Re: Use of DOMXPathEvaluator

[David Lawrence <davidl@jlab.org>]

Hi Richard,

    It looks like XQuery is (or is soon to be) a w3c standard as well:

http://www.w3.org/TR/xquery/

Though I'm still unclear as to whether we'll ever take advantage of all 
that power.

Regards,
-David

Richard Jones wrote:
> Dave,
> I agree, we should discuss it, and look at any alternatives that might 
> look better.  XPath has the advantage of being a w3c-standardized 
> interface.
> -rtj
>
> David Lawrence wrote:
>>
>> Hi Richard,
>>
>>    Thanks for the example and the quick response. Your explanation 
>> makes sense. I was confused since I was accessing the 
>> DOMXPathEvaluator though the DOMDocument class, I though that it was 
>> implementing the xpath evaluations.
>>
>>    We should discuss this at our next software meeting (date and time 
>> as yet not determined). In an attempt to get something working, I had 
>> implemented an extremely simple parser. It works on only a very small 
>> subset of the xpath syntax, but is capable of at least understanding 
>> strings like the one you gave in your example. It would certainly be 
>> desirable to have a fully XPath 1.0 compliant parser instead. 
>> However, as you know, I am very cautious when it comes to package 
>> creep. Also, if we are going to bring in a whole other package in 
>> order to gain XPath compliance, then should we consider bringing in 
>> something to do XQuery instead (i.e. XQilla)? Xalan, of course, has 
>> the advantage that we already have some software in our repository 
>> that uses it. How much power will we need here?
>>
>> Regards,
>> -David
>>
>> Richard Jones wrote:
>>> Dave,
>>>
>>> You may have noticed in the xerces API documentation that references 
>>> to XPath objects do not have explicit types (void *).  This is 
>>> because xerces has hooks for xpath extensions but no support in the 
>>> built-in classes.  The evaluate method of the DOMXPathEvaluator is 
>>> pure virtual, you may have noticed.
>>>
>>> The apache project that implements a full userspace xpath library is 
>>> called xalan.  Like xerces it comes in c++ and java implementations, 
>>> and xalan is implemented on top of xerces.  The download binary 
>>> tarballs and installation instructions are very similar to xerces, 
>>> and found on the web site http://xalan.apache.org.  I installed 
>>> xalan-c a long time ago at the same time as I installed xerces 
>>> because it has a full XSLT translation engine in it.  You may have 
>>> it as well.  The two versions (xerces and xalan) should be kept in 
>>> sync.
>>>
>>> In the xalan distribution there are several sample programs to show 
>>> how the xpath API works.  I took one of them, SimpleXPathAPI.cpp, 
>>> and tweaked it to work with the hdds geometry document tree.  You 
>>> can find it in subversion within the hdds package called 
>>> xpath-example.cpp.  Makefile knows how to build it, provided that 
>>> your setup script defines the environment variable XALANCROOT.
>>>
>>> For example, to extract the atomic number of Antimony from the 
>>> materials db, go into the hdds directory, make xpath-example, and 
>>> try the following command.
>>>
>>> $  ./xpath-example main_HDDS.xml / 
>>> '//hdds:element[@name="Antimony"]/@a'
>>> The string value of the result is:
>>> 121.75
>>>
>>>
>>>
>>> David Lawrence wrote:
>>>>
>>>> Hi Richard,
>>>>
>>>>    I am currently working on the API through which JANA will read 
>>>> in geometry information. My plan is to have it read in from the 
>>>> hdds file using xerces. After doing a little web research, I have 
>>>> come upon something you mentioned some time ago, xpath. It seems 
>>>> this is the appropriate way to specify a node for which one needs 
>>>> the value of a specific attribute. (Unless you have a better 
>>>> suggestion).
>>>>
>>>>    I have been trying to use it with the evaluate method of a 
>>>> DOMDocument object which looks like it inherits from 
>>>> DOMXPathEvaluator. I have not had much luck getting it to work yet 
>>>> and can't seem to find any examples on the Web. Have you used this 
>>>> before? Any insights would be great.
>>>>
>>>> Regards,
>>>> -David
>>>>
>>>
>>
>

-- 

------------------------------------------------------------------------
  David Lawrence Ph.D.
  Staff Scientist                 Office: (757)269-5567   [[[  [   [ [       
  Jefferson Lab                   Pager:  (757)584-5567   [  [ [ [ [ [   
  http://www.jlab.org/~davidl     davidl@jlab.org         [[[  [[ [[ [[[
------------------------------------------------------------------------