Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
Overview
Comment: | Fix a requirement number conflict in fileformat.in. Enhanced and expanded vtab.in. |
---|---|
Downloads: | Tarball | ZIP archive |
Timelines: | family | ancestors | descendants | both | trunk |
Files: | files | file ages | folders |
SHA1: |
9acad193dd73a5cebdbad433be7ed0fa |
User & Date: | drh 2009-04-13 18:04:52.000 |
Context
2009-04-14
| ||
11:45 | Additional virtual table documentation improvements. Fix the "when-to-use" document to omit discussion of the obsolete bitmap size limits. (check-in: 9ef2178fee user: drh tags: trunk) | |
2009-04-13
| ||
18:04 | Fix a requirement number conflict in fileformat.in. Enhanced and expanded vtab.in. (check-in: 9acad193dd user: drh tags: trunk) | |
15:07 | Merge [ebd923dab6] and [491737c7cf]. (check-in: 8f18472bac user: dan tags: trunk) | |
Changes
Changes to pages/fileformat.in.
︙ | ︙ | |||
2239 2240 2241 2242 2243 2244 2245 | [Figure journal_format.gif figure_journal_format "Journal File Format"] <p> The following requirements define a well-formed journal section. This concept is used in section <cite>reading_from_files</cite>. | | | | | | | 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 | [Figure journal_format.gif figure_journal_format "Journal File Format"] <p> The following requirements define a well-formed journal section. This concept is used in section <cite>reading_from_files</cite>. [fileformat_import_requirement2 H32210] [fileformat_import_requirement2 H32220] [fileformat_import_requirement2 H32230] [fileformat_import_requirement2 H32240] <p> Note that a journal section that is not strictly speaking a well-formed journal section often contains important data. For example, many journal files created by SQLite that consist of a single journal section and no master journal pointer contain a journal section that is not well-formed according to requirement H32240. See section <cite>reading_from_files</cite> for details on when well-formedness is an important property of journal sections and when it is not. [h4 "Journal Header Format" journal_header_format] <p> A journal header is sector-size bytes in size, where sector-size is the |
︙ | ︙ | |||
2319 2320 2321 2322 2323 2324 2325 | is defined as a blob of 28 bytes for which the journal magic field is set correctly and for which both the page size and sector size fields are set to power of two values greater than 512. Because there are no restrictions on the values that may be stored in the record count, checksum initializer or database page count fields, they do not enter into the definition of a well-formed journal header. | | | | | | 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 | is defined as a blob of 28 bytes for which the journal magic field is set correctly and for which both the page size and sector size fields are set to power of two values greater than 512. Because there are no restrictions on the values that may be stored in the record count, checksum initializer or database page count fields, they do not enter into the definition of a well-formed journal header. [fileformat_import_requirement2 H32090] [fileformat_import_requirement2 H32180] [fileformat_import_requirement2 H32190] [fileformat_import_requirement2 H32200] [h4 "Journal Record Format" journal_record_format] <p> Each <i>journal record</i> contains the data for a single database page, a page number identifying the page, and a checksum value used to help detect journal file corruption. |
︙ | ︙ | |||
2400 2401 2402 2403 2404 2405 2406 | <p> The set of <i>journal records</i> that follow a <i>journal header</i> in a journal file are packed tightly together. There are no alignment requirements for <i>journal records</i>. | | | | | 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 | <p> The set of <i>journal records</i> that follow a <i>journal header</i> in a journal file are packed tightly together. There are no alignment requirements for <i>journal records</i>. [fileformat_import_requirement2 H32100] [fileformat_import_requirement2 H32110] [fileformat_import_requirement2 H32120] [h4 "Master Journal Pointer" master_journal_ptr] <p> If present, a master journal pointer occurs at the end of a journal file. There may or may not be unused space between the end of the final journal |
︙ | ︙ | |||
2459 2460 2461 2462 2463 2464 2465 | Finally, the <b>journal magic</b> field always contains a well-known 8-byte string value; the same value stored in the first 8 bytes of a <i>journal header</i>. The well-known sequence of bytes is: <pre>0xd9 0xd5 0x05 0xf9 0x20 0xa1 0x63 0xd7</pre> </table> | | | | | | 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 | Finally, the <b>journal magic</b> field always contains a well-known 8-byte string value; the same value stored in the first 8 bytes of a <i>journal header</i>. The well-known sequence of bytes is: <pre>0xd9 0xd5 0x05 0xf9 0x20 0xa1 0x63 0xd7</pre> </table> [fileformat_import_requirement2 H32140] [fileformat_import_requirement2 H32150] [fileformat_import_requirement2 H32160] [fileformat_import_requirement2 H32170] [h3 "Master-Journal File Details" masterjournal_file_format] <p> A <i>master-journal file</i> contains the full paths to two or more <i>journal files</i>, each encoded using UTF-8 encoding and terminated by a single nul character (byte value 0x00). There is no padding |
︙ | ︙ | |||
2590 2591 2592 2593 2594 2595 2596 | [Figure filesystem1.gif figure_filesystem1 "Two ways to store the same database image"] <p> These requirements describe the way a database reader must determine whether or not there is a valid journal file within the file-system. | | | | | | | | | | | | | | | 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 | [Figure filesystem1.gif figure_filesystem1 "Two ways to store the same database image"] <p> These requirements describe the way a database reader must determine whether or not there is a valid journal file within the file-system. [fileformat_import_requirement2 H32000] [fileformat_import_requirement2 H32010] [fileformat_import_requirement2 H32020] <p> If there is a valid journal file within the file-system, the following requirements govern how a reader should determine the set of valid <i>journal records</i> that it contains. [fileformat_import_requirement2 H32250] [fileformat_import_requirement2 H32260] [fileformat_import_requirement2 H32270] [fileformat_import_requirement2 H32280] <p> The following requirements dictate the way in which database <i>page-size</i> and the number of pages in the database image should be determined by the reader. [fileformat_import_requirement2 H32030] [fileformat_import_requirement2 H32040] [fileformat_import_requirement2 H32050] [fileformat_import_requirement2 H32060] <p> The following requirements dictate the way in which the data for each page of the database image can be located within the file-system by a database reader. [fileformat_import_requirement2 H32070] [fileformat_import_requirement2 H32080] [h2 "Writing to an SQLite Database" writing_to_files] <p> When an SQLite user commits a transaction that modifies the contents of the database, the database representation on disk must be modified to reflect the new contents of the database image. SQLite is required |
︙ | ︙ |
Changes to pages/vtab.in.
1 2 3 4 5 6 7 8 | <title>The Virtual Table Mechanism Of SQLite</title> <h1 align="center">The Virtual Table Mechanism Of SQLite</h1> <tcl>hd_keywords {virtual table} {virtual tables}</tcl> <h2>1.0 Introduction</h2> <p>A virtual table is an object that is registered with an open SQLite | | | > | | > > | | > | | | | | | | | > > > | | | | | > | | | 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 | <title>The Virtual Table Mechanism Of SQLite</title> <h1 align="center">The Virtual Table Mechanism Of SQLite</h1> <tcl>hd_keywords {virtual table} {virtual tables}</tcl> <h2>1.0 Introduction</h2> <p>A virtual table is an object that is registered with an open SQLite [database connection]. From the perspective of an SQL statement, the virtual table object looks like any other table or view. But behind the scenes, queries from and updates to a virtual table invoke callback methods on the virtual table object instead of reading and writing to the database file. <p>The virtual table mechanism allows an application to publish interfaces that are accessible from SQL statements as if they were tables. SQL statements can in general do anything to a virtual table that they can do to a real table, with the following exceptions: <p> <ul> <li> One cannot create a trigger on a virtual table. <li> One cannot create additional indices on a virtual table. (Virtual tables can have indices but that must be built into the virtual table implementation. Indices cannot be added separately using [CREATE INDEX] statements.) <li> One cannot run [ALTER TABLE | ALTER TABLE ... ADD COLUMN] commands against a virtual table. <li> Virtual tables cannot be used in a database that makes use of the [shared cache] feature. </ul> <p>Particular virtual table implementations might impose additional constraints. For example, some virtual implementations might provide read-only tables. Or some virtual table implementations might allow [INSERT] or [DELETE] but not [UPDATE]. Or some virtual table implementations might limit the kinds of UPDATEs that can be made. <p>A virtual table might represent an in-memory data structures. Or it might represent a view of data on disk that is not in the SQLite format. Or the application might compute the content of the virtual table on demand. <p>Here are some postulated uses for virtual tables: <ul> <li> A full-text search interface <li> Spatial indices using R-Trees <li> Read and/or write the content of a comma-separated value (CSV) file <li> Access to the filesystem of the host computer <li> Enabling SQL manipulation of data in statistics packages like R </ul> <h3>1.1 Usage</h3> <p>A virtual table is created using using a [CREATE VIRTUAL TABLE] statement. This statement creates a table with a particular name and associates the table with a "module". <blockquote><pre> CREATE VIRTUAL TABLE tablename USING modulename; </pre></blockquote> <p>One can also provide comma-separated arguments to the module following the module name: <blockquote><pre> CREATE VIRTUAL TABLE tablename USING modulename(arg1, arg2, ...); </pre></blockquote> <p>The format of the arguments to the module is very general. Each argument can consist of keywords, string literals, identifiers, numbers, and punctuation. The arguments are passed as written (as text) into the [xCreate | constructor method] of the virtual table implementation when the virtual table is created and that constructor is responsible for parsing and interpreting the arguments. The argument syntax is sufficiently general that a virtual table implementation can, if it wants to, interpret its arguments as column definitions in an ordinary [CREATE TABLE] statement. The implementation could also impose some other interpretation on the arguments. <p>Once a virtual table has been created, it can be used like any other table with the exceptions noted above and imposed by specific virtual table implementations. A virtual table is destroyed using the ordinary [DROP TABLE] syntax. <h2>2.2 Implementation</h2> <p>Several new C-level objects are used by the virtual table implementation: <blockquote><pre> typedef struct sqlite3_vtab sqlite3_vtab; typedef struct sqlite3_index_info sqlite3_index_info; typedef struct sqlite3_vtab_cursor sqlite3_vtab_cursor; typedef struct sqlite3_module sqlite3_module; </pre></blockquote> <p>The [sqlite3_module] structure defines a module object used to implement a virtual table. Think of a module as a class from which one can construct multiple virtual tables having similar properties. For example, one might have a module that provides read-only access to comma-separated-value (CSV) files on disk. That one module can then be used to create several virtual tables where each virtual table refers to a different CSV file. <p>The module structure contains methods that are invoked by SQLite to perform various actions on the virtual table such as creating new instances of a virtual table or destroying old ones, reading and writing data, searching for and deleting, updating, or inserting rows. The module structure is explained in more detail below. <p>Each virtual table instance is represented by an [sqlite3_vtab] structure. The sqlite3_vtab structure looks like this: <blockquote><pre> struct sqlite3_vtab { const sqlite3_module *pModule; int nRef; char *zErrMsg; }; </pre></blockquote> <p>Virtual table implementations will normally subclass this structure to add additional private and implementation-specific fields. The nRef field is used internally by the SQLite core and should not be altered by the virtual table implementation. The virtual table implementation may pass error message text to the core by putting an error message string in zErrMsg. Space to hold this error message string must be obtained from an SQLite memory allocation function such as [sqlite3_mprintf()] or [sqlite3_malloc()]. Prior to assigning a new value to zErrMsg, the virtual table implementation must free any preexisting content of zErrMsg using [sqlite3_free()]. Failure to do this will result in a memory leak. The SQLite core will free and zero the content of zErrMsg when it delivers the error message text to the client application or when it destroys the virtual table. The virtual table implementation only needs to worry about freeing the zErrMsg content when it overwrites the content with a new, different error message. <p>The [sqlite3_vtab_cursor] structure represents a pointer to a specific row of a virtual table. This is what an sqlite3_vtab_cursor looks like: <blockquote><pre> struct sqlite3_vtab_cursor { sqlite3_vtab *pVtab; }; </pre></blockquote> <p>Once again, practical implementations will likely subclass this structure to add additional private fields. <p>The [sqlite3_index_info] structure is used to pass information into and out of the xBestIndex method of the module that implements a virtual table. <p>Before a [CREATE VIRTUAL TABLE] statement can be run, the module specified in that statement must be registered with the database connection. This is accomplished using either of the [sqlite3_create_module()] or [sqlite3_create_module_v2()] interfaces: <blockquote><pre> int sqlite3_create_module( sqlite3 *db, /* SQLite connection to register module with */ const char *zName, /* Name of the module */ const sqlite3_module *, /* Methods for the module */ void * /* Client data for xCreate/xConnect */ ); int sqlite3_create_module_v2( sqlite3 *db, /* SQLite connection to register module with */ const char *zName, /* Name of the module */ const sqlite3_module *, /* Methods for the module */ void *, /* Client data for xCreate/xConnect */ void(*xDestroy)(void*) /* Client data destructor function */ ); </pre></blockquote> <p>The [sqlite3_create_module()] and [sqlite3_create_module_v2()] routines associates a module name with an [sqlite3_module] structure and a separate client data that is specific to each module. The only difference between the two create_module methods is that the _v2 method includes an extra parameter that specifies a destructor for client data pointer. The module structure is what defines the behavior of a virtual table. The module structure looks like this: <blockquote><pre> struct sqlite3_module { |
︙ | ︙ | |||
210 211 212 213 214 215 216 | void **ppArg); int (*Rename)(sqlite3_vtab *pVtab, const char *zNew); }; </pre></blockquote> <p>The module structure defines all of the methods for each virtual table object. The module structure also contains the iVersion field which | | | | | | | | > > | | > > > | > > | > | | | | < | | > > > > | > | | > | | > > > > > | > > > > > > | | 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 | void **ppArg); int (*Rename)(sqlite3_vtab *pVtab, const char *zNew); }; </pre></blockquote> <p>The module structure defines all of the methods for each virtual table object. The module structure also contains the iVersion field which defines the particular edition of the module table structure. Currently, iVersion is always 1, but in future releases of SQLite the module structure definition might be extended with additional methods and in that case the iVersion value will be increased. <p>The rest of the module structure consists of methods used to implement various features of the virtual table. Details on what each of these methods do are provided in the sequel. <h3>1.3 Virtual Tables And Shared Cache</h3> <p>The virtual table mechanism assumes that each [database connection] keeps its own copy of the database schema. Hence, the virtual table mechanism cannot be used in a database that has [shared cache] enabled. The [sqlite3_create_module()] interface will return an error if the [shared cache] feature is enabled. <h3>1.4 Creating New Virtual Table Implementations</h3> <p>Follow these steps to create your own virtual table: <p> <ol> <li> Write all necessary methods. <li> Create an instance of the [sqlite3_module] structure containing pointers to all the methods from step 1. <li> Register your [sqlite3_module] structure using one of the [sqlite3_create_module()] or [sqlite3_create_module_v2()] interfaces. <li> Run a [CREATE VIRTUAL TABLE] command that specifies the new module in the USING clause. </ol> <p>The only really hard part is step 1. You might want to start with an existing virtual table implementation and modify it to suit your needs. There are several virtual table implementations in the SQLite source tree (for testing purposes). You might use one of those as a guide. Locate these test virtual table implementations by searching for "sqlite3_create_module". <p>You might also want to implement your new virtual table as a [sqlite3_load_extension | loadable extension]. <h2>2.0 Virtual Table Methods</h2> <tcl>############################################################### xCreate hd_fragment xcreate {sqlite3_module.xCreate} {xCreate}</tcl> <h3>2.1 The xCreate Method</h3> <blockquote><pre> int (*xCreate)(sqlite3 *db, void *pAux, int argc, char **argv, sqlite3_vtab **ppVTab, char **pzErr); </pre></blockquote> <p>This method is called to create a new instance of a virtual table in response to a [CREATE VIRTUAL TABLE] statement. The db parameter is a pointer to the SQLite [database connection] that is executing the [CREATE VIRTUAL TABLE] statement. The pAux argument is the copy of the client data pointer that was the fourth argument to the [sqlite3_create_module()] or [sqlite3_create_module_v2()] call that registered the [virtual table module]. The argv parameter is an array of argc pointers to null terminated strings. The first string, argv[0], is the name of the module being invoked. The module name is the name provided as the second argument to [sqlite3_create_module()] and as the argument to the USING clause of the [CREATE VIRTUAL TABLE] statement that is running. The second, argv[1], is the name of the database in which the new virtual table is being created. The database name is "main" for the primary database, or "temp" for TEMP database, or the name given at the end of the [ATTACH] statement for attached databases. The third element of the array, argv[2], is the name of the new virtual table, as specified following the TABLE keyword in the [CREATE VIRTUAL TABLE] statement. If present, the fourth and subsquent strings in the argv[] array report the arguments to the module name in the [CREATE VIRTUAL TABLE] statement. <p>The job of this method is to construct the new virtual table object (an [sqlite3_vtab] object) and return a pointer to it in *ppVTab. <p>As part of the task of creating a new [sqlite3_vtab] structure, this method <u>must</u> invoke [sqlite3_declare_vtab()] to tell the SQLite core about the columns and datatypes in the virtual table. The [sqlite3_declare_vtab()] API has the following prototype: <blockquote><pre> int sqlite3_declare_vtab(sqlite3 *db, const char *zCreateTable) </pre></blockquote> <p>The first argument to [sqlite3_declare_vtab()] must be the same [database connection] pointer as the first parameter to this method. The second argument to [sqlite3_declare_vtab()] must a zero-terminated UTF-8 string that contains a well-formed [CREATE TABLE] statement that defines the columns in the virtual table and their data types. The name of the table in this CREATE TABLE statement is ignored, as are all constraints. Only the column names and datatypes matter. The CREATE TABLE statement string need not to be held in persistent memory. The string can be deallocated and/or reused as soon as the [sqlite3_declare_vtab()] routine returns. <p>If a column datatype contains the special keyword "HIDDEN" (in any combination of upper and lower case letters) then that keyword it is omitted from the column datatype name and the column is marked as a hidden column internally. A hidden column differs from a normal column in three respects: <p> <ul> <li> Hidden columns are not listed in the dataset returned by "[PRAGMA table_info]", <li> Hidden columns are not included in the expansion of a "*" expression in the result set of a [SELECT], and <li> Hidden columns are not included in the implicit column-list used by an [INSERT] statement that lacks an explicit column-list. </ul> <p>For example, if the following SQL is passed to [sqlite3_declare_vtab()]: <blockquote><pre> CREATE TABLE x(a HIDDEN VARCHAR(12), b INTEGER, c INTEGER Hidden); </pre></blockquote> <p>Then the virtual table would be created with two hidden columns, and with datatypes of "VARCHAR(12)" and "INTEGER". <p>The xCreate must should return [SQLITE_OK] if it is successful in creating the new virtual table, or [SQLITE_ERROR] if it is not successful. If not successful, the [sqlite3_vtab] structure must not be allocated. An error message may optionally be returned in *pzErr if unsuccessful. Space to hold the error message string must be allocated using an SQLite memory allocation function like [sqlite3_malloc()] or [sqlite3_mprintf()] as the SQLite core will attempt to free the space using [sqlite3_free()] after the error has been reported up to the application. <p>The xCreate method is required for every virtual table implementation, though the xCreate and [xConnect] pointers of the [sqlite3_module] object may point to the same function the virtual table does not need to initialize backing store. <tcl>############################################################# xConnect hd_fragment xconnect {sqlite3_module.xConnect} {xConnect}</tcl> <h3>2.2 The xConnect Method</h3> <blockquote><pre> int (*xConnect)(sqlite3*, void *pAux, int argc, char **argv, sqlite3_vtab **ppVTab, char **pzErr); </pre></blockquote> <p>The xConnect method is very similar to [xCreate]. It has the same parameters and constructs a new [sqlite3_vtab] structure just like xCreate. And it must also call [sqlite3_declare_vtab()] like xCreate. <p>The difference is that xConnect is called to establish a new connection to an existing virtual table whereas xCreate is called to create a new virtual table from scratch. |
︙ | ︙ | |||
364 365 366 367 368 369 370 | <p>Another example is a virtual table that implements a full-text index. The xCreate method must create and initialize data structures to hold the dictionary and posting lists for that index. The xConnect method, on the other hand, only has to locate and use an existing dictionary and posting lists that were created by a prior xCreate call. | | | > > > > > | > > > > > > > > | | 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 | <p>Another example is a virtual table that implements a full-text index. The xCreate method must create and initialize data structures to hold the dictionary and posting lists for that index. The xConnect method, on the other hand, only has to locate and use an existing dictionary and posting lists that were created by a prior xCreate call. <p>The xConnect method must return [SQLITE_OK] if it is successful in creating the new virtual table, or [SQLITE_ERROR] if it is not successful. If not successful, the [sqlite3_vtab] structure must not be allocated. An error message may optionally be returned in *pzErr if unsuccessful. Space to hold the error message string must be allocated using an SQLite memory allocation function like [sqlite3_malloc()] or [sqlite3_mprintf()] as the SQLite core will attempt to free the space using [sqlite3_free()] after the error has been reported up to the application. <p>The xConnect method is required for every virtual table implementation, though the [xCreate] and xConnect pointers of the [sqlite3_module] object may point to the same function the virtual table does not need to initialize backing store. <tcl>############################################################ xBestIndex hd_fragment xbestindex {sqlite3_module.xBestIndex} {xBestIndex}</tcl> <h3>2.3 The xBestIndex Method</h3> <p>SQLite uses the xBestIndex method of a virtual table module to determine the best way to access the virtual table. The xBestIndex method has a prototype like this: <blockquote><pre> int (*xBestIndex)(sqlite3_vtab *pVTab, sqlite3_index_info*); </pre></blockquote> <p>The SQLite core communicates with the xBestIndex method by filling in certain fields of the [sqlite3_index_info] structure and passing a |
︙ | ︙ | |||
426 427 428 429 430 431 432 | #define SQLITE_INDEX_CONSTRAINT_LE 8 #define SQLITE_INDEX_CONSTRAINT_LT 16 #define SQLITE_INDEX_CONSTRAINT_GE 32 #define SQLITE_INDEX_CONSTRAINT_MATCH 64 </pre></blockquote> <p>The SQLite core calls the xBestIndex method when it is compiling a query | | | > < < > > > > > > > > > > > > > > > > | 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 | #define SQLITE_INDEX_CONSTRAINT_LE 8 #define SQLITE_INDEX_CONSTRAINT_LT 16 #define SQLITE_INDEX_CONSTRAINT_GE 32 #define SQLITE_INDEX_CONSTRAINT_MATCH 64 </pre></blockquote> <p>The SQLite core calls the xBestIndex method when it is compiling a query that involves a virtual table. In other words, SQLite calls this method when it is running [sqlite3_prepare()] or the equivalent. By calling this method, the SQLite core is saying to the virtual table that it needs to access some subset of the rows in the virtual table and it wants to know the most efficient way to do that access. The xBestIndex method replies with information that the SQLite core can then use to conduct an efficient search of the virtual table. <p>While compiling a single SQL query, the SQLite core might call xBestIndex multiple times with different settings in [sqlite3_index_info]. The SQLite core will then select the combination that appears to give the best performance. <p>Before calling this method, the SQLite core initializes an instance of the [sqlite3_index_info] structure with information about the query that it is currently trying to process. This information derives mainly from the WHERE clause and ORDER BY or GROUP BY clauses of the query, but also from any ON or USING clauses if the query is a join. The information that the SQLite core provides to the xBestIndex method is held in the part of the structure that is marked as "Inputs". The "Outputs" section is initialized to zero. <p>The information in the [sqlite3_index_info] structure is ephemeral and may be overwritten or deallocated as soon as the xBestIndex method returns. If the xBestIndex method needs to remember any part of the [sqlite3_index_info] structure, it should make a copy. Care must be take to store the copy in a place where it will be deallocated, such as in the idxStr field with needToFreeIdxStr set to 1. <p>Note that xBestIndex will always be called before [xFilter], since the idxNum and idxStr outputs from xBestIndex are required inputs to xFilter. However, there is no guarantee that xFilter will be called following a successful xBestIndex. <p>The xBestIndex method is required for every virtual table implementation. <h4>2.3.1 Inputs</h4> <p>The main thing that the SQLite core is trying to communicate to the virtual table is the constraints that are available to limit the number of rows that need to be searched. The aConstraint[] array contains one entry for each constraint. There will be exactly nConstraint entries in that array. |
︙ | ︙ | |||
472 473 474 475 476 477 478 | if the WHERE clause contained a term like this: <blockquote><pre> a = 5 </pre></blockquote> <p>Then one of the constraints would be on the "a" column with | | | > | | > > > | | | > | | > | | | 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 | if the WHERE clause contained a term like this: <blockquote><pre> a = 5 </pre></blockquote> <p>Then one of the constraints would be on the "a" column with operator "=" and an expression of "5". Constraints need not have a literal representation of the WHERE clause. The query optimizer might make transformations to the WHERE clause in order to extract as many constraints as it can. So, for example, if the WHERE clause contained something like this: <blockquote><pre> x BETWEEN 10 AND 100 AND 999>y </pre></blockquote> <p>The query optimizer might translate this into three separate constraints: <blockquote><pre> x >= 10 x <= 100 y < 999 </pre></blockquote> <p>For each constraint, the aConstraint[].iColumn field indicates which column appears on the left-hand side of the constraint. The first column of the virtual table is column 0. The rowid of the virtual table is column -1. The aConstraint[].op field indicates which operator is used. The SQLITE_INDEX_CONSTRAINT_* constants map integer constants into operator values. Columns occur in the order they were defined by the call to [sqlite3_declare_vtab()] in the [xCreate] or [xConnect] method. Hidden columns are counted when determining the column index. <p>The aConstraint[] array contains information about all constraints that apply to the virtual table. But some of the constraints might not be usable because of the way tables are ordered in a join. The xBestIndex method must therefore only consider constraints that have a aConstraint[].usable flag which is true. <p>In addition to WHERE clause constraints, the SQLite core also tells the xBestIndex method about the ORDER BY clause. (In an aggregate query, the SQLite core might put in GROUP BY clause information in place of the ORDER BY clause information, but this fact should not make any difference to the xBestIndex method.) If all terms of the ORDER BY clause are columns in the virtual table, then nOrderBy will be the number of terms in the ORDER BY clause and the aOrderBy[] array will identify the column for each term in the order by clause and whether or not that column is ASC or DESC. <h4>2.3.2 Outputs</h4> <p>Given all of the information above, the job of the xBestIndex method it to figure out the best way to search the virtual table. <p>The xBestIndex method fills the idxNum and idxStr fields with information that communicates an indexing strategy to the [xFilter] method. The information in idxNum and idxStr is arbitrary as far as the SQLite core is concerned. The SQLite core just copies the information through to the xFilter method. Any desired meaning can be assigned to idxNum and idxStr as long as xBestIndex and xFilter agree on what that meaning is. <p>The idxStr value may be a string obtained from an SQLite memory allocation function such as [sqlite3_mprintf()]. If this is the case, then the needToFreeIdxStr flag must be set to true so that the SQLite core will know to call [sqlite3_free()] on that string when it has finished with it, and thus avoid a memory leak. <p>If the virtual table will output rows in the order specified by the ORDER BY clause, then the orderByConsumed flag may be set to true. If the output is not automatically in the correct order then orderByConsumed must be left in its default false setting. This will indicate to the SQLite core that it will need to do a separate sorting pass over the data after it comes out of the virtual table. <p>The estimatedCost field should be set to the estimated number of disk access operations required to execute this query against the virtual table. The SQLite core will often call xBestIndex multiple times with different constraints, obtain multiple cost |
︙ | ︙ | |||
566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 | the EXPR value of the aConstraint[3] constraint. <p>By default, the SQLite core double checks all constraints on each row of the virtual table that it receives. If such a check is redundant, the xBestFilter method can suppress that check by setting aConstraintUsage[].omit. <h3>2.4 The xDisconnect Method</h3> <blockquote><pre> int (*xDisconnect)(sqlite3_vtab *pVTab); </pre></blockquote> <p>This method releases a connection to a virtual table. The virtual table is not destroyed and any backing store associated with the virtual table persists. | > > > > | > > > > > > > > > | | | > > > > > > | | > > > | | > | > > > | > > | > | | > > > > > > | > > > | | | | < | > | | | > | | | | > > > > | | | > | | | > > > > | > > > > > > | > > > > > > > > | > > > > > > > | > > > > > | > > > | | | | | | | | 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 | the EXPR value of the aConstraint[3] constraint. <p>By default, the SQLite core double checks all constraints on each row of the virtual table that it receives. If such a check is redundant, the xBestFilter method can suppress that check by setting aConstraintUsage[].omit. <tcl>########################################################## xDisconnect hd_fragment xdisconnect {sqlite3_module.xDisconnect} {xDisconnect}</tcl> <h3>2.4 The xDisconnect Method</h3> <blockquote><pre> int (*xDisconnect)(sqlite3_vtab *pVTab); </pre></blockquote> <p>This method releases a connection to a virtual table. Only the [sqlite3_vtab] object is destroyed. The virtual table is not destroyed and any backing store associated with the virtual table persists. This method undoes the work of [xConnect]. <p>This method is a destructor for a connection to the virtual table. Constrast this method with [xDestroy]. The xDestroy is a destructor for the entire virtual table. <p>The xDestroy method is required for every virtual table implementation, though it is acceptable for the [xDisconnect] and xDestroy methods to be the same function if that makes sense for the particular virtual table. <tcl>########################################################## xDestroy hd_fragment {sqlite3_module.xDestroy} {xDestroy}</tcl> <h3>2.5 The xDestroy Method</h3> <blockquote><pre> int (*xDestroy)(sqlite3_vtab *pVTab); </pre></blockquote> <p>This method releases a connection to a virtual table, just like the [xDisconnect] method, and it also destroys the underlying table implementation. This method undoes the work of [xCreate]. <p>The [xDisconnect] method is called whenever a database connection that uses a virtual table is closed. The xDestroy method is only called when a [DROP TABLE] statement is executed against the virtual table. <p>The xDisconnect method is required for every virtual table implementation, though it is acceptable for the xDisconnect and [xDestroy] methods to be the same function if that makes sense for the particular virtual table. <tcl>########################################################## xOpen hd_fragment xopen {sqlite3_module.xOpen}</tcl> <h3>2.6 The xOpen Method</h3> <blockquote><pre> int (*xOpen)(sqlite3_vtab *pVTab, sqlite3_vtab_cursor **ppCursor); </pre></blockquote> <p>The xOpen method creates a new cursor used for accessing (read and/or writing) a virtual table. A successful invocation of this method will allocate the memory for the [sqlite3_vtab_cursor] (or a subclass), initialize the new object, and make *ppCursor point to the new object. The successful call then returns [SQLITE_OK]. <p>For every successful call to this method, the SQLite core will later invoke the [sqlite3_module.xClose | xClose] method to destroy the allocated cursor. <p>A virtual table implementation must be able to support an arbitrary number of simultaneously open cursors. <p>When initially opened, the cursor is in an undefined state. The SQLite core will invoke the [xFilter] method on the cursor prior to any attempt to position or read from the cursor. <p>The xOpen method is required for every virtual table implementation. <tcl>############################################################### xClose hd_fragment xclose {sqlite3_module.xClose}</tcl> <h3>2.7 The xClose Method</h3> <blockquote><pre> int (*xClose)(sqlite3_vtab_cursor*); </pre></blockquote> <p>The xClose method closes a cursor previously opened by [sqlite3_module.xOpen | xOpen]. The SQLite core will always call xClose once for each cursor opened using xOpen. <p>This method must release all resources allocated by the corresponding xOpen call. The routine will not be called again even if it returns an error. The SQLite core will not use the [sqlite3_vtab_cursor] again after it has been closed. <p>The xClose method is required for every virtual table implementation. <tcl>############################################################## xEof hd_fragment xeof {sqlite3_module.xEof} {xEof}</tcl> <h3>2.8 The xEof Method</h3> <blockquote><pre> int (*xEof)(sqlite3_vtab_cursor*); </pre></blockquote> <p>The xEof method must return false (zero) if the specified cursor currently points to a valid row of data, or true (non-zero) otherwise. This method is called by the SQL engine immediately after each [xFilter] and [xNext] invocation. <p>The xEof method is required for every virtual table implementation. <tcl>############################################################## xFilter hd_fragment xfilter {sqlite3_module.xFilter} {xFilter}</tcl> <h3>2.9 The xFilter Method</h3> <blockquote><pre> int (*xFilter)(sqlite3_vtab_cursor*, int idxNum, const char *idxStr, int argc, sqlite3_value **argv); </pre></blockquote> <p>This method begins a search of a virtual table. The first argument is a cursor opened by [sqlite3_module.xOpen | xOpen]. The next two argument define a particular search index previously choosen by [xBestIndex]. The specific meanings of idxNum and idxStr are unimportant as long as xFilter and xBestIndex agree on what that meaning is. <p>The xBestIndex function may have requested the values of certain expressions using the aConstraintUsage[].argvIndex values of the [sqlite3_index_info] structure. Those values are passed to xFilter using the argc and argv parameters. <p>If the virtual table contains one or more rows that match the search criteria, then the cursor must be left point at the first row. Subsequent calls to [xEof] must return false (zero). If there are no rows match, then the cursor must be left in a state that will cause the [xEof] to return true (non-zero). The SQLite engine will use the [xColumn] and [xRowid] methods to access that row content. The [xNext] method will be used to advance to the next row. <p>This method must return [SQLITE_OK] if successful, or an sqlite [error code] if an error occurs. <p>The xFilter method is required for every virtual table implementation. <tcl>############################################################### xNext hd_fragment xnext {sqlite3_module.xNext} {xNext}</tcl> <h3>2.10 The xNext Method</h3> <blockquote><pre> int (*xNext)(sqlite3_vtab_cursor*); </pre></blockquote> <p>The xNext method advances a [virtual table cursor] to the next row of a result set initiated by [xFilter]. If the cursor is already pointing at the last row when this routine is called, then the cursor no longer points to valid data and a subsequent call to the [xEof] method must return true (non-zero). If the cursor is successfully advanced to another row of content, then subsequent calls to [xEof] must return false (zero). <p>This method must return [SQLITE_OK] if successful, or an sqlite [error code] if an error occurs. <p>The xNext method is required for every virtual table implementation. <tcl>############################################################## xColumn hd_fragment xcolumn {sqlite3_module.xColumn} {xColumn}</tcl> <h3>2.11 The xColumn Method</h3> <blockquote><pre> int (*xColumn)(sqlite3_vtab_cursor*, sqlite3_context*, int N); </pre></blockquote> <p>The SQLite core invokes this method in order to find the value for the N-th column of the current row. N is zero-based so the first column is numbered 0. The xColumn method must uses one of the [sqlite3_result_blob | sqlite3_result_*() APIs] to return the result. The xColumn method returns its result back to SQLite using one of the following interface: <p> <li> [sqlite3_result_blob()] <li> [sqlite3_result_double()] <li> [sqlite3_result_int()] <li> [sqlite3_result_int64()] <li> [sqlite3_result_null()] <li> [sqlite3_result_text()] <li> [sqlite3_result_text16()] <li> [sqlite3_result_text16le()] <li> [sqlite3_result_text16be()] <li> [sqlite3_result_zeroblob()] </p> <p>To raise an error, the xColumn method should use one of the result_text() methods to set the error message text, then return an appropriate [error code]. The xColumn method must return [SQLITE_OK] on success. <p>The xColumn method is required for every virtual table implementation. <tcl>############################################################# xRowid hd_fragment xrowid {sqlite3_module.xRowid} {xRowid}</tcl> <h3>2.12 The xRowid Method</h3> <blockquote><pre> int (*xRowid)(sqlite3_vtab_cursor *pCur, sqlite_int64 *pRowid); </pre></blockquote> <p>A successful invocation of this method will cause *pRowid to be filled with the [rowid] of row that the [virtual table cursor] pCur is currently pointing at. This method returns [SQLITE_OK] on success. It returns an appropriate [error code] on failure.</p> <p>The xRowid method is required for every virtual table implementation. <tcl>############################################################# xUpdate hd_fragment xupdate {sqlite3_module.xUpdate} {xUpdate}</tcl> <h3>2.13 The xUpdate Method</h3> <blockquote><pre> int (*xUpdate)( sqlite3_vtab *pVTab, int argc, sqlite3_value **argv, sqlite_int64 *pRowid ); </pre></blockquote> <p>All changes to a virtual table are made using the xUpdate method. This one method can be used to insert, delete, or update. <p>The argc parameter specifies the number of entries in the argv array. Every argv entry will have a non-NULL value in C (but may contain the SQL value NULL). <p>The argv[0] parameter is the [rowid] of a row in the virtual table to be deleted. If argv[0] is NULL, then no deletion occurs. <p>The argv[1] parameter is the rowid of a new row to be inserted into the virtual table. If argv[1] is NULL, then the implementation must choose a rowid for the newly inserted row. Subsequent argv[] entries contain values of the columns of the virtual table, in the order that the columns were declared. The number of columns will match the table declaration that the [xConnect] or [xCreate] method made using the [sqlite3_declare_vtab()] call. All hidden columns are included. <p>When doing an insert without a rowid (argc>1, argv[1]==NULL), the implementation must set *pRowid to the rowid of the newly inserted row; this will become the value returned by the [sqlite3_last_insert_rowid()] function. Setting this value in all the other cases is a harmless no-op; the SQLite engine ignores the *pRowid return value if argc==1 or argv[1]!=NULL. <p>Each call to xUpdate will fall into one of the following cases: <blockquote> <dl> <dt><b>argc == 1</b> <dd><p>The single row argv[0] is deleted; no insert occurs |
︙ | ︙ | |||
762 763 764 765 766 767 768 | when an SQL statement updates a rowid, as in the statement: <blockquote> [UPDATE] table SET rowid=rowid+1 WHERE ...; </blockquote> </dl> </blockquote> | > > > > > | > > > > > > > > > > | > > > > > > > > | | | | > > > > > | > > > > > > > > > | > > > > > > > > > > > > > > > | > > > > > > > > | > > > > | | | > > | 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 | when an SQL statement updates a rowid, as in the statement: <blockquote> [UPDATE] table SET rowid=rowid+1 WHERE ...; </blockquote> </dl> </blockquote> <p>The xUpdate method must return [SQLITE_OK] if and only if it is successful. If a failure occurs, the xUpdate must return an appropriate [error code]. On a failure, the pVTab->zErrMsg element may optionally be replaced with error message text stored in memory allocated from SQLite using functions such as [sqlite3_mprintf()] or [sqlite3_malloc()]. <p>If the xUpdate method violates some constraint of the virtual table (including, but not limited to, attempting to store a value of the wrong datatype, attempting to store a value that is too large or too small, or attempting to change a read-only value) then the xUpdate must fail with an appropriate [error code]. <p>There might be one or more [sqlite3_vtab_cursor] objects open and in use on the virtual table instance and perhaps even on the row of the virtual table when the xUpdate method is invoked. The implementation of xUpdate must be prepared for attempts to delete or modify rows of the table out from other existing cursors. If the virtual table cannot accommodate such changes, the xUpdate method must return an [error code]. <p>The xUpdate method is optional. If the xUpdate pointer in the [sqlite3_module] for a virtual table is a NULL pointer, then the virtual table is read-only. <tcl>########################################################## xFindFunction hd_fragment xfindfunction {sqlite3_module.xFindFunction} {xFindFunction}</tcl> <h3>2.14 The xFindFunction Method</h3> <blockquote><pre> int (*xFindFunction)( sqlite3_vtab *pVtab, int nArg, const char *zName, void (**pxFunc)(sqlite3_context*,int,sqlite3_value**), void **ppArg ); </pre></blockquote> <p>This method is called during [sqlite3_prepare()] to give the virtual table implementation an opportunity to overload functions. This method may be set to NULL in which case no overloading occurs. <p>When a function uses a column from a virtual table as its first argument, this method is called to see if the virtual table would like to overload the function. The first three parameters are inputs: the virtual table, the number of arguments to the function, and the name of the function. If no overloading is desired, this method returns 0. To overload the function, this method writes the new function implementation into *pxFunc and writes user data into *ppArg and returns 1. <p>Note that infix functions ([LIKE], [GLOB], [REGEXP], and [MATCH]) reverse the order of their arguments. So "like(A,B)" is equivalent to "B like A". For the form "B like A" the B term is considered the first argument to the function. But for "like(A,B)" the A term is considered the first argument. <p>The function pointer returned by this routine must be valid for the lifetime of the [sqlite3_vtab] object given in the first parameter. <tcl>############################################################ xBegin hd_fragment xBegin {sqlite3_module.xBegin} {xBegin}</tcl> <h3>2.15 The xBegin Method</h3> <blockquote><pre> int (*xBegin)(sqlite3_vtab *pVTab); </pre></blockquote> <p>This method begins a transaction on a virtual table. This is method is optional. The xBegin pointer of [sqlite3_module] may be NULL. <p>This method is always followed by one call to either the [xCommit] or [xRollback] method. Virtual table transactions do not nest, so the xBegin method will not be invoked more than once on a single virtual table without an intervening call to either [xCommit] or [xRollback]. Multiple calls to other methods can and likely will occur in between the xBegin and the corresponding [xCommit] or [xRollback]. <tcl>############################################################ xSync hd_fragment xsync {sqlite3_module.xSync}</tcl> <h3>2.16 The xSync Method</h3> <blockquote><pre> int (*xSync)(sqlite3_vtab *pVTab); </pre></blockquote> <p>This method signals the start of a two-phase commit on a virtual table. This is method is optional. The xSync pointer of [sqlite3_module] may be NULL. <p>This method is only invoked after call to the [xBegin] method and prior to an [xCommit] or [xRollback]. In order to implement two-phase commit, the xSync method on all virtual tables is invoked prior to invoking the [xCommit] method on any virtual table. If any of the xSync methods fail, the entire transaction is rolled back. <tcl>########################################################### xCommit hd_fragment xcommit {sqlite3_module.xCommit} {xCommit}</tcl> <h3>2.17 The xCommit Method</h3> <blockquote><pre> int (*xCommit)(sqlite3_vtab *pVTab); </pre></blockquote> <p>This method causes a virtual table transaction to commit. This is method is optional. The xCommit pointer of [sqlite3_module] may be NULL. <p>A call to this method always follows a prior call to [xBegin] and [sqlite3_module.xSync|xSync]. <tcl>############################################################## xRollback hd_fragment xrollback {sqlite3_module.xRollback} {xRollback}</tcl> <h3>2.18 The xRollback Method</h3> <blockquote><pre> int (*xRollback)(sqlite3_vtab *pVTab); </pre></blockquote> <p>This method causes a virtual table transaction to rollback. This is method is optional. The xRollback pointer of [sqlite3_module] may be NULL. <p>A call to this method always follows a prior call to [xBegin]. <tcl>############################################################# xRename hd_fragment xrename {sqlite3_module.xRename} {xRename}</tcl> <h3>2.19 The xRename Method</h3> <blockquote><pre> int (*xRename)(sqlite3_vtab *pVtab, const char *zNew); </pre></blockquote> <p>This method provides notification that the virtual table implementation that the virtual table will be given a new name. If this method returns [SQLITE_OK] then SQLite renames the table. If this method returns an [error code] then the renaming is prevented. <p>The xRename method is required for every virtual table implementation. |
Changes to req/hlr30000.txt.
︙ | ︙ | |||
849 850 851 852 853 854 855 | structure (not part of an overflow chain), the page type of the corresponding pointer-map entry is set to the value 0x05 and the parent page number field is set to the page number of the parent node in the B-Tree structure. | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | < < | 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 | structure (not part of an overflow chain), the page type of the corresponding pointer-map entry is set to the value 0x05 and the parent page number field is set to the page number of the parent node in the B-Tree structure. HLR H32000 If a <i>journal file</i> contains a well-formed <i>master-journal pointer</i>, and the named <i>master-journal file</i> either does not exist or does not contain the name of the <i>journal file</i>, then the <i>journal file</i> shall be considered invalid. HLR H32010 If the first 28 bytes of a <i>journal file</i> do not contain a well-formed <i>journal header</i>, then the <i>journal file</i> shall be considered invalid. HLR H32020 If the journal file exists within the file-system and neither H32000 nor H32010 apply, then the journal file shall be considered valid. HLR H32030 If there exists a valid <i>journal file</i> in the file-system, then the database <i>page-size</i> in bytes used to interpret the <i>database image</i> shall be the value stored as a 4-byte big-endian unsigned integer at byte offset 24 of the <i>journal file</i>. HLR H32040 If there exists a valid <i>journal file</i> in the file-system, then the number of pages in the <i>database image</i> shall be the value stored as a 4-byte big-endian unsigned integer at byte offset 24 of the <i>journal file</i>. HLR H32050 If there is no valid <i>journal file</i> in the file-system, then the database <i>page-size</i> in bytes used to interpret the <i>database image</i> shall be the value stored as a 2-byte big-endian unsigned integer at byte offset 16 of the <i>database file</i>. HLR H32060 If there is no valid <i>journal file</i> in the file-system, then the number of pages in the <i>database image</i> shall be calculated by dividing the size of the <i>database file</i> in bytes by the database <i>page-size</i>. HLR H32070 If there exists a valid <i>journal file</i> in the file-system, then the contents of each page of the <i>database image</i> for which there is a valid <i>journal record</i> in the <i>journal file</i> shall be read from the corresponding journal record. HLR H32080 The contents of all <i>database image</i> pages for which there is no valid <i>journal record</i> shall be read from the database file. HLR H32090 A buffer of 28 bytes shall be considered a well-formed journal header if it is not excluded by requirements H32180, H32190 or H32200. HLR H32180 A buffer of 28 bytes shall only be considered a well-formed journal header if the first eight bytes of the buffer contain the values 0xd9, 0xd5, 0x05, 0xf9, 0x20, 0xa1, 0x63, and 0xd7, respectively. HLR H32190 A buffer of 28 bytes shall only be considered a well-formed journal header if the value stored in the sector size field (the 4-byte big-endian unsigned integer at offset 20 of the buffer) contains a value that is an integer power of two greater than 512. HLR H32200 A buffer of 28 bytes shall only be considered a well-formed journal header if the value stored in the page size field (the 4-byte big-endian unsigned integer at offset 24 of the buffer) contains a value that is an integer power of two greater than 512. HLR H32100 A buffer of (8 + page size) bytes shall be considered a well-formed journal record if it is not excluded by requirements H32110 or H32120. HLR H32110 A journal record shall only be considered to be well-formed if the page number field contains a value other than zero and the locking-page number, calculated using the page size found in the first journal header of the journal file that contains the journal record. HLR H32120 A journal record shall only be considered to be well-formed if the checksum field contains a value equal to the sum of the value stored in the checksum-initializer field of the journal header that precedes the record and the value stored in every 200th byte of the page data field, interpreted as an 8-bit unsigned integer), starting with byte offset (page-size % 200) and ending with the byte at byte offset (page-size - 200). HLR H32130 A buffer shall be considered to contain a well-formed master journal pointer record if it is not excluded from this category by requirements H32140, H32150, H32160 or H32170. HLR H32140 A buffer shall only be considered to be a well-formed master journal pointer if the final eight bytes of the buffer contain the values 0xd9, 0xd5, 0x05, 0xf9, 0x20, 0xa1, 0x63, and 0xd7, respectively. HLR H32150 A buffer shall only be considered to be a well-formed master journal pointer if the size of the buffer in bytes is equal to the value stored as a 4-byte big-endian unsigned integer starting 16 bytes before the end of the buffer. HLR H32160 A buffer shall only be considered to be a well-formed master journal pointer if the first four bytes of the buffer, interpreted as a big-endian unsigned integer, contain the page number of the locking page (the value (1 + 2<sup>30</sup> / page-size), where page-size is the value stored in the page-size field of the first journal header of the journal file). HLR H32170 A buffer shall only be considered to be a well-formed master journal pointer if the value stored as a 4-byte big-endian integer starting 12 bytes before the end of the buffer is equal to the sum of all bytes, each interpreted as an 8-bit unsigned integer, starting at offset 4 of the buffer and continuing until offset (buffer-size - 16) (the 17th last byte of the buffer). HLR H32210 A buffer shall be considered to contain a well-formed journal section record if it is not excluded from this category by requirements H32220, H32230 or H32240. HLR H32220 A buffer shall only be considered to contain a well-formed journal section if the first 28 bytes of it contain a well-formed journal header. HLR H32230 A buffer shall only be considered to contain a well-formed journal section if, beginning at byte offset sector-size, it contains a sequence of record-count well-formed journal records. In this case sector-size and record-count are the integer values stored in the sector size and record count fields of the journal section's journal header. HLR H32240 A buffer shall only be considered to contain a well-formed journal section if it is an integer multiple of sector-size bytes in size, where sector-size is the value stored in the sector size field of the journal section's journal header. HLR H32250 A journal record found within a valid journal file shall be considered a valid journal record if it is not excluded from this category by requirement H32260, H32270 or H32280. HLR H32260 A journal record shall only be considered a valid journal record if it and any other journal records that occur before it within the same journal section are well-formed. HLR H32270 A journal record shall only be considered a valid journal record if the journal section to which it belongs begins with a well-formed journal header. HLR H32280 A journal record shall only be considered a valid journal record if all journal sections that occur before the journal section containing the journal record are well-formed journal sections. |