aboutsummaryrefslogtreecommitdiffhomepage
path: root/inc/3rdparty
diff options
context:
space:
mode:
Diffstat (limited to 'inc/3rdparty')
-rw-r--r--inc/3rdparty/Session.class.php40
-rw-r--r--inc/3rdparty/class.messages.php3
-rw-r--r--inc/3rdparty/libraries/feedwriter/FeedItem.php1
-rwxr-xr-x[-rw-r--r--]inc/3rdparty/libraries/feedwriter/FeedWriter.php168
-rwxr-xr-xinc/3rdparty/makefulltextfeed.php353
-rwxr-xr-xinc/3rdparty/makefulltextfeedHelpers.php355
6 files changed, 478 insertions, 442 deletions
diff --git a/inc/3rdparty/Session.class.php b/inc/3rdparty/Session.class.php
index b30a31f3..59dfbe67 100644
--- a/inc/3rdparty/Session.class.php
+++ b/inc/3rdparty/Session.class.php
@@ -31,9 +31,9 @@ class Session
31 public static $sessionName = ''; 31 public static $sessionName = '';
32 // If the user does not access any page within this time, 32 // If the user does not access any page within this time,
33 // his/her session is considered expired (3600 sec. = 1 hour) 33 // his/her session is considered expired (3600 sec. = 1 hour)
34 public static $inactivityTimeout = 86400; 34 public static $inactivityTimeout = 3600;
35 // Extra timeout for long sessions (if enabled) (82800 sec. = 23 hours) 35 // Extra timeout for long sessions (if enabled) (82800 sec. = 23 hours)
36 public static $longSessionTimeout = 31536000; 36 public static $longSessionTimeout = 7776000; // 7776000 = 90 days
37 // If you get disconnected often or if your IP address changes often. 37 // If you get disconnected often or if your IP address changes often.
38 // Let you disable session cookie hijacking protection 38 // Let you disable session cookie hijacking protection
39 public static $disableSessionProtection = false; 39 public static $disableSessionProtection = false;
@@ -48,8 +48,13 @@ class Session
48 /** 48 /**
49 * Initialize session 49 * Initialize session
50 */ 50 */
51 public static function init() 51 public static function init($longlastingsession = false)
52 { 52 {
53 //check if session name is correct
54 if ( (session_id() && !empty(self::$sessionName) && session_name()!=self::$sessionName) || $longlastingsession ) {
55 session_destroy();
56 }
57
53 // Force cookie path (but do not change lifetime) 58 // Force cookie path (but do not change lifetime)
54 $cookie = session_get_cookie_params(); 59 $cookie = session_get_cookie_params();
55 // Default cookie expiration and path. 60 // Default cookie expiration and path.
@@ -61,12 +66,22 @@ class Session
61 if (isset($_SERVER["HTTPS"]) && $_SERVER["HTTPS"] == "on") { 66 if (isset($_SERVER["HTTPS"]) && $_SERVER["HTTPS"] == "on") {
62 $ssl = true; 67 $ssl = true;
63 } 68 }
64 session_set_cookie_params($cookie['lifetime'], $cookiedir, $_SERVER['HTTP_HOST'], $ssl); 69
70 if ( $longlastingsession ) {
71 session_set_cookie_params(self::$longSessionTimeout, $cookiedir, null, $ssl, true);
72 }
73 else {
74 session_set_cookie_params(0, $cookiedir, null, $ssl, true);
75 }
76 //set server side valid session timeout
77 //WARNING! this may not work in shared session environment. See http://www.php.net/manual/en/session.configuration.php#ini.session.gc-maxlifetime about min value: it can be set in any application
78 ini_set('session.gc_maxlifetime', self::$longSessionTimeout);
79
65 // Use cookies to store session. 80 // Use cookies to store session.
66 ini_set('session.use_cookies', 1); 81 ini_set('session.use_cookies', 1);
67 // Force cookies for session (phpsessionID forbidden in URL) 82 // Force cookies for session (phpsessionID forbidden in URL)
68 ini_set('session.use_only_cookies', 1); 83 ini_set('session.use_only_cookies', 1);
69 if (!session_id()) { 84 if ( !session_id() ) {
70 // Prevent php to use sessionID in URL if cookies are disabled. 85 // Prevent php to use sessionID in URL if cookies are disabled.
71 ini_set('session.use_trans_sid', false); 86 ini_set('session.use_trans_sid', false);
72 if (!empty(self::$sessionName)) { 87 if (!empty(self::$sessionName)) {
@@ -115,6 +130,9 @@ class Session
115 if (self::banCanLogin()) { 130 if (self::banCanLogin()) {
116 if ($login === $loginTest && $password === $passwordTest) { 131 if ($login === $loginTest && $password === $passwordTest) {
117 self::banLoginOk(); 132 self::banLoginOk();
133
134 self::init($longlastingsession);
135
118 // Generate unique random number to sign forms (HMAC) 136 // Generate unique random number to sign forms (HMAC)
119 $_SESSION['uid'] = sha1(uniqid('', true).'_'.mt_rand()); 137 $_SESSION['uid'] = sha1(uniqid('', true).'_'.mt_rand());
120 $_SESSION['ip'] = self::_allIPs(); 138 $_SESSION['ip'] = self::_allIPs();
@@ -135,6 +153,7 @@ class Session
135 self::banLoginFailed(); 153 self::banLoginFailed();
136 } 154 }
137 155
156 self::init();
138 return false; 157 return false;
139 } 158 }
140 159
@@ -143,7 +162,14 @@ class Session
143 */ 162 */
144 public static function logout() 163 public static function logout()
145 { 164 {
146 unset($_SESSION['uid'],$_SESSION['ip'],$_SESSION['expires_on'],$_SESSION['tokens'], $_SESSION['login'], $_SESSION['pass'], $_SESSION['longlastingsession'], $_SESSION['poche_user']); 165 // unset($_SESSION['uid'],$_SESSION['ip'],$_SESSION['expires_on'],$_SESSION['tokens'], $_SESSION['login'], $_SESSION['pass'], $_SESSION['longlastingsession'], $_SESSION['poche_user']);
166
167 // Destruction du cookie (le code peut paraître complexe mais c'est pour être certain de reprendre les mêmes paramètres)
168 $args = array_merge(array(session_name(), ''), array_values(session_get_cookie_params()));
169 $args[2] = time() - 3600;
170 call_user_func_array('setcookie', $args);
171 // Suppression physique de la session
172 session_destroy();
147 } 173 }
148 174
149 /** 175 /**
@@ -157,7 +183,7 @@ class Session
157 || (self::$disableSessionProtection === false 183 || (self::$disableSessionProtection === false
158 && $_SESSION['ip'] !== self::_allIPs()) 184 && $_SESSION['ip'] !== self::_allIPs())
159 || time() >= $_SESSION['expires_on']) { 185 || time() >= $_SESSION['expires_on']) {
160 self::logout(); 186 //self::logout();
161 187
162 return false; 188 return false;
163 } 189 }
diff --git a/inc/3rdparty/class.messages.php b/inc/3rdparty/class.messages.php
index e60bd3a1..27c28f43 100644
--- a/inc/3rdparty/class.messages.php
+++ b/inc/3rdparty/class.messages.php
@@ -59,6 +59,7 @@ class Messages {
59 $this->msgId = md5(uniqid()); 59 $this->msgId = md5(uniqid());
60 60
61 // Create the session array if it doesnt already exist 61 // Create the session array if it doesnt already exist
62 settype($_SESSION, 'array');
62 if( !array_key_exists('flash_messages', $_SESSION) ) $_SESSION['flash_messages'] = array(); 63 if( !array_key_exists('flash_messages', $_SESSION) ) $_SESSION['flash_messages'] = array();
63 64
64 } 65 }
@@ -228,4 +229,4 @@ class Messages {
228 229
229 230
230} // end class 231} // end class
231?> \ No newline at end of file 232?>
diff --git a/inc/3rdparty/libraries/feedwriter/FeedItem.php b/inc/3rdparty/libraries/feedwriter/FeedItem.php
index 9373deeb..0eae5e08 100644
--- a/inc/3rdparty/libraries/feedwriter/FeedItem.php
+++ b/inc/3rdparty/libraries/feedwriter/FeedItem.php
@@ -156,6 +156,7 @@
156 if($this->version == RSS2 || $this->version == RSS1) 156 if($this->version == RSS2 || $this->version == RSS1)
157 { 157 {
158 $this->setElement('link', $link); 158 $this->setElement('link', $link);
159 $this->setElement('guid', $link);
159 } 160 }
160 else 161 else
161 { 162 {
diff --git a/inc/3rdparty/libraries/feedwriter/FeedWriter.php b/inc/3rdparty/libraries/feedwriter/FeedWriter.php
index adb2526c..5d16e765 100644..100755
--- a/inc/3rdparty/libraries/feedwriter/FeedWriter.php
+++ b/inc/3rdparty/libraries/feedwriter/FeedWriter.php
@@ -9,9 +9,9 @@ define('JSONP', 3, true);
9 * Genarate RSS2 or JSON (original: RSS 1.0, RSS2.0 and ATOM Feed) 9 * Genarate RSS2 or JSON (original: RSS 1.0, RSS2.0 and ATOM Feed)
10 * 10 *
11 * Modified for FiveFilters.org's Full-Text RSS project 11 * Modified for FiveFilters.org's Full-Text RSS project
12 * to allow for inclusion of hubs, JSON output. 12 * to allow for inclusion of hubs, JSON output.
13 * Stripped RSS1 and ATOM support. 13 * Stripped RSS1 and ATOM support.
14 * 14 *
15 * @package UnivarselFeedWriter 15 * @package UnivarselFeedWriter
16 * @author Anis uddin Ahmad <anisniit@gmail.com> 16 * @author Anis uddin Ahmad <anisniit@gmail.com>
17 * @link http://www.ajaxray.com/projects/rss 17 * @link http://www.ajaxray.com/projects/rss
@@ -26,32 +26,32 @@ define('JSONP', 3, true);
26 private $CDATAEncoding = array(); // The tag names which have to encoded as CDATA 26 private $CDATAEncoding = array(); // The tag names which have to encoded as CDATA
27 private $xsl = null; // stylesheet to render RSS (used by Chrome) 27 private $xsl = null; // stylesheet to render RSS (used by Chrome)
28 private $json = null; // JSON object 28 private $json = null; // JSON object
29 29
30 private $version = null; 30 private $version = null;
31 31
32 /** 32 /**
33 * Constructor 33 * Constructor
34 * 34 *
35 * @param constant the version constant (RSS2 or JSON). 35 * @param constant the version constant (RSS2 or JSON).
36 */ 36 */
37 function __construct($version = RSS2) 37 function __construct($version = RSS2)
38 { 38 {
39 $this->version = $version; 39 $this->version = $version;
40 40
41 // Setting default value for assential channel elements 41 // Setting default value for assential channel elements
42 $this->channels['title'] = $version . ' Feed'; 42 $this->channels['title'] = $version . ' Feed';
43 $this->channels['link'] = 'http://www.ajaxray.com/blog'; 43 $this->channels['link'] = 'http://www.ajaxray.com/blog';
44 44
45 //Tag names to encode in CDATA 45 //Tag names to encode in CDATA
46 $this->CDATAEncoding = array('description', 'content:encoded', 'content', 'subtitle', 'summary'); 46 $this->CDATAEncoding = array('description', 'content:encoded', 'content', 'subtitle', 'summary');
47 } 47 }
48 48
49 public function setFormat($format) { 49 public function setFormat($format) {
50 $this->version = $format; 50 $this->version = $format;
51 } 51 }
52 52
53 // Start # public functions --------------------------------------------- 53 // Start # public functions ---------------------------------------------
54 54
55 /** 55 /**
56 * Set a channel element 56 * Set a channel element
57 * @access public 57 * @access public
@@ -63,11 +63,11 @@ define('JSONP', 3, true);
63 { 63 {
64 $this->channels[$elementName] = $content ; 64 $this->channels[$elementName] = $content ;
65 } 65 }
66 66
67 /** 67 /**
68 * Set multiple channel elements from an array. Array elements 68 * Set multiple channel elements from an array. Array elements
69 * should be 'channelName' => 'channelContent' format. 69 * should be 'channelName' => 'channelContent' format.
70 * 70 *
71 * @access public 71 * @access public
72 * @param array array of channels 72 * @param array array of channels
73 * @return void 73 * @return void
@@ -75,30 +75,30 @@ define('JSONP', 3, true);
75 public function setChannelElementsFromArray($elementArray) 75 public function setChannelElementsFromArray($elementArray)
76 { 76 {
77 if(! is_array($elementArray)) return; 77 if(! is_array($elementArray)) return;
78 foreach ($elementArray as $elementName => $content) 78 foreach ($elementArray as $elementName => $content)
79 { 79 {
80 $this->setChannelElement($elementName, $content); 80 $this->setChannelElement($elementName, $content);
81 } 81 }
82 } 82 }
83 83
84 /** 84 /**
85 * Genarate the actual RSS/JSON file 85 * Genarate the actual RSS/JSON file
86 * 86 *
87 * @access public 87 * @access public
88 * @return void 88 * @return void
89 */ 89 */
90 public function genarateFeed() 90 public function genarateFeed()
91 { 91 {
92 if ($this->version == RSS2) { 92 if ($this->version == RSS2) {
93 header('Content-type: text/xml; charset=UTF-8'); 93// header('Content-type: text/xml; charset=UTF-8');
94 // this line prevents Chrome 20 from prompting download 94 // this line prevents Chrome 20 from prompting download
95 // used by Google: https://news.google.com/news/feeds?ned=us&topic=b&output=rss 95 // used by Google: https://news.google.com/news/feeds?ned=us&topic=b&output=rss
96 header('X-content-type-options: nosniff'); 96// header('X-content-type-options: nosniff');
97 } elseif ($this->version == JSON) { 97 } elseif ($this->version == JSON) {
98 header('Content-type: application/json; charset=UTF-8'); 98// header('Content-type: application/json; charset=UTF-8');
99 $this->json = new stdClass(); 99 $this->json = new stdClass();
100 } elseif ($this->version == JSONP) { 100 } elseif ($this->version == JSONP) {
101 header('Content-type: application/javascript; charset=UTF-8'); 101// header('Content-type: application/javascript; charset=UTF-8');
102 $this->json = new stdClass(); 102 $this->json = new stdClass();
103 } 103 }
104 $this->printHead(); 104 $this->printHead();
@@ -109,10 +109,10 @@ define('JSONP', 3, true);
109 echo json_encode($this->json); 109 echo json_encode($this->json);
110 } 110 }
111 } 111 }
112 112
113 /** 113 /**
114 * Create a new FeedItem. 114 * Create a new FeedItem.
115 * 115 *
116 * @access public 116 * @access public
117 * @return object instance of FeedItem class 117 * @return object instance of FeedItem class
118 */ 118 */
@@ -121,24 +121,24 @@ define('JSONP', 3, true);
121 $Item = new FeedItem($this->version); 121 $Item = new FeedItem($this->version);
122 return $Item; 122 return $Item;
123 } 123 }
124 124
125 /** 125 /**
126 * Add a FeedItem to the main class 126 * Add a FeedItem to the main class
127 * 127 *
128 * @access public 128 * @access public
129 * @param object instance of FeedItem class 129 * @param object instance of FeedItem class
130 * @return void 130 * @return void
131 */ 131 */
132 public function addItem($feedItem) 132 public function addItem($feedItem)
133 { 133 {
134 $this->items[] = $feedItem; 134 $this->items[] = $feedItem;
135 } 135 }
136 136
137 // Wrapper functions ------------------------------------------------------------------- 137 // Wrapper functions -------------------------------------------------------------------
138 138
139 /** 139 /**
140 * Set the 'title' channel element 140 * Set the 'title' channel element
141 * 141 *
142 * @access public 142 * @access public
143 * @param srting value of 'title' channel tag 143 * @param srting value of 'title' channel tag
144 * @return void 144 * @return void
@@ -147,59 +147,59 @@ define('JSONP', 3, true);
147 { 147 {
148 $this->setChannelElement('title', $title); 148 $this->setChannelElement('title', $title);
149 } 149 }
150 150
151 /** 151 /**
152 * Add a hub to the channel element 152 * Add a hub to the channel element
153 * 153 *
154 * @access public 154 * @access public
155 * @param string URL 155 * @param string URL
156 * @return void 156 * @return void
157 */ 157 */
158 public function addHub($hub) 158 public function addHub($hub)
159 { 159 {
160 $this->hubs[] = $hub; 160 $this->hubs[] = $hub;
161 } 161 }
162 162
163 /** 163 /**
164 * Set XSL URL 164 * Set XSL URL
165 * 165 *
166 * @access public 166 * @access public
167 * @param string URL 167 * @param string URL
168 * @return void 168 * @return void
169 */ 169 */
170 public function setXsl($xsl) 170 public function setXsl($xsl)
171 { 171 {
172 $this->xsl = $xsl; 172 $this->xsl = $xsl;
173 } 173 }
174 174
175 /** 175 /**
176 * Set self URL 176 * Set self URL
177 * 177 *
178 * @access public 178 * @access public
179 * @param string URL 179 * @param string URL
180 * @return void 180 * @return void
181 */ 181 */
182 public function setSelf($self) 182 public function setSelf($self)
183 { 183 {
184 $this->self = $self; 184 $this->self = $self;
185 } 185 }
186 186
187 /** 187 /**
188 * Set the 'description' channel element 188 * Set the 'description' channel element
189 * 189 *
190 * @access public 190 * @access public
191 * @param srting value of 'description' channel tag 191 * @param srting value of 'description' channel tag
192 * @return void 192 * @return void
193 */ 193 */
194 public function setDescription($desciption) 194 public function setDescription($desciption)
195 { 195 {
196 $tag = ($this->version == ATOM)? 'subtitle' : 'description'; 196 $tag = ($this->version == ATOM)? 'subtitle' : 'description';
197 $this->setChannelElement($tag, $desciption); 197 $this->setChannelElement($tag, $desciption);
198 } 198 }
199 199
200 /** 200 /**
201 * Set the 'link' channel element 201 * Set the 'link' channel element
202 * 202 *
203 * @access public 203 * @access public
204 * @param srting value of 'link' channel tag 204 * @param srting value of 'link' channel tag
205 * @return void 205 * @return void
@@ -208,10 +208,10 @@ define('JSONP', 3, true);
208 { 208 {
209 $this->setChannelElement('link', $link); 209 $this->setChannelElement('link', $link);
210 } 210 }
211 211
212 /** 212 /**
213 * Set the 'image' channel element 213 * Set the 'image' channel element
214 * 214 *
215 * @access public 215 * @access public
216 * @param srting title of image 216 * @param srting title of image
217 * @param srting link url of the imahe 217 * @param srting link url of the imahe
@@ -222,14 +222,14 @@ define('JSONP', 3, true);
222 { 222 {
223 $this->setChannelElement('image', array('title'=>$title, 'link'=>$link, 'url'=>$url)); 223 $this->setChannelElement('image', array('title'=>$title, 'link'=>$link, 'url'=>$url));
224 } 224 }
225 225
226 // End # public functions ---------------------------------------------- 226 // End # public functions ----------------------------------------------
227 227
228 // Start # private functions ---------------------------------------------- 228 // Start # private functions ----------------------------------------------
229 229
230 /** 230 /**
231 * Prints the xml and rss namespace 231 * Prints the xml and rss namespace
232 * 232 *
233 * @access private 233 * @access private
234 * @return void 234 * @return void
235 */ 235 */
@@ -247,10 +247,10 @@ define('JSONP', 3, true);
247 $this->json->rss = array('@attributes' => array('version' => '2.0')); 247 $this->json->rss = array('@attributes' => array('version' => '2.0'));
248 } 248 }
249 } 249 }
250 250
251 /** 251 /**
252 * Closes the open tags at the end of file 252 * Closes the open tags at the end of file
253 * 253 *
254 * @access private 254 * @access private
255 * @return void 255 * @return void
256 */ 256 */
@@ -258,14 +258,14 @@ define('JSONP', 3, true);
258 { 258 {
259 if ($this->version == RSS2) 259 if ($this->version == RSS2)
260 { 260 {
261 echo '</channel>',PHP_EOL,'</rss>'; 261 echo '</channel>',PHP_EOL,'</rss>';
262 } 262 }
263 // do nothing for JSON 263 // do nothing for JSON
264 } 264 }
265 265
266 /** 266 /**
267 * Creates a single node as xml format 267 * Creates a single node as xml format
268 * 268 *
269 * @access private 269 * @access private
270 * @param string name of the tag 270 * @param string name of the tag
271 * @param mixed tag value as string or array of nested tags in 'tagName' => 'tagValue' format 271 * @param mixed tag value as string or array of nested tags in 'tagName' => 'tagValue' format
@@ -273,22 +273,22 @@ define('JSONP', 3, true);
273 * @return string formatted xml tag 273 * @return string formatted xml tag
274 */ 274 */
275 private function makeNode($tagName, $tagContent, $attributes = null) 275 private function makeNode($tagName, $tagContent, $attributes = null)
276 { 276 {
277 if ($this->version == RSS2) 277 if ($this->version == RSS2)
278 { 278 {
279 $nodeText = ''; 279 $nodeText = '';
280 $attrText = ''; 280 $attrText = '';
281 if (is_array($attributes)) 281 if (is_array($attributes))
282 { 282 {
283 foreach ($attributes as $key => $value) 283 foreach ($attributes as $key => $value)
284 { 284 {
285 $attrText .= " $key=\"$value\" "; 285 $attrText .= " $key=\"$value\" ";
286 } 286 }
287 } 287 }
288 $nodeText .= "<{$tagName}{$attrText}>"; 288 $nodeText .= "<{$tagName}{$attrText}>";
289 if (is_array($tagContent)) 289 if (is_array($tagContent))
290 { 290 {
291 foreach ($tagContent as $key => $value) 291 foreach ($tagContent as $key => $value)
292 { 292 {
293 $nodeText .= $this->makeNode($key, $value); 293 $nodeText .= $this->makeNode($key, $value);
294 } 294 }
@@ -297,7 +297,7 @@ define('JSONP', 3, true);
297 { 297 {
298 //$nodeText .= (in_array($tagName, $this->CDATAEncoding))? $tagContent : htmlentities($tagContent); 298 //$nodeText .= (in_array($tagName, $this->CDATAEncoding))? $tagContent : htmlentities($tagContent);
299 $nodeText .= htmlspecialchars($tagContent); 299 $nodeText .= htmlspecialchars($tagContent);
300 } 300 }
301 //$nodeText .= (in_array($tagName, $this->CDATAEncoding))? "]]></$tagName>" : "</$tagName>"; 301 //$nodeText .= (in_array($tagName, $this->CDATAEncoding))? "]]></$tagName>" : "</$tagName>";
302 $nodeText .= "</$tagName>"; 302 $nodeText .= "</$tagName>";
303 return $nodeText . PHP_EOL; 303 return $nodeText . PHP_EOL;
@@ -321,7 +321,7 @@ define('JSONP', 3, true);
321 } 321 }
322 return ''; // should not get here 322 return ''; // should not get here
323 } 323 }
324 324
325 private function json_keys(array $array) { 325 private function json_keys(array $array) {
326 $new = array(); 326 $new = array();
327 foreach ($array as $key => $val) { 327 foreach ($array as $key => $val) {
@@ -334,7 +334,7 @@ define('JSONP', 3, true);
334 } 334 }
335 return $new; 335 return $new;
336 } 336 }
337 337
338 /** 338 /**
339 * @desc Print channels 339 * @desc Print channels
340 * @access private 340 * @access private
@@ -344,7 +344,7 @@ define('JSONP', 3, true);
344 { 344 {
345 //Start channel tag 345 //Start channel tag
346 if ($this->version == RSS2) { 346 if ($this->version == RSS2) {
347 echo '<channel>' . PHP_EOL; 347 echo '<channel>' . PHP_EOL;
348 // add hubs 348 // add hubs
349 foreach ($this->hubs as $hub) { 349 foreach ($this->hubs as $hub) {
350 //echo $this->makeNode('link', '', array('rel'=>'hub', 'href'=>$hub, 'xmlns'=>'http://www.w3.org/2005/Atom')); 350 //echo $this->makeNode('link', '', array('rel'=>'hub', 'href'=>$hub, 'xmlns'=>'http://www.w3.org/2005/Atom'));
@@ -356,7 +356,7 @@ define('JSONP', 3, true);
356 echo '<link rel="self" href="'.htmlspecialchars($this->self).'" xmlns="http://www.w3.org/2005/Atom" />' . PHP_EOL; 356 echo '<link rel="self" href="'.htmlspecialchars($this->self).'" xmlns="http://www.w3.org/2005/Atom" />' . PHP_EOL;
357 } 357 }
358 //Print Items of channel 358 //Print Items of channel
359 foreach ($this->channels as $key => $value) 359 foreach ($this->channels as $key => $value)
360 { 360 {
361 echo $this->makeNode($key, $value); 361 echo $this->makeNode($key, $value);
362 } 362 }
@@ -364,26 +364,26 @@ define('JSONP', 3, true);
364 $this->json->rss['channel'] = (object)$this->json_keys($this->channels); 364 $this->json->rss['channel'] = (object)$this->json_keys($this->channels);
365 } 365 }
366 } 366 }
367 367
368 /** 368 /**
369 * Prints formatted feed items 369 * Prints formatted feed items
370 * 370 *
371 * @access private 371 * @access private
372 * @return void 372 * @return void
373 */ 373 */
374 private function printItems() 374 private function printItems()
375 { 375 {
376 foreach ($this->items as $item) { 376 foreach ($this->items as $item) {
377 $itemElements = $item->getElements(); 377 $itemElements = $item->getElements();
378 378
379 echo $this->startItem(); 379 echo $this->startItem();
380 380
381 if ($this->version == JSON || $this->version == JSONP) { 381 if ($this->version == JSON || $this->version == JSONP) {
382 $json_item = array(); 382 $json_item = array();
383 } 383 }
384 384
385 foreach ($itemElements as $thisElement) { 385 foreach ($itemElements as $thisElement) {
386 foreach ($thisElement as $instance) { 386 foreach ($thisElement as $instance) {
387 if ($this->version == RSS2) { 387 if ($this->version == RSS2) {
388 echo $this->makeNode($instance['name'], $instance['content'], $instance['attributes']); 388 echo $this->makeNode($instance['name'], $instance['content'], $instance['attributes']);
389 } elseif ($this->version == JSON || $this->version == JSONP) { 389 } elseif ($this->version == JSON || $this->version == JSONP) {
@@ -406,10 +406,10 @@ define('JSONP', 3, true);
406 } 406 }
407 } 407 }
408 } 408 }
409 409
410 /** 410 /**
411 * Make the starting tag of channels 411 * Make the starting tag of channels
412 * 412 *
413 * @access private 413 * @access private
414 * @return void 414 * @return void
415 */ 415 */
@@ -417,14 +417,14 @@ define('JSONP', 3, true);
417 { 417 {
418 if ($this->version == RSS2) 418 if ($this->version == RSS2)
419 { 419 {
420 echo '<item>' . PHP_EOL; 420 echo '<item>' . PHP_EOL;
421 } 421 }
422 // nothing for JSON 422 // nothing for JSON
423 } 423 }
424 424
425 /** 425 /**
426 * Closes feed item tag 426 * Closes feed item tag
427 * 427 *
428 * @access private 428 * @access private
429 * @return void 429 * @return void
430 */ 430 */
@@ -432,10 +432,10 @@ define('JSONP', 3, true);
432 { 432 {
433 if ($this->version == RSS2) 433 if ($this->version == RSS2)
434 { 434 {
435 echo '</item>' . PHP_EOL; 435 echo '</item>' . PHP_EOL;
436 } 436 }
437 // nothing for JSON 437 // nothing for JSON
438 } 438 }
439 439
440 // End # private functions ---------------------------------------------- 440 // End # private functions ----------------------------------------------
441 } \ No newline at end of file 441 } \ No newline at end of file
diff --git a/inc/3rdparty/makefulltextfeed.php b/inc/3rdparty/makefulltextfeed.php
index 2852c4c2..135964f1 100755
--- a/inc/3rdparty/makefulltextfeed.php
+++ b/inc/3rdparty/makefulltextfeed.php
@@ -55,42 +55,8 @@ if (get_magic_quotes_gpc()) {
55 55
56// set include path 56// set include path
57set_include_path(realpath(dirname(__FILE__).'/libraries').PATH_SEPARATOR.get_include_path()); 57set_include_path(realpath(dirname(__FILE__).'/libraries').PATH_SEPARATOR.get_include_path());
58// Autoloading of classes allows us to include files only when they're 58
59// needed. If we've got a cached copy, for example, only Zend_Cache is loaded. 59require_once dirname(__FILE__).'/makefulltextfeedHelpers.php';
60function autoload($class_name) {
61 static $dir = null;
62 if ($dir === null) $dir = dirname(__FILE__).'/libraries/';
63 static $mapping = array(
64 // Include FeedCreator for RSS/Atom creation
65 'FeedWriter' => 'feedwriter/FeedWriter.php',
66 'FeedItem' => 'feedwriter/FeedItem.php',
67 // Include ContentExtractor and Readability for identifying and extracting content from URLs
68 'ContentExtractor' => 'content-extractor/ContentExtractor.php',
69 'SiteConfig' => 'content-extractor/SiteConfig.php',
70 'Readability' => 'readability/Readability.php',
71 // Include Humble HTTP Agent to allow parallel requests and response caching
72 'HumbleHttpAgent' => 'humble-http-agent/HumbleHttpAgent.php',
73 'SimplePie_HumbleHttpAgent' => 'humble-http-agent/SimplePie_HumbleHttpAgent.php',
74 'CookieJar' => 'humble-http-agent/CookieJar.php',
75 // Include Zend Cache to improve performance (cache results)
76 'Zend_Cache' => 'Zend/Cache.php',
77 // Language detect
78 'Text_LanguageDetect' => 'language-detect/LanguageDetect.php',
79 // HTML5 Lib
80 'HTML5_Parser' => 'html5/Parser.php',
81 // htmLawed - used if XSS filter is enabled (xss_filter)
82 'htmLawed' => 'htmLawed/htmLawed.php'
83 );
84 if (isset($mapping[$class_name])) {
85 debug("** Loading class $class_name ({$mapping[$class_name]})");
86 require $dir.$mapping[$class_name];
87 return true;
88 } else {
89 return false;
90 }
91}
92spl_autoload_register('autoload');
93require dirname(__FILE__).'/libraries/simplepie/autoloader.php';
94 60
95//////////////////////////////// 61////////////////////////////////
96// Load config file 62// Load config file
@@ -415,6 +381,7 @@ if (!$debug_mode) {
415////////////////////////////////// 381//////////////////////////////////
416// Set up HTTP agent 382// Set up HTTP agent
417////////////////////////////////// 383//////////////////////////////////
384global $http;
418$http = new HumbleHttpAgent(); 385$http = new HumbleHttpAgent();
419$http->debug = $debug_mode; 386$http->debug = $debug_mode;
420$http->userAgentMap = $options->user_agents; 387$http->userAgentMap = $options->user_agents;
@@ -478,29 +445,6 @@ if ($html_only || !$result) {
478 $isDummyFeed = true; 445 $isDummyFeed = true;
479 unset($feed, $result); 446 unset($feed, $result);
480 // create single item dummy feed object 447 // create single item dummy feed object
481 class DummySingleItemFeed {
482 public $item;
483 function __construct($url) { $this->item = new DummySingleItem($url); }
484 public function get_title() { return ''; }
485 public function get_description() { return 'Content extracted from '.$this->item->url; }
486 public function get_link() { return $this->item->url; }
487 public function get_language() { return false; }
488 public function get_image_url() { return false; }
489 public function get_items($start=0, $max=1) { return array(0=>$this->item); }
490 }
491 class DummySingleItem {
492 public $url;
493 function __construct($url) { $this->url = $url; }
494 public function get_permalink() { return $this->url; }
495 public function get_title() { return null; }
496 public function get_date($format='') { return false; }
497 public function get_author($key=0) { return null; }
498 public function get_authors() { return null; }
499 public function get_description() { return ''; }
500 public function get_enclosure($key=0, $prefer=null) { return null; }
501 public function get_enclosures() { return null; }
502 public function get_categories() { return null; }
503 }
504 $feed = new DummySingleItemFeed($url); 448 $feed = new DummySingleItemFeed($url);
505} 449}
506 450
@@ -903,294 +847,3 @@ if (!$debug_mode) {
903 if ($callback) echo ');'; 847 if ($callback) echo ');';
904} 848}
905 849
906///////////////////////////////
907// HELPER FUNCTIONS
908///////////////////////////////
909
910function url_allowed($url) {
911 global $options;
912 if (!empty($options->allowed_urls)) {
913 $allowed = false;
914 foreach ($options->allowed_urls as $allowurl) {
915 if (stristr($url, $allowurl) !== false) {
916 $allowed = true;
917 break;
918 }
919 }
920 if (!$allowed) return false;
921 } else {
922 foreach ($options->blocked_urls as $blockurl) {
923 if (stristr($url, $blockurl) !== false) {
924 return false;
925 }
926 }
927 }
928 return true;
929}
930
931//////////////////////////////////////////////
932// Convert $html to UTF8
933// (uses HTTP headers and HTML to find encoding)
934// adapted from http://stackoverflow.com/questions/910793/php-detect-encoding-and-make-everything-utf-8
935//////////////////////////////////////////////
936function convert_to_utf8($html, $header=null)
937{
938 $encoding = null;
939 if ($html || $header) {
940 if (is_array($header)) $header = implode("\n", $header);
941 if (!$header || !preg_match_all('/^Content-Type:\s+([^;]+)(?:;\s*charset=["\']?([^;"\'\n]*))?/im', $header, $match, PREG_SET_ORDER)) {
942 // error parsing the response
943 debug('Could not find Content-Type header in HTTP response');
944 } else {
945 $match = end($match); // get last matched element (in case of redirects)
946 if (isset($match[2])) $encoding = trim($match[2], "\"' \r\n\0\x0B\t");
947 }
948 // TODO: check to see if encoding is supported (can we convert it?)
949 // If it's not, result will be empty string.
950 // For now we'll check for invalid encoding types returned by some sites, e.g. 'none'
951 // Problem URL: http://facta.co.jp/blog/archives/20111026001026.html
952 if (!$encoding || $encoding == 'none') {
953 // search for encoding in HTML - only look at the first 50000 characters
954 // Why 50000? See, for example, http://www.lemonde.fr/festival-de-cannes/article/2012/05/23/deux-cretes-en-goguette-sur-la-croisette_1705732_766360.html
955 // TODO: improve this so it looks at smaller chunks first
956 $html_head = substr($html, 0, 50000);
957 if (preg_match('/^<\?xml\s+version=(?:"[^"]*"|\'[^\']*\')\s+encoding=("[^"]*"|\'[^\']*\')/s', $html_head, $match)) {
958 $encoding = trim($match[1], '"\'');
959 } elseif (preg_match('/<meta\s+http-equiv=["\']?Content-Type["\']? content=["\'][^;]+;\s*charset=["\']?([^;"\'>]+)/i', $html_head, $match)) {
960 $encoding = trim($match[1]);
961 } elseif (preg_match_all('/<meta\s+([^>]+)>/i', $html_head, $match)) {
962 foreach ($match[1] as $_test) {
963 if (preg_match('/charset=["\']?([^"\']+)/i', $_test, $_m)) {
964 $encoding = trim($_m[1]);
965 break;
966 }
967 }
968 }
969 }
970 if (isset($encoding)) $encoding = trim($encoding);
971 // trim is important here!
972 if (!$encoding || (strtolower($encoding) == 'iso-8859-1')) {
973 // replace MS Word smart qutoes
974 $trans = array();
975 $trans[chr(130)] = '&sbquo;'; // Single Low-9 Quotation Mark
976 $trans[chr(131)] = '&fnof;'; // Latin Small Letter F With Hook
977 $trans[chr(132)] = '&bdquo;'; // Double Low-9 Quotation Mark
978 $trans[chr(133)] = '&hellip;'; // Horizontal Ellipsis
979 $trans[chr(134)] = '&dagger;'; // Dagger
980 $trans[chr(135)] = '&Dagger;'; // Double Dagger
981 $trans[chr(136)] = '&circ;'; // Modifier Letter Circumflex Accent
982 $trans[chr(137)] = '&permil;'; // Per Mille Sign
983 $trans[chr(138)] = '&Scaron;'; // Latin Capital Letter S With Caron
984 $trans[chr(139)] = '&lsaquo;'; // Single Left-Pointing Angle Quotation Mark
985 $trans[chr(140)] = '&OElig;'; // Latin Capital Ligature OE
986 $trans[chr(145)] = '&lsquo;'; // Left Single Quotation Mark
987 $trans[chr(146)] = '&rsquo;'; // Right Single Quotation Mark
988 $trans[chr(147)] = '&ldquo;'; // Left Double Quotation Mark
989 $trans[chr(148)] = '&rdquo;'; // Right Double Quotation Mark
990 $trans[chr(149)] = '&bull;'; // Bullet
991 $trans[chr(150)] = '&ndash;'; // En Dash
992 $trans[chr(151)] = '&mdash;'; // Em Dash
993 $trans[chr(152)] = '&tilde;'; // Small Tilde
994 $trans[chr(153)] = '&trade;'; // Trade Mark Sign
995 $trans[chr(154)] = '&scaron;'; // Latin Small Letter S With Caron
996 $trans[chr(155)] = '&rsaquo;'; // Single Right-Pointing Angle Quotation Mark
997 $trans[chr(156)] = '&oelig;'; // Latin Small Ligature OE
998 $trans[chr(159)] = '&Yuml;'; // Latin Capital Letter Y With Diaeresis
999 $html = strtr($html, $trans);
1000 }
1001 if (!$encoding) {
1002 debug('No character encoding found, so treating as UTF-8');
1003 $encoding = 'utf-8';
1004 } else {
1005 debug('Character encoding: '.$encoding);
1006 if (strtolower($encoding) != 'utf-8') {
1007 debug('Converting to UTF-8');
1008 $html = SimplePie_Misc::change_encoding($html, $encoding, 'utf-8');
1009 /*
1010 if (function_exists('iconv')) {
1011 // iconv appears to handle certain character encodings better than mb_convert_encoding
1012 $html = iconv($encoding, 'utf-8', $html);
1013 } else {
1014 $html = mb_convert_encoding($html, 'utf-8', $encoding);
1015 }
1016 */
1017 }
1018 }
1019 }
1020 return $html;
1021}
1022
1023function makeAbsolute($base, $elem) {
1024 $base = new SimplePie_IRI($base);
1025 // remove '//' in URL path (used to prevent URLs from resolving properly)
1026 // TODO: check if this is still the case
1027 if (isset($base->path)) $base->path = preg_replace('!//+!', '/', $base->path);
1028 foreach(array('a'=>'href', 'img'=>'src') as $tag => $attr) {
1029 $elems = $elem->getElementsByTagName($tag);
1030 for ($i = $elems->length-1; $i >= 0; $i--) {
1031 $e = $elems->item($i);
1032 //$e->parentNode->replaceChild($articleContent->ownerDocument->createTextNode($e->textContent), $e);
1033 makeAbsoluteAttr($base, $e, $attr);
1034 }
1035 if (strtolower($elem->tagName) == $tag) makeAbsoluteAttr($base, $elem, $attr);
1036 }
1037}
1038function makeAbsoluteAttr($base, $e, $attr) {
1039 if ($e->hasAttribute($attr)) {
1040 // Trim leading and trailing white space. I don't really like this but
1041 // unfortunately it does appear on some sites. e.g. <img src=" /path/to/image.jpg" />
1042 $url = trim(str_replace('%20', ' ', $e->getAttribute($attr)));
1043 $url = str_replace(' ', '%20', $url);
1044 if (!preg_match('!https?://!i', $url)) {
1045 if ($absolute = SimplePie_IRI::absolutize($base, $url)) {
1046 $e->setAttribute($attr, $absolute);
1047 }
1048 }
1049 }
1050}
1051function makeAbsoluteStr($base, $url) {
1052 $base = new SimplePie_IRI($base);
1053 // remove '//' in URL path (causes URLs not to resolve properly)
1054 if (isset($base->path)) $base->path = preg_replace('!//+!', '/', $base->path);
1055 if (preg_match('!^https?://!i', $url)) {
1056 // already absolute
1057 return $url;
1058 } else {
1059 if ($absolute = SimplePie_IRI::absolutize($base, $url)) {
1060 return $absolute;
1061 }
1062 return false;
1063 }
1064}
1065// returns single page response, or false if not found
1066function getSinglePage($item, $html, $url) {
1067 global $http, $extractor;
1068 debug('Looking for site config files to see if single page link exists');
1069 $site_config = $extractor->buildSiteConfig($url, $html);
1070 $splink = null;
1071 if (!empty($site_config->single_page_link)) {
1072 $splink = $site_config->single_page_link;
1073 } elseif (!empty($site_config->single_page_link_in_feed)) {
1074 // single page link xpath is targeted at feed
1075 $splink = $site_config->single_page_link_in_feed;
1076 // so let's replace HTML with feed item description
1077 $html = $item->get_description();
1078 }
1079 if (isset($splink)) {
1080 // Build DOM tree from HTML
1081 $readability = new Readability($html, $url);
1082 $xpath = new DOMXPath($readability->dom);
1083 // Loop through single_page_link xpath expressions
1084 $single_page_url = null;
1085 foreach ($splink as $pattern) {
1086 $elems = @$xpath->evaluate($pattern, $readability->dom);
1087 if (is_string($elems)) {
1088 $single_page_url = trim($elems);
1089 break;
1090 } elseif ($elems instanceof DOMNodeList && $elems->length > 0) {
1091 foreach ($elems as $item) {
1092 if ($item instanceof DOMElement && $item->hasAttribute('href')) {
1093 $single_page_url = $item->getAttribute('href');
1094 break 2;
1095 } elseif ($item instanceof DOMAttr && $item->value) {
1096 $single_page_url = $item->value;
1097 break 2;
1098 }
1099 }
1100 }
1101 }
1102 // If we've got URL, resolve against $url
1103 if (isset($single_page_url) && ($single_page_url = makeAbsoluteStr($url, $single_page_url))) {
1104 // check it's not what we have already!
1105 if ($single_page_url != $url) {
1106 // it's not, so let's try to fetch it...
1107 $_prev_ref = $http->referer;
1108 $http->referer = $single_page_url;
1109 if (($response = $http->get($single_page_url, true)) && $response['status_code'] < 300) {
1110 $http->referer = $_prev_ref;
1111 return $response;
1112 }
1113 $http->referer = $_prev_ref;
1114 }
1115 }
1116 }
1117 return false;
1118}
1119
1120// based on content-type http header, decide what to do
1121// param: HTTP headers string
1122// return: array with keys: 'mime', 'type', 'subtype', 'action', 'name'
1123// e.g. array('mime'=>'image/jpeg', 'type'=>'image', 'subtype'=>'jpeg', 'action'=>'link', 'name'=>'Image')
1124function get_mime_action_info($headers) {
1125 global $options;
1126 // check if action defined for returned Content-Type
1127 $info = array();
1128 if (preg_match('!^Content-Type:\s*(([-\w]+)/([-\w\+]+))!im', $headers, $match)) {
1129 // look for full mime type (e.g. image/jpeg) or just type (e.g. image)
1130 // match[1] = full mime type, e.g. image/jpeg
1131 // match[2] = first part, e.g. image
1132 // match[3] = last part, e.g. jpeg
1133 $info['mime'] = strtolower(trim($match[1]));
1134 $info['type'] = strtolower(trim($match[2]));
1135 $info['subtype'] = strtolower(trim($match[3]));
1136 foreach (array($info['mime'], $info['type']) as $_mime) {
1137 if (isset($options->content_type_exc[$_mime])) {
1138 $info['action'] = $options->content_type_exc[$_mime]['action'];
1139 $info['name'] = $options->content_type_exc[$_mime]['name'];
1140 break;
1141 }
1142 }
1143 }
1144 return $info;
1145}
1146
1147function remove_url_cruft($url) {
1148 // remove google analytics for the time being
1149 // regex adapted from http://navitronic.co.uk/2010/12/removing-google-analytics-cruft-from-urls/
1150 // https://gist.github.com/758177
1151 return preg_replace('/(\?|\&)utm_[a-z]+=[^\&]+/', '', $url);
1152}
1153
1154function make_substitutions($string) {
1155 if ($string == '') return $string;
1156 global $item, $effective_url;
1157 $string = str_replace('{url}', htmlspecialchars($item->get_permalink()), $string);
1158 $string = str_replace('{effective-url}', htmlspecialchars($effective_url), $string);
1159 return $string;
1160}
1161
1162function get_cache() {
1163 global $options, $valid_key;
1164 static $cache = null;
1165 if ($cache === null) {
1166 $frontendOptions = array(
1167 'lifetime' => 10*60, // cache lifetime of 10 minutes
1168 'automatic_serialization' => false,
1169 'write_control' => false,
1170 'automatic_cleaning_factor' => $options->cache_cleanup,
1171 'ignore_user_abort' => false
1172 );
1173 $backendOptions = array(
1174 'cache_dir' => ($valid_key) ? $options->cache_dir.'/rss-with-key/' : $options->cache_dir.'/rss/', // directory where to put the cache files
1175 'file_locking' => false,
1176 'read_control' => true,
1177 'read_control_type' => 'strlen',
1178 'hashed_directory_level' => $options->cache_directory_level,
1179 'hashed_directory_perm' => 0777,
1180 'cache_file_perm' => 0664,
1181 'file_name_prefix' => 'ff'
1182 );
1183 // getting a Zend_Cache_Core object
1184 $cache = Zend_Cache::factory('Core', 'File', $frontendOptions, $backendOptions);
1185 }
1186 return $cache;
1187}
1188
1189function debug($msg) {
1190 global $debug_mode;
1191 if ($debug_mode) {
1192 echo '* ',$msg,"\n";
1193 ob_flush();
1194 flush();
1195 }
1196} \ No newline at end of file
diff --git a/inc/3rdparty/makefulltextfeedHelpers.php b/inc/3rdparty/makefulltextfeedHelpers.php
new file mode 100755
index 00000000..1c11b8f6
--- /dev/null
+++ b/inc/3rdparty/makefulltextfeedHelpers.php
@@ -0,0 +1,355 @@
1<?php
2
3// Autoloading of classes allows us to include files only when they're
4// needed. If we've got a cached copy, for example, only Zend_Cache is loaded.
5function autoload($class_name) {
6 static $dir = null;
7 if ($dir === null) $dir = dirname(__FILE__).'/libraries/';
8 static $mapping = array(
9 // Include FeedCreator for RSS/Atom creation
10 'FeedWriter' => 'feedwriter/FeedWriter.php',
11 'FeedItem' => 'feedwriter/FeedItem.php',
12 // Include ContentExtractor and Readability for identifying and extracting content from URLs
13 'ContentExtractor' => 'content-extractor/ContentExtractor.php',
14 'SiteConfig' => 'content-extractor/SiteConfig.php',
15 'Readability' => 'readability/Readability.php',
16 // Include Humble HTTP Agent to allow parallel requests and response caching
17 'HumbleHttpAgent' => 'humble-http-agent/HumbleHttpAgent.php',
18 'SimplePie_HumbleHttpAgent' => 'humble-http-agent/SimplePie_HumbleHttpAgent.php',
19 'CookieJar' => 'humble-http-agent/CookieJar.php',
20 // Include Zend Cache to improve performance (cache results)
21 'Zend_Cache' => 'Zend/Cache.php',
22 // Language detect
23 'Text_LanguageDetect' => 'language-detect/LanguageDetect.php',
24 // HTML5 Lib
25 'HTML5_Parser' => 'html5/Parser.php',
26 // htmLawed - used if XSS filter is enabled (xss_filter)
27 'htmLawed' => 'htmLawed/htmLawed.php'
28 );
29 if (isset($mapping[$class_name])) {
30 debug("** Loading class $class_name ({$mapping[$class_name]})");
31 require $dir.$mapping[$class_name];
32 return true;
33 } else {
34 return false;
35 }
36}
37spl_autoload_register('autoload');
38require dirname(__FILE__).'/libraries/simplepie/autoloader.php';
39
40
41class DummySingleItemFeed {
42 public $item;
43 function __construct($url) { $this->item = new DummySingleItem($url); }
44 public function get_title() { return ''; }
45 public function get_description() { return 'Content extracted from '.$this->item->url; }
46 public function get_link() { return $this->item->url; }
47 public function get_language() { return false; }
48 public function get_image_url() { return false; }
49 public function get_items($start=0, $max=1) { return array(0=>$this->item); }
50}
51class DummySingleItem {
52 public $url;
53 function __construct($url) { $this->url = $url; }
54 public function get_permalink() { return $this->url; }
55 public function get_title() { return null; }
56 public function get_date($format='') { return false; }
57 public function get_author($key=0) { return null; }
58 public function get_authors() { return null; }
59 public function get_description() { return ''; }
60 public function get_enclosure($key=0, $prefer=null) { return null; }
61 public function get_enclosures() { return null; }
62 public function get_categories() { return null; }
63}
64
65///////////////////////////////
66// HELPER FUNCTIONS
67///////////////////////////////
68
69function url_allowed($url) {
70 global $options;
71 if (!empty($options->allowed_urls)) {
72 $allowed = false;
73 foreach ($options->allowed_urls as $allowurl) {
74 if (stristr($url, $allowurl) !== false) {
75 $allowed = true;
76 break;
77 }
78 }
79 if (!$allowed) return false;
80 } else {
81 foreach ($options->blocked_urls as $blockurl) {
82 if (stristr($url, $blockurl) !== false) {
83 return false;
84 }
85 }
86 }
87 return true;
88}
89
90//////////////////////////////////////////////
91// Convert $html to UTF8
92// (uses HTTP headers and HTML to find encoding)
93// adapted from http://stackoverflow.com/questions/910793/php-detect-encoding-and-make-everything-utf-8
94//////////////////////////////////////////////
95function convert_to_utf8($html, $header=null)
96{
97 $encoding = null;
98 if ($html || $header) {
99 if (is_array($header)) $header = implode("\n", $header);
100 if (!$header || !preg_match_all('/^Content-Type:\s+([^;]+)(?:;\s*charset=["\']?([^;"\'\n]*))?/im', $header, $match, PREG_SET_ORDER)) {
101 // error parsing the response
102 debug('Could not find Content-Type header in HTTP response');
103 } else {
104 $match = end($match); // get last matched element (in case of redirects)
105 if (isset($match[2])) $encoding = trim($match[2], "\"' \r\n\0\x0B\t");
106 }
107 // TODO: check to see if encoding is supported (can we convert it?)
108 // If it's not, result will be empty string.
109 // For now we'll check for invalid encoding types returned by some sites, e.g. 'none'
110 // Problem URL: http://facta.co.jp/blog/archives/20111026001026.html
111 if (!$encoding || $encoding == 'none') {
112 // search for encoding in HTML - only look at the first 50000 characters
113 // Why 50000? See, for example, http://www.lemonde.fr/festival-de-cannes/article/2012/05/23/deux-cretes-en-goguette-sur-la-croisette_1705732_766360.html
114 // TODO: improve this so it looks at smaller chunks first
115 $html_head = substr($html, 0, 50000);
116 if (preg_match('/^<\?xml\s+version=(?:"[^"]*"|\'[^\']*\')\s+encoding=("[^"]*"|\'[^\']*\')/s', $html_head, $match)) {
117 $encoding = trim($match[1], '"\'');
118 } elseif (preg_match('/<meta\s+http-equiv=["\']?Content-Type["\']? content=["\'][^;]+;\s*charset=["\']?([^;"\'>]+)/i', $html_head, $match)) {
119 $encoding = trim($match[1]);
120 } elseif (preg_match_all('/<meta\s+([^>]+)>/i', $html_head, $match)) {
121 foreach ($match[1] as $_test) {
122 if (preg_match('/charset=["\']?([^"\']+)/i', $_test, $_m)) {
123 $encoding = trim($_m[1]);
124 break;
125 }
126 }
127 }
128 }
129 if (isset($encoding)) $encoding = trim($encoding);
130 // trim is important here!
131 if (!$encoding || (strtolower($encoding) == 'iso-8859-1')) {
132 // replace MS Word smart qutoes
133 $trans = array();
134 $trans[chr(130)] = '&sbquo;'; // Single Low-9 Quotation Mark
135 $trans[chr(131)] = '&fnof;'; // Latin Small Letter F With Hook
136 $trans[chr(132)] = '&bdquo;'; // Double Low-9 Quotation Mark
137 $trans[chr(133)] = '&hellip;'; // Horizontal Ellipsis
138 $trans[chr(134)] = '&dagger;'; // Dagger
139 $trans[chr(135)] = '&Dagger;'; // Double Dagger
140 $trans[chr(136)] = '&circ;'; // Modifier Letter Circumflex Accent
141 $trans[chr(137)] = '&permil;'; // Per Mille Sign
142 $trans[chr(138)] = '&Scaron;'; // Latin Capital Letter S With Caron
143 $trans[chr(139)] = '&lsaquo;'; // Single Left-Pointing Angle Quotation Mark
144 $trans[chr(140)] = '&OElig;'; // Latin Capital Ligature OE
145 $trans[chr(145)] = '&lsquo;'; // Left Single Quotation Mark
146 $trans[chr(146)] = '&rsquo;'; // Right Single Quotation Mark
147 $trans[chr(147)] = '&ldquo;'; // Left Double Quotation Mark
148 $trans[chr(148)] = '&rdquo;'; // Right Double Quotation Mark
149 $trans[chr(149)] = '&bull;'; // Bullet
150 $trans[chr(150)] = '&ndash;'; // En Dash
151 $trans[chr(151)] = '&mdash;'; // Em Dash
152 $trans[chr(152)] = '&tilde;'; // Small Tilde
153 $trans[chr(153)] = '&trade;'; // Trade Mark Sign
154 $trans[chr(154)] = '&scaron;'; // Latin Small Letter S With Caron
155 $trans[chr(155)] = '&rsaquo;'; // Single Right-Pointing Angle Quotation Mark
156 $trans[chr(156)] = '&oelig;'; // Latin Small Ligature OE
157 $trans[chr(159)] = '&Yuml;'; // Latin Capital Letter Y With Diaeresis
158 $html = strtr($html, $trans);
159 }
160 if (!$encoding) {
161 debug('No character encoding found, so treating as UTF-8');
162 $encoding = 'utf-8';
163 } else {
164 debug('Character encoding: '.$encoding);
165 if (strtolower($encoding) != 'utf-8') {
166 debug('Converting to UTF-8');
167 $html = SimplePie_Misc::change_encoding($html, $encoding, 'utf-8');
168 /*
169 if (function_exists('iconv')) {
170 // iconv appears to handle certain character encodings better than mb_convert_encoding
171 $html = iconv($encoding, 'utf-8', $html);
172 } else {
173 $html = mb_convert_encoding($html, 'utf-8', $encoding);
174 }
175 */
176 }
177 }
178 }
179 return $html;
180}
181
182function makeAbsolute($base, $elem) {
183 $base = new SimplePie_IRI($base);
184 // remove '//' in URL path (used to prevent URLs from resolving properly)
185 // TODO: check if this is still the case
186 if (isset($base->path)) $base->path = preg_replace('!//+!', '/', $base->path);
187 foreach(array('a'=>'href', 'img'=>'src') as $tag => $attr) {
188 $elems = $elem->getElementsByTagName($tag);
189 for ($i = $elems->length-1; $i >= 0; $i--) {
190 $e = $elems->item($i);
191 //$e->parentNode->replaceChild($articleContent->ownerDocument->createTextNode($e->textContent), $e);
192 makeAbsoluteAttr($base, $e, $attr);
193 }
194 if (strtolower($elem->tagName) == $tag) makeAbsoluteAttr($base, $elem, $attr);
195 }
196}
197function makeAbsoluteAttr($base, $e, $attr) {
198 if ($e->hasAttribute($attr)) {
199 // Trim leading and trailing white space. I don't really like this but
200 // unfortunately it does appear on some sites. e.g. <img src=" /path/to/image.jpg" />
201 $url = trim(str_replace('%20', ' ', $e->getAttribute($attr)));
202 $url = str_replace(' ', '%20', $url);
203 if (!preg_match('!https?://!i', $url)) {
204 if ($absolute = SimplePie_IRI::absolutize($base, $url)) {
205 $e->setAttribute($attr, $absolute);
206 }
207 }
208 }
209}
210function makeAbsoluteStr($base, $url) {
211 $base = new SimplePie_IRI($base);
212 // remove '//' in URL path (causes URLs not to resolve properly)
213 if (isset($base->path)) $base->path = preg_replace('!//+!', '/', $base->path);
214 if (preg_match('!^https?://!i', $url)) {
215 // already absolute
216 return $url;
217 } else {
218 if ($absolute = SimplePie_IRI::absolutize($base, $url)) {
219 return $absolute;
220 }
221 return false;
222 }
223}
224// returns single page response, or false if not found
225function getSinglePage($item, $html, $url) {
226 global $http, $extractor;
227 debug('Looking for site config files to see if single page link exists');
228 $site_config = $extractor->buildSiteConfig($url, $html);
229 $splink = null;
230 if (!empty($site_config->single_page_link)) {
231 $splink = $site_config->single_page_link;
232 } elseif (!empty($site_config->single_page_link_in_feed)) {
233 // single page link xpath is targeted at feed
234 $splink = $site_config->single_page_link_in_feed;
235 // so let's replace HTML with feed item description
236 $html = $item->get_description();
237 }
238 if (isset($splink)) {
239 // Build DOM tree from HTML
240 $readability = new Readability($html, $url);
241 $xpath = new DOMXPath($readability->dom);
242 // Loop through single_page_link xpath expressions
243 $single_page_url = null;
244 foreach ($splink as $pattern) {
245 $elems = @$xpath->evaluate($pattern, $readability->dom);
246 if (is_string($elems)) {
247 $single_page_url = trim($elems);
248 break;
249 } elseif ($elems instanceof DOMNodeList && $elems->length > 0) {
250 foreach ($elems as $item) {
251 if ($item instanceof DOMElement && $item->hasAttribute('href')) {
252 $single_page_url = $item->getAttribute('href');
253 break 2;
254 } elseif ($item instanceof DOMAttr && $item->value) {
255 $single_page_url = $item->value;
256 break 2;
257 }
258 }
259 }
260 }
261 // If we've got URL, resolve against $url
262 if (isset($single_page_url) && ($single_page_url = makeAbsoluteStr($url, $single_page_url))) {
263 // check it's not what we have already!
264 if ($single_page_url != $url) {
265 // it's not, so let's try to fetch it...
266 $_prev_ref = $http->referer;
267 $http->referer = $single_page_url;
268 if (($response = $http->get($single_page_url, true)) && $response['status_code'] < 300) {
269 $http->referer = $_prev_ref;
270 return $response;
271 }
272 $http->referer = $_prev_ref;
273 }
274 }
275 }
276 return false;
277}
278
279// based on content-type http header, decide what to do
280// param: HTTP headers string
281// return: array with keys: 'mime', 'type', 'subtype', 'action', 'name'
282// e.g. array('mime'=>'image/jpeg', 'type'=>'image', 'subtype'=>'jpeg', 'action'=>'link', 'name'=>'Image')
283function get_mime_action_info($headers) {
284 global $options;
285 // check if action defined for returned Content-Type
286 $info = array();
287 if (preg_match('!^Content-Type:\s*(([-\w]+)/([-\w\+]+))!im', $headers, $match)) {
288 // look for full mime type (e.g. image/jpeg) or just type (e.g. image)
289 // match[1] = full mime type, e.g. image/jpeg
290 // match[2] = first part, e.g. image
291 // match[3] = last part, e.g. jpeg
292 $info['mime'] = strtolower(trim($match[1]));
293 $info['type'] = strtolower(trim($match[2]));
294 $info['subtype'] = strtolower(trim($match[3]));
295 foreach (array($info['mime'], $info['type']) as $_mime) {
296 if (isset($options->content_type_exc[$_mime])) {
297 $info['action'] = $options->content_type_exc[$_mime]['action'];
298 $info['name'] = $options->content_type_exc[$_mime]['name'];
299 break;
300 }
301 }
302 }
303 return $info;
304}
305
306function remove_url_cruft($url) {
307 // remove google analytics for the time being
308 // regex adapted from http://navitronic.co.uk/2010/12/removing-google-analytics-cruft-from-urls/
309 // https://gist.github.com/758177
310 return preg_replace('/(\?|\&)utm_[a-z]+=[^\&]+/', '', $url);
311}
312
313function make_substitutions($string) {
314 if ($string == '') return $string;
315 global $item, $effective_url;
316 $string = str_replace('{url}', htmlspecialchars($item->get_permalink()), $string);
317 $string = str_replace('{effective-url}', htmlspecialchars($effective_url), $string);
318 return $string;
319}
320
321function get_cache() {
322 global $options, $valid_key;
323 static $cache = null;
324 if ($cache === null) {
325 $frontendOptions = array(
326 'lifetime' => 10*60, // cache lifetime of 10 minutes
327 'automatic_serialization' => false,
328 'write_control' => false,
329 'automatic_cleaning_factor' => $options->cache_cleanup,
330 'ignore_user_abort' => false
331 );
332 $backendOptions = array(
333 'cache_dir' => ($valid_key) ? $options->cache_dir.'/rss-with-key/' : $options->cache_dir.'/rss/', // directory where to put the cache files
334 'file_locking' => false,
335 'read_control' => true,
336 'read_control_type' => 'strlen',
337 'hashed_directory_level' => $options->cache_directory_level,
338 'hashed_directory_perm' => 0777,
339 'cache_file_perm' => 0664,
340 'file_name_prefix' => 'ff'
341 );
342 // getting a Zend_Cache_Core object
343 $cache = Zend_Cache::factory('Core', 'File', $frontendOptions, $backendOptions);
344 }
345 return $cache;
346}
347
348function debug($msg) {
349 global $debug_mode;
350 if ($debug_mode) {
351 echo '* ',$msg,"\n";
352 ob_flush();
353 flush();
354 }
355}
generated by cgit v1.2.3 (git 2.25.1) at 2024-11-16 06:59:06 +0000