pngcrush.sh

[tt-rss.git] / classes / feedparser.php

diff --git a/classes/feedparser.php b/classes/feedparser.php
index 22052bdb6c8352b5981f5d3ba08918cebf1144fc..a5e406149d46062c620b60a6ce53738f0b0d13d8 100644 (file)
--- a/classes/feedparser.php
+++ b/classes/feedparser.php
@@ -2,6 +2,7 @@
 class FeedParser {
        private $doc;
        private $error;
+       private $libxml_errors = array();
        private $items;
        private $link;
        private $title;
@@ -12,6 +13,20 @@ class FeedParser {
        const FEED_RSS = 1;
        const FEED_ATOM = 2;
 
+       function normalize_encoding($data) {
+               if (preg_match('/^(<\?xml[\t\n\r ].*?encoding[\t\n\r ]*=[\t\n\r ]*["\'])(.+?)(["\'].*?\?>)/s', $data, $matches) === 1) {
+
+                       $encoding = strtolower($matches[2]);
+
+                       if (in_array($encoding, array_map('strtolower', mb_list_encodings())))
+                               $data = mb_convert_encoding($data, 'UTF-8', $encoding);
+
+                       $data = preg_replace('/^<\?xml[\t\n\r ].*?\?>/s', $matches[1] . "UTF-8" . $matches[3] , $data);
+               }
+
+               return $data;
+       }
+
        function __construct($data) {
                libxml_use_internal_errors(true);
                libxml_clear_errors();
@@ -24,29 +39,28 @@ class FeedParser {
 
                // libxml compiled without iconv?
                if ($error && $error->code == 32) {
-                       if (preg_match('/^(<\?xml[\t\n\r ].*?encoding[\t\n\r ]*=[\t\n\r ]*["\'])(.+?)(["\'].*?\?>)/s', $data, $matches) === 1) {
-                               $data = mb_convert_encoding($data, 'UTF-8', $matches[2]);
-
-                               $data = preg_replace('/^<\?xml[\t\n\r ].*?\?>/s', $matches[1] . "UTF-8" . $matches[3] , $data);
+                       $data = $this->normalize_encoding($data);
 
-                               if ($data) {
-                                       libxml_clear_errors();
+                       if ($data) {
+                               libxml_clear_errors();
 
-                                       $this->doc = new DOMDocument();
-                                       $this->doc->loadXML($data);
+                               $this->doc = new DOMDocument();
+                               $this->doc->loadXML($data);
 
-                                       $error = libxml_get_last_error();
-                               }
+                               $error = libxml_get_last_error();
                        }
                }
 
                // some terrible invalid unicode entity?
                if ($error) {
-                       foreach(libxml_get_errors() as $err) {
+                       foreach (libxml_get_errors() as $err) {
                                if ($err->code == 9) {
+                                       // if the source feed is not in utf8, next conversion will fail
+                                       $data = $this->normalize_encoding($data);
+
                                        // remove dangling bytes
                                        $data = mb_convert_encoding($data, 'UTF-8', 'UTF-8');
-                                       
+
                                        // apparently not all UTF-8 characters are valid for XML
                                        $data = preg_replace('/[^\x{0009}\x{000a}\x{000d}\x{0020}-\x{D7FF}\x{E000}-\x{FFFD}]+/u', ' ', $data);
 
@@ -63,7 +77,15 @@ class FeedParser {
                        }
                }
 
-               $this->error = $this->format_error($error);
+               if ($error) {
+                       foreach (libxml_get_errors() as $error) {
+                               if ($error->level == LIBXML_ERR_FATAL) {
+                                       if(!isset($this->error)) //currently only the first error is reported
+                                               $this->error = $this->format_error($error);
+                                       $this->libxml_errors [] = $this->format_error($error);
+                               }
+                       }
+               }
                libxml_clear_errors();
 
                $this->items = array();
@@ -79,12 +101,13 @@ class FeedParser {
                $xpath->registerNamespace('slash', 'http://purl.org/rss/1.0/modules/slash/');
                $xpath->registerNamespace('dc', 'http://purl.org/dc/elements/1.1/');
                $xpath->registerNamespace('content', 'http://purl.org/rss/1.0/modules/content/');
+               $xpath->registerNamespace('thread', 'http://purl.org/syndication/thread/1.0');
 
                $this->xpath = $xpath;
 
                $root = $xpath->query("(//atom03:feed|//atom:feed|//channel|//rdf:rdf|//rdf:RDF)");
 
-               if ($root) {
+               if ($root && $root->length > 0) {
                        $root = $root->item(0);
 
                        if ($root) {
@@ -96,6 +119,7 @@ class FeedParser {
                                        $this->type = $this::FEED_RSS;
                                        break;
                                case "feed":
+                               case "atom:feed":
                                        $this->type = $this::FEED_ATOM;
                                        break;
                                default:
@@ -121,9 +145,14 @@ class FeedParser {
 
                                $link = $xpath->query("//atom:feed/atom:link[not(@rel)]")->item(0);
 
+                               if (!$link)
+                                       $link = $xpath->query("//atom:feed/atom:link[@rel='alternate']")->item(0);
+
                                if (!$link)
                                        $link = $xpath->query("//atom03:feed/atom03:link[not(@rel)]")->item(0);
 
+                               if (!$link)
+                                       $link = $xpath->query("//atom03:feed/atom03:link[@rel='alternate']")->item(0);
 
                                if ($link && $link->hasAttributes()) {
                                        $this->link = $link->getAttribute("href");
@@ -186,6 +215,10 @@ class FeedParser {
                                break;
 
                        }
+
+                       if ($this->title) $this->title = trim($this->title);
+                       if ($this->link) $this->link = trim($this->link);
+
                } else {
                        if( !isset($this->error) ){
                                $this->error = "Unknown/unsupported feed type";
@@ -208,6 +241,10 @@ class FeedParser {
                return $this->error;
        }
 
+       function errors() {
+               return $this->libxml_errors;
+       }
+
        function get_link() {
                return $this->link;
        }
@@ -229,7 +266,7 @@ class FeedParser {
 
                        foreach ($links as $link) {
                                if (!$rel || $link->hasAttribute('rel') && $link->getAttribute('rel') == $rel) {
-                                       array_push($rv, $link->getAttribute('href'));
+                                       array_push($rv, trim($link->getAttribute('href')));
                                }
                        }
                        break;
@@ -238,7 +275,7 @@ class FeedParser {
 
                        foreach ($links as $link) {
                                if (!$rel || $link->hasAttribute('rel') && $link->getAttribute('rel') == $rel) {
-                                       array_push($rv, $link->getAttribute('href'));
+                                       array_push($rv, trim($link->getAttribute('href')));
                                }
                        }
                        break;
@@ -246,4 +283,4 @@ class FeedParser {
 
                return $rv;
        }
-} ?>
+}