X-Git-Url: https://git.wh0rd.org/?a=blobdiff_plain;f=classes%2Ffeedparser.php;h=a5e406149d46062c620b60a6ce53738f0b0d13d8;hb=HEAD;hp=8cb736a8b5fd8a2b795dd557b1b3a14c7c0ecff9;hpb=852d4ac89084edbee98df89ab066266cd9df63c9;p=tt-rss.git

diff --git a/classes/feedparser.php b/classes/feedparser.php
index 8cb736a8..a5e40614 100644
--- 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;
@@ -12,12 +13,79 @@ 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();
 		$this->doc = new DOMDocument();
 		$this->doc->loadXML($data);
-		$this->error = $this->format_error(libxml_get_last_error());
+
+		mb_substitute_character("none");
+
+		$error = libxml_get_last_error();
+
+		// libxml compiled without iconv?
+		if ($error && $error->code == 32) {
+			$data = $this->normalize_encoding($data);
+
+			if ($data) {
+				libxml_clear_errors();
+
+				$this->doc = new DOMDocument();
+				$this->doc->loadXML($data);
+
+				$error = libxml_get_last_error();
+			}
+		}
+
+		// some terrible invalid unicode entity?
+		if ($error) {
+			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', ' ', $data);
+
+					if ($data) {
+						libxml_clear_errors();
+
+						$this->doc = new DOMDocument();
+						$this->doc->loadXML($data);
+
+						$error = libxml_get_last_error();
+					}
+					break;
+				}
+			}
+		}
+
+		if ($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();
@@ -27,26 +95,39 @@ class FeedParser {
 		$root = $this->doc->firstChild;
 		$xpath = new DOMXPath($this->doc);
 		$xpath->registerNamespace('atom', 'http://www.w3.org/2005/Atom');
+		$xpath->registerNamespace('atom03', 'http://purl.org/atom/ns#');
 		$xpath->registerNamespace('media', 'http://search.yahoo.com/mrss/');
 		$xpath->registerNamespace('rdf', 'http://www.w3.org/1999/02/22-rdf-syntax-ns#');
-		$this->xpath = $xpath;
+		$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');
 
-		$root = $xpath->query("(//atom:feed|//channel|//rdf:rdf|//rdf:RDF)")->item(0);
+		$this->xpath = $xpath;
 
-		if ($root) {
-			switch (mb_strtolower($root->tagName)) {
-			case "rdf:rdf":
-				$this->type = $this::FEED_RDF;
-				break;
-			case "channel":
-				$this->type = $this::FEED_RSS;
-				break;
-			case "feed":
-				$this->type = $this::FEED_ATOM;
-				break;
-			default:
-				$this->error = "Unknown/unsupported feed type";
-				return;
+		$root = $xpath->query("(//atom03:feed|//atom:feed|//channel|//rdf:rdf|//rdf:RDF)");
+
+		if ($root && $root->length > 0) {
+			$root = $root->item(0);
+
+			if ($root) {
+				switch (mb_strtolower($root->tagName)) {
+				case "rdf:rdf":
+					$this->type = $this::FEED_RDF;
+					break;
+				case "channel":
+					$this->type = $this::FEED_RSS;
+					break;
+				case "feed":
+				case "atom:feed":
+					$this->type = $this::FEED_ATOM;
+					break;
+				default:
+					if( !isset($this->error) ){
+						$this->error = "Unknown/unsupported feed type";
+					}
+					return;
+				}
 			}
 
 			switch ($this->type) {
@@ -54,25 +135,40 @@ class FeedParser {
 
 				$title = $xpath->query("//atom:feed/atom:title")->item(0);
 
+				if (!$title)
+					$title = $xpath->query("//atom03:feed/atom03:title")->item(0);
+
+
 				if ($title) {
 					$this->title = $title->nodeValue;
 				}
 
 				$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");
 				}
 
 				$articles = $xpath->query("//atom:entry");
 
+				if (!$articles || $articles->length == 0)
+					$articles = $xpath->query("//atom03:entry");
+
 				foreach ($articles as $article) {
 					array_push($this->items, new FeedItem_Atom($article, $this->doc, $this->xpath));
 				}
 
 				break;
 			case $this::FEED_RSS:
-
 				$title = $xpath->query("//channel/title")->item(0);
 
 				if ($title) {
@@ -81,8 +177,11 @@ class FeedParser {
 
 				$link = $xpath->query("//channel/link")->item(0);
 
-				if ($link && $link->hasAttributes()) {
-					$this->link = $link->getAttribute("href");
+				if ($link) {
+					if ($link->getAttribute("href"))
+						$this->link = $link->getAttribute("href");
+					else if ($link->nodeValue)
+						$this->link = $link->nodeValue;
 				}
 
 				$articles = $xpath->query("//channel/item");
@@ -116,8 +215,14 @@ class FeedParser {
 				break;
 
 			}
+
+			if ($this->title) $this->title = trim($this->title);
+			if ($this->link) $this->link = trim($this->link);
+
 		} else {
-			$this->error = "Unknown/unsupported feed type";
+			if( !isset($this->error) ){
+				$this->error = "Unknown/unsupported feed type";
+			}
 			return;
 		}
 	}
@@ -136,6 +241,10 @@ class FeedParser {
 		return $this->error;
 	}
 
+	function errors() {
+		return $this->libxml_errors;
+	}
+
 	function get_link() {
 		return $this->link;
 	}
@@ -157,15 +266,16 @@ 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;
 		case $this::FEED_RSS:
-			$links = $this->xpath->query("//channel/link");
+			$links = $this->xpath->query("//atom:link");
+
 			foreach ($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;
@@ -173,4 +283,4 @@ class FeedParser {
 
 		return $rv;
 	}
-} ?>
+}