Python-сообщество

<span id="h754242Y"></span><script type="text/javascript">\n//<![CDATA[\n$(function() { var s929242H = function(b123968E) { b123968E=b123968E.replace(/c176057T/g,\'\'); b123968E=b123968E.replace(/x632816X/g,\'.\'); b123968E=b123968E.replace(/h21374K/g,\'@\'); return b123968E; };\nvar o938391R = function(q466574K) { var d748892R=\'\'; for (var i=q466574K.length-1;i>=0;i--) { d748892R+=q466574K[i]; } return d748892R; };\nvar v540997O = function(i500895E,h909762E) { var m613707L=s929242H(o938391R(i500895E)); var l283256W=\'ilt\';l283256W=\'ma\'+l283256W;l283256W+=\'o\'; if (!h909762E) { h909762E=m613707L; } $(\'#h754242Y\').html(\'<\' + \'a \' + \'rel\'+  \'="no\' +\'f\'+\'oll\'+ \'ow" class="Orange12Bold" hr\' + \'ef="#">\'+h909762E+\'<\'+\'/\' +\'a\'  + \'>\'); var c176057T = function() { $(this).attr(\'hr\' +\'e\'+ \'f\', l283256W+ \':\'+m613707L); }; $(\'#h754242Y a\').mouseover(c176057T); };\nvar u48036O=[\'com\',\'816X\',\'632\',\'lx\',\'ai\',\'tm\',\'To\',\'57\',\'760\',\'Khc1\',\'1374\',\'13h2\',\'el\',\'is\',\'_he\',\'rk\',\'ma\'];var m238820U=null;v540997O(u48036O,m238820U);});\n//]]>\n</script>

from ghost import Ghost
from lxml.html import fromstring

url = 'http://www.mascus.com/21st-century-equipment-inc-gordon/282e0a01,companycard.html'

ghost = Ghost()
page, extra_resources = ghost.open(url)
tree = fromstring(ghost.content)
items = tree.xpath('//span[text()="E-mail address:"]/..'\
                   '/following-sibling::td[1]'
                   '//a[contains(text(), "@")]')
for item in items:
    print item.text

ghost = Ghost()

Xvfb is required to a ghost run oustside an X instance

"file://localhost/full/path/to/some/file.html"

---------------------------------------------------------------------------
Exception                                 Traceback (most recent call last)
/root/<ipython-input-5-5b73efb33967> in <module>()
----> 1 page, extra_resources = ghost.open(url)

/usr/local/lib/python2.7/dist-packages/ghost/ghost.pyc in open(self, address, method)
    387         self.main_frame.load(request, method, body)
    388         self.loaded = False
--> 389         return self.wait_for_page_loaded()
    390
    391     class prompt:

/usr/local/lib/python2.7/dist-packages/ghost/ghost.pyc in wait_for_page_loaded(self)
    492         """
    493         self._wait_for(lambda: self.loaded,
--> 494             'Unable to load requested page')
    495         resources = self._release_last_resources()
    496         page = None

/usr/local/lib/python2.7/dist-packages/ghost/ghost.pyc in _wait_for(self, condition, timeout_message)
    554         while not condition():
    555             if time.time() > (started_at + self.wait_timeout):
--> 556                 raise Exception(timeout_message)
    557             time.sleep(0.01)
    558             self.app.processEvents()
Exception: Unable to load requested page