def get_catalog_from_artwork(entity):
catalog_ids = find_catalog_id(entity)
catalog_detail = []
for property_id in sorted(catalog_ids):
value = wikibase.first_datavalue(entity, property_id)
# identifier can be 'no value', example: Q26754456
if value is None:
continue
detail = lookup(property_id, value)
catalog_detail.append(detail)
catalog = {
'url': wikibase.first_datavalue(entity, 'P973'),
'detail': catalog_detail,
'ids': catalog_ids,
}
try:
check_catalog(entity, catalog)
except (requests.exceptions.ReadTimeout,
requests.exceptions.ConnectTimeout,
requests.exceptions.ConnectionError,
requests.exceptions.TooManyRedirects):
pass
return catalog
def get_institution(entity, other):
if 'P276' in entity['claims']:
location = wikibase.first_datavalue(entity, 'P276')
if location:
return other[location['id']]
if 'P195' in entity['claims']:
collection = wikibase.first_datavalue(entity, 'P195')
if collection:
return other[collection['id']]
def check_catalog(entity, catalog):
catalog_url = catalog['url']
catalog_ids = catalog['ids']
if 'P4704' in entity['claims']:
saam_id = wikibase.first_datavalue(entity, 'P4704')
cat = saam.get_catalog(saam_id)
if cat:
catalog.update(cat)
return
if 'P4709' in entity['claims']:
catalog_id = wikibase.first_datavalue(entity, 'P4709')
cat = barnesfoundation.get_catalog(catalog_id)
if cat:
catalog.update(cat)
return
institutions = [
('www.dia.org', dia),
('www.rijksmuseum.nl', rijksmuseum),
('www.npg.org.uk', npg),
('www.museodelprado.es', museodelprado),
]
if catalog_url:
for host, module in institutions:
if host in catalog_url:
cat = module.get_catalog(catalog_url)
if not cat:
continue
catalog.update(cat)
return
html = get_catalog_url(catalog_url)
description = get_description_from_page(html)
if description:
catalog['description'] = description,
return
for property_id in sorted(catalog_ids):
if property_id == 'P350':
continue # RKDimages ID
value = wikibase.first_datavalue(entity, property_id)
detail = lookup(property_id, value)
try:
html = get_catalog_page(property_id, value)
except (requests.exceptions.ConnectionError, requests.exceptions.SSLError):
continue # ignore this error
description = get_description_from_page(html)
if not description:
continue
catalog = {
'institution': detail['label'],
'description': description,
}
def add_images_to_depicts_lookup(hits):
qid_to_item = {hit['qid']: hit for hit in hits}
all_qids = [hit['qid'] for hit in hits]
entities = mediawiki.get_entities_with_cache(all_qids)
for entity in entities:
qid = entity['id']
item = qid_to_item[qid]
item.entity = entity
database.session.commit()
for hit in hits:
item = qid_to_item[hit['qid']]
if item.entity:
image_filename = wikibase.first_datavalue(item.entity, 'P18')
hit['image_filename'] = image_filename
filenames = [
hit['image_filename'] for hit in hits if hit.get('image_filename')
]
filenames = filenames[:50]
thumbwidth = 200
detail = commons.image_detail(filenames, thumbwidth=thumbwidth)
for hit in hits:
filename = hit.get('image_filename')
if not filename or filename not in detail:
continue
hit['image'] = detail[filename]
def catalog_page():
params = get_artwork_params()
bindings = filter_artwork(params)
page = utils.get_int_arg('page') or 1
page_size = 45
item_ids = set()
for row in bindings:
item_id = wdqs.row_id(row)
item_ids.add(item_id)
qids = [f'Q{item_id}' for item_id in sorted(item_ids)]
items = [Item.query.get(item_id) for item_id in item_ids]
entities = mediawiki.get_entities_with_cache(qids)
items = []
other_items = set()
for entity in entities:
other_items.update(build_other_set(entity))
continue
item = {
'label': wikibase.get_entity_label(entity),
'qid': entity['id'],
'item_id': int(entity['id'][1:]),
'image_filename': wikibase.first_datavalue(entity, 'P18'),
'entity': entity,
}
items.append(item)
other = get_labels(other_items)
flat = '_'.join(f'{pid}={qid}' for pid, qid in params)
thumbwidth = 400
# FIXME cache_name can be too long for filesystem
cache_name = f'{flat}_{page}_{page_size}_{thumbwidth}'
detail = get_image_detail_with_cache(items,
cache_name,
thumbwidth=thumbwidth)
for item in items:
item['url'] = url_for('item_page', item_id=item['item_id'])
item['image'] = detail[item['image_filename']]
item_labels = get_labels(qid for pid, qid in params)
title = ' / '.join(find_more_props[pid] + ': ' + item_labels[qid]
for pid, qid in params)
return render_template('catalog.html',
labels=find_more_props,
items=items,
other=other,
title=title)
def get_catalog_from_artwork(entity):
catalog_ids = find_catalog_id(entity)
catalog_detail = []
for property_id in sorted(catalog_ids):
value = wikibase.first_datavalue(entity, property_id)
detail = lookup(property_id, value)
catalog_detail.append(detail)
catalog = {
'url': wikibase.first_datavalue(entity, 'P973'),
'detail': catalog_detail,
'ids': catalog_ids,
}
try:
check_catalog(entity, catalog)
except (requests.exceptions.ReadTimeout,
requests.exceptions.ConnectTimeout,
requests.exceptions.ConnectionError):
pass
return catalog
def missing_image_report():
limit = utils.get_int_arg('limit') or 1000
q = DepictsItem.query.order_by(DepictsItem.count.desc()).limit(limit)
qids = [item.qid for item in q]
entities = mediawiki.get_entities_dict_with_cache(qids)
item_list = []
for depicts in q:
entity = entities[depicts.qid]
if any(
wikibase.first_datavalue(entity, prop)
for prop in ('P18', 'P2716')):
continue
item_list.append(depicts)
# TODO: call wikidata search to find images that depict item
return render_template('missing_image.html', item_list=item_list)
def next_page(item_id):
qid = f'Q{item_id}'
entity = mediawiki.get_entity_with_cache(qid)
width = 800
image_filename = wikibase.first_datavalue(entity, 'P18')
image = image_with_cache(qid, image_filename, width)
label = wikibase.get_entity_label(entity)
other = get_other(entity)
other_list = []
for key, prop_label in find_more_props.items():
if key == 'P186': # skip material used
continue # too generic
claims = entity['claims'].get(key)
if not claims:
continue
values = []
for claim in claims:
if 'datavalue' not in claim['mainsnak']:
continue
value = claim['mainsnak']['datavalue']['value']
claim_qid = value['id']
if claim_qid == 'Q4233718':
continue # anonymous artist
numeric_id = value['numeric-id']
href = url_for('find_more_page',
property_id=key[1:],
item_id=numeric_id)
values.append({
'href': href,
'qid': claim_qid,
'label': other.get(claim_qid),
})
if not values:
continue
qid_list = [v['qid'] for v in values]
other_list.append({
'label':
prop_label,
'image_lookup':
url_for('find_more_json', pid=key, qid=qid_list),
'pid':
key,
'values':
values,
'images': [],
})
return render_template('next.html',
qid=qid,
label=label,
image=image,
labels=find_more_props,
other=other,
entity=entity,
other_props=other_list)