| 1 |
<?php
|
| 2 |
// $Id: typogrify.class.php,v 1.2.2.4 2009/04/28 16:01:52 mikl Exp $
|
| 3 |
|
| 4 |
/**
|
| 5 |
* @file typogrify.class.php
|
| 6 |
* Defines a class for providing different typographical tweaks to HTML
|
| 7 |
*/
|
| 8 |
class Typogrify {
|
| 9 |
/**
|
| 10 |
* Enable custom styling of ampersands.
|
| 11 |
*
|
| 12 |
* Wraps apersands in html with '<span class="amp">', so they can be
|
| 13 |
* styled with CSS. Ampersands are also normalized to '&. Requires
|
| 14 |
* ampersands to have whitespace or an ' ' on both sides.
|
| 15 |
*
|
| 16 |
* It won't mess up & that are already wrapped, in entities or URLs
|
| 17 |
* @param string
|
| 18 |
* @return string
|
| 19 |
*/
|
| 20 |
public static function amp($text) {
|
| 21 |
$amp_finder = "/(\s| )(&|&|&\#38;|&)(\s| )/";
|
| 22 |
return preg_replace($amp_finder, '\\1<span class="amp">&</span>\\3', $text);
|
| 23 |
}
|
| 24 |
|
| 25 |
/**
|
| 26 |
* Puts a   before and after an &ndash or —
|
| 27 |
*
|
| 28 |
* Dashes may have whitespace or an `` `` on both sides
|
| 29 |
* @param string
|
| 30 |
* @return string
|
| 31 |
*/
|
| 32 |
public static function dash($text) {
|
| 33 |
$dash_finder = "/(\s| | )*(—|–|–|–|—|—)(\s| | )*/";
|
| 34 |
return preg_replace($dash_finder, ' \\2 ', $text);
|
| 35 |
}
|
| 36 |
|
| 37 |
/**
|
| 38 |
* Helper method for caps method - used for preg_replace_callback
|
| 39 |
*/
|
| 40 |
public static function _cap_wrapper($matchobj) {
|
| 41 |
if (!empty($matchobj[2])) {
|
| 42 |
return sprintf('<span class="caps">%s</span>', $matchobj[2]);
|
| 43 |
}
|
| 44 |
else {
|
| 45 |
$mthree = $matchobj[3];
|
| 46 |
if (($mthree{strlen($mthree)-1}) == " ") {
|
| 47 |
$caps = substr($mthree, 0, -1);
|
| 48 |
$tail = ' ';
|
| 49 |
}
|
| 50 |
else {
|
| 51 |
$caps = $mthree;
|
| 52 |
$tail = '';
|
| 53 |
}
|
| 54 |
return sprintf('<span class="caps">%s</span>%s', $caps, $tail);
|
| 55 |
}
|
| 56 |
}
|
| 57 |
|
| 58 |
/**
|
| 59 |
* Stylable capitals
|
| 60 |
*
|
| 61 |
* Wraps multiple capital letters in ``<span class="caps">``
|
| 62 |
* so they can be styled with CSS.
|
| 63 |
*
|
| 64 |
* Uses the smartypants tokenizer to not screw with HTML or with tags it shouldn't.
|
| 65 |
*/
|
| 66 |
public static function caps($text) {
|
| 67 |
// If _TokenizeHTML from Smartypants is not present, don't do anything.
|
| 68 |
if (!function_exists('_TokenizeHTML')) {
|
| 69 |
return $text;
|
| 70 |
}
|
| 71 |
|
| 72 |
$tokens = _TokenizeHTML($text);
|
| 73 |
$result = array();
|
| 74 |
$in_skipped_tag = false;
|
| 75 |
|
| 76 |
$cap_finder = "/(
|
| 77 |
(\b[A-Z\d]* # Group 2: Any amount of caps and digits
|
| 78 |
[A-Z]\d*[A-Z] # A cap string much at least include two caps (but they can have digits between them)
|
| 79 |
[A-Z\d]*\b) # Any amount of caps and digits
|
| 80 |
| (\b[A-Z]+\.\s? # OR: Group 3: Some caps, followed by a '.' and an optional space
|
| 81 |
(?:[A-Z]+\.\s?)+) # Followed by the same thing at least once more
|
| 82 |
(?:\s|\b|$))/x";
|
| 83 |
|
| 84 |
$tags_to_skip_regex = "/<(\/)?(?:pre|code|kbd|script|math)[^>]*>/i";
|
| 85 |
|
| 86 |
foreach ($tokens as $token) {
|
| 87 |
if ( $token[0] == "tag" ) {
|
| 88 |
// Don't mess with tags.
|
| 89 |
$result[] = $token[1];
|
| 90 |
$close_match = preg_match($tags_to_skip_regex, $token[1]);
|
| 91 |
if ($close_match) {
|
| 92 |
$in_skipped_tag = true;
|
| 93 |
}
|
| 94 |
else {
|
| 95 |
$in_skipped_tag = false;
|
| 96 |
}
|
| 97 |
}
|
| 98 |
else {
|
| 99 |
if ( $in_skipped_tag ) {
|
| 100 |
$result[] = $token[1];
|
| 101 |
}
|
| 102 |
else {
|
| 103 |
$result[] = preg_replace_callback($cap_finder, array('Typogrify', '_cap_wrapper'), $token[1]);
|
| 104 |
}
|
| 105 |
}
|
| 106 |
}
|
| 107 |
return join("", $result);
|
| 108 |
}
|
| 109 |
|
| 110 |
/**
|
| 111 |
* Helper method for initial_quotes method - used for preg_replace_callback
|
| 112 |
*/
|
| 113 |
public static function _quote_wrapper($matchobj) {
|
| 114 |
if (!empty($matchobj[7])) {
|
| 115 |
$classname = "dquo";
|
| 116 |
$quote = $matchobj[7];
|
| 117 |
}
|
| 118 |
else {
|
| 119 |
$classname = "quo";
|
| 120 |
$quote = $matchobj[8];
|
| 121 |
}
|
| 122 |
return sprintf('%s<span class="%s">%s</span>', $matchobj[1], $classname, $quote);
|
| 123 |
}
|
| 124 |
|
| 125 |
/**
|
| 126 |
* initial_quotes
|
| 127 |
*
|
| 128 |
* Wraps initial quotes in ``class="dquo"`` for double quotes or
|
| 129 |
* ``class="quo"`` for single quotes. Works in these block tags ``(h1-h6, p, li)``
|
| 130 |
* and also accounts for potential opening inline elements ``a, em, strong, span, b, i``
|
| 131 |
* Optionally choose to apply quote span tags to Gullemets as well.
|
| 132 |
*/
|
| 133 |
public static function initial_quotes($text, $do_guillemets = false) {
|
| 134 |
$quote_finder = "/((<(p|h[1-6]|li)[^>]*>|^) # start with an opening p, h1-6, li or the start of the string
|
| 135 |
\s* # optional white space!
|
| 136 |
(<(a|em|span|strong|i|b)[^>]*>\s*)*) # optional opening inline tags, with more optional white space for each.
|
| 137 |
((\"|“|&\#8220;)|('|‘|&\#8216;)) # Find me a quote! (only need to find the left quotes and the primes)
|
| 138 |
# double quotes are in group 7, singles in group 8
|
| 139 |
/ix";
|
| 140 |
|
| 141 |
if ($do_guillemets) {
|
| 142 |
$quote_finder = "/((<(p|h[1-6]|li)[^>]*>|^) # start with an opening p, h1-6, li or the start of the string
|
| 143 |
\s* # optional white space!
|
| 144 |
(<(a|em|span|strong|i|b)[^>]*>\s*)*) # optional opening inline tags, with more optional white space for each.
|
| 145 |
((\"|“|&\#8220;|\xAE|&\#171;|«)|('|‘|&\#8216;)) # Find me a quote! (only need to find the left quotes and the primes) - also look for guillemets (>> and << characters))
|
| 146 |
# double quotes are in group 7, singles in group 8
|
| 147 |
/ix";
|
| 148 |
}
|
| 149 |
return preg_replace_callback($quote_finder, array('Typogrify', '_quote_wrapper'), $text);
|
| 150 |
}
|
| 151 |
|
| 152 |
/**
|
| 153 |
* widont
|
| 154 |
*
|
| 155 |
* Replaces the space between the last two words in a string with `` ``
|
| 156 |
* Works in these block tags ``(h1-h6, p, li)`` and also accounts for
|
| 157 |
* potential closing inline elements ``a, em, strong, span, b, i``
|
| 158 |
*
|
| 159 |
* Empty HTMLs shouldn't error
|
| 160 |
*/
|
| 161 |
public static function widont($text) {
|
| 162 |
// This regex is a beast, tread lightly
|
| 163 |
$widont_finder = "/(\s+) # the space to replace
|
| 164 |
([^<>\s]+ # must be flollowed by non-tag non-space characters
|
| 165 |
\s* # optional white space!
|
| 166 |
(<\/(a|em|span|strong|i|b)[^>]*>\s*)* # optional closing inline tags with optional white space after each
|
| 167 |
((<\/(p|h[1-6]|li|dt|dd)>)|$)) # end with a closing p, h1-6, li or the end of the string
|
| 168 |
/x";
|
| 169 |
return preg_replace($widont_finder, ' $2', $text);
|
| 170 |
}
|
| 171 |
|
| 172 |
/**
|
| 173 |
* typogrify
|
| 174 |
*
|
| 175 |
* The super typography filter.
|
| 176 |
* Applies the following filters: widont, smartypants, caps, amp, initial_quotes
|
| 177 |
* Optionally choose to apply quote span tags to Gullemets as well.
|
| 178 |
*/
|
| 179 |
public static function filter($text, $do_guillemets=FALSE) {
|
| 180 |
$text = Typogrify::amp($text);
|
| 181 |
$text = Typogrify::widont($text);
|
| 182 |
$text = SmartyPants($text);
|
| 183 |
$text = Typogrify::caps($text);
|
| 184 |
$text = Typogrify::initial_quotes($text, $do_guillemets);
|
| 185 |
$text = Typogrify::dash($text);
|
| 186 |
return $text;
|
| 187 |
}
|
| 188 |
}
|
| 189 |
|
Parent Directory
| 
Revision Log
| 
Revision Graph