def merge_authors(ol, keys, debug=False):
# print 'merge author %s:"%s" and %s:"%s"' % (author['key'], author['name'], merge_with['key'], merge_with['name'])
# print 'becomes: "%s"' % repr(new_name)
authors = [
a for a in (withKey(k) for k in keys)
if a['type']['key'] != '/type/redirect'
]
not_redirect = set(a['key'] for a in authors)
if debug:
for a in authors:
print(a)
assert all(a['type']['key'] == '/type/author' for a in authors)
name1 = authors[0]['name']
for a in authors:
print(repr(a['key'], a['name']))
assert all(match_with_bad_chars(a['name'], name1) for a in authors[1:])
best_key = pick_best_author(authors)['key']
imgs = [
a['key'] for a in authors
if a['key'] != '/a/OL2688880A' and has_image(a['key'])
]
if len(imgs) == 1:
new_key = imgs[0]
else:
new_key = "/a/OL%dA" % min(key_int(a) for a in authors)
# Molière and O. J. O. Ferreira
if len(imgs) != 0:
print('imgs:', imgs)
return # skip
if not (imgs == [u'/a/OL21848A', u'/a/OL4280680A'] \
or imgs == [u'/a/OL325189A', u'/a/OL266422A'] \
or imgs == [u'/a/OL5160945A', u'/a/OL5776228A']):
print(imgs)
assert len(imgs) == 0
print(new_key)
print(best_key)
do_normalize(new_key, best_key, authors)
old_keys = set(k for k in keys if k != new_key)
print('old keys:', old_keys)
for old in old_keys:
# /b/OL21291659M
switch_author(ol, old, new_key, old_keys, debug=True)
if old in not_redirect:
make_redirect(ol, old, new_key)
q = {
'authors': old,
'type': '/type/edition',
}
if list(get_things(q)) != []:
switch_author(ol, old, new_key, old_keys, debug=True)
def merge_authors(ol, keys, debug=False):
# print 'merge author %s:"%s" and %s:"%s"' % (author['key'], author['name'], merge_with['key'], merge_with['name'])
# print 'becomes: "%s"' % repr(new_name)
authors = [a for a in (withKey(k) for k in keys) if a['type']['key'] != '/type/redirect']
not_redirect = set(a['key'] for a in authors)
if debug:
for a in authors:
print(a)
assert all(a['type']['key'] == '/type/author' for a in authors)
name1 = authors[0]['name']
for a in authors:
print(repr(a['key'], a['name']))
assert all(match_with_bad_chars(a['name'], name1) for a in authors[1:])
best_key = pick_best_author(authors)['key']
imgs = [a['key'] for a in authors if a['key'] != '/a/OL2688880A' and has_image(a['key'])]
if len(imgs) == 1:
new_key = imgs[0]
else:
new_key = "/a/OL%dA" % min(key_int(a) for a in authors)
# Molière and O. J. O. Ferreira
if len(imgs) != 0:
print('imgs:', imgs)
return # skip
if not (imgs == [u'/a/OL21848A', u'/a/OL4280680A'] \
or imgs == [u'/a/OL325189A', u'/a/OL266422A'] \
or imgs == [u'/a/OL5160945A', u'/a/OL5776228A']):
print(imgs)
assert len(imgs) == 0
print(new_key)
print(best_key)
do_normalize(new_key, best_key, authors)
old_keys = set(k for k in keys if k != new_key)
print('old keys:', old_keys)
for old in old_keys:
# /b/OL21291659M
switch_author(ol, old, new_key, old_keys, debug=True)
if old in not_redirect:
make_redirect(ol, old, new_key)
q = { 'authors': old, 'type': '/type/edition', }
if list(get_things(q)) != []:
switch_author(ol, old, new_key, old_keys, debug=True)
def test_pick_best_author():
a1 = {
'name': 'Bretteville, Etienne Dubois abb\xe9 de',
'death_date': '1688',
'key': '/a/OL6398452A',
'birth_date': '1650',
'title': 'abb\xe9 de',
'personal_name': 'Bretteville, Etienne Dubois',
'type': {
'key': '/type/author'
},
}
a2 = {
'name': 'Bretteville, \xc9tienne Dubois abb\xe9 de',
'death_date': '1688',
'key': '/a/OL4953701A',
'birth_date': '1650',
'title': 'abb\xe9 de',
'personal_name': 'Bretteville, \xc9tienne Dubois',
'type': {
'key': '/type/author'
},
}
assert pick_best_author([a1, a2])['key'] == a2['key']
def test_pick_best_author():
a1 = {u'name': u'Bretteville, Etienne Dubois abb\xe9 de', u'death_date': u'1688', 'key': u'/a/OL6398452A', u'birth_date': u'1650', u'title': u'abb\xe9 de', u'personal_name': u'Bretteville, Etienne Dubois', u'type': {u'key': u'/type/author'}, }
a2 = {u'name': u'Bretteville, \xc9tienne Dubois abb\xe9 de', u'death_date': u'1688', u'key': u'/a/OL4953701A', u'birth_date': u'1650', u'title': u'abb\xe9 de', u'personal_name': u'Bretteville, \xc9tienne Dubois', u'type': {u'key': u'/type/author'}, }
assert pick_best_author([a1, a2])['key'] == a2['key']