refactor: renames to CodeMetaSchema/DataCiteSchema

- replace "Metadata" with "Schema" for something slightly shorter and
  better reading than CodeMetaMetadata. Could also go with straight up
  DataCite and CodeMeta and have them serve as the front-end to both
  metadata worlds
- add two subtype classes to handle creating a DataCiteSchema from a
  Codebase or a CodebaseRelease. Should consider @sgfost decorators-based
  design in the future
- quiet down exceptions for degenerate codebases w/o Licenses and return
  partially consistent proxy objects if the codebase is not yet published

django/core/management/commands/discourse_sso_user_import.py

@@ -4,7 +4,6 @@
 from django.core.management.base import BaseCommand
 
 
-
 class Command(BaseCommand):
     help = "Copy CoMSES users to discourse"
 

django/curator/admin.py

@@ -1,2 +1 @@
-
 # Register your models here.

django/curator/management/commands/curator_edit_clusters.py

@@ -1,4 +1,3 @@
-
 from django.core.management.base import BaseCommand
 
 from curator.tag_deduplication import TagClusterManager

django/curator/tag_deduplication.py

@@ -11,7 +11,7 @@
 class AbstractTagDeduper(abc.ABC):
     # TRAINING_FILE = "curator/clustering_training.json"
     TRAINING_FILE = pathlib.Path("curator", "clustering_training.json")
-    FIELDS = [{"field": "name", "type": "String"}]
+    FIELDS = [dedupe.variables.String("name")]
 
     def uncertain_pairs(self):
         return self.deduper.uncertain_pairs()
@@ -222,7 +222,6 @@ def remove_training_file(self):
 class TagGazetteer(AbstractTagDeduper):
     def __init__(self, search_threshold):
         self.search_threshold = search_threshold
-
         self.deduper = dedupe.Gazetteer(AbstractTagDeduper.FIELDS)
         self.prepare_training()
         self.deduper.train()

django/library/admin.py

@@ -1,2 +1 @@
-
 # Register your models here.

django/library/doi.py

@@ -14,7 +14,7 @@
     Codebase,
     CodebaseRelease,
     DataCiteAction,
-    DataCiteMetadata,
+    DataCiteSchema,
     DataCiteRegistrationLog,
 )
 
@@ -182,7 +182,7 @@ def _datacite_heartbeat_url(self):
             else "https://api.test.datacite.org/heartbeat"
         )
 
-    def _validate_metadata(self, datacite_metadata: DataCiteMetadata):
+    def _validate_metadata(self, datacite_metadata: DataCiteSchema):
         metadata_dict = datacite_metadata.to_dict()
         if not schema43.validate(metadata_dict):
             logger.error("Invalid DataCite metadata: %s", metadata_dict)

django/library/management/commands/delete_all_existing_codebase_dois_01.py

@@ -5,6 +5,7 @@
 
 logger = logging.getLogger(__name__)
 
+
 def delete_all_existing_codebase_dois_01(interactive=True, dry_run=True):
     print(get_welcome_message(dry_run))
     codebases_with_dois = Codebase.objects.exclude(doi__isnull=True)
@@ -35,18 +36,28 @@ def delete_all_existing_codebase_dois_01(interactive=True, dry_run=True):
         print(VERIFICATION_MESSAGE)
         logger.info("Checking that all existing codebase DOIs have been deleted...")
         for i, codebase in enumerate(codebases_with_dois):
-            print(f"Processing Codebase {i}/{len(codebases_with_dois)} {'' if (i+1)%8 == 0 else '.'*((i+1)%8)}", end=" \r")
+            print(
+                f"Processing Codebase {i}/{len(codebases_with_dois)} {'' if (i+1)%8 == 0 else '.'*((i+1)%8)}",
+                end=" \r",
+            )
             if codebase.doi is not None:
                 logger.error(f"DOI for codebase {codebase.pk} should be None!")
         logger.info("Success. All existing codebase DOIs deleted.")
 
+
 class Command(BaseCommand):
 
     def add_arguments(self, parser):
-        parser.add_argument('--interactive', action='store_true', help='Wait for user to press enter to continue.')
-        parser.add_argument('--dry-run', action='store_true', help='Output what would have happened.')
+        parser.add_argument(
+            "--interactive",
+            action="store_true",
+            help="Wait for user to press enter to continue.",
+        )
+        parser.add_argument(
+            "--dry-run", action="store_true", help="Output what would have happened."
+        )
 
     def handle(self, *args, **options):
-        interactive = options['interactive']
-        dry_run = options['dry_run']
-        delete_all_existing_codebase_dois_01(interactive, dry_run)
+        interactive = options["interactive"]
+        dry_run = options["dry_run"]
+        delete_all_existing_codebase_dois_01(interactive, dry_run)

django/library/management/commands/fix_existing_dois_03.py

@@ -4,10 +4,16 @@
 from django.conf import settings
 
 from library.models import CodebaseRelease
-from library.doi import DataCiteApi, VERIFICATION_MESSAGE, doi_matches_pattern, get_welcome_message
+from library.doi import (
+    DataCiteApi,
+    VERIFICATION_MESSAGE,
+    doi_matches_pattern,
+    get_welcome_message,
+)
 
 logger = logging.getLogger(__name__)
 
+
 def fix_existing_dois_03(interactive=True, dry_run=True):
     print(get_welcome_message(dry_run))
 
@@ -123,38 +129,58 @@ def fix_existing_dois_03(interactive=True, dry_run=True):
             continue
 
     logger.info(
-        f'Successfully fixed DOIs for existing {len(peer_reviewed_releases_with_dois)} peer reviewed CodebaseReleases with DOIs and their Codebases.'
+        f"Successfully fixed DOIs for existing {len(peer_reviewed_releases_with_dois)} peer reviewed CodebaseReleases with DOIs and their Codebases."
     )
 
     """
     assert correctness
     """
     if not dry_run:
         print(VERIFICATION_MESSAGE)
-        logger.info("Checking that  all existing peer reviewed releases with DOIs (and their parent codebases) have valid DOIs...")
+        logger.info(
+            "Checking that  all existing peer reviewed releases with DOIs (and their parent codebases) have valid DOIs..."
+        )
         for i, release in enumerate(peer_reviewed_releases_with_dois):
-            print(f"Processing Codebase {i}/{len(peer_reviewed_releases_with_dois)} {'' if (i+1)%8 == 0 else '.'*((i+1)%8)}", end=" \r")
+            print(
+                f"Processing Codebase {i}/{len(peer_reviewed_releases_with_dois)} {'' if (i+1)%8 == 0 else '.'*((i+1)%8)}",
+                end=" \r",
+            )
             if release.codebase.doi is None:
-                logger.error(f"Codebase DOI should not be None for codebase {release.codebase.pk}")
+                logger.error(
+                    f"Codebase DOI should not be None for codebase {release.codebase.pk}"
+                )
 
             if release.doi is None:
                 logger.error(f"DOI should not be None for release {release.pk}")
 
             if not doi_matches_pattern(release.codebase.doi):
-                logger.error(f"{release.codebase.doi} Codebase DOI doesn't match DataCite pattern!")
+                logger.error(
+                    f"{release.codebase.doi} Codebase DOI doesn't match DataCite pattern!"
+                )
 
             if not doi_matches_pattern(release.doi):
-                logger.error(f"{release.doi} CodebaseRelease DOI doesn't match DataCite pattern!")
+                logger.error(
+                    f"{release.doi} CodebaseRelease DOI doesn't match DataCite pattern!"
+                )
+
+        logger.info(
+            "Success. All existing peer reviewed releases with DOIs (and their parent codebases) have valid DOIs now."
+        )
 
-        logger.info("Success. All existing peer reviewed releases with DOIs (and their parent codebases) have valid DOIs now.")
 
 class Command(BaseCommand):
 
     def add_arguments(self, parser):
-        parser.add_argument('--interactive', action='store_true', help='Wait for user to press enter to continue.')
-        parser.add_argument('--dry-run', action='store_true', help='Output what would have happened.')
+        parser.add_argument(
+            "--interactive",
+            action="store_true",
+            help="Wait for user to press enter to continue.",
+        )
+        parser.add_argument(
+            "--dry-run", action="store_true", help="Output what would have happened."
+        )
 
     def handle(self, *args, **options):
-        interactive = options['interactive']
-        dry_run = options['dry_run']
-        fix_existing_dois_03(interactive, dry_run)
+        interactive = options["interactive"]
+        dry_run = options["dry_run"]
+        fix_existing_dois_03(interactive, dry_run)

django/library/management/commands/mint_dois.py

@@ -1,10 +1,16 @@
 import logging
 from django.core.management.base import BaseCommand
 from library.models import CodebaseRelease
-from library.doi import DataCiteApi, VERIFICATION_MESSAGE, doi_matches_pattern, get_welcome_message
+from library.doi import (
+    DataCiteApi,
+    VERIFICATION_MESSAGE,
+    doi_matches_pattern,
+    get_welcome_message,
+)
 
 logger = logging.getLogger(__name__)
 
+
 def mint_dois_for_peer_reviewed_releases_without_dois(interactive=True, dry_run=True):
     """
     for ALL peer_reviewed releases without DOIs:
@@ -98,44 +104,68 @@ def mint_dois_for_peer_reviewed_releases_without_dois(interactive=True, dry_run=
                     f"Failed to update metadata for next_release {next_release.pk}"
                 )
 
-    logger.info(f"Minted {len(peer_reviewed_releases_without_dois)} DOIs for peer reviewed releases without DOIs.")
+    logger.info(
+        f"Minted {len(peer_reviewed_releases_without_dois)} DOIs for peer reviewed releases without DOIs."
+    )
 
     """
     assert correctness
     """
     if not dry_run:
         print(VERIFICATION_MESSAGE)
-        logger.info(f"Checking that: all peer reviewed releases (previously) without DOIs (and their parent codebases) have valid DOIs now...")
+        logger.info(
+            f"Checking that: all peer reviewed releases (previously) without DOIs (and their parent codebases) have valid DOIs now..."
+        )
         invalid_codebases = []
         invalid_releases = []
 
         for i, release in enumerate(peer_reviewed_releases_without_dois):
-            print(f"Verifying release: {i}/{len(peer_reviewed_releases_without_dois)} {'' if (i+1)%8 == 0 else '.'*((i+1)%8)}", end=" \r")
+            print(
+                f"Verifying release: {i}/{len(peer_reviewed_releases_without_dois)} {'' if (i+1)%8 == 0 else '.'*((i+1)%8)}",
+                end=" \r",
+            )
 
             if not release.doi or not doi_matches_pattern(release.doi):
                 invalid_releases.append(release.pk)
-            if not release.codebase.doi or not doi_matches_pattern(release.codebase.doi):
+            if not release.codebase.doi or not doi_matches_pattern(
+                release.codebase.doi
+            ):
                 invalid_codebases.append(release.codebase.pk)
 
         if invalid_codebases:
-            logger.error(f"Failure. Codebases with invalid or missing DOIs ({len(invalid_codebases)}): {invalid_codebases}")
+            logger.error(
+                f"Failure. Codebases with invalid or missing DOIs ({len(invalid_codebases)}): {invalid_codebases}"
+            )
         else:
-            logger.info("Success. All parent codebases for peer reviewed releases previously without DOIs have valid DOIs now.")
+            logger.info(
+                "Success. All parent codebases for peer reviewed releases previously without DOIs have valid DOIs now."
+            )
         if invalid_releases:
-            logger.error(f"Failure. CodebaseReleases with invalid or missing DOIs ({len(invalid_releases)}): {invalid_releases}")
+            logger.error(
+                f"Failure. CodebaseReleases with invalid or missing DOIs ({len(invalid_releases)}): {invalid_releases}"
+            )
         else:
-            logger.info("Success. All peer reviewed releases previously without DOIs have valid DOIs now.")
-class Command(BaseCommand):
+            logger.info(
+                "Success. All peer reviewed releases previously without DOIs have valid DOIs now."
+            )
 
+
+class Command(BaseCommand):
     """
     Mints DOIs for all peer reviewed CodebaseReleases.
     """
 
     def add_arguments(self, parser):
-        parser.add_argument('--interactive', action='store_true', help='Wait for user to press enter to continue.')
-        parser.add_argument('--dry-run', action='store_true', help='Output what would have happened.')
+        parser.add_argument(
+            "--interactive",
+            action="store_true",
+            help="Wait for user to press enter to continue.",
+        )
+        parser.add_argument(
+            "--dry-run", action="store_true", help="Output what would have happened."
+        )
 
     def handle(self, *args, **options):
-        interactive = options['interactive']
-        dry_run = options['dry_run']
-        mint_dois_for_peer_reviewed_releases_without_dois(interactive, dry_run)
+        interactive = options["interactive"]
+        dry_run = options["dry_run"]
+        mint_dois_for_peer_reviewed_releases_without_dois(interactive, dry_run)

django/library/management/commands/remove_dois_from_not_peer_reviewed_releases_02.py

@@ -5,6 +5,7 @@
 
 logger = logging.getLogger(__name__)
 
+
 def remove_dois_from_not_peer_reviewed_releases_02(interactive=True, dry_run=True):
     print(get_welcome_message(dry_run))
 
@@ -36,20 +37,35 @@ def remove_dois_from_not_peer_reviewed_releases_02(interactive=True, dry_run=Tru
     if not dry_run:
         print(VERIFICATION_MESSAGE)
         logger.info(
-            "Checking that DOIs for all not peer reviewed releases have been deleted...")
+            "Checking that DOIs for all not peer reviewed releases have been deleted..."
+        )
         for i, release in enumerate(not_peer_reviewed_releases_with_dois):
-            print(f"Processing Codebase {i}/{len(not_peer_reviewed_releases_with_dois)} {'' if (i+1)%8 == 0 else '.'*((i+1)%8)}", end=" \r")
+            print(
+                f"Processing Codebase {i}/{len(not_peer_reviewed_releases_with_dois)} {'' if (i+1)%8 == 0 else '.'*((i+1)%8)}",
+                end=" \r",
+            )
             if release.doi is not None:
-                logger.error(f"DOI for not peer reviewed release {release.pk} should be None!")
-        logger.info("Success. All existing DOIs for non peer reviewed releases have been deleted.")
+                logger.error(
+                    f"DOI for not peer reviewed release {release.pk} should be None!"
+                )
+        logger.info(
+            "Success. All existing DOIs for non peer reviewed releases have been deleted."
+        )
+
 
 class Command(BaseCommand):
 
     def add_arguments(self, parser):
-        parser.add_argument('--interactive', action='store_true', help='Wait for user to press enter to continue.')
-        parser.add_argument('--dry-run', action='store_true', help='Output what would have happened.')
+        parser.add_argument(
+            "--interactive",
+            action="store_true",
+            help="Wait for user to press enter to continue.",
+        )
+        parser.add_argument(
+            "--dry-run", action="store_true", help="Output what would have happened."
+        )
 
     def handle(self, *args, **options):
-        interactive = options['interactive']
-        dry_run = options['dry_run']
-        remove_dois_from_not_peer_reviewed_releases_02(interactive, dry_run)
+        interactive = options["interactive"]
+        dry_run = options["dry_run"]
+        remove_dois_from_not_peer_reviewed_releases_02(interactive, dry_run)