Adds more author identifier fields

This commit is contained in:
Mouse Reeve 2021-04-06 17:46:06 -07:00
parent bfdcf611e7
commit 82c2f2eeb1
7 changed files with 103 additions and 51 deletions

View file

@ -150,11 +150,11 @@ class AbstractConnector(AbstractMinimalConnector):
load_more_data.delay(self.connector.id, work.id)
return edition
def get_book_data(self, remote_id):
def get_book_data(self, remote_id): # pylint: disable=no-self-use
""" this allows connectors to override the default behavior """
return get_data(remote_id)
def get_search_data(self, remote_id):
def get_search_data(self, remote_id): # pylint: disable=no-self-use
""" this allows connectors to override the default behavior """
return get_data(remote_id)
@ -293,7 +293,7 @@ class Mapping:
""" associate a local database field with a field in an external dataset """
def __init__(self, local_field, remote_field=None, formatter=None):
noop = lambda x, *_: x
noop = lambda x: x
self.local_field = local_field
self.remote_field = remote_field or local_field
@ -305,6 +305,6 @@ class Mapping:
if not value:
return None
try:
return self.formatter(value, self.remote_field)
return self.formatter(value)
except: # pylint: disable=bare-except
return None

View file

@ -10,32 +10,45 @@ class Connector(AbstractConnector):
def __init__(self, identifier):
super().__init__(identifier)
get_first = lambda a, *args: a[0]
get_remote_id_list = lambda a, *_: [self.get_remote_id(v) for v in a]
get_first = lambda a: a[0]
shared_mappings = [
Mapping("id", remote_field="uri", formatter=self.get_remote_id),
Mapping("bnfId", remote_field="wdt:P268", formatter=get_first),
Mapping("oclcNumber", remote_field="wdt:P5331", formatter=get_first),
Mapping("openlibraryKey", remote_field="wdt:P648", formatter=get_first),
]
self.book_mappings = [
Mapping("title", remote_field="wdt:P1476", formatter=get_first),
Mapping("subtitle", remote_field="wdt:P1680", formatter=get_first),
Mapping("id", remote_field="uri", formatter=self.get_remote_id),
# Mapping("authors", remote_field="wdt:P50", formatter=get_remote_id_list),
Mapping("inventaireId", remote_field="uri"),
Mapping("cover", remote_field="image", formatter=self.get_cover_url),
Mapping("isbn13", remote_field="wdt:P212", formatter=get_first),
Mapping("isbn10", remote_field="wdt:P957", formatter=get_first),
Mapping("goodreadsKey", remote_field="wdt:P2969", formatter=get_first),
Mapping("librarythingKey", remote_field="wdt:P1085", formatter=get_first),
# Mapping("languages", remote_field="wdt:P407", formatter=get_language),
# Mapping("publishers", remote_field="wdt:P123", formatter=resolve_key),
Mapping("publishedDate", remote_field="wdt:P577", formatter=get_first),
Mapping("pages", remote_field="wdt:P1104", formatter=get_first),
Mapping("goodreadsKey", remote_field="wdt:P2969", formatter=get_first),
Mapping("openlibraryKey", remote_field="wdt:P648", formatter=get_first),
# Mapping("subjectPlaces", remote_field="wdt:P840", formatter=resolve_key),
# Mapping("subjects", remote_field="wdt:P921", formatter=resolve_key),
Mapping("librarythingKey", remote_field="wdt:P1085", formatter=get_first),
Mapping("oclcNumber", remote_field="wdt:P5331", formatter=get_first),
Mapping("asin", remote_field="wdt:P5749", formatter=get_first),
]
# TODO: P136: genre, P268 bnf id, P674 characters, P950 bne
] + shared_mappings
# TODO: P136: genre, P674 characters, P950 bne
def get_remote_id(self, value, *_):
self.author_mappings = [
Mapping("id", remote_field="uri", formatter=self.get_remote_id),
Mapping("name", remote_field="label", formatter=get_language_code),
Mapping("goodreadsKey", remote_field="wdt:P2963", formatter=get_first),
Mapping("isni", remote_field="wdt:P213", formatter=get_first),
Mapping("viafId", remote_field="wdt:P214", formatter=get_first),
Mapping("gutenberg_id", remote_field="wdt:P1938", formatter=get_first),
Mapping("born", remote_field="wdt:P569", formatter=get_first),
Mapping("died", remote_field="wdt:P570", formatter=get_first),
] + shared_mappings
def get_remote_id(self, value):
""" convert an id/uri into a url """
return "{:s}?action=by-uris&uris={:s}".format(self.books_url, value)
@ -102,7 +115,9 @@ class Connector(AbstractConnector):
return self.get_book_data(self.get_remote_id(uri))
def get_authors_from_data(self, data):
return []
authors = data.get("wdt:P50")
for author in authors:
yield self.get_or_create_author(self.get_remote_id(author))
def expand_book_data(self, book):
return
@ -117,11 +132,15 @@ class Connector(AbstractConnector):
return "%s%s" % (self.covers_url, cover_id)
def get_language(wikidata_key, *_):
def get_language(wikidata_key):
""" who here speaks "wd:Q150" """
return wikidata_key # TODO
def resolve_key(wikidata_key, *_):
def resolve_key(wikidata_key):
""" cool, it's "wd:Q3156592" now what the heck does that mean """
return wikidata_key # TODO
def get_language_code(options, code="en"):
""" when there are a bunch of translation but we need a single field """
return options.get(code)

View file

@ -95,7 +95,7 @@ class Connector(AbstractConnector):
url = "%s%s" % (self.base_url, author_id)
yield self.get_or_create_author(url)
def get_cover_url(self, cover_blob, *_, size="L"):
def get_cover_url(self, cover_blob, size="L"):
""" ask openlibrary for the cover """
if not cover_blob:
return None
@ -181,19 +181,19 @@ def ignore_edition(edition_data):
return True
def get_description(description_blob, *_):
def get_description(description_blob):
""" descriptions can be a string or a dict """
if isinstance(description_blob, dict):
return description_blob.get("value")
return description_blob
def get_openlibrary_key(key, *_):
def get_openlibrary_key(key):
""" convert /books/OL27320736M into OL27320736M """
return key.split("/")[-1]
def get_languages(language_blob, *_):
def get_languages(language_blob):
""" /language/eng -> English """
langs = []
for lang in language_blob:
@ -201,7 +201,7 @@ def get_languages(language_blob, *_):
return langs
def pick_default_edition(options, *_):
def pick_default_edition(options):
""" favor physical copies with covers in english """
if not options:
return None

View file

@ -1,28 +0,0 @@
# Generated by Django 3.1.6 on 2021-04-06 17:38
import bookwyrm.models.fields
from django.db import migrations
class Migration(migrations.Migration):
dependencies = [
("bookwyrm", "0062_auto_20210406_1731"),
]
operations = [
migrations.AddField(
model_name="author",
name="inventaire_id",
field=bookwyrm.models.fields.CharField(
blank=True, max_length=255, null=True
),
),
migrations.AddField(
model_name="book",
name="inventaire_id",
field=bookwyrm.models.fields.CharField(
blank=True, max_length=255, null=True
),
),
]

View file

@ -0,0 +1,49 @@
# Generated by Django 3.1.6 on 2021-04-07 00:45
import bookwyrm.models.fields
from django.db import migrations
class Migration(migrations.Migration):
dependencies = [
('bookwyrm', '0062_auto_20210406_1731'),
]
operations = [
migrations.AddField(
model_name='author',
name='bnf_id',
field=bookwyrm.models.fields.CharField(blank=True, max_length=255, null=True),
),
migrations.AddField(
model_name='author',
name='gutenberg_id',
field=bookwyrm.models.fields.CharField(blank=True, max_length=255, null=True),
),
migrations.AddField(
model_name='author',
name='inventaire_id',
field=bookwyrm.models.fields.CharField(blank=True, max_length=255, null=True),
),
migrations.AddField(
model_name='author',
name='isni',
field=bookwyrm.models.fields.CharField(blank=True, max_length=255, null=True),
),
migrations.AddField(
model_name='author',
name='viaf_id',
field=bookwyrm.models.fields.CharField(blank=True, max_length=255, null=True),
),
migrations.AddField(
model_name='book',
name='bnf_id',
field=bookwyrm.models.fields.CharField(blank=True, max_length=255, null=True),
),
migrations.AddField(
model_name='book',
name='inventaire_id',
field=bookwyrm.models.fields.CharField(blank=True, max_length=255, null=True),
),
]

View file

@ -14,6 +14,15 @@ class Author(BookDataModel):
wikipedia_link = fields.CharField(
max_length=255, blank=True, null=True, deduplication_field=True
)
isni = fields.CharField(
max_length=255, blank=True, null=True, deduplication_field=True
)
viaf_id = fields.CharField(
max_length=255, blank=True, null=True, deduplication_field=True
)
gutenberg_id = fields.CharField(
max_length=255, blank=True, null=True, deduplication_field=True
)
# idk probably other keys would be useful here?
born = fields.DateTimeField(blank=True, null=True)
died = fields.DateTimeField(blank=True, null=True)

View file

@ -28,6 +28,9 @@ class BookDataModel(ObjectMixin, BookWyrmModel):
goodreads_key = fields.CharField(
max_length=255, blank=True, null=True, deduplication_field=True
)
bnf_id = fields.CharField( # Bibliothèque nationale de France
max_length=255, blank=True, null=True, deduplication_field=True
)
last_edited_by = models.ForeignKey("User", on_delete=models.PROTECT, null=True)