From: Drew Raines <aaraines@...>
Date: 2007-04-19T05:00:06+09:00
Subject: Re: Hpricot getting a table

lrlebron@gmail.com wrote:

> This gets me a lot closer to what I need.
> I'm having some problems with syntax. If I'm reading the docs
> correctly map returns an array. So I should be able to do something
> like
>
> arrTables = (doc/"tables").map
>
> And then access each table individually. For example
>
> arrTables[0]

Array#map wasn't particularly relevant in my example; I just used it
to iterate #puts over the result of (doc/"table").

The real lesson to glean from my response is that:

  (doc/"table")

is much nicer than:

  doc.search("/html/body/table[5]/tr/td[2]/div[3]/table/tr/td/div[1]/table")

...which, like Peter alluded to, is fairly meaningless to us because
we don't know what the full original HTML looks like.  You can grab
the <table>s from the snippet you provided with just a CSS-style
search[1].

FWIW, if you have any control over the markup, you can simply add
some unique classes to the tables you want:

  <table class="foo">...</table>
  <table class="foo">...</table>
  <table class="foo">...</table>

Then do:

  (doc/"table.foo")

That'll work regardless of how many <table>s are on the page.

-Drew

Footnotes: 
[1]  http://lnk.nu/code.whytheluckystiff.net/e7m