Parsing XML with C# .NET2

[AtW]

particularly functional programming

Yeah I loved that bit most of all - "functional" my



Now seriously - people who made XSL (and probably "functional" languages in general) were either totally stoned (not sure since I never used drugs) or operated in environment that had nothing to do with demands of commercial (or even real world) programming. It should have never left academia - and maybe even single Lab when it was designed, perhaps it should have never left the brains of the person who invented it.

Even databases (and it's natural to deal with sets there) provide options to deal with variables, loops etc - XSL/XPath are pathetic when it comes to sets querying options that are available in SQL - with such stupid set limits they MUST have included ability to do proper programming language stuff to workaround sets limitations.

If I were to list top 10 mistakes of my life then usage of XSL/XPath (that seemed on the surface good so I recommended it) would be in that list, probably closer to the bottom but still that's way too high up.

[NickFitz]

I could have forgiven them that, but calling something that is a constant (can't be changed) a variable really did the trick...

I'd rather sweep streets than do XSL every again...

It's a fundamental principle of declarative programming, and particularly functional programming, that you don't allow mutable values - allowing variables to change their value leads to side-effects, and side-effects cause bugs. The term "variable" in such a language doesn't have the bastardised meaning found in FORTRAN-style languages, up to and including C and C#, of "denoting a chunk of memory" but the mathematical meaning.

In fact, when working with XSLT, it's basically impossible to come up with any circumstance where it would even make sense to think in terms of altering the value of a variable.

[AtW]

Brush up on your set theory, and remember: there is no loop.

I could have forgiven them that, but calling something that is a constant (can't be changed) a variable really did the trick...

I'd rather sweep streets than do XSL every again...

[NickFitz]

Run a board search for Xpath topics - there should be a big one created by me about 2 years ago... yes, it is that

It was about XSL but XPath is closely related -

http://forums.contractoruk.com/gener...ighlight=xpath

I was doing a contract job at the time and it paid well but it would need to be x10 the rate for me to even consider touching it with a bargepole

People who don't understand declarative programming often struggle with XSLT. It's actually a very easy language to use once you understand how to use it.

Forget all that sequential stuff you use in languages like C and Java and assume is essential to programming, because all the languages you've used to date work like that. A procedural approach is not, in fact, a prerequisite for programming, and using a procedural approach to a declarative language is a positive hindrance. Brush up on your set theory, and remember: there is no loop.

Oh, and the problem you cite in the thread you link to: you had a problem because you were doing it wrong.

[AtW]

Forgot to tell us why, like the person afterwards.

Run a board search for Xpath topics - there should be a big one created by me about 2 years ago... yes, it is that

It was about XSL but XPath is closely related -

http://forums.contractoruk.com/gener...ighlight=xpath

I was doing a contract job at the time and it paid well but it would need to be x10 the rate for me to even consider touching it with a bargepole

[jkoder]

XPath ==

Stay away from it...

Forgot to tell us why, like the person afterwards.

[madeoff]

Already Pm'd this to wxman, but here's a possible solution.

Feeling generous at christmas and as this is my first post, I thought I'd post a compilable solutuion.

Basically a set of Info "records" are created one per location, containing the windspeed, min and max temperature for each location.

Each location element, is checked to see if it has a windspeed, mintemperature or maxtemperature sibling and this is used to populate the appropriate property of the info "record".

This can be compiled as a console application, but you will need to change the xml source file from c:\temp\xml.xml to http://www.stormtrack.co.uk/test.xml


Madeoff

=============== Code starts (indentation may be lost)=============
using System;
using System.Collections.Generic;
using System.Text;
using System.Xml;
namespace CUK
{
class Program
{
class Info
{
public string Location = "";
public string Windspeed = "<none>";
public string MinTemp = "<none>";
public string MaxTemp = "<none>";
public Info(string loc)
{
Location = loc;
}
}
static void Main(string[] args)
{
XmlDocument doc=new XmlDocument();
doc.Load(@"C:\temp\xml.xml");

Dictionary<string, Info> locInfoMap = new Dictionary<string, Info>();



XmlNodeList nodeList = doc.SelectNodes("/netweather/*/location");
foreach (XmlElement locElem in nodeList)
{
//Ensure you have a map entry for this location
string loc = locElem.InnerText;
if (locInfoMap.ContainsKey(loc) == false) locInfoMap.Add(loc, new Info(loc));

//get location info entry
Info locInfo = locInfoMap[loc];

//Look for next sibling element named windspeed and set its value if found.
XmlElement windSpeedElem = locElem.SelectSingleNode("following-sibling::windspeed") as XmlElement;
if(windSpeedElem!=null) locInfo.Windspeed=windSpeedElem.InnerText;

//Do as above for min/max temperature

XmlElement minTempElem = locElem.SelectSingleNode("following-sibling::mintemperature") as XmlElement;
if (minTempElem != null) locInfo.MinTemp = minTempElem.InnerText;

XmlElement maxTempElem = locElem.SelectSingleNode("following-sibling::maxtemperature") as XmlElement;
if (maxTempElem != null) locInfo.MaxTemp = maxTempElem.InnerText;

}

int iCount = 0;
foreach (string location in locInfoMap.Keys)
{
++iCount;
Info locInfo = locInfoMap[location];
Console.WriteLine("Location ({0}): {1}; Windspeed: {2}; Min temp {3}; Max temp {4}",
iCount,locInfo.Location, locInfo.Windspeed, locInfo.MinTemp, locInfo.MaxTemp);

}

Console.WriteLine("Total Locations: "+ iCount);
}
}
}

[Jaws]

Generally I prefer the XPath approach as well (although I have previously experienced quite a big performance problem [no doubt due to my query] when dealing with a large result set).

If you are going down the code root in .net, the XmlNode object has a PreviousSibling property which will get you the "location node" for each of the nodes in your GetElementsByTagName loops.

The below works as expected.

            XmlDocument doc = new XmlDocument();
            string xml = "<test><location>Home</location><next>123</next></test>";
            doc.LoadXml(xml);

            XmlNode node = doc.GetElementsByTagName("next")[0];
            
            Console.WriteLine(node.PreviousSibling.Name);
            Console.WriteLine(node.PreviousSibling.InnerText);

[wxman]

Wow many thanks for a full an concise reply - I have no control over the source XML feed so I will do as you suggest and reformat it via an xlst - I *think* then that i can lift out the nodes as required.

One other option is for my to just bung the raw XML into an SQL2005 table and querry out the info that I need for there - but I alway find this approch a pain!

[NickFitz]

XPath FTW

Whatever you do, stay away from things like "InnerXml", which are derived from a broken model of treating XML as if it's text.

Contrary to popular belief, XML isn't a text file format - it's a rather complex data model. It has the specific benefit of being easily serialised as text, but thinking that a text file full of strings that look like the strings in the XML specification is what XML is is the rapid route to epic fail.

Microsoft introduced various techniques to make it easy for people to "do" XML as if XML was text, although this is one of the leakiest abstractions they have yet inflicted on the world (as you are now discovering) in that all it does is allow you to "do" XML in a way that doesn't work unless you already know how to do it, in which case you wouldn't do it that way.

(Sorry if this sounds patronising; I don't mean it that way, but I really should wrap this up and get to bed.)

Right, so your XML at http://www.stormtrack.co.uk/test.xml is well-formed. From here on in I'll be working with W3C standards - MS claim to support these, so this should help, if you abandon MS's proprietary nonsense.

To start with, your XML isn't structured in a useful manner - it's a case if "If you want to go there, don't start from here."

So, for example,

<topwindspeed>
        <location>Wyton Royal Air Force Base</location>
        <windspeed>15</windspeed>
        <location>Sumburgh Cape</location>
        <windspeed>13</windspeed>
</topwindspeed>

is of little value - it should be

<topwindspeed>
        <location>Wyton Royal Air Force Base
            <windspeed>15</windspeed>
        </location>
        <location>Sumburgh Cape
            <windspeed>13</windspeed>
        </location>
</topwindspeed>

although I would be more likely to go with

<topwindspeed>
        <location>
            <name>Wyton Royal Air Force Base</name>
            <windspeed>15</windspeed>
        </location>
        <location>
            <name>Sumburgh Cape</name>
            <windspeed>13</windspeed>
        </location>
</topwindspeed>

as this would allow me, from the top level <netweather> element, to then select the <topwindspeed> and suchlike elements, and easily navigate to the locations, which have names and windspeeds as child elements.

XML follows a hierarchical data model, and you should always seek to present your data in an appropriate hierarchical structure - a datum should not be the sibling of its owner. In this case, the windspeed is a property of the location (as indeed is the location's name) and should be represented as such by making it a child element thereof.

If you have control over the format of the XML, you should change it such that it represents the hierarchical structure of the data.

If you can't control the structure of the source XML then, after telling the provider thereof that they're doing it wrong, you should be able to get by with... actually, scrub that. The best thing to do is to turn the XML into something usable; the following XSLT (1.0, with most belts-and-braces in situ, but watch out for whitespace in the original XML) will do that with your sample file:

<?xml version="1.0" encoding="UTF-8"?>
<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0">

    <xsl:output method="xml" encoding="UTF-8" />
    
    <xsl:template match="/">
        <xsl:apply-templates/>
    </xsl:template>
    
    <xsl:template match="netweather">
        <netweather>
            <xsl:apply-templates/>
        </netweather>
    </xsl:template>

    <xsl:template match="topwindspeed">
        <topwindspeed>
            <xsl:apply-templates select="location"/>
        </topwindspeed>
    </xsl:template>
    
    <xsl:template match="topmintemperature">
        <topmintemperature>
            <xsl:apply-templates select="location"/>
        </topmintemperature>
    </xsl:template>
    
    <xsl:template match="topmaxtemperature">
        <topmaxtemperature>
            <xsl:apply-templates select="location"/>
        </topmaxtemperature>
    </xsl:template>
    
    <xsl:template match="location">
        <location>
            <name><xsl:value-of select="."/></name>
            <value><xsl:value-of select="following-sibling::node()/text()"/></value>
        </location>
    </xsl:template>

</xsl:stylesheet>

giving the following output, if one adds the indent="yes" attribute to the xsl:output element:

<?xml version="1.0" encoding="UTF-8"?>
<netweather>
    
   <topwindspeed>
      <location>
         <name>Wyton Royal Air Force Base</name>
         <value>15</value>
      </location>
      <location>
         <name>Sumburgh Cape</name>
         <value>13</value>
      </location>
      <location>
         <name>Saint Mawgan</name>
         <value>13</value>
      </location>
      <location>
         <name>Holbeach</name>
         <value>12</value>
      </location>
      <location>
         <name>Leeds And Bradford</name>
         <value>12</value>
      </location>
      <location>
         <name>Leuchars</name>
         <value>10</value>
      </location>
      <location>
         <name>Tain Range</name>
         <value>10</value>
      </location>
      <location>
         <name>Cosford Royal Air Force Base</name>
         <value>10</value>
      </location>
      <location>
         <name>Glasgow Airport</name>
         <value>9</value>
      </location>
      <location>
         <name>Lakenheath Royal Air Force Base</name>
         <value>8</value>
      </location>
   </topwindspeed>
    
   <topmintemperature>
      <location>
         <name>Aberdeen / Dyce</name>
         <value>2</value>
      </location>
      <location>
         <name>Disforth</name>
         <value>3</value>
      </location>
      <location>
         <name>Tees-Side</name>
         <value>3</value>
      </location>
      <location>
         <name>Wick</name>
         <value>3</value>
      </location>
      <location>
         <name>Topcliffe Royal Air Force Base</name>
         <value>3</value>
      </location>
      <location>
         <name>Kirkwall Airport</name>
         <value>3</value>
      </location>
      <location>
         <name>Spadeadam</name>
         <value>4</value>
      </location>
      <location>
         <name>Leeming</name>
         <value>4</value>
      </location>
      <location>
         <name>Benbecula</name>
         <value>4</value>
      </location>
      <location>
         <name>Linton-On-Ouse</name>
         <value>5</value>
      </location>
   </topmintemperature>
    
   <topmaxtemperature>
      <location>
         <name>Wyton Royal Air Force Base</name>
         <value>12</value>
      </location>
      <location>
         <name>Cosford Royal Air Force Base</name>
         <value>11</value>
      </location>
      <location>
         <name>Manston, South East</name>
         <value>10</value>
      </location>
      <location>
         <name>Yeovilton</name>
         <value>10</value>
      </location>
      <location>
         <name>Shoreham Airport</name>
         <value>9</value>
      </location>
      <location>
         <name>Middle Wallop</name>
         <value>9</value>
      </location>
      <location>
         <name>Lydd Airport</name>
         <value>9</value>
      </location>
      <location>
         <name>London / Gatwick Airport</name>
         <value>9</value>
      </location>
      <location>
         <name>Eglinton / Londonderr</name>
         <value>9</value>
      </location>
      <location>
         <name>Scilly, Saint MaryS</name>
         <value>9</value>
      </location>
   </topmaxtemperature>

</netweather>

With the document having that structure, it should be a lot easier to find your way around it. If you have existing files but can move to the new XML structure for the future, simply run your existing files through that XSLT and re-save; if you're stuck with the old XML structure, then pre-process the input file with that XSLT. Then select all the location elements into the list you originally wanted, use their child elements to get at the names and values, and get on with the rest of your life.

Oh, and those bits like following-sibling::node()/text()? That's where I began: XPath FTW!

[bogeyman]

XPath ==

Stay away from it...

For once I agree with the squirrily one - it's evil

[AtW]

XPath ==

Stay away from it...

[Cowboy Bob]

What you want is XPath - http://www.w3schools.com/XPath/default.asp

I'm a Java guy, so can't be of more specific help, but I found this article which seems to demonstrate how to do it in .NET - http://www.aspfree.com/c/a/.NET/Work...The-NET-Way/4/

[wxman]

OK this should be easy but I am missing a trick here. I have an XML feed (example file here http://www.stormtrack.co.uk/test.xml )

For each child element, <topwindspeed>, <topmintemperature> and <topmaxtemperature> I need to get the 10 locations and wind/temp values into a list for now

I can get the 30 locations but not by child element.

Can anyone point me in the right direction?? [NB. This is private hobby work]

This is what I have so far...

--------------------------------------------------------
XmlDocument xDoc = new XmlDocument();
xDoc.Load("http://www.stormtrack.co.uk/test.xml");

List<string> locList = new List<string>();
XmlNodeList locNodes = xDoc.GetElementsByTagName("location");
for (int i = 0; i < locNodes.Count; i++)
{
locList.Add(locNodes[i].InnerXml); //gives 30 values
}

List<string> windList = new List<string>();
XmlNodeList windNodes = xDoc.GetElementsByTagName("windspeed");
for (int i = 0; i < windNodes.Count; i++)
{
windList.Add(windNodes[i].InnerXml); //gives 10 values
}


List<string> tHiList = new List<string>();
XmlNodeList tHiNodes = xDoc.GetElementsByTagName("maxtemperature");
for (int i = 0; i < tHiNodes.Count; i++)
{
tHiList.Add(tHiNodes[i].InnerXml); //gives 10 values
}

List<string> tLoList = new List<string>();
XmlNodeList tLoNodes = xDoc.GetElementsByTagName("mintemperature");
for (int i = 0; i < tLoNodes.Count; i++)
{
tLoList.Add(tLoNodes[i].InnerXml); //gives 10 values
}