def find_entity(author):
name = author['name']
things = find_author(name)
if author['entity_type'] != 'person':
if not things:
return None
db_entity = withKey(things[0])
if db_entity['type']['key'] == '/type/redirect':
db_entity = withKey(db_entity['location'])
assert db_entity['type']['key'] == '/type/author'
return db_entity
if ', ' in name:
things += find_author(flip_name(name))
match = []
seen = set()
for key in things:
if key in seen:
continue
seen.add(key)
db_entity = withKey(key)
if db_entity['type']['key'] == '/type/redirect':
key = db_entity['location']
if key in seen:
continue
seen.add(key)
db_entity = withKey(key)
if db_entity['type']['key'] == '/type/delete':
continue
try:
assert db_entity['type']['key'] == '/type/author'
except:
print name, key, db_entity
raise
if 'birth_date' in author and 'birth_date' not in db_entity:
continue
if 'birth_date' not in author and 'birth_date' in db_entity:
continue
if not author_dates_match(author, db_entity):
continue
match.append(db_entity)
if not match:
return None
if len(match) == 1:
return match[0]
try:
return pick_from_matches(author, match)
except ValueError:
print 'author:', author
print 'match:', match
raise
def build_query(loc, rec):
if 'table_of_contents' in rec:
assert not isinstance(rec['table_of_contents'][0], list)
book = {
'type': { 'key': '/type/edition'},
}
try:
east = east_in_by_statement(rec)
except:
pprint(rec)
raise
if east:
print rec
langs = rec.get('languages', [])
print langs
if any(l['key'] == '/languages/zxx' for l in langs):
print 'zxx found in langs'
rec['languages'] = [l for l in langs if l['key'] != '/languages/zxx']
print 'fixed:', langs
for l in rec.get('languages', []):
print l
if l['key'] == '/languages/ser':
l['key'] = '/languages/srp'
if l['key'] in ('/languages/end', '/languages/enk', '/languages/ent'):
l['key'] = '/languages/eng'
if l['key'] == '/languages/cro':
l['key'] = '/languages/chu'
if l['key'] == '/languages/jap':
l['key'] = '/languages/jpn'
if l['key'] == '/languages/fra':
l['key'] = '/languages/fre'
if l['key'] == '/languages/gwr':
l['key'] = '/languages/ger'
if l['key'] == '/languages/fr ':
l['key'] = '/languages/fre'
if l['key'] == '/languages/it ':
l['key'] = '/languages/ita'
if l['key'] == '/languages/fle': # flemish -> dutch
l['key'] = '/languages/dut'
assert withKey(l['key'])
for k, v in rec.iteritems():
if k == 'authors':
book[k] = [import_author(v[0], eastern=east)]
continue
if k in type_map:
t = '/type/' + type_map[k]
if isinstance(v, list):
book[k] = [{'type': t, 'value': i} for i in v]
else:
book[k] = {'type': t, 'value': v}
else:
book[k] = v
assert 'title' in book
return book
def build_query(loc, rec):
if 'table_of_contents' in rec:
assert not isinstance(rec['table_of_contents'][0], list)
book = {
'type': { 'key': '/type/edition'},
}
try:
east = east_in_by_statement(rec)
except:
pprint(rec)
raise
if east:
print rec
langs = rec.get('languages', [])
print langs
if any(l['key'] == '/languages/zxx' for l in langs):
print 'zxx found in langs'
rec['languages'] = [l for l in langs if l['key'] != '/languages/zxx']
print 'fixed:', langs
for l in rec.get('languages', []):
assert withKey(l)
for k, v in rec.iteritems():
if k == 'authors':
book[k] = [import_author(v[0], eastern=east)]
continue
if k in type_map:
t = '/type/' + type_map[k]
if isinstance(v, list):
book[k] = [{'type': t, 'value': i} for i in v]
else:
book[k] = {'type': t, 'value': v}
else:
book[k] = v
assert 'title' in book
return book
def build_query(loc, rec):
if 'table_of_contents' in rec:
assert not isinstance(rec['table_of_contents'][0], list)
book = {
'type': {
'key': '/type/edition'
},
}
try:
east = east_in_by_statement(rec)
except:
pprint(rec)
raise
if east:
print rec
langs = rec.get('languages', [])
print langs
if any(l['key'] == '/languages/zxx' for l in langs):
print 'zxx found in langs'
rec['languages'] = [l for l in langs if l['key'] != '/languages/zxx']
print 'fixed:', langs
for l in rec.get('languages', []):
print l
if l['key'] == '/languages/ser':
l['key'] = '/languages/srp'
if l['key'] in ('/languages/end', '/languages/enk', '/languages/ent',
'/languages/enb'):
l['key'] = '/languages/eng'
if l['key'] == '/languages/emg':
l['key'] = '/languages/eng'
if l['key'] == '/languages/cro':
l['key'] = '/languages/chu'
if l['key'] == '/languages/jap':
l['key'] = '/languages/jpn'
if l['key'] == '/languages/fra':
l['key'] = '/languages/fre'
if l['key'] == '/languages/ila':
l['key'] = '/languages/ita'
if l['key'] == '/languages/gwr':
l['key'] = '/languages/ger'
if l['key'] == '/languages/fr ':
l['key'] = '/languages/fre'
if l['key'] == '/languages/it ':
l['key'] = '/languages/ita'
if l['key'] == '/languages/fle': # flemish -> dutch
l['key'] = '/languages/dut'
assert withKey(l['key'])
for k, v in rec.iteritems():
if k == 'authors':
book[k] = [import_author(v[0], eastern=east)]
continue
if k in type_map:
t = '/type/' + type_map[k]
if isinstance(v, list):
book[k] = [{'type': t, 'value': i} for i in v]
else:
book[k] = {'type': t, 'value': v}
else:
book[k] = v
assert 'title' in book
return book