Fix some problems in fts3 found by address-sanitizer.

FossilOrigin-Name: 16a8e84fa7f67a467f824bdd7f72cbd6a6e95dab8cc7aa1e0e751720b98f3e31

ext/fts3/fts3.c

@@ -349,8 +349,9 @@ int sqlite3Fts3PutVarint(char *p, sqlite_int64 v){
 ** Return the number of bytes read, or 0 on error.
 ** The value is stored in *v.
 */
-int sqlite3Fts3GetVarint(const char *p, sqlite_int64 *v){
-  const char *pStart = p;
+int sqlite3Fts3GetVarint(const char *pBuf, sqlite_int64 *v){
+  const unsigned char *p = (const unsigned char*)pBuf;
+  const unsigned char *pStart = p;
   u32 a;
   u64 b;
   int shift;

ext/fts3/fts3_unicode.c

@@ -136,16 +136,16 @@ static int unicodeAddExceptions(
 ){
   const unsigned char *z = (const unsigned char *)zIn;
   const unsigned char *zTerm = &z[nIn];
-  int iCode;
+  unsigned int iCode;
   int nEntry = 0;
 
   assert( bAlnum==0 || bAlnum==1 );
 
   while( z<zTerm ){
     READ_UTF8(z, zTerm, iCode);
-    assert( (sqlite3FtsUnicodeIsalnum(iCode) & 0xFFFFFFFE)==0 );
-    if( sqlite3FtsUnicodeIsalnum(iCode)!=bAlnum 
-     && sqlite3FtsUnicodeIsdiacritic(iCode)==0 
+    assert( (sqlite3FtsUnicodeIsalnum((int)iCode) & 0xFFFFFFFE)==0 );
+    if( sqlite3FtsUnicodeIsalnum((int)iCode)!=bAlnum 
+     && sqlite3FtsUnicodeIsdiacritic((int)iCode)==0 
     ){
       nEntry++;
     }
@@ -162,13 +162,13 @@ static int unicodeAddExceptions(
     z = (const unsigned char *)zIn;
     while( z<zTerm ){
       READ_UTF8(z, zTerm, iCode);
-      if( sqlite3FtsUnicodeIsalnum(iCode)!=bAlnum 
-       && sqlite3FtsUnicodeIsdiacritic(iCode)==0
+      if( sqlite3FtsUnicodeIsalnum((int)iCode)!=bAlnum 
+       && sqlite3FtsUnicodeIsdiacritic((int)iCode)==0
       ){
         int i, j;
-        for(i=0; i<nNew && aNew[i]<iCode; i++);
+        for(i=0; i<nNew && aNew[i]<(int)iCode; i++);
         for(j=nNew; j>i; j--) aNew[j] = aNew[j-1];
-        aNew[i] = iCode;
+        aNew[i] = (int)iCode;
         nNew++;
       }
     }
@@ -318,7 +318,7 @@ static int unicodeNext(
 ){
   unicode_cursor *pCsr = (unicode_cursor *)pC;
   unicode_tokenizer *p = ((unicode_tokenizer *)pCsr->base.pTokenizer);
-  int iCode = 0;
+  unsigned int iCode = 0;
   char *zOut;
   const unsigned char *z = &pCsr->aInput[pCsr->iOff];
   const unsigned char *zStart = z;
@@ -330,7 +330,7 @@ static int unicodeNext(
   ** the input.  */
   while( z<zTerm ){
     READ_UTF8(z, zTerm, iCode);
-    if( unicodeIsAlnum(p, iCode) ) break;
+    if( unicodeIsAlnum(p, (int)iCode) ) break;
     zStart = z;
   }
   if( zStart>=zTerm ) return SQLITE_DONE;
@@ -350,7 +350,7 @@ static int unicodeNext(
 
     /* Write the folded case of the last character read to the output */
     zEnd = z;
-    iOut = sqlite3FtsUnicodeFold(iCode, p->bRemoveDiacritic);
+    iOut = sqlite3FtsUnicodeFold((int)iCode, p->bRemoveDiacritic);
     if( iOut ){
       WRITE_UTF8(zOut, iOut);
     }
@@ -358,8 +358,8 @@ static int unicodeNext(
     /* If the cursor is not at EOF, read the next character */
     if( z>=zTerm ) break;
     READ_UTF8(z, zTerm, iCode);
-  }while( unicodeIsAlnum(p, iCode) 
-       || sqlite3FtsUnicodeIsdiacritic(iCode)
+  }while( unicodeIsAlnum(p, (int)iCode) 
+       || sqlite3FtsUnicodeIsdiacritic((int)iCode)
   );
 
   /* Set the output variables and return. */

ext/fts3/fts3_unicode2.c

@@ -127,9 +127,9 @@ int sqlite3FtsUnicodeIsalnum(int c){
     0xFFFFFFFF, 0xFC00FFFF, 0xF8000001, 0xF8000001,
   };
 
-  if( c<128 ){
-    return ( (aAscii[c >> 5] & (1 << (c & 0x001F)))==0 );
-  }else if( c<(1<<22) ){
+  if( (unsigned int)c<128 ){
+    return ( (aAscii[c >> 5] & ((unsigned int)1 << (c & 0x001F)))==0 );
+  }else if( (unsigned int)c<(1<<22) ){
     unsigned int key = (((unsigned int)c)<<10) | 0x000003FF;
     int iRes = 0;
     int iHi = sizeof(aEntry)/sizeof(aEntry[0]) - 1;
@@ -322,16 +322,17 @@ int sqlite3FtsUnicodeFold(int c, int bRemoveDiacritic){
 
   int ret = c;
 
-  assert( c>=0 );
   assert( sizeof(unsigned short)==2 && sizeof(unsigned char)==1 );
 
   if( c<128 ){
     if( c>='A' && c<='Z' ) ret = c + ('a' - 'A');
   }else if( c<65536 ){
+    const struct TableEntry *p;
     int iHi = sizeof(aEntry)/sizeof(aEntry[0]) - 1;
     int iLo = 0;
     int iRes = -1;
 
+    assert( c>aEntry[0].iCode );
     while( iHi>=iLo ){
       int iTest = (iHi + iLo) / 2;
       int cmp = (c - aEntry[iTest].iCode);
@@ -342,14 +343,12 @@ int sqlite3FtsUnicodeFold(int c, int bRemoveDiacritic){
         iHi = iTest-1;
       }
     }
-    assert( iRes<0 || c>=aEntry[iRes].iCode );
 
-    if( iRes>=0 ){
-      const struct TableEntry *p = &aEntry[iRes];
-      if( c<(p->iCode + p->nRange) && 0==(0x01 & p->flags & (p->iCode ^ c)) ){
-        ret = (c + (aiOff[p->flags>>1])) & 0x0000FFFF;
-        assert( ret>0 );
-      }
+    assert( iRes>=0 && c>=aEntry[iRes].iCode );
+    p = &aEntry[iRes];
+    if( c<(p->iCode + p->nRange) && 0==(0x01 & p->flags & (p->iCode ^ c)) ){
+      ret = (c + (aiOff[p->flags>>1])) & 0x0000FFFF;
+      assert( ret>0 );
     }
 
     if( bRemoveDiacritic ) ret = remove_diacritic(ret);

ext/fts3/fts3_write.c

@@ -4956,11 +4956,14 @@ int sqlite3Fts3Incrmerge(Fts3Table *p, int nMerge, int nMin){
 ** Convert the text beginning at *pz into an integer and return
 ** its value.  Advance *pz to point to the first character past
 ** the integer.
+**
+** This function used for parameters to merge= and incrmerge=
+** commands. 
 */
 static int fts3Getint(const char **pz){
   const char *z = *pz;
   int i = 0;
-  while( (*z)>='0' && (*z)<='9' ) i = 10*i + *(z++) - '0';
+  while( (*z)>='0' && (*z)<='9' && i<214748363 ) i = 10*i + *(z++) - '0';
   *pz = z;
   return i;
 }

ext/fts3/unicode/mkunicode.tcl

@@ -227,7 +227,7 @@ proc print_isalnum {zFunc lRange} {
   an_print_ascii_bitmap $lRange
   puts {
   if( (unsigned int)c<128 ){
-    return ( (aAscii[c >> 5] & (1 << (c & 0x001F)))==0 );
+    return ( (aAscii[c >> 5] & ((unsigned int)1 << (c & 0x001F)))==0 );
   }else if( (unsigned int)c<(1<<22) ){
     unsigned int key = (((unsigned int)c)<<10) | 0x000003FF;
     int iRes = 0;