Many hyperlinks are disabled.
Use anonymous login to enable hyperlinks.

Overview
Comment:More work on updating an fts5 index. Add a pragma that checks if the index and table contents match.
Downloads: Tarball | ZIP archive
Timelines: family | ancestors | descendants | both | trunk
Files: files | file ages | folders
SHA1: 4693eb7bcc6714f4bd44d03af7f736e98999ad9d
User & Date: dan 2012-12-22 19:59:46.127
Context
2012-12-24
15:32
Fixes for updates and deletes on tables with fts5 indexes. check-in: 8161b13910 user: dan tags: trunk
2012-12-22
19:59
More work on updating an fts5 index. Add a pragma that checks if the index and table contents match. check-in: 4693eb7bcc user: dan tags: trunk
2012-12-21
19:58
Add update function to fts5.c. check-in: 49eff5d82d user: dan tags: trunk
Changes
Unified Diff Ignore Whitespace Patch
Changes to src/fts5.c.
741
742
743
744
745
746
747





748






749

750
751
752
753
754
755
756
757
758
759
760
761
762
static void fts5TokenizerCreate(
  Parse *pParse, 
  Fts5Index *pFts, 
  Fts5Tokenizer **ppTokenizer,
  sqlite4_tokenizer **pp
){
  Fts5Tokenizer *pTok;





  char *zTok = pFts->azTokenizer[0];








  *ppTokenizer = pTok = fts5FindTokenizer(pParse->db, zTok);
  if( !pTok ){
    sqlite4ErrorMsg(pParse, "no such tokenizer: \"%s\"", zTok);
  }else{
    const char **azArg = (const char **)&pFts->azTokenizer[1];
    int rc = pTok->xCreate(pTok->pCtx, azArg, pFts->nTokenizer-1, pp);
    if( rc!=SQLITE4_OK ){
      assert( *pp==0 );
      sqlite4ErrorMsg(pParse, "error creating tokenizer");
    }
  }
}








>
>
>
>
>
|
>
>
>
>
>
>
|
>




<
|







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
static void fts5TokenizerCreate(
  Parse *pParse, 
  Fts5Index *pFts, 
  Fts5Tokenizer **ppTokenizer,
  sqlite4_tokenizer **pp
){
  Fts5Tokenizer *pTok;
  char *zTok;                     /* Tokenizer name */
  const char **azArg;             /* Tokenizer arguments */
  int nArg;                       /* Number of elements in azArg */

  if( pFts->nTokenizer ){
    zTok = pFts->azTokenizer[0];
    azArg = (const char **)&pFts->azTokenizer[1];
    nArg = pFts->nTokenizer-1;
  }else{
    zTok = "simple";
    azArg = 0;
    nArg = 0;
  }
 
  *ppTokenizer = pTok = fts5FindTokenizer(pParse->db, zTok);
  if( !pTok ){
    sqlite4ErrorMsg(pParse, "no such tokenizer: \"%s\"", zTok);
  }else{

    int rc = pTok->xCreate(pTok->pCtx, azArg, nArg, pp);
    if( rc!=SQLITE4_OK ){
      assert( *pp==0 );
      sqlite4ErrorMsg(pParse, "error creating tokenizer");
    }
  }
}

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
}


/*
** Context structure passed to tokenizer callback when tokenizing a document.
**
** The hash table maps between tokens and TokenizeTerm structures.









*/
typedef struct TokenizeCtx TokenizeCtx;
typedef struct TokenizeTerm TokenizeTerm;
struct TokenizeCtx {
  int rc;
  int iCol;
  sqlite4 *db;
  int nMax;
  Hash hash;
};
struct TokenizeTerm {
  int iWeight;                    /* Weight of previous entry */
  int iCol;                       /* Column containing previous entry */
  int iOff;                       /* Token offset of previous entry */

  int nData;                      /* Bytes of data in value */
  int nAlloc;                     /* Bytes of data allocated */
};

TokenizeTerm *fts5TokenizeAppendInt(
  TokenizeCtx *p, 
  TokenizeTerm *pTerm, 
  int iVal
){
  unsigned char *a;

  if( (pTerm->nAlloc - pTerm->nData) < 5 ){

    int nAlloc = (pTerm->nAlloc<256) ? 256 : pTerm->nAlloc * 2;
    pTerm = sqlite4DbReallocOrFree(p->db, pTerm, nAlloc);
    if( !pTerm ) return 0;
    pTerm->nAlloc = nAlloc - sizeof(TokenizeTerm);
  }

  a = &(((unsigned char *)&pTerm[1])[pTerm->nData]);
  pTerm->nData += sqlite4PutVarint32(a, iVal);
  return pTerm;
}


static int fts5TokenizeCb(
  void *pCtx, 
  int iWeight, 
  int iOff,
  const char *zToken, 
  int nToken, 
  int iSrc, 
  int nSrc
){
  TokenizeCtx *p = (TokenizeCtx *)pCtx;
  TokenizeTerm *pTerm = 0;
  TokenizeTerm *pOrig = 0;

  if( nToken>p->nMax ) p->nMax = nToken;

  pTerm = (TokenizeTerm *)sqlite4HashFind(&p->hash, zToken, nToken);
  if( pTerm==0 ){

    const int nAlloc = 100-sizeof(TokenizeTerm);

    pTerm = sqlite4DbMallocZero(p->db, sizeof(TokenizeTerm) + nAlloc);
    if( pTerm ){
      void *pFree;
      pTerm->nAlloc = nAlloc;


      pFree = sqlite4HashInsert(&p->hash, zToken, nToken, pTerm);
      if( pFree ){
        sqlite4DbFree(p->db, pFree);
        pTerm = 0;
      }
      if( pTerm==0 ) goto tokenize_cb_out;
    }
  }







>
>
>
>
>
>
>
>
>














>










>
|
>



|


|
|


<


















>
|
>
|


|
>
>
|







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
}


/*
** Context structure passed to tokenizer callback when tokenizing a document.
**
** The hash table maps between tokens and TokenizeTerm structures.
**
** TokenizeTerm structures are allocated using sqlite4DbMalloc(). Immediately
** following the structure in memory is the token itself (TokenizeTerm.nToken
** bytes of data). Following this is the list of token instances in the same
** format as it is stored in the database. 
**
** All of the above is a single allocation, size TokenizeTerm.nAlloc bytes.
** If the initial allocation is too small, it is extended using
** sqlite4DbRealloc().
*/
typedef struct TokenizeCtx TokenizeCtx;
typedef struct TokenizeTerm TokenizeTerm;
struct TokenizeCtx {
  int rc;
  int iCol;
  sqlite4 *db;
  int nMax;
  Hash hash;
};
struct TokenizeTerm {
  int iWeight;                    /* Weight of previous entry */
  int iCol;                       /* Column containing previous entry */
  int iOff;                       /* Token offset of previous entry */
  int nToken;                     /* Size of token in bytes */
  int nData;                      /* Bytes of data in value */
  int nAlloc;                     /* Bytes of data allocated */
};

TokenizeTerm *fts5TokenizeAppendInt(
  TokenizeCtx *p, 
  TokenizeTerm *pTerm, 
  int iVal
){
  unsigned char *a;
  int nSpace = pTerm->nAlloc-pTerm->nData-pTerm->nToken-sizeof(TokenizeTerm);

  if( nSpace < 5 ){
    int nAlloc = (pTerm->nAlloc<256) ? 256 : pTerm->nAlloc * 2;
    pTerm = sqlite4DbReallocOrFree(p->db, pTerm, nAlloc);
    if( !pTerm ) return 0;
    pTerm->nAlloc = sqlite4DbMallocSize(p->db, pTerm);
  }

  a = &(((unsigned char *)&pTerm[1])[pTerm->nToken+pTerm->nData]);
  pTerm->nData += putVarint32(a, iVal);
  return pTerm;
}


static int fts5TokenizeCb(
  void *pCtx, 
  int iWeight, 
  int iOff,
  const char *zToken, 
  int nToken, 
  int iSrc, 
  int nSrc
){
  TokenizeCtx *p = (TokenizeCtx *)pCtx;
  TokenizeTerm *pTerm = 0;
  TokenizeTerm *pOrig = 0;

  if( nToken>p->nMax ) p->nMax = nToken;

  pTerm = (TokenizeTerm *)sqlite4HashFind(&p->hash, zToken, nToken);
  if( pTerm==0 ){
    /* Size the initial allocation so that it fits in the lookaside buffer */
    int nAlloc = sizeof(TokenizeTerm) + nToken + 32;

    pTerm = sqlite4DbMallocZero(p->db, nAlloc);
    if( pTerm ){
      void *pFree;
      pTerm->nAlloc = sqlite4DbMallocSize(p->db, pTerm);
      pTerm->nToken = nToken;
      memcpy(&pTerm[1], zToken, nToken);
      pFree = sqlite4HashInsert(&p->hash, (char *)&pTerm[1], nToken, pTerm);
      if( pFree ){
        sqlite4DbFree(p->db, pFree);
        pTerm = 0;
      }
      if( pTerm==0 ) goto tokenize_cb_out;
    }
  }
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

  pTerm = fts5TokenizeAppendInt(p, pTerm, (iOff-pTerm->iOff) << 1);
  if( !pTerm ) goto tokenize_cb_out;
  pTerm->iOff = iOff;

tokenize_cb_out:
  if( pTerm!=pOrig ){
    sqlite4HashInsert(&p->hash, zToken, nToken, 0);
  }
  if( !pTerm ){
    p->rc = SQLITE4_NOMEM;
    return 1;
  }

  return 0;
}

/*
** Update an fts index.
*/
int sqlite4Fts5Update(
  sqlite4 *db,                    /* Database handle */
  Fts5Info *pInfo,                /* Description of fts index to update */

  Mem *aArg,                      /* Array of arguments (see above) */
  int bDel,                       /* True for a delete, false for insert */
  char **pzErr                    /* OUT: Error message */
){
  int i;
  int rc = SQLITE4_OK;
  KVStore *pStore;
  TokenizeCtx sCtx;
  u8 *aKey = 0;
  int nKey = 0;



  const void *pPK;
  int nPK;
  HashElem *pElem;

  pStore = db->aDb[pInfo->iDb].pKV;
  sCtx.rc = SQLITE4_OK;
  sCtx.db = db;
  sCtx.nMax = 0;
  sqlite4HashInit(db->pEnv, &sCtx.hash);

  pPK = sqlite4_value_blob(&aArg[0]);
  nPK = sqlite4_value_bytes(&aArg[0]);





  for(i=0; rc==SQLITE4_OK && i<pInfo->nCol; i++){
    sqlite4_value *pArg = (sqlite4_value *)(&aArg[i+1]);
    if( sqlite4_value_type(pArg)==SQLITE4_TEXT ){
      const char *zText;
      int nText;

      zText = (const char *)sqlite4_value_text(pArg);
      nText = sqlite4_value_bytes(pArg);
      sCtx.iCol = i;
      rc = pInfo->pTokenizer->xTokenize(







|















>










>
>

|









|
|
|
>
>
>
>

|
|







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

  pTerm = fts5TokenizeAppendInt(p, pTerm, (iOff-pTerm->iOff) << 1);
  if( !pTerm ) goto tokenize_cb_out;
  pTerm->iOff = iOff;

tokenize_cb_out:
  if( pTerm!=pOrig ){
    sqlite4HashInsert(&p->hash, (char *)&pTerm[1], nToken, pTerm);
  }
  if( !pTerm ){
    p->rc = SQLITE4_NOMEM;
    return 1;
  }

  return 0;
}

/*
** Update an fts index.
*/
int sqlite4Fts5Update(
  sqlite4 *db,                    /* Database handle */
  Fts5Info *pInfo,                /* Description of fts index to update */
  Mem *pKey,                      /* Primary key blob */
  Mem *aArg,                      /* Array of arguments (see above) */
  int bDel,                       /* True for a delete, false for insert */
  char **pzErr                    /* OUT: Error message */
){
  int i;
  int rc = SQLITE4_OK;
  KVStore *pStore;
  TokenizeCtx sCtx;
  u8 *aKey = 0;
  int nKey = 0;
  int nTnum = 0;
  u32 dummy = 0;

  const u8 *pPK;
  int nPK;
  HashElem *pElem;

  pStore = db->aDb[pInfo->iDb].pKV;
  sCtx.rc = SQLITE4_OK;
  sCtx.db = db;
  sCtx.nMax = 0;
  sqlite4HashInit(db->pEnv, &sCtx.hash);

  pPK = (const u8 *)sqlite4_value_blob(pKey);
  nPK = sqlite4_value_bytes(pKey);
  
  nTnum = getVarint32(pPK, dummy);
  nPK -= nTnum;
  pPK += nTnum;

  for(i=0; rc==SQLITE4_OK && i<pInfo->nCol; i++){
    sqlite4_value *pArg = (sqlite4_value *)(&aArg[i]);
    if( pArg->flags & MEM_Str ){
      const char *zText;
      int nText;

      zText = (const char *)sqlite4_value_text(pArg);
      nText = sqlite4_value_bytes(pArg);
      sCtx.iCol = i;
      rc = pInfo->pTokenizer->xTokenize(
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

  for(pElem=sqliteHashFirst(&sCtx.hash); pElem; pElem=sqliteHashNext(pElem)){
    TokenizeTerm *pTerm = (TokenizeTerm *)sqliteHashData(pElem);
    if( rc==SQLITE4_OK ){
      int nToken = sqliteHashKeysize(pElem);
      char *zToken = (char *)sqliteHashKey(pElem);

      nKey = sqlite4PutVarint32(aKey, pInfo->iRoot);
      aKey[nKey++] = 0x24;
      memcpy(&aKey[nKey], zToken, nToken);
      nKey += nToken;
      aKey[nKey++] = 0x00;
      memcpy(&aKey[nKey], pPK, nPK);
      nKey += nPK;

      if( bDel ){
        /* delete key aKey/nKey... */
        assert( 0 );
      }else{
        const KVByteArray *aData = (const KVByteArray *)&pTerm[1];

        rc = sqlite4KVStoreReplace(pStore, aKey, nKey, aData, pTerm->nData);
      }
    }
    sqlite4DbFree(db, pTerm);
  }
  
 fts5_update_out:
  sqlite4DbFree(db, aKey);
  sqlite4HashClear(&sCtx.hash);
  return rc;
}




































































































































































































































/**************************************************************************
***************************************************************************
** Below this point is test code.
*/
#ifdef SQLITE4_TEST
static int fts5PrintExprNode(sqlite4 *, const char **, Fts5ExprNode *, char **);







|












>






<




>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>







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

  for(pElem=sqliteHashFirst(&sCtx.hash); pElem; pElem=sqliteHashNext(pElem)){
    TokenizeTerm *pTerm = (TokenizeTerm *)sqliteHashData(pElem);
    if( rc==SQLITE4_OK ){
      int nToken = sqliteHashKeysize(pElem);
      char *zToken = (char *)sqliteHashKey(pElem);

      nKey = putVarint32(aKey, pInfo->iRoot);
      aKey[nKey++] = 0x24;
      memcpy(&aKey[nKey], zToken, nToken);
      nKey += nToken;
      aKey[nKey++] = 0x00;
      memcpy(&aKey[nKey], pPK, nPK);
      nKey += nPK;

      if( bDel ){
        /* delete key aKey/nKey... */
        assert( 0 );
      }else{
        const KVByteArray *aData = (const KVByteArray *)&pTerm[1];
        aData += pTerm->nToken;
        rc = sqlite4KVStoreReplace(pStore, aKey, nKey, aData, pTerm->nData);
      }
    }
    sqlite4DbFree(db, pTerm);
  }
  

  sqlite4DbFree(db, aKey);
  sqlite4HashClear(&sCtx.hash);
  return rc;
}

static Fts5Info *fts5InfoCreate(Parse *pParse, Index *pIdx){
  sqlite4 *db = pParse->db;
  Fts5Info *pInfo;                /* p4 argument for FtsUpdate opcode */

  pInfo = sqlite4DbMallocZero(db, sizeof(Fts5Info));
  if( pInfo ){
    pInfo->iDb = sqlite4SchemaToIndex(db, pIdx->pSchema);
    pInfo->iRoot = pIdx->tnum;
    pInfo->nCol = pIdx->pTable->nCol;
    fts5TokenizerCreate(pParse, pIdx->pFts, &pInfo->pTokenizer, &pInfo->p);

    if( pInfo->p==0 ){
      assert( pParse->nErr );
      sqlite4DbFree(db, pInfo);
      pInfo = 0;
    }
  }

  return pInfo;
}


void sqlite4Fts5CodeUpdate(
  Parse *pParse, 
  Index *pIdx, 
  int iRegPk, 
  int iRegData
){
  Vdbe *v;
  Fts5Info *pInfo;                /* p4 argument for FtsUpdate opcode */

  if( 0==(pInfo = fts5InfoCreate(pParse, pIdx)) ) return;

  v = sqlite4GetVdbe(pParse);
  sqlite4VdbeAddOp3(v, OP_FtsUpdate, iRegPk, 0, iRegData);
  sqlite4VdbeChangeP4(v, -1, (const char *)pInfo, P4_FTS5INFO);
}

void sqlite4Fts5FreeInfo(sqlite4 *db, Fts5Info *p){
  if( db->pnBytesFreed==0 ){
    if( p->p ) p->pTokenizer->xDestroy(p->p);
    sqlite4DbFree(db, p);
  }
}

void sqlite4Fts5CodeCksum(
  Parse *pParse, 
  Index *pIdx, 
  int iCksum, 
  int iReg,
  int bIdx                        /* True for fts index, false for table */
){
  Vdbe *v;
  Fts5Info *pInfo;                /* p4 argument for FtsCksum opcode */

  if( 0==(pInfo = fts5InfoCreate(pParse, pIdx)) ) return;

  v = sqlite4GetVdbe(pParse);
  sqlite4VdbeAddOp3(v, OP_FtsCksum, iCksum, 0, iReg);
  sqlite4VdbeChangeP4(v, -1, (const char *)pInfo, P4_FTS5INFO);
  sqlite4VdbeChangeP5(v, bIdx);
}

/*
** Calculate a 64-bit checksum for a term instance. The index checksum is
** the XOR of the checksum for each term instance in the table. A term
** instance checksum is calculated based on:
**
**   * the term itself,
**   * the pk of the row the instance appears in,
**   * the weight assigned to the instance,
**   * the column number, and
**   * the term offset.
*/
static i64 fts5TermInstanceCksum(
  const u8 *aTerm, int nTerm,
  const u8 *aPk, int nPk,
  int iWeight,
  int iCol,
  int iOff
){
  int i;
  i64 cksum = 0;

  /* Add the term to the checksum */
  for(i=0; i<nTerm; i++){
    cksum += (cksum << 3) + aTerm[i];
  }

  /* Add the primary key blob to the checksum */
  for(i=0; i<nPk; i++){
    cksum += (cksum << 3) + aPk[i];
  }

  /* Add the weight, column number and offset (in that order) to the checksum */
  cksum += (cksum << 3) + iWeight;
  cksum += (cksum << 3) + iCol;
  cksum += (cksum << 3) + iOff;

  return cksum;
}


int sqlite4Fts5EntryCksum(
  sqlite4 *db,                    /* Database handle */
  Fts5Info *p,                    /* Index description */
  Mem *pKey,                      /* Database key */
  Mem *pVal,                      /* Database value */
  i64 *piCksum                    /* OUT: Checksum value */
){
  i64 cksum = 0;
  u8 const *aKey; int nKey;       /* Key blob */
  u8 const *aVal; int nVal;       /* List of token instances */
  u8 const *aToken; int nToken;   /* Token for this entry */
  u8 const *aPk; int nPk;         /* Entry primary key blob */
  int nTnum;
  u32 tnum;

  int iOff = 0;
  int iCol = 0;
  int iWeight = 0;
  int i = 0;

  aKey = (const u8 *)sqlite4_value_blob(pKey);
  nKey = sqlite4_value_bytes(pKey);
  aVal = (const u8 *)sqlite4_value_blob(pVal);
  nVal = sqlite4_value_bytes(pVal);

  /* Find the token and primary key blobs for this entry. */
  nTnum = getVarint32(aKey, tnum);
  aToken = &aKey[nTnum+1];
  nToken = sqlite4Strlen30((const char *)aToken);
  aPk = &aToken[nToken+1];
  nPk = (&aKey[nKey] - aPk);

  while( i<nVal ){
    u32 iVal;
    i += getVarint32(&aVal[i], iVal);

    if( (iVal & 0x03)==0x01 ){
      iCol = (iVal>>2);
      iOff = 0;
    }
    else if( (iVal & 0x03)==0x03 ){
      iWeight = (iVal>>2);
    }
    else{
      i64 v;
      iOff += (iVal>>1);
      v = fts5TermInstanceCksum(aPk, nPk, aToken, nToken, iWeight, iCol, iOff);
      cksum = cksum ^ v;
    }
  }

  *piCksum = cksum;
  return SQLITE4_OK;
}

typedef struct CksumCtx CksumCtx;
struct CksumCtx {
  const u8 *pPK;
  int nPK;
  int iCol;
  i64 cksum;
};

static int fts5CksumCb(
  void *pCtx, 
  int iWeight, 
  int iOff,
  const char *zToken, 
  int nToken, 
  int iSrc, 
  int nSrc
){
  CksumCtx *p = (CksumCtx *)pCtx;
  i64 cksum;

  cksum = fts5TermInstanceCksum(p->pPK, p->nPK, 
      (const u8 *)zToken, nToken, iWeight, p->iCol, iOff
  );

  p->cksum = (p->cksum ^ cksum);
  return 0;
}

int sqlite4Fts5RowCksum(
  sqlite4 *db,                    /* Database handle */
  Fts5Info *pInfo,                /* Index description */
  Mem *pKey,                      /* Primary key blob */
  Mem *aArg,                      /* Array of column values */
  i64 *piCksum                    /* OUT: Checksum value */
){
  int i;
  int rc = SQLITE4_OK;
  CksumCtx sCtx;
  int nTnum = 0;
  u32 dummy = 0;

  sCtx.cksum = 0;

  sCtx.pPK = (const u8 *)sqlite4_value_blob(pKey);
  sCtx.nPK = sqlite4_value_bytes(pKey);
  nTnum = getVarint32(sCtx.pPK, dummy);
  sCtx.nPK -= nTnum;
  sCtx.pPK += nTnum;

  for(i=0; rc==SQLITE4_OK && i<pInfo->nCol; i++){
    sqlite4_value *pArg = (sqlite4_value *)(&aArg[i]);
    if( pArg->flags & MEM_Str ){
      const char *zText;
      int nText;

      zText = (const char *)sqlite4_value_text(pArg);
      nText = sqlite4_value_bytes(pArg);
      sCtx.iCol = i;
      rc = pInfo->pTokenizer->xTokenize(
          &sCtx, pInfo->p, zText, nText, fts5CksumCb
      );
    }
  }

  *piCksum = sCtx.cksum;
  return rc;
}


/**************************************************************************
***************************************************************************
** Below this point is test code.
*/
#ifdef SQLITE4_TEST
static int fts5PrintExprNode(sqlite4 *, const char **, Fts5ExprNode *, char **);
Changes to src/fts5func.c.
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
}

static int fts5SimpleTokenize(
  void *pCtx, 
  sqlite4_tokenizer *p,
  const char *zDoc,
  int nDoc,
  int(*x)(void*, int, const char*, int, int, int)
){
  sqlite4_env *pEnv = (sqlite4_env *)p;
  char *aBuf;
  int nBuf;
  int iBuf;
  int i;
  int brk = 0;


  nBuf = 128;
  aBuf = (char *)sqlite4_malloc(pEnv, nBuf);
  if( !aBuf ) return SQLITE4_NOMEM;

  iBuf = 0;
  for(i=0; brk==0 && i<nDoc; i++){
    if( sqlite4Isalnum(zDoc[i]) ){
      aBuf[iBuf++] = fts5Tolower(zDoc[i]);
    }else if( iBuf>0 ){
      brk = x(pCtx, 0, aBuf, iBuf, i-iBuf, iBuf);
      iBuf = 0;
    }
  }
  if( iBuf>0 ) x(pCtx, 0, aBuf, iBuf, i-iBuf, iBuf);

  sqlite4_free(pEnv, aBuf);
  return SQLITE4_OK;
}

int sqlite4InitFts5Func(sqlite4 *db){
  int rc;







|







>










|



|







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
}

static int fts5SimpleTokenize(
  void *pCtx, 
  sqlite4_tokenizer *p,
  const char *zDoc,
  int nDoc,
  int(*x)(void*, int, int, const char*, int, int, int)
){
  sqlite4_env *pEnv = (sqlite4_env *)p;
  char *aBuf;
  int nBuf;
  int iBuf;
  int i;
  int brk = 0;
  int iOff = 0;

  nBuf = 128;
  aBuf = (char *)sqlite4_malloc(pEnv, nBuf);
  if( !aBuf ) return SQLITE4_NOMEM;

  iBuf = 0;
  for(i=0; brk==0 && i<nDoc; i++){
    if( sqlite4Isalnum(zDoc[i]) ){
      aBuf[iBuf++] = fts5Tolower(zDoc[i]);
    }else if( iBuf>0 ){
      brk = x(pCtx, 0, iOff++, aBuf, iBuf, i-iBuf, iBuf);
      iBuf = 0;
    }
  }
  if( iBuf>0 ) x(pCtx, 0, iOff++, aBuf, iBuf, i-iBuf, iBuf);

  sqlite4_free(pEnv, aBuf);
  return SQLITE4_OK;
}

int sqlite4InitFts5Func(sqlite4 *db){
  int rc;
Changes to src/insert.c.
1415
1416
1417
1418
1419
1420
1421





1422
1423
1424
1425
1426
1427
1428
1429
  sqlite4VdbeAddOp3(v, OP_MakeRecord, regContent, pTab->nCol, regRec);
  sqlite4TableAffinityStr(v, pTab);
  sqlite4ExprCacheAffinityChange(pParse, regContent, pTab->nCol);

  /* Write the entry to each index. */
  for(i=0, pIdx=pTab->pIndex; pIdx; i++, pIdx=pIdx->pNext){
    assert( pIdx->eIndexType!=SQLITE4_INDEX_PRIMARYKEY || aRegIdx[i] );





    if( aRegIdx[i] ){
      int regData = 0;
      int flags = 0;
      if( pIdx->eIndexType==SQLITE4_INDEX_PRIMARYKEY ){
        regData = regRec;
        flags = pik_flags;
      }
      sqlite4VdbeAddOp3(v, OP_IdxInsert, baseCur+i, regData, aRegIdx[i]);







>
>
>
>
>
|







1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
  sqlite4VdbeAddOp3(v, OP_MakeRecord, regContent, pTab->nCol, regRec);
  sqlite4TableAffinityStr(v, pTab);
  sqlite4ExprCacheAffinityChange(pParse, regContent, pTab->nCol);

  /* Write the entry to each index. */
  for(i=0, pIdx=pTab->pIndex; pIdx; i++, pIdx=pIdx->pNext){
    assert( pIdx->eIndexType!=SQLITE4_INDEX_PRIMARYKEY || aRegIdx[i] );
    if( pIdx->eIndexType==SQLITE4_INDEX_FTS5 ){
      int iPK;
      sqlite4FindPrimaryKey(pTab, &iPK);
      sqlite4Fts5CodeUpdate(pParse, pIdx, aRegIdx[iPK], regContent);
    }
    else if( aRegIdx[i] ){
      int regData = 0;
      int flags = 0;
      if( pIdx->eIndexType==SQLITE4_INDEX_PRIMARYKEY ){
        regData = regRec;
        flags = pik_flags;
      }
      sqlite4VdbeAddOp3(v, OP_IdxInsert, baseCur+i, regData, aRegIdx[i]);
Changes to src/pragma.c.
234
235
236
237
238
239
240































































241
242
243
244
245
246
247

  if( sqlite4StrICmp(zLeft, "lsm_merge")==0 ){
    int nPage = zRight ? sqlite4Atoi(zRight) : 1000;
    sqlite4_kvstore_control(db, zDb, SQLITE4_KVCTRL_LSM_MERGE, &nPage);
    returnSingleInt(pParse, "nWrite", (sqlite4_int64)nPage);
  }else



































































#ifndef SQLITE4_OMIT_SCHEMA_PRAGMAS
  /*
  **   PRAGMA table_info(<table>)
  **







>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>







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

  if( sqlite4StrICmp(zLeft, "lsm_merge")==0 ){
    int nPage = zRight ? sqlite4Atoi(zRight) : 1000;
    sqlite4_kvstore_control(db, zDb, SQLITE4_KVCTRL_LSM_MERGE, &nPage);
    returnSingleInt(pParse, "nWrite", (sqlite4_int64)nPage);
  }else

  /*
  **   PRAGMA fts_check(<index>)
  */
  if( sqlite4StrICmp(zLeft, "fts_check")==0 && zRight ){
    int iCksum1;
    int iCksum2;
    Index *pIdx;
    Table *pTab;
    Vdbe *v = sqlite4GetVdbe(pParse);
    if( v==0 || sqlite4ReadSchema(pParse) ) goto pragma_out;

    iCksum1 = ++pParse->nMem;
    iCksum2 = ++pParse->nMem;
    sqlite4VdbeAddOp2(v, OP_Integer, 0, iCksum1);
    sqlite4VdbeAddOp2(v, OP_Integer, 0, iCksum2);

    pIdx = sqlite4FindIndex(db, zRight, zDb);
    if( pIdx && pIdx->eIndexType==SQLITE4_INDEX_FTS5 ){
      int iTab = pParse->nTab++;
      int iAddr;
      int iReg;
      int i;

      pTab = pIdx->pTable;
      sqlite4OpenPrimaryKey(pParse, iTab, iDb, pTab, OP_OpenRead);
      iAddr = sqlite4VdbeAddOp2(v, OP_Rewind, iTab, 0);

      iReg = pParse->nMem+1;
      pParse->nMem += (1 + pTab->nCol);

      sqlite4VdbeAddOp2(v, OP_RowKey, iTab, iReg);
      for(i=0; i<pTab->nCol; i++){
        sqlite4VdbeAddOp3(v, OP_Column, iTab, i, iReg+1+i);
      }
      sqlite4Fts5CodeCksum(pParse, pIdx, iCksum1, iReg, 0);

      sqlite4VdbeAddOp2(v, OP_Next, iTab, iAddr+1);
      sqlite4VdbeJumpHere(v, iAddr);
      sqlite4VdbeAddOp1(v, OP_Close, iTab);

      sqlite4VdbeAddOp3(v, OP_OpenRead, iTab, pIdx->tnum, iDb);
      iAddr = sqlite4VdbeAddOp2(v, OP_Rewind, iTab, 0);

      iReg = pParse->nMem+1;
      pParse->nMem += 2;
      sqlite4VdbeAddOp2(v, OP_RowKey, iTab, iReg);
      sqlite4VdbeAddOp2(v, OP_RowData, iTab, iReg+1);
      sqlite4Fts5CodeCksum(pParse, pIdx, iCksum2, iReg, 1);

      sqlite4VdbeAddOp2(v, OP_Next, iTab, iAddr+1);
      sqlite4VdbeJumpHere(v, iAddr);
      sqlite4VdbeAddOp1(v, OP_Close, iTab);

      iReg = ++pParse->nMem;
      sqlite4VdbeAddOp4(v, OP_String8, 0, iReg, 0, "ok", 0);
      iAddr = sqlite4VdbeAddOp3(v, OP_Eq, iCksum1, 0, iCksum2);
      sqlite4VdbeAddOp4(v, OP_String8, 0, iReg, 0, "error - cksum mismatch", 0);
      sqlite4VdbeJumpHere(v, iAddr);
      sqlite4VdbeAddOp2(v, OP_ResultRow, iReg, 1);

      sqlite4VdbeSetNumCols(v, 1);
    }
  }



#ifndef SQLITE4_OMIT_SCHEMA_PRAGMAS
  /*
  **   PRAGMA table_info(<table>)
  **
Changes to src/sqliteInt.h.
3265
3266
3267
3268
3269
3270
3271
3272



3273
3274
void sqlite4ShutdownFts5(sqlite4 *db);
void sqlite4CreateUsingIndex(Parse*, CreateIndex*, ExprList*, Token*, Token*);

int sqlite4Fts5IndexSz(void);
void sqlite4Fts5IndexInit(Parse *, Index *, ExprList *);
void sqlite4Fts5IndexFree(sqlite4 *, Index *);

int sqlite4Fts5Update(sqlite4 *, Fts5Info *, Mem *aArg, int, char **);




#endif /* _SQLITEINT_H_ */







|
>
>
>


3265
3266
3267
3268
3269
3270
3271
3272
3273
3274
3275
3276
3277
void sqlite4ShutdownFts5(sqlite4 *db);
void sqlite4CreateUsingIndex(Parse*, CreateIndex*, ExprList*, Token*, Token*);

int sqlite4Fts5IndexSz(void);
void sqlite4Fts5IndexInit(Parse *, Index *, ExprList *);
void sqlite4Fts5IndexFree(sqlite4 *, Index *);

int sqlite4Fts5Update(sqlite4 *, Fts5Info *, Mem *pPk, Mem *aArg, int, char **);
void sqlite4Fts5FreeInfo(sqlite4 *db, Fts5Info *);
void sqlite4Fts5CodeUpdate(Parse *, Index *pIdx, int iRegPk, int iRegData);
void sqlite4Fts5CodeCksum(Parse *, Index *, int, int, int);

#endif /* _SQLITEINT_H_ */
Changes to src/vdbe.c.
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
    sqlite4DebugPrintf("SQL-trace: %s\n", zTrace);
  }
#endif /* SQLITE4_DEBUG */
  break;
}
#endif

/* Opcode: FtsUpdate * * P3 P4 P5
**
** This opcode is used to write to an FTS index. P4 points to an Fts5Info 
** object describing the index.
**
** If argument P5 is non-zero, then entries are removed from the FTS index.
** If it is zero, then entries are inserted. In other words, when a row
** is deleted from a table with an FTS index, this opcode is invoked with
** P5==1. When a row is inserted, it is invoked with P5==0. If an existing
** row is updated, this opcode is invoked twice - once with P5==1 and then
** again with P5==0.
**

** P3 is the first in an array of N+1 registers, where N is the number of
** columns in the indexed table. The first register in the array contains
** the PK (a blob in key-format) of the affected row. The following N
** registers contain the column values.
*/
case OP_FtsUpdate: {
  Fts5Info *pInfo;                /* Description of fts5 index to update */

  Mem *aArg;                      /* Pointer to array of N+1 arguments */

  assert( pOp->p4type==P4_FTS5INFO );
  pInfo = pOp->p4.pFtsInfo;
  aArg = &aMem[pOp->p3];


  rc = sqlite4Fts5Update(db, pInfo, aArg, pOp->p5, &p->zErrMsg);



























  break;
}


/* Opcode: Noop * * * * *
**
** Do nothing.  This instruction is often useful as a jump







|











>
|
|
<
|



>
|




>

|
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>







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
    sqlite4DebugPrintf("SQL-trace: %s\n", zTrace);
  }
#endif /* SQLITE4_DEBUG */
  break;
}
#endif

/* Opcode: FtsUpdate P1 * P3 P4 P5
**
** This opcode is used to write to an FTS index. P4 points to an Fts5Info 
** object describing the index.
**
** If argument P5 is non-zero, then entries are removed from the FTS index.
** If it is zero, then entries are inserted. In other words, when a row
** is deleted from a table with an FTS index, this opcode is invoked with
** P5==1. When a row is inserted, it is invoked with P5==0. If an existing
** row is updated, this opcode is invoked twice - once with P5==1 and then
** again with P5==0.
**
** Register P1 contains the PK (a blob in key format) of the affected row.
** P3 is the first in an array of N registers, where N is the number of
** columns in the indexed table. Each register contains the value for the

** corresponding table column.
*/
case OP_FtsUpdate: {
  Fts5Info *pInfo;                /* Description of fts5 index to update */
  Mem *pKey;                      /* Primary key of indexed row */
  Mem *aArg;                      /* Pointer to array of N arguments */

  assert( pOp->p4type==P4_FTS5INFO );
  pInfo = pOp->p4.pFtsInfo;
  aArg = &aMem[pOp->p3];
  pKey = &aMem[pOp->p1];

  rc = sqlite4Fts5Update(db, pInfo, pKey, aArg, pOp->p5, &p->zErrMsg);
  break;
}

/*
** Opcode: FtsCksum P1 * P3 P4 P5
*/
case OP_FtsCksum: {
  Fts5Info *pInfo;                /* Description of fts5 index to update */
  Mem *pKey;                      /* Primary key of row */
  Mem *aArg;                      /* Pointer to array of N values */
  i64 cksum;

  assert( pOp->p4type==P4_FTS5INFO );
  pInfo = pOp->p4.pFtsInfo;

  pOut = &aMem[pOp->p1];
  pKey = &aMem[pOp->p3];
  aArg = &aMem[pOp->p3+1];
  cksum = 0;

  if( pOp->p5 ){
    sqlite4Fts5EntryCksum(db, pInfo, pKey, aArg, &cksum);
    pOut->u.i = pOut->u.i ^ cksum;
  }else{
    sqlite4Fts5RowCksum(db, pInfo, pKey, aArg, &cksum);
    pOut->u.i = pOut->u.i ^ cksum;
  }
  break;
}


/* Opcode: Noop * * * * *
**
** Do nothing.  This instruction is often useful as a jump
Changes to src/vdbeaux.c.
623
624
625
626
627
628
629




630
631
632
633
634
635
636
          sqlite4DbFree(db, p);
        }
        break;
      }
      case P4_VTAB : {
        if( db->pnBytesFreed==0 ) sqlite4VtabUnlock((VTable *)p4);
        break;




      }
    }
  }
}

/*
** Free the space allocated for aOp and any p4 values allocated for the







>
>
>
>







623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
          sqlite4DbFree(db, p);
        }
        break;
      }
      case P4_VTAB : {
        if( db->pnBytesFreed==0 ) sqlite4VtabUnlock((VTable *)p4);
        break;
      }
      case P4_FTS5INFO : {
        sqlite4Fts5FreeInfo(db, (Fts5Info *)p4);
        break;
      }
    }
  }
}

/*
** Free the space allocated for aOp and any p4 values allocated for the