diff options
| author | David M <davidmj@users.sourceforge.net> | 2007-07-19 20:38:08 +0000 |
|---|---|---|
| committer | David M <davidmj@users.sourceforge.net> | 2007-07-19 20:38:08 +0000 |
| commit | 870991c0608caf7467018245a0a4e1f1d55efd12 (patch) | |
| tree | 95a81e7c78ba914d9970e3e6de8dd1a82d31b534 /phpBB/develop/generate_utf_confusables.php | |
| parent | e9b908174cbeed0c780c56d6d74b761e40d8b594 (diff) | |
| download | forums-870991c0608caf7467018245a0a4e1f1d55efd12.tar forums-870991c0608caf7467018245a0a4e1f1d55efd12.tar.gz forums-870991c0608caf7467018245a0a4e1f1d55efd12.tar.bz2 forums-870991c0608caf7467018245a0a4e1f1d55efd12.tar.xz forums-870991c0608caf7467018245a0a4e1f1d55efd12.zip | |
let's remove 350+ impossible entries
git-svn-id: file:///svn/phpbb/trunk@7908 89ea8834-ac86-4346-8a33-228a782c2dd0
Diffstat (limited to 'phpBB/develop/generate_utf_confusables.php')
| -rw-r--r-- | phpBB/develop/generate_utf_confusables.php | 18 |
1 files changed, 17 insertions, 1 deletions
diff --git a/phpBB/develop/generate_utf_confusables.php b/phpBB/develop/generate_utf_confusables.php index e2fd5bbaa3..908ebbf6f4 100644 --- a/phpBB/develop/generate_utf_confusables.php +++ b/phpBB/develop/generate_utf_confusables.php @@ -30,15 +30,22 @@ $phpEx = substr(strrchr(__FILE__, '.'), 1); echo "Checking for required files\n"; download('http://unicode.org/reports/tr39/data/confusables.txt'); +download('http://unicode.org/Public/UNIDATA/CaseFolding.txt'); echo "\n"; /** -* Load the CaseFolding table +* Load the confusables table */ echo "Loading confusables\n"; $unidata = file_get_contents('confusables.txt'); +/** +* Load the CaseFolding table +*/ +echo "Loading CaseFolding\n"; +$casefolds = file_get_contents('CaseFolding.txt'); + function utf8_chr($cp) { @@ -61,6 +68,7 @@ function utf8_chr($cp) } preg_match_all('/^([0-9A-F]+) ;\s((?:[0-9A-F]+ )*);.*?$/im', $unidata, $array, PREG_SET_ORDER); +preg_match_all('/^([0-9A-F]+); ([CFS]); ([0-9A-F]+(?: [0-9A-F]+)*);/im', $casefolds, $casefold_array); // some that we defined ourselves $uniarray = array( @@ -136,6 +144,14 @@ foreach ($array as $value) $temp_hold = str_replace(utf8_chr(0x0031), utf8_chr(0x006C), $temp_hold); } } + + // uppercased chars that were folded do not exist in this universe, + // no amount of normalization could ever "trick" this into not working + if (in_array($value[1], $casefold_array[1])) + { + continue; + } + $uniarray[utf8_chr(hexdec((string)$value[1]))] = $temp_hold; } |