recursive XML parsing

Tomaszewski · Mar 11, 2009

sooo... i have a question.. i'm tasked with building this app and need to take the below XML and parse it into a multi-dimensional associative array.

Code:

trace(_xmlDataArray["AL"][0])[INDENT]// 87369[/INDENT]
trace(_xmlDataArray["AL"][1])[INDENT]// 12/31/2008[/INDENT]

// etc....

Code:

	<state>
		[INDENT]<name>AL</name>
		<income>87369</income>
		<doctorStats>
			[INDENT]<period id="1">
				[INDENT]<date>12/31/2008</date>
				<participants>485</participants>[/INDENT]
			</period>
			<period id="2">
				<date>12/31/2003</date>
				<participants>294</participants>
			</period>
			<period id="3">
				<date>12/31/1998</date>
				<participants>211</participants>
			</period>[/INDENT]
		</doctorStats>
		<prescription>
			[INDENT]<type>Non-controlled drugs from approved formulary</type>[/INDENT]
		</prescription>
		<programs>
			[INDENT]<prog id="1">
				[INDENT]<name>University of South Alabama in Mobile</name>
				<link>http://southalbama.edu/alliedhealth/pa/</link>[/INDENT]
			</prog>
			<prog id="2">
				[INDENT]<name>University of Alabama at Birmingham Surgical</name>
				<link>http://main.uab.edu/shrp/default.aspx?pid=77392</link>[/INDENT]
			</prog>
			<prog id="3">
				[INDENT]<name></name>
				<link></link>[/INDENT]
			</prog>[/INDENT]
		</programs>
		<organizations>
			[INDENT]<org>
				[INDENT]<type>State Organization</type>
				<name>Alabama Society of Physician Assitants</name>
				<link>http://www.aapacoms.org/aspa</link>[/INDENT]
			</org>
			<org >
				[INDENT]<type>Federal Organization</type>
				<name>Federal Society of Physician Assitants</name>
				<link>http://www.thefeds.gov</link>[/INDENT]
			</org>
			<org >
				[INDENT]<type></type>
				<name></name>
				<link></link>[/NDENT]
			</org>[/INDENT]
		</organizations>
		<fact>Did you know? About 46.3% of Alabama PA's practice in a surgical specialty</fact>
		<source>Source: American Academy of Physicians Assistants</source>[/INDENT]
	</state>

The question I have is how in the heck do I get is so that each text value of each node gets shoved into the appropriate array index?

So each state will have a <state>PA</state> XML structure with its own data just like AL above in the XML. So for (i.e., PA), i need to put the relevant data like income into index [0] of the _xmlDataArray["PA"] array.

Here is what I have so far:

Code:




		private function buildArray(theXML:XML):void
		{
			_xmlDataArray = new Array();

			for (var i:uint; i < theXML.state.length(); i++)
			{
				//trace(theXML.state.name[i]);
				_xmlDataArray[theXML.state[i].name.children()] = [theXML.state[i].income.toString()];
				recurssion(theXML.state[i].name,theXML);
			}
		}
		
		private function recurssion(usaState:String, theXML:XML):void
		{
			if(theXML.children())
			{
				//for each
			}
			trace(_xmlDataArray[usaState]);
		}

I've attached the XML file beacuse the rendering on this post isn't the best!

smoseley · Mar 11, 2009

Try something like this.. Not fully tested, but you see where it's going:

Code:

var arr:Array = convertXmlToArray(xml);

function convertXmlToArray(xml:XML):Array {
	var xmlArray:Array = new Array();
	
	// Populate children
	for (var i:uint = 0; i < xml.length(); i++) {
		var xmlNode:XML = xml.children()[i];
		if (xmlNode.hasComplexContent()) {
			// Recursion
			xmlArray[xmlNode.name().localName] = convertXmlToArray(xmlNode);
		}  else {
			// Text node
			xmlArray[xmlNode.name().localName] = xmlNode.toString();
		}
	}
	
	// Return appropriate element
	if (xml.parent() == null) {
		// Top level element
		var mainArray:Array = new Array();
		mainArray[xml.name().localName] = xmlArray;
		return mainArray;
	} else {
		// Other element
		return xmlArray;
	}
}

Tomaszewski · Mar 12, 2009

smoseley said:
Code:
			xmlArray[xmlNode.name().localName] = convertXmlToArray(xmlNode);
Ok, but this traces state where it needs to be the name of the state (i.e., "AL"). So i converted it to
Code:
_xmlDataArray[xmlNode.child("name")] = "blah blah";
... this works...
Code:
trace(_xmlDataArray["AL"]);[INDENT]//blah blah[/INDENT]
still, i can't figure out a way how to populate the array with the rest of the data

smoseley · Mar 12, 2009

Ok, I see how you want it now.

On a side note, I forgot that ActionScript Arrays don't allow associative keys. You'll have to use the Object Class.

This should do what you want (you can still reference the object values as though they are part of an array):

Code:

var xml:XML = 
	<medicalInfo>
		<state>
			<name>AL</name>
			<source>Source: American Academy of Physicians Assistants</source>
		</state>
		<state>
			<name>GA</name>
			<source>Source: Other Stuff</source>
			<complex>
				<val1>test</val1>
				<val2>another test</val2>
			</complex>
		</state>
	</medicalInfo>;
	

var medInfo:Object = getStateArray(xml);

trace(medInfo);
trace(medInfo["AL"]["source"]);
trace(medInfo.AL.source);
trace(medInfo.GA.complex.val1);
trace(medInfo.GA.complex.val2);

function getStateArray(xml:XML):Object {
	var xmlAssoc:Object = new Object();
	for (var i:uint = 0; i < xml.children().length(); i++) {
		var xmlNode:XML = xml.children()[i];
		xmlAssoc[xmlNode.child("name")] = convertXmlToAssoc(xmlNode);
	}
	return xmlAssoc;
}


function convertXmlToAssoc(xml:XML):Object {
	var xmlAssoc:Object = new Object();
	
	// Populate children
	for (var i:uint = 0; i < xml.children().length(); i++) {
		var xmlNode:XML = xml.children()[i];
		if (xmlNode.hasComplexContent()) {
			// Recursion
			xmlAssoc[xmlNode.name().localName] = convertXmlToAssoc(xmlNode);
		}  else {
			// Text node
			xmlAssoc[xmlNode.name().localName] = xmlNode.toString();
		}
	}
	return xmlAssoc;
}

Shadowfiend · Mar 12, 2009

To be clear, Array does support associative keys, but if you're going to use them there's no reason to use Array. That is to say, since an Array is an Object, it will also take associative keys, but then you've needlessly added the functionality of an Array to something that need only be an Object.

Tomaszewski · Mar 12, 2009

I feel stupid!

ME:

Your a genious. Coulple of question though (surprised?? LOL):

I rewrote your code to learn what its doing. I also integrated as part of my class.

After reading Adobe's description of xml.hasComplexContent() property, I must say that I'm lost as to why you have used it. The only thing I can think of is that if its complex content, than it gets treated like an XMLList and not XML. But then again... if so... than I didn't see you calling anything from the XMLlist.

The same goes for localName. From what I can tell, this is outputting the name contained by the XML (i.e, <name>AL</name> // AL)...

In my XML tree, there are 3 <period> children under <doctorStats> as an example. Currently, the recursive code is pulling in only 1 of those. So if I wanted to pull into the array, as many as there are, would that mean I would have to do a for loop if the recursion sees children()?

Shadowfiend · Mar 12, 2009

hasComplexContent basically returns true if the element has children.

localName is used to retrieve the name of the XML element without a namespace (in your case, there are no XML namespaces anyway, so it would be the same as the name).

Your problem with the three period children is that you would have to check for multiple children of the *same name*, and, instead of assigning them to the same spot in the object, you would need an array to store them. So the way it's currently doing it now, it tries to put the first period object into arr["period"], and then when it sees the next one, it puts it in the same place, replacing the other one. One good way to do it would be this:
Code:
                var content:Object;

                if (xmlNode.hasComplexContent())
                {
			// Recursion
			content = convertXmlToAssoc(xmlNode);
		}
                else
                {
			// Text node
			content = xmlNode.toString();
		}

               var name:String = xmlNode.name().localName;
               var currentElement:Object = xmlAssoc[name];
               if (! currentElement) // if there is nothing there yet
                   xmlAssoc[name] = content;
               else if (currentElement is Array)
                   currentElement.push(content);
               else
                   xmlAssoc[name] = new Array(currentElement, content);
This basically builds the data for the current child, then checks if there is an existing entry in the list for it. If there is, then it makes sure that entry is an array and adds the content of this child to that array.

Tomaszewski · Mar 12, 2009

can you elaborate on this a little more?

so, would I put this after the first recursion?

example:

Code:

		private function buildArray(xml:XML):Object
		{
			var xmlAssoc:Object = new Object();
			for (var i:uint = 0; i < xml.children().length(); i++)
			{
				var xmlNode:XML = xml.children()[i];
				xmlAssoc[xmlNode.child("name")] = recursion(xmlNode);
			}
			
			return xmlAssoc;
		}
		
		private function recursion(xml:XML):Object
		{
			var xmlAssoc:Object = new Object();
			for (var i:uint = 0; i < xml.children().length(); i++)
			{
				var xmlNode:XML = xml.children()[i];
				if (xmlNode.hasComplexContent())
				{
					//Recurssion
					xmlAssoc[xmlNode.name().localName] = recursion(xmlNode);\
[INDENT][COLOR="Red"][B]/* ******************* HERE ******************************** /[/B][/COLOR][/INDENT]
				}
				else
				{
					xmlAssoc[xmlNode.name().localName] = xmlNode.toString();
				}
			}

Shadowfiend · Mar 12, 2009

You would replace the body of the for loop, except for the xmlNode declaration, with it.

As for namespaces. Basically, an XML namespace is a way to include tags from multiple `categories' in a single document. For example, if I wanted to use regular XHTML with FBML (the Facebook Markup Language) in a browser that understood both, I would have an xmlns for the HTML part and one for the fbml part. This means that I could use tags from both categories in my document.

Namespaces are usually of the syntax [minicode]namespace:tagname[/minicode]. In your case, as mentioned, you don't actually have a namespace, so name and localName would likely be the same. Namespaces are used a lot in things like the XMPP protocol and Flex MXML.

smoseley · Mar 12, 2009

Expanding on Shadowfiend's above explanation, here's an example of XML namespaces in use:

Code:

<?xml version="1.0"?>
<[B]myXmlDoc [/B]xmlns="http://mynamespace.mydomain.com/subnamespace" 
        xmlns:steve="http://stevesnamespace.stevesdomain.com/subnamespace">

    <[B]myElement[/B]>
        <[B]name[/B]>This is an element in my namespace</[B]name[/B]>
        <[B]description[/B]>Notice the generic 'name' and 'description' children have no namespace
            defined.  This is because they are part of the default namespace.</[B]description[/B]>
    </[B]myElement[/B]>

    <[B]myOtherElement[/B]>This is another element in my namespace</[B]myOtherElement[/B]>

    <[B]steve:stevesElement[/B]>
        <[B]steve:name[/B]>This is one of Steve's elements</[B]steve:name[/B]>
        <[B]steve:description[/B]>Notice all of the defining parts of the 'steve' XML element are tagged
            with the 'steve' namespace.  This allows us to selectively parse overlapping elements
            of different schemas using the namespace prefix to separate them.</[B]steve:description[/B]>
    </[B]steve:stevesElement[/B]>

</[B]myXmlDoc[/B]>

Neat, no?

Tomaszewski · Mar 13, 2009

smoseley said:
Expanding on Shadowfiend's above explanation, here's an example of XML namespaces in use:
Code:
<?xml version="1.0"?>
<[B]myXmlDoc [/B]xmlns="http://mynamespace.mydomain.com/subnamespace" 
        xmlns:steve="http://stevesnamespace.stevesdomain.com/subnamespace">

    <[B]myElement[/B]>
        <[B]name[/B]>This is an element in my namespace</[B]name[/B]>
        <[B]description[/B]>Notice the generic 'name' and 'description' children have no namespace
            defined.  This is because they are part of the default namespace.</[B]description[/B]>
    </[B]myElement[/B]>

    <[B]myOtherElement[/B]>This is another element in my namespace</[B]myOtherElement[/B]>

    <[B]steve:stevesElement[/B]>
        <[B]steve:name[/B]>This is one of Steve's elements</[B]steve:name[/B]>
        <[B]steve:description[/B]>Notice all of the defining parts of the 'steve' XML element are tagged
            with the 'steve' namespace.  This allows us to selectively parse overlapping elements
            of different schemas using the namespace prefix to separate them.</[B]steve:description[/B]>
    </[B]steve:stevesElement[/B]>

</[B]myXmlDoc[/B]>
Neat, no?
Indeed... and now I understand. As for the code, I got it working thanks to the both of you. Here is the code:

http://friendpaste.com/5h5B416xemqg2zO6UQxVLV

I do have one additional question to this. We are using Objects here to keep the data in a tree like structure. Lets say however, that I wanted to slap all the data and flatten it into an array so that I can call the data like this: _xmlDataArray["PA"][4]... I would have to than push() the data into the Array no?

Tomaszewski · Mar 17, 2009

so is there a way to add the data not into an Object but into a flat array instead? Like [0], [1], [2]... etc... i've been trying to get it to work, but am not having much like. I'm convinced that the line:
Code:
 xmlAssoc[name] = new Array(currentElement, tempObj); 
is what is creating the tree like object, so that instead of having just the values in the array, i have

Object
|
- Project

|
- "value"
- "value"

i'd rather have

Array
|
[0] - "value"
[1] - "value"
[2] - "value"

know what i mean?

Shadowfiend · Mar 17, 2009

Not 100% clear on it, no, but I think that a variant of the array solution I showed above should work.

Tomaszewski · Mar 18, 2009

Ok, thusfar I've gotten this far: http://friendpaste.com/339gSTqjhOV4s1egg4oYhu

... but am seriously stuck... the array _xmlDataArray spits out:

_xmlDataArray["AL"]

[0] > 1234
[1] > 23525
[3] - [0] > sdfsdf
- - - -[1] > 234234
- - - -[2] > 234234
[4] > 234sdfsdf
[5] > 23hsfsf
- - - -[0] sdf222
- - - -[1]234234
[6] sdfsdf

... and I can't flatten the rest of it... like [3][0], [3][1], [3][2], etc. :ichatcry:

... and i think its because line 13 gets a return of an array. So i tried making it get a String, but then the state abbreviation gets messed up.

Tomaszewski · Mar 18, 2009

Let me ask another way... if I wanted to spit out the contents of _xmlDataArray... and if the array has multiple items inside multiple items... (i.e., [0][1], [0][2], [0][3]...etc)... how can I get it to display all those items on individual lines on the screen?

Currently I'm using:
Code:
			for each(var key:String in _xmlDataArray["AL"])
			{
				trace(key + "\n");
			}
but it just gives me the first indexes items, but not any other index inside... know what i mean?

for example
Code:
AL

87369

12/31/2008,485,12/31/2003,294,12/31/1998,211

Non-controlled drugs from approved formulary

University of South Alabama in Mobile,http://southalbama.edu/alliedhealth/pa/,University of Alabama at Birmingham Surgical,http://main.uab.edu/shrp/default.aspx?pid=77392

State Organization,Alabama Society of Physician Assitants,http://www.aapacoms.org/aspa,Federal Organization,Federal Society of Physician Assitants,http://www.thefeds.gov

Did you know? About 46.3% of Alabama PA's practice in a surgical specialty

Source: American Academy of Physicians Assistants
but if you notice the dates
Code:
12/31/2008,485,12/31/2003,294,12/31/1998,211
they are not getting displayed on a new line... grr...

Log in or Sign up

recursive XML parsing

Tomaszewski New Member

Attached Files:

mapInfo.xml

WDF Go Premium, Drop Ads

smoseley Administrator

Tomaszewski New Member

smoseley Administrator

Shadowfiend Code beautifully and honorably

Tomaszewski New Member

Shadowfiend Code beautifully and honorably

Tomaszewski New Member

Shadowfiend Code beautifully and honorably

smoseley Administrator

Tomaszewski New Member

Tomaszewski New Member

Shadowfiend Code beautifully and honorably

Tomaszewski New Member

Tomaszewski New Member

Share This Page