Instead of the 4-byte fields, use regular varints for the poslist-size field in fts5_...

FossilOrigin-Name: 7eb022d7e5fdb180af823c82c47c938e4a7a355f

diff --git a/ext/fts5/fts5Int.h b/ext/fts5/fts5Int.h
index 034e6866dbbda6e7fd4452bac926b6fd33e3c566..90df81a01689db2c25f526bd54a61891062d5652 100644 (file)
--- a/ext/fts5/fts5Int.h
+++ b/ext/fts5/fts5Int.h
@@ -352,6 +352,8 @@ int sqlite3Fts5IndexLoadConfig(Fts5Index *p);
 int sqlite3Fts5GetVarint32(const unsigned char *p, u32 *v);
 #define fts5GetVarint32(a,b) sqlite3Fts5GetVarint32(a,(u32*)&b)
 
+int sqlite3Fts5GetVarintLen(u32 iVal);
+
 /*
 ** End of interface to code in fts5_index.c.
 **************************************************************************/

diff --git a/ext/fts5/fts5_hash.c b/ext/fts5/fts5_hash.c
index 841c4d370960e38d7dde78aa9efa92e261bc3db7..4b85b2af23eb4b60d2debe97934b58dddc1775a9 100644 (file)
--- a/ext/fts5/fts5_hash.c
+++ b/ext/fts5/fts5_hash.c
@@ -184,6 +184,23 @@ static int fts5HashResize(Fts5Hash *pHash){
   return SQLITE_OK;
 }
 
+static void fts5HashAddPoslistSize(Fts5HashEntry *p){
+  if( p->iSzPoslist ){
+    u8 *pPtr = (u8*)p;
+    int nSz = p->nData - p->iSzPoslist - 1;
+
+    if( nSz<=127 ){
+      pPtr[p->iSzPoslist] = nSz;
+    }else{
+      int nByte = sqlite3Fts5GetVarintLen((u32)nSz);
+      memmove(&pPtr[p->iSzPoslist + nByte], &pPtr[p->iSzPoslist + 1], nSz);
+      sqlite3PutVarint(&pPtr[p->iSzPoslist], nSz);
+      p->nData += (nByte-1);
+    }
+    p->iSzPoslist = 0;
+  }
+}
+
 int sqlite3Fts5HashWrite(
   Fts5Hash *pHash,
   i64 iRowid,                     /* Rowid for this entry */
@@ -221,7 +238,7 @@ int sqlite3Fts5HashWrite(
     p->nData = nToken + 1 + sizeof(Fts5HashEntry);
     p->nData += sqlite3PutVarint(&((u8*)p)[p->nData], iRowid);
     p->iSzPoslist = p->nData;
-    p->nData += 4;
+    p->nData += 1;
     p->iRowid = iRowid;
     p->pHashNext = pHash->aSlot[iHash];
     pHash->aSlot[iHash] = p;
@@ -233,7 +250,7 @@ int sqlite3Fts5HashWrite(
   ** is:
   **
   **     + 9 bytes for a new rowid,
-  **     + 4 bytes reserved for the "poslist size" varint.
+  **     + 4 byte reserved for the "poslist size" varint.
   **     + 1 byte for a "new column" byte,
   **     + 3 bytes for a new column number (16-bit max) as a varint,
   **     + 5 bytes for the new position offset (32-bit max).
@@ -255,11 +272,10 @@ int sqlite3Fts5HashWrite(
   /* If this is a new rowid, append the 4-byte size field for the previous
   ** entry, and the new rowid for this entry.  */
   if( iRowid!=p->iRowid ){
-    assert( p->iSzPoslist>0 );
-    fts5Put4ByteVarint(&pPtr[p->iSzPoslist], p->nData - p->iSzPoslist - 4);
+    fts5HashAddPoslistSize(p);
     p->nData += sqlite3PutVarint(&pPtr[p->nData], iRowid - p->iRowid);
     p->iSzPoslist = p->nData;
-    p->nData += 4;
+    p->nData += 1;
     p->iCol = 0;
     p->iPos = 0;
     p->iRowid = iRowid;
@@ -393,14 +409,13 @@ int sqlite3Fts5HashIterate(
     while( pList ){
       Fts5HashEntry *pNext = pList->pScanNext;
       if( rc==SQLITE_OK ){
-        const int nSz = pList->nData - pList->iSzPoslist - 4;
         const int nKey = strlen(pList->zKey);
         i64 iRowid = 0;
         u8 *pPtr = (u8*)pList;
         int iOff = sizeof(Fts5HashEntry) + nKey + 1;
 
         /* Fill in the final poslist size field */
-        fts5Put4ByteVarint(&pPtr[pList->iSzPoslist], nSz);
+        fts5HashAddPoslistSize(pList);
         
         /* Issue the new-term callback */
         rc = xTerm(pCtx, pList->zKey, nKey);
@@ -412,10 +427,9 @@ int sqlite3Fts5HashIterate(
           int nVarint;
           iOff += getVarint(&pPtr[iOff], (u64*)&iDelta);
           iRowid += iDelta;
-          nPoslist = fts5Get4ByteVarint(&pPtr[iOff], &nVarint);
-          iOff += 4;
-          rc = xEntry(pCtx, iRowid, &pPtr[iOff-nVarint], nPoslist+nVarint);
-          iOff += nPoslist;
+          nVarint = fts5GetVarint32(&pPtr[iOff], nPoslist);
+          rc = xEntry(pCtx, iRowid, &pPtr[iOff], nPoslist+nVarint);
+          iOff += nVarint+nPoslist;
         }
 
         /* Issue the term-done callback */
@@ -445,8 +459,7 @@ int sqlite3Fts5HashQuery(
   }
 
   if( p ){
-    u8 *pPtr = (u8*)p;
-    fts5Put4ByteVarint(&pPtr[p->iSzPoslist], p->nData - p->iSzPoslist - 4);
+    fts5HashAddPoslistSize(p);
     *ppDoclist = &p->zKey[nTerm+1];
     *pnDoclist = p->nData - (sizeof(*p) + nTerm + 1);
   }else{
@@ -482,9 +495,8 @@ void sqlite3Fts5HashScanEntry(
 ){
   Fts5HashEntry *p;
   if( (p = pHash->pScan) ){
-    u8 *pPtr = (u8*)p;
     int nTerm = strlen(p->zKey);
-    fts5Put4ByteVarint(&pPtr[p->iSzPoslist], p->nData - p->iSzPoslist - 4);
+    fts5HashAddPoslistSize(p);
     *pzTerm = p->zKey;
     *ppDoclist = &p->zKey[nTerm+1];
     *pnDoclist = p->nData - (sizeof(*p) + nTerm + 1);

diff --git a/ext/fts5/fts5_index.c b/ext/fts5/fts5_index.c
index 6916c4254efd3dd1ca7b78f902ecd75a714e2f7d..b90e5308d637c2f91f78c45ba3e2ba0fb2a0027d 100644 (file)
--- a/ext/fts5/fts5_index.c
+++ b/ext/fts5/fts5_index.c
@@ -371,7 +371,7 @@ struct Fts5Structure {
 struct Fts5PageWriter {
   int pgno;                       /* Page number for this page */
   Fts5Buffer buf;                 /* Buffer containing page data */
-  Fts5Buffer term;              /* Buffer containing previous term on page */
+  Fts5Buffer term;                /* Buffer containing previous term on page */
 };
 struct Fts5SegWriter {
   int iIdx;                       /* Index to write to */
@@ -667,6 +667,14 @@ int sqlite3Fts5GetVarint32(const unsigned char *p, u32 *v){
   }
 }
 
+int sqlite3Fts5GetVarintLen(u32 iVal){
+  if( iVal<(1 << 7 ) ) return 1;
+  if( iVal<(1 << 14) ) return 2;
+  if( iVal<(1 << 21) ) return 3;
+  if( iVal<(1 << 28) ) return 4;
+  return 5;
+}
+
 /*
 ** Allocate and return a buffer at least nByte bytes in size.
 **

diff --git a/ext/fts5/tool/loadfts5.tcl b/ext/fts5/tool/loadfts5.tcl
index 97353ffa3b3c2922f4850d4cf53e180338d03308..feb92ec1622d0a8d64c0f21acab8d1d168a0e3a3 100644 (file)
--- a/ext/fts5/tool/loadfts5.tcl
+++ b/ext/fts5/tool/loadfts5.tcl
@@ -36,6 +36,7 @@ proc usage {} {
   puts stderr "  -fts4        (use fts4 instead of fts5)"
   puts stderr "  -fts5        (use fts5)"
   puts stderr "  -porter      (use porter tokenizer)"
+  puts stderr "  -delete      (delete the database file before starting)"
   puts stderr "  -limit N     (load no more than N documents)"
   puts stderr "  -automerge N (set the automerge parameter to N)"
   puts stderr "  -crisismerge N (set the crisismerge parameter to N)"
@@ -45,6 +46,7 @@ proc usage {} {
 set O(vtab)       fts5
 set O(tok)        ""
 set O(limit)      0
+set O(delete)     0
 set O(automerge)  -1
 set O(crisismerge)  -1
 
@@ -65,6 +67,10 @@ for {set i 0} {$i < $nOpt} {incr i} {
       set O(tok) ", tokenize=porter"
     }
 
+    -delete {
+      set O(delete) 1
+    }
+
     -limit {
       if { [incr i]>=$nOpt } usage
       set O(limit) [lindex $argv $i]
@@ -86,7 +92,9 @@ for {set i 0} {$i < $nOpt} {incr i} {
   }
 }
 
-sqlite3 db [lindex $argv end-1]
+set dbfile [lindex $argv end-1]
+if {$O(delete)} { file delete -force $dbfile }
+sqlite3 db $dbfile
 db func loadfile loadfile
 
 db transaction {

diff --git a/manifest b/manifest
index 9aae583c5cd4c07c9ef2c6d4e1270f5c5d6e4b3c..e43204b80305636dc922675ff5207fe47adc0a99 100644 (file)
--- a/manifest
+++ b/manifest
@@ -1,5 +1,5 @@
-C Ensure\sgenerated\sheader\sfile\sfts5parse.h\sis\sincluded\sin\ssqlite3.c.
-D 2015-02-02T11:58:21.261
+C Instead\sof\sthe\s4-byte\sfields,\suse\sregular\svarints\sfor\sthe\sposlist-size\sfield\sin\sfts5_hash.c.
+D 2015-02-25T19:24:37.378
 F Makefile.arm-wince-mingw32ce-gcc d6df77f1f48d690bd73162294bbba7f59507c72f
 F Makefile.in 5407a688f4d77a05c18a8142be8ae5a2829dd610
 F Makefile.linux-gcc 91d710bdc4998cb015f39edf3cb314ec4f4d7e23
@@ -106,13 +106,13 @@ F ext/fts3/unicode/mkunicode.tcl 159c1194da0bc72f51b3c2eb71022568006dc5ad
 F ext/fts5/extract_api_docs.tcl 55a6d648d516f35d9a1e580ac00de27154e1904a
 F ext/fts5/fts5.c f2e899fba27ca33c8897635752c4c83a40dcb18d
 F ext/fts5/fts5.h f931954065693898d26c51f23f1d27200184a69a
-F ext/fts5/fts5Int.h 34040674eb25f3de8a0e57423a3155aef6312541
+F ext/fts5/fts5Int.h 7c2af493177b0e4e0290b869f19cd6d1d671d5ac
 F ext/fts5/fts5_aux.c 549aef152b0fd46020f5595d861b1fd60b3f9b4f
 F ext/fts5/fts5_buffer.c b92ba0eb67532d174934087f93716caf9a2168c7
 F ext/fts5/fts5_config.c e3421a76c2abd33a05ac09df0c97c64952d1e700
 F ext/fts5/fts5_expr.c eee52c9df84eade48eaa3f50c8876f44b552ff9b
-F ext/fts5/fts5_hash.c 57febfb06e59ae419ee9ba31667635f70d7c4dd0
-F ext/fts5/fts5_index.c beced7a9f360c2bf44a9f987c0a8735b6868ffbf
+F ext/fts5/fts5_hash.c 9032dd35bf8da6f9d4fc8c955c348dd6d229d8e4
+F ext/fts5/fts5_index.c 220321d06ae87496e22bb92fe52b65d84549ef0e
 F ext/fts5/fts5_storage.c f7c12c9f454b2a525827b3d85fd222789236f548
 F ext/fts5/fts5_tcl.c 1293fac2bb26903fd3d5cdee59c5885ba7e620d5
 F ext/fts5/fts5_tokenize.c 0d108148c26132448487926fe683425002aee369
@@ -147,7 +147,7 @@ F ext/fts5/test/fts5rowid.test a1b2a6d76648c734c1aab11ee1a619067e8d90e6
 F ext/fts5/test/fts5tokenizer.test b34ae592db66f6e89546d791ce1f905ba0b3395c
 F ext/fts5/test/fts5unicode.test 79b3e34eb29ce4929628aa514a40cb467fdabe4d
 F ext/fts5/test/fts5unicode2.test 64a5267fd6082fcb46439892ebd0cbaa5c38acee
-F ext/fts5/tool/loadfts5.tcl 310cb6556b74eda5ce7829a539298c3f35003523
+F ext/fts5/tool/loadfts5.tcl 1e126891d14ab85dcdb0fac7755a4cd5ba52e8b8
 F ext/icu/README.txt d9fbbad0c2f647c3fdf715fc9fd64af53aedfc43
 F ext/icu/icu.c d415ccf984defeb9df2c0e1afcfaa2f6dc05eacb
 F ext/icu/sqliteicu.h 728867a802baa5a96de7495e9689a8e01715ef37
@@ -1284,7 +1284,7 @@ F tool/vdbe_profile.tcl 67746953071a9f8f2f668b73fe899074e2c6d8c1
 F tool/warnings-clang.sh f6aa929dc20ef1f856af04a730772f59283631d4
 F tool/warnings.sh 0abfd78ceb09b7f7c27c688c8e3fe93268a13b32
 F tool/win/sqlite.vsix deb315d026cc8400325c5863eef847784a219a2f
-P fb10bbb9f9c4481e6043d323a3018a4ec68eb0ff
-R 5aa72f287ea58e03f6201c3554f79f11
+P bc7be2fcfd29d6f1b567b69b3b20896eceb99798
+R 79d1ace6d8bc7671e8fd6ed98da16ffb
 U dan
-Z faf1a2c366b2fc1b07a53967e5d081da
+Z a35e490814ed578ad02ebd09e2509be2

diff --git a/manifest.uuid b/manifest.uuid
index 419358cbacf556f79315ace1bf16322026479d4a..c4c9f580d3384f3f74230afa5d1af28e20a0cc4a 100644 (file)
--- a/manifest.uuid
+++ b/manifest.uuid
@@ -1 +1 @@
-bc7be2fcfd29d6f1b567b69b3b20896eceb99798
\ No newline at end of file
+7eb022d7e5fdb180af823c82c47c938e4a7a355f
\ No newline at end of file