Update picofeed
This commit is contained in:
Frederic Guillot
parent
c73e3ff9ba
commit
c795d46669
|
|
@ -4,4 +4,4 @@
|
|||
|
||||
require_once __DIR__ . '/composer' . '/autoload_real.php';
|
||||
|
||||
return ComposerAutoloaderInita56cecf18737d4c6655b021e5f21a1a6::getLoader();
|
||||
return ComposerAutoloaderInit8ccf24e95a95febb275803014c1c9a9a::getLoader();
|
||||
|
|
|
|||
|
|
@ -2,7 +2,7 @@
|
|||
|
||||
// autoload_real.php @generated by Composer
|
||||
|
||||
class ComposerAutoloaderInita56cecf18737d4c6655b021e5f21a1a6
|
||||
class ComposerAutoloaderInit8ccf24e95a95febb275803014c1c9a9a
|
||||
{
|
||||
private static $loader;
|
||||
|
||||
|
|
@ -19,9 +19,9 @@ class ComposerAutoloaderInita56cecf18737d4c6655b021e5f21a1a6
|
|||
return self::$loader;
|
||||
}
|
||||
|
||||
spl_autoload_register(array('ComposerAutoloaderInita56cecf18737d4c6655b021e5f21a1a6', 'loadClassLoader'), true, true);
|
||||
spl_autoload_register(array('ComposerAutoloaderInit8ccf24e95a95febb275803014c1c9a9a', 'loadClassLoader'), true, true);
|
||||
self::$loader = $loader = new \Composer\Autoload\ClassLoader();
|
||||
spl_autoload_unregister(array('ComposerAutoloaderInita56cecf18737d4c6655b021e5f21a1a6', 'loadClassLoader'));
|
||||
spl_autoload_unregister(array('ComposerAutoloaderInit8ccf24e95a95febb275803014c1c9a9a', 'loadClassLoader'));
|
||||
|
||||
$map = require __DIR__ . '/autoload_namespaces.php';
|
||||
foreach ($map as $namespace => $path) {
|
||||
|
|
@ -42,14 +42,14 @@ class ComposerAutoloaderInita56cecf18737d4c6655b021e5f21a1a6
|
|||
|
||||
$includeFiles = require __DIR__ . '/autoload_files.php';
|
||||
foreach ($includeFiles as $file) {
|
||||
composerRequirea56cecf18737d4c6655b021e5f21a1a6($file);
|
||||
composerRequire8ccf24e95a95febb275803014c1c9a9a($file);
|
||||
}
|
||||
|
||||
return $loader;
|
||||
}
|
||||
}
|
||||
|
||||
function composerRequirea56cecf18737d4c6655b021e5f21a1a6($file)
|
||||
function composerRequire8ccf24e95a95febb275803014c1c9a9a($file)
|
||||
{
|
||||
require $file;
|
||||
}
|
||||
|
|
|
|||
|
|
@ -162,12 +162,12 @@
|
|||
"source": {
|
||||
"type": "git",
|
||||
"url": "https://github.com/fguillot/picoFeed.git",
|
||||
"reference": "acc16f1a0854fdaeae2416f1b12ee51a9c150b52"
|
||||
"reference": "8973f403ff6c16fb5200cfac44a58111c564b60d"
|
||||
},
|
||||
"dist": {
|
||||
"type": "zip",
|
||||
"url": "https://api.github.com/repos/fguillot/picoFeed/zipball/acc16f1a0854fdaeae2416f1b12ee51a9c150b52",
|
||||
"reference": "acc16f1a0854fdaeae2416f1b12ee51a9c150b52",
|
||||
"url": "https://api.github.com/repos/fguillot/picoFeed/zipball/8973f403ff6c16fb5200cfac44a58111c564b60d",
|
||||
"reference": "8973f403ff6c16fb5200cfac44a58111c564b60d",
|
||||
"shasum": ""
|
||||
},
|
||||
"require": {
|
||||
|
|
@ -181,7 +181,7 @@
|
|||
"suggest": {
|
||||
"ext-curl": "PicoFeed will use cURL if present"
|
||||
},
|
||||
"time": "2015-03-19 22:19:36",
|
||||
"time": "2015-03-25 18:09:25",
|
||||
"bin": [
|
||||
"picofeed"
|
||||
],
|
||||
|
|
|
|||
|
|
@ -205,6 +205,45 @@ $feed->items[0]->getContent(); // Item content (filtered or raw)
|
|||
$feed->items[0]->isRTL(); // Return true if the item language is Right-To-Left
|
||||
```
|
||||
|
||||
Get raw XML tags/attributes or non standard tags for items
|
||||
----------------------------------------------------------
|
||||
|
||||
Get the original `guid` tag for RSS 2.0 feeds:
|
||||
|
||||
```php
|
||||
echo $feed->items[0]->getTag('guid');
|
||||
```
|
||||
|
||||
Get a specific attribute value:
|
||||
|
||||
```php
|
||||
echo $feed->items[1]->getTag('category', 'term');
|
||||
```
|
||||
|
||||
Get value of namespaced tag:
|
||||
|
||||
```php
|
||||
echo $feed->items[1]->getTag('wfw:commentRss');
|
||||
```
|
||||
|
||||
Get attribute value of a namespaced tag:
|
||||
|
||||
```php
|
||||
echo $feed->items[0]->getTag('media:content', 'url');
|
||||
```
|
||||
|
||||
Get the xml of the item (returns a SimpleXMLElement instance):
|
||||
|
||||
```php
|
||||
$simplexml = $feed->items[0]->xml;
|
||||
```
|
||||
|
||||
Get the list of namespaces:
|
||||
|
||||
```php
|
||||
print_r($feed->items[0]->namespaces);
|
||||
```
|
||||
|
||||
RTL language detection
|
||||
----------------------
|
||||
|
||||
|
|
|
|||
|
|
@ -99,6 +99,47 @@ class Item
|
|||
*/
|
||||
public $language = '';
|
||||
|
||||
/**
|
||||
* Raw XML
|
||||
*
|
||||
* @access public
|
||||
* @var \SimpleXMLElement
|
||||
*/
|
||||
public $xml;
|
||||
|
||||
/**
|
||||
* List of namespaces
|
||||
*
|
||||
* @access public
|
||||
* @var array
|
||||
*/
|
||||
public $namespaces = array();
|
||||
|
||||
/**
|
||||
* Get specific XML tag or attribute value
|
||||
*
|
||||
* @access public
|
||||
* @param string $tag Tag name (examples: guid, media:content)
|
||||
* @param string $attribute Tag attribute
|
||||
* @return string
|
||||
*/
|
||||
public function getTag($tag, $attribute = '')
|
||||
{
|
||||
// Get namespaced value
|
||||
if (strpos($tag, ':') !== false) {
|
||||
list(,$tag) = explode(':', $tag);
|
||||
return XmlParser::getNamespaceValue($this->xml, $this->namespaces, $tag, $attribute);
|
||||
}
|
||||
|
||||
// Return attribute value
|
||||
if (! empty($attribute)) {
|
||||
return (string) $this->xml->{$tag}[$attribute];
|
||||
}
|
||||
|
||||
// Return tag content
|
||||
return (string) $this->xml->$tag;
|
||||
}
|
||||
|
||||
/**
|
||||
* Return item information
|
||||
*
|
||||
|
|
|
|||
|
|
@ -153,6 +153,9 @@ abstract class Parser
|
|||
foreach ($this->getItemsTree($xml) as $entry) {
|
||||
|
||||
$item = new Item;
|
||||
$item->xml = $entry;
|
||||
$item->namespaces = $this->namespaces;
|
||||
|
||||
$this->findItemAuthor($xml, $entry, $item);
|
||||
|
||||
$this->findItemUrl($entry, $item);
|
||||
|
|
|
|||
|
|
@ -212,21 +212,7 @@ class XmlParser
|
|||
}
|
||||
|
||||
/**
|
||||
* Extract charset from meta tag
|
||||
*
|
||||
* @static
|
||||
* @access public
|
||||
* @param string $data meta tag content
|
||||
* @return string
|
||||
*/
|
||||
public static function findCharset($data)
|
||||
{
|
||||
$result = explode('charset=', $data);
|
||||
return isset($result[1]) ? $result[1] : $data;
|
||||
}
|
||||
|
||||
/**
|
||||
* Get the encoding from a xml tag
|
||||
* Get the charset from a meta tag
|
||||
*
|
||||
* @static
|
||||
* @access public
|
||||
|
|
@ -237,18 +223,8 @@ class XmlParser
|
|||
{
|
||||
$encoding = '';
|
||||
|
||||
$dom = static::getHtmlDocument($data);
|
||||
$xpath = new DOMXPath($dom);
|
||||
|
||||
$tags = array(
|
||||
'/html/head/meta[translate(@http-equiv, "CENOPTY", "cenopty")="content-type"]/@content', //HTML4, convert upper to lower-case
|
||||
'/html/head/meta/@charset', //HTML5
|
||||
);
|
||||
|
||||
$nodes = $xpath->query(implode(' | ', $tags));
|
||||
|
||||
foreach ($nodes as $node) {
|
||||
$encoding = static::findCharset($node->nodeValue);
|
||||
if (preg_match('/<meta.*?charset\s*=\s*["\']?\s*([^"\'\s\/>;]+)/i', $data, $match) === 1) {
|
||||
$encoding = strtolower($match[1]);
|
||||
}
|
||||
|
||||
return $encoding;
|
||||
|
|
|
|||
|
|
@ -17,8 +17,68 @@ class XmlParserTest extends PHPUnit_Framework_TestCase
|
|||
|
||||
public function testGetEncodingFromMetaTag()
|
||||
{
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<html><head><meta content="text/html; charset=iso-8859-1" http-equiv="Content-Type"/></head></html>'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<html><head><meta charset="iso-8859-1"></head></html>'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv="Content-Type" content="text/html;charset=iso-8859-1"/>'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv="Content-Type" content="text/html;charset=iso-8859-1" />'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv=\'Content-Type\' content=\'text/html;charset=iso-8859-1\'/>'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv=\'Content-Type\' content=\'text/html;charset=iso-8859-1\' />'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv=Content-Type content=text/html;charset=iso-8859-1/>'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv=Content-Type content=text/html;charset=iso-8859-1 />'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv="Content-Type" content="text/html;charset=ISO-8859-1">'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv="Content-Type" content="text/html;charset=iso-8859-1" >'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv=\'Content-Type\' content=\'text/html;charset=iso-8859-1\'>'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv=\'Content-Type\' content=\'text/html;charset=iso-8859-1\' >'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv=Content-Type content=text/html;charset=iso-8859-1>'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv=Content-Type content=text/html;charset=iso-8859-1 >'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv="Content-Type" content="text/html;charset=\'iso-8859-1\'">'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv="Content-Type" content="\'text/html;charset=iso-8859-1\'">'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv="Content-Type" content="\'text/html\';charset=\'iso-8859-1\'">'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv=\'Content-Type\' content=\'text/html;charset="iso-8859-1"\'>'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv=\'Content-Type\' content=\'"text/html;charset=iso-8859-1"\'>'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv=\'Content-Type\' content=\'"text/html";charset="iso-8859-1"\'>'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv="Content-Type" content="text/html;;;charset=iso-8859-1">'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv="Content-Type" content="text/html;;;charset=\'iso-8859-1\'">'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv="Content-Type" content="\'text/html;;;charset=iso-8859-1\'">'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv="Content-Type" content="\'text/html\';;;charset=\'iso-8859-1\'">'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv=\'Content-Type\' content=\'text/html;;;charset=iso-8859-1\'>'));
|
||||
$this->assertEquals('windows-1251', XmlParser::getEncodingFromMetaTag('<meta http-equiv=\'Content-Type\' content=\'text/html;;;charset="windows-1251"\'>'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv=\'Content-Type\' content=\'"text/html;;;charset=iso-8859-1"\'>'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv=\'Content-Type\' content=\'"text/html";;;charset="iso-8859-1"\'>'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv = Content-Type content = text/html;charset=iso-8859-1 >'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta content = text/html;charset=iso-8859-1 http-equiv = Content-Type >'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv = Content-Type content = text/html ; charset = iso-8859-1 >'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta content = text/html ; charset = iso-8859-1 http-equiv = Content-Type >'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv = Content-Type content = text/html ;;; charset = iso-8859-1 >'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta content = text/html ;;; charset = iso-8859-1 http-equiv = Content-Type >'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv = Content-Type content = text/html ; ; ; charset = iso-8859-1 >'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta content = text/html ; ; ; charset = iso-8859-1 http-equiv = Content-Type >'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta charset="uTf-8"/>'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta charset="utf-8" />'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta charset=\'Utf-8\'/>'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta charset=\'utf-8\' />'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta charset=utf-8/>'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta charset=utf-8 />'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta charset="utf-8">'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta charset="utf-8" >'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta charset=\'utf-8\'>'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta charset=\'utf-8\' >'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta charset=utf-8>'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta charset=utf-8 >'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta charset = " utf-8 " >'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta charset = \' utf-8 \' >'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta charset = " utf-8 \' >'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta charset = \' utf-8 " >'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta charset = " utf-8 >'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta charset = \' utf-8 >'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta charset = utf-8 \' >'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta charset = utf-8 " >'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta charset = utf-8 >'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta charset = utf-8 />'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta name="title" value="charset=utf-8 — is it really useful (yep)?">'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta value="charset=utf-8 — is it really useful (yep)?" name="title">'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta name="title" content="charset=utf-8 — is it really useful (yep)?">'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta name="charset=utf-8" content="charset=utf-8 — is it really useful (yep)?">'));
|
||||
$this->assertEquals('utf-8', XmlParser::getEncodingFromMetaTag('<meta content="charset=utf-8 — is it really useful (nope, not here, but gotta admit pretty robust otherwise)?" name="title">'));
|
||||
$this->assertEquals('iso-8859-1', XmlParser::getEncodingFromMetaTag('<meta http-equiv="Content-Type" content="text/html;charset=iSo-8859-1"/><meta charset="invalid" />'));
|
||||
}
|
||||
|
||||
public function testGetEncodingFromXmlTag()
|
||||
|
|
|
|||
Loading…
Reference in New Issue