAdd(PLWriter *pWriter, int iColumn, int iPos, int iStartOffset, int iEndOffset){ /* Worst-case space for POS_COLUMN, iColumn, iPosDelta, ** iStartOffsetDelta, and iEndOffsetDelta. */ char c[5*VARINT_MAX]; int n = 0; /* Ban plwAdd() after plwTerminate(). */ assert( pWriter->iPos!=-1 ); if( pWriter->dlw->iType==DL_DOCIDS ) return; if( iColumn!=pWriter->iColumn ){ n += putVarint(c+n, POS_COLUMN); n += putVarint(c+n, iColumn); pWriter->iColumn = iColumn; pWriter->iPos = 0; pWriter->iOffset = 0; } assert( iPos>=pWriter->iPos ); n += putVarint(c+n, POS_BASE+(iPos-pWriter->iPos)); pWriter->iPos = iPos; if( pWriter->dlw->iType==DL_POSITIONS_OFFSETS ){ assert( iStartOffset>=pWriter->iOffset ); n += putVarint(c+n, iStartOffset-pWriter->iOffset); pWriter->iOffset = iStartOffset; assert( iEndOffset>=iStartOffset ); n += putVarint(c+n, iEndOffset-iStartOffset); } dataBufferAppend(pWriter->dlw->b, c, n); } static void plwCopy(PLWriter *pWriter, PLReader *pReader){ plwAdd(pWriter, plrColumn(pReader), plrPosition(pReader), plrStartOffset(pReader), plrEndOffset(pReader)); } static void plwInit(PLWriter *pWriter, DLWriter *dlw, sqlite_int64 iDocid){ char c[VARINT_MAX]; int n; pWriter->dlw = dlw; /* Docids must ascend. */ assert( !pWriter->dlw->has_iPrevDocid || iDocid>pWriter->dlw->iPrevDocid ); n = putVarint(c, iDocid-pWriter->dlw->iPrevDocid); dataBufferAppend(pWriter->dlw->b, c, n); pWriter->dlw->iPrevDocid = iDocid; #ifndef NDEBUG pWriter->dlw->has_iPrevDocid = 1; #endif pWriter->iColumn = 0; pWriter->iPos = 0; pWriter->iOffset = 0; } /* TODO(shess) Should plwDestroy() also terminate the doclist? But ** then plwDestroy() would no longer be just a destructor, it would ** also be doing work, which isn't consistent with the overall idiom. ** Another option would be for plwAdd() to always append any necessary ** terminator, so that the output is always correct. But that would ** add incremental work to the common case with the only benefit being ** API elegance. Punt for now. */ static void plwTerminate(PLWriter *pWriter){ if( pWriter->dlw->iType>DL_DOCIDS ){ char c[VARINT_MAX]; int n = putVarint(c, POS_END); dataBufferAppend(pWriter->dlw->b, c, n); } #ifndef NDEBUG /* Mark as terminated for assert in plwAdd(). */ pWriter->iPos = -1; #endif } static void plwDestroy(PLWriter *pWriter){ SCRAMBLE(pWriter); } /*******************************************************************/ /* DLCollector wraps PLWriter and DLWriter to provide a ** dynamically-allocated doclist area to use during tokenization. ** ** dlcNew - malloc up and initialize a collector. ** dlcDelete - destroy a collector and all contained items. ** dlcAddPos - append position and offset information. ** dlcAddDoclist - add the collected doclist to the given buffer. ** dlcNext - terminate the current document and open another. */ typedef struct DLCollector { DataBuffer b; DLWriter dlw; PLWriter plw; } DLCollector; /* TODO(shess) This could also be done by calling plwTerminate() and ** dataBufferAppend(). I tried that, expecting nominal performance ** differences, but it seemed to pretty reliably be worth 1% to code ** it this way. I suspect it's the incremental malloc overhead (some ** percentage of the plwTerminate() calls will cause a realloc), so ** this might be worth revisiting if the DataBuffer implementation ** changes. */ static void dlcAddDoclist(DLCollector *pCollector, DataBuffer *b){ if( pCollector->dlw.iType>DL_DOCIDS ){ char c[VARINT_MAX]; int n = putVarint(c, POS_END); dataBufferAppend2(b, pCollector->b.pData, pCollector->b.nData, c, n); }else{ dataBufferAppend(b, pCollector->b.pData, pCollector->b.nData); } } static void dlcNext(DLCollector *pCollector, sqlite_int64 iDocid){ plwTerminate(&pCollector->plw); plwDestroy(&pCollector->plw); plwInit(&pCollector->plw, &pCollector->dlw, iDocid); } static void dlcAddPos(DLCollector *pCollector, int iColumn, int iPos, int iStartOffset, int iEndOffset){ plwAdd(&pCollector->plw, iColumn, iPos, iStartOffset, iEndOffset); } static DLCollector *dlcNew(sqlite_int64 iDocid, DocListType iType){ DLCollector *pCollector = malloc(sizeof(DLCollector)); dataBufferInit(&pCollector->b, 0); dlwInit(&pCollector->dlw, iType, &pCollector->b); plwInit(&pCollector->plw, &pCollector->dlw, iDocid); return pCollector; } static void dlcDelete(DLCollector *pCollector){ plwDestroy(&pCollector->plw); dlwDestroy(&pCollector->dlw); dataBufferDestroy(&pCollector->b); SCRAMBLE(pCollector); free(pCollector); } /* Copy the doclist data of iType in pData/nData into *out, trimming ** unnecessary data as we go. Only columns matching iColumn are ** copied, all columns copied if iColumn is -1. Elements with no ** matching columns are dropped. The output is an iOutType doclist. */ /* NOTE(shess) This code is only valid after all doclists are merged. ** If this is run before merges, then doclist items which represent ** deletion will be trimmed, and will thus not effect a deletion ** during the merge. */ static void docListTrim(DocListType iType, const char *pData, int nData, int iColumn, DocListType iOutType, DataBuffer *out){ DLReader dlReader; DLWriter dlWriter; assert( iOutType<=iType ); dlrInit(&dlReader, iType, pData, nData); dlwInit(&dlWriter, iOutType, out); while( !dlrAtEnd(&dlReader) ){ PLReader plReader; PLWriter plWriter; int match = 0; plrInit(&plReader, &dlReader); while( !plrAtEnd(&plReader) ){ if( iColumn==-1 || plrColumn(&plReader)==iColumn ){ if( !match ){ plwInit(&plWriter, &dlWriter, dlrDocid(&dlReader)); match = 1; } plwAdd(&plWriter, plrColumn(&plReader), plrPosition(&plReader), plrStartOffset(&plReader), plrEndOffset(&plReader)); } plrStep(&plReader); } if( match ){ plwTerminate(&plWriter); plwDestroy(&plWriter); } plrDestroy(&plReader); dlrStep(&dlReader); } dlwDestroy(&dlWriter); dlrDestroy(&dlReader); } /* Used by docListMerge() to keep doclists in the ascending order by ** docid, then ascending order by age (so the newest comes first). */ typedef struct OrderedDLReader { DLReader *pReader; /* TODO(shess) If we assume that docListMerge pReaders is ordered by ** age (which we do), then we could use pReader comparisons to break ** ties. */ int idx; } OrderedDLReader; /* Order eof to end, then by docid asc, idx desc. */ static int orderedDLReaderCmp(OrderedDLReader *r1, OrderedDLReader *r2){ if( dlrAtEnd(r1->pReader) ){ if( dlrAtEnd(r2->pReader) ) return 0; /* Both atEnd(). */ return 1; /* Only r1 atEnd(). */ } if( dlrAtEnd(r2->pReader) ) return -1; /* Only r2 atEnd(). */ if( dlrDocid(r1->pReader)<dlrDocid(r2->pReader) ) return -1; if( dlrDocid(r1->pReader)>dlrDocid(r2->pReader) ) return 1; /* Descending on idx. */ return r2->idx-r1->idx; } /* Bubble p[0] to appropriate place in p[1..n-1]. Assumes that ** p[1..n-1] is already sorted. */ /* TODO(shess) Is this frequent enough to warrant a binary search? ** Before implementing that, instrument the code to check. In most ** current usage, I expect that p[0] will be less than p[1] a very ** high proportion of the time. */ static void orderedDLReaderReorder(OrderedDLReader *p, int n){ while( n>1 && orderedDLReaderCmp(p, p+1)>0 ){ OrderedDLReader tmp = p[0]; p[0] = p[1]; p[1] = tmp; n--; p++; } } /* Given an array of doclist readers, merge their doclist elements ** into out in sorted order (by docid), dropping elements from older ** readers when there is a duplicate docid. pReaders is assumed to be ** ordered by age, oldest first. */ /* TODO(shess) nReaders must be <= MERGE_COUNT. This should probably ** be fixed. */ static void docListMerge(DataBuffer *out, DLReader *pReaders, int nReaders){ OrderedDLReader readers[MERGE_COUNT]; DLWriter writer; int i, n; const char *pStart = 0; int nStart = 0; sqlite_int64 iFirstDocid = 0, iLastDocid = 0; assert( nReaders>0 ); if( nReaders==1 ){ dataBufferAppend(out, dlrDocData(pReaders), dlrAllDataBytes(pReaders)); return; } assert( nReaders<=MERGE_COUNT ); n = 0; for(i=0; i<nReaders; i++){ assert( pReaders[i].iType==pReaders[0].iType ); readers[i].pReader = pReaders+i; readers[i].idx = i; n += dlrAllDataBytes(&pReaders[i]); } /* Conservatively size output to sum of inputs. Output should end ** up strictly smaller than input. */ dataBufferExpand(out, n); /* Get the readers into sorted order. */ while( i-->0 ){ orderedDLReaderReorder(readers+i, nReaders-i); } dlwInit(&writer, pReaders[0].iType, out); while( !dlrAtEnd(readers[0].pReader) ){ sqlite_int64 iDocid = dlrDocid(readers[0].pReader); /* If this is a continuation of the current buffer to copy, extend ** that buffer. memcpy() seems to be more efficient if it has a ** lots of data to copy. */ if( dlrDocData(readers[0].pReader)==pStart+nStart ){ nStart += dlrDocDataBytes(readers[0].pReader); }else{ if( pStart!=0 ){ dlwAppend(&writer, pStart, nStart, iFirstDocid, iLastDocid); } pStart = dlrDocData(readers[0].pReader); nStart = dlrDocDataBytes(readers[0].pReader); iFirstDocid = iDocid; } iLastDocid = iDocid; dlrStep(readers[0].pReader); /* Drop all of the older elements with the same docid. */ for(i=1; i<nReaders && !dlrAtEnd(readers[i].pReader) && dlrDocid(readers[i].pReader)==iDocid; i++){ dlrStep(readers[i].pReader); } /* Get the readers back into order. */ while( i-->0 ){ orderedDLReaderReorder(readers+i, nReaders-i); } } /* Copy over any remaining elements. */ if( nStart>0 ) dlwAppend(&writer, pStart, nStart, iFirstDocid, iLastDocid); dlwDestroy(&writer); } /* Helper function for posListUnion(). Compares the current position ** between left and right, returning as standard C idiom of <0 if ** left<right, >0 if left>right, and 0 if left==right. "End" always ** compares greater. */ static int posListCmp(PLReader *pLeft, PLReader *pRight){ assert( pLeft->iType==pRight->iType ); if( pLeft->iType==DL_DOCIDS ) return 0; if( plrAtEnd(pLeft) ) return plrAtEnd(pRight) ? 0 : 1; if( plrAtEnd(pRight) ) return -1; if( plrColumn(pLeft)<plrColumn(pRight) ) return -1; if( plrColumn(pLeft)>plrColumn(pRight) ) return 1; if( plrPosition(pLeft)<plrPosition(pRight) ) return -1; if( plrPosition(pLeft)>plrPosition(pRight) ) return 1; if( pLeft->iType==DL_POSITIONS ) return 0; if( plrStartOffset(pLeft)<plrStartOffset(pRight) ) return -1; if( plrStartOffset(pLeft)>plrStartOffset(pRight) ) return 1; if( plrEndOffset(pLeft)<plrEndOffset(pRight) ) return -1; if( plrEndOffset(pLeft)>plrEndOffset(pRight) ) return 1; return 0; } /* Write the union of position lists in pLeft and pRight to pOut. ** "Union" in this case meaning "All unique position tuples". Should ** work with any doclist type, though both inputs and the output ** should be the same type. */ static void posListUnion(DLReader *pLeft, DLReader *pRight, DLWriter *pOut){ PLReader left, right; PLWriter writer; assert( dlrDocid(pLeft)==dlrDocid(pRight) ); assert( pLeft->iType==pRight->iType ); assert( pLeft->iType==pOut->iType ); plrInit(&left, pLeft); plrInit(&right, pRight); plwInit(&writer, pOut, dlrDocid(pLeft)); while( !plrAtEnd(&left) || !plrAtEnd(&right) ){ int c = posListCmp(&left, &right); if( c<0 ){ plwCopy(&writer, &left); plrStep(&left); }else if( c>0 ){ plwCopy(&writer, &right); plrStep(&right); }else{ plwCopy(&writer, &left); plrStep(&left); plrStep(&right); } } plwTerminate(&writer); plwDestroy(&writer); plrDestroy(&left); plrDestroy(&right); } /* Write the union of doclists in pLeft and pRight to pOut. For ** docids in common between the inputs, the union of the position ** lists is written. Inputs and outputs are always type DL_DEFAULT. */ static void docListUnion( const char *pLeft, int nLeft, const char *pRight, int nRight, DataBuffer *pOut /* Write the combined doclist here */ ){ DLReader left, right; DLWriter writer; if( nLeft==0 ){ dataBufferAppend(pOut, pRight, nRight); return; } if( nRight==0 ){ dataBufferAppend(pOut, pLeft, nLeft); return; } dlrInit(&left, DL_DEFAULT, pLeft, nLeft); dlrInit(&right, DL_DEFAULT, pRight, nRight); dlwInit(&writer, DL_DEFAULT, pOut); while( !dlrAtEnd(&left) || !dlrAtEnd(&right) ){ if( dlrAtEnd(&right) ){ dlwCopy(&writer, &left); dlrStep(&left); }else if( dlrAtEnd(&left) ){ dlwCopy(&writer, &right); dlrStep(&right); }else if( dlrDocid(&left)<dlrDocid(&right) ){ dlwCopy(&writer, &left); dlrStep(&left); }else if( dlrDocid(&left)>dlrDocid(&right) ){ dlwCopy(&writer, &right); dlrStep(&right); }else{ posListUnion(&left, &right, &writer); dlrStep(&left); dlrStep(&right); } } dlrDestroy(&left); dlrDestroy(&right); dlwDestroy(&writer); } /* pLeft and pRight are DLReaders positioned to the same docid. ** ** If there are no instances in pLeft or pRight where the position ** of pLeft is one less than the position of pRight, then this ** routine adds nothing to pOut. ** ** If there are one or more instances where positions from pLeft ** are exactly one less than positions from pRight, then add a new ** document record to pOut. If pOut wants to hold positions, then ** include the positions from pRight that are one more than a ** position in pLeft. In other words: pRight.iPos==pLeft.iPos+1. */ static void posListPhraseMerge(DLReader *pLeft, DLReader *pRight, DLWriter *pOut){ PLReader left, right; PLWriter writer; int match = 0; assert( dlrDocid(pLeft)==dlrDocid(pRight) ); assert( pOut->iType!=DL_POSITIONS_OFFSETS ); plrInit(&left, pLeft); plrInit(&right, pRight); while( !plrAtEnd(&left) && !plrAtEnd(&right) ){ if( plrColumn(&left)<plrColumn(&right) ){ plrStep(&left); }else if( plrColumn(&left)>plrColumn(&right) ){ plrStep(&right); }else if( plrPosition(&left)+1<plrPosition(&right) ){ plrStep(&left); }else if( plrPosition(&left)+1>plrPosition(&right) ){ plrStep(&right); }else{ if( !match ){ plwInit(&writer, pOut, dlrDocid(pLeft)); match = 1; } plwAdd(&writer, plrColumn(&right), plrPosition(&right), 0, 0); plrStep(&left); plrStep(&right); } } if( match ){ plwTerminate(&writer); plwDestroy(&writer); } plrDestroy(&left); plrDestroy(&right); } /* We have two doclists with positions: pLeft and pRight. ** Write the phrase intersection of these two doclists into pOut. ** ** A phrase intersection means that two documents only match ** if pLeft.iPos+1==pRight.iPos. ** ** iType controls the type of data written to pOut. If iType is ** DL_POSITIONS, the positions are those from pRight. */ static void docListPhraseMerge( const char *pLeft, int nLeft, const char *pRight, int nRight, DocListType iType, DataBuffer *pOut /* Write the combined doclist here */ ){ DLReader left, right; DLWriter writer; if( nLeft==0 || nRight==0 ) return; assert( iType!=DL_POSITIONS_OFFSETS ); dlrInit(&left, DL_POSITIONS, pLeft, nLeft); dlrInit(&right, DL_POSITIONS, pRight, nRight); dlwInit(&writer, iType, pOut); while( !dlrAtEnd(&left) && !dlrAtEnd(&right) ){ if( dlrDocid(&left)<dlrDocid(&right) ){ dlrStep(&left); }else if( dlrDocid(&right)<dlrDocid(&left) ){ dlrStep(&right); }else{ posListPhraseMerge(&left, &right, &writer); dlrStep(&left); dlrStep(&right); } } dlrDestroy(&left); dlrDestroy(&right); dlwDestroy(&writer); } /* We have two DL_DOCIDS doclists: pLeft and pRight. ** Write the intersection of these two doclists into pOut as a ** DL_DOCIDS doclist. */ static void docListAndMerge( const char *pLeft, int nLeft, const char *pRight, int nRight, DataBuffer *pOut /* Write the combined doclist here */ ){ DLReader left, right; DLWriter writer; if( nLeft==0 || nRight==0 ) return; dlrInit(&left, DL_DOCIDS, pLeft, nLeft); dlrInit(&right, DL_DOCIDS, pRight, nRight); dlwInit(&writer, DL_DOCIDS, pOut); while( !dlrAtEnd(&left) && !dlrAtEnd(&right) ){ if( dlrDocid(&left)<dlrDocid(&right) ){ dlrStep(&left); }else if( dlrDocid(&right)<dlrDocid(&left) ){ dlrStep(&right); }else{ dlwAdd(&writer, dlrDocid(&left)); dlrStep(&left); dlrStep(&right); } } dlrDestroy(&left); dlrDestroy(&right); dlwDestroy(&writer); } /* We have two DL_DOCIDS doclists: pLeft and pRight. ** Write the union of these two doclists into pOut as a ** DL_DOCIDS doclist. */ static void docListOrMerge( const char *pLeft, int nLeft, const char *pRight, int nRight, DataBuffer *pOut /* Write the combined doclist here */ ){ DLReader left, right; DLWriter writer; if( nLeft==0 ){ dataBufferAppend(pOut, pRight, nRight); return; } if( nRight==0 ){ dataBufferAppend(pOut, pLeft, nLeft); return; } dlrInit(&left, DL_DOCIDS, pLeft, nLeft); dlrInit(&right, DL_DOCIDS, pRight, nRight); dlwInit(&writer, DL_DOCIDS, pOut); while( !dlrAtEnd(&left) || !dlrAtEnd(&right) ){ if( dlrAtEnd(&right) ){ dlwAdd(&writer, dlrDocid(&left)); dlrStep(&left); }else if( dlrAtEnd(&left) ){ dlwAdd(&writer, dlrDocid(&right)); dlrStep(&right); }else if( dlrDocid(&left)<dlrDocid(&right) ){ dlwAdd(&writer, dlrDocid(&left)); dlrStep(&left); }else if( dlrDocid(&right)<dlrDocid(&left) ){ dlwAdd(&writer, dlrDocid(&right)); dlrStep(&right); }else{ dlwAdd(&writer, dlrDocid(&left)); dlrStep(&left); dlrStep(&right); } } dlrDestroy(&left); dlrDestroy(&right); dlwDestroy(&writer); } /* We have two DL_DOCIDS doclists: pLeft and pRight. ** Write into pOut as DL_DOCIDS doclist containing all documents that ** occur in pLeft but not in pRight. */ static void docListExceptMerge( const char *pLeft, int nLeft, const char *pRight, int nRight, DataBuffer *pOut /* Write the combined doclist here */ ){ DLReader left, right; DLWriter writer; if( nLeft==0 ) return; if( nRight==0 ){ dataBufferAppend(pOut, pLeft, nLeft); return; } dlrInit(&left, DL_DOCIDS, pLeft, nLeft); dlrInit(&right, DL_DOCIDS, pRight, nRight); dlwInit(&writer, DL_DOCIDS, pOut); while( !dlrAtEnd(&left) ){ while( !dlrAtEnd(&right) && dlrDocid(&right)<dlrDocid(&left) ){ dlrStep(&right); } if( dlrAtEnd(&right) || dlrDocid(&left)<dlrDocid(&right) ){ dlwAdd(&writer, dlrDocid(&left)); } dlrStep(&left); } dlrDestroy(&left); dlrDestroy(&right); dlwDestroy(&writer); } static char *string_dup_n(const char *s, int n){ char *str = malloc(n + 1); memcpy(str, s, n); str[n] = '\0'; return str; } /* Duplicate a string; the caller must free() the returned string. * (We don't use strdup() since it's not part of the standard C library and * may not be available everywhere.) */ static char *string_dup(const char *s){ return string_dup_n(s, strlen(s)); } /* Format a string, replacing each occurrence of the % character with * zDb.zName. This may be more convenient than sqlite_mprintf() * when one string is used repeatedly in a format string. * The caller must free() the returned string. */ static char *string_format(const char *zFormat, const char *zDb, const char *zName){ const char *p; size_t len = 0; size_t nDb = strlen(zDb); size_t nName = strlen(zName); size_t nFullTableName = nDb+1+nName; char *result; char *r; /* first compute length needed */ for(p = zFormat ; *p ; ++p){ len += (*p=='%' ? nFullTableName : 1); } len += 1; /* for null terminator */ r = result = malloc(len); for(p = zFormat; *p; ++p){ if( *p=='%' ){ memcpy(r, zDb, nDb); r += nDb; *r++ = '.'; memcpy(r, zName, nName); r += nName; } else { *r++ = *p; } } *r++ = '\0'; assert( r == result + len ); return result; } static int sql_exec(sqlite3 *db, const char *zDb, const char *zName, const char *zFormat){ char *zCommand = string_format(zFormat, zDb, zName); int rc; TRACE(("FTS3 sql: %s\n", zCommand)); rc = sqlite3_exec(db, zCommand, NULL, 0, NULL); free(zCommand); return rc; } static int sql_prepare(sqlite3 *db, const char *zDb, const char *zName, sqlite3_stmt **ppStmt, const char *zFormat){ char *zCommand = string_format(zFormat, zDb, zName); int rc; TRACE(("FTS3 prepare: %s\n", zCommand)); rc = sqlite3_prepare_v2(db, zCommand, -1, ppStmt, NULL); free(zCommand); return rc; } /* end utility functions */ /* Forward reference */ typedef struct fulltext_vtab fulltext_vtab; /* A single term in a query is represented by an instances of ** the following structure. */ typedef struct QueryTerm { short int nPhrase; /* How many following terms are part of the same phrase */ short int iPhrase; /* This is the i-th term of a phrase. */ short int iColumn; /* Column of the index that must match this term */ signed char isOr; /* this term is preceded by "OR" */ signed char isNot; /* this term is preceded by "-" */ signed char isPrefix; /* this term is followed by "*" */ char *pTerm; /* text of the term. '\000' terminated. malloced */ int nTerm; /* Number of bytes in pTerm[] */ } QueryTerm; /* A query string is parsed into a Query structure. * * We could, in theory, allow query strings to be complicated * nested expressions with precedence determined by parentheses. * But none of the major search engines do this. (Perhaps the * feeling is that an parenthesized expression is two complex of * an idea for the average user to grasp.) Taking our lead from * the major search engines, we will allow queries to be a list * of terms (with an implied AND operator) or phrases in double-quotes, * with a single optional "-" before each non-phrase term to designate * negation and an optional OR connector. * * OR binds more tightly than the implied AND, which is what the * major search engines seem to do. So, for example: * * [one two OR three] ==> one AND (two OR three) * [one OR two three] ==> (one OR two) AND three * * A "-" before a term matches all entries that lack that term. * The "-" must occur immediately before the term with in intervening * space. This is how the search engines do it. * * A NOT term cannot be the right-hand operand of an OR. If this * occurs in the query string, the NOT is ignored: * * [one OR -two] ==> one OR two * */ typedef struct Query { fulltext_vtab *pFts; /* The full text index */ int nTerms; /* Number of terms in the query */ QueryTerm *pTerms; /* Array of terms. Space obtained from malloc() */ int nextIsOr; /* Set the isOr flag on the next inserted term */ int nextColumn; /* Next word parsed must be in this column */ int dfltColumn; /* The default column */ } Query; /* ** An instance of the following structure keeps track of generated ** matching-word offset information and snippets. */ typedef struct Snippet { int nMatch; /* Total number of matches */ int nAlloc; /* Space allocated for aMatch[] */ struct snippetMatch { /* One entry for each matching term */ char snStatus; /* Status flag for use while constructing snippets */ short int iCol; /* The column that contains the match */ short int iTerm; /* The index in Query.pTerms[] of the matching term */ short int nByte; /* Number of bytes in the term */ int iStart; /* The offset to the first character of the term */ } *aMatch; /* Points to space obtained from malloc */ char *zOffset; /* Text rendering of aMatch[] */ int nOffset; /* strlen(zOffset) */ char *zSnippet; /* Snippet text */ int nSnippet; /* strlen(zSnippet) */ } Snippet; typedef enum QueryType { QUERY_GENERIC, /* table scan */ QUERY_ROWID, /* lookup by rowid */ QUERY_FULLTEXT /* QUERY_FULLTEXT + [i] is a full-text search for column i*/ } QueryType; typedef enum fulltext_statement { CONTENT_INSERT_STMT, CONTENT_SELECT_STMT, CONTENT_UPDATE_STMT, CONTENT_DELETE_STMT, BLOCK_INSERT_STMT, BLOCK_SELECT_STMT, BLOCK_DELETE_STMT, SEGDIR_MAX_INDEX_STMT, SEGDIR_SET_STMT, SEGDIR_SELECT_STMT, SEGDIR_SPAN_STMT, SEGDIR_DELETE_STMT, SEGDIR_SELECT_ALL_STMT, MAX_STMT /* Always at end! */ } fulltext_statement; /* These must exactly match the enum above. */ /* TODO(shess): Is there some risk that a statement will be used in two ** cursors at once, e.g. if a query joins a virtual table to itself? ** If so perhaps we should move some of these to the cursor object. */ static const char *const fulltext_zStatement[MAX_STMT] = { /* CONTENT_INSERT */ NULL, /* generated in contentInsertStatement() */ /* CONTENT_SELECT */ "select * from %_content where rowid = ?", /* CONTENT_UPDATE */ NULL, /* generated in contentUpdateStatement() */ /* CONTENT_DELETE */ "delete from %_content where rowid = ?", /* BLOCK_INSERT */ "insert into %_segments values (?)", /* BLOCK_SELECT */ "select block from %_segments where rowid = ?", /* BLOCK_DELETE */ "delete from %_segments where rowid between ? and ?", /* SEGDIR_MAX_INDEX */ "select max(idx) from %_segdir where level = ?", /* SEGDIR_SET */ "insert into %_segdir values (?, ?, ?, ?, ?, ?)", /* SEGDIR_SELECT */ "select start_block, leaves_end_block, root from %_segdir " " where level = ? order by idx", /* SEGDIR_SPAN */ "select min(start_block), max(end_block) from %_segdir " " where level = ? and start_block <> 0", /* SEGDIR_DELETE */ "delete from %_segdir where level = ?", /* SEGDIR_SELECT_ALL */ "select root, leaves_end_block from %_segdir order by level desc, idx", }; /* ** A connection to a fulltext index is an instance of the following ** structure. The xCreate and xConnect methods create an instance ** of this structure and xDestroy and xDisconnect free that instance. ** All other methods receive a pointer to the structure as one of their ** arguments. */ struct fulltext_vtab { sqlite3_vtab base; /* Base class used by SQLite core */ sqlite3 *db; /* The database connection */ const char *zDb; /* logical database name */ const char *zName; /* virtual table name */ int nColumn; /* number of columns in virtual table */ char **azColumn; /* column names. malloced */ char **azContentColumn; /* column names in content table; malloced */ sqlite3_tokenizer *pTokenizer; /* tokenizer for inserts and queries */ /* Precompiled statements which we keep as long as the table is ** open. */ sqlite3_stmt *pFulltextStatements[MAX_STMT]; /* Precompiled statements used for segment merges. We run a ** separate select across the leaf level of each tree being merged. */ sqlite3_stmt *pLeafSelectStmts[MERGE_COUNT]; /* The statement used to prepare pLeafSelectStmts. */ #define LEAF_SELECT \ "select block from %_segments where rowid between ? and ? order by rowid" /* These buffer pending index updates during transactions. ** nPendingData estimates the memory size of the pending data. It ** doesn't include the hash-bucket overhead, nor any malloc ** overhead. When nPendingData exceeds kPendingThreshold, the ** buffer is flushed even before the transaction closes. ** pendingTerms stores the data, and is only valid when nPendingData ** is >=0 (nPendingData<0 means pendingTerms has not been ** initialized). iPrevDocid is the last docid written, used to make ** certain we're inserting in sorted order. */ int nPendingData; #define kPendingThreshold (1*1024*1024) sqlite_int64 iPrevDocid; fts3Hash pendingTerms; }; /* ** When the core wants to do a query, it create a cursor using a ** call to xOpen. This structure is an instance of a cursor. It ** is destroyed by xClose. */ typedef struct fulltext_cursor { sqlite3_vtab_cursor base; /* Base class used by SQLite core */ QueryType iCursorType; /* Copy of sqlite3_index_info.idxNum */ sqlite3_stmt *pStmt; /* Prepared statement in use by the cursor */ int eof; /* True if at End Of Results */ Query q; /* Parsed query string */ Snippet snippet; /* Cached snippet for the current row */ int iColumn; /* Column being searched */ DataBuffer result; /* Doclist results from fulltextQuery */ DLReader reader; /* Result reader if result not empty */ } fulltext_cursor; static struct fulltext_vtab *cursor_vtab(fulltext_cursor *c){ return (fulltext_vtab *) c->base.pVtab; } static const sqlite3_module fts3Module; /* forward declaration */ /* Return a dynamically generated statement of the form * insert into %_content (rowid, ...) values (?, ...) */ static const char *contentInsertStatement(fulltext_vtab *v){ StringBuffer sb; int i; initStringBuffer(&sb); append(&sb, "insert into %_content (rowid, "); appendList(&sb, v->nColumn, v->azContentColumn); append(&sb, ") values (?"); for(i=0; i<v->nColumn; ++i) append(&sb, ", ?"); append(&sb, ")"); return stringBufferData(&sb); } /* Return a dynamically generated statement of the form * update %_content set [col_0] = ?, [col_1] = ?, ... * where rowid = ? */ static const char *contentUpdateStatement(fulltext_vtab *v){ StringBuffer sb; int i; initStringBuffer(&sb); append(&sb, "update %_content set "); for(i=0; i<v->nColumn; ++i) { if( i>0 ){ append(&sb, ", "); } append(&sb, v->azContentColumn[i]); append(&sb, " = ?"); } append(&sb, " where rowid = ?"); return stringBufferData(&sb); } /* Puts a freshly-prepared statement determined by iStmt in *ppStmt. ** If the indicated statement has never been prepared, it is prepared ** and cached, otherwise the cached version is reset. */ static int sql_get_statement(fulltext_vtab *v, fulltext_statement iStmt, sqlite3_stmt **ppStmt){ assert( iStmt<MAX_STMT ); if( v->pFulltextStatements[iStmt]==NULL ){ const char *zStmt; int rc; switch( iStmt ){ case CONTENT_INSERT_STMT: zStmt = contentInsertStatement(v); break; case CONTENT_UPDATE_STMT: zStmt = contentUpdateStatement(v); break; default: zStmt = fulltext_zStatement[iStmt]; } rc = sql_prepare(v->db, v->zDb, v->zName, &v->pFulltextStatements[iStmt], zStmt); if( zStmt != fulltext_zStatement[iStmt]) free((void *) zStmt); if( rc!=SQLITE_OK ) return rc; } else { int rc = sqlite3_reset(v->pFulltextStatements[iStmt]); if( rc!=SQLITE_OK ) return rc; } *ppStmt = v->pFulltextStatements[iStmt]; return SQLITE_OK; } /* Like sqlite3_step(), but convert SQLITE_DONE to SQLITE_OK and ** SQLITE_ROW to SQLITE_ERROR. Useful for statements like UPDATE, ** where we expect no results. */ static int sql_single_step(sqlite3_stmt *s){ int rc = sqlite3_step(s); return (rc==SQLITE_DONE) ? SQLITE_OK : rc; } /* Like sql_get_statement(), but for special replicated LEAF_SELECT ** statements. */ /* TODO(shess) Write version for generic statements and then share ** that between the cached-statement functions. */ static int sql_get_leaf_statement(fulltext_vtab *v, int idx, sqlite3_stmt **ppStmt){ assert( idx>=0 && idx<MERGE_COUNT ); if( v->pLeafSelectStmts[idx]==NULL ){ int rc = sql_prepare(v->db, v->zDb, v->zName, &v->pLeafSelectStmts[idx], LEAF_SELECT); if( rc!=SQLITE_OK ) return rc; }else{ int rc = sqlite3_reset(v->pLeafSelectStmts[idx]); if( rc!=SQLITE_OK ) return rc; } *ppStmt = v->pLeafSelectStmts[idx]; return SQLITE_OK; } /* insert into %_content (rowid, ...) values ([rowid], [pValues]) */ static int content_insert(fulltext_vtab *v, sqlite3_value *rowid, sqlite3_value **pValues){ sqlite3_stmt *s; int i; int rc = sql_get_statement(v, CONTENT_INSERT_STMT, &s); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_bind_value(s, 1, rowid); if( rc!=SQLITE_OK ) return rc; for(i=0; i<v->nColumn; ++i){ rc = sqlite3_bind_value(s, 2+i, pValues[i]); if( rc!=SQLITE_OK ) return rc; } return sql_single_step(s); } /* update %_content set col0 = pValues[0], col1 = pValues[1], ... * where rowid = [iRowid] */ static int content_update(fulltext_vtab *v, sqlite3_value **pValues, sqlite_int64 iRowid){ sqlite3_stmt *s; int i; int rc = sql_get_statement(v, CONTENT_UPDATE_STMT, &s); if( rc!=SQLITE_OK ) return rc; for(i=0; i<v->nColumn; ++i){ rc = sqlite3_bind_value(s, 1+i, pValues[i]); if( rc!=SQLITE_OK ) return rc; } rc = sqlite3_bind_int64(s, 1+v->nColumn, iRowid); if( rc!=SQLITE_OK ) return rc; return sql_single_step(s); } static void freeStringArray(int nString, const char **pString){ int i; for (i=0 ; i < nString ; ++i) { if( pString[i]!=NULL ) free((void *) pString[i]); } free((void *) pString); } /* select * from %_content where rowid = [iRow] * The caller must delete the returned array and all strings in it. * null fields will be NULL in the returned array. * * TODO: Perhaps we should return pointer/length strings here for consistency * with other code which uses pointer/length. */ static int content_select(fulltext_vtab *v, sqlite_int64 iRow, const char ***pValues){ sqlite3_stmt *s; const char **values; int i; int rc; *pValues = NULL; rc = sql_get_statement(v, CONTENT_SELECT_STMT, &s); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_bind_int64(s, 1, iRow); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_step(s); if( rc!=SQLITE_ROW ) return rc; values = (const char **) malloc(v->nColumn * sizeof(const char *)); for(i=0; i<v->nColumn; ++i){ if( sqlite3_column_type(s, i)==SQLITE_NULL ){ values[i] = NULL; }else{ values[i] = string_dup((char*)sqlite3_column_text(s, i)); } } /* We expect only one row. We must execute another sqlite3_step() * to complete the iteration; otherwise the table will remain locked. */ rc = sqlite3_step(s); if( rc==SQLITE_DONE ){ *pValues = values; return SQLITE_OK; } freeStringArray(v->nColumn, values); return rc; } /* delete from %_content where rowid = [iRow ] */ static int content_delete(fulltext_vtab *v, sqlite_int64 iRow){ sqlite3_stmt *s; int rc = sql_get_statement(v, CONTENT_DELETE_STMT, &s); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_bind_int64(s, 1, iRow); if( rc!=SQLITE_OK ) return rc; return sql_single_step(s); } /* insert into %_segments values ([pData]) ** returns assigned rowid in *piBlockid */ static int block_insert(fulltext_vtab *v, const char *pData, int nData, sqlite_int64 *piBlockid){ sqlite3_stmt *s; int rc = sql_get_statement(v, BLOCK_INSERT_STMT, &s); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_bind_blob(s, 1, pData, nData, SQLITE_STATIC); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_step(s); if( rc==SQLITE_ROW ) return SQLITE_ERROR; if( rc!=SQLITE_DONE ) return rc; *piBlockid = sqlite3_last_insert_rowid(v->db); return SQLITE_OK; } /* delete from %_segments ** where rowid between [iStartBlockid] and [iEndBlockid] ** ** Deletes the range of blocks, inclusive, used to delete the blocks ** which form a segment. */ static int block_delete(fulltext_vtab *v, sqlite_int64 iStartBlockid, sqlite_int64 iEndBlockid){ sqlite3_stmt *s; int rc = sql_get_statement(v, BLOCK_DELETE_STMT, &s); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_bind_int64(s, 1, iStartBlockid); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_bind_int64(s, 2, iEndBlockid); if( rc!=SQLITE_OK ) return rc; return sql_single_step(s); } /* Returns SQLITE_ROW with *pidx set to the maximum segment idx found ** at iLevel. Returns SQLITE_DONE if there are no segments at ** iLevel. Otherwise returns an error. */ static int segdir_max_index(fulltext_vtab *v, int iLevel, int *pidx){ sqlite3_stmt *s; int rc = sql_get_statement(v, SEGDIR_MAX_INDEX_STMT, &s); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_bind_int(s, 1, iLevel); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_step(s); /* Should always get at least one row due to how max() works. */ if( rc==SQLITE_DONE ) return SQLITE_DONE; if( rc!=SQLITE_ROW ) return rc; /* NULL means that there were no inputs to max(). */ if( SQLITE_NULL==sqlite3_column_type(s, 0) ){ rc = sqlite3_step(s); if( rc==SQLITE_ROW ) return SQLITE_ERROR; return rc; } *pidx = sqlite3_column_int(s, 0); /* We expect only one row. We must execute another sqlite3_step() * to complete the iteration; otherwise the table will remain locked. */ rc = sqlite3_step(s); if( rc==SQLITE_ROW ) return SQLITE_ERROR; if( rc!=SQLITE_DONE ) return rc; return SQLITE_ROW; } /* insert into %_segdir values ( ** [iLevel], [idx], ** [iStartBlockid], [iLeavesEndBlockid], [iEndBlockid], ** [pRootData] ** ) */ static int segdir_set(fulltext_vtab *v, int iLevel, int idx, sqlite_int64 iStartBlockid, sqlite_int64 iLeavesEndBlockid, sqlite_int64 iEndBlockid, const char *pRootData, int nRootData){ sqlite3_stmt *s; int rc = sql_get_statement(v, SEGDIR_SET_STMT, &s); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_bind_int(s, 1, iLevel); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_bind_int(s, 2, idx); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_bind_int64(s, 3, iStartBlockid); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_bind_int64(s, 4, iLeavesEndBlockid); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_bind_int64(s, 5, iEndBlockid); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_bind_blob(s, 6, pRootData, nRootData, SQLITE_STATIC); if( rc!=SQLITE_OK ) return rc; return sql_single_step(s); } /* Queries %_segdir for the block span of the segments in level ** iLevel. Returns SQLITE_DONE if there are no blocks for iLevel, ** SQLITE_ROW if there are blocks, else an error. */ static int segdir_span(fulltext_vtab *v, int iLevel, sqlite_int64 *piStartBlockid, sqlite_int64 *piEndBlockid){ sqlite3_stmt *s; int rc = sql_get_statement(v, SEGDIR_SPAN_STMT, &s); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_bind_int(s, 1, iLevel); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_step(s); if( rc==SQLITE_DONE ) return SQLITE_DONE; /* Should never happen */ if( rc!=SQLITE_ROW ) return rc; /* This happens if all segments at this level are entirely inline. */ if( SQLITE_NULL==sqlite3_column_type(s, 0) ){ /* We expect only one row. We must execute another sqlite3_step() * to complete the iteration; otherwise the table will remain locked. */ int rc2 = sqlite3_step(s); if( rc2==SQLITE_ROW ) return SQLITE_ERROR; return rc2; } *piStartBlockid = sqlite3_column_int64(s, 0); *piEndBlockid = sqlite3_column_int64(s, 1); /* We expect only one row. We must execute another sqlite3_step() * to complete the iteration; otherwise the table will remain locked. */ rc = sqlite3_step(s); if( rc==SQLITE_ROW ) return SQLITE_ERROR; if( rc!=SQLITE_DONE ) return rc; return SQLITE_ROW; } /* Delete the segment blocks and segment directory records for all ** segments at iLevel. */ static int segdir_delete(fulltext_vtab *v, int iLevel){ sqlite3_stmt *s; sqlite_int64 iStartBlockid, iEndBlockid; int rc = segdir_span(v, iLevel, &iStartBlockid, &iEndBlockid); if( rc!=SQLITE_ROW && rc!=SQLITE_DONE ) return rc; if( rc==SQLITE_ROW ){ rc = block_delete(v, iStartBlockid, iEndBlockid); if( rc!=SQLITE_OK ) return rc; } /* Delete the segment directory itself. */ rc = sql_get_statement(v, SEGDIR_DELETE_STMT, &s); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_bind_int64(s, 1, iLevel); if( rc!=SQLITE_OK ) return rc; return sql_single_step(s); } /* TODO(shess) clearPendingTerms() is far down the file because ** writeZeroSegment() is far down the file because LeafWriter is far ** down the file. Consider refactoring the code to move the non-vtab ** code above the vtab code so that we don't need this forward ** reference. */ static int clearPendingTerms(fulltext_vtab *v); /* ** Free the memory used to contain a fulltext_vtab structure. */ static void fulltext_vtab_destroy(fulltext_vtab *v){ int iStmt, i; TRACE(("FTS3 Destroy %p\n", v)); for( iStmt=0; iStmt<MAX_STMT; iStmt++ ){ if( v->pFulltextStatements[iStmt]!=NULL ){ sqlite3_finalize(v->pFulltextStatements[iStmt]); v->pFulltextStatements[iStmt] = NULL; } } for( i=0; i<MERGE_COUNT; i++ ){ if( v->pLeafSelectStmts[i]!=NULL ){ sqlite3_finalize(v->pLeafSelectStmts[i]); v->pLeafSelectStmts[i] = NULL; } } if( v->pTokenizer!=NULL ){ v->pTokenizer->pModule->xDestroy(v->pTokenizer); v->pTokenizer = NULL; } clearPendingTerms(v); free(v->azColumn); for(i = 0; i < v->nColumn; ++i) { sqlite3_free(v->azContentColumn[i]); } free(v->azContentColumn); free(v); } /* ** Token types for parsing the arguments to xConnect or xCreate. */ #define TOKEN_EOF 0 /* End of file */ #define TOKEN_SPACE 1 /* Any kind of whitespace */ #define TOKEN_ID 2 /* An identifier */ #define TOKEN_STRING 3 /* A string literal */ #define TOKEN_PUNCT 4 /* A single punctuation character */ /* ** If X is a character that can be used in an identifier then ** IdChar(X) will be true. Otherwise it is false. ** ** For ASCII, any character with the high-order bit set is ** allowed in an identifier. For 7-bit characters, ** sqlite3IsIdChar[X] must be 1. ** ** Ticket #1066. the SQL standard does not allow '$' in the ** middle of identfiers. But many SQL implementations do. ** SQLite will allow '$' in identifiers for compatibility. ** But the feature is undocumented. */ static const char isIdChar[] = { /* x0 x1 x2 x3 x4 x5 x6 x7 x8 x9 xA xB xC xD xE xF */ 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 2x */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, /* 3x */ 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 4x */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, /* 5x */ 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 6x */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, /* 7x */ }; #define IdChar(C) (((c=C)&0x80)!=0 || (c>0x1f && isIdChar[c-0x20])) /* ** Return the length of the token that begins at z[0]. ** Store the token type in *tokenType before returning. */ static int getToken(const char *z, int *tokenType){ int i, c; switch( *z ){ case 0: { *tokenType = TOKEN_EOF; return 0; } case ' ': case '\t': case '\n': case '\f': case '\r': { for(i=1; safe_isspace(z[i]); i++){} *tokenType = TOKEN_SPACE; return i; } case '`': case '\'': case '"': { int delim = z[0]; for(i=1; (c=z[i])!=0; i++){ if( c==delim ){ if( z[i+1]==delim ){ i++; }else{ break; } } } *tokenType = TOKEN_STRING; return i + (c!=0); } case '[': { for(i=1, c=z[0]; c!=']' && (c=z[i])!=0; i++){} *tokenType = TOKEN_ID; return i; } default: { if( !IdChar(*z) ){ break; } for(i=1; IdChar(z[i]); i++){} *tokenType = TOKEN_ID; return i; } } *tokenType = TOKEN_PUNCT; return 1; } /* ** A token extracted from a string is an instance of the following ** structure. */ typedef struct Token { const char *z; /* Pointer to token text. Not '\000' terminated */ short int n; /* Length of the token text in bytes. */ } Token; /* ** Given a input string (which is really one of the argv[] parameters ** passed into xConnect or xCreate) split the string up into tokens. ** Return an array of pointers to '\000' terminated strings, one string ** for each non-whitespace token. ** ** The returned array is terminated by a single NULL pointer. ** ** Space to hold the returned array is obtained from a single ** malloc and should be freed by passing the return value to free(). ** The individual strings within the token list are all a part of ** the single memory allocation and will all be freed at once. */ static char **tokenizeString(const char *z, int *pnToken){ int nToken = 0; Token *aToken = malloc( strlen(z) * sizeof(aToken[0]) ); int n = 1; int e, i; int totalSize = 0; char **azToken; char *zCopy; while( n>0 ){ n = getToken(z, &e); if( e!=TOKEN_SPACE ){ aToken[nToken].z = z; aToken[nToken].n = n; nToken++; totalSize += n+1; } z += n; } azToken = (char**)malloc( nToken*sizeof(char*) + totalSize ); zCopy = (char*)&azToken[nToken]; nToken--; for(i=0; i<nToken; i++){ azToken[i] = zCopy; n = aToken[i].n; memcpy(zCopy, aToken[i].z, n); zCopy[n] = 0; zCopy += n+1; } azToken[nToken] = 0; free(aToken); *pnToken = nToken; return azToken; } /* ** Convert an SQL-style quoted string into a normal string by removing ** the quote characters. The conversion is done in-place. If the ** input does not begin with a quote character, then this routine ** is a no-op. ** ** Examples: ** ** "abc" becomes abc ** 'xyz' becomes xyz ** [pqr] becomes pqr ** `mno` becomes mno */ static void dequoteString(char *z){ int quote; int i, j; if( z==0 ) return; quote = z[0]; switch( quote ){ case '\'': break; case '"': break; case '`': break; /* For MySQL compatibility */ case '[': quote = ']'; break; /* For MS SqlServer compatibility */ default: return; } for(i=1, j=0; z[i]; i++){ if( z[i]==quote ){ if( z[i+1]==quote ){ z[j++] = quote; i++; }else{ z[j++] = 0; break; } }else{ z[j++] = z[i]; } } } /* ** The input azIn is a NULL-terminated list of tokens. Remove the first ** token and all punctuation tokens. Remove the quotes from ** around string literal tokens. ** ** Example: ** ** input: tokenize chinese ( 'simplifed' , 'mixed' ) ** output: chinese simplifed mixed ** ** Another example: ** ** input: delimiters ( '[' , ']' , '...' ) ** output: [ ] ... */ static void tokenListToIdList(char **azIn){ int i, j; if( azIn ){ for(i=0, j=-1; azIn[i]; i++){ if( safe_isalnum(azIn[i][0]) || azIn[i][1] ){ dequoteString(azIn[i]); if( j>=0 ){ azIn[j] = azIn[i]; } j++; } } azIn[j] = 0; } } /* ** Find the first alphanumeric token in the string zIn. Null-terminate ** this token. Remove any quotation marks. And return a pointer to ** the result. */ static char *firstToken(char *zIn, char **pzTail){ int n, ttype; while(1){ n = getToken(zIn, &ttype); if( ttype==TOKEN_SPACE ){ zIn += n; }else if( ttype==TOKEN_EOF ){ *pzTail = zIn; return 0; }else{ zIn[n] = 0; *pzTail = &zIn[1]; dequoteString(zIn); return zIn; } } /*NOTREACHED*/ } /* Return true if... ** ** * s begins with the string t, ignoring case ** * s is longer than t ** * The first character of s beyond t is not a alphanumeric ** ** Ignore leading space in *s. ** ** To put it another way, return true if the first token of ** s[] is t[]. */ static int startsWith(const char *s, const char *t){ while( safe_isspace(*s) ){ s++; } while( *t ){ if( safe_tolower(*s++)!=safe_tolower(*t++) ) return 0; } return *s!='_' && !safe_isalnum(*s); } /* ** An instance of this structure defines the "spec" of a ** full text index. This structure is populated by parseSpec ** and use by fulltextConnect and fulltextCreate. */ typedef struct TableSpec { const char *zDb; /* Logical database name */ const char *zName; /* Name of the full-text index */ int nColumn; /* Number of columns to be indexed */ char **azColumn; /* Original names of columns to be indexed */ char **azContentColumn; /* Column names for %_content */ char **azTokenizer; /* Name of tokenizer and its arguments */ } TableSpec; /* ** Reclaim all of the memory used by a TableSpec */ static void clearTableSpec(TableSpec *p) { free(p->azColumn); free(p->azContentColumn); free(p->azTokenizer); } /* Parse a CREATE VIRTUAL TABLE statement, which looks like this: * * CREATE VIRTUAL TABLE email * USING fts3(subject, body, tokenize mytokenizer(myarg)) * * We return parsed information in a TableSpec structure. * */ static int parseSpec(TableSpec *pSpec, int argc, const char *const*argv, char**pzErr){ int i, n; char *z, *zDummy; char **azArg; const char *zTokenizer = 0; /* argv[] entry describing the tokenizer */ assert( argc>=3 ); /* Current interface: ** argv[0] - module name ** argv[1] - database name ** argv[2] - table name ** argv[3..] - columns, optionally followed by tokenizer specification ** and snippet delimiters specification. */ /* Make a copy of the complete argv[][] array in a single allocation. ** The argv[][] array is read-only and transient. We can write to the ** copy in order to modify things and the copy is persistent. */ CLEAR(pSpec); for(i=n=0; i<argc; i++){ n += strlen(argv[i]) + 1; } azArg = malloc( sizeof(char*)*argc + n ); if( azArg==0 ){ return SQLITE_NOMEM; } z = (char*)&azArg[argc]; for(i=0; i<argc; i++){ azArg[i] = z; strcpy(z, argv[i]); z += strlen(z)+1; } /* Identify the column names and the tokenizer and delimiter arguments ** in the argv[][] array. */ pSpec->zDb = azArg[1]; pSpec->zName = azArg[2]; pSpec->nColumn = 0; pSpec->azColumn = azArg; zTokenizer = "tokenize simple"; for(i=3; i<argc; ++i){ if( startsWith(azArg[i],"tokenize") ){ zTokenizer = azArg[i]; }else{ z = azArg[pSpec->nColumn] = firstToken(azArg[i], &zDummy); pSpec->nColumn++; } } if( pSpec->nColumn==0 ){ azArg[0] = "content"; pSpec->nColumn = 1; } /* ** Construct the list of content column names. ** ** Each content column name will be of the form cNNAAAA ** where NN is the column number and AAAA is the sanitized ** column name. "sanitized" means that special characters are ** converted to "_". The cNN prefix guarantees that all column ** names are unique. ** ** The AAAA suffix is not strictly necessary. It is included ** for the convenience of people who might examine the generated ** %_content table and wonder what the columns are used for. */ pSpec->azContentColumn = malloc( pSpec->nColumn * sizeof(char *) ); if( pSpec->azContentColumn==0 ){ clearTableSpec(pSpec); return SQLITE_NOMEM; } for(i=0; i<pSpec->nColumn; i++){ char *p; pSpec->azContentColumn[i] = sqlite3_mprintf("c%d%s", i, azArg[i]); for (p = pSpec->azContentColumn[i]; *p ; ++p) { if( !safe_isalnum(*p) ) *p = '_'; } } /* ** Parse the tokenizer specification string. */ pSpec->azTokenizer = tokenizeString(zTokenizer, &n); tokenListToIdList(pSpec->azTokenizer); return SQLITE_OK; } /* ** Generate a CREATE TABLE statement that describes the schema of ** the virtual table. Return a pointer to this schema string. ** ** Space is obtained from sqlite3_mprintf() and should be freed ** using sqlite3_free(). */ static char *fulltextSchema( int nColumn, /* Number of columns */ const char *const* azColumn, /* List of columns */ const char *zTableName /* Name of the table */ ){ int i; char *zSchema, *zNext; const char *zSep = "("; zSchema = sqlite3_mprintf("CREATE TABLE x"); for(i=0; i<nColumn; i++){ zNext = sqlite3_mprintf("%s%s%Q", zSchema, zSep, azColumn[i]); sqlite3_free(zSchema); zSchema = zNext; zSep = ","; } zNext = sqlite3_mprintf("%s,%Q)", zSchema, zTableName); sqlite3_free(zSchema); return zNext; } /* ** Build a new sqlite3_vtab structure that will describe the ** fulltext index defined by spec. */ static int constructVtab( sqlite3 *db, /* The SQLite database connection */ fts3Hash *pHash, /* Hash table containing tokenizers */ TableSpec *spec, /* Parsed spec information from parseSpec() */ sqlite3_vtab **ppVTab, /* Write the resulting vtab structure here */ char **pzErr /* Write any error message here */ ){ int rc; int n; fulltext_vtab *v = 0; const sqlite3_tokenizer_module *m = NULL; char *schema; char const *zTok; /* Name of tokenizer to use for this fts table */ int nTok; /* Length of zTok, including nul terminator */ v = (fulltext_vtab *) malloc(sizeof(fulltext_vtab)); if( v==0 ) return SQLITE_NOMEM; CLEAR(v); /* sqlite will initialize v->base */ v->db = db; v->zDb = spec->zDb; /* Freed when azColumn is freed */ v->zName = spec->zName; /* Freed when azColumn is freed */ v->nColumn = spec->nColumn; v->azContentColumn = spec->azContentColumn; spec->azContentColumn = 0; v->azColumn = spec->azColumn; spec->azColumn = 0; if( spec->azTokenizer==0 ){ return SQLITE_NOMEM; } zTok = spec->azTokenizer[0]; if( !zTok ){ zTok = "simple"; } nTok = strlen(zTok)+1; m = (sqlite3_tokenizer_module *)sqlite3Fts3HashFind(pHash, zTok, nTok); if( !m ){ *pzErr = sqlite3_mprintf("unknown tokenizer: %s", spec->azTokenizer[0]); rc = SQLITE_ERROR; goto err; } for(n=0; spec->azTokenizer[n]; n++){} if( n ){ rc = m->xCreate(n-1, (const char*const*)&spec->azTokenizer[1], &v->pTokenizer); }else{ rc = m->xCreate(0, 0, &v->pTokenizer); } if( rc!=SQLITE_OK ) goto err; v->pTokenizer->pModule = m; /* TODO: verify the existence of backing tables foo_content, foo_term */ schema = fulltextSchema(v->nColumn, (const char*const*)v->azColumn, spec->zName); rc = sqlite3_declare_vtab(db, schema); sqlite3_free(schema); if( rc!=SQLITE_OK ) goto err; memset(v->pFulltextStatements, 0, sizeof(v->pFulltextStatements)); /* Indicate that the buffer is not live. */ v->nPendingData = -1; *ppVTab = &v->base; TRACE(("FTS3 Connect %p\n", v)); return rc; err: fulltext_vtab_destroy(v); return rc; } static int fulltextConnect( sqlite3 *db, void *pAux, int argc, const char *const*argv, sqlite3_vtab **ppVTab, char **pzErr ){ TableSpec spec; int rc = parseSpec(&spec, argc, argv, pzErr); if( rc!=SQLITE_OK ) return rc; rc = constructVtab(db, (fts3Hash *)pAux, &spec, ppVTab, pzErr); clearTableSpec(&spec); return rc; } /* The %_content table holds the text of each document, with ** the rowid used as the docid. */ /* TODO(shess) This comment needs elaboration to match the updated ** code. Work it into the top-of-file comment at that time. */ static int fulltextCreate(sqlite3 *db, void *pAux, int argc, const char * const *argv, sqlite3_vtab **ppVTab, char **pzErr){ int rc; TableSpec spec; StringBuffer schema; TRACE(("FTS3 Create\n")); rc = parseSpec(&spec, argc, argv, pzErr); if( rc!=SQLITE_OK ) return rc; initStringBuffer(&schema); append(&schema, "CREATE TABLE %_content("); appendList(&schema, spec.nColumn, spec.azContentColumn); append(&schema, ")"); rc = sql_exec(db, spec.zDb, spec.zName, stringBufferData(&schema)); stringBufferDestroy(&schema); if( rc!=SQLITE_OK ) goto out; rc = sql_exec(db, spec.zDb, spec.zName, "create table %_segments(block blob);"); if( rc!=SQLITE_OK ) goto out; rc = sql_exec(db, spec.zDb, spec.zName, "create table %_segdir(" " level integer," " idx integer," " start_block integer," " leaves_end_block integer," " end_block integer," " root blob," " primary key(level, idx)" ");"); if( rc!=SQLITE_OK ) goto out; rc = constructVtab(db, (fts3Hash *)pAux, &spec, ppVTab, pzErr); out: clearTableSpec(&spec); return rc; } /* Decide how to handle an SQL query. */ static int fulltextBestIndex(sqlite3_vtab *pVTab, sqlite3_index_info *pInfo){ int i; TRACE(("FTS3 BestIndex\n")); for(i=0; i<pInfo->nConstraint; ++i){ const struct sqlite3_index_constraint *pConstraint; pConstraint = &pInfo->aConstraint[i]; if( pConstraint->usable ) { if( pConstraint->iColumn==-1 && pConstraint->op==SQLITE_INDEX_CONSTRAINT_EQ ){ pInfo->idxNum = QUERY_ROWID; /* lookup by rowid */ TRACE(("FTS3 QUERY_ROWID\n")); } else if( pConstraint->iColumn>=0 && pConstraint->op==SQLITE_INDEX_CONSTRAINT_MATCH ){ /* full-text search */ pInfo->idxNum = QUERY_FULLTEXT + pConstraint->iColumn; TRACE(("FTS3 QUERY_FULLTEXT %d\n", pConstraint->iColumn)); } else continue; pInfo->aConstraintUsage[i].argvIndex = 1; pInfo->aConstraintUsage[i].omit = 1; /* An arbitrary value for now. * TODO: Perhaps rowid matches should be considered cheaper than * full-text searches. */ pInfo->estimatedCost = 1.0; return SQLITE_OK; } } pInfo->idxNum = QUERY_GENERIC; return SQLITE_OK; } static int fulltextDisconnect(sqlite3_vtab *pVTab){ TRACE(("FTS3 Disconnect %p\n", pVTab)); fulltext_vtab_destroy((fulltext_vtab *)pVTab); return SQLITE_OK; } static int fulltextDestroy(sqlite3_vtab *pVTab){ fulltext_vtab *v = (fulltext_vtab *)pVTab; int rc; TRACE(("FTS3 Destroy %p\n", pVTab)); rc = sql_exec(v->db, v->zDb, v->zName, "drop table if exists %_content;" "drop table if exists %_segments;" "drop table if exists %_segdir;" ); if( rc!=SQLITE_OK ) return rc; fulltext_vtab_destroy((fulltext_vtab *)pVTab); return SQLITE_OK; } static int fulltextOpen(sqlite3_vtab *pVTab, sqlite3_vtab_cursor **ppCursor){ fulltext_cursor *c; c = (fulltext_cursor *) calloc(sizeof(fulltext_cursor), 1); /* sqlite will initialize c->base */ *ppCursor = &c->base; TRACE(("FTS3 Open %p: %p\n", pVTab, c)); return SQLITE_OK; } /* Free all of the dynamically allocated memory held by *q */ static void queryClear(Query *q){ int i; for(i = 0; i < q->nTerms; ++i){ free(q->pTerms[i].pTerm); } free(q->pTerms); CLEAR(q); } /* Free all of the dynamically allocated memory held by the ** Snippet */ static void snippetClear(Snippet *p){ free(p->aMatch); free(p->zOffset); free(p->zSnippet); CLEAR(p); } /* ** Append a single entry to the p->aMatch[] log. */ static void snippetAppendMatch( Snippet *p, /* Append the entry to this snippet */ int iCol, int iTerm, /* The column and query term */ int iStart, int nByte /* Offset and size of the match */ ){ int i; struct snippetMatch *pMatch; if( p->nMatch+1>=p->nAlloc ){ p->nAlloc = p->nAlloc*2 + 10; p->aMatch = realloc(p->aMatch, p->nAlloc*sizeof(p->aMatch[0]) ); if( p->aMatch==0 ){ p->nMatch = 0; p->nAlloc = 0; return; } } i = p->nMatch++; pMatch = &p->aMatch[i]; pMatch->iCol = iCol; pMatch->iTerm = iTerm; pMatch->iStart = iStart; pMatch->nByte = nByte; } /* ** Sizing information for the circular buffer used in snippetOffsetsOfColumn() */ #define FTS3_ROTOR_SZ (32) #define FTS3_ROTOR_MASK (FTS3_ROTOR_SZ-1) /* ** Add entries to pSnippet->aMatch[] for every match that occurs against ** document zDoc[0..nDoc-1] which is stored in column iColumn. */ static void snippetOffsetsOfColumn( Query *pQuery, Snippet *pSnippet, int iColumn, const char *zDoc, int nDoc ){ const sqlite3_tokenizer_module *pTModule; /* The tokenizer module */ sqlite3_tokenizer *pTokenizer; /* The specific tokenizer */ sqlite3_tokenizer_cursor *pTCursor; /* Tokenizer cursor */ fulltext_vtab *pVtab; /* The full text index */ int nColumn; /* Number of columns in the index */ const QueryTerm *aTerm; /* Query string terms */ int nTerm; /* Number of query string terms */ int i, j; /* Loop counters */ int rc; /* Return code */ unsigned int match, prevMatch; /* Phrase search bitmasks */ const char *zToken; /* Next token from the tokenizer */ int nToken; /* Size of zToken */ int iBegin, iEnd, iPos; /* Offsets of beginning and end */ /* The following variables keep a circular buffer of the last ** few tokens */ unsigned int iRotor = 0; /* Index of current token */ int iRotorBegin[FTS3_ROTOR_SZ]; /* Beginning offset of token */ int iRotorLen[FTS3_ROTOR_SZ]; /* Length of token */ pVtab = pQuery->pFts; nColumn = pVtab->nColumn; pTokenizer = pVtab->pTokenizer; pTModule = pTokenizer->pModule; rc = pTModule->xOpen(pTokenizer, zDoc, nDoc, &pTCursor); if( rc ) return; pTCursor->pTokenizer = pTokenizer; aTerm = pQuery->pTerms; nTerm = pQuery->nTerms; if( nTerm>=FTS3_ROTOR_SZ ){ nTerm = FTS3_ROTOR_SZ - 1; } prevMatch = 0; while(1){ rc = pTModule->xNext(pTCursor, &zToken, &nToken, &iBegin, &iEnd, &iPos); if( rc ) break; iRotorBegin[iRotor&FTS3_ROTOR_MASK] = iBegin; iRotorLen[iRotor&FTS3_ROTOR_MASK] = iEnd-iBegin; match = 0; for(i=0; i<nTerm; i++){ int iCol; iCol = aTerm[i].iColumn; if( iCol>=0 && iCol<nColumn && iCol!=iColumn ) continue; if( aTerm[i].nTerm>nToken ) continue; if( !aTerm[i].isPrefix && aTerm[i].nTerm<nToken ) continue; assert( aTerm[i].nTerm<=nToken ); if( memcmp(aTerm[i].pTerm, zToken, aTerm[i].nTerm) ) continue; if( aTerm[i].iPhrase>1 && (prevMatch & (1<<i))==0 ) continue; match |= 1<<i; if( i==nTerm-1 || aTerm[i+1].iPhrase==1 ){ for(j=aTerm[i].iPhrase-1; j>=0; j--){ int k = (iRotor-j) & FTS3_ROTOR_MASK; snippetAppendMatch(pSnippet, iColumn, i-j, iRotorBegin[k], iRotorLen[k]); } } } prevMatch = match<<1; iRotor++; } pTModule->xClose(pTCursor); } /* ** Compute all offsets for the current row of the query. ** If the offsets have already been computed, this routine is a no-op. */ static void snippetAllOffsets(fulltext_cursor *p){ int nColumn; int iColumn, i; int iFirst, iLast; fulltext_vtab *pFts; if( p->snippet.nMatch ) return; if( p->q.nTerms==0 ) return; pFts = p->q.pFts; nColumn = pFts->nColumn; iColumn = (p->iCursorType - QUERY_FULLTEXT); if( iColumn<0 || iColumn>=nColumn ){ iFirst = 0; iLast = nColumn-1; }else{ iFirst = iColumn; iLast = iColumn; } for(i=iFirst; i<=iLast; i++){ const char *zDoc; int nDoc; zDoc = (const char*)sqlite3_column_text(p->pStmt, i+1); nDoc = sqlite3_column_bytes(p->pStmt, i+1); snippetOffsetsOfColumn(&p->q, &p->snippet, i, zDoc, nDoc); } } /* ** Convert the information in the aMatch[] array of the snippet ** into the string zOffset[0..nOffset-1]. */ static void snippetOffsetText(Snippet *p){ int i; int cnt = 0; StringBuffer sb; char zBuf[200]; if( p->zOffset ) return; initStringBuffer(&sb); for(i=0; i<p->nMatch; i++){ struct snippetMatch *pMatch = &p->aMatch[i]; zBuf[0] = ' '; sprintf(&zBuf[cnt>0], "%d %d %d %d", pMatch->iCol, pMatch->iTerm, pMatch->iStart, pMatch->nByte); append(&sb, zBuf); cnt++; } p->zOffset = stringBufferData(&sb); p->nOffset = stringBufferLength(&sb); } /* ** zDoc[0..nDoc-1] is phrase of text. aMatch[0..nMatch-1] are a set ** of matching words some of which might be in zDoc. zDoc is column ** number iCol. ** ** iBreak is suggested spot in zDoc where we could begin or end an ** excerpt. Return a value similar to iBreak but possibly adjusted ** to be a little left or right so that the break point is better. */ static int wordBoundary( int iBreak, /* The suggested break point */ const char *zDoc, /* Document text */ int nDoc, /* Number of bytes in zDoc[] */ struct snippetMatch *aMatch, /* Matching words */ int nMatch, /* Number of entries in aMatch[] */ int iCol /* The column number for zDoc[] */ ){ int i; if( iBreak<=10 ){ return 0; } if( iBreak>=nDoc-10 ){ return nDoc; } for(i=0; i<nMatch && aMatch[i].iCol<iCol; i++){} while( i<nMatch && aMatch[i].iStart+aMatch[i].nByte<iBreak ){ i++; } if( i<nMatch ){ if( aMatch[i].iStart<iBreak+10 ){ return aMatch[i].iStart; } if( i>0 && aMatch[i-1].iStart+aMatch[i-1].nByte>=iBreak ){ return aMatch[i-1].iStart; } } for(i=1; i<=10; i++){ if( safe_isspace(zDoc[iBreak-i]) ){ return iBreak - i + 1; } if( safe_isspace(zDoc[iBreak+i]) ){ return iBreak + i + 1; } } return iBreak; } /* ** Allowed values for Snippet.aMatch[].snStatus */ #define SNIPPET_IGNORE 0 /* It is ok to omit this match from the snippet */ #define SNIPPET_DESIRED 1 /* We want to include this match in the snippet */ /* ** Generate the text of a snippet. */ static void snippetText( fulltext_cursor *pCursor, /* The cursor we need the snippet for */ const char *zStartMark, /* Markup to appear before each match */ const char *zEndMark, /* Markup to appear after each match */ const char *zEllipsis /* Ellipsis mark */ ){ int i, j; struct snippetMatch *aMatch; int nMatch; int nDesired; StringBuffer sb; int tailCol; int tailOffset; int iCol; int nDoc; const char *zDoc; int iStart, iEnd; int tailEllipsis = 0; int iMatch; free(pCursor->snippet.zSnippet); pCursor->snippet.zSnippet = 0; aMatch = pCursor->snippet.aMatch; nMatch = pCursor->snippet.nMatch; initStringBuffer(&sb); for(i=0; i<nMatch; i++){ aMatch[i].snStatus = SNIPPET_IGNORE; } nDesired = 0; for(i=0; i<pCursor->q.nTerms; i++){ for(j=0; j<nMatch; j++){ if( aMatch[j].iTerm==i ){ aMatch[j].snStatus = SNIPPET_DESIRED; nDesired++; break; } } } iMatch = 0; tailCol = -1; tailOffset = 0; for(i=0; i<nMatch && nDesired>0; i++){ if( aMatch[i].snStatus!=SNIPPET_DESIRED ) continue; nDesired--; iCol = aMatch[i].iCol; zDoc = (const char*)sqlite3_column_text(pCursor->pStmt, iCol+1); nDoc = sqlite3_column_bytes(pCursor->pStmt, iCol+1); iStart = aMatch[i].iStart - 40; iStart = wordBoundary(iStart, zDoc, nDoc, aMatch, nMatch, iCol); if( iStart<=10 ){ iStart = 0; } if( iCol==tailCol && iStart<=tailOffset+20 ){ iStart = tailOffset; } if( (iCol!=tailCol && tailCol>=0) || iStart!=tailOffset ){ trimWhiteSpace(&sb); appendWhiteSpace(&sb); append(&sb, zEllipsis); appendWhiteSpace(&sb); } iEnd = aMatch[i].iStart + aMatch[i].nByte + 40; iEnd = wordBoundary(iEnd, zDoc, nDoc, aMatch, nMatch, iCol); if( iEnd>=nDoc-10 ){ iEnd = nDoc; tailEllipsis = 0; }else{ tailEllipsis = 1; } while( iMatch<nMatch && aMatch[iMatch].iCol<iCol ){ iMatch++; } while( iStart<iEnd ){ while( iMatch<nMatch && aMatch[iMatch].iStart<iStart && aMatch[iMatch].iCol<=iCol ){ iMatch++; } if( iMatch<nMatch && aMatch[iMatch].iStart<iEnd && aMatch[iMatch].iCol==iCol ){ nappend(&sb, &zDoc[iStart], aMatch[iMatch].iStart - iStart); iStart = aMatch[iMatch].iStart; append(&sb, zStartMark); nappend(&sb, &zDoc[iStart], aMatch[iMatch].nByte); append(&sb, zEndMark); iStart += aMatch[iMatch].nByte; for(j=iMatch+1; j<nMatch; j++){ if( aMatch[j].iTerm==aMatch[iMatch].iTerm && aMatch[j].snStatus==SNIPPET_DESIRED ){ nDesired--; aMatch[j].snStatus = SNIPPET_IGNORE; } } }else{ nappend(&sb, &zDoc[iStart], iEnd - iStart); iStart = iEnd; } } tailCol = iCol; tailOffset = iEnd; } trimWhiteSpace(&sb); if( tailEllipsis ){ appendWhiteSpace(&sb); append(&sb, zEllipsis); } pCursor->snippet.zSnippet = stringBufferData(&sb); pCursor->snippet.nSnippet = stringBufferLength(&sb); } /* ** Close the cursor. For additional information see the documentation ** on the xClose method of the virtual table interface. */ static int fulltextClose(sqlite3_vtab_cursor *pCursor){ fulltext_cursor *c = (fulltext_cursor *) pCursor; TRACE(("FTS3 Close %p\n", c)); sqlite3_finalize(c->pStmt); queryClear(&c->q); snippetClear(&c->snippet); if( c->result.nData!=0 ) dlrDestroy(&c->reader); dataBufferDestroy(&c->result); free(c); return SQLITE_OK; } static int fulltextNext(sqlite3_vtab_cursor *pCursor){ fulltext_cursor *c = (fulltext_cursor *) pCursor; int rc; TRACE(("FTS3 Next %p\n", pCursor)); snippetClear(&c->snippet); if( c->iCursorType < QUERY_FULLTEXT ){ /* TODO(shess) Handle SQLITE_SCHEMA AND SQLITE_BUSY. */ rc = sqlite3_step(c->pStmt); switch( rc ){ case SQLITE_ROW: c->eof = 0; return SQLITE_OK; case SQLITE_DONE: c->eof = 1; return SQLITE_OK; default: c->eof = 1; return rc; } } else { /* full-text query */ rc = sqlite3_reset(c->pStmt); if( rc!=SQLITE_OK ) return rc; if( c->result.nData==0 || dlrAtEnd(&c->reader) ){ c->eof = 1; return SQLITE_OK; } rc = sqlite3_bind_int64(c->pStmt, 1, dlrDocid(&c->reader)); dlrStep(&c->reader); if( rc!=SQLITE_OK ) return rc; /* TODO(shess) Handle SQLITE_SCHEMA AND SQLITE_BUSY. */ rc = sqlite3_step(c->pStmt); if( rc==SQLITE_ROW ){ /* the case we expect */ c->eof = 0; return SQLITE_OK; } /* an error occurred; abort */ return rc==SQLITE_DONE ? SQLITE_ERROR : rc; } } /* TODO(shess) If we pushed LeafReader to the top of the file, or to ** another file, term_select() could be pushed above ** docListOfTerm(). */ static int termSelect(fulltext_vtab *v, int iColumn, const char *pTerm, int nTerm, int isPrefix, DocListType iType, DataBuffer *out); /* Return a DocList corresponding to the query term *pTerm. If *pTerm ** is the first term of a phrase query, go ahead and evaluate the phrase ** query and return the doclist for the entire phrase query. ** ** The resulting DL_DOCIDS doclist is stored in pResult, which is ** overwritten. */ static int docListOfTerm( fulltext_vtab *v, /* The full text index */ int iColumn, /* column to restrict to. No restriction if >=nColumn */ QueryTerm *pQTerm, /* Term we are looking for, or 1st term of a phrase */ DataBuffer *pResult /* Write the result here */ ){ DataBuffer left, right, new; int i, rc; /* No phrase search if no position info. */ assert( pQTerm->nPhrase==0 || DL_DEFAULT!=DL_DOCIDS ); /* This code should never be called with buffered updates. */ assert( v->nPendingData<0 ); dataBufferInit(&left, 0); rc = termSelect(v, iColumn, pQTerm->pTerm, pQTerm->nTerm, pQTerm->isPrefix, 0<pQTerm->nPhrase ? DL_POSITIONS : DL_DOCIDS, &left); if( rc ) return rc; for(i=1; i<=pQTerm->nPhrase && left.nData>0; i++){ dataBufferInit(&right, 0); rc = termSelect(v, iColumn, pQTerm[i].pTerm, pQTerm[i].nTerm, pQTerm[i].isPrefix, DL_POSITIONS, &right); if( rc ){ dataBufferDestroy(&left); return rc; } dataBufferInit(&new, 0); docListPhraseMerge(left.pData, left.nData, right.pData, right.nData, i<pQTerm->nPhrase ? DL_POSITIONS : DL_DOCIDS, &new); dataBufferDestroy(&left); dataBufferDestroy(&right); left = new; } *pResult = left; return SQLITE_OK; } /* Add a new term pTerm[0..nTerm-1] to the query *q. */ static void queryAdd(Query *q, const char *pTerm, int nTerm){ QueryTerm *t; ++q->nTerms; q->pTerms = realloc(q->pTerms, q->nTerms * sizeof(q->pTerms[0])); if( q->pTerms==0 ){ q->nTerms = 0; return; } t = &q->pTerms[q->nTerms - 1]; CLEAR(t); t->pTerm = malloc(nTerm+1); memcpy(t->pTerm, pTerm, nTerm); t->pTerm[nTerm] = 0; t->nTerm = nTerm; t->isOr = q->nextIsOr; t->isPrefix = 0; q->nextIsOr = 0; t->iColumn = q->nextColumn; q->nextColumn = q->dfltColumn; } /* ** Check to see if the string zToken[0...nToken-1] matches any ** column name in the virtual table. If it does, ** return the zero-indexed column number. If not, return -1. */ static int checkColumnSpecifier( fulltext_vtab *pVtab, /* The virtual table */ const char *zToken, /* Text of the token */ int nToken /* Number of characters in the token */ ){ int i; for(i=0; i<pVtab->nColumn; i++){ if( memcmp(pVtab->azColumn[i], zToken, nToken)==0 && pVtab->azColumn[i][nToken]==0 ){ return i; } } return -1; } /* ** Parse the text at pSegment[0..nSegment-1]. Add additional terms ** to the query being assemblied in pQuery. ** ** inPhrase is true if pSegment[0..nSegement-1] is contained within ** double-quotes. If inPhrase is true, then the first term ** is marked with the number of terms in the phrase less one and ** OR and "-" syntax is ignored. If inPhrase is false, then every ** term found is marked with nPhrase=0 and OR and "-" syntax is significant. */ static int tokenizeSegment( sqlite3_tokenizer *pTokenizer, /* The tokenizer to use */ const char *pSegment, int nSegment, /* Query expression being parsed */ int inPhrase, /* True if within "..." */ Query *pQuery /* Append results here */ ){ const sqlite3_tokenizer_module *pModule = pTokenizer->pModule; sqlite3_tokenizer_cursor *pCursor; int firstIndex = pQuery->nTerms; int iCol; int nTerm = 1; int rc = pModule->xOpen(pTokenizer, pSegment, nSegment, &pCursor); if( rc!=SQLITE_OK ) return rc; pCursor->pTokenizer = pTokenizer; while( 1 ){ const char *pToken; int nToken, iBegin, iEnd, iPos; rc = pModule->xNext(pCursor, &pToken, &nToken, &iBegin, &iEnd, &iPos); if( rc!=SQLITE_OK ) break; if( !inPhrase && pSegment[iEnd]==':' && (iCol = checkColumnSpecifier(pQuery->pFts, pToken, nToken))>=0 ){ pQuery->nextColumn = iCol; continue; } if( !inPhrase && pQuery->nTerms>0 && nToken==2 && pSegment[iBegin]=='O' && pSegment[iBegin+1]=='R' ){ pQuery->nextIsOr = 1; continue; } queryAdd(pQuery, pToken, nToken); if( !inPhrase && iBegin>0 && pSegment[iBegin-1]=='-' ){ pQuery->pTerms[pQuery->nTerms-1].isNot = 1; } if( iEnd<nSegment && pSegment[iEnd]=='*' ){ pQuery->pTerms[pQuery->nTerms-1].isPrefix = 1; } pQuery->pTerms[pQuery->nTerms-1].iPhrase = nTerm; if( inPhrase ){ nTerm++; } } if( inPhrase && pQuery->nTerms>firstIndex ){ pQuery->pTerms[firstIndex].nPhrase = pQuery->nTerms - firstIndex - 1; } return pModule->xClose(pCursor); } /* Parse a query string, yielding a Query object pQuery. ** ** The calling function will need to queryClear() to clean up ** the dynamically allocated memory held by pQuery. */ static int parseQuery( fulltext_vtab *v, /* The fulltext index */ const char *zInput, /* Input text of the query string */ int nInput, /* Size of the input text */ int dfltColumn, /* Default column of the index to match against */ Query *pQuery /* Write the parse results here. */ ){ int iInput, inPhrase = 0; if( zInput==0 ) nInput = 0; if( nInput<0 ) nInput = strlen(zInput); pQuery->nTerms = 0; pQuery->pTerms = NULL; pQuery->nextIsOr = 0; pQuery->nextColumn = dfltColumn; pQuery->dfltColumn = dfltColumn; pQuery->pFts = v; for(iInput=0; iInput<nInput; ++iInput){ int i; for(i=iInput; i<nInput && zInput[i]!='"'; ++i){} if( i>iInput ){ tokenizeSegment(v->pTokenizer, zInput+iInput, i-iInput, inPhrase, pQuery); } iInput = i; if( i<nInput ){ assert( zInput[i]=='"' ); inPhrase = !inPhrase; } } if( inPhrase ){ /* unmatched quote */ queryClear(pQuery); return SQLITE_ERROR; } return SQLITE_OK; } /* TODO(shess) Refactor the code to remove this forward decl. */ static int flushPendingTerms(fulltext_vtab *v); /* Perform a full-text query using the search expression in ** zInput[0..nInput-1]. Return a list of matching documents ** in pResult. ** ** Queries must match column iColumn. Or if iColumn>=nColumn ** they are allowed to match against any column. */ static int fulltextQuery( fulltext_vtab *v, /* The full text index */ int iColumn, /* Match against this column by default */ const char *zInput, /* The query string */ int nInput, /* Number of bytes in zInput[] */ DataBuffer *pResult, /* Write the result doclist here */ Query *pQuery /* Put parsed query string here */ ){ int i, iNext, rc; DataBuffer left, right, or, new; int nNot = 0; QueryTerm *aTerm; /* TODO(shess) Instead of flushing pendingTerms, we could query for ** the relevant term and merge the doclist into what we receive from ** the database. Wait and see if this is a common issue, first. ** ** A good reason not to flush is to not generate update-related ** error codes from here. */ /* Flush any buffered updates before executing the query. */ rc = flushPendingTerms(v); if( rc!=SQLITE_OK ) return rc; /* TODO(shess) I think that the queryClear() calls below are not ** necessary, because fulltextClose() already clears the query. */ rc = parseQuery(v, zInput, nInput, iColumn, pQuery); if( rc!=SQLITE_OK ) return rc; /* Empty or NULL queries return no results. */ if( pQuery->nTerms==0 ){ dataBufferInit(pResult, 0); return SQLITE_OK; } /* Merge AND terms. */ /* TODO(shess) I think we can early-exit if( i>nNot && left.nData==0 ). */ aTerm = pQuery->pTerms; for(i = 0; i<pQuery->nTerms; i=iNext){ if( aTerm[i].isNot ){ /* Handle all NOT terms in a separate pass */ nNot++; iNext = i + aTerm[i].nPhrase+1; continue; } iNext = i + aTerm[i].nPhrase + 1; rc = docListOfTerm(v, aTerm[i].iColumn, &aTerm[i], &right); if( rc ){ if( i!=nNot ) dataBufferDestroy(&left); queryClear(pQuery); return rc; } while( iNext<pQuery->nTerms && aTerm[iNext].isOr ){ rc = docListOfTerm(v, aTerm[iNext].iColumn, &aTerm[iNext], &or); iNext += aTerm[iNext].nPhrase + 1; if( rc ){ if( i!=nNot ) dataBufferDestroy(&left); dataBufferDestroy(&right); queryClear(pQuery); return rc; } dataBufferInit(&new, 0); docListOrMerge(right.pData, right.nData, or.pData, or.nData, &new); dataBufferDestroy(&right); dataBufferDestroy(&or); right = new; } if( i==nNot ){ /* first term processed. */ left = right; }else{ dataBufferInit(&new, 0); docListAndMerge(left.pData, left.nData, right.pData, right.nData, &new); dataBufferDestroy(&right); dataBufferDestroy(&left); left = new; } } if( nNot==pQuery->nTerms ){ /* We do not yet know how to handle a query of only NOT terms */ return SQLITE_ERROR; } /* Do the EXCEPT terms */ for(i=0; i<pQuery->nTerms; i += aTerm[i].nPhrase + 1){ if( !aTerm[i].isNot ) continue; rc = docListOfTerm(v, aTerm[i].iColumn, &aTerm[i], &right); if( rc ){ queryClear(pQuery); dataBufferDestroy(&left); return rc; } dataBufferInit(&new, 0); docListExceptMerge(left.pData, left.nData, right.pData, right.nData, &new); dataBufferDestroy(&right); dataBufferDestroy(&left); left = new; } *pResult = left; return rc; } /* ** This is the xFilter interface for the virtual table. See ** the virtual table xFilter method documentation for additional ** information. ** ** If idxNum==QUERY_GENERIC then do a full table scan against ** the %_content table. ** ** If idxNum==QUERY_ROWID then do a rowid lookup for a single entry ** in the %_content table. ** ** If idxNum>=QUERY_FULLTEXT then use the full text index. The ** column on the left-hand side of the MATCH operator is column ** number idxNum-QUERY_FULLTEXT, 0 indexed. argv[0] is the right-hand ** side of the MATCH operator. */ /* TODO(shess) Upgrade the cursor initialization and destruction to ** account for fulltextFilter() being called multiple times on the ** same cursor. The current solution is very fragile. Apply fix to ** fts3 as appropriate. */ static int fulltextFilter( sqlite3_vtab_cursor *pCursor, /* The cursor used for this query */ int idxNum, const char *idxStr, /* Which indexing scheme to use */ int argc, sqlite3_value **argv /* Arguments for the indexing scheme */ ){ fulltext_cursor *c = (fulltext_cursor *) pCursor; fulltext_vtab *v = cursor_vtab(c); int rc; char *zSql; TRACE(("FTS3 Filter %p\n",pCursor)); zSql = sqlite3_mprintf("select rowid, * from %%_content %s", idxNum==QUERY_GENERIC ? "" : "where rowid=?"); sqlite3_finalize(c->pStmt); rc = sql_prepare(v->db, v->zDb, v->zName, &c->pStmt, zSql); sqlite3_free(zSql); if( rc!=SQLITE_OK ) return rc; c->iCursorType = idxNum; switch( idxNum ){ case QUERY_GENERIC: break; case QUERY_ROWID: rc = sqlite3_bind_int64(c->pStmt, 1, sqlite3_value_int64(argv[0])); if( rc!=SQLITE_OK ) return rc; break; default: /* full-text search */ { const char *zQuery = (const char *)sqlite3_value_text(argv[0]); assert( idxNum<=QUERY_FULLTEXT+v->nColumn); assert( argc==1 ); queryClear(&c->q); if( c->result.nData!=0 ){ /* This case happens if the same cursor is used repeatedly. */ dlrDestroy(&c->reader); dataBufferReset(&c->result); }else{ dataBufferInit(&c->result, 0); } rc = fulltextQuery(v, idxNum-QUERY_FULLTEXT, zQuery, -1, &c->result, &c->q); if( rc!=SQLITE_OK ) return rc; if( c->result.nData!=0 ){ dlrInit(&c->reader, DL_DOCIDS, c->result.pData, c->result.nData); } break; } } return fulltextNext(pCursor); } /* This is the xEof method of the virtual table. The SQLite core ** calls this routine to find out if it has reached the end of ** a query's results set. */ static int fulltextEof(sqlite3_vtab_cursor *pCursor){ fulltext_cursor *c = (fulltext_cursor *) pCursor; return c->eof; } /* This is the xColumn method of the virtual table. The SQLite ** core calls this method during a query when it needs the value ** of a column from the virtual table. This method needs to use ** one of the sqlite3_result_*() routines to store the requested ** value back in the pContext. */ static int fulltextColumn(sqlite3_vtab_cursor *pCursor, sqlite3_context *pContext, int idxCol){ fulltext_cursor *c = (fulltext_cursor *) pCursor; fulltext_vtab *v = cursor_vtab(c); if( idxCol<v->nColumn ){ sqlite3_value *pVal = sqlite3_column_value(c->pStmt, idxCol+1); sqlite3_result_value(pContext, pVal); }else if( idxCol==v->nColumn ){ /* The extra column whose name is the same as the table. ** Return a blob which is a pointer to the cursor */ sqlite3_result_blob(pContext, &c, sizeof(c), SQLITE_TRANSIENT); } return SQLITE_OK; } /* This is the xRowid method. The SQLite core calls this routine to ** retrive the rowid for the current row of the result set. The ** rowid should be written to *pRowid. */ static int fulltextRowid(sqlite3_vtab_cursor *pCursor, sqlite_int64 *pRowid){ fulltext_cursor *c = (fulltext_cursor *) pCursor; *pRowid = sqlite3_column_int64(c->pStmt, 0); return SQLITE_OK; } /* Add all terms in [zText] to pendingTerms table. If [iColumn] > 0, ** we also store positions and offsets in the hash table using that ** column number. */ static int buildTerms(fulltext_vtab *v, sqlite_int64 iDocid, const char *zText, int iColumn){ sqlite3_tokenizer *pTokenizer = v->pTokenizer; sqlite3_tokenizer_cursor *pCursor; const char *pToken; int nTokenBytes; int iStartOffset, iEndOffset, iPosition; int rc; rc = pTokenizer->pModule->xOpen(pTokenizer, zText, -1, &pCursor); if( rc!=SQLITE_OK ) return rc; pCursor->pTokenizer = pTokenizer; while( SQLITE_OK==pTokenizer->pModule->xNext(pCursor, &pToken, &nTokenBytes, &iStartOffset, &iEndOffset, &iPosition) ){ DLCollector *p; int nData; /* Size of doclist before our update. */ /* Positions can't be negative; we use -1 as a terminator internally. */ if( iPosition<0 ){ pTokenizer->pModule->xClose(pCursor); return SQLITE_ERROR; } p = fts3HashFind(&v->pendingTerms, pToken, nTokenBytes); if( p==NULL ){ nData = 0; p = dlcNew(iDocid, DL_DEFAULT); fts3HashInsert(&v->pendingTerms, pToken, nTokenBytes, p); /* Overhead for our hash table entry, the key, and the value. */ v->nPendingData += sizeof(struct fts3HashElem)+sizeof(*p)+nTokenBytes; }else{ nData = p->b.nData; if( p->dlw.iPrevDocid!=iDocid ) dlcNext(p, iDocid); } if( iColumn>=0 ){ dlcAddPos(p, iColumn, iPosition, iStartOffset, iEndOffset); } /* Accumulate data added by dlcNew or dlcNext, and dlcAddPos. */ v->nPendingData += p->b.nData-nData; } /* TODO(shess) Check return? Should this be able to cause errors at ** this point? Actually, same question about sqlite3_finalize(), ** though one could argue that failure there means that the data is ** not durable. *ponder* */ pTokenizer->pModule->xClose(pCursor); return rc; } /* Add doclists for all terms in [pValues] to pendingTerms table. */ static int insertTerms(fulltext_vtab *v, sqlite_int64 iRowid, sqlite3_value **pValues){ int i; for(i = 0; i < v->nColumn ; ++i){ char *zText = (char*)sqlite3_value_text(pValues[i]); int rc = buildTerms(v, iRowid, zText, i); if( rc!=SQLITE_OK ) return rc; } return SQLITE_OK; } /* Add empty doclists for all terms in the given row's content to ** pendingTerms. */ static int deleteTerms(fulltext_vtab *v, sqlite_int64 iRowid){ const char **pValues; int i, rc; /* TODO(shess) Should we allow such tables at all? */ if( DL_DEFAULT==DL_DOCIDS ) return SQLITE_ERROR; rc = content_select(v, iRowid, &pValues); if( rc!=SQLITE_OK ) return rc; for(i = 0 ; i < v->nColumn; ++i) { rc = buildTerms(v, iRowid, pValues[i], -1); if( rc!=SQLITE_OK ) break; } freeStringArray(v->nColumn, pValues); return SQLITE_OK; } /* TODO(shess) Refactor the code to remove this forward decl. */ static int initPendingTerms(fulltext_vtab *v, sqlite_int64 iDocid); /* Insert a row into the %_content table; set *piRowid to be the ID of the ** new row. Add doclists for terms to pendingTerms. */ static int index_insert(fulltext_vtab *v, sqlite3_value *pRequestRowid, sqlite3_value **pValues, sqlite_int64 *piRowid){ int rc; rc = content_insert(v, pRequestRowid, pValues); /* execute an SQL INSERT */ if( rc!=SQLITE_OK ) return rc; *piRowid = sqlite3_last_insert_rowid(v->db); rc = initPendingTerms(v, *piRowid); if( rc!=SQLITE_OK ) return rc; return insertTerms(v, *piRowid, pValues); } /* Delete a row from the %_content table; add empty doclists for terms ** to pendingTerms. */ static int index_delete(fulltext_vtab *v, sqlite_int64 iRow){ int rc = initPendingTerms(v, iRow); if( rc!=SQLITE_OK ) return rc; rc = deleteTerms(v, iRow); if( rc!=SQLITE_OK ) return rc; return content_delete(v, iRow); /* execute an SQL DELETE */ } /* Update a row in the %_content table; add delete doclists to ** pendingTerms for old terms not in the new data, add insert doclists ** to pendingTerms for terms in the new data. */ static int index_update(fulltext_vtab *v, sqlite_int64 iRow, sqlite3_value **pValues){ int rc = initPendingTerms(v, iRow); if( rc!=SQLITE_OK ) return rc; /* Generate an empty doclist for each term that previously appeared in this * row. */ rc = deleteTerms(v, iRow); if( rc!=SQLITE_OK ) return rc; rc = content_update(v, pValues, iRow); /* execute an SQL UPDATE */ if( rc!=SQLITE_OK ) return rc; /* Now add positions for terms which appear in the updated row. */ return insertTerms(v, iRow, pValues); } /*******************************************************************/ /* InteriorWriter is used to collect terms and block references into ** interior nodes in %_segments. See commentary at top of file for ** format. */ /* How large interior nodes can grow. */ #define INTERIOR_MAX 2048 /* Minimum number of terms per interior node (except the root). This ** prevents large terms from making the tree too skinny - must be >0 ** so that the tree always makes progress. Note that the min tree ** fanout will be INTERIOR_MIN_TERMS+1. */ #define INTERIOR_MIN_TERMS 7 #if INTERIOR_MIN_TERMS<1 # error INTERIOR_MIN_TERMS must be greater than 0. #endif /* ROOT_MAX controls how much data is stored inline in the segment ** directory. */ /* TODO(shess) Push ROOT_MAX down to whoever is writing things. It's ** only here so that interiorWriterRootInfo() and leafWriterRootInfo() ** can both see it, but if the caller passed it in, we wouldn't even ** need a define. */ #define ROOT_MAX 1024 #if ROOT_MAX<VARINT_MAX*2 # error ROOT_MAX must have enough space for a header. #endif /* InteriorBlock stores a linked-list of interior blocks while a lower ** layer is being constructed. */ typedef struct InteriorBlock { DataBuffer term; /* Leftmost term in block's subtree. */ DataBuffer data; /* Accumulated data for the block. */ struct InteriorBlock *next; } InteriorBlock; static InteriorBlock *interiorBlockNew(int iHeight, sqlite_int64 iChildBlock, const char *pTerm, int nTerm){ InteriorBlock *block = calloc(1, sizeof(InteriorBlock)); char c[VARINT_MAX+VARINT_MAX]; int n; dataBufferInit(&block->term, 0); dataBufferReplace(&block->term, pTerm, nTerm); n = putVarint(c, iHeight); n += putVarint(c+n, iChildBlock); dataBufferInit(&block->data, INTERIOR_MAX); dataBufferReplace(&block->data, c, n); return block; } #ifndef NDEBUG /* Verify that the data is readable as an interior node. */ static void interiorBlockValidate(InteriorBlock *pBlock){ const char *pData = pBlock->data.pData; int nData = pBlock->data.nData; int n, iDummy; sqlite_int64 iBlockid; assert( nData>0 ); assert( pData!=0 ); assert( pData+nData>pData ); /* Must lead with height of node as a varint(n), n>0 */ n = getVarint32(pData, &iDummy); assert( n>0 ); assert( iDummy>0 ); assert( n<nData ); pData += n; nData -= n; /* Must contain iBlockid. */ n = getVarint(pData, &iBlockid); assert( n>0 ); assert( n<=nData ); pData += n; nData -= n; /* Zero or more terms of positive length */ if( nData!=0 ){ /* First term is not delta-encoded. */ n = getVarint32(pData, &iDummy); assert( n>0 ); assert( iDummy>0 ); assert( n+iDummy>0); assert( n+iDummy<=nData ); pData += n+iDummy; nData -= n+iDummy; /* Following terms delta-encoded. */ while( nData!=0 ){ /* Length of shared prefix. */ n = getVarint32(pData, &iDummy); assert( n>0 ); assert( iDummy>=0 ); assert( n<nData ); pData += n; nData -= n; /* Length and data of distinct suffix. */ n = getVarint32(pData, &iDummy); assert( n>0 ); assert( iDummy>0 ); assert( n+iDummy>0); assert( n+iDummy<=nData ); pData += n+iDummy; nData -= n+iDummy; } } } #define ASSERT_VALID_INTERIOR_BLOCK(x) interiorBlockValidate(x) #else #define ASSERT_VALID_INTERIOR_BLOCK(x) assert( 1 ) #endif typedef struct InteriorWriter { int iHeight; /* from 0 at leaves. */ InteriorBlock *first, *last; struct InteriorWriter *parentWriter; DataBuffer term; /* Last term written to block "last". */ sqlite_int64 iOpeningChildBlock; /* First child block in block "last". */ #ifndef NDEBUG sqlite_int64 iLastChildBlock; /* for consistency checks. */ #endif } InteriorWriter; /* Initialize an interior node where pTerm[nTerm] marks the leftmost ** term in the tree. iChildBlock is the leftmost child block at the ** next level down the tree. */ static void interiorWriterInit(int iHeight, const char *pTerm, int nTerm, sqlite_int64 iChildBlock, InteriorWriter *pWriter){ InteriorBlock *block; assert( iHeight>0 ); CLEAR(pWriter); pWriter->iHeight = iHeight; pWriter->iOpeningChildBlock = iChildBlock; #ifndef NDEBUG pWriter->iLastChildBlock = iChildBlock; #endif block = interiorBlockNew(iHeight, iChildBlock, pTerm, nTerm); pWriter->last = pWriter->first = block; ASSERT_VALID_INTERIOR_BLOCK(pWriter->last); dataBufferInit(&pWriter->term, 0); } /* Append the child node rooted at iChildBlock to the interior node, ** with pTerm[nTerm] as the leftmost term in iChildBlock's subtree. */ static void interiorWriterAppend(InteriorWriter *pWriter, const char *pTerm, int nTerm, sqlite_int64 iChildBlock){ char c[VARINT_MAX+VARINT_MAX]; int n, nPrefix = 0; ASSERT_VALID_INTERIOR_BLOCK(pWriter->last); /* The first term written into an interior node is actually ** associated with the second child added (the first child was added ** in interiorWriterInit, or in the if clause at the bottom of this ** function). That term gets encoded straight up, with nPrefix left ** at 0. */ if( pWriter->term.nData==0 ){ n = putVarint(c, nTerm); }else{ while( nPrefix<pWriter->term.nData && pTerm[nPrefix]==pWriter->term.pData[nPrefix] ){ nPrefix++; } n = putVarint(c, nPrefix); n += putVarint(c+n, nTerm-nPrefix); } #ifndef NDEBUG pWriter->iLastChildBlock++; #endif assert( pWriter->iLastChildBlock==iChildBlock ); /* Overflow to a new block if the new term makes the current block ** too big, and the current block already has enough terms. */ if( pWriter->last->data.nData+n+nTerm-nPrefix>INTERIOR_MAX && iChildBlock-pWriter->iOpeningChildBlock>INTERIOR_MIN_TERMS ){ pWriter->last->next = interiorBlockNew(pWriter->iHeight, iChildBlock, pTerm, nTerm); pWriter->last = pWriter->last->next; pWriter->iOpeningChildBlock = iChildBlock; dataBufferReset(&pWriter->term); }else{ dataBufferAppend2(&pWriter->last->data, c, n, pTerm+nPrefix, nTerm-nPrefix); dataBufferReplace(&pWriter->term, pTerm, nTerm); } ASSERT_VALID_INTERIOR_BLOCK(pWriter->last); } /* Free the space used by pWriter, including the linked-list of ** InteriorBlocks, and parentWriter, if present. */ static int interiorWriterDestroy(InteriorWriter *pWriter){ InteriorBlock *block = pWriter->first; while( block!=NULL ){ InteriorBlock *b = block; block = block->next; dataBufferDestroy(&b->term); dataBufferDestroy(&b->data); free(b); } if( pWriter->parentWriter!=NULL ){ interiorWriterDestroy(pWriter->parentWriter); free(pWriter->parentWriter); } dataBufferDestroy(&pWriter->term); SCRAMBLE(pWriter); return SQLITE_OK; } /* If pWriter can fit entirely in ROOT_MAX, return it as the root info ** directly, leaving *piEndBlockid unchanged. Otherwise, flush ** pWriter to %_segments, building a new layer of interior nodes, and ** recursively ask for their root into. */ static int interiorWriterRootInfo(fulltext_vtab *v, InteriorWriter *pWriter, char **ppRootInfo, int *pnRootInfo, sqlite_int64 *piEndBlockid){ InteriorBlock *block = pWriter->first; sqlite_int64 iBlockid = 0; int rc; /* If we can fit the segment inline */ if( block==pWriter->last && block->data.nData<ROOT_MAX ){ *ppRootInfo = block->data.pData; *pnRootInfo = block->data.nData; return SQLITE_OK; } /* Flush the first block to %_segments, and create a new level of ** interior node. */ ASSERT_VALID_INTERIOR_BLOCK(block); rc = block_insert(v, block->data.pData, block->data.nData, &iBlockid); if( rc!=SQLITE_OK ) return rc; *piEndBlockid = iBlockid; pWriter->parentWriter = malloc(sizeof(*pWriter->parentWriter)); interiorWriterInit(pWriter->iHeight+1, block->term.pData, block->term.nData, iBlockid, pWriter->parentWriter); /* Flush additional blocks and append to the higher interior ** node. */ for(block=block->next; block!=NULL; block=block->next){ ASSERT_VALID_INTERIOR_BLOCK(block); rc = block_insert(v, block->data.pData, block->data.nData, &iBlockid); if( rc!=SQLITE_OK ) return rc; *piEndBlockid = iBlockid; interiorWriterAppend(pWriter->parentWriter, block->term.pData, block->term.nData, iBlockid); } /* Parent node gets the chance to be the root. */ return interiorWriterRootInfo(v, pWriter->parentWriter, ppRootInfo, pnRootInfo, piEndBlockid); } /****************************************************************/ /* InteriorReader is used to read off the data from an interior node ** (see comment at top of file for the format). */ typedef struct InteriorReader { const char *pData; int nData; DataBuffer term; /* previous term, for decoding term delta. */ sqlite_int64 iBlockid; } InteriorReader; static void interiorReaderDestroy(InteriorReader *pReader){ SCRAMBLE(pReader); } /* TODO(shess) The assertions are great, but what if we're in NDEBUG ** and the blob is empty or otherwise contains suspect data? */ static void interiorReaderInit(const char *pData, int nData, InteriorReader *pReader){ int n, nTerm; /* Require at least the leading flag byte */ assert( nData>0 ); assert( pData[0]!='\0' ); CLEAR(pReader); /* Decode the base blockid, and set the cursor to the first term. */ n = getVarint(pData+1, &pReader->iBlockid); assert( 1+n<=nData ); pReader->pData = pData+1+n; pReader->nData = nData-(1+n); /* A single-child interior node (such as when a leaf node was too ** large for the segment directory) won't have any terms. ** Otherwise, decode the first term. */ if( pReader->nData==0 ){ dataBufferInit(&pReader->term, 0); }else{ n = getVarint32(pReader->pData, &nTerm); dataBufferInit(&pReader->term, nTerm); dataBufferReplace(&pReader->term, pReader->pData+n, nTerm); assert( n+nTerm<=pReader->nData ); pReader->pData += n+nTerm; pReader->nData -= n+nTerm; } } static int interiorReaderAtEnd(InteriorReader *pReader){ return pReader->term.nData==0; } static sqlite_int64 interiorReaderCurrentBlockid(InteriorReader *pReader){ return pReader->iBlockid; } static int interiorReaderTermBytes(InteriorReader *pReader){ assert( !interiorReaderAtEnd(pReader) ); return pReader->term.nData; } static const char *interiorReaderTerm(InteriorReader *pReader){ assert( !interiorReaderAtEnd(pReader) ); return pReader->term.pData; } /* Step forward to the next term in the node. */ static void interiorReaderStep(InteriorReader *pReader){ assert( !interiorReaderAtEnd(pReader) ); /* If the last term has been read, signal eof, else construct the ** next term. */ if( pReader->nData==0 ){ dataBufferReset(&pReader->term); }else{ int n, nPrefix, nSuffix; n = getVarint32(pReader->pData, &nPrefix); n += getVarint32(pReader->pData+n, &nSuffix); /* Truncate the current term and append suffix data. */ pReader->term.nData = nPrefix; dataBufferAppend(&pReader->term, pReader->pData+n, nSuffix); assert( n+nSuffix<=pReader->nData ); pReader->pData += n+nSuffix; pReader->nData -= n+nSuffix; } pReader->iBlockid++; } /* Compare the current term to pTerm[nTerm], returning strcmp-style ** results. If isPrefix, equality means equal through nTerm bytes. */ static int interiorReaderTermCmp(InteriorReader *pReader, const char *pTerm, int nTerm, int isPrefix){ const char *pReaderTerm = interiorReaderTerm(pReader); int nReaderTerm = interiorReaderTermBytes(pReader); int c, n = nReaderTerm<nTerm ? nReaderTerm : nTerm; if( n==0 ){ if( nReaderTerm>0 ) return -1; if( nTerm>0 ) return 1; return 0; } c = memcmp(pReaderTerm, pTerm, n); if( c!=0 ) return c; if( isPrefix && n==nTerm ) return 0; return nReaderTerm - nTerm; } /****************************************************************/ /* LeafWriter is used to collect terms and associated doclist data ** into leaf blocks in %_segments (see top of file for format info). ** Expected usage is: ** ** LeafWriter writer; ** leafWriterInit(0, 0, &writer); ** while( sorted_terms_left_to_process ){ ** // data is doclist data for that term. ** rc = leafWriterStep(v, &writer, pTerm, nTerm, pData, nData); ** if( rc!=SQLITE_OK ) goto err; ** } ** rc = leafWriterFinalize(v, &writer); **err: ** leafWriterDestroy(&writer); ** return rc; ** ** leafWriterStep() may write a collected leaf out to %_segments. ** leafWriterFinalize() finishes writing any buffered data and stores ** a root node in %_segdir. leafWriterDestroy() frees all buffers and ** InteriorWriters allocated as part of writing this segment. ** ** TODO(shess) Document leafWriterStepMerge(). */ /* Put terms with data this big in their own block. */ #define STANDALONE_MIN 1024 /* Keep leaf blocks below this size. */ #define LEAF_MAX 2048 typedef struct LeafWriter { int iLevel; int idx; sqlite_int64 iStartBlockid; /* needed to create the root info */ sqlite_int64 iEndBlockid; /* when we're done writing. */ DataBuffer term; /* previous encoded term */ DataBuffer data; /* encoding buffer */ /* bytes of first term in the current node which distinguishes that ** term from the last term of the previous node. */ int nTermDistinct; InteriorWriter parentWriter; /* if we overflow */ int has_parent; } LeafWriter; static void leafWriterInit(int iLevel, int idx, LeafWriter *pWriter){ CLEAR(pWriter); pWriter->iLevel = iLevel; pWriter->idx = idx; dataBufferInit(&pWriter->term, 32); /* Start out with a reasonably sized block, though it can grow. */ dataBufferInit(&pWriter->data, LEAF_MAX); } #ifndef NDEBUG /* Verify that the data is readable as a leaf node. */ static void leafNodeValidate(const char *pData, int nData){ int n, iDummy; if( nData==0 ) return; assert( nData>0 ); assert( pData!=0 ); assert( pData+nData>pData ); /* Must lead with a varint(0) */ n = getVarint32(pData, &iDummy); assert( iDummy==0 ); assert( n>0 ); assert( n<nData ); pData += n; nData -= n; /* Leading term length and data must fit in buffer. */ n = getVarint32(pData, &iDummy); assert( n>0 ); assert( iDummy>0 ); assert( n+iDummy>0 ); assert( n+iDummy<nData ); pData += n+iDummy; nData -= n+iDummy; /* Leading term's doclist length and data must fit. */ n = getVarint32(pData, &iDummy); assert( n>0 ); assert( iDummy>0 ); assert( n+iDummy>0 ); assert( n+iDummy<=nData ); ASSERT_VALID_DOCLIST(DL_DEFAULT, pData+n, iDummy, NULL); pData += n+iDummy; nData -= n+iDummy; /* Verify that trailing terms and doclists also are readable. */ while( nData!=0 ){ n = getVarint32(pData, &iDummy); assert( n>0 ); assert( iDummy>=0 ); assert( n<nData ); pData += n; nData -= n; n = getVarint32(pData, &iDummy); assert( n>0 ); assert( iDummy>0 ); assert( n+iDummy>0 ); assert( n+iDummy<nData ); pData += n+iDummy; nData -= n+iDummy; n = getVarint32(pData, &iDummy); assert( n>0 ); assert( iDummy>0 ); assert( n+iDummy>0 ); assert( n+iDummy<=nData ); ASSERT_VALID_DOCLIST(DL_DEFAULT, pData+n, iDummy, NULL); pData += n+iDummy; nData -= n+iDummy; } } #define ASSERT_VALID_LEAF_NODE(p, n) leafNodeValidate(p, n) #else #define ASSERT_VALID_LEAF_NODE(p, n) assert( 1 ) #endif /* Flush the current leaf node to %_segments, and adding the resulting ** blockid and the starting term to the interior node which will ** contain it. */ static int leafWriterInternalFlush(fulltext_vtab *v, LeafWriter *pWriter, int iData, int nData){ sqlite_int64 iBlockid = 0; const char *pStartingTerm; int nStartingTerm, rc, n; /* Must have the leading varint(0) flag, plus at least some ** valid-looking data. */ assert( nData>2 ); assert( iData>=0 ); assert( iData+nData<=pWriter->data.nData ); ASSERT_VALID_LEAF_NODE(pWriter->data.pData+iData, nData); rc = block_insert(v, pWriter->data.pData+iData, nData, &iBlockid); if( rc!=SQLITE_OK ) return rc; assert( iBlockid!=0 ); /* Reconstruct the first term in the leaf for purposes of building ** the interior node. */ n = getVarint32(pWriter->data.pData+iData+1, &nStartingTerm); pStartingTerm = pWriter->data.pData+iData+1+n; assert( pWriter->data.nData>iData+1+n+nStartingTerm ); assert( pWriter->nTermDistinct>0 ); assert( pWriter->nTermDistinct<=nStartingTerm ); nStartingTerm = pWriter->nTermDistinct; if( pWriter->has_parent ){ interiorWriterAppend(&pWriter->parentWriter, pStartingTerm, nStartingTerm, iBlockid); }else{ interiorWriterInit(1, pStartingTerm, nStartingTerm, iBlockid, &pWriter->parentWriter); pWriter->has_parent = 1; } /* Track the span of this segment's leaf nodes. */ if( pWriter->iEndBlockid==0 ){ pWriter->iEndBlockid = pWriter->iStartBlockid = iBlockid; }else{ pWriter->iEndBlockid++; assert( iBlockid==pWriter->iEndBlockid ); } return SQLITE_OK; } static int leafWriterFlush(fulltext_vtab *v, LeafWriter *pWriter){ int rc = leafWriterInternalFlush(v, pWriter, 0, pWriter->data.nData); if( rc!=SQLITE_OK ) return rc; /* Re-initialize the output buffer. */ dataBufferReset(&pWriter->data); return SQLITE_OK; } /* Fetch the root info for the segment. If the entire leaf fits ** within ROOT_MAX, then it will be returned directly, otherwise it ** will be flushed and the root info will be returned from the ** interior node. *piEndBlockid is set to the blockid of the last ** interior or leaf node written to disk (0 if none are written at ** all). */ static int leafWriterRootInfo(fulltext_vtab *v, LeafWriter *pWriter, char **ppRootInfo, int *pnRootInfo, sqlite_int64 *piEndBlockid){ /* we can fit the segment entirely inline */ if( !pWriter->has_parent && pWriter->data.nData<ROOT_MAX ){ *ppRootInfo = pWriter->data.pData; *pnRootInfo = pWriter->data.nData; *piEndBlockid = 0; return SQLITE_OK; } /* Flush remaining leaf data. */ if( pWriter->data.nData>0 ){ int rc = leafWriterFlush(v, pWriter); if( rc!=SQLITE_OK ) return rc; } /* We must have flushed a leaf at some point. */ assert( pWriter->has_parent ); /* Tenatively set the end leaf blockid as the end blockid. If the ** interior node can be returned inline, this will be the final ** blockid, otherwise it will be overwritten by ** interiorWriterRootInfo(). */ *piEndBlockid = pWriter->iEndBlockid; return interiorWriterRootInfo(v, &pWriter->parentWriter, ppRootInfo, pnRootInfo, piEndBlockid); } /* Collect the rootInfo data and store it into the segment directory. ** This has the effect of flushing the segment's leaf data to ** %_segments, and also flushing any interior nodes to %_segments. */ static int leafWriterFinalize(fulltext_vtab *v, LeafWriter *pWriter){ sqlite_int64 iEndBlockid; char *pRootInfo; int rc, nRootInfo; rc = leafWriterRootInfo(v, pWriter, &pRootInfo, &nRootInfo, &iEndBlockid); if( rc!=SQLITE_OK ) return rc; /* Don't bother storing an entirely empty segment. */ if( iEndBlockid==0 && nRootInfo==0 ) return SQLITE_OK; return segdir_set(v, pWriter->iLevel, pWriter->idx, pWriter->iStartBlockid, pWriter->iEndBlockid, iEndBlockid, pRootInfo, nRootInfo); } static void leafWriterDestroy(LeafWriter *pWriter){ if( pWriter->has_parent ) interiorWriterDestroy(&pWriter->parentWriter); dataBufferDestroy(&pWriter->term); dataBufferDestroy(&pWriter->data); } /* Encode a term into the leafWriter, delta-encoding as appropriate. ** Returns the length of the new term which distinguishes it from the ** previous term, which can be used to set nTermDistinct when a node ** boundary is crossed. */ static int leafWriterEncodeTerm(LeafWriter *pWriter, const char *pTerm, int nTerm){ char c[VARINT_MAX+VARINT_MAX]; int n, nPrefix = 0; assert( nTerm>0 ); while( nPrefix<pWriter->term.nData && pTerm[nPrefix]==pWriter->term.pData[nPrefix] ){ nPrefix++; /* Failing this implies that the terms weren't in order. */ assert( nPrefix<nTerm ); } if( pWriter->data.nData==0 ){ /* Encode the node header and leading term as: ** varint(0) ** varint(nTerm) ** char pTerm[nTerm] */ n = putVarint(c, '\0'); n += putVarint(c+n, nTerm); dataBufferAppend2(&pWriter->data, c, n, pTerm, nTerm); }else{ /* Delta-encode the term as: ** varint(nPrefix) ** varint(nSuffix) ** char pTermSuffix[nSuffix] */ n = putVarint(c, nPrefix); n += putVarint(c+n, nTerm-nPrefix); dataBufferAppend2(&pWriter->data, c, n, pTerm+nPrefix, nTerm-nPrefix); } dataBufferReplace(&pWriter->term, pTerm, nTerm); return nPrefix+1; } /* Used to avoid a memmove when a large amount of doclist data is in ** the buffer. This constructs a node and term header before ** iDoclistData and flushes the resulting complete node using ** leafWriterInternalFlush(). */ static int leafWriterInlineFlush(fulltext_vtab *v, LeafWriter *pWriter, const char *pTerm, int nTerm, int iDoclistData){ char c[VARINT_MAX+VARINT_MAX]; int iData, n = putVarint(c, 0); n += putVarint(c+n, nTerm); /* There should always be room for the header. Even if pTerm shared ** a substantial prefix with the previous term, the entire prefix ** could be constructed from earlier data in the doclist, so there ** should be room. */ assert( iDoclistData>=n+nTerm ); iData = iDoclistData-(n+nTerm); memcpy(pWriter->data.pData+iData, c, n); memcpy(pWriter->data.pData+iData+n, pTerm, nTerm); return leafWriterInternalFlush(v, pWriter, iData, pWriter->data.nData-iData); } /* Push pTerm[nTerm] along with the doclist data to the leaf layer of ** %_segments. */ static int leafWriterStepMerge(fulltext_vtab *v, LeafWriter *pWriter, const char *pTerm, int nTerm, DLReader *pReaders, int nReaders){ char c[VARINT_MAX+VARINT_MAX]; int iTermData = pWriter->data.nData, iDoclistData; int i, nData, n, nActualData, nActual, rc, nTermDistinct; ASSERT_VALID_LEAF_NODE(pWriter->data.pData, pWriter->data.nData); nTermDistinct = leafWriterEncodeTerm(pWriter, pTerm, nTerm); /* Remember nTermDistinct if opening a new node. */ if( iTermData==0 ) pWriter->nTermDistinct = nTermDistinct; iDoclistData = pWriter->data.nData; /* Estimate the length of the merged doclist so we can leave space ** to encode it. */ for(i=0, nData=0; i<nReaders; i++){ nData += dlrAllDataBytes(&pReaders[i]); } n = putVarint(c, nData); dataBufferAppend(&pWriter->data, c, n); docListMerge(&pWriter->data, pReaders, nReaders); ASSERT_VALID_DOCLIST(DL_DEFAULT, pWriter->data.pData+iDoclistData+n, pWriter->data.nData-iDoclistData-n, NULL); /* The actual amount of doclist data at this point could be smaller ** than the length we encoded. Additionally, the space required to ** encode this length could be smaller. For small doclists, this is ** not a big deal, we can just use memmove() to adjust things. */ nActualData = pWriter->data.nData-(iDoclistData+n); nActual = putVarint(c, nActualData); assert( nActualData<=nData ); assert( nActual<=n ); /* If the new doclist is big enough for force a standalone leaf ** node, we can immediately flush it inline without doing the ** memmove(). */ /* TODO(shess) This test matches leafWriterStep(), which does this ** test before it knows the cost to varint-encode the term and ** doclist lengths. At some point, change to ** pWriter->data.nData-iTermData>STANDALONE_MIN. */ if( nTerm+nActualData>STANDALONE_MIN ){ /* Push leaf node from before this term. */ if( iTermData>0 ){ rc = leafWriterInternalFlush(v, pWriter, 0, iTermData); if( rc!=SQLITE_OK ) return rc; pWriter->nTermDistinct = nTermDistinct; } /* Fix the encoded doclist length. */ iDoclistData += n - nActual; memcpy(pWriter->data.pData+iDoclistData, c, nActual); /* Push the standalone leaf node. */ rc = leafWriterInlineFlush(v, pWriter, pTerm, nTerm, iDoclistData); if( rc!=SQLITE_OK ) return rc; /* Leave the node empty. */ dataBufferReset(&pWriter->data); return rc; } /* At this point, we know that the doclist was small, so do the ** memmove if indicated. */ if( nActual<n ){ memmove(pWriter->data.pData+iDoclistData+nActual, pWriter->data.pData+iDoclistData+n, pWriter->data.nData-(iDoclistData+n)); pWriter->data.nData -= n-nActual; } /* Replace written length with actual length. */ memcpy(pWriter->data.pData+iDoclistData, c, nActual); /* If the node is too large, break things up. */ /* TODO(shess) This test matches leafWriterStep(), which does this ** test before it knows the cost to varint-encode the term and ** doclist lengths. At some point, change to ** pWriter->data.nData>LEAF_MAX. */ if( iTermData+nTerm+nActualData>LEAF_MAX ){ /* Flush out the leading data as a node */ rc = leafWriterInternalFlush(v, pWriter, 0, iTermData); if( rc!=SQLITE_OK ) return rc; pWriter->nTermDistinct = nTermDistinct; /* Rebuild header using the current term */ n = putVarint(pWriter->data.pData, 0); n += putVarint(pWriter->data.pData+n, nTerm); memcpy(pWriter->data.pData+n, pTerm, nTerm); n += nTerm; /* There should always be room, because the previous encoding ** included all data necessary to construct the term. */ assert( n<iDoclistData ); /* So long as STANDALONE_MIN is half or less of LEAF_MAX, the ** following memcpy() is safe (as opposed to needing a memmove). */ assert( 2*STANDALONE_MIN<=LEAF_MAX ); assert( n+pWriter->data.nData-iDoclistData<iDoclistData ); memcpy(pWriter->data.pData+n, pWriter->data.pData+iDoclistData, pWriter->data.nData-iDoclistData); pWriter->data.nData -= iDoclistData-n; } ASSERT_VALID_LEAF_NODE(pWriter->data.pData, pWriter->data.nData); return SQLITE_OK; } /* Push pTerm[nTerm] along with the doclist data to the leaf layer of ** %_segments. */ /* TODO(shess) Revise writeZeroSegment() so that doclists are ** constructed directly in pWriter->data. */ static int leafWriterStep(fulltext_vtab *v, LeafWriter *pWriter, const char *pTerm, int nTerm, const char *pData, int nData){ int rc; DLReader reader; dlrInit(&reader, DL_DEFAULT, pData, nData); rc = leafWriterStepMerge(v, pWriter, pTerm, nTerm, &reader, 1); dlrDestroy(&reader); return rc; } /****************************************************************/ /* LeafReader is used to iterate over an individual leaf node. */ typedef struct LeafReader { DataBuffer term; /* copy of current term. */ const char *pData; /* data for current term. */ int nData; } LeafReader; static void leafReaderDestroy(LeafReader *pReader){ dataBufferDestroy(&pReader->term); SCRAMBLE(pReader); } static int leafReaderAtEnd(LeafReader *pReader){ return pReader->nData<=0; } /* Access the current term. */ static int leafReaderTermBytes(LeafReader *pReader){ return pReader->term.nData; } static const char *leafReaderTerm(LeafReader *pReader){ assert( pReader->term.nData>0 ); return pReader->term.pData; } /* Access the doclist data for the current term. */ static int leafReaderDataBytes(LeafReader *pReader){ int nData; assert( pReader->term.nData>0 ); getVarint32(pReader->pData, &nData); return nData; } static const char *leafReaderData(LeafReader *pReader){ int n, nData; assert( pReader->term.nData>0 ); n = getVarint32(pReader->pData, &nData); return pReader->pData+n; } static void leafReaderInit(const char *pData, int nData, LeafReader *pReader){ int nTerm, n; assert( nData>0 ); assert( pData[0]=='\0' ); CLEAR(pReader); /* Read the first term, skipping the header byte. */ n = getVarint32(pData+1, &nTerm); dataBufferInit(&pReader->term, nTerm); dataBufferReplace(&pReader->term, pData+1+n, nTerm); /* Position after the first term. */ assert( 1+n+nTerm<nData ); pReader->pData = pData+1+n+nTerm; pReader->nData = nData-1-n-nTerm; } /* Step the reader forward to the next term. */ static void leafReaderStep(LeafReader *pReader){ int n, nData, nPrefix, nSuffix; assert( !leafReaderAtEnd(pReader) ); /* Skip previous entry's data block. */ n = getVarint32(pReader->pData, &nData); assert( n+nData<=pReader->nData ); pReader->pData += n+nData; pReader->nData -= n+nData; if( !leafReaderAtEnd(pReader) ){ /* Construct the new term using a prefix from the old term plus a ** suffix from the leaf data. */ n = getVarint32(pReader->pData, &nPrefix); n += getVarint32(pReader->pData+n, &nSuffix); assert( n+nSuffix<pReader->nData ); pReader->term.nData = nPrefix; dataBufferAppend(&pReader->term, pReader->pData+n, nSuffix); pReader->pData += n+nSuffix; pReader->nData -= n+nSuffix; } } /* strcmp-style comparison of pReader's current term against pTerm. ** If isPrefix, equality means equal through nTerm bytes. */ static int leafReaderTermCmp(LeafReader *pReader, const char *pTerm, int nTerm, int isPrefix){ int c, n = pReader->term.nData<nTerm ? pReader->term.nData : nTerm; if( n==0 ){ if( pReader->term.nData>0 ) return -1; if(nTerm>0 ) return 1; return 0; } c = memcmp(pReader->term.pData, pTerm, n); if( c!=0 ) return c; if( isPrefix && n==nTerm ) return 0; return pReader->term.nData - nTerm; } /****************************************************************/ /* LeavesReader wraps LeafReader to allow iterating over the entire ** leaf layer of the tree. */ typedef struct LeavesReader { int idx; /* Index within the segment. */ sqlite3_stmt *pStmt; /* Statement we're streaming leaves from. */ int eof; /* we've seen SQLITE_DONE from pStmt. */ LeafReader leafReader; /* reader for the current leaf. */ DataBuffer rootData; /* root data for inline. */ } LeavesReader; /* Access the current term. */ static int leavesReaderTermBytes(LeavesReader *pReader){ assert( !pReader->eof ); return leafReaderTermBytes(&pReader->leafReader); } static const char *leavesReaderTerm(LeavesReader *pReader){ assert( !pReader->eof ); return leafReaderTerm(&pReader->leafReader); } /* Access the doclist data for the current term. */ static int leavesReaderDataBytes(LeavesReader *pReader){ assert( !pReader->eof ); return leafReaderDataBytes(&pReader->leafReader); } static const char *leavesReaderData(LeavesReader *pReader){ assert( !pReader->eof ); return leafReaderData(&pReader->leafReader); } static int leavesReaderAtEnd(LeavesReader *pReader){ return pReader->eof; } /* loadSegmentLeaves() may not read all the way to SQLITE_DONE, thus ** leaving the statement handle open, which locks the table. */ /* TODO(shess) This "solution" is not satisfactory. Really, there ** should be check-in function for all statement handles which ** arranges to call sqlite3_reset(). This most likely will require ** modification to control flow all over the place, though, so for now ** just punt. ** ** Note the the current system assumes that segment merges will run to ** completion, which is why this particular probably hasn't arisen in ** this case. Probably a brittle assumption. */ static int leavesReaderReset(LeavesReader *pReader){ return sqlite3_reset(pReader->pStmt); } static void leavesReaderDestroy(LeavesReader *pReader){ leafReaderDestroy(&pReader->leafReader); dataBufferDestroy(&pReader->rootData); SCRAMBLE(pReader); } /* Initialize pReader with the given root data (if iStartBlockid==0 ** the leaf data was entirely contained in the root), or from the ** stream of blocks between iStartBlockid and iEndBlockid, inclusive. */ static int leavesReaderInit(fulltext_vtab *v, int idx, sqlite_int64 iStartBlockid, sqlite_int64 iEndBlockid, const char *pRootData, int nRootData, LeavesReader *pReader){ CLEAR(pReader); pReader->idx = idx; dataBufferInit(&pReader->rootData, 0); if( iStartBlockid==0 ){ /* Entire leaf level fit in root data. */ dataBufferReplace(&pReader->rootData, pRootData, nRootData); leafReaderInit(pReader->rootData.pData, pReader->rootData.nData, &pReader->leafReader); }else{ sqlite3_stmt *s; int rc = sql_get_leaf_statement(v, idx, &s); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_bind_int64(s, 1, iStartBlockid); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_bind_int64(s, 2, iEndBlockid); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_step(s); if( rc==SQLITE_DONE ){ pReader->eof = 1; return SQLITE_OK; } if( rc!=SQLITE_ROW ) return rc; pReader->pStmt = s; leafReaderInit(sqlite3_column_blob(pReader->pStmt, 0), sqlite3_column_bytes(pReader->pStmt, 0), &pReader->leafReader); } return SQLITE_OK; } /* Step the current leaf forward to the next term. If we reach the ** end of the current leaf, step forward to the next leaf block. */ static int leavesReaderStep(fulltext_vtab *v, LeavesReader *pReader){ assert( !leavesReaderAtEnd(pReader) ); leafReaderStep(&pReader->leafReader); if( leafReaderAtEnd(&pReader->leafReader) ){ int rc; if( pReader->rootData.pData ){ pReader->eof = 1; return SQLITE_OK; } rc = sqlite3_step(pReader->pStmt); if( rc!=SQLITE_ROW ){ pReader->eof = 1; return rc==SQLITE_DONE ? SQLITE_OK : rc; } leafReaderDestroy(&pReader->leafReader); leafReaderInit(sqlite3_column_blob(pReader->pStmt, 0), sqlite3_column_bytes(pReader->pStmt, 0), &pReader->leafReader); } return SQLITE_OK; } /* Order LeavesReaders by their term, ignoring idx. Readers at eof ** always sort to the end. */ static int leavesReaderTermCmp(LeavesReader *lr1, LeavesReader *lr2){ if( leavesReaderAtEnd(lr1) ){ if( leavesReaderAtEnd(lr2) ) return 0; return 1; } if( leavesReaderAtEnd(lr2) ) return -1; return leafReaderTermCmp(&lr1->leafReader, leavesReaderTerm(lr2), leavesReaderTermBytes(lr2), 0); } /* Similar to leavesReaderTermCmp(), with additional ordering by idx ** so that older segments sort before newer segments. */ static int leavesReaderCmp(LeavesReader *lr1, LeavesReader *lr2){ int c = leavesReaderTermCmp(lr1, lr2); if( c!=0 ) return c; return lr1->idx-lr2->idx; } /* Assume that pLr[1]..pLr[nLr] are sorted. Bubble pLr[0] into its ** sorted position. */ static void leavesReaderReorder(LeavesReader *pLr, int nLr){ while( nLr>1 && leavesReaderCmp(pLr, pLr+1)>0 ){ LeavesReader tmp = pLr[0]; pLr[0] = pLr[1]; pLr[1] = tmp; nLr--; pLr++; } } /* Initializes pReaders with the segments from level iLevel, returning ** the number of segments in *piReaders. Leaves pReaders in sorted ** order. */ static int leavesReadersInit(fulltext_vtab *v, int iLevel, LeavesReader *pReaders, int *piReaders){ sqlite3_stmt *s; int i, rc = sql_get_statement(v, SEGDIR_SELECT_STMT, &s); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_bind_int(s, 1, iLevel); if( rc!=SQLITE_OK ) return rc; i = 0; while( (rc = sqlite3_step(s))==SQLITE_ROW ){ sqlite_int64 iStart = sqlite3_column_int64(s, 0); sqlite_int64 iEnd = sqlite3_column_int64(s, 1); const char *pRootData = sqlite3_column_blob(s, 2); int nRootData = sqlite3_column_bytes(s, 2); assert( i<MERGE_COUNT ); rc = leavesReaderInit(v, i, iStart, iEnd, pRootData, nRootData, &pReaders[i]); if( rc!=SQLITE_OK ) break; i++; } if( rc!=SQLITE_DONE ){ while( i-->0 ){ leavesReaderDestroy(&pReaders[i]); } return rc; } *piReaders = i; /* Leave our results sorted by term, then age. */ while( i-- ){ leavesReaderReorder(pReaders+i, *piReaders-i); } return SQLITE_OK; } /* Merge doclists from pReaders[nReaders] into a single doclist, which ** is written to pWriter. Assumes pReaders is ordered oldest to ** newest. */ /* TODO(shess) Consider putting this inline in segmentMerge(). */ static int leavesReadersMerge(fulltext_vtab *v, LeavesReader *pReaders, int nReaders, LeafWriter *pWriter){ DLReader dlReaders[MERGE_COUNT]; const char *pTerm = leavesReaderTerm(pReaders); int i, nTerm = leavesReaderTermBytes(pReaders); assert( nReaders<=MERGE_COUNT ); for(i=0; i<nReaders; i++){ dlrInit(&dlReaders[i], DL_DEFAULT, leavesReaderData(pReaders+i), leavesReaderDataBytes(pReaders+i)); } return leafWriterStepMerge(v, pWriter, pTerm, nTerm, dlReaders, nReaders); } /* Forward ref due to mutual recursion with segdirNextIndex(). */ static int segmentMerge(fulltext_vtab *v, int iLevel); /* Put the next available index at iLevel into *pidx. If iLevel ** already has MERGE_COUNT segments, they are merged to a higher ** level to make room. */ static int segdirNextIndex(fulltext_vtab *v, int iLevel, int *pidx){ int rc = segdir_max_index(v, iLevel, pidx); if( rc==SQLITE_DONE ){ /* No segments at iLevel. */ *pidx = 0; }else if( rc==SQLITE_ROW ){ if( *pidx==(MERGE_COUNT-1) ){ rc = segmentMerge(v, iLevel); if( rc!=SQLITE_OK ) return rc; *pidx = 0; }else{ (*pidx)++; } }else{ return rc; } return SQLITE_OK; } /* Merge MERGE_COUNT segments at iLevel into a new segment at ** iLevel+1. If iLevel+1 is already full of segments, those will be ** merged to make room. */ static int segmentMerge(fulltext_vtab *v, int iLevel){ LeafWriter writer; LeavesReader lrs[MERGE_COUNT]; int i, rc, idx = 0; /* Determine the next available segment index at the next level, ** merging as necessary. */ rc = segdirNextIndex(v, iLevel+1, &idx); if( rc!=SQLITE_OK ) return rc; /* TODO(shess) This assumes that we'll always see exactly ** MERGE_COUNT segments to merge at a given level. That will be ** broken if we allow the developer to request preemptive or ** deferred merging. */ memset(&lrs, '\0', sizeof(lrs)); rc = leavesReadersInit(v, iLevel, lrs, &i); if( rc!=SQLITE_OK ) return rc; assert( i==MERGE_COUNT ); leafWriterInit(iLevel+1, idx, &writer); /* Since leavesReaderReorder() pushes readers at eof to the end, ** when the first reader is empty, all will be empty. */ while( !leavesReaderAtEnd(lrs) ){ /* Figure out how many readers share their next term. */ for(i=1; i<MERGE_COUNT && !leavesReaderAtEnd(lrs+i); i++){ if( 0!=leavesReaderTermCmp(lrs, lrs+i) ) break; } rc = leavesReadersMerge(v, lrs, i, &writer); if( rc!=SQLITE_OK ) goto err; /* Step forward those that were merged. */ while( i-->0 ){ rc = leavesReaderStep(v, lrs+i); if( rc!=SQLITE_OK ) goto err; /* Reorder by term, then by age. */ leavesReaderReorder(lrs+i, MERGE_COUNT-i); } } for(i=0; i<MERGE_COUNT; i++){ leavesReaderDestroy(&lrs[i]); } rc = leafWriterFinalize(v, &writer); leafWriterDestroy(&writer); if( rc!=SQLITE_OK ) return rc; /* Delete the merged segment data. */ return segdir_delete(v, iLevel); err: for(i=0; i<MERGE_COUNT; i++){ leavesReaderDestroy(&lrs[i]); } leafWriterDestroy(&writer); return rc; } /* Scan pReader for pTerm/nTerm, and merge the term's doclist over ** *out (any doclists with duplicate docids overwrite those in *out). ** Internal function for loadSegmentLeaf(). */ static int loadSegmentLeavesInt(fulltext_vtab *v, LeavesReader *pReader, const char *pTerm, int nTerm, int isPrefix, DataBuffer *out){ assert( nTerm>0 ); /* Process while the prefix matches. */ while( !leavesReaderAtEnd(pReader) ){ /* TODO(shess) Really want leavesReaderTermCmp(), but that name is ** already taken to compare the terms of two LeavesReaders. Think ** on a better name. [Meanwhile, break encapsulation rather than ** use a confusing name.] */ int rc; int c = leafReaderTermCmp(&pReader->leafReader, pTerm, nTerm, isPrefix); if( c==0 ){ const char *pData = leavesReaderData(pReader); int nData = leavesReaderDataBytes(pReader); if( out->nData==0 ){ dataBufferReplace(out, pData, nData); }else{ DataBuffer result; dataBufferInit(&result, out->nData+nData); docListUnion(out->pData, out->nData, pData, nData, &result); dataBufferDestroy(out); *out = result; /* TODO(shess) Rather than destroy out, we could retain it for ** later reuse. */ } } if( c>0 ) break; /* Past any possible matches. */ rc = leavesReaderStep(v, pReader); if( rc!=SQLITE_OK ) return rc; } return SQLITE_OK; } /* Call loadSegmentLeavesInt() with pData/nData as input. */ static int loadSegmentLeaf(fulltext_vtab *v, const char *pData, int nData, const char *pTerm, int nTerm, int isPrefix, DataBuffer *out){ LeavesReader reader; int rc; assert( nData>1 ); assert( *pData=='\0' ); rc = leavesReaderInit(v, 0, 0, 0, pData, nData, &reader); if( rc!=SQLITE_OK ) return rc; rc = loadSegmentLeavesInt(v, &reader, pTerm, nTerm, isPrefix, out); leavesReaderReset(&reader); leavesReaderDestroy(&reader); return rc; } /* Call loadSegmentLeavesInt() with the leaf nodes from iStartLeaf to ** iEndLeaf (inclusive) as input, and merge the resulting doclist into ** out. */ static int loadSegmentLeaves(fulltext_vtab *v, sqlite_int64 iStartLeaf, sqlite_int64 iEndLeaf, const char *pTerm, int nTerm, int isPrefix, DataBuffer *out){ int rc; LeavesReader reader; assert( iStartLeaf<=iEndLeaf ); rc = leavesReaderInit(v, 0, iStartLeaf, iEndLeaf, NULL, 0, &reader); if( rc!=SQLITE_OK ) return rc; rc = loadSegmentLeavesInt(v, &reader, pTerm, nTerm, isPrefix, out); leavesReaderReset(&reader); leavesReaderDestroy(&reader); return rc; } /* Taking pData/nData as an interior node, find the sequence of child ** nodes which could include pTerm/nTerm/isPrefix. Note that the ** interior node terms logically come between the blocks, so there is ** one more blockid than there are terms (that block contains terms >= ** the last interior-node term). */ /* TODO(shess) The calling code may already know that the end child is ** not worth calculating, because the end may be in a later sibling ** node. Consider whether breaking symmetry is worthwhile. I suspect ** it's not worthwhile. */ static void getChildrenContaining(const char *pData, int nData, const char *pTerm, int nTerm, int isPrefix, sqlite_int64 *piStartChild, sqlite_int64 *piEndChild){ InteriorReader reader; assert( nData>1 ); assert( *pData!='\0' ); interiorReaderInit(pData, nData, &reader); /* Scan for the first child which could contain pTerm/nTerm. */ while( !interiorReaderAtEnd(&reader) ){ if( interiorReaderTermCmp(&reader, pTerm, nTerm, 0)>0 ) break; interiorReaderStep(&reader); } *piStartChild = interiorReaderCurrentBlockid(&reader); /* Keep scanning to find a term greater than our term, using prefix ** comparison if indicated. If isPrefix is false, this will be the ** same blockid as the starting block. */ while( !interiorReaderAtEnd(&reader) ){ if( interiorReaderTermCmp(&reader, pTerm, nTerm, isPrefix)>0 ) break; interiorReaderStep(&reader); } *piEndChild = interiorReaderCurrentBlockid(&reader); interiorReaderDestroy(&reader); /* Children must ascend, and if !prefix, both must be the same. */ assert( *piEndChild>=*piStartChild ); assert( isPrefix || *piStartChild==*piEndChild ); } /* Read block at iBlockid and pass it with other params to ** getChildrenContaining(). */ static int loadAndGetChildrenContaining( fulltext_vtab *v, sqlite_int64 iBlockid, const char *pTerm, int nTerm, int isPrefix, sqlite_int64 *piStartChild, sqlite_int64 *piEndChild ){ sqlite3_stmt *s = NULL; int rc; assert( iBlockid!=0 ); assert( pTerm!=NULL ); assert( nTerm!=0 ); /* TODO(shess) Why not allow this? */ assert( piStartChild!=NULL ); assert( piEndChild!=NULL ); rc = sql_get_statement(v, BLOCK_SELECT_STMT, &s); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_bind_int64(s, 1, iBlockid); if( rc!=SQLITE_OK ) return rc; rc = sqlite3_step(s); if( rc==SQLITE_DONE ) return SQLITE_ERROR; if( rc!=SQLITE_ROW ) return rc; getChildrenContaining(sqlite3_column_blob(s, 0), sqlite3_column_bytes(s, 0), pTerm, nTerm, isPrefix, piStartChild, piEndChild); /* We expect only one row. We must execute another sqlite3_step() * to complete the iteration; otherwise the table will remain * locked. */ rc = sqlite3_step(s); if( rc==SQLITE_ROW ) return SQLITE_ERROR; if( rc!=SQLITE_DONE ) return rc; return SQLITE_OK; } /* Traverse the tree represented by pData[nData] looking for ** pTerm[nTerm], placing its doclist into *out. This is internal to ** loadSegment() to make error-handling cleaner. */ static int loadSegmentInt(fulltext_vtab *v, const char *pData, int nData, sqlite_int64 iLeavesEnd, const char *pTerm, int nTerm, int isPrefix, DataBuffer *out){ /* Special case where root is a leaf. */ if( *pData=='\0' ){ return loadSegmentLeaf(v, pData, nData, pTerm, nTerm, isPrefix, out); }else{ int rc; sqlite_int64 iStartChild, iEndChild; /* Process pData as an interior node, then loop down the tree ** until we find the set of leaf nodes to scan for the term. */ getChildrenContaining(pData, nData, pTerm, nTerm, isPrefix, &iStartChild, &iEndChild); while( iStartChild>iLeavesEnd ){ sqlite_int64 iNextStart, iNextEnd; rc = loadAndGetChildrenContaining(v, iStartChild, pTerm, nTerm, isPrefix, &iNextStart, &iNextEnd); if( rc!=SQLITE_OK ) return rc; /* If we've branched, follow the end branch, too. */ if( iStartChild!=iEndChild ){ sqlite_int64 iDummy; rc = loadAndGetChildrenContaining(v, iEndChild, pTerm, nTerm, isPrefix, &iDummy, &iNextEnd); if( rc!=SQLITE_OK ) return rc; } assert( iNextStart<=iNextEnd ); iStartChild = iNextStart; iEndChild = iNextEnd; } assert( iStartChild<=iLeavesEnd ); assert( iEndChild<=iLeavesEnd ); /* Scan through the leaf segments for doclists. */ return loadSegmentLeaves(v, iStartChild, iEndChild, pTerm, nTerm, isPrefix, out); } } /* Call loadSegmentInt() to collect the doclist for pTerm/nTerm, then ** merge its doclist over *out (any duplicate doclists read from the ** segment rooted at pData will overwrite those in *out). */ /* TODO(shess) Consider changing this to determine the depth of the ** leaves using either the first characters of interior nodes (when ** ==1, we're one level above the leaves), or the first character of ** the root (which will describe the height of the tree directly). ** Either feels somewhat tricky to me. */ /* TODO(shess) The current merge is likely to be slow for large ** doclists (though it should process from newest/smallest to ** oldest/largest, so it may not be that bad). It might be useful to ** modify things to allow for N-way merging. This could either be ** within a segment, with pairwise merges across segments, or across ** all segments at once. */ static int loadSegment(fulltext_vtab *v, const char *pData, int nData, sqlite_int64 iLeavesEnd, const char *pTerm, int nTerm, int isPrefix, DataBuffer *out){ DataBuffer result; int rc; assert( nData>1 ); /* This code should never be called with buffered updates. */ assert( v->nPendingData<0 ); dataBufferInit(&result, 0); rc = loadSegmentInt(v, pData, nData, iLeavesEnd, pTerm, nTerm, isPrefix, &result); if( rc==SQLITE_OK && result.nData>0 ){ if( out->nData==0 ){ DataBuffer tmp = *out; *out = result; result = tmp; }else{ DataBuffer merged; DLReader readers[2]; dlrInit(&readers[0], DL_DEFAULT, out->pData, out->nData); dlrInit(&readers[1], DL_DEFAULT, result.pData, result.nData); dataBufferInit(&merged, out->nData+result.nData); docListMerge(&merged, readers, 2); dataBufferDestroy(out); *out = merged; dlrDestroy(&readers[0]); dlrDestroy(&readers[1]); } } dataBufferDestroy(&result); return rc; } /* Scan the database and merge together the posting lists for the term ** into *out. */ static int termSelect(fulltext_vtab *v, int iColumn, const char *pTerm, int nTerm, int isPrefix, DocListType iType, DataBuffer *out){ DataBuffer doclist; sqlite3_stmt *s; int rc = sql_get_statement(v, SEGDIR_SELECT_ALL_STMT, &s); if( rc!=SQLITE_OK ) return rc; /* This code should never be called with buffered updates. */ assert( v->nPendingData<0 ); dataBufferInit(&doclist, 0); /* Traverse the segments from oldest to newest so that newer doclist ** elements for given docids overwrite older elements. */ while( (rc = sqlite3_step(s))==SQLITE_ROW ){ const char *pData = sqlite3_column_blob(s, 0); const int nData = sqlite3_column_bytes(s, 0); const sqlite_int64 iLeavesEnd = sqlite3_column_int64(s, 1); rc = loadSegment(v, pData, nData, iLeavesEnd, pTerm, nTerm, isPrefix, &doclist); if( rc!=SQLITE_OK ) goto err; } if( rc==SQLITE_DONE ){ if( doclist.nData!=0 ){ /* TODO(shess) The old term_select_all() code applied the column ** restrict as we merged segments, leading to smaller buffers. ** This is probably worthwhile to bring back, once the new storage ** system is checked in. */ if( iColumn==v->nColumn) iColumn = -1; docListTrim(DL_DEFAULT, doclist.pData, doclist.nData, iColumn, iType, out); } rc = SQLITE_OK; } err: dataBufferDestroy(&doclist); return rc; } /****************************************************************/ /* Used to hold hashtable data for sorting. */ typedef struct TermData { const char *pTerm; int nTerm; DLCollector *pCollector; } TermData; /* Orders TermData elements in strcmp fashion ( <0 for less-than, 0 ** for equal, >0 for greater-than). */ static int termDataCmp(const void *av, const void *bv){ const TermData *a = (const TermData *)av; const TermData *b = (const TermData *)bv; int n = a->nTerm<b->nTerm ? a->nTerm : b->nTerm; int c = memcmp(a->pTerm, b->pTerm, n); if( c!=0 ) return c; return a->nTerm-b->nTerm; } /* Order pTerms data by term, then write a new level 0 segment using ** LeafWriter. */ static int writeZeroSegment(fulltext_vtab *v, fts3Hash *pTerms){ fts3HashElem *e; int idx, rc, i, n; TermData *pData; LeafWriter writer; DataBuffer dl; /* Determine the next index at level 0, merging as necessary. */ rc = segdirNextIndex(v, 0, &idx); if( rc!=SQLITE_OK ) return rc; n = fts3HashCount(pTerms); pData = malloc(n*sizeof(TermData)); for(i = 0, e = fts3HashFirst(pTerms); e; i++, e = fts3HashNext(e)){ assert( i<n ); pData[i].pTerm = fts3HashKey(e); pData[i].nTerm = fts3HashKeysize(e); pData[i].pCollector = fts3HashData(e); } assert( i==n ); /* TODO(shess) Should we allow user-defined collation sequences, ** here? I think we only need that once we support prefix searches. */ if( n>1 ) qsort(pData, n, sizeof(*pData), termDataCmp); /* TODO(shess) Refactor so that we can write directly to the segment ** DataBuffer, as happens for segment merges. */ leafWriterInit(0, idx, &writer); dataBufferInit(&dl, 0); for(i=0; i<n; i++){ dataBufferReset(&dl); dlcAddDoclist(pData[i].pCollector, &dl); rc = leafWriterStep(v, &writer, pData[i].pTerm, pData[i].nTerm, dl.pData, dl.nData); if( rc!=SQLITE_OK ) goto err; } rc = leafWriterFinalize(v, &writer); err: dataBufferDestroy(&dl); free(pData); leafWriterDestroy(&writer); return rc; } /* If pendingTerms has data, free it. */ static int clearPendingTerms(fulltext_vtab *v){ if( v->nPendingData>=0 ){ fts3HashElem *e; for(e=fts3HashFirst(&v->pendingTerms); e; e=fts3HashNext(e)){ dlcDelete(fts3HashData(e)); } fts3HashClear(&v->pendingTerms); v->nPendingData = -1; } return SQLITE_OK; } /* If pendingTerms has data, flush it to a level-zero segment, and ** free it. */ static int flushPendingTerms(fulltext_vtab *v){ if( v->nPendingData>=0 ){ int rc = writeZeroSegment(v, &v->pendingTerms); if( rc==SQLITE_OK ) clearPendingTerms(v); return rc; } return SQLITE_OK; } /* If pendingTerms is "too big", or docid is out of order, flush it. ** Regardless, be certain that pendingTerms is initialized for use. */ static int initPendingTerms(fulltext_vtab *v, sqlite_int64 iDocid){ /* TODO(shess) Explore whether partially flushing the buffer on ** forced-flush would provide better performance. I suspect that if ** we ordered the doclists by size and flushed the largest until the ** buffer was half empty, that would let the less frequent terms ** generate longer doclists. */ if( iDocid<=v->iPrevDocid || v->nPendingData>kPendingThreshold ){ int rc = flushPendingTerms(v); if( rc!=SQLITE_OK ) return rc; } if( v->nPendingData<0 ){ fts3HashInit(&v->pendingTerms, FTS3_HASH_STRING, 1); v->nPendingData = 0; } v->iPrevDocid = iDocid; return SQLITE_OK; } /* This function implements the xUpdate callback; it's the top-level entry * point for inserting, deleting or updating a row in a full-text table. */ static int fulltextUpdate(sqlite3_vtab *pVtab, int nArg, sqlite3_value **ppArg, sqlite_int64 *pRowid){ fulltext_vtab *v = (fulltext_vtab *) pVtab; int rc; TRACE(("FTS3 Update %p\n", pVtab)); if( nArg<2 ){ rc = index_delete(v, sqlite3_value_int64(ppArg[0])); } else if( sqlite3_value_type(ppArg[0]) != SQLITE_NULL ){ /* An update: * ppArg[0] = old rowid * ppArg[1] = new rowid * ppArg[2..2+v->nColumn-1] = values * ppArg[2+v->nColumn] = value for magic column (we ignore this) */ sqlite_int64 rowid = sqlite3_value_int64(ppArg[0]); if( sqlite3_value_type(ppArg[1]) != SQLITE_INTEGER || sqlite3_value_int64(ppArg[1]) != rowid ){ rc = SQLITE_ERROR; /* we don't allow changing the rowid */ } else { assert( nArg==2+v->nColumn+1); rc = index_update(v, rowid, &ppArg[2]); } } else { /* An insert: * ppArg[1] = requested rowid * ppArg[2..2+v->nColumn-1] = values * ppArg[2+v->nColumn] = value for magic column (we ignore this) */ assert( nArg==2+v->nColumn+1); rc = index_insert(v, ppArg[1], &ppArg[2], pRowid); } return rc; } static int fulltextSync(sqlite3_vtab *pVtab){ TRACE(("FTS3 xSync()\n")); return flushPendingTerms((fulltext_vtab *)pVtab); } static int fulltextBegin(sqlite3_vtab *pVtab){ fulltext_vtab *v = (fulltext_vtab *) pVtab; TRACE(("FTS3 xBegin()\n")); /* Any buffered updates should have been cleared by the previous ** transaction. */ assert( v->nPendingData<0 ); return clearPendingTerms(v); } static int fulltextCommit(sqlite3_vtab *pVtab){ fulltext_vtab *v = (fulltext_vtab *) pVtab; TRACE(("FTS3 xCommit()\n")); /* Buffered updates should have been cleared by fulltextSync(). */ assert( v->nPendingData<0 ); return clearPendingTerms(v); } static int fulltextRollback(sqlite3_vtab *pVtab){ TRACE(("FTS3 xRollback()\n")); return clearPendingTerms((fulltext_vtab *)pVtab); } /* ** Implementation of the snippet() function for FTS3 */ static void snippetFunc( sqlite3_context *pContext, int argc, sqlite3_value **argv ){ fulltext_cursor *pCursor; if( argc<1 ) return; if( sqlite3_value_type(argv[0])!=SQLITE_BLOB || sqlite3_value_bytes(argv[0])!=sizeof(pCursor) ){ sqlite3_result_error(pContext, "illegal first argument to html_snippet",-1); }else{ const char *zStart = "<b>"; const char *zEnd = "</b>"; const char *zEllipsis = "<b>...</b>"; memcpy(&pCursor, sqlite3_value_blob(argv[0]), sizeof(pCursor)); if( argc>=2 ){ zStart = (const char*)sqlite3_value_text(argv[1]); if( argc>=3 ){ zEnd = (const char*)sqlite3_value_text(argv[2]); if( argc>=4 ){ zEllipsis = (const char*)sqlite3_value_text(argv[3]); } } } snippetAllOffsets(pCursor); snippetText(pCursor, zStart, zEnd, zEllipsis); sqlite3_result_text(pContext, pCursor->snippet.zSnippet, pCursor->snippet.nSnippet, SQLITE_STATIC); } } /* ** Implementation of the offsets() function for FTS3 */ static void snippetOffsetsFunc( sqlite3_context *pContext, int argc, sqlite3_value **argv ){ fulltext_cursor *pCursor; if( argc<1 ) return; if( sqlite3_value_type(argv[0])!=SQLITE_BLOB || sqlite3_value_bytes(argv[0])!=sizeof(pCursor) ){ sqlite3_result_error(pContext, "illegal first argument to offsets",-1); }else{ memcpy(&pCursor, sqlite3_value_blob(argv[0]), sizeof(pCursor)); snippetAllOffsets(pCursor); snippetOffsetText(&pCursor->snippet); sqlite3_result_text(pContext, pCursor->snippet.zOffset, pCursor->snippet.nOffset, SQLITE_STATIC); } } /* ** This routine implements the xFindFunction method for the FTS3 ** virtual table. */ static int fulltextFindFunction( sqlite3_vtab *pVtab, int nArg, const char *zName, void (**pxFunc)(sqlite3_context*,int,sqlite3_value**), void **ppArg ){ if( strcmp(zName,"snippet")==0 ){ *pxFunc = snippetFunc; return 1; }else if( strcmp(zName,"offsets")==0 ){ *pxFunc = snippetOffsetsFunc; return 1; } return 0; } /* ** Rename an fts3 table. */ static int fulltextRename( sqlite3_vtab *pVtab, const char *zName ){ fulltext_vtab *p = (fulltext_vtab *)pVtab; int rc = SQLITE_NOMEM; char *zSql = sqlite3_mprintf( "ALTER TABLE %Q.'%q_content' RENAME TO '%q_content';" "ALTER TABLE %Q.'%q_segments' RENAME TO '%q_segments';" "ALTER TABLE %Q.'%q_segdir' RENAME TO '%q_segdir';" , p->zDb, p->zName, zName , p->zDb, p->zName, zName , p->zDb, p->zName, zName ); if( zSql ){ rc = sqlite3_exec(p->db, zSql, 0, 0, 0); sqlite3_free(zSql); } return rc; } static const sqlite3_module fts3Module = { /* iVersion */ 0, /* xCreate */ fulltextCreate, /* xConnect */ fulltextConnect, /* xBestIndex */ fulltextBestIndex, /* xDisconnect */ fulltextDisconnect, /* xDestroy */ fulltextDestroy, /* xOpen */ fulltextOpen, /* xClose */ fulltextClose, /* xFilter */ fulltextFilter, /* xNext */ fulltextNext, /* xEof */ fulltextEof, /* xColumn */ fulltextColumn, /* xRowid */ fulltextRowid, /* xUpdate */ fulltextUpdate, /* xBegin */ fulltextBegin, /* xSync */ fulltextSync, /* xCommit */ fulltextCommit, /* xRollback */ fulltextRollback, /* xFindFunction */ fulltextFindFunction, /* xRename */ fulltextRename, }; static void hashDestroy(void *p){ fts3Hash *pHash = (fts3Hash *)p; sqlite3Fts3HashClear(pHash); sqlite3_free(pHash); } /* ** The fts3 built-in tokenizers - "simple" and "porter" - are implemented ** in files fts3_tokenizer1.c and fts3_porter.c respectively. The following ** two forward declarations are for functions declared in these files ** used to retrieve the respective implementations. ** ** Calling sqlite3Fts3SimpleTokenizerModule() sets the value pointed ** to by the argument to point a the "simple" tokenizer implementation. ** Function ...PorterTokenizerModule() sets *pModule to point to the ** porter tokenizer/stemmer implementation. */ void sqlite3Fts3SimpleTokenizerModule(sqlite3_tokenizer_module const**ppModule); void sqlite3Fts3PorterTokenizerModule(sqlite3_tokenizer_module const**ppModule); void sqlite3Fts3IcuTokenizerModule(sqlite3_tokenizer_module const**ppModule); int sqlite3Fts3InitHashTable(sqlite3 *, fts3Hash *, const char *); /* ** Initialise the fts3 extension. If this extension is built as part ** of the sqlite library, then this function is called directly by ** SQLite. If fts3 is built as a dynamically loadable extension, this ** function is called by the sqlite3_extension_init() entry point. */ int sqlite3Fts3Init(sqlite3 *db){ int rc = SQLITE_OK; fts3Hash *pHash = 0; const sqlite3_tokenizer_module *pSimple = 0; const sqlite3_tokenizer_module *pPorter = 0; const sqlite3_tokenizer_module *pIcu = 0; sqlite3Fts3SimpleTokenizerModule(&pSimple); sqlite3Fts3PorterTokenizerModule(&pPorter); #ifdef SQLITE_ENABLE_ICU sqlite3Fts3IcuTokenizerModule(&pIcu); #endif /* Allocate and initialise the hash-table used to store tokenizers. */ pHash = sqlite3_malloc(sizeof(fts3Hash)); if( !pHash ){ rc = SQLITE_NOMEM; }else{ sqlite3Fts3HashInit(pHash, FTS3_HASH_STRING, 1); } /* Load the built-in tokenizers into the hash table */ if( rc==SQLITE_OK ){ if( sqlite3Fts3HashInsert(pHash, "simple", 7, (void *)pSimple) || sqlite3Fts3HashInsert(pHash, "porter", 7, (void *)pPorter) || (pIcu && sqlite3Fts3HashInsert(pHash, "icu", 4, (void *)pIcu)) ){ rc = SQLITE_NOMEM; } } /* Create the virtual table wrapper around the hash-table and overload ** the two scalar functions. If this is successful, register the ** module with sqlite. */ if( SQLITE_OK==rc && SQLITE_OK==(rc = sqlite3Fts3InitHashTable(db, pHash, "fts3_tokenizer")) && SQLITE_OK==(rc = sqlite3_overload_function(db, "snippet", -1)) && SQLITE_OK==(rc = sqlite3_overload_function(db, "offsets", -1)) ){ return sqlite3_create_module_v2( db, "fts3", &fts3Module, (void *)pHash, hashDestroy ); } /* An error has occured. Delete the hash table and return the error code. */ assert( rc!=SQLITE_OK ); if( pHash ){ sqlite3Fts3HashClear(pHash); sqlite3_free(pHash); } return rc; } #if !SQLITE_CORE int sqlite3_extension_init( sqlite3 *db, char **pzErrMsg, const sqlite3_api_routines *pApi ){ SQLITE_EXTENSION_INIT2(pApi) return sqlite3Fts3Init(db); } #endif #endif /* !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS3) */ |
Added ext/fts3/fts3.h.
> > > > > > > > > > > > > > > > > > > > > > > > > > | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 | /* ** 2006 Oct 10 ** ** The author disclaims copyright to this source code. In place of ** a legal notice, here is a blessing: ** ** May you do good and not evil. ** May you find forgiveness for yourself and forgive others. ** May you share freely, never taking more than you give. ** ****************************************************************************** ** ** This header file is used by programs that want to link against the ** FTS3 library. All it does is declare the sqlite3Fts3Init() interface. */ #include "sqlite3.h" #ifdef __cplusplus extern "C" { #endif /* __cplusplus */ int sqlite3Fts3Init(sqlite3 *db); #ifdef __cplusplus } /* extern "C" */ #endif /* __cplusplus */ |
Added ext/fts3/fts3_hash.c.
> > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 | /* ** 2001 September 22 ** ** The author disclaims copyright to this source code. In place of ** a legal notice, here is a blessing: ** ** May you do good and not evil. ** May you find forgiveness for yourself and forgive others. ** May you share freely, never taking more than you give. ** ************************************************************************* ** This is the implementation of generic hash-tables used in SQLite. ** We've modified it slightly to serve as a standalone hash table ** implementation for the full-text indexing module. */ /* ** The code in this file is only compiled if: ** ** * The FTS3 module is being built as an extension ** (in which case SQLITE_CORE is not defined), or ** ** * The FTS3 module is being built into the core of ** SQLite (in which case SQLITE_ENABLE_FTS3 is defined). */ #if !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS3) #include <assert.h> #include <stdlib.h> #include <string.h> #include "fts3_hash.h" static void *malloc_and_zero(int n){ void *p = malloc(n); if( p ){ memset(p, 0, n); } return p; } /* Turn bulk memory into a hash table object by initializing the ** fields of the Hash structure. ** ** "pNew" is a pointer to the hash table that is to be initialized. ** keyClass is one of the constants ** FTS3_HASH_BINARY or FTS3_HASH_STRING. The value of keyClass ** determines what kind of key the hash table will use. "copyKey" is ** true if the hash table should make its own private copy of keys and ** false if it should just use the supplied pointer. */ void sqlite3Fts3HashInit(fts3Hash *pNew, int keyClass, int copyKey){ assert( pNew!=0 ); assert( keyClass>=FTS3_HASH_STRING && keyClass<=FTS3_HASH_BINARY ); pNew->keyClass = keyClass; pNew->copyKey = copyKey; pNew->first = 0; pNew->count = 0; pNew->htsize = 0; pNew->ht = 0; pNew->xMalloc = malloc_and_zero; pNew->xFree = free; } /* Remove all entries from a hash table. Reclaim all memory. ** Call this routine to delete a hash table or to reset a hash table ** to the empty state. */ void sqlite3Fts3HashClear(fts3Hash *pH){ fts3HashElem *elem; /* For looping over all elements of the table */ assert( pH!=0 ); elem = pH->first; pH->first = 0; if( pH->ht ) pH->xFree(pH->ht); pH->ht = 0; pH->htsize = 0; while( elem ){ fts3HashElem *next_elem = elem->next; if( pH->copyKey && elem->pKey ){ pH->xFree(elem->pKey); } pH->xFree(elem); elem = next_elem; } pH->count = 0; } /* ** Hash and comparison functions when the mode is FTS3_HASH_STRING */ static int strHash(const void *pKey, int nKey){ const char *z = (const char *)pKey; int h = 0; if( nKey<=0 ) nKey = (int) strlen(z); while( nKey > 0 ){ h = (h<<3) ^ h ^ *z++; nKey--; } return h & 0x7fffffff; } static int strCompare(const void *pKey1, int n1, const void *pKey2, int n2){ if( n1!=n2 ) return 1; return strncmp((const char*)pKey1,(const char*)pKey2,n1); } /* ** Hash and comparison functions when the mode is FTS3_HASH_BINARY */ static int binHash(const void *pKey, int nKey){ int h = 0; const char *z = (const char *)pKey; while( nKey-- > 0 ){ h = (h<<3) ^ h ^ *(z++); } return h & 0x7fffffff; } static int binCompare(const void *pKey1, int n1, const void *pKey2, int n2){ if( n1!=n2 ) return 1; return memcmp(pKey1,pKey2,n1); } /* ** Return a pointer to the appropriate hash function given the key class. ** ** The C syntax in this function definition may be unfamilar to some ** programmers, so we provide the following additional explanation: ** ** The name of the function is "hashFunction". The function takes a ** single parameter "keyClass". The return value of hashFunction() ** is a pointer to another function. Specifically, the return value ** of hashFunction() is a pointer to a function that takes two parameters ** with types "const void*" and "int" and returns an "int". */ static int (*hashFunction(int keyClass))(const void*,int){ if( keyClass==FTS3_HASH_STRING ){ return &strHash; }else{ assert( keyClass==FTS3_HASH_BINARY ); return &binHash; } } /* ** Return a pointer to the appropriate hash function given the key class. ** ** For help in interpreted the obscure C code in the function definition, ** see the header comment on the previous function. */ static int (*compareFunction(int keyClass))(const void*,int,const void*,int){ if( keyClass==FTS3_HASH_STRING ){ return &strCompare; }else{ assert( keyClass==FTS3_HASH_BINARY ); return &binCompare; } } /* Link an element into the hash table */ static void insertElement( fts3Hash *pH, /* The complete hash table */ struct _fts3ht *pEntry, /* The entry into which pNew is inserted */ fts3HashElem *pNew /* The element to be inserted */ ){ fts3HashElem *pHead; /* First element already in pEntry */ pHead = pEntry->chain; if( pHead ){ pNew->next = pHead; pNew->prev = pHead->prev; if( pHead->prev ){ pHead->prev->next = pNew; } else { pH->first = pNew; } pHead->prev = pNew; }else{ pNew->next = pH->first; if( pH->first ){ pH->first->prev = pNew; } pNew->prev = 0; pH->first = pNew; } pEntry->count++; pEntry->chain = pNew; } /* Resize the hash table so that it cantains "new_size" buckets. ** "new_size" must be a power of 2. The hash table might fail ** to resize if sqliteMalloc() fails. */ static void rehash(fts3Hash *pH, int new_size){ struct _fts3ht *new_ht; /* The new hash table */ fts3HashElem *elem, *next_elem; /* For looping over existing elements */ int (*xHash)(const void*,int); /* The hash function */ assert( (new_size & (new_size-1))==0 ); new_ht = (struct _fts3ht *)pH->xMalloc( new_size*sizeof(struct _fts3ht) ); if( new_ht==0 ) return; if( pH->ht ) pH->xFree(pH->ht); pH->ht = new_ht; pH->htsize = new_size; xHash = hashFunction(pH->keyClass); for(elem=pH->first, pH->first=0; elem; elem = next_elem){ int h = (*xHash)(elem->pKey, elem->nKey) & (new_size-1); next_elem = elem->next; insertElement(pH, &new_ht[h], elem); } } /* This function (for internal use only) locates an element in an ** hash table that matches the given key. The hash for this key has ** already been computed and is passed as the 4th parameter. */ static fts3HashElem *findElementGivenHash( const fts3Hash *pH, /* The pH to be searched */ const void *pKey, /* The key we are searching for */ int nKey, int h /* The hash for this key. */ ){ fts3HashElem *elem; /* Used to loop thru the element list */ int count; /* Number of elements left to test */ int (*xCompare)(const void*,int,const void*,int); /* comparison function */ if( pH->ht ){ struct _fts3ht *pEntry = &pH->ht[h]; elem = pEntry->chain; count = pEntry->count; xCompare = compareFunction(pH->keyClass); while( count-- && elem ){ if( (*xCompare)(elem->pKey,elem->nKey,pKey,nKey)==0 ){ return elem; } elem = elem->next; } } return 0; } /* Remove a single entry from the hash table given a pointer to that ** element and a hash on the element's key. */ static void removeElementGivenHash( fts3Hash *pH, /* The pH containing "elem" */ fts3HashElem* elem, /* The element to be removed from the pH */ int h /* Hash value for the element */ ){ struct _fts3ht *pEntry; if( elem->prev ){ elem->prev->next = elem->next; }else{ pH->first = elem->next; } if( elem->next ){ elem->next->prev = elem->prev; } pEntry = &pH->ht[h]; if( pEntry->chain==elem ){ pEntry->chain = elem->next; } pEntry->count--; if( pEntry->count<=0 ){ pEntry->chain = 0; } if( pH->copyKey && elem->pKey ){ pH->xFree(elem->pKey); } pH->xFree( elem ); pH->count--; if( pH->count<=0 ){ assert( pH->first==0 ); assert( pH->count==0 ); fts3HashClear(pH); } } /* Attempt to locate an element of the hash table pH with a key ** that matches pKey,nKey. Return the data for this element if it is ** found, or NULL if there is no match. */ void *sqlite3Fts3HashFind(const fts3Hash *pH, const void *pKey, int nKey){ int h; /* A hash on key */ fts3HashElem *elem; /* The element that matches key */ int (*xHash)(const void*,int); /* The hash function */ if( pH==0 || pH->ht==0 ) return 0; xHash = hashFunction(pH->keyClass); assert( xHash!=0 ); h = (*xHash)(pKey,nKey); assert( (pH->htsize & (pH->htsize-1))==0 ); elem = findElementGivenHash(pH,pKey,nKey, h & (pH->htsize-1)); return elem ? elem->data : 0; } /* Insert an element into the hash table pH. The key is pKey,nKey ** and the data is "data". ** ** If no element exists with a matching key, then a new ** element is created. A copy of the key is made if the copyKey ** flag is set. NULL is returned. ** ** If another element already exists with the same key, then the ** new data replaces the old data and the old data is returned. ** The key is not copied in this instance. If a malloc fails, then ** the new data is returned and the hash table is unchanged. ** ** If the "data" parameter to this function is NULL, then the ** element corresponding to "key" is removed from the hash table. */ void *sqlite3Fts3HashInsert( fts3Hash *pH, /* The hash table to insert into */ const void *pKey, /* The key */ int nKey, /* Number of bytes in the key */ void *data /* The data */ ){ int hraw; /* Raw hash value of the key */ int h; /* the hash of the key modulo hash table size */ fts3HashElem *elem; /* Used to loop thru the element list */ fts3HashElem *new_elem; /* New element added to the pH */ int (*xHash)(const void*,int); /* The hash function */ assert( pH!=0 ); xHash = hashFunction(pH->keyClass); assert( xHash!=0 ); hraw = (*xHash)(pKey, nKey); assert( (pH->htsize & (pH->htsize-1))==0 ); h = hraw & (pH->htsize-1); elem = findElementGivenHash(pH,pKey,nKey,h); if( elem ){ void *old_data = elem->data; if( data==0 ){ removeElementGivenHash(pH,elem,h); }else{ elem->data = data; } return old_data; } if( data==0 ) return 0; new_elem = (fts3HashElem*)pH->xMalloc( sizeof(fts3HashElem) ); if( new_elem==0 ) return data; if( pH->copyKey && pKey!=0 ){ new_elem->pKey = pH->xMalloc( nKey ); if( new_elem->pKey==0 ){ pH->xFree(new_elem); return data; } memcpy((void*)new_elem->pKey, pKey, nKey); }else{ new_elem->pKey = (void*)pKey; } new_elem->nKey = nKey; pH->count++; if( pH->htsize==0 ){ rehash(pH,8); if( pH->htsize==0 ){ pH->count = 0; pH->xFree(new_elem); return data; } } if( pH->count > pH->htsize ){ rehash(pH,pH->htsize*2); } assert( pH->htsize>0 ); assert( (pH->htsize & (pH->htsize-1))==0 ); h = hraw & (pH->htsize-1); insertElement(pH, &pH->ht[h], new_elem); new_elem->data = data; return 0; } #endif /* !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS3) */ |
Added ext/fts3/fts3_hash.h.
> > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 | /* ** 2001 September 22 ** ** The author disclaims copyright to this source code. In place of ** a legal notice, here is a blessing: ** ** May you do good and not evil. ** May you find forgiveness for yourself and forgive others. ** May you share freely, never taking more than you give. ** ************************************************************************* ** This is the header file for the generic hash-table implemenation ** used in SQLite. We've modified it slightly to serve as a standalone ** hash table implementation for the full-text indexing module. ** */ #ifndef _FTS3_HASH_H_ #define _FTS3_HASH_H_ /* Forward declarations of structures. */ typedef struct fts3Hash fts3Hash; typedef struct fts3HashElem fts3HashElem; /* A complete hash table is an instance of the following structure. ** The internals of this structure are intended to be opaque -- client ** code should not attempt to access or modify the fields of this structure ** directly. Change this structure only by using the routines below. ** However, many of the "procedures" and "functions" for modifying and ** accessing this structure are really macros, so we can't really make ** this structure opaque. */ struct fts3Hash { char keyClass; /* HASH_INT, _POINTER, _STRING, _BINARY */ char copyKey; /* True if copy of key made on insert */ int count; /* Number of entries in this table */ fts3HashElem *first; /* The first element of the array */ void *(*xMalloc)(int); /* malloc() function to use */ void (*xFree)(void *); /* free() function to use */ int htsize; /* Number of buckets in the hash table */ struct _fts3ht { /* the hash table */ int count; /* Number of entries with this hash */ fts3HashElem *chain; /* Pointer to first entry with this hash */ } *ht; }; /* Each element in the hash table is an instance of the following ** structure. All elements are stored on a single doubly-linked list. ** ** Again, this structure is intended to be opaque, but it can't really ** be opaque because it is used by macros. */ struct fts3HashElem { fts3HashElem *next, *prev; /* Next and previous elements in the table */ void *data; /* Data associated with this element */ void *pKey; int nKey; /* Key associated with this element */ }; /* ** There are 2 different modes of operation for a hash table: ** ** FTS3_HASH_STRING pKey points to a string that is nKey bytes long ** (including the null-terminator, if any). Case ** is respected in comparisons. ** ** FTS3_HASH_BINARY pKey points to binary data nKey bytes long. ** memcmp() is used to compare keys. ** ** A copy of the key is made if the copyKey parameter to fts3HashInit is 1. */ #define FTS3_HASH_STRING 1 #define FTS3_HASH_BINARY 2 /* ** Access routines. To delete, insert a NULL pointer. */ void sqlite3Fts3HashInit(fts3Hash*, int keytype, int copyKey); void *sqlite3Fts3HashInsert(fts3Hash*, const void *pKey, int nKey, void *pData); void *sqlite3Fts3HashFind(const fts3Hash*, const void *pKey, int nKey); void sqlite3Fts3HashClear(fts3Hash*); /* ** Shorthand for the functions above */ #define fts3HashInit sqlite3Fts3HashInit #define fts3HashInsert sqlite3Fts3HashInsert #define fts3HashFind sqlite3Fts3HashFind #define fts3HashClear sqlite3Fts3HashClear /* ** Macros for looping over all elements of a hash table. The idiom is ** like this: ** ** fts3Hash h; ** fts3HashElem *p; ** ... ** for(p=fts3HashFirst(&h); p; p=fts3HashNext(p)){ ** SomeStructure *pData = fts3HashData(p); ** // do something with pData ** } */ #define fts3HashFirst(H) ((H)->first) #define fts3HashNext(E) ((E)->next) #define fts3HashData(E) ((E)->data) #define fts3HashKey(E) ((E)->pKey) #define fts3HashKeysize(E) ((E)->nKey) /* ** Number of entries in a hash table */ #define fts3HashCount(H) ((H)->count) #endif /* _FTS3_HASH_H_ */ |
Added ext/fts3/fts3_icu.c.
> > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 | /* ** 2007 June 22 ** ** The author disclaims copyright to this source code. In place of ** a legal notice, here is a blessing: ** ** May you do good and not evil. ** May you find forgiveness for yourself and forgive others. ** May you share freely, never taking more than you give. ** ************************************************************************* ** This file implements a tokenizer for fts3 based on the ICU library. ** ** $Id: fts3_icu.c,v 1.1 2007/08/20 17:37:04 shess Exp $ */ #if !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS3) #ifdef SQLITE_ENABLE_ICU #include <assert.h> #include <string.h> #include "fts3_tokenizer.h" #include <unicode/ubrk.h> #include <unicode/ucol.h> #include <unicode/ustring.h> #include <unicode/utf16.h> typedef struct IcuTokenizer IcuTokenizer; typedef struct IcuCursor IcuCursor; struct IcuTokenizer { sqlite3_tokenizer base; char *zLocale; }; struct IcuCursor { sqlite3_tokenizer_cursor base; UBreakIterator *pIter; /* ICU break-iterator object */ int nChar; /* Number of UChar elements in pInput */ UChar *aChar; /* Copy of input using utf-16 encoding */ int *aOffset; /* Offsets of each character in utf-8 input */ int nBuffer; char *zBuffer; int iToken; }; /* ** Create a new tokenizer instance. */ static int icuCreate( int argc, /* Number of entries in argv[] */ const char * const *argv, /* Tokenizer creation arguments */ sqlite3_tokenizer **ppTokenizer /* OUT: Created tokenizer */ ){ IcuTokenizer *p; int n = 0; if( argc>0 ){ n = strlen(argv[0])+1; } p = (IcuTokenizer *)sqlite3_malloc(sizeof(IcuTokenizer)+n); if( !p ){ return SQLITE_NOMEM; } memset(p, 0, sizeof(IcuTokenizer)); if( n ){ p->zLocale = (char *)&p[1]; memcpy(p->zLocale, argv[0], n); } *ppTokenizer = (sqlite3_tokenizer *)p; return SQLITE_OK; } /* ** Destroy a tokenizer */ static int icuDestroy(sqlite3_tokenizer *pTokenizer){ IcuTokenizer *p = (IcuTokenizer *)pTokenizer; sqlite3_free(p); return SQLITE_OK; } /* ** Prepare to begin tokenizing a particular string. The input ** string to be tokenized is pInput[0..nBytes-1]. A cursor ** used to incrementally tokenize this string is returned in ** *ppCursor. */ static int icuOpen( sqlite3_tokenizer *pTokenizer, /* The tokenizer */ const char *zInput, /* Input string */ int nInput, /* Length of zInput in bytes */ sqlite3_tokenizer_cursor **ppCursor /* OUT: Tokenization cursor */ ){ IcuTokenizer *p = (IcuTokenizer *)pTokenizer; IcuCursor *pCsr; const int32_t opt = U_FOLD_CASE_DEFAULT; UErrorCode status = U_ZERO_ERROR; int nChar; UChar32 c; int iInput = 0; int iOut = 0; *ppCursor = 0; nChar = nInput+1; pCsr = (IcuCursor *)sqlite3_malloc( sizeof(IcuCursor) + /* IcuCursor */ nChar * sizeof(UChar) + /* IcuCursor.aChar[] */ (nChar+1) * sizeof(int) /* IcuCursor.aOffset[] */ ); if( !pCsr ){ return SQLITE_NOMEM; } memset(pCsr, 0, sizeof(IcuCursor)); pCsr->aChar = (UChar *)&pCsr[1]; pCsr->aOffset = (int *)&pCsr->aChar[nChar]; pCsr->aOffset[iOut] = iInput; U8_NEXT(zInput, iInput, nInput, c); while( c>0 ){ int isError = 0; c = u_foldCase(c, opt); U16_APPEND(pCsr->aChar, iOut, nChar, c, isError); if( isError ){ sqlite3_free(pCsr); return SQLITE_ERROR; } pCsr->aOffset[iOut] = iInput; if( iInput<nInput ){ U8_NEXT(zInput, iInput, nInput, c); }else{ c = 0; } } pCsr->pIter = ubrk_open(UBRK_WORD, p->zLocale, pCsr->aChar, iOut, &status); if( !U_SUCCESS(status) ){ sqlite3_free(pCsr); return SQLITE_ERROR; } pCsr->nChar = iOut; ubrk_first(pCsr->pIter); *ppCursor = (sqlite3_tokenizer_cursor *)pCsr; return SQLITE_OK; } /* ** Close a tokenization cursor previously opened by a call to icuOpen(). */ static int icuClose(sqlite3_tokenizer_cursor *pCursor){ IcuCursor *pCsr = (IcuCursor *)pCursor; ubrk_close(pCsr->pIter); sqlite3_free(pCsr->zBuffer); sqlite3_free(pCsr); return SQLITE_OK; } /* ** Extract the next token from a tokenization cursor. */ static int icuNext( sqlite3_tokenizer_cursor *pCursor, /* Cursor returned by simpleOpen */ const char **ppToken, /* OUT: *ppToken is the token text */ int *pnBytes, /* OUT: Number of bytes in token */ int *piStartOffset, /* OUT: Starting offset of token */ int *piEndOffset, /* OUT: Ending offset of token */ int *piPosition /* OUT: Position integer of token */ ){ IcuCursor *pCsr = (IcuCursor *)pCursor; int iStart = 0; int iEnd = 0; int nByte = 0; while( iStart==iEnd ){ UChar32 c; iStart = ubrk_current(pCsr->pIter); iEnd = ubrk_next(pCsr->pIter); if( iEnd==UBRK_DONE ){ return SQLITE_DONE; } while( iStart<iEnd ){ int iWhite = iStart; U8_NEXT(pCsr->aChar, iWhite, pCsr->nChar, c); if( u_isspace(c) ){ iStart = iWhite; }else{ break; } } assert(iStart<=iEnd); } do { UErrorCode status = U_ZERO_ERROR; if( nByte ){ char *zNew = sqlite3_realloc(pCsr->zBuffer, nByte); if( !zNew ){ return SQLITE_NOMEM; } pCsr->zBuffer = zNew; pCsr->nBuffer = nByte; } u_strToUTF8( pCsr->zBuffer, pCsr->nBuffer, &nByte, /* Output vars */ &pCsr->aChar[iStart], iEnd-iStart, /* Input vars */ &status /* Output success/failure */ ); } while( nByte>pCsr->nBuffer ); *ppToken = pCsr->zBuffer; *pnBytes = nByte; *piStartOffset = pCsr->aOffset[iStart]; *piEndOffset = pCsr->aOffset[iEnd]; *piPosition = pCsr->iToken++; return SQLITE_OK; } /* ** The set of routines that implement the simple tokenizer */ static const sqlite3_tokenizer_module icuTokenizerModule = { 0, /* iVersion */ icuCreate, /* xCreate */ icuDestroy, /* xCreate */ icuOpen, /* xOpen */ icuClose, /* xClose */ icuNext, /* xNext */ }; /* ** Set *ppModule to point at the implementation of the ICU tokenizer. */ void sqlite3Fts3IcuTokenizerModule( sqlite3_tokenizer_module const**ppModule ){ *ppModule = &icuTokenizerModule; } #endif /* defined(SQLITE_ENABLE_ICU) */ #endif /* !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS3) */ |
Added ext/fts3/fts3_porter.c.
> > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 | /* ** 2006 September 30 ** ** The author disclaims copyright to this source code. In place of ** a legal notice, here is a blessing: ** ** May you do good and not evil. ** May you find forgiveness for yourself and forgive others. ** May you share freely, never taking more than you give. ** ************************************************************************* ** Implementation of the full-text-search tokenizer that implements ** a Porter stemmer. */ /* ** The code in this file is only compiled if: ** ** * The FTS3 module is being built as an extension ** (in which case SQLITE_CORE is not defined), or ** ** * The FTS3 module is being built into the core of ** SQLite (in which case SQLITE_ENABLE_FTS3 is defined). */ #if !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS3) #include <assert.h> #include <stdlib.h> #include <stdio.h> #include <string.h> #include <ctype.h> #include "fts3_tokenizer.h" /* ** Class derived from sqlite3_tokenizer */ typedef struct porter_tokenizer { sqlite3_tokenizer base; /* Base class */ } porter_tokenizer; /* ** Class derived from sqlit3_tokenizer_cursor */ typedef struct porter_tokenizer_cursor { sqlite3_tokenizer_cursor base; const char *zInput; /* input we are tokenizing */ int nInput; /* size of the input */ int iOffset; /* current position in zInput */ int iToken; /* index of next token to be returned */ char *zToken; /* storage for current token */ int nAllocated; /* space allocated to zToken buffer */ } porter_tokenizer_cursor; /* Forward declaration */ static const sqlite3_tokenizer_module porterTokenizerModule; /* ** Create a new tokenizer instance. */ static int porterCreate( int argc, const char * const *argv, sqlite3_tokenizer **ppTokenizer ){ porter_tokenizer *t; t = (porter_tokenizer *) calloc(sizeof(*t), 1); if( t==NULL ) return SQLITE_NOMEM; *ppTokenizer = &t->base; return SQLITE_OK; } /* ** Destroy a tokenizer */ static int porterDestroy(sqlite3_tokenizer *pTokenizer){ free(pTokenizer); return SQLITE_OK; } /* ** Prepare to begin tokenizing a particular string. The input ** string to be tokenized is zInput[0..nInput-1]. A cursor ** used to incrementally tokenize this string is returned in ** *ppCursor. */ static int porterOpen( sqlite3_tokenizer *pTokenizer, /* The tokenizer */ const char *zInput, int nInput, /* String to be tokenized */ sqlite3_tokenizer_cursor **ppCursor /* OUT: Tokenization cursor */ ){ porter_tokenizer_cursor *c; c = (porter_tokenizer_cursor *) malloc(sizeof(*c)); if( c==NULL ) return SQLITE_NOMEM; c->zInput = zInput; if( zInput==0 ){ c->nInput = 0; }else if( nInput<0 ){ c->nInput = (int)strlen(zInput); }else{ c->nInput = nInput; } c->iOffset = 0; /* start tokenizing at the beginning */ c->iToken = 0; c->zToken = NULL; /* no space allocated, yet. */ c->nAllocated = 0; *ppCursor = &c->base; return SQLITE_OK; } /* ** Close a tokenization cursor previously opened by a call to ** porterOpen() above. */ static int porterClose(sqlite3_tokenizer_cursor *pCursor){ porter_tokenizer_cursor *c = (porter_tokenizer_cursor *) pCursor; free(c->zToken); free(c); return SQLITE_OK; } /* ** Vowel or consonant */ static const char cType[] = { 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 0, 1, 1, 1, 2, 1 }; /* ** isConsonant() and isVowel() determine if their first character in ** the string they point to is a consonant or a vowel, according ** to Porter ruls. ** ** A consonate is any letter other than 'a', 'e', 'i', 'o', or 'u'. ** 'Y' is a consonant unless it follows another consonant, ** in which case it is a vowel. ** ** In these routine, the letters are in reverse order. So the 'y' rule ** is that 'y' is a consonant unless it is followed by another ** consonent. */ static int isVowel(const char*); static int isConsonant(const char *z){ int j; char x = *z; if( x==0 ) return 0; assert( x>='a' && x<='z' ); j = cType[x-'a']; if( j<2 ) return j; return z[1]==0 || isVowel(z + 1); } static int isVowel(const char *z){ int j; char x = *z; if( x==0 ) return 0; assert( x>='a' && x<='z' ); j = cType[x-'a']; if( j<2 ) return 1-j; return isConsonant(z + 1); } /* ** Let any sequence of one or more vowels be represented by V and let ** C be sequence of one or more consonants. Then every word can be ** represented as: ** ** [C] (VC){m} [V] ** ** In prose: A word is an optional consonant followed by zero or ** vowel-consonant pairs followed by an optional vowel. "m" is the ** number of vowel consonant pairs. This routine computes the value ** of m for the first i bytes of a word. ** ** Return true if the m-value for z is 1 or more. In other words, ** return true if z contains at least one vowel that is followed ** by a consonant. ** ** In this routine z[] is in reverse order. So we are really looking ** for an instance of of a consonant followed by a vowel. */ static int m_gt_0(const char *z){ while( isVowel(z) ){ z++; } if( *z==0 ) return 0; while( isConsonant(z) ){ z++; } return *z!=0; } /* Like mgt0 above except we are looking for a value of m which is ** exactly 1 */ static int m_eq_1(const char *z){ while( isVowel(z) ){ z++; } if( *z==0 ) return 0; while( isConsonant(z) ){ z++; } if( *z==0 ) return 0; while( isVowel(z) ){ z++; } if( *z==0 ) return 1; while( isConsonant(z) ){ z++; } return *z==0; } /* Like mgt0 above except we are looking for a value of m>1 instead ** or m>0 */ static int m_gt_1(const char *z){ while( isVowel(z) ){ z++; } if( *z==0 ) return 0; while( isConsonant(z) ){ z++; } if( *z==0 ) return 0; while( isVowel(z) ){ z++; } if( *z==0 ) return 0; while( isConsonant(z) ){ z++; } return *z!=0; } /* ** Return TRUE if there is a vowel anywhere within z[0..n-1] */ static int hasVowel(const char *z){ while( isConsonant(z) ){ z++; } return *z!=0; } /* ** Return TRUE if the word ends in a double consonant. ** ** The text is reversed here. So we are really looking at ** the first two characters of z[]. */ static int doubleConsonant(const char *z){ return isConsonant(z) && z[0]==z[1] && isConsonant(z+1); } /* ** Return TRUE if the word ends with three letters which ** are consonant-vowel-consonent and where the final consonant ** is not 'w', 'x', or 'y'. ** ** The word is reversed here. So we are really checking the ** first three letters and the first one cannot be in [wxy]. */ static int star_oh(const char *z){ return z[0]!=0 && isConsonant(z) && z[0]!='w' && z[0]!='x' && z[0]!='y' && z[1]!=0 && isVowel(z+1) && z[2]!=0 && isConsonant(z+2); } /* ** If the word ends with zFrom and xCond() is true for the stem ** of the word that preceeds the zFrom ending, then change the ** ending to zTo. ** ** The input word *pz and zFrom are both in reverse order. zTo ** is in normal order. ** ** Return TRUE if zFrom matches. Return FALSE if zFrom does not ** match. Not that TRUE is returned even if xCond() fails and ** no substitution occurs. */ static int stem( char **pz, /* The word being stemmed (Reversed) */ const char *zFrom, /* If the ending matches this... (Reversed) */ const char *zTo, /* ... change the ending to this (not reversed) */ int (*xCond)(const char*) /* Condition that must be true */ ){ char *z = *pz; while( *zFrom && *zFrom==*z ){ z++; zFrom++; } if( *zFrom!=0 ) return 0; if( xCond && !xCond(z) ) return 1; while( *zTo ){ *(--z) = *(zTo++); } *pz = z; return 1; } /* ** This is the fallback stemmer used when the porter stemmer is ** inappropriate. The input word is copied into the output with ** US-ASCII case folding. If the input word is too long (more ** than 20 bytes if it contains no digits or more than 6 bytes if ** it contains digits) then word is truncated to 20 or 6 bytes ** by taking 10 or 3 bytes from the beginning and end. */ static void copy_stemmer(const char *zIn, int nIn, char *zOut, int *pnOut){ int i, mx, j; int hasDigit = 0; for(i=0; i<nIn; i++){ int c = zIn[i]; if( c>='A' && c<='Z' ){ zOut[i] = c - 'A' + 'a'; }else{ if( c>='0' && c<='9' ) hasDigit = 1; zOut[i] = c; } } mx = hasDigit ? 3 : 10; if( nIn>mx*2 ){ for(j=mx, i=nIn-mx; i<nIn; i++, j++){ zOut[j] = zOut[i]; } i = j; } zOut[i] = 0; *pnOut = i; } /* ** Stem the input word zIn[0..nIn-1]. Store the output in zOut. ** zOut is at least big enough to hold nIn bytes. Write the actual ** size of the output word (exclusive of the '\0' terminator) into *pnOut. ** ** Any upper-case characters in the US-ASCII character set ([A-Z]) ** are converted to lower case. Upper-case UTF characters are ** unchanged. ** ** Words that are longer than about 20 bytes are stemmed by retaining ** a few bytes from the beginning and the end of the word. If the ** word contains digits, 3 bytes are taken from the beginning and ** 3 bytes from the end. For long words without digits, 10 bytes ** are taken from each end. US-ASCII case folding still applies. ** ** If the input word contains not digits but does characters not ** in [a-zA-Z] then no stemming is attempted and this routine just ** copies the input into the input into the output with US-ASCII ** case folding. ** ** Stemming never increases the length of the word. So there is ** no chance of overflowing the zOut buffer. */ static void porter_stemmer(const char *zIn, int nIn, char *zOut, int *pnOut){ int i, j, c; char zReverse[28]; char *z, *z2; if( nIn<3 || nIn>=sizeof(zReverse)-7 ){ /* The word is too big or too small for the porter stemmer. ** Fallback to the copy stemmer */ copy_stemmer(zIn, nIn, zOut, pnOut); return; } for(i=0, j=sizeof(zReverse)-6; i<nIn; i++, j--){ c = zIn[i]; if( c>='A' && c<='Z' ){ zReverse[j] = c + 'a' - 'A'; }else if( c>='a' && c<='z' ){ zReverse[j] = c; }else{ /* The use of a character not in [a-zA-Z] means that we fallback ** to the copy stemmer */ copy_stemmer(zIn, nIn, zOut, pnOut); return; } } memset(&zReverse[sizeof(zReverse)-5], 0, 5); z = &zReverse[j+1]; /* Step 1a */ if( z[0]=='s' ){ if( !stem(&z, "sess", "ss", 0) && !stem(&z, "sei", "i", 0) && !stem(&z, "ss", "ss", 0) ){ z++; } } /* Step 1b */ z2 = z; if( stem(&z, "dee", "ee", m_gt_0) ){ /* Do nothing. The work was all in the test */ }else if( (stem(&z, "gni", "", hasVowel) || stem(&z, "de", "", hasVowel)) && z!=z2 ){ if( stem(&z, "ta", "ate", 0) || stem(&z, "lb", "ble", 0) || stem(&z, "zi", "ize", 0) ){ /* Do nothing. The work was all in the test */ }else if( doubleConsonant(z) && (*z!='l' && *z!='s' && *z!='z') ){ z++; }else if( m_eq_1(z) && star_oh(z) ){ *(--z) = 'e'; } } /* Step 1c */ if( z[0]=='y' && hasVowel(z+1) ){ z[0] = 'i'; } /* Step 2 */ switch( z[1] ){ case 'a': stem(&z, "lanoita", "ate", m_gt_0) || stem(&z, "lanoit", "tion", m_gt_0); break; case 'c': stem(&z, "icne", "ence", m_gt_0) || stem(&z, "icna", "ance", m_gt_0); break; case 'e': stem(&z, "rezi", "ize", m_gt_0); break; case 'g': stem(&z, "igol", "log", m_gt_0); break; case 'l': stem(&z, "ilb", "ble", m_gt_0) || stem(&z, "illa", "al", m_gt_0) || stem(&z, "iltne", "ent", m_gt_0) || stem(&z, "ile", "e", m_gt_0) || stem(&z, "ilsuo", "ous", m_gt_0); break; case 'o': stem(&z, "noitazi", "ize", m_gt_0) || stem(&z, "noita", "ate", m_gt_0) || stem(&z, "rota", "ate", m_gt_0); break; case 's': stem(&z, "msila", "al", m_gt_0) || stem(&z, "ssenevi", "ive", m_gt_0) || stem(&z, "ssenluf", "ful", m_gt_0) || stem(&z, "ssensuo", "ous", m_gt_0); break; case 't': stem(&z, "itila", "al", m_gt_0) || stem(&z, "itivi", "ive", m_gt_0) || stem(&z, "itilib", "ble", m_gt_0); break; } /* Step 3 */ switch( z[0] ){ case 'e': stem(&z, "etaci", "ic", m_gt_0) || stem(&z, "evita", "", m_gt_0) || stem(&z, "ezila", "al", m_gt_0); break; case 'i': stem(&z, "itici", "ic", m_gt_0); break; case 'l': stem(&z, "laci", "ic", m_gt_0) || stem(&z, "luf", "", m_gt_0); break; case 's': stem(&z, "ssen", "", m_gt_0); break; } /* Step 4 */ switch( z[1] ){ case 'a': if( z[0]=='l' && m_gt_1(z+2) ){ z += 2; } break; case 'c': if( z[0]=='e' && z[2]=='n' && (z[3]=='a' || z[3]=='e') && m_gt_1(z+4) ){ z += 4; } break; case 'e': if( z[0]=='r' && m_gt_1(z+2) ){ z += 2; } break; case 'i': if( z[0]=='c' && m_gt_1(z+2) ){ z += 2; } break; case 'l': if( z[0]=='e' && z[2]=='b' && (z[3]=='a' || z[3]=='i') && m_gt_1(z+4) ){ z += 4; } break; case 'n': if( z[0]=='t' ){ if( z[2]=='a' ){ if( m_gt_1(z+3) ){ z += 3; } }else if( z[2]=='e' ){ stem(&z, "tneme", "", m_gt_1) || stem(&z, "tnem", "", m_gt_1) || stem(&z, "tne", "", m_gt_1); } } break; case 'o': if( z[0]=='u' ){ if( m_gt_1(z+2) ){ z += 2; } }else if( z[3]=='s' || z[3]=='t' ){ stem(&z, "noi", "", m_gt_1); } break; case 's': if( z[0]=='m' && z[2]=='i' && m_gt_1(z+3) ){ z += 3; } break; case 't': stem(&z, "eta", "", m_gt_1) || stem(&z, "iti", "", m_gt_1); break; case 'u': if( z[0]=='s' && z[2]=='o' && m_gt_1(z+3) ){ z += 3; } break; case 'v': case 'z': if( z[0]=='e' && z[2]=='i' && m_gt_1(z+3) ){ z += 3; } break; } /* Step 5a */ if( z[0]=='e' ){ if( m_gt_1(z+1) ){ z++; }else if( m_eq_1(z+1) && !star_oh(z+1) ){ z++; } } /* Step 5b */ if( m_gt_1(z) && z[0]=='l' && z[1]=='l' ){ z++; } /* z[] is now the stemmed word in reverse order. Flip it back ** around into forward order and return. */ *pnOut = i = strlen(z); zOut[i] = 0; while( *z ){ zOut[--i] = *(z++); } } /* ** Characters that can be part of a token. We assume any character ** whose value is greater than 0x80 (any UTF character) can be ** part of a token. In other words, delimiters all must have ** values of 0x7f or lower. */ static const char porterIdChar[] = { /* x0 x1 x2 x3 x4 x5 x6 x7 x8 x9 xA xB xC xD xE xF */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, /* 3x */ 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 4x */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, /* 5x */ 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 6x */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, /* 7x */ }; #define isDelim(C) (((ch=C)&0x80)==0 && (ch<0x30 || !porterIdChar[ch-0x30])) /* ** Extract the next token from a tokenization cursor. The cursor must ** have been opened by a prior call to porterOpen(). */ static int porterNext( sqlite3_tokenizer_cursor *pCursor, /* Cursor returned by porterOpen */ const char **pzToken, /* OUT: *pzToken is the token text */ int *pnBytes, /* OUT: Number of bytes in token */ int *piStartOffset, /* OUT: Starting offset of token */ int *piEndOffset, /* OUT: Ending offset of token */ int *piPosition /* OUT: Position integer of token */ ){ porter_tokenizer_cursor *c = (porter_tokenizer_cursor *) pCursor; const char *z = c->zInput; while( c->iOffset<c->nInput ){ int iStartOffset, ch; /* Scan past delimiter characters */ while( c->iOffset<c->nInput && isDelim(z[c->iOffset]) ){ c->iOffset++; } /* Count non-delimiter characters. */ iStartOffset = c->iOffset; while( c->iOffset<c->nInput && !isDelim(z[c->iOffset]) ){ c->iOffset++; } if( c->iOffset>iStartOffset ){ int n = c->iOffset-iStartOffset; if( n>c->nAllocated ){ c->nAllocated = n+20; c->zToken = realloc(c->zToken, c->nAllocated); if( c->zToken==NULL ) return SQLITE_NOMEM; } porter_stemmer(&z[iStartOffset], n, c->zToken, pnBytes); *pzToken = c->zToken; *piStartOffset = iStartOffset; *piEndOffset = c->iOffset; *piPosition = c->iToken++; return SQLITE_OK; } } return SQLITE_DONE; } /* ** The set of routines that implement the porter-stemmer tokenizer */ static const sqlite3_tokenizer_module porterTokenizerModule = { 0, porterCreate, porterDestroy, porterOpen, porterClose, porterNext, }; /* ** Allocate a new porter tokenizer. Return a pointer to the new ** tokenizer in *ppModule */ void sqlite3Fts3PorterTokenizerModule( sqlite3_tokenizer_module const**ppModule ){ *ppModule = &porterTokenizerModule; } #endif /* !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS3) */ |
Added ext/fts3/fts3_tokenizer.c.
> > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 | /* ** 2007 June 22 ** ** The author disclaims copyright to this source code. In place of ** a legal notice, here is a blessing: ** ** May you do good and not evil. ** May you find forgiveness for yourself and forgive others. ** May you share freely, never taking more than you give. ** ****************************************************************************** ** ** This is part of an SQLite module implementing full-text search. ** This particular file implements the generic tokenizer interface. */ /* ** The code in this file is only compiled if: ** ** * The FTS3 module is being built as an extension ** (in which case SQLITE_CORE is not defined), or ** ** * The FTS3 module is being built into the core of ** SQLite (in which case SQLITE_ENABLE_FTS3 is defined). */ #if !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS3) #include "sqlite3.h" #include "sqlite3ext.h" SQLITE_EXTENSION_INIT1 #include "fts3_hash.h" #include "fts3_tokenizer.h" #include <assert.h> /* ** Implementation of the SQL scalar function for accessing the underlying ** hash table. This function may be called as follows: ** ** SELECT <function-name>(<key-name>); ** SELECT <function-name>(<key-name>, <pointer>); ** ** where <function-name> is the name passed as the second argument ** to the sqlite3Fts3InitHashTable() function (e.g. 'fts3_tokenizer'). ** ** If the <pointer> argument is specified, it must be a blob value ** containing a pointer to be stored as the hash data corresponding ** to the string <key-name>. If <pointer> is not specified, then ** the string <key-name> must already exist in the has table. Otherwise, ** an error is returned. ** ** Whether or not the <pointer> argument is specified, the value returned ** is a blob containing the pointer stored as the hash data corresponding ** to string <key-name> (after the hash-table is updated, if applicable). */ static void scalarFunc( sqlite3_context *context, int argc, sqlite3_value **argv ){ fts3Hash *pHash; void *pPtr = 0; const unsigned char *zName; int nName; assert( argc==1 || argc==2 ); pHash = (fts3Hash *)sqlite3_user_data(context); zName = sqlite3_value_text(argv[0]); nName = sqlite3_value_bytes(argv[0])+1; if( argc==2 ){ void *pOld; int n = sqlite3_value_bytes(argv[1]); if( n!=sizeof(pPtr) ){ sqlite3_result_error(context, "argument type mismatch", -1); return; } pPtr = *(void **)sqlite3_value_blob(argv[1]); pOld = sqlite3Fts3HashInsert(pHash, (void *)zName, nName, pPtr); if( pOld==pPtr ){ sqlite3_result_error(context, "out of memory", -1); return; } }else{ pPtr = sqlite3Fts3HashFind(pHash, zName, nName); if( !pPtr ){ char *zErr = sqlite3_mprintf("unknown tokenizer: %s", zName); sqlite3_result_error(context, zErr, -1); sqlite3_free(zErr); return; } } sqlite3_result_blob(context, (void *)&pPtr, sizeof(pPtr), SQLITE_TRANSIENT); } #ifdef SQLITE_TEST #include <tcl.h> #include <string.h> /* ** Implementation of a special SQL scalar function for testing tokenizers ** designed to be used in concert with the Tcl testing framework. This ** function must be called with two arguments: ** ** SELECT <function-name>(<key-name>, <input-string>); ** SELECT <function-name>(<key-name>, <pointer>); ** ** where <function-name> is the name passed as the second argument ** to the sqlite3Fts3InitHashTable() function (e.g. 'fts3_tokenizer') ** concatenated with the string '_test' (e.g. 'fts3_tokenizer_test'). ** ** The return value is a string that may be interpreted as a Tcl ** list. For each token in the <input-string>, three elements are ** added to the returned list. The first is the token position, the ** second is the token text (folded, stemmed, etc.) and the third is the ** substring of <input-string> associated with the token. For example, ** using the built-in "simple" tokenizer: ** ** SELECT fts_tokenizer_test('simple', 'I don't see how'); ** ** will return the string: ** ** "{0 i I 1 dont don't 2 see see 3 how how}" ** */ static void testFunc( sqlite3_context *context, int argc, sqlite3_value **argv ){ fts3Hash *pHash; sqlite3_tokenizer_module *p; sqlite3_tokenizer *pTokenizer = 0; sqlite3_tokenizer_cursor *pCsr = 0; const char *zErr = 0; const char *zName; int nName; const char *zInput; int nInput; const char *zArg = 0; const char *zToken; int nToken; int iStart; int iEnd; int iPos; Tcl_Obj *pRet; assert( argc==2 || argc==3 ); nName = sqlite3_value_bytes(argv[0]); zName = (const char *)sqlite3_value_text(argv[0]); nInput = sqlite3_value_bytes(argv[argc-1]); zInput = (const char *)sqlite3_value_text(argv[argc-1]); if( argc==3 ){ zArg = (const char *)sqlite3_value_text(argv[1]); } pHash = (fts3Hash *)sqlite3_user_data(context); p = (sqlite3_tokenizer_module *)sqlite3Fts3HashFind(pHash, zName, nName+1); if( !p ){ char *zErr = sqlite3_mprintf("unknown tokenizer: %s", zName); sqlite3_result_error(context, zErr, -1); sqlite3_free(zErr); return; } pRet = Tcl_NewObj(); Tcl_IncrRefCount(pRet); if( SQLITE_OK!=p->xCreate(zArg ? 1 : 0, &zArg, &pTokenizer) ){ zErr = "error in xCreate()"; goto finish; } pTokenizer->pModule = p; if( SQLITE_OK!=p->xOpen(pTokenizer, zInput, nInput, &pCsr) ){ zErr = "error in xOpen()"; goto finish; } pCsr->pTokenizer = pTokenizer; while( SQLITE_OK==p->xNext(pCsr, &zToken, &nToken, &iStart, &iEnd, &iPos) ){ Tcl_ListObjAppendElement(0, pRet, Tcl_NewIntObj(iPos)); Tcl_ListObjAppendElement(0, pRet, Tcl_NewStringObj(zToken, nToken)); zToken = &zInput[iStart]; nToken = iEnd-iStart; Tcl_ListObjAppendElement(0, pRet, Tcl_NewStringObj(zToken, nToken)); } if( SQLITE_OK!=p->xClose(pCsr) ){ zErr = "error in xClose()"; goto finish; } if( SQLITE_OK!=p->xDestroy(pTokenizer) ){ zErr = "error in xDestroy()"; goto finish; } finish: if( zErr ){ sqlite3_result_error(context, zErr, -1); }else{ sqlite3_result_text(context, Tcl_GetString(pRet), -1, SQLITE_TRANSIENT); } Tcl_DecrRefCount(pRet); } static int registerTokenizer( sqlite3 *db, char *zName, const sqlite3_tokenizer_module *p ){ int rc; sqlite3_stmt *pStmt; const char zSql[] = "SELECT fts3_tokenizer(?, ?)"; rc = sqlite3_prepare_v2(db, zSql, -1, &pStmt, 0); if( rc!=SQLITE_OK ){ return rc; } sqlite3_bind_text(pStmt, 1, zName, -1, SQLITE_STATIC); sqlite3_bind_blob(pStmt, 2, &p, sizeof(p), SQLITE_STATIC); sqlite3_step(pStmt); return sqlite3_finalize(pStmt); } static int queryTokenizer( sqlite3 *db, char *zName, const sqlite3_tokenizer_module **pp ){ int rc; sqlite3_stmt *pStmt; const char zSql[] = "SELECT fts3_tokenizer(?)"; *pp = 0; rc = sqlite3_prepare_v2(db, zSql, -1, &pStmt, 0); if( rc!=SQLITE_OK ){ return rc; } sqlite3_bind_text(pStmt, 1, zName, -1, SQLITE_STATIC); if( SQLITE_ROW==sqlite3_step(pStmt) ){ if( sqlite3_column_type(pStmt, 0)==SQLITE_BLOB ){ memcpy(pp, sqlite3_column_blob(pStmt, 0), sizeof(*pp)); } } return sqlite3_finalize(pStmt); } void sqlite3Fts3SimpleTokenizerModule(sqlite3_tokenizer_module const**ppModule); /* ** Implementation of the scalar function fts3_tokenizer_internal_test(). ** This function is used for testing only, it is not included in the ** build unless SQLITE_TEST is defined. ** ** The purpose of this is to test that the fts3_tokenizer() function ** can be used as designed by the C-code in the queryTokenizer and ** registerTokenizer() functions above. These two functions are repeated ** in the README.tokenizer file as an example, so it is important to ** test them. ** ** To run the tests, evaluate the fts3_tokenizer_internal_test() scalar ** function with no arguments. An assert() will fail if a problem is ** detected. i.e.: ** ** SELECT fts3_tokenizer_internal_test(); ** */ static void intTestFunc( sqlite3_context *context, int argc, sqlite3_value **argv ){ int rc; const sqlite3_tokenizer_module *p1; const sqlite3_tokenizer_module *p2; sqlite3 *db = (sqlite3 *)sqlite3_user_data(context); /* Test the query function */ sqlite3Fts3SimpleTokenizerModule(&p1); rc = queryTokenizer(db, "simple", &p2); assert( rc==SQLITE_OK ); assert( p1==p2 ); rc = queryTokenizer(db, "nosuchtokenizer", &p2); assert( rc==SQLITE_ERROR ); assert( p2==0 ); assert( 0==strcmp(sqlite3_errmsg(db), "unknown tokenizer: nosuchtokenizer") ); /* Test the storage function */ rc = registerTokenizer(db, "nosuchtokenizer", p1); assert( rc==SQLITE_OK ); rc = queryTokenizer(db, "nosuchtokenizer", &p2); assert( rc==SQLITE_OK ); assert( p2==p1 ); sqlite3_result_text(context, "ok", -1, SQLITE_STATIC); } #endif /* ** Set up SQL objects in database db used to access the contents of ** the hash table pointed to by argument pHash. The hash table must ** been initialised to use string keys, and to take a private copy ** of the key when a value is inserted. i.e. by a call similar to: ** ** sqlite3Fts3HashInit(pHash, FTS3_HASH_STRING, 1); ** ** This function adds a scalar function (see header comment above ** scalarFunc() in this file for details) and, if ENABLE_TABLE is ** defined at compilation time, a temporary virtual table (see header ** comment above struct HashTableVtab) to the database schema. Both ** provide read/write access to the contents of *pHash. ** ** The third argument to this function, zName, is used as the name ** of both the scalar and, if created, the virtual table. */ int sqlite3Fts3InitHashTable( sqlite3 *db, fts3Hash *pHash, const char *zName ){ int rc = SQLITE_OK; void *p = (void *)pHash; const int any = SQLITE_ANY; char *zTest = 0; char *zTest2 = 0; #ifdef SQLITE_TEST void *pdb = (void *)db; zTest = sqlite3_mprintf("%s_test", zName); zTest2 = sqlite3_mprintf("%s_internal_test", zName); if( !zTest || !zTest2 ){ rc = SQLITE_NOMEM; } #endif if( rc!=SQLITE_OK || (rc = sqlite3_create_function(db, zName, 1, any, p, scalarFunc, 0, 0)) || (rc = sqlite3_create_function(db, zName, 2, any, p, scalarFunc, 0, 0)) #ifdef SQLITE_TEST || (rc = sqlite3_create_function(db, zTest, 2, any, p, testFunc, 0, 0)) || (rc = sqlite3_create_function(db, zTest, 3, any, p, testFunc, 0, 0)) || (rc = sqlite3_create_function(db, zTest2, 0, any, pdb, intTestFunc, 0, 0)) #endif ); sqlite3_free(zTest); sqlite3_free(zTest2); return rc; } #endif /* !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS3) */ |
Added ext/fts3/fts3_tokenizer.h.
> > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 | /* ** 2006 July 10 ** ** The author disclaims copyright to this source code. ** ************************************************************************* ** Defines the interface to tokenizers used by fulltext-search. There ** are three basic components: ** ** sqlite3_tokenizer_module is a singleton defining the tokenizer ** interface functions. This is essentially the class structure for ** tokenizers. ** ** sqlite3_tokenizer is used to define a particular tokenizer, perhaps ** including customization information defined at creation time. ** ** sqlite3_tokenizer_cursor is generated by a tokenizer to generate ** tokens from a particular input. */ #ifndef _FTS3_TOKENIZER_H_ #define _FTS3_TOKENIZER_H_ /* TODO(shess) Only used for SQLITE_OK and SQLITE_DONE at this time. ** If tokenizers are to be allowed to call sqlite3_*() functions, then ** we will need a way to register the API consistently. */ #include "sqlite3.h" /* ** Structures used by the tokenizer interface. When a new tokenizer ** implementation is registered, the caller provides a pointer to ** an sqlite3_tokenizer_module containing pointers to the callback ** functions that make up an implementation. ** ** When an fts3 table is created, it passes any arguments passed to ** the tokenizer clause of the CREATE VIRTUAL TABLE statement to the ** sqlite3_tokenizer_module.xCreate() function of the requested tokenizer ** implementation. The xCreate() function in turn returns an ** sqlite3_tokenizer structure representing the specific tokenizer to ** be used for the fts3 table (customized by the tokenizer clause arguments). ** ** To tokenize an input buffer, the sqlite3_tokenizer_module.xOpen() ** method is called. It returns an sqlite3_tokenizer_cursor object ** that may be used to tokenize a specific input buffer based on ** the tokenization rules supplied by a specific sqlite3_tokenizer ** object. */ typedef struct sqlite3_tokenizer_module sqlite3_tokenizer_module; typedef struct sqlite3_tokenizer sqlite3_tokenizer; typedef struct sqlite3_tokenizer_cursor sqlite3_tokenizer_cursor; struct sqlite3_tokenizer_module { /* ** Structure version. Should always be set to 0. */ int iVersion; /* ** Create a new tokenizer. The values in the argv[] array are the ** arguments passed to the "tokenizer" clause of the CREATE VIRTUAL ** TABLE statement that created the fts3 table. For example, if ** the following SQL is executed: ** ** CREATE .. USING fts3( ... , tokenizer <tokenizer-name> arg1 arg2) ** ** then argc is set to 2, and the argv[] array contains pointers ** to the strings "arg1" and "arg2". ** ** This method should return either SQLITE_OK (0), or an SQLite error ** code. If SQLITE_OK is returned, then *ppTokenizer should be set ** to point at the newly created tokenizer structure. The generic ** sqlite3_tokenizer.pModule variable should not be initialised by ** this callback. The caller will do so. */ int (*xCreate)( int argc, /* Size of argv array */ const char *const*argv, /* Tokenizer argument strings */ sqlite3_tokenizer **ppTokenizer /* OUT: Created tokenizer */ ); /* ** Destroy an existing tokenizer. The fts3 module calls this method ** exactly once for each successful call to xCreate(). */ int (*xDestroy)(sqlite3_tokenizer *pTokenizer); /* ** Create a tokenizer cursor to tokenize an input buffer. The caller ** is responsible for ensuring that the input buffer remains valid ** until the cursor is closed (using the xClose() method). */ int (*xOpen)( sqlite3_tokenizer *pTokenizer, /* Tokenizer object */ const char *pInput, int nBytes, /* Input buffer */ sqlite3_tokenizer_cursor **ppCursor /* OUT: Created tokenizer cursor */ ); /* ** Destroy an existing tokenizer cursor. The fts3 module calls this ** method exactly once for each successful call to xOpen(). */ int (*xClose)(sqlite3_tokenizer_cursor *pCursor); /* ** Retrieve the next token from the tokenizer cursor pCursor. This ** method should either return SQLITE_OK and set the values of the ** "OUT" variables identified below, or SQLITE_DONE to indicate that ** the end of the buffer has been reached, or an SQLite error code. ** ** *ppToken should be set to point at a buffer containing the ** normalized version of the token (i.e. after any case-folding and/or ** stemming has been performed). *pnBytes should be set to the length ** of this buffer in bytes. The input text that generated the token is ** identified by the byte offsets returned in *piStartOffset and ** *piEndOffset. ** ** The buffer *ppToken is set to point at is managed by the tokenizer ** implementation. It is only required to be valid until the next call ** to xNext() or xClose(). */ /* TODO(shess) current implementation requires pInput to be ** nul-terminated. This should either be fixed, or pInput/nBytes ** should be converted to zInput. */ int (*xNext)( sqlite3_tokenizer_cursor *pCursor, /* Tokenizer cursor */ const char **ppToken, int *pnBytes, /* OUT: Normalized text for token */ int *piStartOffset, /* OUT: Byte offset of token in input buffer */ int *piEndOffset, /* OUT: Byte offset of end of token in input buffer */ int *piPosition /* OUT: Number of tokens returned before this one */ ); }; struct sqlite3_tokenizer { const sqlite3_tokenizer_module *pModule; /* The module for this tokenizer */ /* Tokenizer implementations will typically add additional fields */ }; struct sqlite3_tokenizer_cursor { sqlite3_tokenizer *pTokenizer; /* Tokenizer for this cursor. */ /* Tokenizer implementations will typically add additional fields */ }; #endif /* _FTS3_TOKENIZER_H_ */ |
Added ext/fts3/fts3_tokenizer1.c.
> > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 | /* ** 2006 Oct 10 ** ** The author disclaims copyright to this source code. In place of ** a legal notice, here is a blessing: ** ** May you do good and not evil. ** May you find forgiveness for yourself and forgive others. ** May you share freely, never taking more than you give. ** ****************************************************************************** ** ** Implementation of the "simple" full-text-search tokenizer. */ /* ** The code in this file is only compiled if: ** ** * The FTS3 module is being built as an extension ** (in which case SQLITE_CORE is not defined), or ** ** * The FTS3 module is being built into the core of ** SQLite (in which case SQLITE_ENABLE_FTS3 is defined). */ #if !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS3) #include <assert.h> #include <stdlib.h> #include <stdio.h> #include <string.h> #include <ctype.h> #include "fts3_tokenizer.h" typedef struct simple_tokenizer { sqlite3_tokenizer base; char delim[128]; /* flag ASCII delimiters */ } simple_tokenizer; typedef struct simple_tokenizer_cursor { sqlite3_tokenizer_cursor base; const char *pInput; /* input we are tokenizing */ int nBytes; /* size of the input */ int iOffset; /* current position in pInput */ int iToken; /* index of next token to be returned */ char *pToken; /* storage for current token */ int nTokenAllocated; /* space allocated to zToken buffer */ } simple_tokenizer_cursor; /* Forward declaration */ static const sqlite3_tokenizer_module simpleTokenizerModule; static int simpleDelim(simple_tokenizer *t, unsigned char c){ return c<0x80 && t->delim[c]; } /* ** Create a new tokenizer instance. */ static int simpleCreate( int argc, const char * const *argv, sqlite3_tokenizer **ppTokenizer ){ simple_tokenizer *t; t = (simple_tokenizer *) calloc(sizeof(*t), 1); if( t==NULL ) return SQLITE_NOMEM; /* TODO(shess) Delimiters need to remain the same from run to run, ** else we need to reindex. One solution would be a meta-table to ** track such information in the database, then we'd only want this ** information on the initial create. */ if( argc>1 ){ int i, n = strlen(argv[1]); for(i=0; i<n; i++){ unsigned char ch = argv[1][i]; /* We explicitly don't support UTF-8 delimiters for now. */ if( ch>=0x80 ){ free(t); return SQLITE_ERROR; } t->delim[ch] = 1; } } else { /* Mark non-alphanumeric ASCII characters as delimiters */ int i; for(i=1; i<0x80; i++){ t->delim[i] = !isalnum(i); } } *ppTokenizer = &t->base; return SQLITE_OK; } /* ** Destroy a tokenizer */ static int simpleDestroy(sqlite3_tokenizer *pTokenizer){ free(pTokenizer); return SQLITE_OK; } /* ** Prepare to begin tokenizing a particular string. The input ** string to be tokenized is pInput[0..nBytes-1]. A cursor ** used to incrementally tokenize this string is returned in ** *ppCursor. */ static int simpleOpen( sqlite3_tokenizer *pTokenizer, /* The tokenizer */ const char *pInput, int nBytes, /* String to be tokenized */ sqlite3_tokenizer_cursor **ppCursor /* OUT: Tokenization cursor */ ){ simple_tokenizer_cursor *c; c = (simple_tokenizer_cursor *) malloc(sizeof(*c)); if( c==NULL ) return SQLITE_NOMEM; c->pInput = pInput; if( pInput==0 ){ c->nBytes = 0; }else if( nBytes<0 ){ c->nBytes = (int)strlen(pInput); }else{ c->nBytes = nBytes; } c->iOffset = 0; /* start tokenizing at the beginning */ c->iToken = 0; c->pToken = NULL; /* no space allocated, yet. */ c->nTokenAllocated = 0; *ppCursor = &c->base; return SQLITE_OK; } /* ** Close a tokenization cursor previously opened by a call to ** simpleOpen() above. */ static int simpleClose(sqlite3_tokenizer_cursor *pCursor){ simple_tokenizer_cursor *c = (simple_tokenizer_cursor *) pCursor; free(c->pToken); free(c); return SQLITE_OK; } /* ** Extract the next token from a tokenization cursor. The cursor must ** have been opened by a prior call to simpleOpen(). */ static int simpleNext( sqlite3_tokenizer_cursor *pCursor, /* Cursor returned by simpleOpen */ const char **ppToken, /* OUT: *ppToken is the token text */ int *pnBytes, /* OUT: Number of bytes in token */ int *piStartOffset, /* OUT: Starting offset of token */ int *piEndOffset, /* OUT: Ending offset of token */ int *piPosition /* OUT: Position integer of token */ ){ simple_tokenizer_cursor *c = (simple_tokenizer_cursor *) pCursor; simple_tokenizer *t = (simple_tokenizer *) pCursor->pTokenizer; unsigned char *p = (unsigned char *)c->pInput; while( c->iOffset<c->nBytes ){ int iStartOffset; /* Scan past delimiter characters */ while( c->iOffset<c->nBytes && simpleDelim(t, p[c->iOffset]) ){ c->iOffset++; } /* Count non-delimiter characters. */ iStartOffset = c->iOffset; while( c->iOffset<c->nBytes && !simpleDelim(t, p[c->iOffset]) ){ c->iOffset++; } if( c->iOffset>iStartOffset ){ int i, n = c->iOffset-iStartOffset; if( n>c->nTokenAllocated ){ c->nTokenAllocated = n+20; c->pToken = realloc(c->pToken, c->nTokenAllocated); if( c->pToken==NULL ) return SQLITE_NOMEM; } for(i=0; i<n; i++){ /* TODO(shess) This needs expansion to handle UTF-8 ** case-insensitivity. */ unsigned char ch = p[iStartOffset+i]; c->pToken[i] = ch<0x80 ? tolower(ch) : ch; } *ppToken = c->pToken; *pnBytes = n; *piStartOffset = iStartOffset; *piEndOffset = c->iOffset; *piPosition = c->iToken++; return SQLITE_OK; } } return SQLITE_DONE; } /* ** The set of routines that implement the simple tokenizer */ static const sqlite3_tokenizer_module simpleTokenizerModule = { 0, simpleCreate, simpleDestroy, simpleOpen, simpleClose, simpleNext, }; /* ** Allocate a new simple tokenizer. Return a pointer to the new ** tokenizer in *ppModule */ void sqlite3Fts3SimpleTokenizerModule( sqlite3_tokenizer_module const**ppModule ){ *ppModule = &simpleTokenizerModule; } #endif /* !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS3) */ |
Added ext/fts3/mkfts3amal.tcl.
> > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 | #!/usr/bin/tclsh # # This script builds a single C code file holding all of FTS3 code. # The name of the output file is fts3amal.c. To build this file, # first do: # # make target_source # # The make target above moves all of the source code files into # a subdirectory named "tsrc". (This script expects to find the files # there and will not work if they are not found.) # # After the "tsrc" directory has been created and populated, run # this script: # # tclsh mkfts3amal.tcl # # The amalgamated FTS3 code will be written into fts3amal.c # # Open the output file and write a header comment at the beginning # of the file. # set out [open fts3amal.c w] set today [clock format [clock seconds] -format "%Y-%m-%d %H:%M:%S UTC" -gmt 1] puts $out [subst \ {/****************************************************************************** ** This file is an amalgamation of separate C source files from the SQLite ** Full Text Search extension 2 (fts3). By combining all the individual C ** code files into this single large file, the entire code can be compiled ** as a one translation unit. This allows many compilers to do optimizations ** that would not be possible if the files were compiled separately. It also ** makes the code easier to import into other projects. ** ** This amalgamation was generated on $today. */}] # These are the header files used by FTS3. The first time any of these # files are seen in a #include statement in the C code, include the complete # text of the file in-line. The file only needs to be included once. # foreach hdr { fts3.h fts3_hash.h fts3_tokenizer.h sqlite3.h sqlite3ext.h } { set available_hdr($hdr) 1 } # 78 stars used for comment formatting. set s78 \ {*****************************************************************************} # Insert a comment into the code # proc section_comment {text} { global out s78 set n [string length $text] set nstar [expr {60 - $n}] set stars [string range $s78 0 $nstar] puts $out "/************** $text $stars/" } # Read the source file named $filename and write it into the # sqlite3.c output file. If any #include statements are seen, # process them approprately. # proc copy_file {filename} { global seen_hdr available_hdr out set tail [file tail $filename] section_comment "Begin file $tail" set in [open $filename r] while {![eof $in]} { set line [gets $in] if {[regexp {^#\s*include\s+["<]([^">]+)[">]} $line all hdr]} { if {[info exists available_hdr($hdr)]} { if {$available_hdr($hdr)} { section_comment "Include $hdr in the middle of $tail" copy_file tsrc/$hdr section_comment "Continuing where we left off in $tail" } } elseif {![info exists seen_hdr($hdr)]} { set seen_hdr($hdr) 1 puts $out $line } } elseif {[regexp {^#ifdef __cplusplus} $line]} { puts $out "#if 0" } elseif {[regexp {^#line} $line]} { # Skip #line directives. } else { puts $out $line } } close $in section_comment "End of $tail" } # Process the source files. Process files containing commonly # used subroutines first in order to help the compiler find # inlining opportunities. # foreach file { fts3.c fts3_hash.c fts3_porter.c fts3_tokenizer.c fts3_tokenizer1.c fts3_icu.c } { copy_file tsrc/$file } close $out |