Documentation Source Text

<tcl>hd_keywords *fts3 FTS3 {full-text search}</tcl>
<title>SQLite FTS3 and FTS4 Extensions</title>

<table_of_contents>

<h2 style="margin-left:1.0em" notoc> Overview</h2>

<p>
  FTS3 and FTS4 are SQLite virtual table modules that allows users to perform 
  full-text searches on a set of documents. The most common (and effective) 
  way to describe full-text searches is "what Google, Yahoo, and Bing do
  with documents placed on the World Wide Web". Users input a term, or series 
  of terms, perhaps connected by a binary operator or grouped together into a 

<tcl>hd_keywords *fts3 FTS3 {full-text search}</tcl>
<title>SQLite FTS3 and FTS4 Extensions</title>

<table_of_contents>

<h2 id=overview style="margin-left:1.0em" notoc> Overview</h2>

<p>
  FTS3 and FTS4 are SQLite virtual table modules that allows users to perform 
  full-text searches on a set of documents. The most common (and effective) 
  way to describe full-text searches is "what Google, Yahoo, and Bing do
  with documents placed on the World Wide Web". Users input a term, or series 
  of terms, perhaps connected by a binary operator or grouped together into a 

      set nosearch(releaselog/current.html) 1


      # As of version 3.7.16, sub-release changelogs duplicated the entries
      # from the major release. This block does the following:
      #
      #   * sets the weight of a changelog containing superceded content
      #     to 25%
      #   * sets the weights of other changelogs to 50%.
      #
      foreach f [glob releaselog/*.html] { 
        set tail [file tail $f]
        set ::weight($f) 50
        if {[regexp {^(3_8_[0-9]*).*} $tail -> prefix]
         || [regexp {^(3_7_16).*} $tail -> prefix]
         || [regexp {^(3_9_).*} $tail -> prefix]
         || [regexp {^(3_[1-9][0-9]).*} $tail -> prefix]
        } {
          set f1 [lindex [lsort -decreasing [glob releaselog/$prefix*.html]] 0]
          if {$f!=$f1} { set ::weight($f) 25 }
        } 
      }

      foreach f [glob releaselog/*.html] { 
        if {[info exists nosearch($f)]==0} { 
          lappend lFiles $f 
        }

      set nosearch(releaselog/current.html) 1


      # As of version 3.7.16, sub-release changelogs duplicated the entries
      # from the major release. This block does the following:
      #
      #   * sets the weight of a changelog containing superceded content
      #     to 10%
      #   * sets the weights of other changelogs to 25%.
      #
      foreach f [glob releaselog/*.html] { 
        set tail [file tail $f]
        set ::weight($f) 25
        if {[regexp {^(3_8_[0-9]*).*} $tail -> prefix]
         || [regexp {^(3_7_16).*} $tail -> prefix]
         || [regexp {^(3_9_).*} $tail -> prefix]
         || [regexp {^(3_[1-9][0-9]).*} $tail -> prefix]
        } {
          set f1 [lindex [lsort -decreasing [glob releaselog/$prefix*.html]] 0]
          if {$f!=$f1} { set ::weight($f) 10 }
        } 
      }

      foreach f [glob releaselog/*.html] { 
        if {[info exists nosearch($f)]==0} { 
          lappend lFiles $f 
        }

    memcpy(&pRet, sqlite3_column_blob(pStmt, 0), sizeof(pRet));
  }
  sqlite3_finalize(pStmt);
  return pRet;
}
/************************************************************************/
































































typedef struct STokenizer STokenizer;
typedef struct STokenCtx STokenCtx;

/*
** Tokenizer type. Casts to Fts5Tokenizer.
*/
................................................................................

  return rc;
}

static int register_tokenizer(sqlite3 *db, char **pzErr, void *p){
  fts5_api *pApi;
  fts5_tokenizer t;


  pApi = fts5_api_from_db(db);
  if( pApi==0 ){
    *pzErr = sqlite3_mprintf("fts5_api_from_db: %s", sqlite3_errmsg(db));
    return SQLITE_ERROR;
  }

  t.xCreate = stokenCreate;
  t.xDelete = stokenDelete;
  t.xTokenize = stokenTokenize;


  return pApi->xCreateTokenizer(pApi, "stoken", (void*)pApi, &t, 0);


}





int Fts5ext_Init(Tcl_Interp *interp){
#ifdef USE_TCL_STUBS
  if (Tcl_InitStubs(interp, "8.4", 0) == 0) {
    return TCL_ERROR;
  }
#endif
  sqlite3_auto_extension((void (*)(void))register_tokenizer);
  return TCL_OK;
}

    memcpy(&pRet, sqlite3_column_blob(pStmt, 0), sizeof(pRet));
  }
  sqlite3_finalize(pStmt);
  return pRet;
}
/************************************************************************/

/*
** Simple ranking function used by search script. Assumes the queried
** table has the following 5 indexed columns:
**
**     apis,                      -- C APIs 
**     keywords,                  -- Keywords
**     title1,                    -- Document title
**     title2,                    -- Heading title, if any
**     content,                   -- Document text
**
** This function returns the following integer values:
**
**   10000 - all phrases present in (the combination of) "apis" or "keywords".
**    1000 - all phrases present in (the combination of) "apis", "keywords"
**           or either "title[12] column.
**
** It adds a bonus of 100 if either of the above and the condition 
** (xRowid()>1000 && (xRowid() % 1000)==1) is true.
**
*/
void srankFunc(
  const Fts5ExtensionApi *pApi,   /* API offered by current FTS version */
  Fts5Context *pFts,              /* First arg to pass to pApi functions */
  sqlite3_context *pCtx,          /* Context for returning result/error */
  int nVal,                       /* Number of values in apVal[] array */
  sqlite3_value **apVal           /* Array of trailing arguments */
){
  int nPhrase;                    /* Number of phrases in query */
  int i;                          /* Used to iterate through phrases */
  int rc;                         /* Return code */
  int n1 = 0;
  int n2 = 0;
  int iScore = 0;                 /* Returned value */
  sqlite3_int64 iRowid;           /* Rowid for current row */

  iRowid = pApi->xRowid(pFts);
  if( iRowid<1000 ) return;
  nPhrase = pApi->xPhraseCount(pFts);
  for(i=0; i<nPhrase; i++){
    Fts5PhraseIter iter;
    int ic, io;
    rc = pApi->xPhraseFirst(pFts, i, &iter, &ic, &io);
    if( rc!=SQLITE_OK ){
      sqlite3_result_error(pCtx, "Error in xPhraseFirst", -1);
      return;
    }

    if( ic==0 || ic==1 ) n1++;
    if( ic==2 || ic==3 ) n2++;
  }

  if( n1==nPhrase ){ iScore = 10000; }
  else if( n1+n2==nPhrase ){ iScore = 1000; }

  if( iScore && iRowid>1000 && (iRowid % 1000)==1 ){
    iScore += 100;
  }

  sqlite3_result_int(pCtx, iScore);
}



typedef struct STokenizer STokenizer;
typedef struct STokenCtx STokenCtx;

/*
** Tokenizer type. Casts to Fts5Tokenizer.
*/
................................................................................

  return rc;
}

static int register_tokenizer(sqlite3 *db, char **pzErr, void *p){
  fts5_api *pApi;
  fts5_tokenizer t;
  int rc;

  pApi = fts5_api_from_db(db);
  if( pApi==0 ){
    *pzErr = sqlite3_mprintf("fts5_api_from_db: %s", sqlite3_errmsg(db));
    return SQLITE_ERROR;
  }

  t.xCreate = stokenCreate;
  t.xDelete = stokenDelete;
  t.xTokenize = stokenTokenize;
  rc = pApi->xCreateTokenizer(pApi, "stoken", (void*)pApi, &t, 0);


  if( rc==SQLITE_OK ){
    rc = pApi->xCreateFunction(pApi, "srank", 0, srankFunc, 0);
  }

  return rc;
}


int Fts5ext_Init(Tcl_Interp *interp){
#ifdef USE_TCL_STUBS
  if (Tcl_InitStubs(interp, "8.4", 0) == 0) {
    return TCL_ERROR;
  }
#endif
  sqlite3_auto_extension((void (*)(void))register_tokenizer);
  return TCL_OK;
}

  #
  set iStart [expr {([info exists ::A(i)] ? $::A(i) : 0)*10}]

  # Grab a list of rowid results.
  #
  set q {
    SELECT rowid FROM page WHERE page MATCH $::A(q) 

    ORDER BY rank * COALESCE(
      (SELECT percent FROM weight WHERE id=page.rowid), 100
    );
  }
  if {[catch { set lRowid [db eval $q] }]} {
    set x ""
    foreach word [split $::A(q) " "] {
      append x " \"[string map [list "\"" "\"\""] $word]\""
................................................................................

      append s_content " $data($childid,s_content)"
    }

    append ret [subst -nocommands {<tr>
      <td valign=top style="line-height:150%">
        <div style="white-space:wrap;font-size:larger" class=nounderline>
          <a href="$url">$s_title1</a> 
          <div style="float:right;font-size:smaller;color:#BBB">($url)</div>
        </div>
          <div style="margin-left: 10ex; font:larger monospace">$s_apis</div>
        <div style="margin-left: 4ex; margin-bottom:1.5em">
           $s_content 
        </div>
      </td>

  #
  set iStart [expr {([info exists ::A(i)] ? $::A(i) : 0)*10}]

  # Grab a list of rowid results.
  #
  set q {
    SELECT rowid FROM page WHERE page MATCH $::A(q) 
    ORDER BY srank(page) DESC,
    rank * COALESCE(
      (SELECT percent FROM weight WHERE id=page.rowid), 100
    );
  }
  if {[catch { set lRowid [db eval $q] }]} {
    set x ""
    foreach word [split $::A(q) " "] {
      append x " \"[string map [list "\"" "\"\""] $word]\""
................................................................................

      append s_content " $data($childid,s_content)"
    }

    append ret [subst -nocommands {<tr>
      <td valign=top style="line-height:150%">
        <div style="white-space:wrap;font-size:larger" class=nounderline>
          <a href="$url">$s_title1 </a> 
          <div style="float:right;font-size:smaller;color:#BBB">($url)</div>
        </div>
          <div style="margin-left: 10ex; font:larger monospace">$s_apis</div>
        <div style="margin-left: 4ex; margin-bottom:1.5em">
           $s_content 
        </div>
      </td>