|
1 # 2007 March 28 |
|
2 # |
|
3 # The author disclaims copyright to this source code. |
|
4 # |
|
5 #************************************************************************* |
|
6 # This file implements regression tests for SQLite library. The focus |
|
7 # of this script is testing isspace/isalnum/tolower problems with the |
|
8 # FTS1 module. Unfortunately, this code isn't a really principled set |
|
9 # of tests, because it is impossible to know where new uses of these |
|
10 # functions might appear. |
|
11 # |
|
12 # $Id: fts1k.test,v 1.2 2007/12/13 21:54:11 drh Exp $ |
|
13 # |
|
14 |
|
15 set testdir [file dirname $argv0] |
|
16 source $testdir/tester.tcl |
|
17 |
|
18 # If SQLITE_ENABLE_FTS1 is defined, omit this file. |
|
19 ifcapable !fts1 { |
|
20 finish_test |
|
21 return |
|
22 } |
|
23 |
|
24 # Tests that startsWith() (calls isspace, tolower, isalnum) can handle |
|
25 # hi-bit chars. parseSpec() also calls isalnum here. |
|
26 do_test fts1k-1.1 { |
|
27 execsql "CREATE VIRTUAL TABLE t1 USING fts1(content, \x80)" |
|
28 } {} |
|
29 |
|
30 # Additionally tests isspace() call in getToken(), and isalnum() call |
|
31 # in tokenListToIdList(). |
|
32 do_test fts1k-1.2 { |
|
33 catch { |
|
34 execsql "CREATE VIRTUAL TABLE t2 USING fts1(content, tokenize \x80)" |
|
35 } |
|
36 sqlite3_errmsg $DB |
|
37 } "unknown tokenizer: \x80" |
|
38 |
|
39 # Additionally test final isalnum() in startsWith(). |
|
40 do_test fts1k-1.3 { |
|
41 execsql "CREATE VIRTUAL TABLE t3 USING fts1(content, tokenize\x80)" |
|
42 } {} |
|
43 |
|
44 # The snippet-generation code has calls to isspace() which are sort of |
|
45 # hard to get to. It finds convenient breakpoints by starting ~40 |
|
46 # chars before and after the matched term, and scanning ~10 chars |
|
47 # around that position for isspace() characters. The long word with |
|
48 # embedded hi-bit chars causes one of these isspace() calls to be |
|
49 # exercised. The version with a couple extra spaces should cause the |
|
50 # other isspace() call to be exercised. [Both cases have been tested |
|
51 # in the debugger, but I'm hoping to continue to catch it if simple |
|
52 # constant changes change things slightly. |
|
53 # |
|
54 # The trailing and leading hi-bit chars help with code which tests for |
|
55 # isspace() to coalesce multiple spaces. |
|
56 |
|
57 set word "\x80xxxxx\x80xxxxx\x80xxxxx\x80xxxxx\x80xxxxx\x80xxxxx\x80" |
|
58 set phrase1 "$word $word $word target $word $word $word" |
|
59 set phrase2 "$word $word $word target $word $word $word" |
|
60 |
|
61 db eval {CREATE VIRTUAL TABLE t4 USING fts1(content)} |
|
62 db eval "INSERT INTO t4 (content) VALUES ('$phrase1')" |
|
63 db eval "INSERT INTO t4 (content) VALUES ('$phrase2')" |
|
64 |
|
65 do_test fts1k-1.4 { |
|
66 execsql {SELECT rowid, length(snippet(t4)) FROM t4 WHERE t4 MATCH 'target'} |
|
67 } {1 111 2 117} |
|
68 |
|
69 finish_test |