| commit | author | age | ||
| eda92e | 1 | <?php |
| AM | 2 | |
| a95874 | 3 | /** |
| eda92e | 4 | +-----------------------------------------------------------------------+ |
| AM | 5 | | This file is part of the Roundcube Webmail client | |
| 6 | | Copyright (C) 2008-2014, The Roundcube Dev Team | | |
| 7 | | | | |
| 8 | | Licensed under the GNU General Public License version 3 or | | |
| 9 | | any later version with exceptions for skins & plugins. | | |
| 10 | | See the README file for a full license statement. | | |
| 11 | | | | |
| 12 | | PURPOSE: | | |
| 13 | | Converts plain text to HTML | | |
| 14 | +-----------------------------------------------------------------------+ | |
| 15 | | Author: Aleksander Machniak <alec@alec.pl> | | |
| 16 | +-----------------------------------------------------------------------+ | |
| 17 | */ | |
| 18 | ||
| 19 | /** | |
| 20 | * Converts plain text to HTML | |
| 21 | * | |
| 22 | * @package Framework | |
| 23 | * @subpackage Utils | |
| 24 | */ | |
| 25 | class rcube_text2html | |
| 26 | { | |
| 27 | /** | |
| 28 | * Contains the HTML content after conversion. | |
| 29 | * | |
| 30 | * @var string $html | |
| 31 | */ | |
| 32 | protected $html; | |
| 33 | ||
| 34 | /** | |
| 35 | * Contains the plain text. | |
| 36 | * | |
| 37 | * @var string $text | |
| 38 | */ | |
| 39 | protected $text; | |
| 40 | ||
| 41 | /** | |
| 42 | * Configuration | |
| 43 | * | |
| 44 | * @var array $config | |
| 45 | */ | |
| 46 | protected $config = array( | |
| 47 | // non-breaking space | |
| d86ff9 | 48 | 'space' => "\xC2\xA0", |
| eda92e | 49 | // enables format=flowed parser |
| AM | 50 | 'flowed' => false, |
| 51 | // enables wrapping for non-flowed text | |
| d86ff9 | 52 | 'wrap' => true, |
| eda92e | 53 | // line-break tag |
| d86ff9 | 54 | 'break' => "<br>\n", |
| eda92e | 55 | // prefix and suffix (wrapper element) |
| d86ff9 | 56 | 'begin' => '<div class="pre">', |
| AM | 57 | 'end' => '</div>', |
| eda92e | 58 | // enables links replacement |
| d86ff9 | 59 | 'links' => true, |
| 9cc5a5 | 60 | // string replacer class |
| AM | 61 | 'replacer' => 'rcube_string_replacer', |
| c4ad7e | 62 | // prefix and suffix of unwrappable line |
| AM | 63 | 'nobr_start' => '<span style="white-space:nowrap">', |
| 64 | 'nobr_end' => '</span>', | |
| eda92e | 65 | ); |
| AM | 66 | |
| 67 | ||
| 68 | /** | |
| 69 | * Constructor. | |
| 70 | * | |
| 71 | * If the plain text source string (or file) is supplied, the class | |
| 72 | * will instantiate with that source propagated, all that has | |
| 73 | * to be done it to call get_html(). | |
| 74 | * | |
| 75 | * @param string $source Plain text | |
| 76 | * @param boolean $from_file Indicates $source is a file to pull content from | |
| 77 | * @param array $config Class configuration | |
| 78 | */ | |
| 79 | function __construct($source = '', $from_file = false, $config = array()) | |
| 80 | { | |
| 81 | if (!empty($source)) { | |
| 82 | $this->set_text($source, $from_file); | |
| 83 | } | |
| 84 | ||
| 85 | if (!empty($config) && is_array($config)) { | |
| 86 | $this->config = array_merge($this->config, $config); | |
| 87 | } | |
| 88 | } | |
| 89 | ||
| 90 | /** | |
| 91 | * Loads source text into memory, either from $source string or a file. | |
| 92 | * | |
| 93 | * @param string $source Plain text | |
| 94 | * @param boolean $from_file Indicates $source is a file to pull content from | |
| 95 | */ | |
| 96 | function set_text($source, $from_file = false) | |
| 97 | { | |
| 98 | if ($from_file && file_exists($source)) { | |
| 99 | $this->text = file_get_contents($source); | |
| 100 | } | |
| 101 | else { | |
| 102 | $this->text = $source; | |
| 103 | } | |
| 104 | ||
| 105 | $this->_converted = false; | |
| 106 | } | |
| 107 | ||
| 108 | /** | |
| 109 | * Returns the HTML content. | |
| 110 | * | |
| 111 | * @return string HTML content | |
| 112 | */ | |
| 113 | function get_html() | |
| 114 | { | |
| 115 | if (!$this->_converted) { | |
| 116 | $this->_convert(); | |
| 117 | } | |
| 118 | ||
| 119 | return $this->html; | |
| 120 | } | |
| 121 | ||
| 122 | /** | |
| 123 | * Prints the HTML. | |
| 124 | */ | |
| 125 | function print_html() | |
| 126 | { | |
| 127 | print $this->get_html(); | |
| 128 | } | |
| 129 | ||
| 130 | /** | |
| 131 | * Workhorse function that does actual conversion (calls _converter() method). | |
| 132 | */ | |
| 133 | protected function _convert() | |
| 134 | { | |
| 135 | // Convert TXT to HTML | |
| efc470 | 136 | $this->html = $this->_converter($this->text); |
| eda92e | 137 | $this->_converted = true; |
| AM | 138 | } |
| 139 | ||
| 140 | /** | |
| 141 | * Workhorse function that does actual conversion. | |
| 142 | * | |
| 143 | * @param string Plain text | |
| 144 | */ | |
| 145 | protected function _converter($text) | |
| 146 | { | |
| 147 | // make links and email-addresses clickable | |
| 148 | $attribs = array('link_attribs' => array('rel' => 'noreferrer', 'target' => '_blank')); | |
| 9cc5a5 | 149 | $replacer = new $this->config['replacer']($attribs); |
| eda92e | 150 | |
| AM | 151 | if ($this->config['flowed']) { |
| 152 | $flowed_char = 0x01; | |
| 153 | $text = rcube_mime::unfold_flowed($text, chr($flowed_char)); | |
| 154 | } | |
| 155 | ||
| 156 | // search for patterns like links and e-mail addresses and replace with tokens | |
| 157 | if ($this->config['links']) { | |
| 158 | $text = $replacer->replace($text); | |
| 159 | } | |
| 160 | ||
| 161 | // split body into single lines | |
| 162 | $text = preg_split('/\r?\n/', $text); | |
| 163 | $quote_level = 0; | |
| c0a5aa | 164 | $last = null; |
| eda92e | 165 | |
| c0a5aa | 166 | // wrap quoted lines with <blockquote> |
| AM | 167 | for ($n = 0, $cnt = count($text); $n < $cnt; $n++) { |
| eda92e | 168 | $flowed = false; |
| d41367 | 169 | if ($this->config['flowed'] && ord($text[$n][0]) == $flowed_char) { |
| eda92e | 170 | $flowed = true; |
| AM | 171 | $text[$n] = substr($text[$n], 1); |
| 172 | } | |
| 173 | ||
| 174 | if ($text[$n][0] == '>' && preg_match('/^(>+ {0,1})+/', $text[$n], $regs)) { | |
| 175 | $q = substr_count($regs[0], '>'); | |
| 176 | $text[$n] = substr($text[$n], strlen($regs[0])); | |
| 177 | $text[$n] = $this->_convert_line($text[$n], $flowed || $this->config['wrap']); | |
| c0a5aa | 178 | $_length = strlen(str_replace(' ', '', $text[$n])); |
| eda92e | 179 | |
| AM | 180 | if ($q > $quote_level) { |
| c0a5aa | 181 | if ($last !== null) { |
| AM | 182 | $text[$last] .= (!$length ? "\n" : '') |
| 183 | . $replacer->get_replacement($replacer->add( | |
| 184 | str_repeat('<blockquote>', $q - $quote_level))) | |
| 185 | . $text[$n]; | |
| 186 | ||
| 187 | unset($text[$n]); | |
| 188 | } | |
| 189 | else { | |
| 190 | $text[$n] = $replacer->get_replacement($replacer->add( | |
| 191 | str_repeat('<blockquote>', $q - $quote_level))) . $text[$n]; | |
| 192 | ||
| 193 | $last = $n; | |
| 194 | } | |
| eda92e | 195 | } |
| AM | 196 | else if ($q < $quote_level) { |
| c0a5aa | 197 | $text[$last] .= (!$length ? "\n" : '') |
| AM | 198 | . $replacer->get_replacement($replacer->add( |
| 199 | str_repeat('</blockquote>', $quote_level - $q))) | |
| 200 | . $text[$n]; | |
| 201 | ||
| 202 | unset($text[$n]); | |
| 203 | } | |
| 204 | else { | |
| eda92e | 205 | $last = $n; |
| AM | 206 | } |
| 207 | } | |
| 208 | else { | |
| 209 | $text[$n] = $this->_convert_line($text[$n], $flowed || $this->config['wrap']); | |
| c0a5aa | 210 | $q = 0; |
| AM | 211 | $_length = strlen(str_replace(' ', '', $text[$n])); |
| eda92e | 212 | |
| AM | 213 | if ($quote_level > 0) { |
| c0a5aa | 214 | $text[$last] .= (!$length ? "\n" : '') |
| AM | 215 | . $replacer->get_replacement($replacer->add( |
| 216 | str_repeat('</blockquote>', $quote_level))) | |
| 217 | . $text[$n]; | |
| 218 | ||
| 219 | unset($text[$n]); | |
| 220 | } | |
| 221 | else { | |
| 222 | $last = $n; | |
| eda92e | 223 | } |
| AM | 224 | } |
| 225 | ||
| 226 | $quote_level = $q; | |
| c0a5aa | 227 | $length = $_length; |
| eda92e | 228 | } |
| AM | 229 | |
| 230 | if ($quote_level > 0) { | |
| c0a5aa | 231 | $text[$last] .= $replacer->get_replacement($replacer->add( |
| AM | 232 | str_repeat('</blockquote>', $quote_level))); |
| eda92e | 233 | } |
| AM | 234 | |
| 235 | $text = join("\n", $text); | |
| 236 | ||
| 237 | // colorize signature (up to <sig_max_lines> lines) | |
| c0a5aa | 238 | $len = strlen($text); |
| AM | 239 | $sig_sep = "--" . $this->config['space'] . "\n"; |
| eda92e | 240 | $sig_max_lines = rcube::get_instance()->config->get('sig_max_lines', 15); |
| AM | 241 | |
| c0a5aa | 242 | while (($sp = strrpos($text, $sig_sep, $sp ? -$len+$sp-1 : 0)) !== false) { |
| eda92e | 243 | if ($sp == 0 || $text[$sp-1] == "\n") { |
| AM | 244 | // do not touch blocks with more that X lines |
| 245 | if (substr_count($text, "\n", $sp) < $sig_max_lines) { | |
| 246 | $text = substr($text, 0, max(0, $sp)) | |
| 247 | .'<span class="sig">'.substr($text, $sp).'</span>'; | |
| 248 | } | |
| 249 | ||
| 250 | break; | |
| 251 | } | |
| 252 | } | |
| 253 | ||
| 254 | // insert url/mailto links and citation tags | |
| 255 | $text = $replacer->resolve($text); | |
| 256 | ||
| 257 | // replace line breaks | |
| 258 | $text = str_replace("\n", $this->config['break'], $text); | |
| 259 | ||
| 260 | return $this->config['begin'] . $text . $this->config['end']; | |
| 261 | } | |
| 262 | ||
| 263 | /** | |
| 264 | * Converts spaces in line of text | |
| 265 | */ | |
| 266 | protected function _convert_line($text, $is_flowed) | |
| 267 | { | |
| 268 | static $table; | |
| 269 | ||
| 270 | if (empty($table)) { | |
| 271 | $table = get_html_translation_table(HTML_SPECIALCHARS); | |
| 272 | unset($table['?']); | |
| afd090 | 273 | |
| AM | 274 | // replace some whitespace characters |
| 275 | $table["\r"] = ''; | |
| 276 | $table["\t"] = ' '; | |
| eda92e | 277 | } |
| AM | 278 | |
| 279 | // skip signature separator | |
| 280 | if ($text == '-- ') { | |
| c0a5aa | 281 | return '--' . $this->config['space']; |
| eda92e | 282 | } |
| AM | 283 | |
| afd090 | 284 | // replace HTML special and whitespace characters |
| eda92e | 285 | $text = strtr($text, $table); |
| c4ad7e | 286 | |
| AM | 287 | $nbsp = $this->config['space']; |
| eda92e | 288 | |
| AM | 289 | // replace spaces with non-breaking spaces |
| 290 | if ($is_flowed) { | |
| d9d276 | 291 | $pos = 0; |
| AM | 292 | $diff = 0; |
| 293 | $len = strlen($nbsp); | |
| 294 | $copy = $text; | |
| 295 | ||
| 296 | while (($pos = strpos($text, ' ', $pos)) !== false) { | |
| 297 | if ($pos == 0 || $text[$pos-1] == ' ') { | |
| 298 | $copy = substr_replace($copy, $nbsp, $pos + $diff, 1); | |
| 299 | $diff += $len - 1; | |
| 300 | } | |
| 301 | $pos++; | |
| 302 | } | |
| 303 | ||
| 304 | $text = $copy; | |
| eda92e | 305 | } |
| c4ad7e | 306 | // make the whole line non-breakable if needed |
| AM | 307 | else if ($text !== '' && preg_match('/[^a-zA-Z0-9_]/', $text)) { |
| 308 | // use non-breakable spaces to correctly display | |
| 309 | // trailing/leading spaces and multi-space inside | |
| 310 | $text = str_replace(' ', $nbsp, $text); | |
| 311 | // wrap in nobr element, so it's not wrapped on e.g. - or / | |
| 312 | $text = $this->config['nobr_start'] . $text . $this->config['nobr_end']; | |
| eda92e | 313 | } |
| AM | 314 | |
| 315 | return $text; | |
| 316 | } | |
| 317 | } | |
