Merge pull request #2 from RSS-Bridge/master

Update from base
2016-09-25 12:21:27 +02:00 · 2016-09-25 12:21:27 +02:00 · c5831e384f
commit c5831e384f
parent 47200752fb ad534444fa
3 changed files with 193 additions and 23 deletions
--- a/bridges/GithubIssueBridge.php
+++ b/bridges/GithubIssueBridge.php
@ -18,7 +18,12 @@ class GithubIssueBridge extends BridgeAbstract{
      )
    ),

-    'Project Issues'=>array(),
+    'Project Issues'=>array(
+      'c'=>array(
+        'name'=>'Show Issues Comments',
+        'type'=>'checkbox'
+      )
+    ),
    'Issue comments'=>array(
      'i'=>array(
        'name'=>'Issue number',
@ -32,7 +37,12 @@ class GithubIssueBridge extends BridgeAbstract{
    $name=$this->getInput('u').'/'.$this->getInput('p');
    switch($this->queriedContext){
    case 'Project Issues':
-      $name=static::NAME.'s '.$name;
+      if($this->getInput('c')){
+        $prefix=static::NAME.'s comments for ';
+      }else{
+        $prefix=static::NAME.'s for ';
+      }
+      $name=$prefix.$name;
      break;
    case 'Issue comments':
      $name=static::NAME.' '.$name.' #'.$this->getInput('i');
@ -42,48 +52,139 @@ class GithubIssueBridge extends BridgeAbstract{
  }

  public function getURI(){
-    $uri = static::URI.$this->getInput('u').'/'.$this->getInput('p').'/issues/';
+    $uri = static::URI.$this->getInput('u').'/'.$this->getInput('p').'/issues';
    if($this->queriedContext==='Issue comments'){
-      $uri.=$this->getInput('i');
+      $uri.='/'.$this->getInput('i');
+    }else if($this->getInput('c')){
+      $uri.='?q=is%3Aissue+sort%3Aupdated-desc';
    }
    return $uri;
  }

+  protected function extractIssueComment($issueNbr,$title,$comment){
+    $class=$comment->getAttribute('class');
+    $classes=explode(' ',$class);
+    $event=false;
+    if(in_array('discussion-item',$classes)){
+      $event=true;
+    }
+
+    $author='unknown';
+    if($comment->find('.author',0)){
+      $author=$comment->find('.author',0)->plaintext;
+    }
+
+    $uri=static::URI.$this->getInput('u').'/'.$this->getInput('p').'/issues/'
+      .$issueNbr;
+
+    $comment=$comment->firstChild();
+    if(!$event){
+      $comment=$comment->nextSibling();
+    }
+
+    if($event){
+      $title.=' / '.substr($class,strpos($class,'discussion-item-')+strlen('discussion-item-'));
+      if(!$comment->hasAttribute('id')){
+        $items=array();
+        $timestamp=strtotime($comment->find('relative-time',0)->getAttribute('datetime'));
+        $content=$comment->innertext;
+        while($comment=$comment->nextSibling()){
+          $item=array();
+          $item['author']=$author;
+          $item['title']=html_entity_decode($title,ENT_QUOTES,'UTF-8');
+          $item['timestamp']=$timestamp;
+          $item['content']=$content.'<p>'.$comment->children(1)->innertext.'</p>';
+          $item['uri']=$uri.'#'.$comment->children(1)->getAttribute('id');
+          $items[]=$item;
+        }
+        return $items;
+      }
+      $content=$comment->parent()->innertext;
+    }else{
+      $title.=' / '.trim($comment->firstChild()->plaintext);
+      $content="<pre>".$comment->find('.comment-body',0)->innertext."</pre>";
+    }
+
+    $item = array();
+    $item['author']=$author;
+    $item['uri']= $uri.'#'.$comment->getAttribute('id');
+    $item['title']=html_entity_decode($title,ENT_QUOTES,'UTF-8');
+    $item['timestamp']=strtotime($comment->find('relative-time',0)->getAttribute('datetime'));
+    $item['content']=$content;
+    return $item;
+  }
+
+  protected function extractIssueComments($issue){
+    $items=array();
+    $title=$issue->find('.gh-header-title',0)->plaintext;
+    $issueNbr=trim(substr($issue->find('.gh-header-number',0)->plaintext,1));
+    $comments=$issue->find('.js-discussion',0);
+    foreach($comments->children() as $comment){
+      $classes=explode(' ',$comment->getAttribute('class'));
+      if(in_array('discussion-item',$classes) ||
+        in_array('timeline-comment-wrapper',$classes)
+      ){
+        $item=$this->extractIssueComment($issueNbr,$title,$comment);
+        if(array_keys($item)!==range(0,count($item)-1)){
+          $item=array($item);
+        }
+        $items=array_merge($items,$item);
+      }
+    }
+    return $items;
+  }
+
  public function collectData(){
    $html = $this->getSimpleHTMLDOM($this->getURI())
-      or $this->returnServerError('No results for Github Issue '.$this->getInput('i').' in project '.$this->getInput('u').'/'.$this->getInput('p'));
+      or $this->returnServerError('No results for Github Issue '.$this->getURI());

    switch($this->queriedContext){
    case 'Issue comments':
-      foreach($html->find('.js-comment-container') as $comment){
-
-        $item = array();
-        $item['author']=$comment->find('img',0)->getAttribute('alt');
-
-        $comment=$comment->firstChild()->nextSibling();
-
-        $item['uri']=$this->getURI().'#'.$comment->getAttribute('id');
-        $item['title']=trim($comment->firstChild()->plaintext);
-        $item['timestamp']=strtotime($comment->find('relative-time',0)->getAttribute('datetime'));
-        $item['content']=$comment->find('.comment-body',0)->innertext;
-
-        $this->items[]=$item;
-      }
+      $this->items=$this->extractIssueComments($html);
      break;
    case 'Project Issues':
      foreach($html->find('.js-active-navigation-container .js-navigation-item') as $issue){
-        $item=array();
        $info=$issue->find('.opened-by',0);
+        $issueNbr=substr(trim($info->plaintext),1,strpos(trim($info->plaintext),' '));
+
+        $item=array();
+        $item['content']='';
+
+        if($this->getInput('c')){
+          $uri=static::URI.$this->getInput('u').'/'.$this->getInput('p').'/issues/'.$issueNbr;
+          $issue=$this->getSimpleHTMLDOMCached($uri,1800);
+          if($issue){
+            $this->items=array_merge($this->items,$this->extractIssueComments($issue));
+            continue;
+          }
+          $item['content']='Can not extract comments from '.$uri;
+        }
+
        $item['author']=$info->find('a',0)->plaintext;
        $item['timestamp']=strtotime($info->find('relative-time',0)->getAttribute('datetime'));
-        $item['title']=$issue->find('.js-navigation-open',0)->plaintext;
+        $item['title']=html_entity_decode(
+          $issue->find('.js-navigation-open',0)->plaintext,
+          ENT_QUOTES,
+          'UTF-8'
+        );
        $comments=$issue->find('.col-5',0)->plaintext;
-        $item['content']='Comments: '.($comments?$comments:'0');
+        $item['content'].="\n".'Comments: '.($comments?$comments:'0');
        $item['uri']=self::URI.$issue->find('.js-navigation-open',0)->getAttribute('href');
        $this->items[]=$item;
      }
      break;
    }
+
+    array_walk($this->items, function(&$item){
+      $item['content']=preg_replace('/\s+/',' ',$item['content']);
+      $item['content']=str_replace('href="/','href="'.static::URI,$item['content']);
+      $item['content']=str_replace(
+        'href="#',
+        'href="'.substr($item['uri'],0,strpos($item['uri'],'#')+1),
+        $item['content']
+      );
+      $item['title']=preg_replace('/\s+/',' ',$item['title']);
+    });
  }

  public function getCacheDuration(){
--- a/bridges/LegifranceJOBridge.php
+++ b/bridges/LegifranceJOBridge.php
@ -0,0 +1,70 @@
+<?php
+class LegifranceJOBridge extends BridgeAbstract{
+
+  const MAINTAINER = 'Pierre Mazière';
+  const NAME = 'Journal Officiel de la République Française';
+  const URI = 'https://www.legifrance.gouv.fr/affichJO.do';
+  const DESCRIPTION = 'Returns the laws and decrees officially registered daily in France';
+
+  const PARAMETERS=array();
+
+  private $author;
+  private $timestamp;
+  private $uri;
+
+  private function extractItem($section,$subsection=null,$origin=null){
+    $item=array();
+    $item['author']=$this->author;
+    $item['timestamp']=$this->timestamp;
+    $item['uri']=$this->uri.'#'.count($this->items);
+    $item['title']=$section->plaintext;
+
+    if(!is_null($origin)){
+      $item['title']='[ '.$item['title'].' / '.$subsection->plaintext.' ] '.$origin->plaintext;
+      $data=$origin;
+    }elseif(!is_null($subsection)){
+      $item['title']='[ '.$item['title'].' ] '.$subsection->plaintext;
+      $data=$subsection;
+    }else{
+      $data=$section;
+    }
+
+    $item['content']='';
+    foreach($data->nextSibling()->find('a') as $content){
+      $text=$content->plaintext;
+      $href=$content->nextSibling()->getAttribute('resource');
+      $item['content'].='<p><a href="'.$href.'">'.$text.'</a></p>';
+    }
+    return $item;
+  }
+
+  public function collectData(){
+    $html=$this->getSimpleHTMLDOM(self::URI)
+      or $this->returnServer('Unable to download '.self::URI);
+
+    $this->author=trim($html->find('h2.title',0)->plaintext);
+    $uri=$html->find('h2.titleELI',0)->plaintext;
+    $this->uri=trim(substr($uri,strpos($uri,'https')));
+    $this->timestamp=strtotime(substr($this->uri,strpos($this->uri,'eli/jo/')+strlen('eli/jo/')));
+
+    foreach($html->find('h3') as $section){
+      $subsections=$section->nextSibling()->find('h4');
+      foreach($subsections as $subsection){
+        $origins=$subsection->nextSibling()->find('h5');
+        foreach($origins as $origin){
+          $this->items[]=$this->extractItem($section,$subsection,$origin);
+        }
+        if(!empty($origins)){
+          continue;
+        }
+        $this->items[]=$this->extractItem($section,$subsection);
+      }
+      if(!empty($subsections)){
+        continue;
+      }
+      $this->items[]=$this->extractItem($section);
+    }
+  }
+}
+
+
--- a/bridges/SexactuBridge.php
+++ b/bridges/SexactuBridge.php
@ -4,7 +4,6 @@ class SexactuBridge extends BridgeAbstract{
 	const MAINTAINER = "Riduidel";
 	const NAME = "Sexactu";
 	const URI = "https://www.gqmagazine.fr";
-    const HTTPS_STATUS = HTTPS_STATUS_VALID;
 	const DESCRIPTION = "Sexactu via rss-bridge";

    public function collectData(){