miniflux-legacy/vendor/fguillot/picofeed/lib/PicoFeed/Reader/Favicon.php

239 lines
5.3 KiB
PHP
Raw Normal View History

2014-10-19 20:42:31 +02:00
<?php
namespace PicoFeed\Reader;
2014-10-19 20:42:31 +02:00
use DOMXpath;
use PicoFeed\Client\Client;
use PicoFeed\Client\ClientException;
use PicoFeed\Client\Url;
use PicoFeed\Config\Config;
use PicoFeed\Logging\Logger;
use PicoFeed\Parser\XmlParser;
2014-10-19 20:42:31 +02:00
/**
* Favicon class
*
* https://en.wikipedia.org/wiki/Favicon
*
* @author Frederic Guillot
* @package Reader
2014-10-19 20:42:31 +02:00
*/
class Favicon
{
2014-12-24 23:54:27 +01:00
/**
* Valid types for favicon (supported by browsers)
*
* @access private
* @var array
*/
private $types = array(
'image/png',
'image/gif',
'image/x-icon',
'image/jpeg',
'image/jpg',
);
2014-10-19 20:42:31 +02:00
/**
* Config class instance
*
* @access private
* @var \PicoFeed\Config\Config
2014-10-19 20:42:31 +02:00
*/
private $config;
2014-10-19 20:42:31 +02:00
/**
* Icon binary content
2014-10-19 20:42:31 +02:00
*
* @access private
* @var string
*/
private $content = '';
/**
* Icon content type
*
* @access private
* @var string
*/
private $content_type = '';
2014-10-19 20:42:31 +02:00
/**
* Constructor
*
* @access public
* @param \PicoFeed\Config\Config $config Config class instance
2014-10-19 20:42:31 +02:00
*/
public function __construct(Config $config = null)
{
$this->config = $config ?: new Config;
}
/**
* Get the icon file content (available only after the download)
*
* @access public
* @return string
*/
public function getContent()
{
return $this->content;
}
/**
* Get the icon file type (available only after the download)
*
* @access public
* @return string
*/
public function getType()
{
2014-12-24 23:54:27 +01:00
foreach ($this->types as $type) {
if (strpos($this->content_type, $type) === 0) {
return $type;
}
}
return 'image/x-icon';
}
/**
* Get data URI (http://en.wikipedia.org/wiki/Data_URI_scheme)
*
* @access public
* @return string
*/
public function getDataUri()
{
2014-12-25 00:12:41 +01:00
if (empty($this->content)) {
return '';
}
return sprintf(
'data:%s;base64,%s',
2014-12-24 23:54:27 +01:00
$this->getType(),
base64_encode($this->content)
);
}
2014-10-19 20:42:31 +02:00
/**
* Download and check if a resource exists
*
* @access public
* @param string $url URL
* @return \PicoFeed\Client Client instance
2014-10-19 20:42:31 +02:00
*/
public function download($url)
{
$client = Client::getInstance();
$client->setConfig($this->config);
Logger::setMessage(get_called_class().' Download => '.$url);
try {
$client->execute($url);
}
catch (ClientException $e) {
Logger::setMessage(get_called_class().' Download Failed => '.$e->getMessage());
2014-10-19 20:42:31 +02:00
}
return $client;
2014-10-19 20:42:31 +02:00
}
/**
* Check if a remote file exists
*
* @access public
* @param string $url URL
* @return boolean
*/
public function exists($url)
{
return $this->download($url)->getContent() !== '';
2014-10-19 20:42:31 +02:00
}
/**
* Get the icon link for a website
*
* @access public
* @param string $website_link URL
* @param string $favicon_link optional URL
2014-10-19 20:42:31 +02:00
* @return string
*/
public function find($website_link, $favicon_link = '')
2014-10-19 20:42:31 +02:00
{
$website = new Url($website_link);
if ($favicon_link !== '') {
$icons = array($favicon_link);
} else {
$icons = $this->extract($this->download($website->getBaseUrl('/'))->getContent());
$icons[] = $website->getBaseUrl('/favicon.ico');
}
2014-10-19 20:42:31 +02:00
foreach ($icons as $icon_link) {
$icon_link = $this->convertLink($website, new Url($icon_link));
$resource = $this->download($icon_link);
$this->content = $resource->getContent();
$this->content_type = $resource->getContentType();
2014-10-19 20:42:31 +02:00
if ($this->content !== '') {
return $icon_link;
} elseif ($favicon_link !== '') {
return $this->find($website_link);
2014-10-19 20:42:31 +02:00
}
}
return '';
}
/**
* Convert icon links to absolute url
*
* @access public
* @param \PicoFeed\Client\Url $website Website url
* @param \PicoFeed\Client\Url $icon Icon url
2014-10-19 20:42:31 +02:00
* @return string
*/
public function convertLink(Url $website, Url $icon)
{
$base_url = '';
if ($icon->isRelativeUrl()) {
$base_url = $website->getBaseUrl();
}
else if ($icon->isProtocolRelative()) {
$icon->setScheme($website->getScheme());
}
return $icon->getAbsoluteUrl($base_url);
}
/**
* Extract the icon links from the HTML
*
* @access public
* @param string $html HTML
* @return array
*/
public function extract($html)
{
$icons = array();
if (empty($html)) {
return $icons;
}
$dom = XmlParser::getHtmlDocument($html);
$xpath = new DOMXpath($dom);
$elements = $xpath->query("//link[contains(@rel, 'icon') and not(contains(@rel, 'apple'))]");
for ($i = 0; $i < $elements->length; $i++) {
$icons[] = $elements->item($i)->getAttribute('href');
}
return $icons;
}
}