|
0
|
1 <?php
|
|
|
2
|
|
|
3 /**
|
|
|
4 +-----------------------------------------------------------------------+
|
|
|
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
|
|
|
48 'space' => "\xC2\xA0",
|
|
|
49 // enables format=flowed parser
|
|
|
50 'flowed' => false,
|
|
|
51 // enables delsp=yes parser
|
|
|
52 'delsp' => false,
|
|
|
53 // enables wrapping for non-flowed text
|
|
|
54 'wrap' => true,
|
|
|
55 // line-break tag
|
|
|
56 'break' => "<br>\n",
|
|
|
57 // prefix and suffix (wrapper element)
|
|
|
58 'begin' => '<div class="pre">',
|
|
|
59 'end' => '</div>',
|
|
|
60 // enables links replacement
|
|
|
61 'links' => true,
|
|
|
62 // string replacer class
|
|
|
63 'replacer' => 'rcube_string_replacer',
|
|
|
64 // prefix and suffix of unwrappable line
|
|
|
65 'nobr_start' => '<span style="white-space:nowrap">',
|
|
|
66 'nobr_end' => '</span>',
|
|
|
67 );
|
|
|
68
|
|
19
|
69 /**
|
|
|
70 * Indicates whether content in the $text variable has been converted yet.
|
|
|
71 *
|
|
|
72 * @var boolean $_converted
|
|
|
73 * @see $html, $text
|
|
|
74 */
|
|
|
75 protected $_converted = false;
|
|
0
|
76
|
|
|
77 /**
|
|
|
78 * Constructor.
|
|
|
79 *
|
|
|
80 * If the plain text source string (or file) is supplied, the class
|
|
|
81 * will instantiate with that source propagated, all that has
|
|
|
82 * to be done it to call get_html().
|
|
|
83 *
|
|
|
84 * @param string $source Plain text
|
|
|
85 * @param boolean $from_file Indicates $source is a file to pull content from
|
|
|
86 * @param array $config Class configuration
|
|
|
87 */
|
|
|
88 function __construct($source = '', $from_file = false, $config = array())
|
|
|
89 {
|
|
|
90 if (!empty($source)) {
|
|
|
91 $this->set_text($source, $from_file);
|
|
|
92 }
|
|
|
93
|
|
|
94 if (!empty($config) && is_array($config)) {
|
|
|
95 $this->config = array_merge($this->config, $config);
|
|
|
96 }
|
|
|
97 }
|
|
|
98
|
|
|
99 /**
|
|
|
100 * Loads source text into memory, either from $source string or a file.
|
|
|
101 *
|
|
|
102 * @param string $source Plain text
|
|
|
103 * @param boolean $from_file Indicates $source is a file to pull content from
|
|
|
104 */
|
|
|
105 function set_text($source, $from_file = false)
|
|
|
106 {
|
|
|
107 if ($from_file && file_exists($source)) {
|
|
|
108 $this->text = file_get_contents($source);
|
|
|
109 }
|
|
|
110 else {
|
|
|
111 $this->text = $source;
|
|
|
112 }
|
|
|
113
|
|
|
114 $this->_converted = false;
|
|
|
115 }
|
|
|
116
|
|
|
117 /**
|
|
|
118 * Returns the HTML content.
|
|
|
119 *
|
|
|
120 * @return string HTML content
|
|
|
121 */
|
|
|
122 function get_html()
|
|
|
123 {
|
|
|
124 if (!$this->_converted) {
|
|
|
125 $this->_convert();
|
|
|
126 }
|
|
|
127
|
|
|
128 return $this->html;
|
|
|
129 }
|
|
|
130
|
|
|
131 /**
|
|
|
132 * Prints the HTML.
|
|
|
133 */
|
|
|
134 function print_html()
|
|
|
135 {
|
|
|
136 print $this->get_html();
|
|
|
137 }
|
|
|
138
|
|
|
139 /**
|
|
|
140 * Workhorse function that does actual conversion (calls _converter() method).
|
|
|
141 */
|
|
|
142 protected function _convert()
|
|
|
143 {
|
|
|
144 // Convert TXT to HTML
|
|
|
145 $this->html = $this->_converter($this->text);
|
|
|
146 $this->_converted = true;
|
|
|
147 }
|
|
|
148
|
|
|
149 /**
|
|
|
150 * Workhorse function that does actual conversion.
|
|
|
151 *
|
|
|
152 * @param string Plain text
|
|
|
153 */
|
|
|
154 protected function _converter($text)
|
|
|
155 {
|
|
|
156 // make links and email-addresses clickable
|
|
|
157 $attribs = array('link_attribs' => array('rel' => 'noreferrer', 'target' => '_blank'));
|
|
|
158 $replacer = new $this->config['replacer']($attribs);
|
|
|
159
|
|
|
160 if ($this->config['flowed']) {
|
|
|
161 $flowed_char = 0x01;
|
|
|
162 $delsp = $this->config['delsp'];
|
|
|
163 $text = rcube_mime::unfold_flowed($text, chr($flowed_char), $delsp);
|
|
|
164 }
|
|
|
165
|
|
|
166 // search for patterns like links and e-mail addresses and replace with tokens
|
|
|
167 if ($this->config['links']) {
|
|
|
168 $text = $replacer->replace($text);
|
|
|
169 }
|
|
|
170
|
|
|
171 // split body into single lines
|
|
|
172 $text = preg_split('/\r?\n/', $text);
|
|
|
173 $quote_level = 0;
|
|
|
174 $last = null;
|
|
|
175
|
|
|
176 // wrap quoted lines with <blockquote>
|
|
|
177 for ($n = 0, $cnt = count($text); $n < $cnt; $n++) {
|
|
|
178 $flowed = false;
|
|
27
|
179 if ($this->config['flowed'] && ord($text[$n][0]??null) == $flowed_char) {
|
|
0
|
180 $flowed = true;
|
|
|
181 $text[$n] = substr($text[$n], 1);
|
|
|
182 }
|
|
|
183
|
|
19
|
184 if ($text[$n] && $text[$n][0] == '>' && preg_match('/^(>+ {0,1})+/', $text[$n], $regs)) {
|
|
0
|
185 $q = substr_count($regs[0], '>');
|
|
|
186 $text[$n] = substr($text[$n], strlen($regs[0]));
|
|
|
187 $text[$n] = $this->_convert_line($text[$n], $flowed || $this->config['wrap']);
|
|
|
188 $_length = strlen(str_replace(' ', '', $text[$n]));
|
|
|
189
|
|
|
190 if ($q > $quote_level) {
|
|
|
191 if ($last !== null) {
|
|
|
192 $text[$last] .= (!$length ? "\n" : '')
|
|
|
193 . $replacer->get_replacement($replacer->add(
|
|
|
194 str_repeat('<blockquote>', $q - $quote_level)))
|
|
|
195 . $text[$n];
|
|
|
196
|
|
|
197 unset($text[$n]);
|
|
|
198 }
|
|
|
199 else {
|
|
|
200 $text[$n] = $replacer->get_replacement($replacer->add(
|
|
|
201 str_repeat('<blockquote>', $q - $quote_level))) . $text[$n];
|
|
|
202
|
|
|
203 $last = $n;
|
|
|
204 }
|
|
|
205 }
|
|
|
206 else if ($q < $quote_level) {
|
|
|
207 $text[$last] .= (!$length ? "\n" : '')
|
|
|
208 . $replacer->get_replacement($replacer->add(
|
|
|
209 str_repeat('</blockquote>', $quote_level - $q)))
|
|
|
210 . $text[$n];
|
|
|
211
|
|
|
212 unset($text[$n]);
|
|
|
213 }
|
|
|
214 else {
|
|
|
215 $last = $n;
|
|
|
216 }
|
|
|
217 }
|
|
|
218 else {
|
|
|
219 $text[$n] = $this->_convert_line($text[$n], $flowed || $this->config['wrap']);
|
|
|
220 $q = 0;
|
|
|
221 $_length = strlen(str_replace(' ', '', $text[$n]));
|
|
|
222
|
|
|
223 if ($quote_level > 0) {
|
|
|
224 $text[$last] .= (!$length ? "\n" : '')
|
|
|
225 . $replacer->get_replacement($replacer->add(
|
|
|
226 str_repeat('</blockquote>', $quote_level)))
|
|
|
227 . $text[$n];
|
|
|
228
|
|
|
229 unset($text[$n]);
|
|
|
230 }
|
|
|
231 else {
|
|
|
232 $last = $n;
|
|
|
233 }
|
|
|
234 }
|
|
|
235
|
|
|
236 $quote_level = $q;
|
|
|
237 $length = $_length;
|
|
|
238 }
|
|
|
239
|
|
|
240 if ($quote_level > 0) {
|
|
|
241 $text[$last] .= $replacer->get_replacement($replacer->add(
|
|
|
242 str_repeat('</blockquote>', $quote_level)));
|
|
|
243 }
|
|
|
244
|
|
|
245 $text = join("\n", $text);
|
|
|
246
|
|
|
247 // colorize signature (up to <sig_max_lines> lines)
|
|
|
248 $len = strlen($text);
|
|
|
249 $sig_sep = "--" . $this->config['space'] . "\n";
|
|
|
250 $sig_max_lines = rcube::get_instance()->config->get('sig_max_lines', 15);
|
|
|
251
|
|
22
|
252 while (($sp = strrpos($text, $sig_sep, isset($sp) ? -$len+$sp-1 : 0)) !== false) {
|
|
0
|
253 if ($sp == 0 || $text[$sp-1] == "\n") {
|
|
|
254 // do not touch blocks with more that X lines
|
|
|
255 if (substr_count($text, "\n", $sp) < $sig_max_lines) {
|
|
|
256 $text = substr($text, 0, max(0, $sp))
|
|
|
257 .'<span class="sig">'.substr($text, $sp).'</span>';
|
|
|
258 }
|
|
|
259
|
|
|
260 break;
|
|
|
261 }
|
|
|
262 }
|
|
|
263
|
|
|
264 // insert url/mailto links and citation tags
|
|
|
265 $text = $replacer->resolve($text);
|
|
|
266
|
|
|
267 // replace line breaks
|
|
|
268 $text = str_replace("\n", $this->config['break'], $text);
|
|
|
269
|
|
|
270 return $this->config['begin'] . $text . $this->config['end'];
|
|
|
271 }
|
|
|
272
|
|
|
273 /**
|
|
|
274 * Converts spaces in line of text
|
|
|
275 */
|
|
|
276 protected function _convert_line($text, $is_flowed)
|
|
|
277 {
|
|
|
278 static $table;
|
|
|
279
|
|
|
280 if (empty($table)) {
|
|
|
281 $table = get_html_translation_table(HTML_SPECIALCHARS);
|
|
|
282 unset($table['?']);
|
|
|
283
|
|
|
284 // replace some whitespace characters
|
|
|
285 $table["\r"] = '';
|
|
|
286 $table["\t"] = ' ';
|
|
|
287 }
|
|
|
288
|
|
|
289 // skip signature separator
|
|
|
290 if ($text == '-- ') {
|
|
|
291 return '--' . $this->config['space'];
|
|
|
292 }
|
|
|
293
|
|
|
294 // replace HTML special and whitespace characters
|
|
|
295 $text = strtr($text, $table);
|
|
|
296
|
|
|
297 $nbsp = $this->config['space'];
|
|
|
298
|
|
|
299 // replace spaces with non-breaking spaces
|
|
|
300 if ($is_flowed) {
|
|
|
301 $pos = 0;
|
|
|
302 $diff = 0;
|
|
|
303 $len = strlen($nbsp);
|
|
|
304 $copy = $text;
|
|
|
305
|
|
|
306 while (($pos = strpos($text, ' ', $pos)) !== false) {
|
|
|
307 if ($pos == 0 || $text[$pos-1] == ' ') {
|
|
|
308 $copy = substr_replace($copy, $nbsp, $pos + $diff, 1);
|
|
|
309 $diff += $len - 1;
|
|
|
310 }
|
|
|
311 $pos++;
|
|
|
312 }
|
|
|
313
|
|
|
314 $text = $copy;
|
|
|
315 }
|
|
|
316 // make the whole line non-breakable if needed
|
|
|
317 else if ($text !== '' && preg_match('/[^a-zA-Z0-9_]/', $text)) {
|
|
|
318 // use non-breakable spaces to correctly display
|
|
|
319 // trailing/leading spaces and multi-space inside
|
|
|
320 $text = str_replace(' ', $nbsp, $text);
|
|
|
321 // wrap in nobr element, so it's not wrapped on e.g. - or /
|
|
|
322 $text = $this->config['nobr_start'] . $text . $this->config['nobr_end'];
|
|
|
323 }
|
|
|
324
|
|
|
325 return $text;
|
|
|
326 }
|
|
|
327 }
|
