## I wanted to work on something like the following example string
require 'hpricot'
string = '<html><a></a><a href="/123/456" title="2009-04-06">posted on April
2009</a></html>'
h = Hpricot(string)
t = "2009-04-06"
## Here it goes: confusion No.1
h.at('a[@title*="2009-04-06"]')
##=> returns the 2nd anchor element, as expected.
h.at('a[@title*=Time.now.strftime("%Y-%m-%d")]')
##=> *1st anchor element. Why is that??*
I'm not sure why it is returning {emptyelem <a>}, but I can tell you
why its not returning the element you expect: because you didn't use
string interpolation so that the call to Time.now.strftime(...) would
be evaluated and inserted into the string. This selects the expected
element:
h.at("a[@title*=#{Time.now.strftime('%Y-%m-%d')}]")
h.at("a[@title*=#{t}]")
##=> 2nd anchor. works fine
h.at('a[@title*="#{t}"]')
##=> *nil. Because of the single quote?*
Exactly, that's just ruby single- versus double-quote string behavior.
With the same setup as you used:
irb(main):037:0> "#{t}"
=> "2009-04-06"
irb(main):038:0> '#{t}'
=> "\#{t}"
## And here comes another confusion:
year = "2009"
h.at("a[@title*=#{t}][text()*='2009']")
##=> 2nd anchor, as expected.
h.at("a[@title*=#{t}][text()*=#{year}]")
##=> *nil. Why is that? Hpricot can't handle #{} more than once?*
Do you mean for these to pass different strings to h.at()? Look at the
strings you are using.
irb(main):048:0> puts [ "a[@title*=#{t}][text()*='2009']",
irb(main):049:1* "a[@title*=#{t}][text()*=#{year}]" ]
a[@title*=2009-04-06][text()*='2009']
a[@title*=2009-04-06][text()*=2009]
So, you are just getting unreliable results when you aren't using
quotes around the values you are searching for. This version works,
where the second one above did not:
h.at("a[@title*='#{t}'][text()*='#{year}']")
Note that I've put quotes on both values, though at least in this
example the title appears to work without them.
···
On Mon, Apr 6, 2009 at 4:11 AM, Wang Jian <jwang376@gmail.com> wrote:
