xopp2py/src/xopp2py/parser.py

import abstract_syntax

def parseFile(path) -> abstract_syntax.XournalFile:
    """Parse a .xopp file."""

    def parseLayer(context):
        elements = []
        for (event, element) in context:
            if event == "start":
                if element.tag == "text":
                    elements.append(
                        abstract_syntax.Text(
                            text=element.text, attributes=element.attrib))
                elif element.tag == "stroke":
                    elements.append(
                        abstract_syntax.Stroke(
                            values=element.text, attributes=element.attrib))
                else:
                    raise Error("Unsupported tag:" + element.tag)
            elif event == "end":
                if element.tag == "layer":
                    return abstract_syntax.Layer(elements=elements)

    def parsePage(element, context):
        from decimal import Decimal
        width = Decimal(element.get("width"))
        height = Decimal(element.get("height"))
        layers = []
        for (event, element) in context:
            if event == "start":
                if element.tag == "background":
                    background = abstract_syntax.Background(
                        type=element.get("type"),
                        color=element.get("color"),
                        style=element.get("plain"))
                elif element.tag == "layer":
                    layers.append(parseLayer(context))
                else:
                    raise Error("Unsupported tag:" + element.tag)
            elif event == "end":
                if element.tag == "page":
                    return abstract_syntax.Page(
                        width=width,
                        height=height,
                        background=background,
                        layers=layers)

    def parseXournal(context):
        pages = []
        for (event, element) in context:
            if event == "start":
                if element.tag == "xournal":
                    creator=element.get("creator")
                    fileversion=int(element.get("fileversion"))
                elif element.tag == "title":
                    title = element.text
                elif element.tag == "preview":
                    import base64
                    preview = base64.b64decode(element.text)
                elif element.tag == "page":
                    pages.append(parsePage(element, context))
                else:
                    raise Error("Unsupported tag:" + element.tag)

        return abstract_syntax.XournalFile(
            creator=creator,
            fileversion=fileversion,
            title=title,
            preview=preview,
            pages=pages)

    import gzip
    with gzip.open(path, mode='rt') as f:
        from xml.etree import ElementTree
        context = ElementTree.iterparse(f, events=["start", "end"])
        return parseXournal(context)