Benchmark for https://github.com/earwig/mwparserfromhell/issues/195

mwph_benchmark.py

#! /usr/bin/env python3

import re
from itertools import chain

import mwparserfromhell

FLAGS = re.IGNORECASE | re.DOTALL | re.UNICODE

def indexed_ifilter(wikicode, recursive=True, matches=None, flags=FLAGS,
                    forcetype=None):
    """Iterate over nodes and their corresponding indices and parents.

    The arguments are interpreted as for :meth:`ifilter`. For each tuple
    ``(parent, i, node)`` yielded by this method, ``parent`` is the direct
    parent wikicode of ``node`` and ``parent.index(node) == i``.
    """
    match = wikicode._build_matcher(matches, flags)
    if recursive:
        restrict = forcetype if recursive == wikicode.RECURSE_OTHERS else None
        def getter(node):
            for parent, ch in wikicode._get_children(node, restrict=restrict, contexts=True, parent=wikicode):
                i = parent.index(ch)
                yield (parent, i, ch)
        inodes = chain(*(getter(n) for n in wikicode.nodes))
    else:
        inodes = ((wikicode, i, node) for i, node in enumerate(wikicode.nodes))
    for parent, i, node in inodes:
        if (not forcetype or isinstance(node, forcetype)) and match(node):
            yield (parent, i, node)

def parented_ifilter(wikicode, recursive=True, matches=None, flags=FLAGS,
                     forcetype=None):
    """Iterate over nodes and their corresponding parents.

    The arguments are interpreted as for :meth:`ifilter`. For each tuple
    ``(parent, node)`` yielded by this method, ``parent`` is the direct
    parent wikicode of ``node``.
    """
    match = wikicode._build_matcher(matches, flags)
    if recursive:
        restrict = forcetype if recursive == wikicode.RECURSE_OTHERS else None
        def getter(node):
            for parent, ch in wikicode._get_children(node, restrict=restrict, contexts=True, parent=wikicode):
                yield (parent, ch)
        inodes = chain(*(getter(n) for n in wikicode.nodes))
    else:
        inodes = ((wikicode, node) for node in wikicode.nodes)
    for parent, node in inodes:
        if (not forcetype or isinstance(node, forcetype)) and match(node):
            yield (parent, node)

@profile
def expand_1(wikicode):
    for template in wikicode.ifilter_templates(recursive=wikicode.RECURSE_OTHERS):
        if template.has(1):
            replacement = template.get(1).value
            expand_1(replacement)
        else:
            replacement = ""
        wikicode.replace(template, replacement)

@profile
def expand_2(wikicode):
    for parent, i, template in indexed_ifilter(wikicode, forcetype=mwparserfromhell.nodes.template.Template, recursive=wikicode.RECURSE_OTHERS):
        if template.has(1):
            replacement = template.get(1).value
            expand_2(replacement)
        else:
            replacement = ""
        assert parent.get(i) is template, (parent.nodes[i], i, template)
        parent.nodes.pop(i)
        parent.insert(i, replacement)

@profile
def expand_3(wikicode):
    for parent, template in parented_ifilter(wikicode, forcetype=mwparserfromhell.nodes.template.Template, recursive=wikicode.RECURSE_OTHERS):
        if template.has(1):
            replacement = template.get(1).value
            expand_3(replacement)
        else:
            replacement = ""
        parent.replace(template, replacement, recursive=False)

wikicode = mwparserfromhell.parse(open("wireless.mediawiki").read())
expand_1(wikicode)
result_1 = str(wikicode)

wikicode = mwparserfromhell.parse(open("wireless.mediawiki").read())
expand_2(wikicode)
result_2 = str(wikicode)

wikicode = mwparserfromhell.parse(open("wireless.mediawiki").read())
expand_3(wikicode)
result_3 = str(wikicode)

assert result_1 == result_2
assert result_1 == result_3


wikicode = mwparserfromhell.parse(open("wireless.mediawiki").read())
#print(wikicode)

indexed_templates = []
for parent, i, node in indexed_ifilter(wikicode, forcetype=mwparserfromhell.nodes.template.Template):
    indexed_templates.append(node)
#    if parent is wikicode:
#        print("<wikicode>", i, node)
#    else:
#        print(parent, i, node)
    assert parent.get(i) is node
assert indexed_templates == wikicode.filter_templates()

parented_templates = []
for parent, node in parented_ifilter(wikicode, forcetype=mwparserfromhell.nodes.template.Template):
    parented_templates.append(node)
#    if parent is wikicode:
#        print("<wikicode>", i, node)
#    else:
#        print(parent, i, node)
    assert parent.index(node) >= 0
assert parented_templates == wikicode.filter_templates()