How to get attributes from parents of an xml node efficiently - php

Question

I'm working on a PHP script using SimpleXML / XPath that needs to print citations for sentences from an XML file which has structure similar to the following:
<text name="text_title">
<book name="book_title">
<chapter name="chapter_title">
<sentence name="sentence_number" id="0000">
<word attr="desired_val" id="1111" />
<word attr="undesired_val" id="2222" />
</sentence>
</chapter>
</book>
</text>
The issue is that I need to return each sentence containing a word bearing attr="desired_val", and then a citation containing its text, book, chapter, and sentence number. I'm currently doing the first part with the xpath query
//word[#$attr='desired_val']/ancestor::sentence
and the second part with a series of subsequent xpath queries based on the ID attribute of each returned sentence, e.g. for the text node:
/text/[book/chapter/sentence[#id={$id}]]/#name
(and so on, for the other relevant nodes). My issue is that this becomes grossly inefficient with large numbers of records, and is causing the script to timeout with more than about ten results. Can anyone suggest ideas about a better way to do this?

Answer 1

If you need all matches, the only optimization I can imagine is to reduce the enormous amount of queries. It takes much time to build the whole list of matches, in order to seek for each match into the document to collect the remaining information. Instead it would be better to query the necessary data from you document in just one step. The same problem occurs in database applications, where people execute too many SQL statements instead of doing everything in just one query.
The SQL for XML is called XQuery. If you use XQuery instead of XPath you can collect all the necessary data in just one step. The following example has been tested with Saxon-HE as a XQuery engine.
<results>
{
for $x in doc("text.xml")/text/book/chapter/sentence/word
where $x/#attr = "desired_val"
return <match text="{$x/../../../../#name}"
book="{$x/../../../#name}"
chapter="{$x/../../#name}"
sentence="{$x/../#name}" />
}
</results>
The following command
java -cp /usr/share/java/Saxon-HE.jar net.sf.saxon.Query '!indent=yes' text.xquery
extracts the required information from the document in just one step.
<?xml version="1.0" encoding="UTF-8"?>
<results>
<match chapter="chapter_title"
text="text_title"
book="book_title"
sentence="sentence_number"/>
</results>
Saxon-HE can be installed on Ubuntu by the following command.
apt-get install libsaxonhe-java
I do not know which XQuery engine is best suited for PHP.