From c3d7d830b4989c9ec4d56ec49b6b3ca3c44a9d20 Mon Sep 17 00:00:00 2001 From: schneefux Date: Tue, 2 Sep 2014 13:49:17 +0200 Subject: first commit --- tests/charset.php | 60 --------------------------------------------------- tests/charsetcurl.php | 11 ---------- tests/showmods.php | 9 -------- tests/xpath.php | 33 ---------------------------- 4 files changed, 113 deletions(-) delete mode 100644 tests/charset.php delete mode 100644 tests/charsetcurl.php delete mode 100644 tests/showmods.php delete mode 100644 tests/xpath.php (limited to 'tests') diff --git a/tests/charset.php b/tests/charset.php deleted file mode 100644 index f429241..0000000 --- a/tests/charset.php +++ /dev/null @@ -1,60 +0,0 @@ - 'xpath', - 'xpath' => 'div[@class="meldung_wrapper"]', -); - -// http://www.heise.de/newsticker/heise-atom.xml - -$article = array( - 'link' => 'http://www.heise.de/newsticker/meldung/Fruehjahrspatches-Microsoft-9-Adobe-3-1838175.html/from/atom10', - 'content' => 'This is the feed content', - 'plugin_data' => '', -); - -$doc = new DOMDocument(); - -$html = file_get_contents($article['link']); - -// BEGIN --- New code -$headers = $http_response_header; -$content_type = false; -foreach ($headers as $h) { - if (substr(strtolower($h), 0, 13) == 'content-type:') { - $content_type = substr($h, 14); - // don't break here to find LATEST (if redirected) entry - } -} - -$charset = false; -if ($content_type) { - preg_match('/charset=(\S+)/', $content_type, $matches); - if (isset($matches[1]) && !empty($matches[1])) $charset = $matches[1]; -} - -// END --- New code - -echo 'CHARSET: ' . $charset . PHP_EOL; - - -$doc->loadHTML('' . $html); - -echo 'ENCODING: ' . $doc->encoding . PHP_EOL; - -if ($doc) { - $basenode = false; - $xpath = new DOMXPath($doc); - $entries = $xpath->query('(//'.$config['xpath'].')'); // find main DIV according to config - - var_dump($entries); - - if ($entries->length > 0) $basenode = $entries->item(0); - - if ($basenode) { - $article['content'] = $doc->saveXML($basenode); - $article['plugin_data'] = "feedmod,$owner_uid:" . $article['plugin_data']; - } -} - -print_r($article); diff --git a/tests/charsetcurl.php b/tests/charsetcurl.php deleted file mode 100644 index 9db93e8..0000000 --- a/tests/charsetcurl.php +++ /dev/null @@ -1,11 +0,0 @@ - 'xpath', - 'xpath' => 'div[@itemprop="articleBody"]', -); - -$article = array( - 'link' => 'http://www.der-postillon.com/2013/04/nordkoreas-armee-nach-wochenlangem.html', - 'content' => 'This is the feed content', - 'plugin_data' => '', -); - -$doc = new DOMDocument(); -$html = file_get_contents($article['link']); -$doc->loadHTML($html); - -if ($doc) { - $basenode = false; - $xpath = new DOMXPath($doc); - $entries = $xpath->query('(//'.$config['xpath'].')'); // find main DIV according to config - - var_dump($entries); - - if ($entries->length > 0) $basenode = $entries->item(0); - - if ($basenode) { - $article['content'] = $doc->saveXML($basenode); - $article['plugin_data'] = "feedmod,$owner_uid:" . $article['plugin_data']; - } -} - -print_r($article); \ No newline at end of file -- cgit v1.3.1