def create_season(series_id, label, descr, ordinal, dry):
dry_str = "[DRY-RUN] " if dry else ""
repoutil = RepoUtils(Site().data_repository())
print(
f"{dry_str}Creating season with\n\tlabel='{label}'\n\tdescription='{descr}'"
)
if not dry:
season = ItemPage(repoutil.repo)
season.editLabels({"en": label}, summary="Setting label")
season.editDescriptions({"en": descr}, summary="Setting description")
print(f"Created a new Item: {season.getID()}")
print(f"{dry_str}Setting {wp.INSTANCE_OF}={wp.TELEVISION_SERIES_SEASON}")
if not dry:
instance_claim = repoutil.new_claim(wp.INSTANCE_OF.pid)
instance_claim.setTarget(
ItemPage(repoutil.repo, wp.TELEVISION_SERIES_SEASON))
season.addClaim(instance_claim,
summary=f"Setting {wp.INSTANCE_OF.pid}")
print(
f"{dry_str}Setting {wp.PART_OF_THE_SERIES}={series_id}, with {wp.SERIES_ORDINAL.pid}={ordinal}"
)
if not dry:
series_claim = repoutil.new_claim(wp.PART_OF_THE_SERIES.pid)
series_claim.setTarget(ItemPage(repoutil.repo, series_id))
season_ordinal = repoutil.new_claim(wp.SERIES_ORDINAL.pid)
season_ordinal.setTarget(str(ordinal))
series_claim.addQualifier(season_ordinal)
season.addClaim(series_claim,
summary=f"Setting {wp.PART_OF_THE_SERIES.pid}")
def copy(
self,
src_item: ItemPage,
dest_item: ItemPage,
props: Iterable[wp.WikidataProperty],
):
"""Copy properties from the source item to the destination item
Returns a tuple of (successes, failures)
"""
src_item.get()
dest_item.get()
failures = 0
successes = 0
for prop in props:
if prop.pid not in src_item.claims:
print(f"{prop} not found in {src_item.title()}")
failures += 1
continue
src_claims = src_item.claims[prop.pid]
if len(src_claims) > 1:
copy_multiple = click.confirm(
f"There are {len(src_claims)} values for {prop}. Are you sure you want to copy all of them?"
)
# copy_multiple = False
if not copy_multiple:
print(
f"Cannot copy {prop} from {format(src_item)} to {format(dest_item)}. Only scalar properties can be copied"
)
failures += 1
continue
if prop.pid in dest_item.claims:
print(f"{prop} already has a value in {format(dest_item)}")
failures += 1
continue
targets = [claim.getTarget() for claim in src_claims]
for target in targets:
if hasattr(target, "get"):
target.get()
target_str = printable_target_value(target)
print(
f"Copying {prop}={target_str} from {format(src_item)} to {format(dest_item)}"
)
new_claim = Claim(self.repo, prop.pid)
new_claim.setTarget(target)
dest_item.addClaim(
new_claim, summary=f"Setting {prop.pid} ({prop.name})"
)
successes += 1
return (successes, failures)
def normalize_repo_url(
item: ItemPage,
url_normalized: str,
url_raw: str,
q_value: str,
edit_group_hash: str,
):
""" Canonicalize the github url
This use the format https://github.com/[owner]/[repo]
Note: This apparently only works with a bot account
"""
if url_raw == url_normalized:
return
logger.info("Normalizing {} to {}".format(url_raw, url_normalized))
source_p = Properties.source_code_repository
urls = item.claims[source_p]
if source_p in item.claims and len(urls) == 2:
if urls[0].getTarget() == url_normalized and urls[1].getTarget(
) == url_raw:
logger.info(
"The old and the new url are already set, removing the old")
item.removeClaims(urls[1], summary=get_sumary(edit_group_hash))
return
if urls[0].getTarget() == url_raw and urls[1].getTarget(
) == url_normalized:
logger.info(
"The old and the new url are already set, removing the old")
item.removeClaims(urls[0], summary=get_sumary(edit_group_hash))
return
if source_p in item.claims and len(urls) > 1:
logger.info(
"Multiple source code repositories for {} not supported".format(
q_value))
return
if urls[0].getTarget() != url_raw:
logger.error(
f"The url on the object ({urls[0].getTarget()}) doesn't match the url from the sparql query ({url_raw}) for {q_value}"
)
return
# Editing is in this case actually remove the old value and adding the new one
claim = Claim(Settings.wikidata_repo, source_p)
claim.setTarget(url_normalized)
claim.setSnakType("value")
item.addClaim(claim, summary=get_sumary(edit_group_hash))
item.removeClaims(urls[0], summary=get_sumary(edit_group_hash))
# Add git as protocol
git = ItemPage(Settings.wikidata_repo, "Q186055")
get_or_create_qualifiers(claim, Properties.protocol, git, edit_group_hash)
def main(dry=False):
dry_str = ""
if dry:
print("Running in dry-run mode, will not implement any changes")
dry_str = "[DRY-RUN MODE] "
repo = Site().data_repository()
for movie_id, movie_label in movies_with_missing_titles():
print(
f"{dry_str}Setting title='{movie_label}' for {movie_id} ( https://www.wikidata.org/wiki/{movie_id} )"
)
if not dry:
movie_item = ItemPage(repo, movie_id)
movie_item.get()
claim = Claim(repo, wp.TITLE.pid)
claim.setTarget(WbMonolingualText(movie_label, "en"))
movie_item.addClaim(claim)
def get_or_create_claim(item: ItemPage, p_value: str, value: Any,
edit_group_hash: str) -> Tuple[Claim, bool]:
"""
Gets or creates a claim with `value` under the property `p_value` to `item`
"""
all_claims = item.claims.get(p_value, [])
for claim in all_claims:
if claim.target_equals(value):
return claim, False
claim = Claim(Settings.wikidata_repo, p_value)
claim.setTarget(value)
item.addClaim(claim, summary=get_sumary(edit_group_hash))
return claim, True
def create_episode(series_id, season_id, title, ordinal, dry):
dry_str = "[DRY-RUN] " if dry else ""
print(f"{dry_str}Creating episode with label='{title}'")
if not dry:
repoutil = RepoUtils(Site().data_repository())
episode = ItemPage(repoutil.repo)
episode.editLabels({"en": title}, summary="Setting label")
print(f"Created a new Item: {episode.getID()}")
print(f"{dry_str}Setting {wp.INSTANCE_OF}={wp.TELEVISION_SERIES_EPISODE}")
if not dry:
instance_claim = repoutil.new_claim(wp.INSTANCE_OF.pid)
instance_claim.setTarget(
ItemPage(repoutil.repo, wp.TELEVISION_SERIES_EPISODE))
episode.addClaim(instance_claim,
summary=f"Setting {wp.INSTANCE_OF.pid}")
print(f"{dry_str}Setting {wp.PART_OF_THE_SERIES}={series_id}")
if not dry:
series_claim = repoutil.new_claim(wp.PART_OF_THE_SERIES.pid)
series_claim.setTarget(ItemPage(repoutil.repo, series_id))
episode.addClaim(series_claim,
summary=f"Setting {wp.PART_OF_THE_SERIES.pid}")
print(
f"{dry_str}Setting {wp.SEASON}={season_id}, with {wp.SERIES_ORDINAL}={ordinal}"
)
if not dry:
season_claim = repoutil.new_claim(wp.SEASON.pid)
season_claim.setTarget(ItemPage(repoutil.repo, season_id))
season_ordinal = repoutil.new_claim(wp.SERIES_ORDINAL.pid)
season_ordinal.setTarget(ordinal)
season_claim.addQualifier(season_ordinal)
episode.addClaim(season_claim, summary=f"Setting {wp.SEASON.pid}")
def create_episode(series_id, season_id, title, series_ordinal, season_ordinal,
dry):
"""Creates a season item on WikiData
Arguments
---------
series_id: str
The Wiki ID of the series ItemPage
season_id: str
The Wiki ID of the season ItemPage
title: str
The title of this episode. This is used to set the label.
series_ordinal: int
The ordinal of this episode, within the series
season_ordinal: int
The ordinal of this episode, within the season
dry: bool
Whether or not this function should run in dry-run mode.
In dry-run mode, no real changes are made to WikiData, they are only
logged to stdout.
Returns
-------
episode_id: str
The Wiki ID of the episode item
"""
dry_str = "[DRY-RUN] " if dry else ""
print(f"{dry_str}Creating episode with label='{title}'")
episode = None
if not dry:
repoutil = RepoUtils(Site().data_repository())
season = ItemPage(repoutil.repo, season_id)
season.get()
# Check if season has part_of_the_series set to series_id
if wp.PART_OF_THE_SERIES.pid not in season.claims:
raise ValueError(
f"The season {season_id} does not have a PART_OF_THE_SERIES ({wp.PART_OF_THE_SERIES.pid} property). Check the input series and season IDs for correctness."
)
actual_series_id = str(
season.claims[wp.PART_OF_THE_SERIES.pid][0].getTarget().getID())
if actual_series_id != series_id:
raise ValueError(
f"The season {season_id} has PART_OF_THE_SERIES={actual_series_id} but expected={series_id}. Check the input series and season IDs for correctness."
)
episode = ItemPage(repoutil.repo)
episode.editLabels({"en": title}, summary="Setting label")
print(f"Created a new Item: {episode.getID()}")
print(f"{dry_str}Setting {wp.INSTANCE_OF}={wp.TELEVISION_SERIES_EPISODE}")
if not dry:
instance_claim = repoutil.new_claim(wp.INSTANCE_OF.pid)
instance_claim.setTarget(
ItemPage(repoutil.repo, wp.TELEVISION_SERIES_EPISODE))
episode.addClaim(instance_claim,
summary=f"Setting {wp.INSTANCE_OF.pid}")
print(
f"{dry_str}Setting {wp.PART_OF_THE_SERIES}={series_id}, with {wp.SERIES_ORDINAL}={series_ordinal}"
)
if not dry:
series_claim = repoutil.new_claim(wp.PART_OF_THE_SERIES.pid)
series_claim.setTarget(ItemPage(repoutil.repo, series_id))
series_ordinal_claim = repoutil.new_claim(wp.SERIES_ORDINAL.pid)
series_ordinal_claim.setTarget(series_ordinal)
series_claim.addQualifier(series_ordinal_claim)
episode.addClaim(series_claim,
summary=f"Setting {wp.PART_OF_THE_SERIES.pid}")
print(
f"{dry_str}Setting {wp.SEASON}={season_id}, with {wp.SERIES_ORDINAL}={season_ordinal}"
)
if not dry:
season_claim = repoutil.new_claim(wp.SEASON.pid)
season_claim.setTarget(ItemPage(repoutil.repo, season_id))
season_ordinal_claim = repoutil.new_claim(wp.SERIES_ORDINAL.pid)
season_ordinal_claim.setTarget(season_ordinal)
season_claim.addQualifier(season_ordinal_claim)
episode.addClaim(season_claim, summary=f"Setting {wp.SEASON.pid}")
return episode.getID() if episode is not None else "Q-1"