[Python] XML parser benchmark

bench-xml-parser.py

"""
http://www.tiger-222.fr/index.php?d=2020/05/03/12/59/13
"""
import xml.etree.ElementTree as ET
import xml.sax
import xml.sax.handler
from datetime import timedelta
from time import monotonic
from typing import Any, Dict, Generator

import xmltodict


def timer(func):
    def inner(file):
        start = monotonic()
        try:
            func(file)
        finally:
            end = monotonic()
            delta = timedelta(seconds=end - start)
            print(f"{func.__name__}(): {delta}", flush=True)
    return inner


#
# --- xmltodict
#


def handle_element(_, page: Dict[str, Any]) -> bool:
    return True


@timer
def f1(file: str) -> None:
    with open(file, "rb") as fh:
        xmltodict.parse(fh, item_depth=2, item_callback=handle_element)


#
# --- ElementTree
#


def xml_iter_parse(file: str) -> Generator[ET.Element, None, None]:
    tag_page = "{http://www.mediawiki.org/xml/export-0.10/}page"
    start_tag = None
    doc = ET.iterparse(file, events=("start", "end"))
    _, root = next(doc)

    for event, element in doc:
        if start_tag is None and event == "start" and element.tag == tag_page:
            start_tag = element.tag
        elif start_tag is not None and event == "end" and element.tag == start_tag:
            yield element
            start_tag = None
            root.clear()  # Keep memory low


def xml_parse_element(element: ET.Element) -> None:
    return None


@timer
def f2(file: str) -> None:
    for element in xml_iter_parse(file):
        xml_parse_element(element)


#
# --- SAX2
#


class PageHandler(xml.sax.handler.ContentHandler):
    __slots__ = ("_current_tag", "_is_page", "code", "word", "words")

    def __init__(self) -> None:
        self.code = ""
        self.word = ""
        self._current_tag = ""
        self._is_page = False

    def reset(self) -> None:
        self.code = ""
        self.word = ""
        self._is_page = False

    def startElement(self, tag: str, attributes: Dict[str, str]) -> None:
        if tag == "page":
            self._is_page = True
        elif tag in ("title", "text"):
            self._current_tag = tag

    def endElement(self, tag: str) -> None:
        if self._current_tag == "text":
            if self.word:
                self.process()
            self.reset()
        self._current_tag = ""

    def process(self) -> None:
        pass

    def characters(self, content: str) -> None:
        if not self._is_page:
            pass
        elif self._current_tag == "title":
            self.word = content
        elif self._current_tag == "text":
            self.code += content


@timer
def f3(file: str) -> None:
    parser = xml.sax.make_parser()
    parser.setFeature(xml.sax.handler.feature_string_interning, 1)
    handler = PageHandler()
    parser.setContentHandler(handler)
    parser.parse(file)


# https://dumps.wikimedia.org/frwiktionary/20200401/frwiktionary-20200401-pages-meta-current.xml.bz2
# 4,107,154 <page> elements => 4,107,154 words to handle
file = "data/fr/pages-20200420.xml"
f1(file)  # 0:07:27
f2(file)  # 0:02:08
f3(file)  # 0:05:46