def custom_generator(self):
query = self.store.build_query('dupes',
dupe=' wd:'.join(self.dupe_items),
offset=self.offset)
return WikidataSPARQLPageGenerator(query,
site=self.repo,
result_type=list)
def custom_generator(self):
for prop in self.opt['props']:
for key in ('duplicate_dates', 'unmerged_dates'):
query = self.store.build_query(key,
prop=prop,
limit=self.opt['limit'])
yield from WikidataSPARQLPageGenerator(query, site=self.repo)
def check_tv_show(tvshow_id=None,
child_type="all",
autofix=False,
accumulate=False,
interactive=False,
filter=""):
"""Check constraints for season/episodes of this TV show
Arguments
---------
tvshow_id: str
the Wiki ID of the television series, in the format Q######.
child_type: str
one of "episode", "season", "series", or "all"
autofix: bool
whether or not to attempt auto-fixing constraint failures
accumulate: bool
whether or not to accumulate all fixes before applying them to WikiData
interactive: bool
whether or not to prompt for confirmation before making edits
filter: str
a comma-separated list of properties in the format P###.
Only edits for these properties will be applied.
"""
if child_type == "episode":
instance_types = [wp.TELEVISION_SERIES_EPISODE]
elif child_type == "season":
instance_types = [wp.TELEVISION_SERIES_SEASON]
elif child_type == "series":
instance_types = [wp.TELEVISION_SERIES]
elif child_type == "all":
instance_types = [
wp.TELEVISION_SERIES, wp.TELEVISION_SERIES_SEASON,
wp.TELEVISION_SERIES_EPISODE
]
for instance_of_type in instance_types:
key_val_pairs = {
wp.PART_OF_THE_SERIES.pid: tvshow_id,
wp.INSTANCE_OF.pid: instance_of_type
}
query = generate_sparql_query(key_val_pairs)
gen = WikidataSPARQLPageGenerator(query)
if instance_of_type == wp.TELEVISION_SERIES:
gen = [ItemPage(Site().data_repository(), tvshow_id)]
bot = getbot(gen,
autofix=autofix,
accumulate=accumulate,
always=(not interactive),
property_filter=filter)
bot.run()
def previous_in_series(self) -> Optional[Season]:
"""The previous season from the same series"""
if self.ordinal_in_series is None:
return None
query = f"""SELECT ?item WHERE {{
?item wdt:{wp.INSTANCE_OF.pid} wd:{wp.TELEVISION_SERIES_SEASON}.
?item wdt:{wp.PART_OF_THE_SERIES.pid} wd:{self.series_qid}.
?item p:{wp.PART_OF_THE_SERIES.pid}/pq:{wp.SERIES_ORDINAL.pid} "{self.ordinal_in_series - 1}"
}}
"""
gen = WikidataSPARQLPageGenerator(query)
previous_season_itempage = next(gen, None)
if previous_season_itempage is None:
return None
return Season(previous_season_itempage)
def next_in_series(self) -> Optional[Episode]:
"""The next Episode from the same series"""
if self.ordinal_in_series is None:
return None
query = f"""SELECT ?item WHERE {{
?item wdt:{wp.INSTANCE_OF.pid} wd:{wp.TELEVISION_SERIES_EPISODE}.
?item wdt:{wp.PART_OF_THE_SERIES.pid} wd:{self.series_qid}.
?item p:{wp.PART_OF_THE_SERIES.pid}/pq:{wp.SERIES_ORDINAL.pid} "{self.ordinal_in_series + 1}"
}}
"""
gen = WikidataSPARQLPageGenerator(query)
next_episode_itempage = next(gen, None)
if next_episode_itempage is None:
return None
return Episode(next_episode_itempage)
def generator(self):
step = self.getOption('step')
opts = dict(blacklist=' wd:'.join(self.blacklist), limit=step)
offset = self.getOption('offset')
while True:
pywikibot.output('\nLoading items (offset %i)...' % offset)
opts['offset'] = offset
ask = self.store.build_query('ask_externalid_props', **opts)
if not self.sparql.ask(ask):
break
query = self.store.build_query('external-ids', **opts)
gen = PreloadingEntityGenerator(
WikidataSPARQLPageGenerator(query, site=self.repo))
for item in gen:
yield item
offset += step
def generator(self):
step = self.opt['step']
opts = {
# fixme: don't use this word
'blacklist': ' wd:'.join(self.blacklist),
'limit': step,
}
offset = self.opt['offset']
while True:
pywikibot.output('\nLoading items (offset %i)...' % offset)
opts['offset'] = offset
ask = self.store.build_query('ask_externalid_props', **opts)
if not self.sparql.ask(ask):
break
query = self.store.build_query('external-ids', **opts)
gen = PreloadingEntityGenerator(
WikidataSPARQLPageGenerator(query, site=self.repo))
yield from gen
offset += step
def previous(self) -> Optional[Season]:
"""The previous season, if any"""
# Check if it has the FOLLOWS field set
previous_season_itempage = self.first_claim(wp.FOLLOWS.pid)
if previous_season_itempage is not None:
return Season(previous_season_itempage)
# Find the item that has the FOLLOWED_BY field set to this item
query = generate_sparql_query({wp.FOLLOWED_BY.pid: self.qid})
gen = WikidataSPARQLPageGenerator(query)
follows = next(gen, None)
if follows is not None:
return Season(follows)
# Find the item whose ordinal is one lower for this series
if self.ordinal_in_series is not None:
return self.previous_in_series
return None
def next(self) -> Optional[Season]:
"""The next season, if any"""
# Check if it has the FOLLOWED_BY field set
next_season_itempage = self.first_claim(wp.FOLLOWED_BY.pid)
if next_season_itempage is not None:
return Season(next_season_itempage)
# Find the item that has the FOLLOWS field set to this item
query = generate_sparql_query({wp.FOLLOWS.pid: self.qid})
gen = WikidataSPARQLPageGenerator(query)
is_followed_by = next(gen, None)
if is_followed_by is not None:
return Season(is_followed_by)
# Find the item whose ordinal is one higher for this series
if self.ordinal_in_series is not None:
return self.next_in_series
return None
def check_tv_show(tvshow_id=None,
child_type="episode",
autofix=False,
accumulate=False,
always=False,
filter=""):
"""Check constraints for season/episodes of this TV show
TVSHOW_ID is the ID of the television series, in the format Q######.
"""
if child_type == "episode":
instance_types = [wp.TELEVISION_SERIES_EPISODE]
elif child_type == "season":
instance_types = [wp.TELEVISION_SERIES_SEASON]
elif child_type == "series":
instance_types = [wp.TELEVISION_SERIES]
elif child_type == "all":
instance_types = [
wp.TELEVISION_SERIES, wp.TELEVISION_SERIES_SEASON,
wp.TELEVISION_SERIES_EPISODE
]
for instance_of_type in instance_types:
key_val_pairs = {
wp.PART_OF_THE_SERIES.pid: tvshow_id,
wp.INSTANCE_OF.pid: instance_of_type
}
query = generate_sparql_query(key_val_pairs)
gen = WikidataSPARQLPageGenerator(query)
if instance_of_type == wp.TELEVISION_SERIES:
gen = [ItemPage(Site().data_repository(), tvshow_id)]
bot = getbot(gen,
autofix=autofix,
accumulate=accumulate,
always=always,
property_filter=filter)
bot.run()
def custom_generator(self):
query = self.store.build_query('redirects', days=self.opt['days'])
return WikidataSPARQLPageGenerator(query, site=self.repo)
def generator(self):
query = self.store.build_query('missing_descriptions',
hostname=self.site.hostname(),
lang=self.site.lang)
return PreloadingEntityGenerator(
WikidataSPARQLPageGenerator(query, site=self.repo))
def custom_generator(self):
for prop in self.props:
query = self.store.build_query('duplicate_dates', prop=prop)
for item in WikidataSPARQLPageGenerator(query, site=self.repo):
yield item