]> git.immae.eu Git - github/wallabag/wallabag.git/blame - inc/functions.php
add tidy call to clean html output
[github/wallabag/wallabag.git] / inc / functions.php
CommitLineData
24619534 1<?php
6f87a197 2/**
3 * poche, a read it later open source system
4 *
5 * @category poche
6 * @author Nicolas Lœuillet <support@inthepoche.com>
7 * @copyright 2013
8 * @license http://www.wtfpl.net/ see COPYING file
9 */
24619534 10
c8bbe19b 11/**
12 * Permet de générer l'URL de poche pour le bookmarklet
13 */
8046748b 14function get_poche_url()
c8bbe19b 15{
16 $protocol = "http";
17 if(isset($_SERVER['HTTPS'])) {
f0fc5011 18 if($_SERVER['HTTPS'] != "off" && $_SERVER['HTTPS'] != "") {
c8bbe19b 19 $protocol = "https";
20 }
21 }
24619534 22
c8bbe19b 23 return $protocol . "://" . $_SERVER['HTTP_HOST'] . $_SERVER['REQUEST_URI'];
24619534 24}
25
24619534 26// function define to retrieve url content
1c182b6c 27function get_external_file($url)
c8bbe19b 28{
1c182b6c 29 $timeout = 15;
24619534 30 // spoofing FireFox 18.0
31 $useragent="Mozilla/5.0 (Windows NT 5.1; rv:18.0) Gecko/20100101 Firefox/18.0";
32
33 if (in_array ('curl', get_loaded_extensions())) {
34 // Fetch feed from URL
35 $curl = curl_init();
36 curl_setopt($curl, CURLOPT_URL, $url);
37 curl_setopt($curl, CURLOPT_TIMEOUT, $timeout);
38 curl_setopt($curl, CURLOPT_FOLLOWLOCATION, true);
39 curl_setopt($curl, CURLOPT_RETURNTRANSFER, true);
40 curl_setopt($curl, CURLOPT_HEADER, false);
41
2987031b 42 // FOR SSL do not verified certificate
4ddbd267 43 curl_setopt($curl, CURLOPT_SSL_VERIFYPEER, FALSE);
2987031b 44 curl_setopt($curl, CURLOPT_AUTOREFERER, TRUE );
4ddbd267 45
24619534 46 // FeedBurner requires a proper USER-AGENT...
47 curl_setopt($curl, CURL_HTTP_VERSION_1_1, true);
48 curl_setopt($curl, CURLOPT_ENCODING, "gzip, deflate");
49 curl_setopt($curl, CURLOPT_USERAGENT, $useragent);
50
51 $data = curl_exec($curl);
52
53 $httpcode = curl_getinfo($curl, CURLINFO_HTTP_CODE);
54
55 $httpcodeOK = isset($httpcode) and ($httpcode == 200 or $httpcode == 301);
56
57 curl_close($curl);
58 } else {
59
60 // create http context and add timeout and user-agent
4ddbd267 61 $context = stream_context_create(array(
2987031b 62 'http'=>array('timeout' => $timeout,
63 'header'=> "User-Agent: ".$useragent, /*spoot Mozilla Firefox*/
64 'follow_location' => true),
65 // FOR SSL do not verified certificate
66 'ssl' => array('verify_peer' => false,
67 'allow_self_signed' => true)
68 )
69 );
24619534 70
71 // only download page lesser than 4MB
72 $data = @file_get_contents($url, false, $context, -1, 4000000); // We download at most 4 MB from source.
24619534 73
74 if(isset($http_response_header) and isset($http_response_header[0])) {
75 $httpcodeOK = isset($http_response_header) and isset($http_response_header[0]) and ((strpos($http_response_header[0], '200 OK') !== FALSE) or (strpos($http_response_header[0], '301 Moved Permanently') !== FALSE));
76 }
77 }
78
79 // if response is not empty and response is OK
80 if (isset($data) and isset($httpcodeOK) and $httpcodeOK ) {
81
82 // take charset of page and get it
83 preg_match('#<meta .*charset=.*>#Usi', $data, $meta);
84
85 // if meta tag is found
86 if (!empty($meta[0])) {
87 // retrieve encoding in $enc
88 preg_match('#charset="?(.*)"#si', $meta[0], $enc);
89
90 // if charset is found set it otherwise, set it to utf-8
91 $html_charset = (!empty($enc[1])) ? strtolower($enc[1]) : 'utf-8';
92
93 } else {
94 $html_charset = 'utf-8';
95 $enc[1] = '';
96 }
97
98 // replace charset of url to charset of page
99 $data = str_replace('charset='.$enc[1], 'charset='.$html_charset, $data);
100
101 return $data;
102 }
103 else {
104 return FALSE;
105 }
3c8d80ae 106}
107
8046748b 108/**
109 * Préparation de l'URL avec récupération du contenu avant insertion en base
110 */
d06f30ef 111function prepare_url($url)
3c8d80ae 112{
113 $parametres = array();
e4d2565e 114 $url = html_entity_decode(trim($url));
3c8d80ae 115
116 // We remove the annoying parameters added by FeedBurner and GoogleFeedProxy (?utm_source=...)
117 // from shaarli, by sebsauvage
118 $i=strpos($url,'&utm_source='); if ($i!==false) $url=substr($url,0,$i);
119 $i=strpos($url,'?utm_source='); if ($i!==false) $url=substr($url,0,$i);
120 $i=strpos($url,'#xtor=RSS-'); if ($i!==false) $url=substr($url,0,$i);
121
e4d2565e 122 $title = $url;
2987031b 123 $html = Encoding::toUTF8(get_external_file($url,15));
124 // If get_external_file if not able to retrieve HTTPS content try the same URL with HTTP protocol
125 if (!preg_match('!^https?://!i', $url) && (!isset($html) || strlen($html) <= 0)) {
126 $url = 'http://' . $url;
127 $html = Encoding::toUTF8(get_external_file($url,15));
128 }
3c8d80ae 129
494e21b4 130 if (function_exists('tidy_parse_string')) {
131 $tidy = tidy_parse_string($html, array(), 'UTF8');
132 $tidy->cleanRepair();
133 $html = $tidy->value;
134 }
135
3c8d80ae 136 if (isset($html) and strlen($html) > 0)
137 {
138 $r = new Readability($html, $url);
2987031b 139
d06f30ef 140 $r->convertLinksToFootnotes = CONVERT_LINKS_FOOTNOTES;
2987031b 141 $r->revertForcedParagraphElements = REVERT_FORCED_PARAGRAPH_ELEMENTS;
4ddbd267 142
3c8d80ae 143 if($r->init())
144 {
1c182b6c 145 $content = $r->articleContent->innerHTML;
146 $parametres['title'] = $r->articleTitle->innerHTML;
64458521 147 $parametres['content'] = $content;
1c182b6c 148 return $parametres;
3c8d80ae 149 }
150 }
151
1c182b6c 152 return FALSE;
153}
154
155/**
156 * On modifie les URLS des images dans le corps de l'article
157 */
158function filtre_picture($content, $url, $id)
159{
160 $matches = array();
161 preg_match_all('#<\s*(img)[^>]+src="([^"]*)"[^>]*>#Si', $content, $matches, PREG_SET_ORDER);
162 foreach($matches as $i => $link)
163 {
164 $link[1] = trim($link[1]);
165 if (!preg_match('#^(([a-z]+://)|(\#))#', $link[1]) )
166 {
167 $absolute_path = get_absolute_link($link[2],$url);
168 $filename = basename(parse_url($absolute_path, PHP_URL_PATH));
169 $directory = create_assets_directory($id);
170 $fullpath = $directory . '/' . $filename;
171 download_pictures($absolute_path, $fullpath);
172 $content = str_replace($matches[$i][2], $fullpath, $content);
173 }
174
175 }
176
177 return $content;
178}
179
180/**
181 * Retourne le lien absolu
182 */
183function get_absolute_link($relative_link, $url)
184{
185 /* return if already absolute URL */
186 if (parse_url($relative_link, PHP_URL_SCHEME) != '') return $relative_link;
187
188 /* queries and anchors */
189 if ($relative_link[0]=='#' || $relative_link[0]=='?') return $url . $relative_link;
190
191 /* parse base URL and convert to local variables:
192 $scheme, $host, $path */
193 extract(parse_url($url));
194
195 /* remove non-directory element from path */
196 $path = preg_replace('#/[^/]*$#', '', $path);
197
198 /* destroy path if relative url points to root */
199 if ($relative_link[0] == '/') $path = '';
200
201 /* dirty absolute URL */
202 $abs = $host . $path . '/' . $relative_link;
203
204 /* replace '//' or '/./' or '/foo/../' with '/' */
205 $re = array('#(/\.?/)#', '#/(?!\.\.)[^/]+/\.\./#');
206 for($n=1; $n>0; $abs=preg_replace($re, '/', $abs, -1, $n)) {}
207
208 /* absolute URL is ready! */
209 return $scheme.'://'.$abs;
210}
211
212/**
213 * Téléchargement des images
214 */
215
216function download_pictures($absolute_path, $fullpath)
217{
218 $rawdata = get_external_file($absolute_path);
3c8d80ae 219
1c182b6c 220 if(file_exists($fullpath)) {
221 unlink($fullpath);
222 }
223 $fp = fopen($fullpath, 'x');
224 fwrite($fp, $rawdata);
225 fclose($fp);
226}
227
228/**
229 * Crée un répertoire de médias pour l'article
230 */
231function create_assets_directory($id)
232{
233 $assets_path = ABS_PATH;
234 if(!is_dir($assets_path)) {
235 mkdir($assets_path, 0705);
236 }
237
238 $article_directory = $assets_path . $id;
239 if(!is_dir($article_directory)) {
240 mkdir($article_directory, 0705);
241 }
242
243 return $article_directory;
244}
245
246/**
247 * Suppression du répertoire d'images
248 */
249function remove_directory($directory)
250{
251 if(is_dir($directory)) {
252 $files = array_diff(scandir($directory), array('.','..'));
253 foreach ($files as $file) {
07a19252 254 (is_dir("$directory/$file")) ? remove_directory("$directory/$file") : unlink("$directory/$file");
1c182b6c 255 }
256 return rmdir($directory);
257 }
263d6c67 258}
259
a1953dff 260function display_view($view, $id = 0, $full_head = 'yes')
261{
f0070a15 262 global $tpl, $store, $msg;
a1953dff 263
264 switch ($view)
265 {
44e77bfa 266 case 'export':
267 $entries = $store->retrieveAll();
6f87a197 268 $tpl->assign('export', myTool::renderJson($entries));
44e77bfa 269 $tpl->draw('export');
270 logm('export view');
271 break;
016989b7 272 case 'config':
273 $tpl->assign('load_all_js', 0);
274 $tpl->draw('head');
275 $tpl->draw('home');
276 $tpl->draw('config');
277 $tpl->draw('js');
278 $tpl->draw('footer');
279 logm('config view');
44e77bfa 280 break;
a1953dff 281 case 'view':
14890de3 282 $entry = $store->retrieveOneById($id);
a1953dff 283
284 if ($entry != NULL) {
14890de3 285 $tpl->assign('id', $entry['id']);
286 $tpl->assign('url', $entry['url']);
287 $tpl->assign('title', $entry['title']);
494e21b4 288 $content = $entry['content'];
289 if (function_exists('tidy_parse_string')) {
290 $tidy = tidy_parse_string($content, array('indent'=>true, 'show-body-only' => true), 'UTF8');
291 $tidy->cleanRepair();
292 $content = $tidy->value;
293 }
294 $tpl->assign('content', $content);
14890de3 295 $tpl->assign('is_fav', $entry['is_fav']);
296 $tpl->assign('is_read', $entry['is_read']);
a1953dff 297 $tpl->assign('load_all_js', 0);
298 $tpl->draw('view');
299 }
300 else {
301 logm('error in view call : entry is NULL');
302 }
303
304 logm('view link #' . $id);
305 break;
306 default: # home view
14890de3 307 $entries = $store->getEntriesByView($view);
a1953dff 308
309 $tpl->assign('entries', $entries);
310
311 if ($full_head == 'yes') {
312 $tpl->assign('load_all_js', 1);
313 $tpl->draw('head');
314 $tpl->draw('home');
315 }
316
317 $tpl->draw('entries');
318
319 if ($full_head == 'yes') {
320 $tpl->draw('js');
321 $tpl->draw('footer');
322 }
323 break;
324 }
325}
326
263d6c67 327/**
328 * Appel d'une action (mark as fav, archive, delete)
329 */
e4d2565e 330function action_to_do($action, $url, $id = 0)
263d6c67 331{
f0070a15 332 global $store, $msg;
263d6c67 333
334 switch ($action)
335 {
336 case 'add':
337 if ($url == '')
338 continue;
339
6f87a197 340 if (MyTool::isUrl($url)) {
341 if($parametres_url = prepare_url($url)) {
29c6fd46 342 if ($store->add($url, $parametres_url['title'], $parametres_url['content'])) {
343 $last_id = $store->getLastId();
344 if (DOWNLOAD_PICTURES) {
345 $content = filtre_picture($parametres_url['content'], $url, $last_id);
346 }
347 $msg->add('s', 'the link has been added successfully');
6f87a197 348 }
29c6fd46 349 else {
350 $msg->add('e', 'error during insertion : the link wasn\'t added');
351 }
352 }
353 else {
354 $msg->add('e', 'error during url preparation : the link wasn\'t added');
355 logm('error during url preparation');
14890de3 356 }
1c182b6c 357 }
6f87a197 358 else {
29c6fd46 359 $msg->add('e', 'error during url preparation : the link is not valid');
6f87a197 360 logm($url . ' is not a valid url');
361 }
1c182b6c 362
a81cd067 363 logm('add link ' . $url);
263d6c67 364 break;
365 case 'delete':
29c6fd46 366 if ($store->deleteById($id)) {
367 remove_directory(ABS_PATH . $id);
368 $msg->add('s', 'the link has been deleted successfully');
369 logm('delete link #' . $id);
370 }
371 else {
372 $msg->add('e', 'the link wasn\'t deleted');
373 logm('error : can\'t delete link #' . $id);
374 }
263d6c67 375 break;
139769aa 376 case 'toggle_fav' :
14890de3 377 $store->favoriteById($id);
f0070a15 378 $msg->add('s', 'the favorite toggle has been done successfully');
e4d2565e 379 logm('mark as favorite link #' . $id);
139769aa 380 break;
381 case 'toggle_archive' :
14890de3 382 $store->archiveById($id);
f0070a15 383 $msg->add('s', 'the archive toggle has been done successfully');
e4d2565e 384 logm('archive link #' . $id);
139769aa 385 break;
263d6c67 386 default:
387 break;
388 }
cf3180f6 389}
390
713b2d69 391function logm($message)
392{
393 $t = strval(date('Y/m/d_H:i:s')).' - '.$_SERVER["REMOTE_ADDR"].' - '.strval($message)."\n";
a81cd067 394 file_put_contents('./log.txt',$t,FILE_APPEND);
cdcc8d25 395}