From 933c59c405a303bbbc2f7e5b87d913b4be06f95b Mon Sep 17 00:00:00 2001
From: Laurent Destailleur <eldy@users.sourceforge.net>
Date: Fri, 26 Aug 2011 22:38:27 +0000
Subject: [PATCH] Rss parser can use xml_parse or simplexml functions
---
htdocs/core/class/rssparser.class.php | 442 +++++++++++++++++++++++---
1 file changed, 399 insertions(+), 43 deletions(-)
diff --git a/htdocs/core/class/rssparser.class.php b/htdocs/core/class/rssparser.class.php
index 4a8277f52bf..d0d1af6f895 100755
--- a/htdocs/core/class/rssparser.class.php
+++ b/htdocs/core/class/rssparser.class.php
@@ -19,7 +19,7 @@
* \file htdocs/core/class/rssparser.class.php
* \ingroup core
* \brief File of class to parse rss feeds
- * \version $Id: rssparser.class.php,v 1.3 2011/08/26 19:09:02 eldy Exp $
+ * \version $Id: rssparser.class.php,v 1.4 2011/08/26 22:38:27 eldy Exp $
*/
class RssParser
{
@@ -53,6 +53,11 @@ class RssParser
public function getLastFetchDate() { return $this->_lastfetchdate; }
public function getItems() { return $this->_rssarray; }
+
+ // For parsing with xmlparser
+ var $stack = array(); // parser stack
+
+
/**
* Constructor
*/
@@ -75,6 +80,8 @@ class RssParser
{
include_once(DOL_DOCUMENT_ROOT.'/lib/files.lib.php');
+ $str=''; // This will contain content of feed
+
// Check parameters
if (! dol_is_url($urlRSS))
{
@@ -107,41 +114,54 @@ class RssParser
}
}
- // Load file into $rss
+ // Load file into $str
if ($foundintocache) // Cache file found and is not too old
{
$str = file_get_contents($newpathofdestfile);
- $rss = simplexml_load_string(unserialize($str));
}
else
{
try {
ini_set("user_agent","Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.0)");
ini_set("max_execution_time", 10);
- if (! empty($conf->global->MAIN_SIMPLEXMLLOAD_DEBUG)) $rss = simplexml_load_file($this->_urlRSS);
- else
- {
- //libxml_use_internal_errors(false);
- $rss = @simplexml_load_file($this->_urlRSS);
- }
+ $str = file_get_contents($this->_urlRSS);
}
catch (Exception $e) {
print 'Error retrieving URL '.$this->urlRSS.' - '.$e->getMessage();
}
}
+ // Convert $str into xml
+ if (! empty($conf->global->EXTERNALRSS_USE_SIMPLEXML))
+ {
+ //print 'xx'.LIBXML_NOCDATA;
+ libxml_use_internal_errors(false);
+ $rss = simplexml_load_string($str, "SimpleXMLElement", LIBXML_NOCDATA);
+ }
+ else
+ {
+ $xmlparser=xml_parser_create('');
+ if (!is_resource($xmlparser)) { $this->error="ErrorFailedToCreateParser"; return -1; }
+
+ xml_set_object( $xmlparser, $this );
+ xml_set_element_handler($xmlparser, 'feed_start_element', 'feed_end_element' );
+ xml_set_character_data_handler( $xmlparser, 'feed_cdata' );
+ $status = xml_parse( $xmlparser, $str );
+ xml_parser_free( $xmlparser );
+ $rss=$this;
+ //var_dump($this);exit;
+ }
+
// If $rss loaded
if ($rss)
{
- $items=array();
-
// Save file into cache
if (empty($foundintocache) && $cachedir)
{
dol_syslog("RssParser::parser cache file ".$newpathofdestfile." is saved onto disk.");
if (! dol_is_dir($cachedir)) dol_mkdir($cachedir);
$fp = fopen($newpathofdestfile, 'w');
- fwrite($fp, serialize($rss->asXML()));
+ fwrite($fp, $str);
fclose($fp);
if (! empty($conf->global->MAIN_UMASK)) $newmask=$conf->global->MAIN_UMASK;
@chmod($newpathofdestfile, octdec($newmask));
@@ -149,60 +169,131 @@ class RssParser
$this->_lastfetchdate=$nowgmt;
}
- $rss->_format='rss';
- if (empty($rss->channel)) $rss->_format='atom';
+ unset($str); // Free memory
+
+ if (empty($rss->_format)) // If format not detected automatically
+ {
+ $rss->_format='rss';
+ if (empty($rss->channel)) $rss->_format='atom';
+ }
+
+ $items=array();
// Save description entries
if ($rss->_format == 'rss')
{
- if (!empty($rss->channel->language)) $this->_language = (string) $rss->channel->language;
- if (!empty($rss->channel->generator)) $this->_generator = (string) $rss->channel->generator;
- if (!empty($rss->channel->copyright)) $this->_copyright = (string) $rss->channel->copyright;
- if (!empty($rss->channel->lastbuilddate)) $this->_lastbuilddate = (string) $rss->channel->lastbuilddate;
- if (!empty($rss->channel->image->url[0])) $this->_imageurl = (string) $rss->channel->image->url[0];
- if (!empty($rss->channel->link)) $this->_link = (string) $rss->channel->link;
- if (!empty($rss->channel->title)) $this->_title = (string) $rss->channel->title;
- if (!empty($rss->channel->description)) $this->_description = (string) $rss->channel->description;
- $items=$rss->channel->item;
+ //var_dump($rss);
+ if (! empty($conf->global->EXTERNALRSS_USE_SIMPLEXML))
+ {
+ if (!empty($rss->channel->language)) $this->_language = (string) $rss->channel->language;
+ if (!empty($rss->channel->generator)) $this->_generator = (string) $rss->channel->generator;
+ if (!empty($rss->channel->copyright)) $this->_copyright = (string) $rss->channel->copyright;
+ if (!empty($rss->channel->lastbuilddate)) $this->_lastbuilddate = (string) $rss->channel->lastbuilddate;
+ if (!empty($rss->channel->image->url[0])) $this->_imageurl = (string) $rss->channel->image->url[0];
+ if (!empty($rss->channel->link)) $this->_link = (string) $rss->channel->link;
+ if (!empty($rss->channel->title)) $this->_title = (string) $rss->channel->title;
+ if (!empty($rss->channel->description)) $this->_description = (string) $rss->channel->description;
+ }
+ else
+ {
+ if (!empty($rss->channel['rss_language'])) $this->_language = (string) $rss->channel['rss_language'];
+ if (!empty($rss->channel['rss_generator'])) $this->_generator = (string) $rss->channel['rss_generator'];
+ if (!empty($rss->channel['rss_copyright'])) $this->_copyright = (string) $rss->channel['rss_copyright'];
+ if (!empty($rss->channel['rss_lastbuilddate'])) $this->_lastbuilddate = (string) $rss->channel['rss_lastbuilddate'];
+ if (!empty($rss->image['rss_url'])) $this->_imageurl = (string) $rss->image['rss_url'];
+ if (!empty($rss->channel['rss_link'])) $this->_link = (string) $rss->channel['rss_link'];
+ if (!empty($rss->channel['rss_title'])) $this->_title = (string) $rss->channel['rss_title'];
+ if (!empty($rss->channel['rss_description'])) $this->_description = (string) $rss->channel['rss_description'];
+ }
+
+ if (! empty($conf->global->EXTERNALRSS_USE_SIMPLEXML)) $items=$rss->channel->item; // With simplexml
+ else $items=$rss->items; // With xmlparse
+ //var_dump($items);exit;
}
else if ($rss->_format == 'atom')
{
- if (!empty($rss->generator)) $this->_generator = (string) $rss->generator;
- if (!empty($rss->lastbuilddate)) $this->_lastbuilddate = (string) $rss->modified;
- if (!empty($rss->link->href)) $this->_link = (string) $rss->link->href;
- if (!empty($rss->title)) $this->_title = (string) $rss->title;
- if (!empty($rss->description)) $this->_description = (string) $rss->description;
- $tmprss=xml2php($rss);
- $items=$tmprss['entry'];
+ if (! empty($conf->global->EXTERNALRSS_USE_SIMPLEXML))
+ {
+ if (!empty($rss->generator)) $this->_generator = (string) $rss->generator;
+ if (!empty($rss->lastbuilddate)) $this->_lastbuilddate = (string) $rss->modified;
+ if (!empty($rss->link->href)) $this->_link = (string) $rss->link->href;
+ if (!empty($rss->title)) $this->_title = (string) $rss->title;
+ if (!empty($rss->description)) $this->_description = (string) $rss->description;
+ }
+ else
+ {
+ if (!empty($rss->channel['rss_language'])) $this->_language = (string) $rss->channel['rss_language'];
+ if (!empty($rss->channel['rss_generator'])) $this->_generator = (string) $rss->channel['rss_generator'];
+ if (!empty($rss->channel['rss_copyright'])) $this->_copyright = (string) $rss->channel['rss_copyright'];
+ if (!empty($rss->channel['rss_lastbuilddate'])) $this->_lastbuilddate = (string) $rss->channel['rss_lastbuilddate'];
+ if (!empty($rss->image['rss_url'])) $this->_imageurl = (string) $rss->image['rss_url'];
+ if (!empty($rss->channel['rss_link'])) $this->_link = (string) $rss->channel['rss_link'];
+ if (!empty($rss->channel['rss_title'])) $this->_title = (string) $rss->channel['rss_title'];
+ if (!empty($rss->channel['rss_description'])) $this->_description = (string) $rss->channel['rss_description'];
+ }
+ if (! empty($conf->global->EXTERNALRSS_USE_SIMPLEXML)) { $tmprss=xml2php($rss); $items=$tmprss['entry'];} // With simplexml
+ else $items=$rss->items; // With xmlparse
+ //var_dump($items);exit;
}
$i = 0;
// Loop on each record
foreach($items as $item)
{
+ //var_dump($item);exit;
if ($rss->_format == 'rss')
{
- $itemLink = (string) $item->link;
- $itemTitle = (string) $item->title;
- $itemDescription = (string) $item->description;
- $itemPubDate = (string) $item->pubDate;
- $itemId = '';
+ if (! empty($conf->global->EXTERNALRSS_USE_SIMPLEXML))
+ {
+ $itemLink = (string) $item->link;
+ $itemTitle = (string) $item->title;
+ $itemDescription = (string) $item->description;
+ $itemPubDate = (string) $item->pubDate;
+ $itemId = '';
+ $itemAuthor = '';
+ }
+ else
+ {
+ $itemLink = (string) $item['rss_link'];
+ $itemTitle = (string) $item['rss_title'];
+ $itemDescription = (string) $item['rss_description'];
+ $itemPubDate = (string) $item['rss_pubdate'];
+ $itemId = (string) $item['rss_guid'];
+ $itemAuthor = (string) $item['rss_author'];
+ }
// Loop on each category
$itemCategory=array();
- foreach ($item->category as $cat)
+ if (is_array($item->category))
{
- $itemCategory[] = (string) $cat;
+ foreach ($item->category as $cat)
+ {
+ $itemCategory[] = (string) $cat;
+ }
}
}
else if ($rss->_format == 'atom')
{
- $itemLink = (string) $item['link']['href'];
- $itemTitle = (string) $item['title'];
- $itemDescription = (string) $item['summary'];
- $itemPubDate = (string) $item['created'];
- $itemId = (string) $item['id'];
+ if (! empty($conf->global->EXTERNALRSS_USE_SIMPLEXML))
+ {
+ $itemLink = (string) $item['link']['href'];
+ $itemTitle = (string) $item['title'];
+ $itemDescription = (string) $item['summary'];
+ $itemPubDate = (string) $item['created'];
+ $itemId = (string) $item['id'];
+ $itemAuthor = '';
+ }
+ else
+ {
+ $itemLink = (string) $item['rss_link'];
+ $itemTitle = (string) $item['rss_title'];
+ $itemDescription = (string) $item['rss_description'];
+ $itemPubDate = (string) $item['rss_pubdate'];
+ $itemId = (string) $item['rss_guid'];
+ $itemAuthor = (string) $item['rss_author'];
+ }
}
+ else print 'ErrorBadFeedFormat';
// Add record to result array
$this->_rssarray[$i] = array(
@@ -211,7 +302,8 @@ class RssParser
'description'=>$itemDescription,
'pubDate'=>$itemPubDate,
'category'=>$itemCategory,
- 'id'=>$itemId);
+ 'id'=>$itemId,
+ 'author'=>$itemAuthor);
$i++;
@@ -227,6 +319,270 @@ class RssParser
}
}
+
+
+ /**
+ * Triggered when opened tag is found
+ *
+ * @param $p
+ * @param $element Tag
+ * @param $attrs Attributes of tags
+ */
+ function feed_start_element($p, $element, &$attrs)
+ {
+ $el = $element = strtolower($element);
+ $attrs = array_change_key_case($attrs, CASE_LOWER);
+
+ // check for a namespace, and split if found
+ $ns = false;
+ if ( strpos( $element, ':' ) ) {
+ list($ns, $el) = explode( ':', $element, 2);
+ }
+ if ( $ns and $ns != 'rdf' ) {
+ $this->current_namespace = $ns;
+ }
+
+ # if feed type isn't set, then this is first element of feed
+ # identify feed from root element
+ #
+ if (!isset($this->_format) ) {
+ if ( $el == 'rdf' ) {
+ $this->_format = 'rss';
+ $this->feed_version = '1.0';
+ }
+ elseif ( $el == 'rss' ) {
+ $this->_format = 'rss';
+ $this->feed_version = $attrs['version'];
+ }
+ elseif ( $el == 'feed' ) {
+ $this->_format = 'atom';
+ $this->feed_version = $attrs['version'];
+ $this->inchannel = true;
+ }
+ return;
+ }
+
+ if ( $el == 'channel' )
+ {
+ $this->inchannel = true;
+ }
+ elseif ($el == 'item' or $el == 'entry' )
+ {
+ $this->initem = true;
+ if ( isset($attrs['rdf:about']) ) {
+ $this->current_item['about'] = $attrs['rdf:about'];
+ }
+ }
+
+ // if we're in the default namespace of an RSS feed,
+ // record textinput or image fields
+ elseif (
+ $this->_format == 'rss' and
+ $this->current_namespace == '' and
+ $el == 'textinput' )
+ {
+ $this->intextinput = true;
+ }
+
+ elseif (
+ $this->_format == 'rss' and
+ $this->current_namespace == '' and
+ $el == 'image' )
+ {
+ $this->inimage = true;
+ }
+
+ # handle atom content constructs
+ elseif ( $this->_format == 'atom' and in_array($el, $this->_CONTENT_CONSTRUCTS) )
+ {
+ // avoid clashing w/ RSS mod_content
+ if ($el == 'content' ) {
+ $el = 'atom_content';
+ }
+
+ $this->incontent = $el;
+
+
+ }
+
+ // if inside an Atom content construct (e.g. content or summary) field treat tags as text
+ elseif ($this->_format == 'atom' and $this->incontent )
+ {
+ // if tags are inlined, then flatten
+ $attrs_str = join(' ',
+ array_map('map_attrs',
+ array_keys($attrs),
+ array_values($attrs) ) );
+
+ $this->append_content( "<$element $attrs_str>" );
+
+ array_unshift( $this->stack, $el );
+ }
+
+ // Atom support many links per containging element.
+ // Magpie treats link elements of type rel='alternate'
+ // as being equivalent to RSS's simple link element.
+ //
+ elseif ($this->_format == 'atom' and $el == 'link' )
+ {
+ if ( isset($attrs['rel']) and $attrs['rel'] == 'alternate' )
+ {
+ $link_el = 'link';
+ }
+ else {
+ $link_el = 'link_' . $attrs['rel'];
+ }
+
+ $this->append($link_el, $attrs['href']);
+ }
+ // set stack[0] to current element
+ else {
+ array_unshift($this->stack, $el);
+ }
+ }
+
+
+ /**
+ * Triggered when CDATA is found
+ *
+ * @param $p
+ * @param $element Tag
+ * @param $attrs Attributes of tags
+ */
+ function feed_cdata ($p, $text) {
+ if ($this->_format == 'atom' and $this->incontent)
+ {
+ $this->append_content( $text );
+ }
+ else {
+ $current_el = join('_', array_reverse($this->stack));
+ $this->append($current_el, $text);
+ }
+ }
+
+ /**
+ * Triggered when closed tag is found
+ *
+ * @param $p
+ * @param $element Tag
+ */
+ function feed_end_element ($p, $el) {
+ $el = strtolower($el);
+
+ if ( $el == 'item' or $el == 'entry' )
+ {
+ $this->items[] = $this->current_item;
+ $this->current_item = array();
+ $this->initem = false;
+ }
+ elseif ($this->_format == 'rss' and $this->current_namespace == '' and $el == 'textinput' )
+ {
+ $this->intextinput = false;
+ }
+ elseif ($this->_format == 'rss' and $this->current_namespace == '' and $el == 'image' )
+ {
+ $this->inimage = false;
+ }
+ elseif ($this->_format == 'atom' and in_array($el, $this->_CONTENT_CONSTRUCTS) )
+ {
+ $this->incontent = false;
+ }
+ elseif ($el == 'channel' or $el == 'feed' )
+ {
+ $this->inchannel = false;
+ }
+ elseif ($this->_format == 'atom' and $this->incontent ) {
+ // balance tags properly
+ // note: i don't think this is actually neccessary
+ if ( $this->stack[0] == $el )
+ {
+ $this->append_content("</$el>");
+ }
+ else {
+ $this->append_content("<$el />");
+ }
+
+ array_shift( $this->stack );
+ }
+ else {
+ array_shift( $this->stack );
+ }
+
+ $this->current_namespace = false;
+ }
+
+
+ /**
+ * To concat 2 string with no warning if an operand is not defined
+ *
+ * @param $str1
+ * @param $str2
+ */
+ function concat (&$str1, $str2="") {
+ if (!isset($str1) ) {
+ $str1="";
+ }
+ $str1 .= $str2;
+ }
+
+ /**
+ */
+ function append_content($text) {
+ if ( $this->initem ) {
+ $this->concat( $this->current_item[ $this->incontent ], $text );
+ }
+ elseif ( $this->inchannel ) {
+ $this->concat( $this->channel[ $this->incontent ], $text );
+ }
+ }
+
+ /**
+ * smart append - field and namespace aware
+ */
+ function append($el, $text) {
+ if (!$el) {
+ return;
+ }
+ if ( $this->current_namespace )
+ {
+ if ( $this->initem ) {
+ $this->concat(
+ $this->current_item[ $this->current_namespace ][ $el ], $text);
+ }
+ elseif ($this->inchannel) {
+ $this->concat(
+ $this->channel[ $this->current_namespace][ $el ], $text );
+ }
+ elseif ($this->intextinput) {
+ $this->concat(
+ $this->textinput[ $this->current_namespace][ $el ], $text );
+ }
+ elseif ($this->inimage) {
+ $this->concat(
+ $this->image[ $this->current_namespace ][ $el ], $text );
+ }
+ }
+ else {
+ if ( $this->initem ) {
+ $this->concat(
+ $this->current_item[ $el ], $text);
+ }
+ elseif ($this->intextinput) {
+ $this->concat(
+ $this->textinput[ $el ], $text );
+ }
+ elseif ($this->inimage) {
+ $this->concat(
+ $this->image[ $el ], $text );
+ }
+ elseif ($this->inchannel) {
+ $this->concat(
+ $this->channel[ $el ], $text );
+ }
+
+ }
+ }
+
}
--
GitLab