xmlStor.py

import sys
from lxml import etree

CARD_MULT=1
CARD_SINGLE=0

class xmlStor:
    
    properties = {}

    def __init__(self):
        self.tree = None
        for i in self.properties.keys():
            if not self.properties[i].has_key('card'):
                self.properties[i]['card'] = CARD_SINGLE
        
    @property
    def blob(self):
        if self.tree is not None:
            return etree.tostring(self.tree)
        return None

    # @blob.setter
    def setBlob(self, blob):
        self.tree = etree.fromstring(blob)

    def _checkproperty(self, name):
        """ Check if a valid property with the given name exists """
        if self.properties.has_key(name):
            if self.properties[name].has_key('xpath'):
                return True
        return False
        
    def __getattr__(self, name):
        if name == "blob":
            return self.blob()
        elif name == "tree":
            return self.__dict__["tree"]

        if self._checkproperty(name):
            try:
                # execute xpath
                nodes = self.tree.xpath(self.properties[name]['xpath'])
                # check for results 
                if nodes and len(nodes) > 0:
                    # if this property is an attribute 
                    if self.properties[name].has_key('attribute'):
                        # check cardinality
                        if self.properties[name]['card'] == CARD_MULT: 
                            return [ n.get(self.properties[name]['attribute'], u"") for n in nodes ]
                        else:
                            return  nodes[0].get(self.properties[name]['attribute'], u"")
                    else:
                        if self.properties[name]['card'] == CARD_MULT:
                            return [ n.text for n in nodes ]
                        else:
                            return nodes[0].text
                # no results 
                else:
                    return None

            # xpath failed 
            except Exception, e:
                return None
        # this property does not exists
        else:
            return None

    def __setattr__(self, name, value):
        if name == "blob":
            self.setBlob(value)
            return 
        if name == "tree":
            self.__dict__["tree"] = value
            return 
        
        if self._checkproperty(name):
            try:
                # execute xpath
                nodes = self.tree.xpath(self.properties[name]['xpath'])
                # check for results 
                if nodes and len(nodes) > 0:
                    # check cardinality
                    if type(value).__name__ == "list" and self.properties[name]['card'] == CARD_MULT:
                        l = min(len(name), len(nodes))
                        for nodeNum in range(0, l-1):
                            # for an attribute
                            if self.properties[name].has_key('attribute'):
                                nodes[nodeNum].set(self.properties[name]['attribute'], value[nodeNum])
                            else:
                                nodes[nodeNum].text = value[nodeNum]
                    else:
                        if self.properties[name].has_key("attribute"):
                            nodes[0].set(self.properties[name]['attribute'], value)
                        else:
                            nodes[0].text = value 
                # no results 
                else:
                    return None

                if type(value).__name__ == "list":
                    if len(value) > len(nodes):
                        # take the parent node of the last node 
                        parentNode = nodes[len(nodes)-1].getparent()
                        for i in range(len(nodes)-1, len(value)-1):
                            n = etree.SubElement(parentNode, nodes[len(nodes)-1].tag) 
                            if self.properties[name].has_key('attribute'):
                                n.set(self.properties[name]['attribute'], value[i])
                            else:
                                n.text = value[i]

            # xpath failed 
            except Exception, e:
                return None
        # this property does not exists
        else:
            return None


class testBlob(xmlStor):
    properties = {"update": { "xpath": "//span[@class='update']" }, 
                  "uuid": { "xpath": "//div[@class='product']", "attribute": "id"}, 
                  "status": { "xpath": "//span[@class='status']" }, 
                  "links": { "xpath": "//a", "card": CARD_MULT, "attribute": "href" },
                  "uriLink": { "xpath": "//a[@class='uri']" } }


if __name__ == "__main__":
    
    t = etree.parse(sys.argv[1])
    print "Parsed"
    blobs = []
    for b in t.xpath("//div[@class='product']"):
        print "---------------"
        blob = testBlob()
        blob.blob = etree.tostring(b)
        blob.uuid = "urn:uuid:"+blob.uriLink
        blob.links.append("http://foobar")
        links = blob.links
        links.append("http://foobar")
        blob.links = links
        blobs.append(blob)

    for b in blobs:
        print "--"
        print """
Update: %s
uuid: %s
uriLink: %s
status: %s
links: %s"""%(b.update, b.uuid, b.uriLink, b.status, str(b.links))
    
    f = open("out.html", "w")
    for i in blobs:
        f.write(i.blob)
    f.close()