boyska/rss-bridge
1
0
Fork 0
forked from blallo/rss-bridge
Code Releases Activity

WikipediaBridge: improved title detection

Browse source 
* improved title detection through changed AddTodaysFeaturedArticleGeneric function call
* added Dutch Wikipedia support
This commit is contained in:
Frans de Jonge 2016-08-28 11:22:37 +02:00
parent 737942ae7a
commit 0fcde196b9
1 changed files with 31 additions and 9 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

40
bridges/WikipediaBridge.php
View file

@ -19,9 +19,10 @@ class WikipediaBridge extends BridgeAbstract{
'exampleValue'=>'English', 'exampleValue'=>'English',
'values'=>array( 'values'=>array(
'English'=>'en', 'English'=>'en',
'German'=>'de', 'Dutch'=>'nl',
'Esperanto'=>'es',
'French'=>'fr', 'French'=>'fr',
'Esperanto'=>'es' 'German'=>'de',
) )
), ),
'subject'=>array( 'subject'=>array(
@ -101,7 +102,7 @@ class WikipediaBridge extends BridgeAbstract{
$function = 'GetContents' . strtoupper($params['language']['value']); $function = 'GetContents' . strtoupper($params['language']['value']);
if(!method_exists($this, $function)) if(!method_exists($this, $function))
$this->returnServerError('A function to get the contents for your langauage is missing (\'' . $function . '\')!'); $this->returnServerError('A function to get the contents for your language is missing (\'' . $function . '\')!');
/* /*
* The method takes care of creating all items. * The method takes care of creating all items.
@ -134,15 +135,18 @@ class WikipediaBridge extends BridgeAbstract{
/* /*
* Adds a new item to $items using a generic operation (should work for most (all?) wikis) * Adds a new item to $items using a generic operation (should work for most (all?) wikis)
* $anchorText can be specified if the wiki in question doesn't use '...' (like Dutch, French and Italian)
* $anchorFallbackIndex can be used to specify a different fallback link than the first (e.g., -1 for the last)
*/ */
private function AddTodaysFeaturedArticleGeneric($element, $fullArticle){ private function AddTodaysFeaturedArticleGeneric($element, $fullArticle, $anchorText = '...', $anchorFallbackIndex = 0){
// Clean the bottom of the featured article // Clean the bottom of the featured article
$element->find('div', -1)->outertext = ''; if ($element->find('div', -1))
$element->find('div', -1)->outertext = '';
// The title and URI of the article is best defined in an anchor containint the string '...' ('full article ...') // The title and URI of the article can be found in an anchor containing the string '...' in most wikis ('full article ...')
$target = $element->find('p/a', 0); // We'll use the first anchor as fallback $target = $element->find('p/a', $anchorFallbackIndex);
foreach($element->find('//a') as $anchor){ foreach($element->find('//a') as $anchor){
if(strpos($anchor->innertext, '...') !== false){ if(strpos($anchor->innertext, $anchorText) !== false){
$target = $anchor; $target = $anchor;
break; break;
} }
@ -230,7 +234,7 @@ class WikipediaBridge extends BridgeAbstract{
switch($subject){ switch($subject){
case WIKIPEDIA_SUBJECT_TFA: case WIKIPEDIA_SUBJECT_TFA:
$element = $html->find('div[id=accueil-lumieresur]', 0); $element = $html->find('div[id=accueil-lumieresur]', 0);
$this->AddTodaysFeaturedArticleGeneric($element, $fullArticle); $this->AddTodaysFeaturedArticleGeneric($element, $fullArticle, 'Lire la suite');
break; break;
case WIKIPEDIA_SUBJECT_DYK: case WIKIPEDIA_SUBJECT_DYK:
$element = $html->find('div[id=SaviezVous]', 0); $element = $html->find('div[id=SaviezVous]', 0);
@ -276,4 +280,22 @@ class WikipediaBridge extends BridgeAbstract{
break; break;
} }
} }
/**
* Implementation for nl.wikipedia.org
*/
private function GetContentsNL($html, $subject, $fullArticle){
switch($subject){
case WIKIPEDIA_SUBJECT_TFA:
$element = $html->find('div[id=mf-uitgelicht]', 0);
$this->AddTodaysFeaturedArticleGeneric($element, $fullArticle, 'Lees meer');
break;
case WIKIPEDIA_SUBJECT_DYK:
$element = $html->find('div[id=mw-content-text]', 0)->find('table', 4)->find('td', 2);
$this->AddDidYouKnowGeneric($element, $fullArticle);
break;
default:
break;
}
}
} }