wip

invenio_workflows_tugraz/openaccess/convert.py

@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 #
-# Copyright (C) 2022-2023 Graz University of Technology.
+# Copyright (C) 2022-2024 Graz University of Technology.
 #
 # invenio-workflows-tugraz is free software; you can redistribute it and/or
 # modify it under the terms of the MIT License; see LICENSE file for more
@@ -43,7 +43,7 @@ def convert_attribute(
         record: Marc21Metadata,
     ) -> None:
         """Traverse first level elements of dictionary and extract attributes."""
-        convert_function = getattr(self, f"convert_{attribute}", lambda _: None)
+        convert_function = getattr(self, f"convert_{attribute}", lambda _, __: None)
         convert_function(value, record)
 
 
@@ -52,46 +52,65 @@ class Pure2Marc21(Converter):
 
     def convert_abstract(self, value: dict, record: Marc21Metadata) -> None:
         """Add the abstract to the Marc21Metadata."""
-        for abstract in value["text"]:
-            record.emplace_field("520...", value=abstract["value"])
+        try:
+            abstract = value["de_DE"]
+        except KeyError:
+            try:
+                abstract = value["en_GB"]
+            except KeyError:
+                abstract = None
+
+        if abstract:
+            record.emplace_datafield("520...", value=abstract)
 
     def convert_additionalLinks(self, value: list, record: Marc21Metadata) -> None:
         """Add the additionalLinks attribute to the Marc21Metadata."""
         for link in value:
             if "url" in link:
-                record.emplace_field("856.4.1.u", value=link["url"])
+                record.emplace_datafield("856.4.1.u", value=link["url"])
 
     def convert_bibliographicalNote(self, value: dict, record: Marc21Metadata) -> None:
         """Add the bibliographicalNote attribute to the Marc21Metadata."""
         for text in value["text"]:
-            record.emplace_field("500...", value=text["value"])
+            record.emplace_datafield("500...", value=text["value"])
 
     def convert_edition(self, value: str, record: Marc21Metadata) -> None:
         """Add the edition attribute to the Marc21Metadata."""
-        record.emplace_field("250...", value=value)
+        record.emplace_datafield("250...", value=value)
 
     def convert_electronicIsbns(self, value: list, record: Marc21Metadata) -> None:
         """Add the electronicIsbns attribute to the Marc21Metadata."""
-        record.emplace_field("020...", value=str(value[0]).strip())
+        record.emplace_datafield("020...", value=str(value[0]).strip())
 
     def convert_event(self, value: dict, record: Marc21Metadata) -> None:
         """Add the event attribute to the Marc21Metadata."""
         for event_name in value["name"]["text"]:
-            record.emplace_field("711.2..", value=event_name["value"])
+            record.emplace_datafield("711.2..", value=event_name["value"])
 
     def convert_isbns(self, value: list, record: Marc21Metadata) -> None:
         """Add the isbns attribute to the Marc21Metadata."""
         for isbn in value:
-            record.emplace_field("020...", value=isbn)
+            record.emplace_datafield("020...", value=isbn)
 
     def convert_journalAssociation(self, value: dict, record: Marc21Metadata) -> None:
         """Add the journalAssociation attribute to the Marc21Metadata."""
-        value = value["title"]["value"]
-        record.emplace_field("773.0.8.t", value=value)
+        subfs = {}
+        try:
+            subfs["t"] = value["title"]["title"]
+        except KeyError:
+            pass
+
+        try:
+            subfs["x"] = value["issn"]["issn"]
+        except KeyError:
+            pass
+
+        if subfs:
+            record.emplace_datafield("773.0.8.", subfs=subfs)
 
     def convert_journalNumber(self, value: str, record: Marc21Metadata) -> None:
         """Add the journalNumber attribute to the Marc21Metadata."""
-        record.emplace_field("773.0.8.g", value=value)
+        record.emplace_datafield("773.0.8.g", value=value)
 
     def convert_keywordGroups(self, value: list, record: Marc21Metadata) -> None:
         """Add the keywordGroups attribute to the Marc21Metadata."""
@@ -101,10 +120,9 @@ def convert_keywordGroups(self, value: list, record: Marc21Metadata) -> None:
 
     def convert_language(self, value: dict, record: Marc21Metadata) -> None:
         """Add the language attribute to the Marc21Metadata."""
-        for locale in value["term"]["text"]:
-            language = locale["value"]
-            language_iso6393 = self.languages[language]
-            record.emplace_field("041...", value=language_iso6393)
+        language = value["term"]["en_GB"]
+        language_iso6393 = self.languages[language]
+        record.emplace_datafield("041...", value=language_iso6393)
 
     def convert_managingOrganisationalUnit(
         self,
@@ -118,7 +136,7 @@ def convert_managingOrganisationalUnit(
 
     def convert_numberOfPages(self, value: int, record: Marc21Metadata) -> None:
         """Add the numberOfPages attribute to the Marc21Metadata."""
-        record.emplace_field("300...", value=str(value))
+        record.emplace_datafield("300...", value=str(value))
 
     def convert_organisationalUnits(self, value: list, record: Marc21Metadata) -> None:
         """Add the organisationalUnits attribute to the Marc21Metadata."""
@@ -130,11 +148,11 @@ def convert_organisationalUnits(self, value: list, record: Marc21Metadata) -> No
     def convert_pages(self, value: str, record: Marc21Metadata) -> None:
         """Add the pages attriute to the Marc21Metadata."""
         pages = value
-        record.emplace_field("300...", value=pages)
+        record.emplace_datafield("300...", value=pages)
 
     def convert_patentNumber(self, value: str, record: Marc21Metadata) -> None:
         """Add the patentNumber attribute to the Marc21Metadata."""
-        record.emplace_field("013...", value=value)
+        record.emplace_datafield("013...", value=value)
 
     def convert_peerReview(
         self,
@@ -144,16 +162,16 @@ def convert_peerReview(
         """Add the peerReview attribute to the Marc21Metadata."""
         if value:
             status = "Refereed/Peer-reviewed"
-            record.emplace_field("500...", value=status)
+            record.emplace_datafield("500...", value=status)
 
     def convert_placeOfPublication(self, value: str, record: Marc21Metadata) -> None:
         """Add the placeOfPublication attribute to the Marc21Metadata."""
-        record.emplace_field("264...", value=value)
+        record.emplace_datafield("264..1.", value=value)
 
     def convert_publicationSeries(self, value: list, record: Marc21Metadata) -> None:
         """Add the publicationSeries attribute to the Marc21Metadata."""
         for series in value:
-            record.emplace_field("490.0..", value=series["name"])
+            record.emplace_datafield("490.0..", value=series["name"])
 
     def convert_publicationStatuses(self, value: list, record: Marc21Metadata) -> None:
         """Add the publicationStatuses attribute to the Marc21Metadata."""
@@ -164,26 +182,26 @@ def convert_publicationStatuses(self, value: list, record: Marc21Metadata) -> No
     def convert_publisher(self, value: dict, record: Marc21Metadata) -> None:
         """Add the publisher attribute to the Marc21Metadata."""
         for text in value["name"]["text"]:
-            record.emplace_field("264...b", value=text["value"])
+            record.emplace_datafield("264..1.b", value=text["value"])
 
     def convert_relatedProjects(self, value: list, record: Marc21Metadata) -> None:
         """Add the relatedProjects attribute to the Marc21Metadata."""
         for entry in value:
             for locale in entry["name"]["text"]:
-                record.emplace_field("536...", value=locale["value"])
+                record.emplace_datafield("536...", value=locale["value"])
 
     def convert_subTitle(self, value: dict, record: Marc21Metadata) -> None:
         """Add the subTitle attribute to the Marc21Metadata."""
-        record.emplace_field("245.1.0.b", value=value["value"])
+        record.emplace_datafield("245.1.0.b", value=value["value"])
 
     def convert_title(self, value: dict, record: Marc21Metadata) -> None:
         """Add the title attribute to the Marc21Metadata."""
-        record.emplace_field("245.1.0.", value=value["value"])
+        record.emplace_datafield("245.1.0.", value=value["value"])
 
     def convert_volume(self, value: str, record: Marc21Metadata) -> None:
         """Add the volume attribute to the Marc21Metadata."""
-        record.emplace_field("490.0..", value=value)
-        record.emplace_field("773.0.8.g", value=value)
+        record.emplace_datafield("490.0..", value=value)
+        record.emplace_datafield("773.0.8.g", value=value)
 
 
 class KeywordGroup(Converter):
@@ -198,22 +216,22 @@ def convert_freeKeywords(self, value: list, record: Marc21Metadata) -> None:
         """Add free keywords."""
         for free_keyword in value:
             for word in free_keyword["freeKeywords"]:
-                record.emplace_field("650..4.g", value=word)
+                record.emplace_datafield("650..4.g", value=word)
 
     def convert_structuredKeyword(self, value: dict, record: Marc21Metadata) -> None:
         """Add free keywords."""
         for word in value["term"]["text"]:
-            record.emplace_field("650..4.", value=word["value"])
+            record.emplace_datafield("650..4.", value=word["value"])
 
 
 class PublicationStatus(Converter):
     """Class to convert publication status."""
 
     def convert_publicationDate(self, value: dict, record: Marc21Metadata) -> None:
         """Add the publication date to the Marc21Metadata."""
-        record.emplace_field("264...c", value=value["year"])
+        # without cast not serializable
+        record.emplace_datafield("264..1.c", value=str(value["year"]))
 
     def convert_publicationStatus(self, value: dict, record: Marc21Metadata) -> None:
         """Add the publication status to the Marc21Metadata."""
-        for text in value["term"]["text"]:
-            record.emplace_field("250...", value=text["value"])
+        record.emplace_datafield("250...", value=value["term"]["de_DE"])

invenio_workflows_tugraz/openaccess/utils.py

@@ -23,37 +23,62 @@ def _(value: PureId) -> None:
 def access_type(electronic_version: dict) -> str:
     """Get Access type."""
     try:
-        return electronic_version["accessType"]["term"]["text"][0]["value"]
+        return electronic_version["accessType"]["uri"]
     except (KeyError, TypeError):
         return False
 
 
 def license_type(electronic_version: dict) -> str:
     """Get license type."""
     try:
-        return electronic_version["licenseType"]["term"]["text"][0]["value"]
+        return electronic_version["licenseType"]["uri"]
     except (KeyError, TypeError):
         return False
 
 
-def extract_file_urls(pure_record) -> URL:
+def extract_files(pure_record) -> URL:
     """Extract file url."""
 
     def condition(item: dict) -> bool:
-        condition_1 = access_type(item) in ["Open", "Offen"]
-        condition_2 = license_type(item).startswith("CC BY")
+        condition_1 = (
+            access_type(item) == "/dk/atira/pure/core/openaccesspermission/open"
+        )
+        condition_2 = "cc_by" in license_type(item)
         return condition_1 and condition_2
 
-    file_urls = []
+    files = []
     for electronic_version in pure_record["electronicVersions"]:
         try:
-            file_url = electronic_version["file"]["fileURL"]
             if condition(electronic_version):
-                file_urls.append(file_url)
+                files.append(electronic_version["file"])
         except (KeyError, TypeError):
             continue
 
-    if len(file_urls) == 0:
+    if len(files) == 0:
         raise PureRuntimeError(pure_record)
 
-    return file_urls
+    return files
+
+
+def change_to_exported(pure_record: dict) -> dict:
+    """Replace the keyword group."""
+
+    # pure_record["keywordGroups"]
+    # new_keyword_group = {
+    #   "typeDiscriminator": "ClassificationsKeywordGroup",
+    #   "pureId": 69857868,
+    #   "logicalName": new_logical_name,
+    #   "name": {
+    #     "en_GB": "Export to Repository",
+    #     "de_DE": "Export ins Repository"
+    #   },
+    #   "classifications": [
+    #     {
+    #       "uri": "dk/atira/pure/researchoutput/keywords/export2repo/validated",
+    #       "term": {
+    #         "en_GB": "Validated",
+    #         "de_DE": "Validiert"
+    #       }
+    #     }
+    #   ]
+    # }

invenio_workflows_tugraz/openaccess/workflow.py

@@ -27,7 +27,7 @@
 
 from .convert import Pure2Marc21
 from .types import PureId
-from .utils import extract_file_urls
+from .utils import change_to_exported, extract_files
 
 
 def openaccess_filter() -> dict:
@@ -52,13 +52,15 @@ def import_func(
 
     try:
         pure_record = pure_service.get_metadata(identity, pure_id)
-        file_urls = extract_file_urls(pure_record)
-        file_path = pure_service.download_file(identity, file_urls)
+        files = extract_files(pure_record)
+        file_paths = []
+        for file_ in files:
+            file_paths.append(pure_service.download_file(identity, file_))
     except (PureRESTError, PureRuntimeError) as error:
         raise RuntimeError(str(error)) from error
 
     marc21_record = Marc21Metadata()
-    converter = Pure2Marc21(marc21_record)
+    converter = Pure2Marc21()
     converter.convert(pure_record, marc21_record)
 
     data = marc21_record.json
@@ -71,10 +73,12 @@ def import_func(
         record = create_record(
             marc21_service,
             data,
-            [file_path],
+            file_paths,
             identity,
             do_publish=False,
         )
+        # validate the draft here so that the record could be marked as exported
+        # because the publish will go through without problems
         marc21_service.validate_draft(
             identity,
             id_=record.id,
@@ -88,6 +92,10 @@ def import_func(
         raise RuntimeError(msg) from error
 
     try:
+        print(f"import_func pure_record: {pure_record}")
+
+        change_to_exported(pure_record)
+
         pure_service.mark_as_exported(pure_id, pure_record)
     except PureRESTError as error:
         raise RuntimeError(str(error)) from error

setup.cfg

@@ -48,6 +48,8 @@ flask.commands =
     workflows = invenio_workflows_tugraz.cli:workflows
 invenio_base.apps =
     invenio_workflows_tugraz = invenio_workflows_tugraz:InvenioWorkflowsTugraz
+invenio_base.api_apps =
+    invenio_workflows_tugraz = invenio_workflows_tugraz:InvenioWorkflowsTugraz
 invenio_base.blueprints =
     invenio_workflows_tugraz_theses = invenio_workflows_tugraz.theses:create_blueprint
 invenio_celery.tasks =

tests/conftest.py

@@ -18,6 +18,7 @@
 
 import pytest
 from flask import Flask
+from invenio_app.factory import create_api
 from invenio_db import InvenioDB
 from invenio_records_marc21 import InvenioRecordsMARC21
 from invenio_search import InvenioSearch
@@ -26,19 +27,25 @@
 
 
 @pytest.fixture(scope="module")
-def create_app(instance_path: str) -> Callable:
+def create_app(instance_path, entry_points):
     """Application factory fixture."""
+    return create_api
 
-    def factory(**config: dict) -> Flask:
-        app = Flask("testapp", instance_path=instance_path)
-        app.config.update(**config)
-        InvenioWorkflowsTugraz(app)
-        InvenioSearch(app)
-        InvenioDB(app)
-        InvenioRecordsMARC21(app)
-        return app
 
-    return factory
+# @pytest.fixture(scope="module")
+# def create_app(instance_path: str) -> Callable:
+#     """Application factory fixture."""
+
+#     def factory(**config: dict) -> Flask:
+#         app = Flask("testapp", instance_path=instance_path)
+#         app.config.update(**config)
+#         InvenioWorkflowsTugraz(app)
+#         InvenioSearch(app)
+#         InvenioDB(app)
+#         InvenioRecordsMARC21(app)
+#         return app
+
+#     return factory
 
 
 @pytest.fixture()

tests/theses/test_theses.py

@@ -21,7 +21,7 @@
 
 
 def test_update_func(
-    running_app: Flask,  # noqa: ARG001
+    app: Flask,  # noqa: ARG001
     embargoed_record_xml: str,
 ) -> None:
     """Test update func."""