with open(
(create_new_path(filename)).with_suffix(f".{JSON}"),
"w",
newline="",
encoding="utf-8",
) as file:
json.dump(items, file, ensure_ascii=False)
write_state(ETL_STATES.GET_WIKIPEDIA_EXTRACTS.EXTRACT_ABSTRACTS + filename)
except Exception as error:
print(
f"Error when opening following file: {filename}. Error: {error}. Skipping file now."
)
continue
print(
datetime.datetime.now(),
"Finished extracting wikipedia extracts with",
filename,
)
if __name__ == "__main__":
if len(sys.argv) > 1 and "-r" in sys.argv:
RECOVER_MODE = True
if RECOVER_MODE and check_state(ETL_STATES.GET_WIKIPEDIA_EXTRACTS.STATE):
exit(0)
logger.info("Extracting Wikipedia Abstracts")
add_wikipedia_extracts()
write_state(ETL_STATES.GET_WIKIPEDIA_EXTRACTS.STATE)
def add_wikipedia_extracts(language_keys: Optional[List[str]] = lang_keys, ) -> None:
"""Add the wikipedia extracts to the already existing files
Args:
language_keys: Language keys to extract wikipedia abstracts for. Defaults to languageconfig.csv
"""
for filename in [
ARTWORK[PLURAL],
MOTIF[PLURAL],
GENRE[PLURAL],
MATERIAL[PLURAL],
MOVEMENT[PLURAL],
ARTIST[PLURAL],
LOCATION[PLURAL],
CLASS[PLURAL],
]:
print(
datetime.datetime.now(),
"Starting extracting wikipedia extracts with",
filename,
)
try:
with open(
(create_new_path(filename)).with_suffix(f".{JSON}"), encoding="utf-8"
) as file:
if RECOVER_MODE and check_state(ETL_STATES.GET_WIKIPEDIA_EXTRACTS.EXTRACT_ABSTRACTS + filename):
continue
items = json.load(file)
for key in language_keys:
item_indices_with_wiki_link_for_lang = [
items.index(item)
for item in items
if item[f"{WIKIPEDIA_LINK}_{key}"] != ""
]
print(
f"There are {len(item_indices_with_wiki_link_for_lang)} {key}.wikipedia links within the {len(items)} {filename} items"
)
# retry operation until its done
done = False
# ToDo: The limit for extracts seems to be 20, there is an excontinue parameter which
# could be used to increase the performance and load more at once (50 is allowed by the API) if needed
# The request method has to be adjusted for this
# Further information https://stackoverflow.com/questions/9846795/prop-extracts-not-returning-all-extracts-in-the-wikimedia-api
chunk_size = 20
while not done:
try:
item_indices_chunks = chunks(
item_indices_with_wiki_link_for_lang, chunk_size
)
extracted_count = 0
# Fill json objects without wikilink to an abstract with empty key-value pairs (could be removed if frontend is adjusted)
for j in range(len(items)):
if j not in item_indices_with_wiki_link_for_lang:
items[j][f"{ABSTRACT}_{key}"] = ""
for chunk in item_indices_chunks:
# Get PageIds from URL https://en.wikipedia.org/w/api.php?action=query&titles=Jean_Wauquelin_presenting_his_'Chroniques_de_Hainaut'_to_Philip_the_Good
page_id_indices_dictionary = get_wikipedia_page_ids(
items, chunk, key
)
# Get Extracts from PageId https://en.wikipedia.org/w/api.php?format=json&action=query&prop=extracts&exintro&explaintext&pageids=70889|1115370
raw_response = get_wikipedia_extracts(
items, page_id_indices_dictionary, key
)
# add extracted abstracts to json objects
for i in chunk:
items[i][f"{ABSTRACT}_{key}"] = raw_response[i]
extracted_count += len(chunk)
print(
f"Extracts for {filename} and language {key} status: {extracted_count}/{len(item_indices_with_wiki_link_for_lang)}",
end="\r",
flush=True,
)
# If a chunk is finished and the chunk size is < 20 (e.g. the previous chunk failed but the current one succeeded): increase the chunk size
chunk_size = chunk_size + 5 if chunk_size < 20 else chunk_size
# set done to true after all items have been processed
done = True
except Exception as error:
logger.error(f"Fetching wikipedia extracs for {filename}, lang:{key} and chunk size:{chunk_size} failed!")
logger.error(error)
# lower chunk size and try again in while loop
chunk_size -= 5
if chunk_size > 0:
logger.info(f"Trying the wikipedia extracts again with chunk size:{chunk_size}")
continue
else:
logger.exception(error)
raise error
# overwrite file
with open(
(create_new_path(filename)).with_suffix(f".{JSON}"),
"w",
newline="",
encoding="utf-8",
) as file:
json.dump(items, file, ensure_ascii=False)
write_state(ETL_STATES.GET_WIKIPEDIA_EXTRACTS.EXTRACT_ABSTRACTS + filename)
except Exception as error:
print(
f"Error when opening following file: {filename}. Error: {error}. Skipping file now."
)
continue
print(
datetime.datetime.now(),
"Finished extracting wikipedia extracts with",
filename,
)
else:
inceptions[art_mov] = {
"start": artwork["inception"],
"end": artwork["inception"],
}
return
if __name__ == "__main__":
"""Gets for all movements all first and last inceptions from the artworks file and
sets them to each movement if start_time/end_time is missing.
"""
if len(sys.argv) > 1 and "-r" in sys.argv:
RECOVER_MODE = True
if RECOVER_MODE and check_state(
ETL_STATES.DATA_TRANSFORMATION.ESTIMATE_MOVEMENT_PERIOD):
exit(0)
# start = datetime.now()
movements_file = (Path(__file__).resolve().parent.parent /
"crawler_output" / "intermediate_files" / "json" /
"movements.json")
artworks_file = (Path(__file__).resolve().parent.parent /
"crawler_output" / "intermediate_files" / "json" /
"artworks.json")
movements_output_file = movements_file # Change here for different output file
movements_list = ijson.items(open(movements_file, "r", encoding="utf-8"),
"item")
find_start_end_in_artworks(artworks_file)
for qid in movement[attribute]:
# Add the current movement id to the inverse attribute list
if (movement[ID] not in movements[qid_index_dict[qid]]
[inverse_attribute]):
movements[qid_index_dict[qid]][inverse_attribute].append(
movement[ID])
return movements
if __name__ == "__main__":
if len(sys.argv) > 1 and "-r" in sys.argv:
RECOVER_MODE = True
if RECOVER_MODE and check_state(
ETL_STATES.DATA_TRANSFORMATION.HAS_PART_PART_OF_ENHANCEMENT):
exit(0)
print("Starting part of, has part enhancement on movements",
datetime.datetime.now())
movements_file = (Path(__file__).resolve().parent.parent /
"crawler_output" / "intermediate_files" / "json" /
"movements.json")
with open(movements_file, encoding="utf-8") as file:
movements = json.load(file)
movements = inverse_attribute_enhancement(HAS_PART, PART_OF, movements)
movements = inverse_attribute_enhancement(PART_OF, HAS_PART, movements)
with open(movements_file, "w", newline="", encoding="utf-8") as file:
file.write(json.dumps(movements, ensure_ascii=False))
for entity in language_file:
if EXHIBITION_HISTORY in entity and entity[EXHIBITION_HISTORY]:
for attribute in [LABEL[SINGULAR], DESCRIPTION[SINGULAR]]:
for key in lang_keys:
for exhibition in entity[EXHIBITION_HISTORY]:
if f"{attribute}_{key}" in exhibition:
del exhibition[f"{attribute}_{key}"]
return language_file
if __name__ == "__main__":
if len(sys.argv) > 1 and "-r" in sys.argv:
RECOVER_MODE = True
if RECOVER_MODE and check_state(
ETL_STATES.DATA_TRANSFORMATION.SPLIT_LANGUAGES,
Path(__file__).parent.parent):
exit(0)
art_ontology_file = (Path(__file__).resolve().parent.parent /
CRAWLER_OUTPUT / "art_ontology.json")
print(
datetime.datetime.now(),
"Starting with splitting art_ontology.json to its language files",
)
for lang_key in language_keys:
with open(art_ontology_file, encoding="utf-8") as json_file:
art_ontology = ijson.items(json_file, 'item')
art_ontology_for_lang = []
print(f"Start generating art_ontology_{lang_key}.{JSON}")
for item in art_ontology:
def extract_art_ontology() -> None:
"""Extracts *.csv and *.json files for artworks and subjects (e. g. motifs, movements) from wikidata
"""
# Array of already crawled wikidata items
already_crawled_wikidata_items = set(BLOCKLIST)
for source in SOURCE_TYPES if not TEST_MODE else SOURCE_TYPES[:CLASS_LIM]:
if RECOVER_MODE and check_state(ETL_STATES.GET_WIKIDATA_ITEMS.EXTRACT_SOURCE + source[PLURAL]):
continue
extracted_artwork = load_wd_entities.extract_artworks(
source[PLURAL], source[ID], already_crawled_wikidata_items, DEV, DEV_CHUNK_LIMIT
)
path_name = create_new_path(ARTWORK[PLURAL], source[PLURAL], CSV)
generate_csv(extracted_artwork, get_fields(source[PLURAL]), path_name)
path_name = create_new_path(ARTWORK[PLURAL], source[PLURAL], JSON)
generate_json(extracted_artwork, path_name)
write_state(ETL_STATES.GET_WIKIDATA_ITEMS.EXTRACT_SOURCE + source[PLURAL])
if RECOVER_MODE and check_state(ETL_STATES.GET_WIKIDATA_ITEMS.MERGED_ARTWORKS):
return
merged_artworks = merge_artworks()
path_name = create_new_path(ARTWORK[PLURAL], file_type=CSV)
# Get motifs and main subjects
motifs = load_wd_entities.extract_motifs_and_main_subjects(merged_artworks)
[motif.update({TYPE: MOTIF[SINGULAR]}) for motif in motifs]
# Get extracted genres, materials, etc.
(
genres,
materials,
movements,
artists,
locations,
classes,
) = load_wd_entities.bundle_extract_subjects_calls(
[
GENRE[PLURAL],
MATERIAL[PLURAL],
MOVEMENT[PLURAL],
ARTIST[PLURAL],
LOCATION[PLURAL],
CLASS[PLURAL],
],
merged_artworks,
)
print("Total movements after transitive closure loading: ", len(movements))
for subject, type_name in [
(genres, GENRE[SINGULAR]),
(materials, MATERIAL[SINGULAR]),
(movements, MOVEMENT[SINGULAR]),
(artists, ARTIST[SINGULAR]),
(locations, LOCATION[SINGULAR]),
(classes, CLASS[SINGULAR]),
]:
[entity.update({TYPE: type_name}) for entity in subject]
# Get distinct classes from artworks, motifs, etc.
extracted_classes = load_wd_entities.get_distinct_extracted_classes(
merged_artworks, motifs, genres, materials, movements, artists, locations, classes,
)
[c.update({TYPE: CLASS[SINGULAR]}) for c in extracted_classes]
# Add the "subclass_of" parameter from the extracted_classes to the crawled classes
existing_classes = []
for class_itm in classes:
extracted_class = [d for i, d in enumerate(extracted_classes) if class_itm[ID] in d.values()]
class_itm.update({SUBCLASS_OF: extracted_class[0][SUBCLASS_OF]}) if len(extracted_class) > 0 else ""
existing_classes.append(class_itm[ID])
# append classes that are missing from our first list
for class_itm in extracted_classes:
if class_itm[ID] not in existing_classes:
classes.append(class_itm)
print("Total classes after transitive closure loading: ", len(classes))
# Get country labels for merged artworks and locations
(
locations,
merged_artworks,
movements,
) = load_wd_entities.get_country_labels_for_merged_artworks_and_locations(
locations, merged_artworks, movements
)
# Get labels for artists
artists = load_wd_entities.get_labels_for_artists(
artists, [GENDER, PLACE_OF_BIRTH, PLACE_OF_DEATH, CITIZENSHIP]
)
# Get unit symbols from qid for artworks
distinct_unit_qids = load_wd_entities.get_distinct_unit_symbol_qids(merged_artworks)
unit_symbols = load_wd_entities.get_unit_symbols(distinct_unit_qids)
load_wd_entities.resolve_unit_id_to_unit_symbol(merged_artworks, unit_symbols)
# Get exhibition histories as subdict
merged_artworks = load_wd_entities.resolve_exhibition_ids_to_exhibition_entities(
merged_artworks
)
# Significant events as subdict
merged_artworks = load_wd_entities.resolve_significant_event_id_entities_to_labels(
merged_artworks
)
# Write to JSON
write_data_to_json_and_csv(
motifs,
genres,
extracted_classes,
materials,
movements,
locations,
merged_artworks,
artists,
classes,
)
write_state(ETL_STATES.GET_WIKIDATA_ITEMS.MERGED_ARTWORKS)
classes,
)
write_state(ETL_STATES.GET_WIKIDATA_ITEMS.MERGED_ARTWORKS)
if __name__ == "__main__":
if len(sys.argv) > 1:
print(sys.argv)
dev_count_set = False
if "-d" in sys.argv:
if len(sys.argv) > sys.argv.index('-d') + 1 and any([c.isdigit() for c in sys.argv]):
DEV_CHUNK_LIMIT = int(sys.argv[sys.argv.index('-d') + 1])
dev_count_set = True
print("DEV MODE: on, DEV_LIM={0}".format(DEV_CHUNK_LIMIT))
DEV = True
if "-r" in sys.argv:
RECOVER_MODE = True
if "-t" in sys.argv:
if len(sys.argv) > sys.argv.index('-t') + 1 and sys.argv[sys.argv.index('-t') + 1].isdigit():
CLASS_LIM = int(sys.argv[sys.argv.index('-t') + 1])
print("TEST MODE: on, CLASS_LIM={0}".format(CLASS_LIM))
TEST_MODE = True
DEV = True
if not dev_count_set:
DEV_CHUNK_LIMIT = 3
if RECOVER_MODE and check_state(ETL_STATES.GET_WIKIDATA_ITEMS.STATE):
exit(0)
logger.info("Extracting Art Ontology")
extract_art_ontology()
write_state(ETL_STATES.GET_WIKIDATA_ITEMS.STATE)
entries_added_count = 0
for entity in entities:
if entity[ID] in videos:
entries_added_count += 1
entity[VIDEOS] = videos[entity[ID]]
print("Added videos for {} entries. Saving the file..".format(
entries_added_count))
return entities
if __name__ == "__main__":
if len(sys.argv) > 1 and "-r" in sys.argv:
RECOVER_MODE = True
if RECOVER_MODE and check_state(
ETL_STATES.DATA_TRANSFORMATION.ADD_YOUTUBE_VIDEOS):
exit(0)
check = "-c" in sys.argv
for entity_type in [ARTWORK[PLURAL], ARTIST[PLURAL], MOVEMENT[PLURAL]]:
print(
datetime.datetime.now(),
f"Starting with adding youtube videos for file: {entity_type}",
)
try:
# Open file
with open((create_new_path(entity_type)).with_suffix(f".{JSON}"),
encoding="utf-8") as file:
items = json.load(file)
entities = add_youtube_videos(items, check_ids=check)
return calc_relative_rank(subjects)
def calc_relative_rank(entities: List[Dict]) -> List[Dict]:
sorted_entities = sorted(entities, key=lambda k: k[ABSOLUTE_RANK])
for i, entity in enumerate(sorted_entities):
entity.update({RELATIVE_RANK: float(i / len(entities))})
return sorted_entities
if __name__ == "__main__":
if len(sys.argv) > 1 and "-r" in sys.argv:
RECOVER_MODE = True
if RECOVER_MODE and check_state(ETL_STATES.DATA_TRANSFORMATION.RANKING):
exit(0)
artworks = []
for filename in [
ARTWORK[
PLURAL], # Artworks has to be first otherwise the ranking doesn't work
MOTIF[PLURAL], # Main subjects are not considered
GENRE[PLURAL],
MATERIAL[PLURAL],
MOVEMENT[PLURAL],
ARTIST[PLURAL],
LOCATION[PLURAL],
CLASS[PLURAL],
]:
print(
datetime.datetime.now(),
