tcljava-user

[tcljava-user] jacl 1.3.1 regexp

[Johannes K. <joh...@kl...>]

Hi all,

I look for a way to parse an XML-File in IBMs wsadmin, which uses jacl
1.3.1 as a scripting language.

I found a project called sa4was [1] on IBMs developerworks which does
parsing XML with regexp. sa4was works with WebSpher 5.x (which used jacl
1.2.x) however in WebSphere 6.0 (which uses jacl 1.3.1) it doesn't work
anymore.

Some code in sa4was does the following:

============================================
    while {[regexp {([^=]+)="([^"]*?)"(.*)} $restOfTag dontCare
attributeName attributeValue restOfTag]} {
        if {$attributeName == "id"} {
            set idValue $attributeValue
            break
        }
    }
============================================

and there I get the error 
"couldn't compile regular expression pattern: nested *?+"


I found out that regexp changed in jacl 1.3.1 and non-greedy regexp
(with this "*?" expression) doesn't work anymore. Is that right?  

Questions:
a) So are there workarounds for something like that? 
b) Or are there some better ways to parse XML in jacl?

Any help would be appreciated!!

best regards,
Johannes

Links:

1: http://www.alphaworks.ibm.com/tech/sa4was

Re: [tcljava-user] jacl 1.3.1 regexp

[Patrick F. <pfi...@oz...>]

On Monday 09 April 2007 13:32, Johannes Kleinlercher wrote:
> Hi all,
>
> I look for a way to parse an XML-File in IBMs wsadmin, which uses jacl
> 1.3.1 as a scripting language.
>
> I found a project called sa4was [1] on IBMs developerworks which does
> parsing XML with regexp. sa4was works with WebSpher 5.x (which used jacl
> 1.2.x) however in WebSphere 6.0 (which uses jacl 1.3.1) it doesn't work
> anymore.
>
> Some code in sa4was does the following:
>
> ============================================
>     while {[regexp {([^=]+)="([^"]*?)"(.*)} $restOfTag dontCare
> attributeName attributeValue restOfTag]} {
>         if {$attributeName == "id"} {
>             set idValue $attributeValue
>             break
>         }
>     }
> ============================================
>
> and there I get the error
> "couldn't compile regular expression pattern: nested *?+"
>
>
> I found out that regexp changed in jacl 1.3.1 and non-greedy regexp
> (with this "*?" expression) doesn't work anymore. Is that right?
>
> Questions:
> a) So are there workarounds for something like that?
> b) Or are there some better ways to parse XML in jacl?
>

The quick work around for existing sa4was scripts in a WAS 6 environment is to 
downgrade JACL from 1.3.1 to 1.2.X by replacing the 1.3.1 jacl.jar and 
tcljava.jar jar files in the was/installdirectory/lib directory. 

However I would not recommend using sa4was with 6.0 because it's is not 
supported by IBM in 6.x.  I have a JACL script library for WAS which I am 
posting on the internet this week that covers most of the WAS admin 
functions.  You should have a look at that. 

The Tcl xml parser TclXml has not been ported to JACL so you need to use 
'regexp' or 'string match' to parse XML.  

If you want to use Tcl packages that have not been ported to JACL in a 
scripted WebSphere admin environment then you will need to use TclBlend as a 
java client with the WebSphere JMX java library.  There are plenty of 
articles on Developerworks that describe WAS admin using java clients rather 
than Wasadmin clients written in Jython or Jacl.  I am re-writing most of my 
stuff to use TclBlend because while IBM might drop support for Jacl and 
Jython in future releases the JVM layer will always be there. 

Re: [tcljava-user] jacl 1.3.1 regexp

[Johannes K. <joh...@kl...>]

sa4was isn't supported anyway by IBM so I don't think that is the
point. 

I am developing my own deployment-scripts so I just wanted to take some
code of sa4was and use it in my scripts. Parsing XML files was some of
it. I didn't find any code to do that and I don't think it is easy
without a good regexp-engine. Non-greedy regexp is one important thing
for parsing XML, I think. Do you have some piece of code which can parse
XML-Files with the regexp-engine in jacl 1.3.1?

I will look at your homepage the next time and look forward to your jacl
libraries.


Am Montag, den 09.04.2007, 14:22 +0100 schrieb Patrick Finnegan:
> On Monday 09 April 2007 13:32, Johannes Kleinlercher wrote:
> > Hi all,
> >
> > I look for a way to parse an XML-File in IBMs wsadmin, which uses jacl
> > 1.3.1 as a scripting language.
> >
> > I found a project called sa4was [1] on IBMs developerworks which does
> > parsing XML with regexp. sa4was works with WebSpher 5.x (which used jacl
> > 1.2.x) however in WebSphere 6.0 (which uses jacl 1.3.1) it doesn't work
> > anymore.
> >
> > Some code in sa4was does the following:
> >
> > ============================================
> >     while {[regexp {([^=]+)="([^"]*?)"(.*)} $restOfTag dontCare
> > attributeName attributeValue restOfTag]} {
> >         if {$attributeName == "id"} {
> >             set idValue $attributeValue
> >             break
> >         }
> >     }
> > ============================================
> >
> > and there I get the error
> > "couldn't compile regular expression pattern: nested *?+"
> >
> >
> > I found out that regexp changed in jacl 1.3.1 and non-greedy regexp
> > (with this "*?" expression) doesn't work anymore. Is that right?
> >
> > Questions:
> > a) So are there workarounds for something like that?
> > b) Or are there some better ways to parse XML in jacl?
> >
> 
> The quick work around for existing sa4was scripts in a WAS 6 environment is to 
> downgrade JACL from 1.3.1 to 1.2.X by replacing the 1.3.1 jacl.jar and 
> tcljava.jar jar files in the was/installdirectory/lib directory. 
> 
> However I would not recommend using sa4was with 6.0 because it's is not 
> supported by IBM in 6.x.  I have a JACL script library for WAS which I am 
> posting on the internet this week that covers most of the WAS admin 
> functions.  You should have a look at that. 
> 

> The Tcl xml parser TclXml has not been ported to JACL so you need to use 
> 'regexp' or 'string match' to parse XML.  
> 
> If you want to use Tcl packages that have not been ported to JACL in a 
> scripted WebSphere admin environment then you will need to use TclBlend as a 
> java client with the WebSphere JMX java library.  There are plenty of 
> articles on Developerworks that describe WAS admin using java clients rather 
> than Wasadmin clients written in Jython or Jacl.  I am re-writing most of my 
> stuff to use TclBlend because while IBM might drop support for Jacl and 
> Jython in future releases the JVM layer will always be there. 

Re: [tcljava-user] jacl 1.3.1 regexp

[Patrick F. <pfi...@oz...>]

On Monday 09 April 2007 15:37, Johannes Kleinlercher wrote:
> sa4was isn't supported anyway by IBM so I don't think that is the
> point.

Parts of the underlying JMX implementation have changed in WS6.0.  I had to 
modify some of my WAS 5.1 jacl scripts to cope with this.  Since sa4was is a 
simplification of the WAS 5.1 JMX layer it may not cope with these changes so 
if you are starting from scratch I would avoid using sa4was with 6. 

The sarwas author is aware of these issues.  See thread on the se4was forum. 

http://www-128.ibm.com/developerworks/forums/dw_thread.jsp?message=13714004&cat=28&thread=78335&treeDisplayType=threadmode1&forum=443#13714004 

>
> I am developing my own deployment-scripts so I just wanted to take some
> code of sa4was and use it in my scripts. Parsing XML files was some of
> it. I didn't find any code to do that and I don't think it is easy
> without a good regexp-engine. Non-greedy regexp is one important thing
> for parsing XML, I think. Do you have some piece of code which can parse
> XML-Files with the regexp-engine in jacl 1.3.1?
>

I have never had the need to parse XML files in JACL mainly because IBM do not 
recommend manual modification of the underlying WAS xml config files.  

I don't know whether jacl 1.4.0 supports non greedy regexp but if it does then 
you can upgrade the WAS 6.0 jacl environment by simply downloading the jacl 
1.4 binary and replacing the jacl.jar and tcljava.jar files in the 
wasinstall/appserver/lib directory with the 1.4 jars.  I have tested 1.4 with 
WAS 6.0 and have not had a single problem.  It's rock solid. 

You can test the regexp in jacl 1.4 independently.  See this link.  
 
link:http://saloon.javaranch.com/cgi-bin/ubb/ultimatebb.cgi?ubb=get_topic&f=60&t=000208

You can also run wsadmin with Jacl 1.4 as an external tool in eclipse.  See 
this link.

http://www-128.ibm.com/developerworks/forums/dw_thread.jsp?message=13714004&cat=28&thread=78335&treeDisplayType=threadmode1&forum=443#13714004

Re: [tcljava-user] jacl 1.3.1 regexp

[Mo D. <mo...@mo...>]

Johannes Kleinlercher wrote:
> Hi all,
>
> I look for a way to parse an XML-File in IBMs wsadmin, which uses jacl
> 1.3.1 as a scripting language.
>
> I found a project called sa4was [1] on IBMs developerworks which does
> parsing XML with regexp. sa4was works with WebSpher 5.x (which used jacl
> 1.2.x) however in WebSphere 6.0 (which uses jacl 1.3.1) it doesn't work
> anymore.
>
> Some code in sa4was does the following:
>
> ============================================
>     while {[regexp {([^=]+)="([^"]*?)"(.*)} $restOfTag dontCare
> attributeName attributeValue restOfTag]} {
>         if {$attributeName == "id"} {
>             set idValue $attributeValue
>             break
>         }
>     }
> ============================================
>
> and there I get the error 
> "couldn't compile regular expression pattern: nested *?+"
>
>
> I found out that regexp changed in jacl 1.3.1 and non-greedy regexp
> (with this "*?" expression) doesn't work anymore. Is that right?  
>
> Questions:
> a) So are there workarounds for something like that? 
> b) Or are there some better ways to parse XML in jacl?
>   
Currently, there is no workaround except porting regexp code back to use 
the older Tcl 8.1
style regexp syntax. Very old versions of Jacl made use or the Oro 
regexp package, but it
was non free and had to be removed from Jacl.

But, lets back up a minute. I would not even suggest that you use regexp 
commands to parse
your XML code. Using a series of regexp call like that is going to be 
SLOW! There is just
no way around it, you would be much better off using a XML parsing 
engine written in Java.
There are lots and lot of them available. You could create your XML 
parser in Java and build
up a DOM in memory and then pass a handle to the DOM back to your Tcl 
scripts. It is actually
quite easy to examine and decode a DOM tree with Tcl code once you have 
your own utility
procs that examine subtrees and extract data from nodes.

I hope that helps
Mo DeJong