sl@0: # 2007 March 28
sl@0: #
sl@0: # The author disclaims copyright to this source code.
sl@0: #
sl@0: #*************************************************************************
sl@0: # This file implements regression tests for SQLite library.  The focus
sl@0: # of this script is testing isspace/isalnum/tolower problems with the
sl@0: # FTS2 module.  Unfortunately, this code isn't a really principled set
sl@0: # of tests, because it is impossible to know where new uses of these
sl@0: # functions might appear.
sl@0: #
sl@0: # $Id: fts2l.test,v 1.2 2007/12/13 21:54:11 drh Exp $
sl@0: #
sl@0: 
sl@0: set testdir [file dirname $argv0]
sl@0: source $testdir/tester.tcl
sl@0: 
sl@0: # If SQLITE_ENABLE_FTS2 is defined, omit this file.
sl@0: ifcapable !fts2 {
sl@0:   finish_test
sl@0:   return
sl@0: }
sl@0: 
sl@0: # Tests that startsWith() (calls isspace, tolower, isalnum) can handle
sl@0: # hi-bit chars.  parseSpec() also calls isalnum here.
sl@0: do_test fts2l-1.1 {
sl@0:   execsql "CREATE VIRTUAL TABLE t1 USING fts2(content, \x80)"
sl@0: } {}
sl@0: 
sl@0: # Additionally tests isspace() call in getToken(), and isalnum() call
sl@0: # in tokenListToIdList().
sl@0: do_test fts2l-1.2 {
sl@0:   catch {
sl@0:     execsql "CREATE VIRTUAL TABLE t2 USING fts2(content, tokenize \x80)"
sl@0:   }
sl@0:   sqlite3_errmsg $DB
sl@0: } "unknown tokenizer: \x80"
sl@0: 
sl@0: # Additionally test final isalnum() in startsWith().
sl@0: do_test fts2l-1.3 {
sl@0:   execsql "CREATE VIRTUAL TABLE t3 USING fts2(content, tokenize\x80)"
sl@0: } {}
sl@0: 
sl@0: # The snippet-generation code has calls to isspace() which are sort of
sl@0: # hard to get to.  It finds convenient breakpoints by starting ~40
sl@0: # chars before and after the matched term, and scanning ~10 chars
sl@0: # around that position for isspace() characters.  The long word with
sl@0: # embedded hi-bit chars causes one of these isspace() calls to be
sl@0: # exercised.  The version with a couple extra spaces should cause the
sl@0: # other isspace() call to be exercised.  [Both cases have been tested
sl@0: # in the debugger, but I'm hoping to continue to catch it if simple
sl@0: # constant changes change things slightly.
sl@0: #
sl@0: # The trailing and leading hi-bit chars help with code which tests for
sl@0: # isspace() to coalesce multiple spaces.
sl@0: 
sl@0: set word "\x80xxxxx\x80xxxxx\x80xxxxx\x80xxxxx\x80xxxxx\x80xxxxx\x80"
sl@0: set phrase1 "$word $word $word target $word $word $word"
sl@0: set phrase2 "$word $word $word    target    $word $word $word"
sl@0: 
sl@0: db eval {CREATE VIRTUAL TABLE t4 USING fts2(content)}
sl@0: db eval "INSERT INTO t4 (content) VALUES ('$phrase1')"
sl@0: db eval "INSERT INTO t4 (content) VALUES ('$phrase2')"
sl@0: 
sl@0: do_test fts2l-1.4 {
sl@0:   execsql {SELECT rowid, length(snippet(t4)) FROM t4 WHERE t4 MATCH 'target'}
sl@0: } {1 111 2 117}
sl@0: 
sl@0: finish_test