Das deutsche Python-Forum

Code: Alles auswählen

import xml.sax


test = """
<root>
<zone>
<seg xml:id='001' a='62' b='90' c='75' d='92'>aba</seg>
<seg xml:id='002' a='62' b='90' c='75' d='92'>abc</seg>
<seg xml:id='003' a='62' b='90' c='75' d='92'>abd</seg>
</zone>
<text>
<div>
<seg sameAs='#001'>aba</seg>
<seg sameAs='#002'>abc</seg>
<seg sameAs='#003'>abd</seg>
</div>
</text>
</root>
"""

class my_handler(xml.sax.ContentHandler):
    def __init__(self):
        self.id_mappings = {}
        self.in_zone = False
        self.in_text = False
        

    def startDocument(self):
        pass

    def endDocument(self):
        pass

    def startElement(self,name,attrs):
        print "startElement", name, attrs

        if name == "zone":
            assert not self.in_zone
            self.in_zone = True
        if name == "seg" and self.in_zone:
            self.id_mappings[attrs["xml:id"]] = dict((k, v) for k, v in attrs.items())
                             

    def endElement(self,name):
        if name == "zone":
            assert self.in_zone
            self.in_zone = False

    def startElementNS(self,name,qname,attrs):
        print "startElementNS", name, qname, attrs


    def endElementNS(self,name,qname):
        print "end", name, qname

    def characters(self,data):
        print "characters", data

    def ignorableWhitespace(self):
        print "ignorableWhitespace"


h = my_handler()
xml.sax.parseString(test, h)
print h.id_mappings