metadata: add codemeta

Author: tmorrell

invenio_github/api.py

@@ -25,6 +25,7 @@
 """Invenio module that adds GitHub integration to the platform."""
 
 import github3
+from convert_codemeta import crosswalk
 from flask import current_app
 from invenio_db import db
 from invenio_oauth2server.models import Token as ProviderToken
@@ -41,7 +42,8 @@
 from .errors import RepositoryAccessError
 from .models import ReleaseStatus, Repository
 from .tasks import sync_hooks
-from .utils import get_extra_metadata, iso_utcnow, parse_timestamp, utcnow
+from .utils import get_extra_metadata, iso_utcnow, \
+    parse_timestamp, utcnow
 
 
 class GitHubAPI(object):
@@ -406,14 +408,29 @@ def defaults(self):
 
     @cached_property
     def extra_metadata(self):
-        """Get extra metadata for file in repository."""
+        """Get extra metadata from file in repository."""
         return get_extra_metadata(
             self.gh.api,
             self.repository['owner']['login'],
             self.repository['name'],
             self.release['tag_name'],
+            current_app.config['GITHUB_METADATA_FILE']
         )
 
+
+    @cached_property
+    def codemeta(self):
+        """Get extra metadata from codemeta file in repository."""
+        metadata = get_extra_metadata(
+            self.gh.api,
+            self.repository['owner']['login'],
+            self.repository['name'],
+            self.release['tag_name'],
+            'codemeta.json'
+        )
+        return crosswalk(metadata, "codemeta", "Zenodo")
+
+
     @cached_property
     def files(self):
         """Extract files to download from GitHub payload."""
@@ -434,6 +451,7 @@ def metadata(self):
         """Return extracted metadata."""
         output = dict(self.defaults)
         output.update(self.extra_metadata)
+        output.update(self.codemeta)
         return output
 
     @cached_property

invenio_github/utils.py

@@ -52,19 +52,19 @@ def parse_timestamp(x):
     return dt
 
 
-def get_extra_metadata(gh, owner, repo_name, ref):
+def get_extra_metadata(gh, owner, repo_name, ref, metadata_file):
     """Get the metadata file."""
     try:
         content = gh.repository(owner, repo_name).file_contents(
-            path=current_app.config['GITHUB_METADATA_FILE'], ref=ref
+            path=metadata_file, ref=ref
         )
         if not content:
             # File does not exists in the given ref
             return {}
         return json.loads(content.decoded.decode('utf-8'))
     except ValueError:
         raise CustomGitHubMetadataError(
-            file=current_app.config['GITHUB_METADATA_FILE'])
+            file=metadata_file)
 
 
 def get_owner(gh, owner):

setup.py

@@ -69,6 +69,7 @@
     'Flask-Menu>=0.5.0',
     'Flask>=1.0.4',
     'WTForms-Alchemy>=0.15.0,<0.17',
+    'convert-codemeta>=0.2.0',
     'email-validator>=1.0.5',
     'github3.py==1.0.0a4',
     'humanize>=0.5.1',