Merge pull request #785 from Gluejar/doab-errata

Doab errata

core/loaders/doab.py

@@ -393,6 +393,20 @@ def add_by_doab(doab_id, record=None):
                 url_to_provider(dl_url) if dl_url else None,
                 **metadata
             )
+        else:
+            if 'format' in metadata:
+                del metadata['format']
+            edition = load_doab_edition(
+                title,
+                doab_id,
+                '',
+                '',
+                license,
+                language,
+                isbns,
+                None,
+                **metadata
+            )
         return edition
     except IdDoesNotExistError:
         return None
@@ -411,8 +425,8 @@ def load_doab_oai(from_year=None, limit=100000):
     if from_year:
         from_ = datetime.datetime(year=from_year, month=1, day=1)
     else: 
-        # last 45 days
-        from_ = datetime.datetime.now() - datetime.timedelta(days=45)
+        # last 15 days
+        from_ = datetime.datetime.now() - datetime.timedelta(days=15)
     doab_ids = []
     for record in doab_client.listRecords(metadataPrefix='oai_dc', from_=from_):
         if not record[1]:

core/loaders/doab_utils.py

@@ -103,6 +103,8 @@ FRONTIERSIN = re.compile(r'frontiersin.org/books/[^/]+/(\d+)')
 
 def online_to_download(url):
     urls = []
+    if not url:
+        return urls
     if url.find(u'mdpi.com/books/pdfview/book/') >= 0:
         doc = get_soup(url)
         if doc: