import xml.sax

class EpiserverFeedValidator(ContentHandler):
    def __init__(self):
        contentTypes = {"link":checkUrl, "p:imageLink":checkImageUrl, "title":checkTitle, "p:unitPrice":checkPrice,}
        self.elementType = ""
        self.elementContent = ""
        self.processElement = False
        self.guid = ""

        linkSet = {}
        imageLinkSet = {}

    def startElement(self, name, attrs):

        self.elementType = ""
        self.elementContent = ""
        self.processElement = False

        if name is "guid":
            self.guid = name

        if name in contentTypes:
            self.elementType = name
            self.processElement = True

    def characters(self, ch):
        if self.processElement:
            self.elementContent = self.elementContent + ch

    def endElement(self, name):
        if self.processElement:
            processElements(name)

    def processElements(name):
        contentTypes(name)(self)

    def checkUrl(self):

        if len(self.elementContent) > 300:
            print("LINK TOO LONG: " + self.elementContent)

        if self.elementContent in linkSet:
            print("DUPLICATE LINK: " + self.elementContent)
        else:
            linkSet.add(self.elementContent)


    def checkImageUrl(self):

        if len(self.elementContent) > 300:
            print("IMAGE LINK TOO LONG FOR " + self.guid + " : " + self.elementContent)

        if self.elementContent in imageLinkSet:
            print("DUPLICATE IMAGE LINK FOR " + self.guid + " : " + self.elementContent)
        else:
            imageLinkSet.add(self.elementContent)

    def checkTitle(self):
        if len(self.elementContent) > 200:
            print("IMAGE LINK TOO LONG FOR " + self.guid + " : " + self.elementContent)

    def checkPrice(self):
        if self.elementContent is "":
            print("UNIT PRICE MISSING FOR " + self.guid + " : " + self.elementContent)


parser = xml.sax.make_parser()
parser.setContentHandler(EpiserverFeedValidator)