Manipulating list from lxml xpath queries

Question

Today I tried lxml as I got very nasty html output from particular web service, and I didn't want to go with re module, just for change and to learn something new. And I did, browsing http://codespeak.net/lxml/ and http://stackoverflow.com in parallel

I won't try to explain above html template, but just for overview it's full of deliberately nested tables.

I extracted part of interest with html parser then find_class() and iterating through TR with xpath (and even this TRs have tables inside). Now I'm trying to extract data pairs based on class and id attributes:

• name child has class "title"
• value child has id "text"

---

Code looks something like this:

fragment = root.find_class('foo')

for node in fragment[0].xpath('table[2]/tr'):
    name = node.xpath('//div[@id="title"]')
    value = node.xpath('//td[@class="text"]')

Problem is that not every TR, that I'm iterating, has those pairs: some are only with name (id "title") so later when I try to zip them I get wrongly paired data.

I tried couple of things that came to my mind but nothing successful: I tried to compare list length (for name and value) and if they don't match skip name lookup, then if they don't match, delete last list item (in many ways) but nothing worked. For example:

if not len(name) == len(value):
    name.pop()

or

if len(name) == len(value):
    name = node.xpath('//div[@id="title"]')

value = node.xpath('//td[@class="text"]')

Some comments from more experienced?

Answer 1

How's this?

from lxml import etree
doc = etree.HTML(open('test.data').read())

for t in doc.xpath('//table[.//div[@id="title"] and .//td[@class="text"]]'):
    print etree.tostring(t.xpath('.//div[@id="title"]')[0])
    print etree.tostring(t.xpath('.//td[@class="text"]')[0])
    print "--"

Yielding:

<div id="title">
              <span class="Browse">string</span>
            </div>

<td class="text" style="padding-left:5px;">
            <a href="/***/***.dll?p=***&amp;sql=xxx:yyy">string</a>
          </td>

--
<div id="title">
              <span>string</span>
            </div>

<td class="text" style="padding-left:5px;">
            <a href="/***/***.dll?p=***&amp;sql=xxx:yyy">string</a>
          </td>

--
<div id="title">
              <span>string</span>
            </div>

<td class="text" style="padding-left:5px;">
            Gospodar of Lutaka
          </td>

--
<div id="title">
              <span>string</span>
            </div>

<td class="text" style="padding-left:5px;">
            1986
          </td>

--
<div id="title">
              <span>string</span>
            </div>

<td class="text" style="padding-left:5px;">
            Sep 1985-Dec 1985
          </td>

--
<div id="title">
              <span>string</span>
            </div>

<td class="text" style="padding-left:5px;">
            Elektra
          </td>

--
<div id="title">
              <span>string</span>
            </div>

<td class="text" style="padding-left:5px;">
            54:51
          </td>

--
<div id="title">
              <span>string</span>
            </div>

<td class="text" style="padding-left:5px;">
          </td>

--

---

Update, extended the leading portion of the xpath expression to eliminate an undesired result. Thanks to Alejandro for pointing this out and suggesting a fix that didn't seem to work out for otrov.

from urllib2 import urlopen
from lxml import etree
doc = etree.HTML(urlopen('http://pastebin.com/download.php?i=cg5HHJ6x').read())

for t in doc.xpath('//table/tr/td/table[.//div[@id="title"] and .//td[@class="text"]]'):
    print etree.tostring(t.xpath('.//div[@id="title"]')[0])
    print etree.tostring(t.xpath('.//td[@class="text"]')[0])
    print "--"

Answer 2

Now, with input sample, is more clear what you are asking.

Just this one XPath 1.0 expression return a node set with div and td pair (in document order):

/table/tr/td/table[tr/td/div[@id='title']]
                  [tr/td[@class='text']]
                  /tr//*[self::div[@id='title'] or self::td[@class='text']]

As proof, this stylesheet:

<xsl:stylesheet version="1.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform"&gt;
    <xsl:template match="/">
        <result>
            <xsl:copy-of 
                 select="/table/tr/td/table[tr/td/div[@id='title']]
                                           [tr/td[@class='text']]
                                           /tr//*[self::div[@id='title'] or
                                                  self::td[@class='text']]"/>
        </result>
    </xsl:template>
</xsl:stylesheet>

Output (with proper input sample, because you miss a closing td):

<result>
    <div id="title">
        <span>string</span>
    </div>
    <td class="text" style="padding-left:5px;">
        <a href="/***/***.dll?p=***&amp;sql=xxx:yyy">string</a>
    </td>
    <div id="title">
        <span>string</span>
    </div>
    <td class="text" style="padding-left:5px;">
            Gospodar of Lutaka
    </td>
    <div id="title">
        <span>string</span>
    </div>
    <td class="text" style="padding-left:5px;">
            1986
    </td>
    <div id="title">
        <span>string</span>
    </div>
    <td class="text" style="padding-left:5px;">
            Sep 1985-Dec 1985
    </td>
    <div id="title">
        <span>string</span>
    </div>
    <td class="text" style="padding-left:5px;">
            Elektra
    </td>
    <div id="title">
        <span>string</span>
    </div>
    <td class="text" style="padding-left:5px;">
            54:51
    </td>
    <div id="title">
        <span>string</span>
    </div>
    <td class="text" style="padding-left:5px;"></td>
</result>