How to extract data from a wikilinks?

Question

I want to extract data from the wikilinks returned by the mwparserfromhell lib. I want for instance to parse the following string:

[[File:Warszawa, ul. Freta 16 20170516 002.jpg|thumb|upright=1.18|[[Maria Skłodowska-Curie Museum|Birthplace]] of Marie Curie, at 16 Freta Street, in [[Warsaw]], [[Poland]].]]

If I split the string using the character |, it doesn't work as there is a link inside the description of the image that uses the | as well: [[Maria Skłodowska-Curie Museum|Birthplace]].

I'm using regexp to first replace all links in the string before spliting it. It works (in this case) but it doesn't feel clean (see code bellow). Is there a better way to extract information from such a string?

import re

wiki_code = "[[File:Warszawa, ul. Freta 16 20170516 002.jpg|thumb|upright=1.18|[[Maria Skłodowska-Curie Museum|Birthplace]] of Marie Curie, at 16 Freta Street, in [[Warsaw]], [[Poland]].]]"

# Remove [[File: at the begining of the string
prefix = "[[File:"
if (wiki_code.startswith(prefix)):
    wiki_code = wiki_code[len(prefix):]

# Remove ]] at the end of the string
suffix = "]]"
if (wiki_code.endswith(suffix)):
    wiki_code = wiki_code[:-len(suffix)]

# Replace links with their
link_pattern = re.compile(r'\[\[.*?\]\]')
matches = link_pattern.findall(wiki_code)
for match in matches:
    content = match[2:-2]
    arr = content.split("|")
    label = arr[-1]
    wiki_code = wiki_code.replace(match, label)

print(wiki_code.split("|"))

Answer 1

The links returned by .filter_wikilinks() are instances of the Wikilink class, which have title and text properties.

• title returns the title of the link: File:Warszawa, ul. Freta 16 20170516 002.jpg
• text returns the rest of the link: thumb|upright=1.18|[[Maria Skłodowska-Curie Museum|Birthplace]] of Marie Curie, at 16 Freta Street, in [[Warsaw]], [[Poland]].

These are returned as Wikicode objects.

Since the actual text is always the last fragment, first you need to find other fragments with the following regex:

([^\[\]|]*\|)+

• ( ): Group of
  • [^\[\]|]*: 0 or more characters that is not square brackets or pipes
  • \|: a literal pipe
• +: 1 or more

Everything else from the ending index of the last match until the end of the string is the last fragment.

>>> import mwparserfromhell
>>> import re
>>> wikitext = mwparserfromhell.parse('[[File:Warszawa, ul. Freta 16 20170516 002.jpg|thumb|upright=1.18|[[Maria Skłodowska-Curie Museum|Birthplace]] of Marie Curie, at 16 Freta Street, in [[Warsaw]], [[Poland]].]]')
>>> image_link = wikitext.filter_wikilinks()[0]
>>> image_link
'[[File:Warszawa, ul. Freta 16 20170516 002.jpg|thumb|upright=1.18|[[Maria Skłodowska-Curie Museum|Birthplace]] of Marie Curie, at 16 Freta Street, in [[Warsaw]], [[Poland]].]]'
>>> image_link.title
'File:Warszawa, ul. Freta 16 20170516 002.jpg'
>>> text = str(image_link.text)
>>> text
'thumb|upright=1.18|[[Maria Skłodowska-Curie Museum|Birthplace]] of Marie Curie, at 16 Freta Street, in [[Warsaw]], [[Poland]].'
>>> other_fragments = re.match(r'([^\[\]|]*\|)+', text)
>>> other_fragments
<re.Match object; span=(0, 19), match='thumb|upright=1.18|'>
>>> other_fragments.span(0)[1]
19
>>> text[19:]
'[[Maria Skłodowska-Curie Museum|Birthplace]] of Marie Curie, at 16 Freta Street, in [[Warsaw]], [[Poland]].'

When the caption is not the last fragment

For such edge cases we can parse the text property again using itertools functions:

>>> import mwparserfromhell
>>> import re
>>> from itertools import chain, groupby
>>> wikitext = mwparserfromhell.parse('[[File:Marie Curie - Mobile X-Ray-Unit.jpg|thumb|Curie in a mobile X-ray vehicle, {{circa|1915}}|alt=]]')
>>> image_link = wikitext.filter_wikilinks()[0]
>>> image_link.text
'thumb|Curie in a mobile X-ray vehicle, {{circa|1915}}|alt='
>>> child_nodes = image_link.text.filter(recursive = False)
>>> child_nodes
['thumb|Curie in a mobile X-ray vehicle, ', '{{circa|1915}}', '|alt=']
>>> isinstance(child_nodes[0], mwparserfromhell.nodes.Text)
True
>>> isinstance(child_nodes[1], mwparserfromhell.nodes.Template)
True
>>> tokens = list(chain.from_iterable(re.split(r'(\|)', str(node)) if isinstance(node, mwparserfromhell.nodes.Text) else [node] for node in child_nodes))
>>> tokens
['thumb', '|', 'Curie in a mobile X-ray vehicle, ', '{{circa|1915}}', '', '|', 'alt=']
>>> fragments = []
>>> for is_not_pipe, group in groupby(tokens, key = lambda token: token != '|'):
...   if is_not_pipe:
...     fragments.append(''.join(map(str, group)))
...
>>> fragments
['thumb', 'Curie in a mobile X-ray vehicle, {{circa|1915}}', 'alt=']