XML Parser class SCOPING issue

[frizzo]

Hello,

I am a little stumped here! I wrote a class to parse some XML and populate a news article object. Here's what's happening:

-- When it encounters a <newsarticle> node, it creates a new NewsStory object, which is scoped to the entire parser class.

-- For every subsequent node, it populates the news article from the XML into the appropriate members (author, story, etc). I've verified that this is working, but only at the specific instance of the function.

-- when it encounters a </newsarticle> tag, it calls the insert() function for the news story to get put into the database.

THE PROBLEM: by this time, when it inserts, none of the data that has been populated into the object is remaining.

This seems like a simple coding issue, I've done this sort of thing before, but not with XML. Is there something I need to do to retain the scope of the news story?? Any ideas?

PLEASE HELP, THANKS A LOT!
- Frizzo

CODE:

The XML:

<mergemedia_news>
<newsstory>
<title><![CDATA[79 Percent Of Americans Missing The Point Entirely]]></title>
<author><![CDATA[Joe Smo]]></author>
<author_desc><![CDATA[Blaha]]></author_desc>
<headline><![CDATA[Headline here ]]></headline>
<sub_headline><![CDATA[subheadline here ]]></sub_headline>
<teaser_txt><![CDATA[Teaser Text here]]></teaser_txt>
<story_txt><![CDATA[ Story text here]]></story_txt>
<ext_txt><![CDATA[ extended text here]]></ext_txt>
</newsstory>
</mergemedia_news>

The XML Parser Class!

class news_xml_parser{
	var $xml_parser;
	var $xml_current; // tree-path to the current element
	var $element; // the current element name
	var $element_data;
	var $story; // the story object

function news_xml_parser(){
	$this->xml_parser = xml_parser_create("UTF-8");
	xml_parser_set_option($this->xml_parser,  XML_OPTION_CASE_FOLDING, false);
	xml_set_element_handler($this->xml_parser, "tag_open", "tag_close"); 
	xml_set_character_data_handler($this->xml_parser, "cdata"); 
}

function xml2db($xml){
	global $ret_msg;
	xml_set_object($this->xml_parser, &$this); 
	if (!xml_parse($this->xml_parser, $xml)) {
		$ret_msg->add_err_msg("Could not parse XML!");
	} else {
		$ret_msg->add_ret_msg("Parsed OK");
	}
	xml_parser_free($this->xml_parser);
}

function tag_open($parser, $tag, $attributes){
	global $ret_msg;
	$this->element = $tag;
	$this->xml_current.="/~".$tag;
	switch ($tag){
		case "newsstory":
			// instantiate new story to $this->story
			$this->story = new NewsStory();
			break;
		 default:
			break;
	}
}

function tag_close($parser, $tag){
	global $ret_msg;
	$xml_caret_pos=strrpos($this->xml_current,"/~");
	$this->xml_current=substr($this->xml_current,0,$xml_caret_pos);
	switch ($tag){
		case "newsstory":
			// insert story into the database
			$this->story->insert();
			break;
	}
}

function cdata($parser, $cdata){
	global $ret_msg;
	$x = $this->element;
	$s = $this->story;
	switch ($x){
		case "title":
			$s->set_title($cdata);
			break;
		case "headline":
			$s->set_headline($cdata);
			break;
		case "sub_headline":
			$s->set_sub_headline($cdata);
			break;
		case "author":
			$s->set_author($cdata);
			break;
		case "author_desc":
			$s->set_author_desc($cdata);
			break;
		case "teaser_txt":
			$s->set_teaser_txt($cdata);
			break;
		case "story_txt":
			$s->set_story_txt($cdata);
			break;
		case "ext_txt":
			$s->set_ext_txt($cdata);
			break;
	}
}

}

[Jack_Slocum]

Changing the line

$s = $this->story;

to

$s = &$this->story;

in your cdata() function should fix your problem.

PHP will automatically make a copy of your object when it's on the right of an "=" operator. So the data you are assigning is never accually going to the object you have stored as a member variable, it's going to the copy (which is destroyed when it goes out of scope). Putting the "&" in front makes $s just an alias and PHP won't make a copy.

[frizzo]

Thanks for the suggestion but it didn't work. I tried that before, and it doesn't make any change. Still losing scope!

I also tried changing $this->story = new NewsStory() to $this->story =& new NewsStory(); in the tag_close() function and still no good.

This is really weird!

Any other ideas?

btw the code is being invoked through the xml2db() function.

-Frizzo

[frizzo]

What was happening is the cdata handler function was for some reason gettting called TWICE, the second time it was receiving an empty string and overwriting the one that was supposed to be there.

BUT I have no idea where or how it's getting called twice.

I made a temporary quick fix by checking the value of the string, if it's "" then the cdata function doesn't set any values. I'd like to know why it's getting called twice, and how to get around that!!

Thanks!
Frizzo

<?php

class news_xml_parser{
	var $xml_parser;
	var $xml_current; // tree-path to the current element
	var $element; // the current element name
	var $element_data;
	var $story ; // the story object

function news_xml_parser(){
	$this->xml_parser = xml_parser_create("UTF-8");
	xml_parser_set_option($this->xml_parser,  XML_OPTION_CASE_FOLDING, false);
	xml_set_element_handler($this->xml_parser, "tag_open", "tag_close"); 
	xml_set_character_data_handler($this->xml_parser, "cdata"); 
	$this->story = new NewsStory();
}

function xml2db($xml){
	global $ret_msg;
	xml_set_object($this->xml_parser, &$this); 
	if (!xml_parse($this->xml_parser, $xml)) {
		$ret_msg->add_err_msg("xml2db:  Could not parse XML!");
		return false;
	} else {
		return true;
	}
	xml_parser_free($this->xml_parser);
}

function tag_open($parser, $tag, $attributes){
	$this->element = $tag;
	$this->xml_current.="/~".$tag;
	switch ($tag){
		case "newsstory":
			// instantiate new story to $this->story
			$this->story = new NewsStory();
			break;
		 default:
			break;
	}
}

function tag_close($parser, $tag){
	$xml_caret_pos=strrpos($this->xml_current,"/~");
	$this->xml_current=substr($this->xml_current,0,$xml_caret_pos);
	switch ($tag){
		case "newsstory":
			// insert story into the database
			$this->story->insert();
			break;
	}
}


function cdata($parser, $cdata){

	//HERE IS THE HACK:
	if (trim($cdata) == "" ){
		return false;
	}

	$d = trim($cdata);
	switch ($this->element){
		case "title":	
			$this->story->set_title($d);
			break;
		case "headline":
			$this->story->set_headline($d);
			break;
		case "sub_headline":
			$this->story->set_sub_headline($d);
			break;
		case "author":
			$this->story->set_author($d);
			break;
		case "author_desc":
			$this->story->set_author_desc($d);
			break;
		case "teaser_txt":
			$this->story->set_teaser_txt($d);
			break;
		case "story_txt":
			$this->story->set_story_txt($d);
			break;
		case "ext_txt":
			$this->story->set_ext_txt($d);
			break;
	}
	return true;
}

}

?>

[ahundiak]

You probably have some extra whitespace in your xml file.
Whitespace will trigger a call to cdata.
So just keep filtering out the empty strings.
http://www.php.net/manual/en/ref.xml.php

[128kb_com]

What was happening is the cdata handler function was for some reason gettting called TWICE, the second time it was receiving an empty string and overwriting the one that was supposed to be there.

Do you mean twice on the same node/cdata? Because your cdata handler should get called once for each node. Also, in your constructor you instantiate $this->story to a new NewsStory() object but your overwriting it with the same line in tag_open(). Not sure if thats your only problem though, it seems like you only have a case statement for for "newsstory" in tag_open() but not your other nodes.

What does your NewsStory class look like?

[Jack_Slocum]

cdata is getting called twice because your xml document has two cdata sections for each tag. We will examine this piece of your xml file and I will explain:

<title><![CDATA[79 Percent Of Americans Missing The Point Entirely]]></title>

Anything between <title></title> is one cdata section. For example, your xml should look like this:

<title>79 Percent Of Americans Missing The Point Entirely</title>

The text between the two tags would be your cdata section.

Since you have ANOTHER cdata section in between the two tags (from wrapping the text in <![CDATA[]]>), you have a second cdata node.

You don't need to put the text in between two tags in it's own cdata section, it already is a cdata section. Just xml encode it to avoid any parser errors.