Fix a bug in merging FTS language tables for languages other than language 0.

FossilOrigin-Name: d281cb8984c911a4c0cce2ec299e1351d8e580e4

diff --git a/ext/fts3/fts3_write.c b/ext/fts3/fts3_write.c
index ef2436531cd227e5250235a8c085f860cfd8e3a2..b158cc451024382dfa1c258a6257720db5e2c435 100644 (file)
--- a/ext/fts3/fts3_write.c
+++ b/ext/fts3/fts3_write.c
@@ -488,7 +488,7 @@ int sqlite3Fts3AllSegdirs(
     /* "SELECT * FROM %_segdir WHERE level = ? ORDER BY ..." */
     rc = fts3SqlStmt(p, SQL_SELECT_LEVEL, &pStmt, 0);
     if( rc==SQLITE_OK ){ 
-      sqlite3_bind_int(pStmt, 1, iLevel+iIndex*FTS3_SEGDIR_MAXLEVEL);
+      sqlite3_bind_int(pStmt, 1, getAbsoluteLevel(p, iLangid, iIndex, iLevel));
     }
   }
   *ppStmt = pStmt;

diff --git a/manifest b/manifest
index c4b93f56a01b1f4f7105d26f429cb4e55c8d06a1..e051362a62164432dd564ee620352e910b1b4363 100644 (file)
--- a/manifest
+++ b/manifest
@@ -1,5 +1,5 @@
-C Add\sthe\s"languageid="\soption\sto\sfts4.\sThis\scode\sis\sstill\slargely\suntested\sand\salsmost\scertainly\sbuggy.
-D 2012-03-01T19:44:20.362
+C Fix\sa\sbug\sin\smerging\sFTS\slanguage\stables\sfor\slanguages\sother\sthan\slanguage\s0.
+D 2012-03-02T11:48:50.564
 F Makefile.arm-wince-mingw32ce-gcc d6df77f1f48d690bd73162294bbba7f59507c72f
 F Makefile.in 3f79a373e57c3b92dabf76f40b065e719d31ac34
 F Makefile.linux-gcc 91d710bdc4998cb015f39edf3cb314ec4f4d7e23
@@ -78,7 +78,7 @@ F ext/fts3/fts3_test.c 24fa13f330db011500acb95590da9eee24951894
 F ext/fts3/fts3_tokenizer.c 9ff7ec66ae3c5c0340fa081958e64f395c71a106
 F ext/fts3/fts3_tokenizer.h 13ffd9fcb397fec32a05ef5cd9e0fa659bf3dbd3
 F ext/fts3/fts3_tokenizer1.c 0dde8f307b8045565cf63797ba9acfaff1c50c68
-F ext/fts3/fts3_write.c 489d262b1ee9ab1dbb4da48bd8737fac15d0f58f
+F ext/fts3/fts3_write.c 36fc2e3a28f51ee135a344877c1e4be0a9f45e6e
 F ext/fts3/fts3speed.tcl b54caf6a18d38174f1a6e84219950d85e98bb1e9
 F ext/fts3/mkfts3amal.tcl 252ecb7fe6467854f2aa237bf2c390b74e71f100
 F ext/icu/README.txt bf8461d8cdc6b8f514c080e4e10dc3b2bbdfefa9
@@ -496,7 +496,7 @@ F test/fts3snippet.test 8e956051221a34c7daeb504f023cb54d5fa5a8b2
 F test/fts3sort.test 95be0b19d7e41c44b29014f13ea8bddd495fd659
 F test/fts4aa.test 6e7f90420b837b2c685f3bcbe84c868492d40a68
 F test/fts4content.test 17b2360f7d1a9a7e5aa8022783f5c5731b6dfd4f
-F test/fts4langid.test 3d968b7c0afb8be1609794267f34b89d378a81ea
+F test/fts4langid.test 7ab7be619d3acb3727e4bef3230ba3dbcf2e0556
 F test/func.test 6c5ce11e3a0021ca3c0649234e2d4454c89110ca
 F test/func2.test 772d66227e4e6684b86053302e2d74a2500e1e0f
 F test/func3.test 001021e5b88bd02a3b365a5c5fd8f6f49d39744a
@@ -992,10 +992,7 @@ F tool/tostr.awk e75472c2f98dd76e06b8c9c1367f4ab07e122d06
 F tool/vdbe-compress.tcl d70ea6d8a19e3571d7ab8c9b75cba86d1173ff0f
 F tool/warnings-clang.sh 9f406d66e750e8ac031c63a9ef3248aaa347ef2a
 F tool/warnings.sh fbc018d67fd7395f440c28f33ef0f94420226381
-P 16330a2f7262173a32ae48a72c0ee2522b6dc554
-R 86036df8ba11902f17395620671e5794
-T *branch * fts4-languageid
-T *sym-fts4-languageid *
-T -sym-trunk *
+P bea257f70f10dd1111d79cabd1e1462dc651704d
+R a3a9247d2c76c9d90f9fc486f3311f0d
 U dan
-Z 6902c01b6e8a000d5e06f8fe8778490f
+Z f1e998b56e58f712fe6da1411961b8ef

diff --git a/manifest.uuid b/manifest.uuid
index d2201717b8e9ccb1b4cb9dee738a46c8fddb99e4..ce8a1c3b82ff02b6e0f868a0628723d13b881771 100644 (file)
--- a/manifest.uuid
+++ b/manifest.uuid
@@ -1 +1 @@
-bea257f70f10dd1111d79cabd1e1462dc651704d
\ No newline at end of file
+d281cb8984c911a4c0cce2ec299e1351d8e580e4
\ No newline at end of file

diff --git a/test/fts4langid.test b/test/fts4langid.test
index a65ccedac311a19eb60aafbc155d9fb620f3f705..51c42ddb54c909af8e58a950cf579e8bab1c77c4 100644 (file)
--- a/test/fts4langid.test
+++ b/test/fts4langid.test
@@ -24,6 +24,27 @@ ifcapable !fts3 {
 
 set ::testprefix fts4langid
 
+#---------------------------------------------------------------------------
+# Test plan:
+#
+#   1.* - Warm-body tests created for specific purposes during development.
+#         Passing these doesn't really prove much.
+#
+#   2.* - Test that FTS queries only ever return rows associated with
+#         the requested language.
+#
+#   3.* - Test that the 'optimize' and 'rebuild' commands work correctly.
+#
+#   4.* - Test that if one is provided, the tokenizer xLanguage method
+#         is called to configure the tokenizer before tokenizing query
+#         or document text.
+#
+#   5.* - Test the fts4aux table when the associated FTS4 table contains
+#         multiple languages.
+#
+#   6.* - Tests with content= tables. Both where there is a real 
+#         underlying content table and where there is not.
+#
 
 
 do_execsql_test 1.1 {
@@ -74,5 +95,165 @@ do_catchsql_test 1.17 {
   INSERT INTO t1(content, lang_id) VALUES('123', -1);
 } {1 {constraint failed}}
 
+do_execsql_test 1.18 {
+  DROP TABLE t1;
+  CREATE VIRTUAL TABLE t1 USING fts4(languageid=lang_id);
+  INSERT INTO t1(content, lang_id) VALUES('A', 13);
+  INSERT INTO t1(content, lang_id) VALUES('B', 13);
+  INSERT INTO t1(content, lang_id) VALUES('C', 13);
+  INSERT INTO t1(content, lang_id) VALUES('D', 13);
+  INSERT INTO t1(content, lang_id) VALUES('E', 13);
+  INSERT INTO t1(content, lang_id) VALUES('F', 13);
+  INSERT INTO t1(content, lang_id) VALUES('G', 13);
+  INSERT INTO t1(content, lang_id) VALUES('H', 13);
+  INSERT INTO t1(content, lang_id) VALUES('I', 13);
+  INSERT INTO t1(content, lang_id) VALUES('J', 13);
+  INSERT INTO t1(content, lang_id) VALUES('K', 13);
+  INSERT INTO t1(content, lang_id) VALUES('L', 13);
+  INSERT INTO t1(content, lang_id) VALUES('M', 13);
+  INSERT INTO t1(content, lang_id) VALUES('N', 13);
+  INSERT INTO t1(content, lang_id) VALUES('O', 13);
+  INSERT INTO t1(content, lang_id) VALUES('P', 13);
+  INSERT INTO t1(content, lang_id) VALUES('Q', 13);
+  INSERT INTO t1(content, lang_id) VALUES('R', 13);
+  INSERT INTO t1(content, lang_id) VALUES('S', 13);
+  SELECT rowid FROM t1 WHERE t1 MATCH 'A';
+} {}
+
+
+#-------------------------------------------------------------------------
+# Test cases 2.*
+#
+
+proc build_multilingual_db_1 {db} {
+  $db eval { CREATE VIRTUAL TABLE t2 USING fts4(x, y, languageid=l) }
+
+  set xwords [list zero one two three four five six seven eight nine ten]
+  set ywords [list alpha beta gamma delta epsilon zeta eta theta iota kappa]
+
+  for {set i 0} {$i < 1000} {incr i} {
+    set iLangid [expr $i%9]
+    set x ""
+    set y ""
+
+    set x [list]
+    lappend x [lindex $xwords [expr ($i / 1000) % 10]]
+    lappend x [lindex $xwords [expr ($i / 100)  % 10]]
+    lappend x [lindex $xwords [expr ($i / 10)   % 10]]
+    lappend x [lindex $xwords [expr ($i / 1)   % 10]]
+
+    set y [list]
+    lappend y [lindex $ywords [expr ($i / 1000) % 10]]
+    lappend y [lindex $ywords [expr ($i / 100)  % 10]]
+    lappend y [lindex $ywords [expr ($i / 10)   % 10]]
+    lappend y [lindex $ywords [expr ($i / 1)   % 10]]
+
+    $db eval { INSERT INTO t2(docid, x, y, l) VALUES($i, $x, $y, $iLangid) }
+  }
+}
+
+proc rowid_list_set_langid {langid} {
+  set ::rowid_list_langid $langid
+}
+proc rowid_list {pattern} {
+  set langid $::rowid_list_langid
+  set res [list]
+  db eval {SELECT docid, x, y FROM t2 WHERE l = $langid ORDER BY docid ASC} {
+    if {[string match "*$pattern*" $x] || [string match "*$pattern*" $y]} {
+      lappend res $docid
+    }
+  }
+  return $res
+}
+
+proc or_merge_list {list1 list2} {
+  set res [list]
+
+  set i1 0
+  set i2 0
+
+  set n1 [llength $list1]
+  set n2 [llength $list2]
+
+  while {$i1 < $n1 && $i2 < $n2} {
+    set e1 [lindex $list1 $i1]
+    set e2 [lindex $list2 $i2]
+
+    if {$e1==$e2} {
+      lappend res $e1
+      incr i1
+      incr i2
+    } elseif {$e1 < $e2} {
+      lappend res $e1
+      incr i1
+    } else {
+      lappend res $e2
+      incr i2
+    }
+  }
+
+  concat $res [lrange $list1 $i1 end] [lrange $list2 $i2 end]
+}
+
+proc or_merge_lists {args} {
+  set res [lindex $args 0]
+  for {set i 1} {$i < [llength $args]} {incr i} {
+    set res [or_merge_list $res [lindex $args $i]]
+  }
+  set res
+}
+
+proc and_merge_list {list1 list2} {
+  foreach i $list2 { set a($i) 1 }
+  set res [list]
+  foreach i $list1 {
+    if {[info exists a($i)]} {lappend res $i}
+  }
+  set res
+}
+
+
+proc and_merge_lists {args} {
+  set res [lindex $args 0]
+  for {set i 1} {$i < [llength $args]} {incr i} {
+    set res [and_merge_list $res [lindex $args $i]]
+  }
+  set res
+}
+
+proc filter_list {list langid} {
+  set res [list]
+  foreach i $list {
+    if {($i % 9) == $langid} {lappend res $i}
+  }
+  set res
+}
+
+do_test 2.0 { 
+  reset_db
+  build_multilingual_db_1 db
+} {}
+
+proc do_test_2.1 {tn query res_script} {
+  for {set langid 0} {$langid < 10} {incr langid} {
+    rowid_list_set_langid $langid
+    set res [eval $res_script]
+
+    set actual [
+      execsql {SELECT docid FROM t2 WHERE t2 MATCH $query AND l = $langid}
+    ]
+    do_test 2.1.$tn.$langid [list set {} $actual] $res
+  }
+}
+
+do_test_2.1 1  {delta}          { rowid_list delta }
+do_test_2.1 2  {"zero one two"} { rowid_list "zero one two" }
+do_test_2.1 3  {zero one two} {
+  and_merge_lists [rowid_list zero] [rowid_list one] [rowid_list two]
+}
+do_test_2.1 4  {"zero one" OR "one two"} {
+  or_merge_lists [rowid_list "zero one"] [rowid_list "one two"]
+}
+
 finish_test