From: ruby talk <rubytalk@...>
Date: 2006-12-16T03:55:15+09:00
Subject: Re: Hpricot html parsing

Ask:

http://code.whytheluckystiff.net/hpricot/ticket/32

text in xpath should return a text node if present. For example:
(doc/"/html/body/div[1]/*/table[0]/tr[0]/*/b[9]/text")

Currently I am using the search and next_node:

    doc.search("/html/body/div[1]/*/table[0]/tr[0]/td/b"){|x|
@movie_plot=x.next_node.to_s.strip if x.inner_html=="Plot Outline:" }


And receive

Author:
    why
Message:

        * lib/hpricot/elements.rb: added support for selecting text
nodes with text(): //p/text(), //p[a]//text(), etc.
        * lib/hpricot/traverse.rb: ditto.
        * lib/hpricot/tag.rb: the pathname method reports the path
fragment needed to get to this node.
        * lib/hpricot/parse.rb: handle possible empty processing instruction.
http://code.whytheluckystiff.net/hpricot/changeset/87




On 12/13/06, David Vallner <david@vallner.net> wrote:
> Peter Szinek wrote:
> > paragraphs = doc/'p'
> >
> > good_elems = paragraphs.map.reject {|elem|  ((elem/"img").empty?) }
>
> Which once again makes me wish paragraphs = doc/'//p[img]/text()'
> worked. This could be doable if you asked Hpricot to provide you with
> the REXML document (it's probably out of scope for the intendedly simple
> XPath engine Hpricot uses natively), but unfortunately I can't for the
> heck of it figure out how to make REXML accept the final /text(), even
> though the parser claims to support XPath 1.0 except a few exceptions,
> that one not being noted.
>
> David Vallner
>
>
>
>