Merge pull request #118 from NatLibFi/EKIRJASTO-132-Fix-import-fail-with-empty-string

Ekirjasto 132 fix import fail with empty string

Author: natlibfi-kaisa

core/opds2_import.py

@@ -438,9 +438,8 @@ def _extract_contributors(
                 roles=contributor.roles if contributor.roles else default_role,
             )
             # If the feed is missing contributor name information, record the information to our metadata
-            if (
-                contributor_metadata.sort_name is None
-                and contributor_metadata.display_name is None
+            if not (
+                contributor_metadata.sort_name or contributor_metadata.display_name
             ):
                 contributor_metadata.sort_name = Edition.UNKNOWN_AUTHOR
                 contributor_metadata.display_name = Edition.UNKNOWN_AUTHOR

tests/core/files/opds2/feed.json

@@ -275,6 +275,76 @@
           "width": 800
         }
       ]
+    },
+    {
+      "metadata": {
+        "@type": "http://schema.org/Book",
+        "title": "Test Book with author empty string",
+        "author": {
+          "name": ""
+      },
+        "description": "Test Book description with author empty string",
+        "identifier": "urn:isbn:978-0-06-112008-4",
+        "language": [
+          "eng"
+        ],
+        "publisher": {
+          "name": "Test Publisher"
+        },
+        "published": "2014-09-28T00:00:00Z",
+        "modified": "2015-09-29T17:00:00Z",
+        "subject": [
+          {
+            "scheme": "http://schema.org/audience",
+            "code": "juvenile-fiction",
+            "name": "Juvenile Fiction",
+            "links": []
+          }
+        ]
+      },
+      "links": [
+        {
+          "type": "application/opds-publication+json",
+          "rel": "http://opds-spec.org/acquisition/borrow",
+          "href": "http://example.org/huckleberry-finn",
+          "properties": {
+            "availability": {
+              "state": "available"
+            },
+            "indirectAcquisition": [
+              {
+                "type": "application/vnd.adobe.adept+xml",
+                "child": [
+                  {
+                    "type": "application/epub+zip"
+                  }
+                ]
+              },
+              {
+                "type": "application/vnd.readium.lcp.license.v1.0+json",
+                "child": [
+                  {
+                    "type": "application/epub+zip"
+                  }
+                ]
+              }
+            ]
+          }
+        },
+        {
+          "rel": "http://opds-spec.org/acquisition/sample",
+          "type": "application/epub+zip",
+          "href": "https://example.com/medias/e5/318061475b11cf8c8e3752da2a1cf68384d8bf.epub"
+        }
+      ],
+      "images": [
+        {
+          "href": "http://example.org/cover.jpg",
+          "type": "image/jpeg",
+          "height": 1400,
+          "width": 800
+        }
+      ]
     }
   ]
 }

tests/core/test_opds2_import.py

@@ -123,6 +123,7 @@ class TestOPDS2Importer(OPDS2Test):
         "urn:librarysimplified.org/terms/id/ProQuest%20Doc%20ID/181639"
     )
     BOOK_WITHOUT_AUTHOR_IDENTIFIER = "urn:isbn:9789523565593"
+    BOOK_AUTHOR_EMPTY_STRING_IDENTIFIER = "urn:isbn:978-0-06-112008-4"
 
     @pytest.mark.parametrize(
         "name,manifest_type",
@@ -164,7 +165,7 @@ def test_opds2_importer_correctly_imports_valid_opds2_feed(
 
         # 1. Make sure that editions contain all required metadata
         assert isinstance(imported_editions, list)
-        assert 4 == len(imported_editions)
+        assert 5 == len(imported_editions)
 
         # 1.1. Edition with open-access links (Moby-Dick)
         moby_dick_edition = self._get_edition_by_identifier(
@@ -258,7 +259,7 @@ def test_opds2_importer_correctly_imports_valid_opds2_feed(
 
         # 2. Make sure that license pools have correct configuration
         assert isinstance(pools, list)
-        assert 4 == len(pools)
+        assert 5 == len(pools)
 
         # 2.1. Edition with open-access links (Moby-Dick)
         moby_dick_license_pool = self._get_license_pool_by_identifier(
@@ -365,7 +366,7 @@ def test_opds2_importer_correctly_imports_valid_opds2_feed(
 
         # 3. Make sure that work objects contain all the required metadata
         assert isinstance(works, list)
-        assert 4 == len(works)
+        assert 5 == len(works)
 
         # 3.1. Work (Moby-Dick)
         moby_dick_work = self._get_work_by_identifier(
@@ -390,7 +391,7 @@ def test_opds2_importer_correctly_imports_valid_opds2_feed(
             == huckleberry_finn_work.summary_text
         )
 
-        # 4.1 Author name is null
+        # 4.1 Author name is null or empty string
         edition_author_null = self._get_edition_by_identifier(
             imported_editions, self.BOOK_WITHOUT_AUTHOR_IDENTIFIER
         )
@@ -408,6 +409,12 @@ def test_opds2_importer_correctly_imports_valid_opds2_feed(
         assert isinstance(book_without_author, Work)
         assert "[Unknown]" == book_without_author.author
 
+        edition_author_empty_string = self._get_edition_by_identifier(
+            imported_editions, self.BOOK_AUTHOR_EMPTY_STRING_IDENTIFIER
+        )
+        assert isinstance(edition_author_empty_string, Edition)
+        assert "[Unknown]" == edition_author_empty_string.author
+
     @pytest.mark.parametrize(
         "this_identifier_type,ignore_identifier_type,identifier",
         [
@@ -473,7 +480,7 @@ def test_opds2_importer_skips_publications_with_unsupported_identifier_types(
         assert isinstance(imported_editions, list)
 
         if this_identifier_type == IdentifierType.ISBN:
-            assert 2 == len(imported_editions)
+            assert 3 == len(imported_editions)
             assert (
                 imported_editions[0].primary_identifier.type
                 == this_identifier_type.value