bin/check-implementer-notes.py

   1 #!/usr/bin/env python
   2
   3 import json, re, subprocess, sys, urllib3
   4
   5 http = urllib3.PoolManager()
   6
   7 # TDF implementer notes pages for LibreOffice
   8 wiki_pages = [
   9     'https://wiki.documentfoundation.org/api.php?action=parse&format=json&page=Development/ODF_Implementer_Notes/List_of_LibreOffice_ODF_Extensions&prop=wikitext',
  10     'https://wiki.documentfoundation.org/api.php?action=parse&format=json&page=Development/ODF_Implementer_Notes/List_of_LibreOffice_OpenFormula_Extensions&prop=wikitext']
  11
  12 # get all commit hashes mentioned in implementer notes
  13 wiki_commit_hashes = {}
  14 query = re.compile(r'\{\{commit\|(\w+)\|\w*\|\w*\}\}', re.IGNORECASE)
  15 for page in wiki_pages:
  16     r = http.request('GET', page)
  17     data = json.loads(r.data.decode('utf-8'))
  18     for line in data['parse']['wikitext']['*'].split('\n'):
  19         for res in query.finditer(line):
  20             wiki_commit_hashes[res.group(1)] = ''
  21
  22 # get all commits that change core/schema/* - and are _not_ mentioned
  23 # in the wiki page
  24 # Cut-off is May 18th 2020, when Michael Stahl had finished cleaning this up
  25 for commit in subprocess.check_output(
  26         ['git', '--no-pager', '-C', sys.path[0]+'/..', 'log',
  27          '--since=2020-05-18', '--format=%H', '--', 'schema/'],
  28         stderr=subprocess.STDOUT).decode("utf-8").split("\n"):
  29     if commit != '' and commit not in wiki_commit_hashes:
  30         print('missing commit: %s' % commit)
  31