aboutsummaryrefslogtreecommitdiffhomepage
path: root/inc
diff options
context:
space:
mode:
authornicosomb <nicolas@loeuillet.org>2013-04-17 15:11:57 +0200
committernicosomb <nicolas@loeuillet.org>2013-04-17 15:11:57 +0200
commit1c182b6c9d74875054dc8efe75564ceab1bc1abe (patch)
tree97ab86be512a518ec3fac8d629211bb666741004 /inc
parent4456315b217f4eb253fc4403e43cdcb0012356e6 (diff)
downloadwallabag-1c182b6c9d74875054dc8efe75564ceab1bc1abe.tar.gz
wallabag-1c182b6c9d74875054dc8efe75564ceab1bc1abe.tar.zst
wallabag-1c182b6c9d74875054dc8efe75564ceab1bc1abe.zip
téléchargement des images d'un article
Diffstat (limited to 'inc')
-rw-r--r--inc/config.php5
-rw-r--r--inc/db.php7
-rwxr-xr-xinc/functions.php138
3 files changed, 134 insertions, 16 deletions
diff --git a/inc/config.php b/inc/config.php
index d468228d..51a25d36 100644
--- a/inc/config.php
+++ b/inc/config.php
@@ -7,7 +7,10 @@
7 * @copyright 2013 7 * @copyright 2013
8 * @license http://www.wtfpl.net/ see COPYING file 8 * @license http://www.wtfpl.net/ see COPYING file
9 */ 9 */
10
11if(!is_dir('db/')){mkdir('db/',0705);}
10define ('DB_PATH', 'sqlite:./db/poche.sqlite'); 12define ('DB_PATH', 'sqlite:./db/poche.sqlite');
13define ('ABS_PATH', 'archiveImg/');
11 14
12include 'db.php'; 15include 'db.php';
13include 'functions.php'; 16include 'functions.php';
@@ -44,4 +47,4 @@ $token = (isset ($_REQUEST['token'])) ? $_REQUEST['token'] : '';
44 47
45if ($action != '') { 48if ($action != '') {
46 action_to_do($action, $id, $url, $token); 49 action_to_do($action, $id, $url, $token);
47} \ No newline at end of file 50}
diff --git a/inc/db.php b/inc/db.php
index 705fc0c4..a1674ea7 100644
--- a/inc/db.php
+++ b/inc/db.php
@@ -10,12 +10,15 @@
10 10
11class db { 11class db {
12 var $handle; 12 var $handle;
13 function __construct($path) { 13 function __construct($path)
14 {
14 $this->handle = new PDO($path); 15 $this->handle = new PDO($path);
16 $this->handle->exec('CREATE TABLE IF NOT EXISTS "entries" ("id" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL UNIQUE , "title" VARCHAR, "url" VARCHAR UNIQUE , "is_read" INTEGER DEFAULT 0, "is_fav" INTEGER DEFAULT 0, "content" BLOB)');
15 $this->handle->setAttribute(PDO::ATTR_ERRMODE, PDO::ERRMODE_EXCEPTION); 17 $this->handle->setAttribute(PDO::ATTR_ERRMODE, PDO::ERRMODE_EXCEPTION);
16 } 18 }
17 19
18 public function getHandle() { 20 public function getHandle()
21 {
19 return $this->handle; 22 return $this->handle;
20 } 23 }
21} \ No newline at end of file 24} \ No newline at end of file
diff --git a/inc/functions.php b/inc/functions.php
index 02d2c9dc..37abafd4 100755
--- a/inc/functions.php
+++ b/inc/functions.php
@@ -16,8 +16,9 @@ function get_poche_url()
16} 16}
17 17
18// function define to retrieve url content 18// function define to retrieve url content
19function get_external_file($url, $timeout) 19function get_external_file($url)
20{ 20{
21 $timeout = 15;
21 // spoofing FireFox 18.0 22 // spoofing FireFox 18.0
22 $useragent="Mozilla/5.0 (Windows NT 5.1; rv:18.0) Gecko/20100101 Firefox/18.0"; 23 $useragent="Mozilla/5.0 (Windows NT 5.1; rv:18.0) Gecko/20100101 Firefox/18.0";
23 24
@@ -45,10 +46,7 @@ function get_external_file($url, $timeout)
45 } else { 46 } else {
46 47
47 // create http context and add timeout and user-agent 48 // create http context and add timeout and user-agent
48 $context = stream_context_create(array('http'=>array('timeout' => $timeout, // Timeout : time until we stop waiting for the response. 49 $context = stream_context_create(array('http'=>array('timeout' => $timeout,'header'=> "User-Agent: ".$useragent,/*spoot Mozilla Firefox*/'follow_location' => true)));
49 'header'=> "User-Agent: ".$useragent, // spoot Mozilla Firefox
50 'follow_location' => true
51 )));
52 50
53 // only download page lesser than 4MB 51 // only download page lesser than 4MB
54 $data = @file_get_contents($url, false, $context, -1, 4000000); // We download at most 4 MB from source. 52 $data = @file_get_contents($url, false, $context, -1, 4000000); // We download at most 4 MB from source.
@@ -90,7 +88,7 @@ function get_external_file($url, $timeout)
90/** 88/**
91 * Préparation de l'URL avec récupération du contenu avant insertion en base 89 * Préparation de l'URL avec récupération du contenu avant insertion en base
92 */ 90 */
93function prepare_url($url) 91function prepare_url($url, $id)
94{ 92{
95 $parametres = array(); 93 $parametres = array();
96 $url = html_entity_decode(trim($url)); 94 $url = html_entity_decode(trim($url));
@@ -112,19 +110,127 @@ function prepare_url($url)
112 $r->convertLinksToFootnotes = TRUE; 110 $r->convertLinksToFootnotes = TRUE;
113 if($r->init()) 111 if($r->init())
114 { 112 {
115 $title = $r->articleTitle->innerHTML; 113 $content = $r->articleContent->innerHTML;
114 $parametres['title'] = $r->articleTitle->innerHTML;
115 $parametres['content'] = filtre_picture($content, $url, $id);
116 return $parametres;
116 } 117 }
117 } 118 }
118 119
119 $parametres['title'] = $title; 120 logm('error during url preparation');
120 $parametres['content'] = $r->articleContent->innerHTML; 121 return FALSE;
122}
123
124/**
125 * On modifie les URLS des images dans le corps de l'article
126 */
127function filtre_picture($content, $url, $id)
128{
129 $matches = array();
130 preg_match_all('#<\s*(img)[^>]+src="([^"]*)"[^>]*>#Si', $content, $matches, PREG_SET_ORDER);
131 foreach($matches as $i => $link)
132 {
133 $link[1] = trim($link[1]);
134 if (!preg_match('#^(([a-z]+://)|(\#))#', $link[1]) )
135 {
136 $absolute_path = get_absolute_link($link[2],$url);
137 $filename = basename(parse_url($absolute_path, PHP_URL_PATH));
138 $directory = create_assets_directory($id);
139 $fullpath = $directory . '/' . $filename;
140 download_pictures($absolute_path, $fullpath);
141 $content = str_replace($matches[$i][2], $fullpath, $content);
142 }
143
144 }
145
146 return $content;
147}
148
149/**
150 * Retourne le lien absolu
151 */
152function get_absolute_link($relative_link, $url)
153{
154 /* return if already absolute URL */
155 if (parse_url($relative_link, PHP_URL_SCHEME) != '') return $relative_link;
156
157 /* queries and anchors */
158 if ($relative_link[0]=='#' || $relative_link[0]=='?') return $url . $relative_link;
159
160 /* parse base URL and convert to local variables:
161 $scheme, $host, $path */
162 extract(parse_url($url));
163
164 /* remove non-directory element from path */
165 $path = preg_replace('#/[^/]*$#', '', $path);
166
167 /* destroy path if relative url points to root */
168 if ($relative_link[0] == '/') $path = '';
169
170 /* dirty absolute URL */
171 $abs = $host . $path . '/' . $relative_link;
172
173 /* replace '//' or '/./' or '/foo/../' with '/' */
174 $re = array('#(/\.?/)#', '#/(?!\.\.)[^/]+/\.\./#');
175 for($n=1; $n>0; $abs=preg_replace($re, '/', $abs, -1, $n)) {}
176
177 /* absolute URL is ready! */
178 return $scheme.'://'.$abs;
179}
180
181/**
182 * Téléchargement des images
183 */
184
185function download_pictures($absolute_path, $fullpath)
186{
187 $rawdata = get_external_file($absolute_path);
121 188
122 return $parametres; 189 if(file_exists($fullpath)) {
190 unlink($fullpath);
191 }
192 $fp = fopen($fullpath, 'x');
193 fwrite($fp, $rawdata);
194 fclose($fp);
195}
196
197/**
198 * Crée un répertoire de médias pour l'article
199 */
200function create_assets_directory($id)
201{
202 $assets_path = ABS_PATH;
203 if(!is_dir($assets_path)) {
204 mkdir($assets_path, 0705);
205 }
206
207 $article_directory = $assets_path . $id;
208 if(!is_dir($article_directory)) {
209 mkdir($article_directory, 0705);
210 }
211
212 return $article_directory;
213}
214
215/**
216 * Suppression du répertoire d'images
217 */
218function remove_directory($directory)
219{
220 if(is_dir($directory)) {
221 $files = array_diff(scandir($directory), array('.','..'));
222 foreach ($files as $file) {
223 // FIXME c'est quoi delTree ??
224 (is_dir("$directory/$file")) ? delTree("$directory/$file") : unlink("$directory/$file");
225 }
226 return rmdir($directory);
227 }
123} 228}
124 229
125/** 230/**
126 * Appel d'une action (mark as fav, archive, delete) 231 * Appel d'une action (mark as fav, archive, delete)
127 */ 232 */
233
128function action_to_do($action, $id, $url, $token) 234function action_to_do($action, $id, $url, $token)
129{ 235{
130 global $db; 236 global $db;
@@ -135,13 +241,19 @@ function action_to_do($action, $id, $url, $token)
135 if ($url == '') 241 if ($url == '')
136 continue; 242 continue;
137 243
138 $parametres_url = prepare_url($url); 244 $req = $db->getHandle()->query("SELECT id FROM entries ORDER BY id DESC");
139 $sql_action = 'INSERT INTO entries ( url, title, content ) VALUES (?, ?, ?)'; 245 $id = $req->fetchColumn()+1;
140 $params_action = array($url, $parametres_url['title'], $parametres_url['content']); 246
247 if($parametres_url = prepare_url($url, $id)) {
248 $sql_action = 'INSERT INTO entries ( id, url, title, content ) VALUES (?,?, ?, ?)';
249 $params_action = array($id,$url, $parametres_url['title'], $parametres_url['content']);
250 }
251
141 logm('add link ' . $url); 252 logm('add link ' . $url);
142 break; 253 break;
143 case 'delete': 254 case 'delete':
144 if (verif_token($token)) { 255 if (verif_token($token)) {
256 remove_directory(ABS_PATH . $id);
145 $sql_action = "DELETE FROM entries WHERE id=?"; 257 $sql_action = "DELETE FROM entries WHERE id=?";
146 $params_action = array($id); 258 $params_action = array($id);
147 logm('delete link #' . $id); 259 logm('delete link #' . $id);