mirror of
https://github.com/RSS-Bridge/rss-bridge.git
synced 2025-08-03 17:44:48 +02:00
[core] extract BridgeAbstract methods to make them functions
- returnError, returnServerError, returnClientError ,debugMessage are moved to lib/error.php - getContents, getSimpleHTMLDOM, getSimpleHTMLDOMCached are moved to lib/contents.php Signed-off-by: Pierre Mazière <pierre.maziere@gmx.com>
This commit is contained in:
Pierre Mazière
parent
ad534444fa
commit
f1fb95b257
111 changed files with 485 additions and 477 deletions
|
|
@ -8,25 +8,11 @@ abstract class BridgeAbstract implements BridgeInterface {
|
|||
const MAINTAINER = 'No maintainer';
|
||||
const PARAMETERS = array();
|
||||
|
||||
public $useProxy = true;
|
||||
|
||||
protected $cache;
|
||||
protected $items = array();
|
||||
protected $inputs = array();
|
||||
protected $queriedContext = '';
|
||||
|
||||
protected function returnError($message, $code){
|
||||
throw new \HttpException($message, $code);
|
||||
}
|
||||
|
||||
protected function returnClientError($message){
|
||||
$this->returnError($message, 400);
|
||||
}
|
||||
|
||||
protected function returnServerError($message){
|
||||
$this->returnError($message, 500);
|
||||
}
|
||||
|
||||
/**
|
||||
* Return items stored in the bridge
|
||||
* @return mixed
|
||||
|
|
@ -260,7 +246,7 @@ abstract class BridgeAbstract implements BridgeInterface {
|
|||
|
||||
if(empty(static::PARAMETERS)){
|
||||
if(!empty($inputs)){
|
||||
$this->returnClientError('Invalid parameters value(s)');
|
||||
returnClientError('Invalid parameters value(s)');
|
||||
}
|
||||
|
||||
$this->collectData();
|
||||
|
|
@ -271,15 +257,15 @@ abstract class BridgeAbstract implements BridgeInterface {
|
|||
}
|
||||
|
||||
if(!$this->validateData($inputs)){
|
||||
$this->returnClientError('Invalid parameters value(s)');
|
||||
returnClientError('Invalid parameters value(s)');
|
||||
}
|
||||
|
||||
// Guess the paramter context from input data
|
||||
$this->queriedContext = $this->getQueriedContext($inputs);
|
||||
if(is_null($this->queriedContext)){
|
||||
$this->returnClientError('Required parameter(s) missing');
|
||||
returnClientError('Required parameter(s) missing');
|
||||
} elseif($this->queriedContext === false){
|
||||
$this->returnClientError('Mixed context parameters');
|
||||
returnClientError('Mixed context parameters');
|
||||
}
|
||||
|
||||
$this->setInputs($inputs, $this->queriedContext);
|
||||
|
|
@ -313,157 +299,4 @@ abstract class BridgeAbstract implements BridgeInterface {
|
|||
public function setCache(\CacheAbstract $cache){
|
||||
$this->cache = $cache;
|
||||
}
|
||||
|
||||
public function debugMessage($text){
|
||||
if(!file_exists('DEBUG')) {
|
||||
return;
|
||||
}
|
||||
|
||||
$backtrace = debug_backtrace(DEBUG_BACKTRACE_IGNORE_ARGS, 3);
|
||||
$calling = $backtrace[2];
|
||||
$message = $calling['file'] . ':'
|
||||
. $calling['line'] . ' class '
|
||||
. get_class($this) . '->'
|
||||
. $calling['function'] . ' - '
|
||||
. $text;
|
||||
|
||||
error_log($message);
|
||||
}
|
||||
|
||||
protected function getContents($url
|
||||
, $use_include_path = false
|
||||
, $context = null
|
||||
, $offset = 0
|
||||
, $maxlen = null){
|
||||
$contextOptions = array(
|
||||
'http' => array(
|
||||
'user_agent' => ini_get('user_agent')
|
||||
)
|
||||
);
|
||||
|
||||
if(defined('PROXY_URL') && $this->useProxy){
|
||||
$contextOptions['http']['proxy'] = PROXY_URL;
|
||||
$contextOptions['http']['request_fulluri'] = true;
|
||||
|
||||
if(is_null($context)){
|
||||
$context = stream_context_create($contextOptions);
|
||||
} else {
|
||||
$prevContext = $context;
|
||||
if(!stream_context_set_option($context, $contextOptions)){
|
||||
$context = $prevContext;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
if(is_null($maxlen)){
|
||||
$content = @file_get_contents($url, $use_include_path, $context, $offset);
|
||||
} else {
|
||||
$content = @file_get_contents($url, $use_include_path, $context, $offset, $maxlen);
|
||||
}
|
||||
|
||||
if($content === false)
|
||||
$this->debugMessage('Cant\'t download ' . $url);
|
||||
|
||||
// handle compressed data
|
||||
foreach($http_response_header as $header){
|
||||
if(stristr($header, 'content-encoding')){
|
||||
switch(true){
|
||||
case stristr($header, 'gzip'):
|
||||
$content = gzinflate(substr($content, 10, -8));
|
||||
break;
|
||||
case stristr($header, 'compress'):
|
||||
//TODO
|
||||
case stristr($header, 'deflate'):
|
||||
//TODO
|
||||
case stristr($header, 'brotli'):
|
||||
//TODO
|
||||
$this->returnServerError($header . '=> Not implemented yet');
|
||||
break;
|
||||
case stristr($header, 'identity'):
|
||||
break;
|
||||
default:
|
||||
$this->returnServerError($header . '=> Unknown compression');
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return $content;
|
||||
}
|
||||
|
||||
protected function getSimpleHTMLDOM($url
|
||||
, $use_include_path = false
|
||||
, $context = null
|
||||
, $offset = 0
|
||||
, $maxLen = null
|
||||
, $lowercase = true
|
||||
, $forceTagsClosed = true
|
||||
, $target_charset = DEFAULT_TARGET_CHARSET
|
||||
, $stripRN = true
|
||||
, $defaultBRText = DEFAULT_BR_TEXT
|
||||
, $defaultSpanText = DEFAULT_SPAN_TEXT){
|
||||
$content = $this->getContents($url, $use_include_path, $context, $offset, $maxLen);
|
||||
return str_get_html($content
|
||||
, $lowercase
|
||||
, $forceTagsClosed
|
||||
, $target_charset
|
||||
, $stripRN
|
||||
, $defaultBRText
|
||||
, $defaultSpanText);
|
||||
}
|
||||
|
||||
/**
|
||||
* Maintain locally cached versions of pages to avoid multiple downloads.
|
||||
* @param url url to cache
|
||||
* @param duration duration of the cache file in seconds (default: 24h/86400s)
|
||||
* @return content of the file as string
|
||||
*/
|
||||
public function getSimpleHTMLDOMCached($url
|
||||
, $duration = 86400
|
||||
, $use_include_path = false
|
||||
, $context = null
|
||||
, $offset = 0
|
||||
, $maxLen = null
|
||||
, $lowercase = true
|
||||
, $forceTagsClosed = true
|
||||
, $target_charset = DEFAULT_TARGET_CHARSET
|
||||
, $stripRN = true
|
||||
, $defaultBRText = DEFAULT_BR_TEXT
|
||||
, $defaultSpanText = DEFAULT_SPAN_TEXT){
|
||||
$this->debugMessage('Caching url ' . $url . ', duration ' . $duration);
|
||||
|
||||
$filepath = __DIR__ . '/../cache/pages/' . sha1($url) . '.cache';
|
||||
$this->debugMessage('Cache file ' . $filepath);
|
||||
|
||||
if(file_exists($filepath) && filectime($filepath) < time() - $duration){
|
||||
unlink ($filepath);
|
||||
$this->debugMessage('Cached file deleted: ' . $filepath);
|
||||
}
|
||||
|
||||
if(file_exists($filepath)){
|
||||
$this->debugMessage('Loading cached file ' . $filepath);
|
||||
touch($filepath);
|
||||
$content = file_get_contents($filepath);
|
||||
} else {
|
||||
$this->debugMessage('Caching ' . $url . ' to ' . $filepath);
|
||||
$dir = substr($filepath, 0, strrpos($filepath, '/'));
|
||||
|
||||
if(!is_dir($dir)){
|
||||
$this->debugMessage('Creating directory ' . $dir);
|
||||
mkdir($dir, 0777, true);
|
||||
}
|
||||
|
||||
$content = $this->getContents($url, $use_include_path, $context, $offset, $maxLen);
|
||||
if($content !== false){
|
||||
file_put_contents($filepath, $content);
|
||||
}
|
||||
}
|
||||
|
||||
return str_get_html($content
|
||||
, $lowercase
|
||||
, $forceTagsClosed
|
||||
, $target_charset
|
||||
, $stripRN
|
||||
, $defaultBRText
|
||||
, $defaultSpanText);
|
||||
}
|
||||
}
|
||||
|
|
|
|||
|
|
@ -9,46 +9,46 @@ abstract class FeedExpander extends BridgeAbstract {
|
|||
|
||||
public function collectExpandableDatas($url, $maxItems = -1){
|
||||
if(empty($url)){
|
||||
$this->returnServerError('There is no $url for this RSS expander');
|
||||
returnServerError('There is no $url for this RSS expander');
|
||||
}
|
||||
|
||||
$this->debugMessage('Loading from ' . $url);
|
||||
debugMessage('Loading from ' . $url);
|
||||
|
||||
/* Notice we do not use cache here on purpose:
|
||||
* we want a fresh view of the RSS stream each time
|
||||
*/
|
||||
$content = $this->getContents($url)
|
||||
or $this->returnServerError('Could not request ' . $url);
|
||||
$content = getContents($url)
|
||||
or returnServerError('Could not request ' . $url);
|
||||
$rssContent = simplexml_load_string($content);
|
||||
|
||||
$this->debugMessage('Detecting feed format/version');
|
||||
debugMessage('Detecting feed format/version');
|
||||
switch(true){
|
||||
case isset($rssContent->item[0]):
|
||||
$this->debugMessage('Detected RSS 1.0 format');
|
||||
debugMessage('Detected RSS 1.0 format');
|
||||
$this->feedType = "RSS_1_0";
|
||||
break;
|
||||
case isset($rssContent->channel[0]):
|
||||
$this->debugMessage('Detected RSS 0.9x or 2.0 format');
|
||||
debugMessage('Detected RSS 0.9x or 2.0 format');
|
||||
$this->feedType = "RSS_2_0";
|
||||
break;
|
||||
case isset($rssContent->entry[0]):
|
||||
$this->debugMessage('Detected ATOM format');
|
||||
debugMessage('Detected ATOM format');
|
||||
$this->feedType = "ATOM_1_0";
|
||||
break;
|
||||
default:
|
||||
$this->debugMessage('Unknown feed format/version');
|
||||
$this->returnServerError('The feed format is unknown!');
|
||||
debugMessage('Unknown feed format/version');
|
||||
returnServerError('The feed format is unknown!');
|
||||
break;
|
||||
}
|
||||
|
||||
$this->debugMessage('Calling function "collect_' . $this->feedType . '_data"');
|
||||
debugMessage('Calling function "collect_' . $this->feedType . '_data"');
|
||||
$this->{'collect_' . $this->feedType . '_data'}($rssContent, $maxItems);
|
||||
}
|
||||
|
||||
protected function collect_RSS_1_0_data($rssContent, $maxItems){
|
||||
$this->load_RSS_2_0_feed_data($rssContent->channel[0]);
|
||||
foreach($rssContent->item as $item){
|
||||
$this->debugMessage('parsing item ' . var_export($item, true));
|
||||
debugMessage('parsing item ' . var_export($item, true));
|
||||
$this->items[] = $this->parseItem($item);
|
||||
if($maxItems !== -1 && count($this->items) >= $maxItems) break;
|
||||
}
|
||||
|
|
@ -56,13 +56,13 @@ abstract class FeedExpander extends BridgeAbstract {
|
|||
|
||||
protected function collect_RSS_2_0_data($rssContent, $maxItems){
|
||||
$rssContent = $rssContent->channel[0];
|
||||
$this->debugMessage('RSS content is ===========\n'
|
||||
debugMessage('RSS content is ===========\n'
|
||||
. var_export($rssContent, true)
|
||||
. '===========');
|
||||
|
||||
$this->load_RSS_2_0_feed_data($rssContent);
|
||||
foreach($rssContent->item as $item){
|
||||
$this->debugMessage('parsing item ' . var_export($item, true));
|
||||
debugMessage('parsing item ' . var_export($item, true));
|
||||
$this->items[] = $this->parseItem($item);
|
||||
if($maxItems !== -1 && count($this->items) >= $maxItems) break;
|
||||
}
|
||||
|
|
@ -71,7 +71,7 @@ abstract class FeedExpander extends BridgeAbstract {
|
|||
protected function collect_ATOM_1_0_data($content, $maxItems){
|
||||
$this->load_ATOM_feed_data($content);
|
||||
foreach($content->entry as $item){
|
||||
$this->debugMessage('parsing item ' . var_export($item, true));
|
||||
debugMessage('parsing item ' . var_export($item, true));
|
||||
$this->items[] = $this->parseItem($item);
|
||||
if($maxItems !== -1 && count($this->items) >= $maxItems) break;
|
||||
}
|
||||
|
|
@ -190,7 +190,7 @@ abstract class FeedExpander extends BridgeAbstract {
|
|||
case 'ATOM_1_0':
|
||||
return $this->parseATOMItem($item);
|
||||
break;
|
||||
default: $this->returnClientError('Unknown version ' . $this->getInput('version') . '!');
|
||||
default: returnClientError('Unknown version ' . $this->getInput('version') . '!');
|
||||
}
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -16,6 +16,8 @@ require __DIR__ . '/FeedExpander.php';
|
|||
require __DIR__ . '/Cache.php';
|
||||
require __DIR__ . '/CacheAbstract.php';
|
||||
require __DIR__ . '/HTMLUtils.php';
|
||||
require __DIR__ . '/error.php';
|
||||
require __DIR__ . '/contents.php';
|
||||
|
||||
$vendorLibSimpleHtmlDom = __DIR__ . PATH_VENDOR . '/simplehtmldom/simple_html_dom.php';
|
||||
if( !file_exists($vendorLibSimpleHtmlDom) ){
|
||||
|
|
|
|||
142
lib/contents.php
Normal file
142
lib/contents.php
Normal file
|
|
@ -0,0 +1,142 @@
|
|||
<?php
|
||||
function getContents($url
|
||||
, $use_include_path = false
|
||||
, $context = null
|
||||
, $offset = 0
|
||||
, $maxlen = null
|
||||
){
|
||||
$contextOptions = array(
|
||||
'http' => array(
|
||||
'user_agent' => ini_get('user_agent')
|
||||
)
|
||||
);
|
||||
|
||||
if(defined('PROXY_URL') && !defined('NOPROXY')){
|
||||
$contextOptions['http']['proxy'] = PROXY_URL;
|
||||
$contextOptions['http']['request_fulluri'] = true;
|
||||
|
||||
if(is_null($context)){
|
||||
$context = stream_context_create($contextOptions);
|
||||
} else {
|
||||
$prevContext = $context;
|
||||
if(!stream_context_set_option($context, $contextOptions)){
|
||||
$context = $prevContext;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
if(is_null($maxlen)){
|
||||
$content = @file_get_contents($url, $use_include_path, $context, $offset);
|
||||
} else {
|
||||
$content = @file_get_contents($url, $use_include_path, $context, $offset, $maxlen);
|
||||
}
|
||||
|
||||
if($content === false)
|
||||
debugMessage('Cant\'t download ' . $url);
|
||||
|
||||
// handle compressed data
|
||||
foreach($http_response_header as $header){
|
||||
if(stristr($header, 'content-encoding')){
|
||||
switch(true){
|
||||
case stristr($header, 'gzip'):
|
||||
$content = gzinflate(substr($content, 10, -8));
|
||||
break;
|
||||
case stristr($header, 'compress'):
|
||||
//TODO
|
||||
case stristr($header, 'deflate'):
|
||||
//TODO
|
||||
case stristr($header, 'brotli'):
|
||||
//TODO
|
||||
returnServerError($header . '=> Not implemented yet');
|
||||
break;
|
||||
case stristr($header, 'identity'):
|
||||
break;
|
||||
default:
|
||||
returnServerError($header . '=> Unknown compression');
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return $content;
|
||||
}
|
||||
|
||||
function getSimpleHTMLDOM($url
|
||||
, $use_include_path = false
|
||||
, $context = null
|
||||
, $offset = 0
|
||||
, $maxLen = null
|
||||
, $lowercase = true
|
||||
, $forceTagsClosed = true
|
||||
, $target_charset = DEFAULT_TARGET_CHARSET
|
||||
, $stripRN = true
|
||||
, $defaultBRText = DEFAULT_BR_TEXT
|
||||
, $defaultSpanText = DEFAULT_SPAN_TEXT
|
||||
){
|
||||
$content = getContents($url, $use_include_path, $context, $offset, $maxLen);
|
||||
return str_get_html($content
|
||||
, $lowercase
|
||||
, $forceTagsClosed
|
||||
, $target_charset
|
||||
, $stripRN
|
||||
, $defaultBRText
|
||||
, $defaultSpanText);
|
||||
}
|
||||
|
||||
/**
|
||||
* Maintain locally cached versions of pages to avoid multiple downloads.
|
||||
* @param url url to cache
|
||||
* @param duration duration of the cache file in seconds (default: 24h/86400s)
|
||||
* @return content of the file as string
|
||||
*/
|
||||
function getSimpleHTMLDOMCached($url
|
||||
, $duration = 86400
|
||||
, $use_include_path = false
|
||||
, $context = null
|
||||
, $offset = 0
|
||||
, $maxLen = null
|
||||
, $lowercase = true
|
||||
, $forceTagsClosed = true
|
||||
, $target_charset = DEFAULT_TARGET_CHARSET
|
||||
, $stripRN = true
|
||||
, $defaultBRText = DEFAULT_BR_TEXT
|
||||
, $defaultSpanText = DEFAULT_SPAN_TEXT
|
||||
){
|
||||
debugMessage('Caching url ' . $url . ', duration ' . $duration);
|
||||
|
||||
$filepath = __DIR__ . '/../cache/pages/' . sha1($url) . '.cache';
|
||||
debugMessage('Cache file ' . $filepath);
|
||||
|
||||
if(file_exists($filepath) && filectime($filepath) < time() - $duration){
|
||||
unlink ($filepath);
|
||||
debugMessage('Cached file deleted: ' . $filepath);
|
||||
}
|
||||
|
||||
if(file_exists($filepath)){
|
||||
debugMessage('Loading cached file ' . $filepath);
|
||||
touch($filepath);
|
||||
$content = file_get_contents($filepath);
|
||||
} else {
|
||||
debugMessage('Caching ' . $url . ' to ' . $filepath);
|
||||
$dir = substr($filepath, 0, strrpos($filepath, '/'));
|
||||
|
||||
if(!is_dir($dir)){
|
||||
debugMessage('Creating directory ' . $dir);
|
||||
mkdir($dir, 0777, true);
|
||||
}
|
||||
|
||||
$content = getContents($url, $use_include_path, $context, $offset, $maxLen);
|
||||
if($content !== false){
|
||||
file_put_contents($filepath, $content);
|
||||
}
|
||||
}
|
||||
|
||||
return str_get_html($content
|
||||
, $lowercase
|
||||
, $forceTagsClosed
|
||||
, $target_charset
|
||||
, $stripRN
|
||||
, $defaultBRText
|
||||
, $defaultSpanText);
|
||||
}
|
||||
|
||||
?>
|
||||
30
lib/error.php
Normal file
30
lib/error.php
Normal file
|
|
@ -0,0 +1,30 @@
|
|||
<?php
|
||||
function returnError($message, $code){
|
||||
throw new \HttpException($message, $code);
|
||||
}
|
||||
|
||||
function returnClientError($message){
|
||||
returnError($message, 400);
|
||||
}
|
||||
|
||||
function returnServerError($message){
|
||||
returnError($message, 500);
|
||||
}
|
||||
|
||||
function debugMessage($text){
|
||||
if(!file_exists('DEBUG')) {
|
||||
return;
|
||||
}
|
||||
|
||||
$backtrace = debug_backtrace(DEBUG_BACKTRACE_IGNORE_ARGS, 3);
|
||||
$calling = $backtrace[2];
|
||||
$message = $calling['file'] . ':'
|
||||
. $calling['line'] . ' class '
|
||||
. $calling['class'] . '->'
|
||||
. $calling['function'] . ' - '
|
||||
. $text;
|
||||
|
||||
error_log($message);
|
||||
}
|
||||
|
||||
?>
|
||||
Loading…
Add table
Add a link
Reference in a new issue