Use author activitypub in OL connector

bookwyrm/connectors/abstract_connector.py

@@ -98,11 +98,12 @@ class AbstractConnector(AbstractMinimalConnector):
 
         # load the json
         data = get_data(remote_id)
-        mapped_data = self.dict_from_mappings(data)
+        mapped_data = dict_from_mappings(data, self.book_mappings)
         if self.is_work_data(data):
             try:
                 edition_data = self.get_edition_from_work_data(data)
-                edition_data = self.dict_from_mappings(edition_data)
+                edition_data = dict_from_mappings(\
+                        edition_data, self.book_mappings)
             except KeyError:
                 # hack: re-use the work data as the edition data
                 # this is why remote ids aren't necessarily unique
@@ -111,7 +112,7 @@ class AbstractConnector(AbstractMinimalConnector):
         else:
             try:
                 work_data = self.get_work_from_edition_data(data)
-                work_data = self.dict_from_mappings(work_data)
+                work_data = dict_from_mappings(work_data, self.book_mappings)
             except KeyError:
                 work_data = mapped_data
             edition_data = mapped_data
@@ -140,13 +141,17 @@ class AbstractConnector(AbstractMinimalConnector):
         return edition
 
 
-    def dict_from_mappings(self, data):
-        ''' create a dict in Activitypub format, using mappings supplies by
-        the subclass '''
-        result = {}
-        for mapping in self.book_mapping:
-            result[mapping.local_field] = mapping.get_value(data)
-        return result
+    def get_or_create_author(self, remote_id):
+        ''' load that author '''
+        existing = models.Author.find_exising_by_remote_id(remote_id)
+        if existing:
+            return existing
+
+        data = get_data(remote_id)
+
+        author_activity = dict_from_mappings(data, self.author_mappings)
+        # this will dedupe
+        return activitypub.Author(**author_activity).to_model()
 
 
     @abstractmethod
@@ -178,20 +183,13 @@ class AbstractConnector(AbstractMinimalConnector):
         ''' get more info on a book '''
 
 
-def get_date(date_string):
-    ''' helper function to try to interpret dates '''
-    if not date_string:
-        return None
-
-    try:
-        return pytz.utc.localize(parser.parse(date_string))
-    except ValueError:
-        pass
-
-    try:
-        return parser.parse(date_string)
-    except ValueError:
-        return None
+def dict_from_mappings(self, data, mappings):
+    ''' create a dict in Activitypub format, using mappings supplies by
+    the subclass '''
+    result = {}
+    for mapping in mappings:
+        result[mapping.local_field] = mapping.get_value(data)
+    return result
 
 
 def get_data(url):

bookwyrm/connectors/openlibrary.py

@@ -6,8 +6,8 @@ from django.core.files.base import ContentFile
 
 from bookwyrm import models
 from .abstract_connector import AbstractConnector, SearchResult, Mapping
-from .abstract_connector import ConnectorException
-from .abstract_connector import get_date, get_data, update_from_mappings
+from .abstract_connector import ConnectorException, dict_from_mappings
+from .abstract_connector import get_data, update_from_mappings
 from .openlibrary_languages import languages
 
 
@@ -51,8 +51,8 @@ class Connector(AbstractConnector):
 
         self.author_mappings = [
             Mapping('name'),
-            Mapping('born', remote_field='birth_date', formatter=get_date),
-            Mapping('died', remote_field='death_date', formatter=get_date),
+            Mapping('born', remote_field='birth_date'),
+            Mapping('died', remote_field='death_date'),
             Mapping('bio', formatter=get_description),
         ]
 
@@ -92,9 +92,10 @@ class Connector(AbstractConnector):
         ''' parse author json and load or create authors '''
         for author_blob in data.get('authors', []):
             author_blob = author_blob.get('author', author_blob)
-            # this id is "/authors/OL1234567A" and we want just "OL1234567A"
+            # this id is "/authors/OL1234567A"
             author_id = author_blob['key'].split('/')[-1]
-            yield self.get_or_create_author(author_id)
+            url = '%s/%s.json' % (self.base_url, author_id)
+            yield self.get_or_create_author(url)
 
 
     def get_cover_from_data(self, data):
@@ -158,24 +159,6 @@ class Connector(AbstractConnector):
                 edition.authors.set(work.authors.all())
 
 
-    def get_or_create_author(self, olkey):
-        ''' load that author '''
-        if not re.match(r'^OL\d+A$', olkey):
-            raise ValueError('Invalid OpenLibrary author ID')
-        author = models.Author.objects.filter(openlibrary_key=olkey).first()
-        if author:
-            return author
-
-        url = '%s/authors/%s.json' % (self.base_url, olkey)
-        data = get_data(url)
-
-        author = models.Author(openlibrary_key=olkey)
-        author = update_from_mappings(author, data, self.author_mappings)
-        author.save()
-
-        return author
-
-
 def get_description(description_blob):
     ''' descriptions can be a string or a dict '''
     if isinstance(description_blob, dict):