Function parse_xml

pattern/text/tree.py:1185–1290

Returns the given Sentence object as an XML-string (plain bytestring, UTF-8 encoded). The tab delimiter is used as indendation for nested elements. The id can be used as a unique identifier per sentence for chunk id's and anchors. For example: "I eat pizza with a fork." =>

(sentence, tab="\t", id="")

_UID_SEPARATOR = "."

def parse_xml(sentence, tab="\t", id=""):
    """ Returns the given Sentence object as an XML-string (plain bytestring, UTF-8 encoded).
        The tab delimiter is used as indendation for nested elements.
        The id can be used as a unique identifier per sentence for chunk id's and anchors.
        For example: "I eat pizza with a fork." =>
        
        <sentence token="word, part-of-speech, chunk, preposition, relation, anchor, lemma" language="en">
            <chunk type="NP" relation="SBJ" of="1">
                <word type="PRP" lemma="i">I</word>
            </chunk>
            <chunk type="VP" relation="VP" id="1" anchor="A1">
                <word type="VBP" lemma="eat">eat</word>
            </chunk>
            <chunk type="NP" relation="OBJ" of="1">
                <word type="NN" lemma="pizza">pizza</word>
            </chunk>
            <chunk type="PNP" of="A1">
                <chunk type="PP">
                    <word type="IN" lemma="with">with</word>
                </chunk>
                <chunk type="NP">
                    <word type="DT" lemma="a">a</word>
                    <word type="NN" lemma="fork">fork</word>
                </chunk>
            </chunk>
            <chink>
                <word type="." lemma=".">.</word>
            </chink>
        </sentence>
    """
    uid  = lambda *parts: "".join([str(id), _UID_SEPARATOR ]+[str(x) for x in parts]).lstrip(_UID_SEPARATOR)
    push = lambda indent: indent+tab         # push() increases the indentation.
    pop  = lambda indent: indent[:-len(tab)] # pop() decreases the indentation.
    indent = tab
    xml = []
    # Start the sentence element:
    # <sentence token="word, part-of-speech, chunk, preposition, relation, anchor, lemma">
    xml.append('<%s%s %s="%s" %s="%s">' % (
        XML_SENTENCE,
        XML_ID and " %s=\"%s\"" % (XML_ID, str(id)) or "",
        XML_TOKEN, ", ".join(sentence.token),
        XML_LANGUAGE, sentence.language
    ))
    # Collect chunks that are PNP anchors and assign id.
    anchors = {}
    for chunk in sentence.chunks:
        if chunk.attachments:
            anchors[chunk.start] = len(anchors) + 1
    # Traverse all words in the sentence.
    for word in sentence.words:
        chunk = word.chunk
        pnp   = word.chunk and word.chunk.pnp or None
        # Start the PNP element if the chunk is the first chunk in PNP:
        # <chunk type="PNP" of="A1">
        if pnp and pnp.start == chunk.start:
            a = pnp.anchor and ' %s="%s"' % (XML_OF, uid("A", anchors.get(pnp.anchor.start, ""))) or ""
            xml.append(indent + '<%s %s="PNP"%s>' % (XML_CHUNK, XML_TYPE, a))
            indent = push(indent)