|
|
@@ -4262,4 +4262,122 @@ function sanitizeMSCutPaste($string)
|
|
|
return $string;
|
|
|
}
|
|
|
|
|
|
+/**
|
|
|
+ * Decode numeric html entities to their ascii or UTF8 equivalent character.
|
|
|
+ *
|
|
|
+ * Callback function for preg_replace_callback in subs-members
|
|
|
+ * Uses capture group 2 in the supplied array
|
|
|
+ * Does basic scan to ensure characters are inside a valid range
|
|
|
+ *
|
|
|
+ * @param array $matches
|
|
|
+ * @return string $string
|
|
|
+*/
|
|
|
+function replaceEntities__callback($matches)
|
|
|
+{
|
|
|
+ global $context;
|
|
|
+
|
|
|
+ if (!isset($matches[2]))
|
|
|
+ return '';
|
|
|
+
|
|
|
+ $num = $matches[2][0] === 'x' ? hexdec(substr($matches[2], 1)) : (int) $matches[2];
|
|
|
+
|
|
|
+ // remove left to right / right to left overrides
|
|
|
+ if ($num === 0x202D || $num === 0x202E)
|
|
|
+ return '';
|
|
|
+
|
|
|
+ // Quote, Ampersand, Apostrophe, Less/Greater Than get html replaced
|
|
|
+ if (in_array($num, array(0x22, 0x26, 0x27, 0x3C, 0x3E)))
|
|
|
+ return '&#' . $num . ';';
|
|
|
+
|
|
|
+ if (empty($context['utf8']))
|
|
|
+ {
|
|
|
+ // no control characters
|
|
|
+ if ($num < 0x20)
|
|
|
+ return '';
|
|
|
+ // text is text
|
|
|
+ elseif ($num < 0x80)
|
|
|
+ return chr($num);
|
|
|
+ // all others get html-ised
|
|
|
+ else
|
|
|
+ return '&#' . $matches[2] . ';';
|
|
|
+ }
|
|
|
+ else
|
|
|
+ {
|
|
|
+ // <0x20 are control characters, 0x20 is a space, > 0x10FFFF is past the end of the utf8 character set
|
|
|
+ // 0xD800 >= $num <= 0xDFFF are surrogate markers (not valid for utf8 text)
|
|
|
+ if ($num < 0x20 || $num > 0x10FFFF || ($num >= 0xD800 && $num <= 0xDFFF))
|
|
|
+ return '';
|
|
|
+ // <0x80 (or less than 128) are standard ascii characters a-z A-Z 0-9 and puncuation
|
|
|
+ elseif ($num < 0x80)
|
|
|
+ return chr($num);
|
|
|
+ // <0x800 (2048)
|
|
|
+ elseif ($num < 0x800)
|
|
|
+ return chr(($num >> 6) + 192) . chr(($num & 63) + 128);
|
|
|
+ // < 0x10000 (65536)
|
|
|
+ elseif ($num < 0x10000)
|
|
|
+ return chr(($num >> 12) + 224) . chr((($num >> 6) & 63) + 128) . chr(($num & 63) + 128);
|
|
|
+ // <= 0x10FFFF (1114111)
|
|
|
+ else
|
|
|
+ return chr(($num >> 18) + 240) . chr((($num >> 12) & 63) + 128) . chr((($num >> 6) & 63) + 128) . chr(($num & 63) + 128);
|
|
|
+ }
|
|
|
+}
|
|
|
+
|
|
|
+/**
|
|
|
+ * Converts html entities to utf8 equivalents
|
|
|
+ *
|
|
|
+ * Callback function for preg_replace_callback
|
|
|
+ * Uses capture group 1 in the supplied array
|
|
|
+ * Does basic checks to keep characters inside a viewable range.
|
|
|
+ *
|
|
|
+ * @param array $matches
|
|
|
+ * @return string $string
|
|
|
+*/
|
|
|
+function fixchar__callback($matches)
|
|
|
+{
|
|
|
+ if (!isset($matches[1]))
|
|
|
+ return '';
|
|
|
+
|
|
|
+ $num = $matches[1][0] === 'x' ? hexdec(substr($matches[1], 1)) : (int) $matches[1];
|
|
|
+
|
|
|
+ // <0x20 are control characters, > 0x10FFFF is past the end of the utf8 character set
|
|
|
+ // 0xD800 >= $num <= 0xDFFF are surrogate markers (not valid for utf8 text), 0x202D-E are left to right overrides
|
|
|
+ if ($num < 0x20 || $num > 0x10FFFF || ($num >= 0xD800 && $num <= 0xDFFF) || $num === 0x202D || $num === 0x202E)
|
|
|
+ return '';
|
|
|
+ // <0x80 (or less than 128) are standard ascii characters a-z A-Z 0-9 and puncuation
|
|
|
+ elseif ($num < 0x80)
|
|
|
+ return chr($num);
|
|
|
+ // <0x800 (2048)
|
|
|
+ elseif ($num < 0x800)
|
|
|
+ return chr(($num >> 6) + 192) . chr(($num & 63) + 128);
|
|
|
+ // < 0x10000 (65536)
|
|
|
+ elseif ($num < 0x10000)
|
|
|
+ return chr(($num >> 12) + 224) . chr((($num >> 6) & 63) + 128) . chr(($num & 63) + 128);
|
|
|
+ // <= 0x10FFFF (1114111)
|
|
|
+ else
|
|
|
+ return chr(($num >> 18) + 240) . chr((($num >> 12) & 63) + 128) . chr((($num >> 6) & 63) + 128) . chr(($num & 63) + 128);
|
|
|
+}
|
|
|
+
|
|
|
+/**
|
|
|
+ * Strips out invalid html entities, replaces others with html style { codes
|
|
|
+ *
|
|
|
+ * Callback function used of preg_replace_callback in smcFunc $ent_checks, for example
|
|
|
+ * strpos, strlen, substr etc
|
|
|
+ *
|
|
|
+ * @param array $matches
|
|
|
+ * @return string $string
|
|
|
+*/
|
|
|
+function entity_fix__callback($matches)
|
|
|
+{
|
|
|
+ if (!isset($matches[2]))
|
|
|
+ return '';
|
|
|
+
|
|
|
+ $num = $matches[2][0] === 'x' ? hexdec(substr($matches[2], 1)) : (int) $matches[2];
|
|
|
+
|
|
|
+ // we don't allow control characters, characters out of range, byte markers, etc
|
|
|
+ if ($num < 0x20 || $num > 0x10FFFF || ($num >= 0xD800 && $num <= 0xDFFF) || $num == 0x202D || $num == 0x202E)
|
|
|
+ return '';
|
|
|
+ else
|
|
|
+ return '&#' . $num . ';';
|
|
|
+}
|
|
|
+
|
|
|
?>
|
Spuds