SQLite

** row:
**     CREATE TABLE vocab(term, doc, cnt, PRIMARY KEY(term));
**
**   One row for each term in the database. The value of $doc is set to
**   the number of fts5 rows that contain at least one instance of term
**   $term. Field $cnt is set to the total number of instances of term 
**   $term in the database.
**
** instance:
**     CREATE TABLE vocab(term, doc, col, offset, PRIMARY KEY(<all-fields>));
**
**   One row for each term instance in the database. 
*/


#include "fts5Int.h"


typedef struct Fts5VocabTable Fts5VocabTable;
typedef struct Fts5VocabCursor Fts5VocabCursor;

struct Fts5VocabTable {
  sqlite3_vtab base;
  char *zFts5Tbl;                 /* Name of fts5 table */
  char *zFts5Db;                  /* Db containing fts5 table */
  sqlite3 *db;                    /* Database handle */
  Fts5Global *pGlobal;            /* FTS5 global object for this database */
  int eType;                      /* FTS5_VOCAB_COL, ROW or INSTANCE */
};

struct Fts5VocabCursor {
  sqlite3_vtab_cursor base;
  sqlite3_stmt *pStmt;            /* Statement holding lock on pIndex */
  Fts5Index *pIndex;              /* Associated FTS5 index */

  int bEof;                       /* True if this cursor is at EOF */
  Fts5IndexIter *pIter;           /* Term/rowid iterator object */

  int nLeTerm;                    /* Size of zLeTerm in bytes */
  char *zLeTerm;                  /* (term <= $zLeTerm) paramater, or NULL */

  /* These are used by 'col' tables only */
  Fts5Config *pConfig;            /* Fts5 table configuration */
  int iCol;
  i64 *aCnt;
  i64 *aDoc;

  /* Output values used by all tables. */
  i64 rowid;                      /* This table's current rowid value */
  Fts5Buffer term;                /* Current value of 'term' column */

  /* Output values Used by 'instance' tables only */
  i64 iInstPos;
  int iInstOff;
};

#define FTS5_VOCAB_COL      0
#define FTS5_VOCAB_ROW      1
#define FTS5_VOCAB_INSTANCE 2

#define FTS5_VOCAB_COL_SCHEMA  "term, col, doc, cnt"
#define FTS5_VOCAB_ROW_SCHEMA  "term, doc, cnt"
#define FTS5_VOCAB_INST_SCHEMA "term, doc, col, offset"

/*
** Bits for the mask used as the idxNum value by xBestIndex/xFilter.
*/
#define FTS5_VOCAB_TERM_EQ 0x01
#define FTS5_VOCAB_TERM_GE 0x02
#define FTS5_VOCAB_TERM_LE 0x04

    sqlite3Fts5Dequote(zCopy);
    if( sqlite3_stricmp(zCopy, "col")==0 ){
      *peType = FTS5_VOCAB_COL;
    }else

    if( sqlite3_stricmp(zCopy, "row")==0 ){
      *peType = FTS5_VOCAB_ROW;
    }else
    if( sqlite3_stricmp(zCopy, "instance")==0 ){
      *peType = FTS5_VOCAB_INSTANCE;
    }else
    {
      *pzErr = sqlite3_mprintf("fts5vocab: unknown table type: %Q", zCopy);
      rc = SQLITE_ERROR;
    }
    sqlite3_free(zCopy);
  }

  int argc,                       /* Number of elements in argv array */
  const char * const *argv,       /* xCreate/xConnect argument array */
  sqlite3_vtab **ppVTab,          /* Write the resulting vtab structure here */
  char **pzErr                    /* Write any error message here */
){
  const char *azSchema[] = { 
    "CREATE TABlE vocab(" FTS5_VOCAB_COL_SCHEMA  ")", 
    "CREATE TABlE vocab(" FTS5_VOCAB_ROW_SCHEMA  ")",
    "CREATE TABlE vocab(" FTS5_VOCAB_INST_SCHEMA ")"
  };

  Fts5VocabTable *pRet = 0;
  int rc = SQLITE_OK;             /* Return code */
  int bDb;

  bDb = (argc==6 && strlen(argv[1])==4 && memcmp("temp", argv[1], 4)==0);

  char **pzErr                    /* OUT: sqlite3_malloc'd error message */
){
  return fts5VocabInitVtab(db, pAux, argc, argv, ppVtab, pzErr);
}

/* 
** Implementation of the xBestIndex method.
**
** Only constraints of the form:
**
**     term <= ?
**     term == ?
**     term >= ?
**
** are interpreted. Less-than and less-than-or-equal are treated 
** identically, as are greater-than and greater-than-or-equal.
*/
static int fts5VocabBestIndexMethod(
  sqlite3_vtab *pUnused,
  sqlite3_index_info *pInfo
){
  int i;
  int iTermEq = -1;

  fts5VocabResetCursor(pCsr);
  sqlite3Fts5BufferFree(&pCsr->term);
  sqlite3_finalize(pCsr->pStmt);
  sqlite3_free(pCsr);
  return SQLITE_OK;
}

static int fts5VocabInstanceNewTerm(Fts5VocabCursor *pCsr){
  int rc = SQLITE_OK;
  
  if( sqlite3Fts5IterEof(pCsr->pIter) ){
    pCsr->bEof = 1;
  }else{
    const char *zTerm;
    int nTerm;
    zTerm = sqlite3Fts5IterTerm(pCsr->pIter, &nTerm);
    if( pCsr->nLeTerm>=0 ){
      int nCmp = MIN(nTerm, pCsr->nLeTerm);
      int bCmp = memcmp(pCsr->zLeTerm, zTerm, nCmp);
      if( bCmp<0 || (bCmp==0 && pCsr->nLeTerm<nTerm) ){
        pCsr->bEof = 1;
      }
    }

    sqlite3Fts5BufferSet(&rc, &pCsr->term, nTerm, (const u8*)zTerm);
  }
  return rc;
}

static int fts5VocabInstanceNext(Fts5VocabCursor *pCsr){
  int eDetail = pCsr->pConfig->eDetail;
  int rc = SQLITE_OK;
  Fts5IndexIter *pIter = pCsr->pIter;
  i64 *pp = &pCsr->iInstPos;
  int *po = &pCsr->iInstOff;
  
  while( eDetail==FTS5_DETAIL_NONE
      || sqlite3Fts5PoslistNext64(pIter->pData, pIter->nData, po, pp) 
  ){
    pCsr->iInstPos = 0;
    pCsr->iInstOff = 0;

    rc = sqlite3Fts5IterNextScan(pCsr->pIter);
    if( rc==SQLITE_OK ){
      rc = fts5VocabInstanceNewTerm(pCsr);
      if( eDetail==FTS5_DETAIL_NONE ) break;
    }
    if( rc ){
      pCsr->bEof = 1;
      break;
    }
  }

  return rc;
}

/*
** Advance the cursor to the next row in the table.
*/
static int fts5VocabNextMethod(sqlite3_vtab_cursor *pCursor){
  Fts5VocabCursor *pCsr = (Fts5VocabCursor*)pCursor;
  Fts5VocabTable *pTab = (Fts5VocabTable*)pCursor->pVtab;
  int rc = SQLITE_OK;
  int nCol = pCsr->pConfig->nCol;

  pCsr->rowid++;

  if( pTab->eType==FTS5_VOCAB_INSTANCE ){
    return fts5VocabInstanceNext(pCsr);
  }

  if( pTab->eType==FTS5_VOCAB_COL ){
    for(pCsr->iCol++; pCsr->iCol<nCol; pCsr->iCol++){
      if( pCsr->aDoc[pCsr->iCol] ) break;
    }
  }

  if( pTab->eType!=FTS5_VOCAB_COL || pCsr->iCol>=nCol ){
    if( sqlite3Fts5IterEof(pCsr->pIter) ){
      pCsr->bEof = 1;
    }else{
      const char *zTerm;
      int nTerm;

      zTerm = sqlite3Fts5IterTerm(pCsr->pIter, &nTerm);

      sqlite3Fts5BufferSet(&rc, &pCsr->term, nTerm, (const u8*)zTerm);
      memset(pCsr->aCnt, 0, nCol * sizeof(i64));
      memset(pCsr->aDoc, 0, nCol * sizeof(i64));
      pCsr->iCol = 0;

      assert( pTab->eType==FTS5_VOCAB_COL || pTab->eType==FTS5_VOCAB_ROW );
      while( rc==SQLITE_OK ){
        int eDetail = pCsr->pConfig->eDetail;
        const u8 *pPos; int nPos;   /* Position list */
        i64 iPos = 0;               /* 64-bit position read from poslist */
        int iOff = 0;               /* Current offset within position list */

        pPos = pCsr->pIter->pData;
        nPos = pCsr->pIter->nData;

        switch( pTab->eType ){
          case FTS5_VOCAB_ROW:


            if( eDetail==FTS5_DETAIL_FULL ){
              while( 0==sqlite3Fts5PoslistNext64(pPos, nPos, &iOff, &iPos) ){
                pCsr->aCnt[0]++;
              }
            }
            pCsr->aDoc[0]++;
            break;

          case FTS5_VOCAB_COL:
            if( eDetail==FTS5_DETAIL_FULL ){
              int iCol = -1;
              while( 0==sqlite3Fts5PoslistNext64(pPos, nPos, &iOff, &iPos) ){
                int ii = FTS5_POS2COLUMN(iPos);
                pCsr->aCnt[ii]++;
                if( iCol!=ii ){
                  if( ii>=nCol ){
                    rc = FTS5_CORRUPT;
                    break;
                  }
                  pCsr->aDoc[ii]++;
                  iCol = ii;
                }
              }



            }else if( eDetail==FTS5_DETAIL_COLUMNS ){



              while( 0==sqlite3Fts5PoslistNext64(pPos, nPos, &iOff,&iPos) ){
                assert_nc( iPos>=0 && iPos<nCol );
                if( iPos>=nCol ){
                  rc = FTS5_CORRUPT;
                  break;
                }
                pCsr->aDoc[iPos]++;
              }
            }else{
              assert( eDetail==FTS5_DETAIL_NONE );
              pCsr->aDoc[0]++;
            }
            break;

          default:
            assert( pTab->eType==FTS5_VOCAB_INSTANCE );

            break;
        }

        if( rc==SQLITE_OK ){
          rc = sqlite3Fts5IterNextScan(pCsr->pIter);
        }
        if( pTab->eType==FTS5_VOCAB_INSTANCE ) break;

        if( rc==SQLITE_OK ){
          zTerm = sqlite3Fts5IterTerm(pCsr->pIter, &nTerm);
          if( nTerm!=pCsr->term.n || memcmp(zTerm, pCsr->term.p, nTerm) ){
            break;
          }
          if( sqlite3Fts5IterEof(pCsr->pIter) ) break;

static int fts5VocabFilterMethod(
  sqlite3_vtab_cursor *pCursor,   /* The cursor used for this query */
  int idxNum,                     /* Strategy index */
  const char *zUnused,            /* Unused */
  int nUnused,                    /* Number of elements in apVal */
  sqlite3_value **apVal           /* Arguments for the indexing scheme */
){
  Fts5VocabTable *pTab = (Fts5VocabTable*)pCursor->pVtab;
  Fts5VocabCursor *pCsr = (Fts5VocabCursor*)pCursor;
  int eType = pTab->eType;
  int rc = SQLITE_OK;

  int iVal = 0;
  int f = FTS5INDEX_QUERY_SCAN;
  const char *zTerm = 0;
  int nTerm = 0;

        rc = SQLITE_NOMEM;
      }else{
        memcpy(pCsr->zLeTerm, zCopy, pCsr->nLeTerm+1);
      }
    }
  }


  if( rc==SQLITE_OK ){
    rc = sqlite3Fts5IndexQuery(pCsr->pIndex, zTerm, nTerm, f, 0, &pCsr->pIter);
  }
  if( rc==SQLITE_OK && eType==FTS5_VOCAB_INSTANCE ){
    rc = fts5VocabInstanceNewTerm(pCsr);
  }
  if( rc==SQLITE_OK 
   && !pCsr->bEof 
   && (eType!=FTS5_VOCAB_INSTANCE || pCsr->pConfig->eDetail!=FTS5_DETAIL_NONE)
  ){
    rc = fts5VocabNextMethod(pCursor);
  }

  return rc;
}

/* 

        sqlite3_result_text(pCtx, z, -1, SQLITE_STATIC);
      }
    }else if( iCol==2 ){
      iVal = pCsr->aDoc[pCsr->iCol];
    }else{
      iVal = pCsr->aCnt[pCsr->iCol];
    }
  }else if( eType==FTS5_VOCAB_ROW ){
    assert( iCol==1 || iCol==2 );
    if( iCol==1 ){
      iVal = pCsr->aDoc[0];
    }else{
      iVal = pCsr->aCnt[0];
    }
  }else{
    int eDetail = pCsr->pConfig->eDetail;
    assert( eType==FTS5_VOCAB_INSTANCE );
    switch( iCol ){
      case 1:
        sqlite3_result_int64(pCtx, pCsr->pIter->iRowid);
        break;
      case 2: {
        int ii = -1;
        if( eDetail==FTS5_DETAIL_FULL ){
          ii = FTS5_POS2COLUMN(pCsr->iInstPos);
        }else if( eDetail==FTS5_DETAIL_COLUMNS ){
          ii = pCsr->iInstPos;
        }
        if( ii>=0 && ii<pCsr->pConfig->nCol ){
          const char *z = pCsr->pConfig->azCol[ii];
          sqlite3_result_text(pCtx, z, -1, SQLITE_STATIC);
        }
        break;
      }
      default: {
        assert( iCol==3 );
        if( eDetail==FTS5_DETAIL_FULL ){
          int ii = FTS5_POS2OFFSET(pCsr->iInstPos);
          sqlite3_result_int(pCtx, ii);
        }
        break;
      }
    }
  }

  if( iVal>0 ) sqlite3_result_int64(pCtx, iVal);
  return SQLITE_OK;
}