root / drupal7 / sites / all / libraries / simplepie / build / charset.php @ 41cc1b08
| 1 |
<?php
|
|---|---|
| 2 |
|
| 3 |
require_once '../autoloader.php'; |
| 4 |
|
| 5 |
function normalize_character_set($charset) |
| 6 |
{
|
| 7 |
return strtolower(preg_replace('/(?:[^a-zA-Z0-9]+|([^0-9])0+)/', '\1', $charset)); |
| 8 |
} |
| 9 |
|
| 10 |
function build_character_set_list() |
| 11 |
{
|
| 12 |
$file = new SimplePie_File('http://www.iana.org/assignments/character-sets'); |
| 13 |
if (!$file->success && !($file->method & SIMPLEPIE_FILE_SOURCE_REMOTE === 0 || ($file->status_code === 200 || $file->status_code > 206 && $file->status_code < 300))) |
| 14 |
{
|
| 15 |
return false; |
| 16 |
} |
| 17 |
else
|
| 18 |
{
|
| 19 |
$data = explode("\n", $file->body); |
| 20 |
unset($file); |
| 21 |
|
| 22 |
foreach ($data as $line) |
| 23 |
{
|
| 24 |
// New character set
|
| 25 |
if (preg_match('/^Name:\s+(\S+)/', $line, $match)) |
| 26 |
{
|
| 27 |
// If we already have one, push it on to the array
|
| 28 |
if (isset($aliases)) |
| 29 |
{
|
| 30 |
foreach ($aliases as &$alias) |
| 31 |
{
|
| 32 |
$alias = normalize_character_set($alias); |
| 33 |
} |
| 34 |
$charsets[$preferred] = array_unique($aliases); |
| 35 |
natsort($charsets[$preferred]); |
| 36 |
} |
| 37 |
|
| 38 |
$aliases = array($match[1]); |
| 39 |
$preferred = $match[1]; |
| 40 |
} |
| 41 |
// Another alias
|
| 42 |
elseif (preg_match('/^Alias:\s+(\S+)(\s+\(preferred MIME name\))?\s*$/', $line, $match)) |
| 43 |
{
|
| 44 |
if ($match[1] !== 'None') |
| 45 |
{
|
| 46 |
$aliases[] = $match[1]; |
| 47 |
if (isset($match[2])) |
| 48 |
{
|
| 49 |
$preferred = $match[1]; |
| 50 |
} |
| 51 |
} |
| 52 |
} |
| 53 |
} |
| 54 |
|
| 55 |
// Compatibility replacements
|
| 56 |
// From http://www.whatwg.org/specs/web-apps/current-work/multipage/parsing.html#misinterpreted-for-compatibility
|
| 57 |
$compat = array( |
| 58 |
'EUC-KR' => 'windows-949', |
| 59 |
'GB2312' => 'GBK', |
| 60 |
'GB_2312-80' => 'GBK', |
| 61 |
'ISO-8859-1' => 'windows-1252', |
| 62 |
'ISO-8859-9' => 'windows-1254', |
| 63 |
'ISO-8859-11' => 'windows-874', |
| 64 |
'KS_C_5601-1987' => 'windows-949', |
| 65 |
'Shift_JIS' => 'Windows-31J', |
| 66 |
'TIS-620' => 'windows-874', |
| 67 |
//'US-ASCII' => 'windows-1252',
|
| 68 |
); |
| 69 |
|
| 70 |
foreach ($compat as $real => $replace) |
| 71 |
{
|
| 72 |
if (isset($charsets[$real]) && isset($charsets[$replace])) |
| 73 |
{
|
| 74 |
$charsets[$replace] = array_merge($charsets[$replace], $charsets[$real]); |
| 75 |
unset($charsets[$real]); |
| 76 |
} |
| 77 |
elseif (isset($charsets[$real])) |
| 78 |
{
|
| 79 |
$charsets[$replace] = $charsets[$real]; |
| 80 |
$charsets[$replace][] = normalize_character_set($replace); |
| 81 |
unset($charsets[$real]); |
| 82 |
} |
| 83 |
else
|
| 84 |
{
|
| 85 |
$charsets[$replace][] = normalize_character_set($real); |
| 86 |
} |
| 87 |
$charsets[$replace] = array_unique($charsets[$replace]); |
| 88 |
natsort($charsets[$replace]); |
| 89 |
} |
| 90 |
|
| 91 |
// Sort it
|
| 92 |
uksort($charsets, 'strnatcasecmp'); |
| 93 |
|
| 94 |
// Check that nothing matches more than one
|
| 95 |
$all = call_user_func_array('array_merge', $charsets); |
| 96 |
$all_count = array_count_values($all); |
| 97 |
if (max($all_count) > 1) |
| 98 |
{
|
| 99 |
echo "Duplicated charsets:\n"; |
| 100 |
foreach ($all_count as $charset => $count) |
| 101 |
{
|
| 102 |
if ($count > 1) |
| 103 |
{
|
| 104 |
echo "$charset\n"; |
| 105 |
} |
| 106 |
} |
| 107 |
} |
| 108 |
|
| 109 |
// And we're done!
|
| 110 |
return $charsets; |
| 111 |
} |
| 112 |
} |
| 113 |
|
| 114 |
function charset($charset) |
| 115 |
{
|
| 116 |
$normalized_charset = normalize_character_set($charset); |
| 117 |
if ($charsets = build_character_set_list()) |
| 118 |
{
|
| 119 |
foreach ($charsets as $preferred => $aliases) |
| 120 |
{
|
| 121 |
if (in_array($normalized_charset, $aliases)) |
| 122 |
{
|
| 123 |
return $preferred; |
| 124 |
} |
| 125 |
} |
| 126 |
return $charset; |
| 127 |
} |
| 128 |
else
|
| 129 |
{
|
| 130 |
return false; |
| 131 |
} |
| 132 |
} |
| 133 |
|
| 134 |
function build_function() |
| 135 |
{
|
| 136 |
if ($charsets = build_character_set_list()) |
| 137 |
{
|
| 138 |
$return = <<<EOF |
| 139 |
public static function encoding(\$charset)
|
| 140 |
{
|
| 141 |
// Normalization from UTS #22
|
| 142 |
switch (strtolower(preg_replace('/(?:[^a-zA-Z0-9]+|([^0-9])0+)/', '\\1', \$charset)))
|
| 143 |
{
|
| 144 |
|
| 145 |
EOF;
|
| 146 |
foreach ($charsets as $preferred => $aliases) |
| 147 |
{
|
| 148 |
foreach ($aliases as $alias) |
| 149 |
{
|
| 150 |
$return .= "\t\tcase " . var_export($alias, true) . ":\n"; |
| 151 |
} |
| 152 |
$return .= "\t\t\treturn " . var_export($preferred, true) . ";\n\n"; |
| 153 |
} |
| 154 |
$return .= <<<EOF |
| 155 |
default:
|
| 156 |
return \$charset;
|
| 157 |
}
|
| 158 |
}
|
| 159 |
EOF;
|
| 160 |
return $return; |
| 161 |
} |
| 162 |
else
|
| 163 |
{
|
| 164 |
return false; |
| 165 |
} |
| 166 |
} |
| 167 |
|
| 168 |
if (php_sapi_name() === 'cli' && realpath($_SERVER['argv'][0]) === __FILE__) |
| 169 |
{
|
| 170 |
echo build_function();
|
| 171 |
} |
| 172 |
|
| 173 |
?>
|