# $Id$
# a simple interface to the W3C validator

import urllib, urllib2

try:
    import xml.etree.cElementTree as ET
except ImportError:
    try:
        import cElementTree as ET
    except ImportError:
        import elementtree.ElementTree as ET

NS = "{http://www.w3.org/2005/10/markup-validator}"

class Result(object):

    uri = None
    validity = None
    checkedby = None
    doctype = None
    charset = None
    validity = None
    errorcount = 0
    errorlist = ()
    warningcount = 0
    warninglist = ()
    recursion = None # only set by check

    def __nonzero__(self):
        return self.validity

class Validator(object):

    def __init__(self, validator="http://validator.w3.org/check"):
        self.validator = validator

    ##
    # (internal) Convert parameter set to validator URI.

    def geturi(self, **params):
        return self.validator + "?" + urllib.urlencode(params)

    ##
    # Do a quick check, using a HTTP HEAD request.  This call sets
    # the <b>uri</b>, <b>validity</b> and <b>errorcount</b> result
    # attributes only.

    def check(self, uri, **params):
        # just do a HEAD request
        # FIXME: enforce 1 second delay!
        request = urllib2.Request(self.geturi(uri=uri, **params))
        request.get_method = lambda: "HEAD"
        http_file = urllib2.urlopen(request)
        res = Result()
        res.uri = uri
        res.status = http_file.headers["X-W3C-Validator-Status"]
        res.validity = res.status == "Valid"
        res.errorcount = int(http_file.headers["X-W3C-Validator-Errors"])
        res.recursion = http_file.headers["X-W3C-Validator-Recursion"]
        return res

    ##
    # Do full check, using the SOAP interface.  This call sets as many
    # result attributes as possible, including <b>errorlist</b> and
    # <b>warninglist</b>.

    def check_full(self, uri, **params):
        # FIXME: enforce 1 second delay!
        params["output"] = "soap12"
        request = urllib2.Request(self.geturi(uri=uri, **params))
        http_file = urllib2.urlopen(request)
        tree = ET.parse(http_file)
        body = tree.find("{http://www.w3.org/2003/05/soap-envelope}Body")
        response = body.find(NS + "markupvalidationresponse")
        def getlist(elem):
            out = []
            for elem in elem:
                line = int(elem.findtext(NS + "line"))
                column = int(elem.findtext(NS + "col"))
                message = elem.findtext(NS + "message")
                out.append(((line, column), message))
            return out
        res = Result()
        for elem in response:
            if elem.tag.startswith(NS):
                tag = elem.tag[len(NS):]
                if tag == "validity":
                    res.validity = elem.text == "true"
                elif tag == "warnings":
                    res.warningcount = int(elem.findtext(NS + "warningcount"))
                    res.warninglist = getlist(elem.find(NS + "warninglist"))
                elif tag == "errors":
                    res.errorcount = int(elem.findtext(NS + "errorcount"))
                    res.errorlist = getlist(elem.find(NS + "errorlist"))
                else:
                    setattr(res, tag, elem.text)
        return res

if __name__ == "__main__":
    v = Validator()
    print vars(v.check("http://www.python.org"))
    print vars(v.check_full("http://www.cnn.com"))