def main(*args):
wikipedia_site = pywikibot.Site() # Use the site configured in params/user-config
commons_site = pywikibot.Site("commons", "commons")
checker = TemplateChecker()
checker.load_config("config/templates.json")
commons_bot = CommonsBot(wikipedia_site, checker)
callbacks = ArticleIteratorCallbacks(
logging_callback=pywikibot.log,
article_callback=commons_bot.cb_check_article
)
article_iterator = ArticleIterator(callbacks)
article_iterator.log_every_n = 1
parser = ArticleIteratorArgumentParser(article_iterator, None)
run_cmd = commons_bot.run_continuously
category_name = u"Images from Wiki Loves Monuments 2015 in Germany"
start_time = first_day_of_month()
for argument in pywikibot.handle_args(args):
if argument.find("-category:") == 0:
category_name = argument[10:]
continue
elif parser.check_argument(argument):
continue
elif argument.find("-start-at:") == 0:
start_time_iso = argument[10:] + "T0:00:00Z"
start_time = pywikibot.Timestamp.fromISOformat(start_time_iso)
elif argument.find("-sleep-seconds:") == 0 and int(argument[15:]) > 0:
commons_bot.sleep_seconds = int(argument[15:])
elif argument == "-once":
run_cmd = commons_bot.run_once
elif argument == "-local-media":
commons_site = wikipedia_site
category = pywikibot.Category(commons_site, category_name)
run_cmd(article_iterator, start_time, category)
def main(*args):
site = pywikibot.Site()
fetcher = CategoryFetcher(site)
checker = TemplateChecker()
checker.load_config("config/templates.json")
checker_bot = CheckerBot(checker, site)
all_categories = fetcher.get_categories()
callbacks = ArticleIteratorCallbacks(
category_callback=checker_bot.cb_store_category_result,
article_callback=checker_bot.cb_check_article,
logging_callback=pywikibot.log,
)
article_iterator = ArticleIterator(callbacks, categories=all_categories)
parser = ArticleIteratorArgumentParser(article_iterator, fetcher)
for argument in pywikibot.handle_args(list(args)):
if parser.check_argument(argument):
continue
elif argument.find("-outputpage:") == 0:
checker_bot.outputpage = argument[12:]
elif argument.find("-exclude-articles:") == 0:
page = pywikibot.Page(site, argument[18:])
article_iterator.excluded_articles = load_excluded_articles_from_wiki(page)
article_iterator.iterate_categories()
if article_iterator.categories != all_categories: # Don't update summary page if only single categories were crawled
return
summary = checker_bot.generate_summary_page()
if checker_bot.outputpage:
checker_bot.save_wikipage(summary, checker_bot.outputpage + u"/Zusammenfassung")
else:
pywikibot.output(u"Zusammenfassung")
pywikibot.output(u"===============")
pywikibot.output(summary)
pywikibot.output(checker_bot.generate_config_table())
def main():
parser = argparse.ArgumentParser(description='Generate JSON info about monument data in wiki text.')
parser.add_argument('--monument_id', '-i', help='Unique ID of the monument. Validity will be checked.',
default='', metavar='ID')
parser.add_argument('infile', nargs='?', type=argparse.FileType('r'), default=sys.stdin)
args = parser.parse_args()
checker = TemplateChecker()
checker.load_config("config/templates.json")
mapper = CommonscatMapper()
info = get_template_info(checker, mapper, args.infile.read(), args.monument_id)
utf8_writer = codecs.getwriter('utf8')
json.dump(info, utf8_writer(sys.stdout))
def setup_instances():
g.site_commons = mwclient.Site("commons.wikimedia.org")
g.site_wikipedia = mwclient.Site("de.wikipedia.org")
g.campaign_validator = CampaignValidator(g.site_commons)
checker = TemplateChecker()
checker.load_config("config/templates.json")
mapper = CommonscatMapper()
mapper.load_mapping("config/commonscat_mapping.json")
g.page_information_collector = PageInformationCollector(checker, mapper)
if app.config["REDIS_CACHE_PREFIX"]:
g.campaign_cache = RedisCache(host=app.config["REDIS_HOST"], key_prefix=app.config["REDIS_CACHE_PREFIX"])
else:
g.campaign_cache = SimpleCache()
def test_is_allowed_template_normalizes_underscores(self):
template = Mock()
template.name = u"Denkmalliste_Sachsen_Tabellenzeile"
self.assertTrue(self.checker.is_allowed_template(template), "Template not found")
# Check if config is normalized
self.config[u"Denkmalliste_Brandenburg_Tabellenzeile"] = {
"id": "ID",
"id_check": "\\d{4,}",
"id_check_description": u"Nummer, mindestens vierstellig"
}
self.checker = TemplateChecker(self.config)
template.name = u"Denkmalliste Brandenburg Tabellenzeile"
self.assertTrue(self.checker.is_allowed_template(template), "Template not found")
def setUp(self):
self.config = {
u"Denkmalliste Sachsen Tabellenzeile": {
"id": "ID",
"id_check": "\\d{4,}",
"id_check_description": u"Nummer, mindestens vierstellig"
},
u"Denkmalliste Bayern Tabellenzeile": {
"id": "Nummer",
"id_check": "D-\\d-\\d{3}",
"id_check_description": u"Nummer im Format D-n-nnn"
}
}
self.checker = TemplateChecker(self.config)
class TestTemplateChecker(unittest.TestCase):
def setUp(self):
self.config = {
u"Denkmalliste Sachsen Tabellenzeile": {
"id": "ID",
"id_check": "\\d{4,}",
"id_check_description": u"Nummer, mindestens vierstellig"
},
u"Denkmalliste Bayern Tabellenzeile": {
"id": "Nummer",
"id_check": "D-\\d-\\d{3}",
"id_check_description": u"Nummer im Format D-n-nnn"
}
}
self.checker = TemplateChecker(self.config)
def create_article_with_text(self, text):
""" Build an Article fixture """
article = Mock()
article.get.return_value = text
article.isRedirectPage.return_value = False
return article
def test_text_contains_templates_finds_template_name(self):
text = "{{Denkmalliste Sachsen Tabellenzeile|}}"
self.assertTrue(self.checker.text_contains_templates(text))
def test_get_id_returns_id(self):
template = Mock()
template.get.return_value = u"ID=12345"
template.name = u"Denkmalliste Sachsen Tabellenzeile"
self.assertEqual(self.checker.get_id(template), u"12345")
def test_get_id_returns_empty_string_if_is_empty(self):
template = Mock()
template.get.return_value = u"ID="
template.name = u"Denkmalliste Sachsen Tabellenzeile"
self.assertEqual(self.checker.get_id(template), u"")
def test_has_valid_id_true_for_valid_ids(self):
template = Mock()
template.get.return_value = u"ID=12345"
template.name = u"Denkmalliste Sachsen Tabellenzeile"
self.assertTrue(self.checker.has_valid_id(template))
def test_has_valid_id_true_for_invalid_ids(self):
template = Mock()
template.get.return_value = u"ID=123"
template.name = u"Denkmalliste Sachsen Tabellenzeile"
self.assertFalse(self.checker.has_valid_id(template))
def test_setting_configuration_compiles_regex_patterns(self):
self.checker.config = {
"Denkmalliste Bayern Tabellenzeile": {
"id": "ID",
"id_check": "D-d{3}"
}
}
expected_class = type(re.compile("test"))
self.assertIsInstance(self.checker.config["Denkmalliste Bayern Tabellenzeile"]["id_check"], expected_class)
def test_is_allowed_template_checks_if_template_name_is_configured(self):
template = Mock()
template.name = u"Denkmalliste Sachsen Tabellenzeile"
self.assertTrue(self.checker.is_allowed_template(template))
template.name = u"Denkmalliste Kleinkleckersdorf Tabellenzeile"
self.assertFalse(self.checker.is_allowed_template(template))
def test_is_allowed_template_normalizes_underscores(self):
template = Mock()
template.name = u"Denkmalliste_Sachsen_Tabellenzeile"
self.assertTrue(self.checker.is_allowed_template(template), "Template not found")
# Check if config is normalized
self.config[u"Denkmalliste_Brandenburg_Tabellenzeile"] = {
"id": "ID",
"id_check": "\\d{4,}",
"id_check_description": u"Nummer, mindestens vierstellig"
}
self.checker = TemplateChecker(self.config)
template.name = u"Denkmalliste Brandenburg Tabellenzeile"
self.assertTrue(self.checker.is_allowed_template(template), "Template not found")
def test_check_for_errors_skips_redirect_pages(self):
article = Mock()
article.isRedirectPage.return_value = True
self.assertEqual(None, self.checker.check_article_for_errors(article))
def test_check_for_errors_reports_pages_without_templates(self):
article = self.create_article_with_text(u"Just some test text")
errors = self.checker.check_article_for_errors(article)
self.assertEqual({TemplateChecker.ERROR_MISSING_TEMPLATE: True}, errors)
def test_check_for_errors_reports_invalid_ids(self):
article = self.create_article_with_text(u"{{Denkmalliste Sachsen Tabellenzeile|ID=1}}")
errors = self.checker.check_article_for_errors(article)
self.assertEqual({TemplateChecker.ERROR_INVALID_IDS: 1}, errors)
def test_check_for_errors_returns_empty_dict_for_valid_text(self):
article = self.create_article_with_text(u"{{Denkmalliste Sachsen Tabellenzeile|ID=1234}}")
errors = self.checker.check_article_for_errors(article)
self.assertEqual({}, errors)
def test_check_for_errors_reports_duplicate_ids(self):
article = self.create_article_with_text(
u"{{Denkmalliste Sachsen Tabellenzeile|ID=1234}}{{Denkmalliste Sachsen Tabellenzeile|ID=1234}}{{Denkmalliste Sachsen Tabellenzeile|ID=1223}}")
errors = self.checker.check_article_for_errors(article)
self.assertEqual({TemplateChecker.ERROR_DUPLICATE_IDS: {u"1234": 2}}, errors)
def test_check_for_errors_reports_too_many_templates(self):
article = self.create_article_with_text(u"{{Denkmalliste Sachsen Tabellenzeile|ID=1234}}" * 600)
errors = self.checker.check_article_for_errors(article)
self.assertEqual({
TemplateChecker.ERROR_DUPLICATE_IDS: {u"1234": 600},
TemplateChecker.ERROR_TOO_MANY_TEMPLATES: 600
}, errors)
def test_check_for_errors_can_report_multiple_errors(self):
article = self.create_article_with_text(
u"{{Denkmalliste Sachsen Tabellenzeile|ID=1}}{{Denkmalliste Sachsen Tabellenzeile|ID=1}}{{Denkmalliste Sachsen Tabellenzeile|}}")
errors = self.checker.check_article_for_errors(article)
expected_errors = {
TemplateChecker.ERROR_INVALID_IDS: 2,
TemplateChecker.ERROR_MISSING_IDS: 1,
TemplateChecker.ERROR_DUPLICATE_IDS: {u"1": 2}
}
self.assertEqual(expected_errors, errors)