Bug fix for RSS2 detection

xeoncross · Dec 16, 2013 · 1d3f58c · 1d3f58c
1 parent d0b7e9c
commit 1d3f58c
Show file tree

Hide file tree

Showing 4 changed files with 1,125 additions and 2 deletions.
diff --git a/lib/PicoFeed/Filter.php b/lib/PicoFeed/Filter.php
@@ -460,10 +460,20 @@ public function removeHTMLTags($data)
 
     public static function stripXmlTag($data)
     {
-        if (strpos($data, '<?xml') !== false) {
-            $data = substr($data, strrpos($data, '?>') + 2);
+        if (strpos($data, '<?xml ') !== false) {
+            $data = ltrim(substr($data, strpos($data, '?>') + 2));
         }
 
+        do {
+
+            $pos = strpos($data, '<?xml-stylesheet ');
+
+            if ($pos !== false) {
+                $data = ltrim(substr($data, strpos($data, '?>') + 2));
+            }
+
+        } while ($pos !== false && $pos < 200);
+
         return $data;
     }
 

diff --git a/tests/FilterTest.php b/tests/FilterTest.php
@@ -8,6 +8,49 @@
 
 class FilterTest extends PHPUnit_Framework_TestCase
 {
+    public function testStripXmlTag()
+    {
+        $data = file_get_contents('tests/fixtures/ezrss.it');
+        $this->assertEquals('<!DOC', substr(Filter::stripXmlTag($data), 0, 5));
+
+        $data = file_get_contents('tests/fixtures/fulltextrss.xml');
+        $this->assertEquals('<rss', substr(Filter::stripXmlTag($data), 0, 4));
+
+        $data = file_get_contents('tests/fixtures/sametmax.xml');
+        $this->assertEquals('<rss', substr(Filter::stripXmlTag($data), 0, 4));
+
+        $data = file_get_contents('tests/fixtures/grotte_barbu.xml');
+        $this->assertEquals('<rss', substr(Filter::stripXmlTag($data), 0, 4));
+
+        $data = file_get_contents('tests/fixtures/ibash.ru.xml');
+        $this->assertEquals('<rss', substr(Filter::stripXmlTag($data), 0, 4));
+
+        $data = file_get_contents('tests/fixtures/pcinpact.xml');
+        $this->assertEquals('<rss', substr(Filter::stripXmlTag($data), 0, 4));
+
+        $data = file_get_contents('tests/fixtures/resorts.xml');
+        $this->assertEquals('<rss', substr(Filter::stripXmlTag($data), 0, 4));
+
+        $data = file_get_contents('tests/fixtures/rue89.xml');
+        $this->assertEquals('<rss', substr(Filter::stripXmlTag($data), 0, 4));
+
+        $data = file_get_contents('tests/fixtures/cercle.psy.xml');
+        $this->assertEquals('<rss', substr(Filter::stripXmlTag($data), 0, 4));
+
+        $data = file_get_contents('tests/fixtures/lagrange.xml');
+        $this->assertEquals('<feed', substr(Filter::stripXmlTag($data), 0, 5));
+
+        $data = file_get_contents('tests/fixtures/atom.xml');
+        $this->assertEquals('<feed', substr(trim(Filter::stripXmlTag($data)), 0, 5));
+
+        $data = file_get_contents('tests/fixtures/atomsample.xml');
+        $this->assertEquals('<feed', substr(trim(Filter::stripXmlTag($data)), 0, 5));
+
+        $data = file_get_contents('tests/fixtures/planete-jquery.xml');
+        $this->assertEquals('<rdf:RDF', trim(substr(trim(Filter::stripXmlTag($data)), 0, 8)));
+    }
+
+
     public function testRelativeScheme()
     {
         $f = new Filter('<a href="//linuxfr.org">link</a>', 'http://blabla');

diff --git a/tests/ReaderTest.php b/tests/ReaderTest.php
@@ -32,6 +32,9 @@ public function testDownloadWithCache()
 
     public function testDetectFormat()
     {
+        $reader = new Reader(file_get_contents('tests/fixtures/sametmax.xml'));
+        $this->assertInstanceOf('PicoFeed\Parsers\Rss20', $reader->getParser());
+
         $reader = new Reader(file_get_contents('tests/fixtures/rss_0.92.xml'));
         $this->assertInstanceOf('PicoFeed\Parsers\Rss92', $reader->getParser());