def try_edition_match(self,
work=None, title=None, author_key=None,
publisher=None, publish_year=None, id_name=None, id_value=None):
# insufficient data
if not publisher and not publish_year and not id_value:
return
q = {}
work and q.setdefault('key', work.key.split("/")[-1])
title and q.setdefault('title', title)
author_key and q.setdefault('author_key', author_key.split('/')[-1])
publisher and q.setdefault('publisher', publisher)
# There are some errors indexing of publish_year. Use publish_date until it is fixed
publish_year and q.setdefault('publish_date', publish_year)
mapping = {
'isbn_10': 'isbn',
'isbn_13': 'isbn',
'lccn': 'lccn',
'oclc_numbers': 'oclc',
'ocaid': 'ia'
}
if id_value and id_name in mapping:
if id_name.startswith('isbn'):
id_value = id_value.replace('-', '')
q[mapping[id_name]] = id_value
solr = get_works_solr()
result = solr.select(q, doc_wrapper=make_work, q_op="AND")
if len(result.docs) > 1:
return result.docs
elif len(result.docs) == 1:
# found one edition match
work = result.docs[0]
publisher = publisher and fuzzy_find(publisher, work.publisher,
stopwords=("publisher", "publishers", "and"))
editions = web.ctx.site.get_many(["/books/" + key for key in work.edition_key])
for e in editions:
d = {}
if publisher:
if not e.publishers or e.publishers[0] != publisher:
continue
if publish_year:
if not e.publish_date or publish_year != self.extract_year(e.publish_date):
continue
if id_value and id_name in mapping:
if not id_name in e or id_value not in e[id_name]:
continue
return e
score = (fuzz.ratio(w, word) + fuzz.partial_ratio(w, word)) / 2
if score > max_ratio:
max_ratio = score
ret = [(t, t + len(word))]
elif score == max_ratio:
ret.append((t, t + len(word)))
else:
pass
t += len(word)
return ret if max_ratio > 85 else []
print(
list(
fuzzy_find([
'Miami Gardens, Florida', 'WSCV', 'Hard Rock Stadium'
], r"Hard Rock Stadium is a multipurpose football stadium located in Miami Gardens, a city north of Miami. It is the home stadium of the Miami Dolphins of the National Football League (NFL)."
)))
# construct cognitive graph in training data
from utils import judge_question_type
def find_fact_content(bundle, title, sen_num):
for x in bundle['context']:
if x[0] == title:
return x[1][sen_num]
test = copy.deepcopy(train_set)
for bundle in tqdm(test):
entities = set([title for title, sen_num in bundle['supporting_facts']])
def try_edition_match(self,
work=None,
title=None,
author_key=None,
publisher=None,
publish_year=None,
id_name=None,
id_value=None):
"""
Searches solr for potential edition matches.
:param web.Storage work:
:param str title:
:param str author_key: e.g. /author/OL1234A
:param str publisher:
:param str publish_year: yyyy
:param str id_name: from list of values in mapping below
:param str id_value:
:rtype: None or Edition or list
:return: None, an Edition, or a list of Works
"""
# insufficient data
if not publisher and not publish_year and not id_value:
return
q = {}
work and q.setdefault('key', work.key.split("/")[-1])
title and q.setdefault('title', title)
author_key and q.setdefault('author_key', author_key.split('/')[-1])
publisher and q.setdefault('publisher', publisher)
# There are some errors indexing of publish_year. Use publish_date until it is fixed
publish_year and q.setdefault('publish_date', publish_year)
mapping = {
'isbn_10': 'isbn',
'isbn_13': 'isbn',
'lccn': 'lccn',
'oclc_numbers': 'oclc',
'ocaid': 'ia'
}
if id_value and id_name in mapping:
if id_name.startswith('isbn'):
id_value = id_value.replace('-', '')
q[mapping[id_name]] = id_value
solr = get_solr()
result = solr.select(q, doc_wrapper=make_work, q_op="AND")
if len(result.docs) > 1:
# found multiple work matches
return result.docs
elif len(result.docs) == 1:
# found one work match
work = result.docs[0]
publisher = publisher and fuzzy_find(
publisher,
work.publisher,
stopwords=("publisher", "publishers", "and"))
editions = web.ctx.site.get_many(
["/books/" + key for key in work.edition_key])
for e in editions:
d = {}
if publisher:
if not e.publishers or e.publishers[0] != publisher:
continue
if publish_year:
if not e.publish_date or publish_year != self.extract_year(
e.publish_date):
continue
if id_value and id_name in mapping:
if not id_name in e or id_value not in e[id_name]:
continue
# return the first good likely matching Edition
return e
