source: products/quintagroup.seoptimizer/trunk/quintagroup/seoptimizer/browser/keywords.py @ 1313

import urllib, re, os, commands
from xml.dom import minidom, Node

from zope.interface import implements
from Products.Five.browser import BrowserView

from Products.CMFCore.utils import getToolByName
from Products.CMFPlone import PloneMessageFactory as _

from interfaces import IValidateSEOKeywordsView

class ValidateSEOKeywordsView(BrowserView):

    implements(IValidateSEOKeywordsView)

    def validateKeywords(self, text):
        """ see interface """
        # extract keywords from text
        if not text.strip():
            return _(u'Keywords list is empty!')

        keywords = map(lambda x: x.strip(), text.lower().split('\n'))
        if not keywords:
            return _(u'Keywords list is empty!')

        # request html page of context object
        url = '%s?qseo_without_additional_keywords=1' % self.context.absolute_url()
        #try:
            #page = urllib.urlopen(url)
        #except IOError:
            #return _('Could not find requested page')

        #page_html = page.read()
        #if not page_html:
            #return _('Page is empty')

        # extract words from body from html page

        # this block work only with valid html
        #doc = minidom.parseString(page_html)
        #rootNode = doc.documentElement
        #bodies = rootNode.getElementsByTagName('body')
        #if len(bodies) > 0:
            #body = bodies[0]
        #else:
            #return _(u'Invalid page html')
        #page_words = []
        #self.walkTextNodes(body, page_words)

        # this block work even with invalid html
        #pattern = re.compile('<\s*body[^>]*>(.*?)<\s*/\s*body\s*>', re.S|re.M|re.I)
        #search = pattern.search(page_html)
        #if search:
            #body_html = search.group(1)
        #else:
            #return _('Invalid html code on page')

        #page_text = self.strip_tags(body_html)
        #page_words = page_text.lower().split()

        # extract words from url page using lynx browser
        page_text = commands.getoutput('lynx --dump --nolist %s' % url).lower()
        if page_text and page_text != 'sh: lynx: command not found':
            #page_words = page_text.lower().split()
            page_text = page_text
        else:
            return _(u'Could not find lynx browser!')

        # check every keyword on appearing in body of html page
        missing = []
        added = {}
        for keyword in keywords:
            if keyword not in added.keys() and not re.compile(r'\b%s\b' % keyword, re.I).search(page_text):
                missing.append(keyword)
                added[keyword] = 1

        # return list of missing keywords
        if missing:
            msg = u"""Next keywords did not appear on the page:\n%s""" % '\n'.join(missing)
        else:
            msg = u"""All keywords found on the page!"""
        return _(msg)

    def walkTextNodes(self, parent, page_words=[]):
        for node in parent.childNodes:
            if node.nodeType == Node.ELEMENT_NODE:
                self.walkTextNodes(node, page_words)
            elif node.nodeType == Node.TEXT_NODE:
                value = node.nodeValue
                if value is not None:
                    page_words.extend(map(lambda x: x.lower(), value.split()))

    def strip_tags(self, in_text):
        s_list = list(in_text)
        i,j = 0,0

        while i < len(s_list):
            if s_list[i] == '<':
                while s_list[i] != '>':
                    # pop everything from the the left-angle bracket until the right-angle bracket
                    s_list.pop(i)
                    
                # pops the right-angle bracket, too
                s_list.pop(i)
            else:
                i=i+1
                
        # convert the list back into text
        join_char=''
        return join_char.join(s_list)