aboutsummaryrefslogtreecommitdiffhomepage
path: root/inc/poche
diff options
context:
space:
mode:
authorNicolas Lœuillet <nicolas.loeuillet@gmail.com>2013-08-02 22:40:51 +0200
committerNicolas Lœuillet <nicolas.loeuillet@gmail.com>2013-08-02 22:40:51 +0200
commita4565e88edbc8e3bd092a475469769c86a4c350c (patch)
treea6a3c935b03a23ff87575c8c315cf8ba78fe68c2 /inc/poche
parentf6c9baab3efeec1d0efa151e276fc08d5b58f9e9 (diff)
downloadwallabag-a4565e88edbc8e3bd092a475469769c86a4c350c.tar.gz
wallabag-a4565e88edbc8e3bd092a475469769c86a4c350c.tar.zst
wallabag-a4565e88edbc8e3bd092a475469769c86a4c350c.zip
add Twig & refactor poche
Diffstat (limited to 'inc/poche')
-rw-r--r--inc/poche/pocheCore.php257
-rw-r--r--inc/poche/pochePictures.php114
-rw-r--r--inc/poche/pocheTool.class.php124
3 files changed, 495 insertions, 0 deletions
diff --git a/inc/poche/pocheCore.php b/inc/poche/pocheCore.php
new file mode 100644
index 00000000..52c603ac
--- /dev/null
+++ b/inc/poche/pocheCore.php
@@ -0,0 +1,257 @@
1<?php
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 */
10
11function encode_string($string)
12{
13 return sha1($string . SALT);
14}
15
16function get_external_file($url)
17{
18 $timeout = 15;
19 $useragent = "Mozilla/5.0 (Windows NT 5.1; rv:18.0) Gecko/20100101 Firefox/18.0";
20
21 if (in_array ('curl', get_loaded_extensions())) {
22 # Fetch feed from URL
23 $curl = curl_init();
24 curl_setopt($curl, CURLOPT_URL, $url);
25 curl_setopt($curl, CURLOPT_TIMEOUT, $timeout);
26 curl_setopt($curl, CURLOPT_FOLLOWLOCATION, true);
27 curl_setopt($curl, CURLOPT_RETURNTRANSFER, true);
28 curl_setopt($curl, CURLOPT_HEADER, false);
29
30 # for ssl, do not verified certificate
31 curl_setopt($curl, CURLOPT_SSL_VERIFYPEER, FALSE);
32 curl_setopt($curl, CURLOPT_AUTOREFERER, TRUE );
33
34 # FeedBurner requires a proper USER-AGENT...
35 curl_setopt($curl, CURL_HTTP_VERSION_1_1, true);
36 curl_setopt($curl, CURLOPT_ENCODING, "gzip, deflate");
37 curl_setopt($curl, CURLOPT_USERAGENT, $useragent);
38
39 $data = curl_exec($curl);
40 $httpcode = curl_getinfo($curl, CURLINFO_HTTP_CODE);
41 $httpcodeOK = isset($httpcode) and ($httpcode == 200 or $httpcode == 301);
42 curl_close($curl);
43 } else {
44 # create http context and add timeout and user-agent
45 $context = stream_context_create(
46 array(
47 'http' => array(
48 'timeout' => $timeout,
49 'header' => "User-Agent: " . $useragent,
50 'follow_location' => true
51 ),
52 'ssl' => array(
53 'verify_peer' => false,
54 'allow_self_signed' => true
55 )
56 )
57 );
58
59 # only download page lesser than 4MB
60 $data = @file_get_contents($url, false, $context, -1, 4000000);
61
62 if (isset($http_response_header) and isset($http_response_header[0])) {
63 $httpcodeOK = isset($http_response_header) and isset($http_response_header[0]) and ((strpos($http_response_header[0], '200 OK') !== FALSE) or (strpos($http_response_header[0], '301 Moved Permanently') !== FALSE));
64 }
65 }
66
67 # if response is not empty and response is OK
68 if (isset($data) and isset($httpcodeOK) and $httpcodeOK) {
69
70 # take charset of page and get it
71 preg_match('#<meta .*charset=.*>#Usi', $data, $meta);
72
73 # if meta tag is found
74 if (!empty($meta[0])) {
75 preg_match('#charset="?(.*)"#si', $meta[0], $encoding);
76 # if charset is found set it otherwise, set it to utf-8
77 $html_charset = (!empty($encoding[1])) ? strtolower($encoding[1]) : 'utf-8';
78 } else {
79 $html_charset = 'utf-8';
80 $encoding[1] = '';
81 }
82
83 # replace charset of url to charset of page
84 $data = str_replace('charset=' . $encoding[1], 'charset=' . $html_charset, $data);
85
86 return $data;
87 }
88 else {
89 return FALSE;
90 }
91}
92
93function fetch_url_content($url)
94{
95 $url = base64_decode($url);
96 if (pocheTool::isUrl($url)) {
97 $url = pocheTool::cleanURL($url);
98 $html = Encoding::toUTF8(get_external_file($url));
99
100 # if get_external_file if not able to retrieve HTTPS content, try the same URL with HTTP protocol
101 if (!preg_match('!^https?://!i', $url) && (!isset($html) || strlen($html) <= 0)) {
102 $url = 'http://' . $url;
103 $html = Encoding::toUTF8(get_external_file($url));
104 }
105
106 if (function_exists('tidy_parse_string')) {
107 $tidy = tidy_parse_string($html, array(), 'UTF8');
108 $tidy->cleanRepair();
109 $html = $tidy->value;
110 }
111
112 $parameters = array();
113 if (isset($html) and strlen($html) > 0)
114 {
115 $readability = new Readability($html, $url);
116 $readability->convertLinksToFootnotes = CONVERT_LINKS_FOOTNOTES;
117 $readability->revertForcedParagraphElements = REVERT_FORCED_PARAGRAPH_ELEMENTS;
118
119 if($readability->init())
120 {
121 $content = $readability->articleContent->innerHTML;
122 $parameters['title'] = $readability->articleTitle->innerHTML;
123 $parameters['content'] = $content;
124
125 return $parameters;
126 }
127 }
128 }
129 else {
130 #$msg->add('e', _('error during url preparation : the link is not valid'));
131 pocheTool::logm($url . ' is not a valid url');
132 }
133
134 return FALSE;
135}
136
137function display_view($view, $id = 0, $full_head = 'yes')
138{
139 global $tpl, $store, $msg;
140
141 switch ($view)
142 {
143 case 'export':
144 $entries = $store->retrieveAll();
145 $tpl->assign('export', pocheTool::renderJson($entries));
146 $tpl->draw('export');
147 pocheTool::logm('export view');
148 break;
149 case 'config':
150 $tpl->assign('load_all_js', 0);
151 $tpl->draw('head');
152 $tpl->draw('home');
153 $tpl->draw('config');
154 $tpl->draw('js');
155 $tpl->draw('footer');
156 pocheTool::logm('config view');
157 break;
158 case 'view':
159 $entry = $store->retrieveOneById($id);
160
161 if ($entry != NULL) {
162 $tpl->assign('id', $entry['id']);
163 $tpl->assign('url', $entry['url']);
164 $tpl->assign('title', $entry['title']);
165 $content = $entry['content'];
166 if (function_exists('tidy_parse_string')) {
167 $tidy = tidy_parse_string($content, array('indent'=>true, 'show-body-only' => true), 'UTF8');
168 $tidy->cleanRepair();
169 $content = $tidy->value;
170 }
171 $tpl->assign('content', $content);
172 $tpl->assign('is_fav', $entry['is_fav']);
173 $tpl->assign('is_read', $entry['is_read']);
174 $tpl->assign('load_all_js', 0);
175 $tpl->draw('view');
176 }
177 else {
178 pocheTool::logm('error in view call : entry is NULL');
179 }
180
181 pocheTool::logm('view link #' . $id);
182 break;
183 default: # home view
184 $entries = $store->getEntriesByView($view);
185
186 $tpl->assign('entries', $entries);
187
188 if ($full_head == 'yes') {
189 $tpl->assign('load_all_js', 1);
190 $tpl->draw('head');
191 $tpl->draw('home');
192 }
193
194 $tpl->draw('entries');
195
196 if ($full_head == 'yes') {
197 $tpl->draw('js');
198 $tpl->draw('footer');
199 }
200 break;
201 }
202}
203
204/**
205 * Appel d'une action (mark as fav, archive, delete)
206 */
207function action_to_do($action, $url, $id = 0)
208{
209 global $store, $msg;
210
211 switch ($action)
212 {
213 case 'add':
214 if($parametres_url = fetch_url_content($url)) {
215 if ($store->add($url, $parametres_url['title'], $parametres_url['content'])) {
216 pocheTool::logm('add link ' . $url);
217 $last_id = $store->getLastId();
218 if (DOWNLOAD_PICTURES) {
219 $content = filtre_picture($parametres_url['content'], $url, $last_id);
220 }
221 #$msg->add('s', _('the link has been added successfully'));
222 }
223 else {
224 #$msg->add('e', _('error during insertion : the link wasn\'t added'));
225 pocheTool::logm('error during insertion : the link wasn\'t added');
226 }
227 }
228 else {
229 #$msg->add('e', _('error during url preparation : the link wasn\'t added'));
230 pocheTool::logm('error during content fetch');
231 }
232 break;
233 case 'delete':
234 if ($store->deleteById($id)) {
235 if (DOWNLOAD_PICTURES) {
236 remove_directory(ABS_PATH . $id);
237 }
238 #$msg->add('s', _('the link has been deleted successfully'));
239 pocheTool::logm('delete link #' . $id);
240 }
241 else {
242 #$msg->add('e', _('the link wasn\'t deleted'));
243 pocheTool::logm('error : can\'t delete link #' . $id);
244 }
245 break;
246 case 'toggle_fav' :
247 $store->favoriteById($id);
248 pocheTool::logm('mark as favorite link #' . $id);
249 break;
250 case 'toggle_archive' :
251 $store->archiveById($id);
252 pocheTool::logm('archive link #' . $id);
253 break;
254 default:
255 break;
256 }
257}
diff --git a/inc/poche/pochePictures.php b/inc/poche/pochePictures.php
new file mode 100644
index 00000000..bfc80657
--- /dev/null
+++ b/inc/poche/pochePictures.php
@@ -0,0 +1,114 @@
1<?php
2/**
3 * poche, a read it later open source system
4 *
5 * @category poche
6 * @author Nicolas Lœuillet <support@inthepoche.com>
7 * @copyright 2013
8 * @license http://www.wtfpl.net/ see COPYING file
9 */
10
11/**
12 * On modifie les URLS des images dans le corps de l'article
13 */
14function filtre_picture($content, $url, $id)
15{
16 $matches = array();
17 preg_match_all('#<\s*(img)[^>]+src="([^"]*)"[^>]*>#Si', $content, $matches, PREG_SET_ORDER);
18 foreach($matches as $i => $link)
19 {
20 $link[1] = trim($link[1]);
21 if (!preg_match('#^(([a-z]+://)|(\#))#', $link[1]) )
22 {
23 $absolute_path = get_absolute_link($link[2],$url);
24 $filename = basename(parse_url($absolute_path, PHP_URL_PATH));
25 $directory = create_assets_directory($id);
26 $fullpath = $directory . '/' . $filename;
27 download_pictures($absolute_path, $fullpath);
28 $content = str_replace($matches[$i][2], $fullpath, $content);
29 }
30
31 }
32
33 return $content;
34}
35
36/**
37 * Retourne le lien absolu
38 */
39function get_absolute_link($relative_link, $url)
40{
41 /* return if already absolute URL */
42 if (parse_url($relative_link, PHP_URL_SCHEME) != '') return $relative_link;
43
44 /* queries and anchors */
45 if ($relative_link[0]=='#' || $relative_link[0]=='?') return $url . $relative_link;
46
47 /* parse base URL and convert to local variables:
48 $scheme, $host, $path */
49 extract(parse_url($url));
50
51 /* remove non-directory element from path */
52 $path = preg_replace('#/[^/]*$#', '', $path);
53
54 /* destroy path if relative url points to root */
55 if ($relative_link[0] == '/') $path = '';
56
57 /* dirty absolute URL */
58 $abs = $host . $path . '/' . $relative_link;
59
60 /* replace '//' or '/./' or '/foo/../' with '/' */
61 $re = array('#(/\.?/)#', '#/(?!\.\.)[^/]+/\.\./#');
62 for($n=1; $n>0; $abs=preg_replace($re, '/', $abs, -1, $n)) {}
63
64 /* absolute URL is ready! */
65 return $scheme.'://'.$abs;
66}
67
68/**
69 * Téléchargement des images
70 */
71
72function download_pictures($absolute_path, $fullpath)
73{
74 $rawdata = get_external_file($absolute_path);
75
76 if(file_exists($fullpath)) {
77 unlink($fullpath);
78 }
79 $fp = fopen($fullpath, 'x');
80 fwrite($fp, $rawdata);
81 fclose($fp);
82}
83
84/**
85 * Crée un répertoire de médias pour l'article
86 */
87function create_assets_directory($id)
88{
89 $assets_path = ABS_PATH;
90 if(!is_dir($assets_path)) {
91 mkdir($assets_path, 0705);
92 }
93
94 $article_directory = $assets_path . $id;
95 if(!is_dir($article_directory)) {
96 mkdir($article_directory, 0705);
97 }
98
99 return $article_directory;
100}
101
102/**
103 * Suppression du répertoire d'images
104 */
105function remove_directory($directory)
106{
107 if(is_dir($directory)) {
108 $files = array_diff(scandir($directory), array('.','..'));
109 foreach ($files as $file) {
110 (is_dir("$directory/$file")) ? remove_directory("$directory/$file") : unlink("$directory/$file");
111 }
112 return rmdir($directory);
113 }
114}
diff --git a/inc/poche/pocheTool.class.php b/inc/poche/pocheTool.class.php
new file mode 100644
index 00000000..cade115e
--- /dev/null
+++ b/inc/poche/pocheTool.class.php
@@ -0,0 +1,124 @@
1<?php
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 */
10
11class pocheTools
12{
13 public static function initPhp()
14 {
15 define('START_TIME', microtime(true));
16
17 if (phpversion() < 5) {
18 die(_('Oops, it seems you don\'t have PHP 5.'));
19 }
20
21 error_reporting(E_ALL);
22
23 function stripslashesDeep($value) {
24 return is_array($value)
25 ? array_map('stripslashesDeep', $value)
26 : stripslashes($value);
27 }
28
29 if (get_magic_quotes_gpc()) {
30 $_POST = array_map('stripslashesDeep', $_POST);
31 $_GET = array_map('stripslashesDeep', $_GET);
32 $_COOKIE = array_map('stripslashesDeep', $_COOKIE);
33 }
34
35 ob_start();
36 register_shutdown_function('ob_end_flush');
37 }
38
39 public static function isUrl($url)
40 {
41 $pattern = '|^http(s)?://[a-z0-9-]+(.[a-z0-9-]+)*(:[0-9]+)?(/.*)?$|i';
42
43 return preg_match($pattern, $url);
44 }
45
46 public static function getUrl()
47 {
48 $https = (!empty($_SERVER['HTTPS'])
49 && (strtolower($_SERVER['HTTPS']) == 'on'))
50 || (isset($_SERVER["SERVER_PORT"])
51 && $_SERVER["SERVER_PORT"] == '443'); // HTTPS detection.
52 $serverport = (!isset($_SERVER["SERVER_PORT"])
53 || $_SERVER["SERVER_PORT"] == '80'
54 || ($https && $_SERVER["SERVER_PORT"] == '443')
55 ? '' : ':' . $_SERVER["SERVER_PORT"]);
56
57 $scriptname = str_replace('/index.php', '/', $_SERVER["SCRIPT_NAME"]);
58
59 if (!isset($_SERVER["SERVER_NAME"])) {
60 return $scriptname;
61 }
62
63 return 'http' . ($https ? 's' : '') . '://'
64 . $_SERVER["SERVER_NAME"] . $serverport . $scriptname;
65 }
66
67 public static function redirect($url = '')
68 {
69 if ($url === '') {
70 $url = (empty($_SERVER['HTTP_REFERER'])?'?':$_SERVER['HTTP_REFERER']);
71 if (isset($_POST['returnurl'])) {
72 $url = $_POST['returnurl'];
73 }
74 }
75
76 # prevent loop
77 if (empty($url) || parse_url($url, PHP_URL_QUERY) === $_SERVER['QUERY_STRING']) {
78 $url = pocheTool::getUrl();
79 }
80
81 if (substr($url, 0, 1) !== '?') {
82 $ref = pocheTool::getUrl();
83 if (substr($url, 0, strlen($ref)) !== $ref) {
84 $url = $ref;
85 }
86 }
87 header('Location: '.$url);
88 exit();
89 }
90
91 public static function cleanURL($url)
92 {
93
94 $url = html_entity_decode(trim($url));
95
96 $stuff = strpos($url,'&utm_source=');
97 if ($stuff !== FALSE)
98 $url = substr($url, 0, $stuff);
99 $stuff = strpos($url,'?utm_source=');
100 if ($stuff !== FALSE)
101 $url = substr($url, 0, $stuff);
102 $stuff = strpos($url,'#xtor=RSS-');
103 if ($stuff !== FALSE)
104 $url = substr($url, 0, $stuff);
105
106 return $url;
107 }
108
109 public static function renderJson($data)
110 {
111 header('Cache-Control: no-cache, must-revalidate');
112 header('Expires: Sat, 26 Jul 1997 05:00:00 GMT');
113 header('Content-type: application/json; charset=UTF-8');
114
115 echo json_encode($data);
116 exit();
117 }
118
119 public static function logm($message)
120 {
121 $t = strval(date('Y/m/d_H:i:s')).' - '.$_SERVER["REMOTE_ADDR"].' - '.strval($message)."\n";
122 file_put_contents('./log.txt',$t,FILE_APPEND);
123 }
124} \ No newline at end of file