paulhorn/sqlite
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Fix problems with combining content= and languageid= in a single fts4 table.

Browse Source 
FossilOrigin-Name: 22491e7bc38aee43819b888e04241cb6a6ef73a3
This commit is contained in:
dan
2012-03-02 19:53:02 +00:00
parent c0db214b10
commit 7395599031
4 changed files with 106 additions and 32 deletions
Download Patch File Download Diff File Expand all files Collapse all files
test/fts4langid.test
+77 -22
View File
@@ -37,17 +37,17 @@ set ::testprefix fts4langid
#
# 2.3.* - Same as 2.1.*, after a 'rebuild' command.
#
# 3.* - Test that if one is provided, the tokenizer xLanguage method
# 3.* - Tests with content= tables. Both where there is a real
# underlying content table and where there is not.
#
#
# 4.* - Test that if one is provided, the tokenizer xLanguage method
# is called to configure the tokenizer before tokenizing query
# or document text.
#
# 4.* - Test the fts4aux table when the associated FTS4 table contains
# 5.* - Test the fts4aux table when the associated FTS4 table contains
# multiple languages.
#
# 5.* - Tests with content= tables. Both where there is a real
# underlying content table and where there is not.
#
do_execsql_test 1.1 {
CREATE VIRTUAL TABLE t1 USING fts4(a, b, languageid=lang_id);
@@ -151,6 +151,11 @@ proc build_multilingual_db_1 {db} {
$db eval { INSERT INTO t2(docid, x, y, l) VALUES($i, $x, $y, $iLangid) }
}
$db eval {
CREATE TABLE data(x, y, l);
INSERT INTO data(rowid, x, y, l) SELECT docid, x, y, l FROM t2;
}
}
proc rowid_list_set_langid {langid} {
@@ -159,9 +164,9 @@ proc rowid_list_set_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} {
db eval {SELECT rowid, x, y FROM data WHERE l = $langid ORDER BY rowid ASC} {
if {[string match "*$pattern*" $x] || [string match "*$pattern*" $y]} {
lappend res $docid
lappend res $rowid
}
}
return $res
@@ -235,7 +240,7 @@ do_test 2.0 {
build_multilingual_db_1 db
} {}
proc do_test_2 {tn query res_script} {
proc do_test_query1 {tn query res_script} {
for {set langid 0} {$langid < 10} {incr langid} {
rowid_list_set_langid $langid
set res [eval $res_script]
@@ -243,17 +248,17 @@ proc do_test_2 {tn query res_script} {
set actual [
execsql {SELECT docid FROM t2 WHERE t2 MATCH $query AND l = $langid}
]
do_test 2.$tn.$langid [list set {} $actual] $res
do_test $tn.$langid [list set {} $actual] $res
}
}
# Run some queries.
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} {
do_test_query1 2.1.1 {delta} { rowid_list delta }
do_test_query1 2.1.2 {"zero one two"} { rowid_list "zero one two" }
do_test_query1 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"} {
do_test_query1 2.1.4 {"zero one" OR "one two"} {
or_merge_lists [rowid_list "zero one"] [rowid_list "one two"]
}
@@ -264,12 +269,12 @@ do_execsql_test 2.2 {
INSERT INTO t2(t2) VALUES('optimize');
SELECT count(*) FROM t2_segdir;
} {9}
do_test_2 2.1 {delta} { rowid_list delta }
do_test_2 2.2 {"zero one two"} { rowid_list "zero one two" }
do_test_2 2.3 {zero one two} {
do_test_query1 2.2.1 {delta} { rowid_list delta }
do_test_query1 2.2.2 {"zero one two"} { rowid_list "zero one two" }
do_test_query1 2.2.3 {zero one two} {
and_merge_lists [rowid_list zero] [rowid_list one] [rowid_list two]
}
do_test_2 2.4 {"zero one" OR "one two"} {
do_test_query1 2.2.4 {"zero one" OR "one two"} {
or_merge_lists [rowid_list "zero one"] [rowid_list "one two"]
}
@@ -280,14 +285,64 @@ do_test 2.3 {
build_multilingual_db_1 db
execsql { INSERT INTO t2(t2) VALUES('rebuild') }
} {}
do_test_2 3.1 {delta} { rowid_list delta }
do_test_2 3.2 {"zero one two"} { rowid_list "zero one two" }
do_test_2 3.3 {zero one two} {
do_test_query1 2.3.1 {delta} { rowid_list delta }
do_test_query1 2.3.2 {"zero one two"} { rowid_list "zero one two" }
do_test_query1 2.3.3 {zero one two} {
and_merge_lists [rowid_list zero] [rowid_list one] [rowid_list two]
}
do_test_2 3.4 {"zero one" OR "one two"} {
do_test_query1 2.3.4 {"zero one" OR "one two"} {
or_merge_lists [rowid_list "zero one"] [rowid_list "one two"]
}
#-------------------------------------------------------------------------
# Test cases 3.*
#
do_test 3.0 {
reset_db
build_multilingual_db_1 db
execsql {
CREATE TABLE t3_data(l, x, y);
INSERT INTO t3_data(rowid, l, x, y) SELECT docid, l, x, y FROM t2;
DROP TABLE t2;
}
} {}
do_execsql_test 3.1 {
CREATE VIRTUAL TABLE t2 USING fts4(content=t3_data, languageid=l);
INSERT INTO t2(t2) VALUES('rebuild');
}
do_test_query1 3.1.1 {delta} { rowid_list delta }
do_test_query1 3.1.2 {"zero one two"} { rowid_list "zero one two" }
do_test_query1 3.1.3 {zero one two} {
and_merge_lists [rowid_list zero] [rowid_list one] [rowid_list two]
}
do_test_query1 3.1.4 {"zero one" OR "one two"} {
or_merge_lists [rowid_list "zero one"] [rowid_list "one two"]
}
do_execsql_test 3.2.1 {
DROP TABLE t2;
CREATE VIRTUAL TABLE t2 USING fts4(x, y, languageid=l, content=nosuchtable);
}
do_execsql_test 3.2.2 {
INSERT INTO t2(docid, x, y, l) SELECT rowid, x, y, l FROM t3_data;
}
do_execsql_test 3.2.3 {
DROP TABLE t3_data;
}
do_test_query1 3.3.1 {delta} { rowid_list delta }
do_test_query1 3.3.2 {"zero one two"} { rowid_list "zero one two" }
do_test_query1 3.3.3 {zero one two} {
and_merge_lists [rowid_list zero] [rowid_list one] [rowid_list two]
}
do_test_query1 3.3.4 {"zero one" OR "one two"} {
or_merge_lists [rowid_list "zero one"] [rowid_list "one two"]
}
finish_test