doc_status_mdn.py

'''
emulating doc status pages
'''
import urllib.request
import json
from datetime import datetime

PREFIX = 'https://developer.mozilla.org/en-US/docs/Web/'
SUFFIX = '$children?expand'
DATE_FORMAT = '%Y-%m-%dT%H:%M:%S'
SECTIONS = ['JavaScript', 'HTML', 'CSS']


def get_delta(section, target_locale):
    """ returns a list of outdated pages under a given page """
    ref_date = section['last_edit']
    ref_url = section['url']
    locale_url = ref_url
    locale_date = ref_date
    if 'translations' in section.keys():
        for translation in section['translations']:
            if translation['locale'] == target_locale:
                locale_date = translation['last_edit']
                locale_url = translation['url']
                break
    dico = dict()
    dico['refURL'] = ref_url
    dico['refDate'] = ref_date
    dico['localeURL'] = locale_url
    dico['localeDate'] = locale_date
    list_pages = []
    if 'subpages' in section.keys():
        for subpage in section['subpages']:
            list_pages.extend(get_delta(subpage, target_locale))
    # Seems like some timestamp have microseconds: stripping
    if datetime.strptime(ref_date[:19], DATE_FORMAT) > datetime.strptime(locale_date[:19], DATE_FORMAT):
        list_pages.append(dico)
    return list_pages

def get_delta_section(section, target_locale='fr'):
    """ returns a list of outdate page for a given section """
    page_json = urllib.request.urlopen(PREFIX + section + SUFFIX).read().decode('utf-8')
    root_section = json.loads(page_json)
    return get_delta(root_section, target_locale)

for sect in SECTIONS:
    delta = get_delta_section(sect)
    print(sect)
    for page in delta:
        print(page)