Forum Moderators: phranque
I have an xml file that is an export from a central server (in other words I can't get the way it is dumped changed :-) ). Unfortunately this is not quite in the form I'd like it. I've heard I can transform one XML file into another using XSL but im not sure how to do it (ideally it needs to be done server side aswell).
The current XML layout is as follows:-
serverdump.xml
<?xml version="1.0" encoding="UTF-8"?>
<all_server_dump>
<clusters>
<cluster name="X4452L" cluster_id="1">
<servers>
<server name="H2212A" server_id="1">
<users>
<user user_id="77397610">
<name>Tony Harper</name>
<grade>Administrative Assistant</grade>
<grade_type_id>1</grade_type_id>
<legacy>reviews</legacy>
<email>outlook</email>
<office>msoffice</office>
<update_timestamp>199093</update_timestamp>
</user>
<user user_id="88495343">
<name>Nicola Smith</name>
etc
each server cluster (of which there are 20) has 4 servers within it. Users can exist on more than one server within the cluster.
what I need to do is divide the main dump into a dump for each server cluster, and make each cluster dump look like
serverdump_X4452L.xml
<?xml version="1.0" encoding="UTF-8"?>
<users>
<user user_id="77397610">
<name>Tony Harper</name>
<grade>Administrative Assistant</grade>
<grade_type_id>1</grade_type_id>
<legacy>reviews</legacy>
<email>outlook</email>
<office>msoffice</office>
<update_timestamp>199093</update_timestamp>
<servers_on>"H2212A,H2212C,H2212D"</servers_on>
</user>
<user user_id="88495343">
<name>Nicola Smith</name>
etc
as can be seen aswell as splitting the server clusters up I need to rearrange the data slightly, sorting it primarily by user instead of server, and by adding an element to each user that contains a concatonated list of the servers the user is on.
It looks like you're trying to output for a different file for each server, is that correct? If so, this is very doable, but you're running up against one of the incompatibilities between XSL parsers.
What server-side programming technology do you plan to use? Is it Windows or Linux? Theoretically, it shouldn't matter, because XSL parsing is XSL parsing. But Microsoft's MSXML object doesn't support the xsl:document tag. I usually use on oldie but goodie called XT [blnz.com] for this but you will need to be able to run Java.
Let me know if you want to take the next step. If you've never done this before, it can be a strange trip. XSL template matching is somewhat mind-blowing at first (probably the reason it's not more widely talked about).
It looks like you're trying to output for a different file for each server, is that correct? If so, this is very doable, but you're running up against one of the incompatibilities between XSL parsers.
Yes I am trying to output a different file for each server cluster.
Welcome to WebmasterWorld.
My limited knowledge of XSLT suggests using XSLT 2.0, which has xsl:result-document that can generate multiple output documents from many inputs. You'd need a Saxon 7.0 processor which I expect is avaliable on a variety of platforms.
If you don't get a complete answer here, you could ask the experts at Mulberry Technologies XSL-List [mulberrytech.com].
(note the XML example isn't the full XML file, I can't post that for security reasons, and it includes some time sensitive information).
I misread your first post and took it that you had multiple inputs files. Seeing there's only one I thought I'd have a go.
Bare in mind this is the forefront of my experience of XSLT but it works for me on a file similar to the one you suggest. I don't know about how well optimised it is and there is probably similar avaliable in XSLT 1.0 but I've jumped into XSLT 2.0 and know nothing of the workarounds you'd need.
Note: it is XSLT 2.0 and you will need the Saxon processor from [saxon.sourceforge.net ] (currently the only processor for XSLT 2.0). I don't know about which platforms it works on but expect you'll find one to suit.
Also: you need to replace PATHTODIRECTORY with the path to the output directory.. eg 'C:/Output/'
XSL follows..
Hope it helps.
<?xml version="1.0"?>
<xsl:transform xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="2.0">
<xsl:output method="xml" indent='yes' encoding="utf-8"/>
<xsl:template match="cluster">
<xsl:variable name="filename" select="concat('file:///','PATHTODIRECTORY','serverdump_',@name,'.html')"/>
<xsl:result-document href="{$filename}">
<users>
<xsl:for-each select=".//user[not(@user_id=preceding::*/@user_id)]">
<xsl:sort select="."/>
<user><xsl:attribute name="user_id" select="./@user_id"/>
<xsl:copy-of select="name"/>
<xsl:copy-of select="grade"/>
<xsl:copy-of select="grade_type_id"/>
<xsl:copy-of select="legacy"/>
<xsl:copy-of select="email"/>
<xsl:copy-of select="office"/>
<xsl:copy-of select="update_timestamp"/>
<servers_on>
<xsl:variable name="user" select="./@user_id" />
<xsl:for-each select="../../..//user[@user_id=$user]">
<xsl:value-of select="../../@name"/>
<xsl:if test="position()!=last()">, </xsl:if>
</xsl:for-each>
</servers_on>
</user>
</xsl:for-each>
</users>
</xsl:result-document>
</xsl:template>
</xsl:transform>
--added--
I've just added <xsl:sort select="."/> which should sort by user_id ascending.
Also, one flaw I'm aware of is that if a user did exist on more than one cluster they don't show on any but the first.
(I did paste the code in with indents but they've been stripped off.)
--/added--
Another ~flaw.. obviously, there's no conflict management. If data in a user's records were to differ the output would give the first occurence of it.
Couple of quick questions though...
1. The Saxon processor needs to be located in what directory? (root of webpage I assume)
2. The output appears to be an html file, what would I need to do if I wanted the output as an XML file? (Ideally I'd like the output as an XML file and there's further as I'd like the functionality on the website for the user to be able to sort the info how they'd like (plus I can also run the simpler transforms on it for displaying for example users who have been last updated within the last day etc)).
Install Saxon in the directory of your choice and so long as your PATH can find java, you can call by command line as
--------
java -jar C:\saxon\saxon7.jar input.xml transform.xsl >output.html
where C:\saxon\saxon7.jar is your location for saxon7.jar
--------
I don't know how you'd call from script.
2.
The output is XML. My mistake - the file is named .html if you correct that to .xml it will represent what the file actually is.
ie. 'serverdump_',@name,'.html')"/>
should be 'serverdump_',@name,'.xml')"/>
If you then want HTML you can do a transform that contains an html template and xsl:value-of
eg
<?xml version="1.0"?>
<xsl:transform xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="2.0">
<xsl:output method="html" indent='yes' encoding="utf-8"/>
<xsl:template match="/">
<html><body>
<xsl:for-each select="//name">
<td><xsl:value-of select="."/></td>
</xsl:for-each>
</body></html>
</xsl:template>
</xsl:transform>
On my to-do list is finding how to call Saxon without generating this redundant file.
when i do:- run java -jar saxon7.jar input.xml transform.xsl >output.html
I get the following errors
Set uncaught java.lang.Throwable
Set deferred uncaught java.lang.Throwable
>
VM Started:
Exception occurred: java.lang.ClassNotFoundException (uncaught)"thread=main", java.net.URLClassLoader$1(), line=199 bci=72
It's been a while since I've installed Java but goto [java.sun.com ]
under Download Bundles click on Separate Bundles = J2SE 1.4.2 and accept the t&c and you'll be offered different flavours.
Once installed as suggested above you'll need to set your PATH to include the path to Java.. which if memory serves, means setting to x/y/z/java/bin/ but it should be clear from the docs.. if not try Google for similar to 'setting path to java 1.4.2 Linux'
HTH
I don't care about the output.html (i'm not getting that far, and could live with it if i was:) )
I have jdk 1.4.2sdk (win XP) installed at H:\j2sdk1.4.2_04\, with java.exe and jav.exe in H:\j2sdk1.4.2_04\bin.
When i run jdb.exe and put run java -jav path\saxon7.jar path\serverdump.xml path\transform.xsl >output.html
i get the errors as detailed above.
I think its tied up with java not being able to find the classes stored in the subdirectories of saxon7-9-1\source\, but I don't know where they should be or how to direct java at them
java -jav path\saxon7.jar path\serverdump.xml path\transform.xsl >output.html
should be
java -jar path\saxon7.jar path\serverdump.xml path\transform.xsl >output.html
-----
added
Also have you set PATH
eg.
PATH = "C:\WINDOWS;C:\WINDOWS\COMMAND;C:\j2sdk1.4.2_02\bin;C:\XML-MSV"
<servers_on>
<xsl:variable name="user" select="./@user_id" />
<xsl:for-each select="../../..//user[@user_id=$user]">
<xsl:value-of select="../../@name"/>
<xsl:if test="position()!=last()">, </xsl:if>
</xsl:for-each>
leaves me with the servers squashed together (eg h2332hh3332hh2223h) how do i get it so the string is at least space deliminated (h2332h h3332h) or comma/space deliminated (h2332h, h3332h).
Not quite sure how this bit works so cant do it myself:)
I suppose you could try explicitly suggesting the output there is text..
<xsl:if test="position()!=last()">
<xsl:text>, </xsl:text>
</xsl:if>
