[Feature] External Identifier/References

Co-authored-by: Tim Lorsbach <tim@lorsba.ch>
Reviewed-on: enviPath/enviPy#139

epdb/management/commands/bootstrap.py

@@ -116,7 +116,7 @@ class Command(BaseCommand):
                 'full_name': 'KEGG Reaction Database',
                 'description': 'Database of biochemical reactions',
                 'base_url': 'https://www.genome.jp',
-                'url_pattern': 'https://www.genome.jp/entry/reaction+{id}'
+                'url_pattern': 'https://www.genome.jp/entry/{id}'
             },
             {
                 'name': 'UniProt',

epdb/management/commands/import_external_identifiers.py

@@ -0,0 +1,57 @@
+from csv import DictReader
+
+from django.core.management.base import BaseCommand
+
+from epdb.models import *
+
+
+class Command(BaseCommand):
+    STR_TO_MODEL = {
+        'Compound': Compound,
+        'CompoundStructure': CompoundStructure,
+        'Reaction': Reaction,
+    }
+
+    STR_TO_DATABASE = {
+        'ChEBI': ExternalDatabase.objects.get(name='ChEBI'),
+        'RHEA': ExternalDatabase.objects.get(name='RHEA'),
+        'KEGG Reaction': ExternalDatabase.objects.get(name='KEGG Reaction'),
+        'PubChem Compound': ExternalDatabase.objects.get(name='PubChem Compound'),
+        'PubChem Substance': ExternalDatabase.objects.get(name='PubChem Substance'),
+    }
+
+    def add_arguments(self, parser):
+        parser.add_argument(
+            '--data',
+            type=str,
+            help='Path of the ID Mapping file.',
+            required=True,
+        )
+
+        parser.add_argument(
+            '--replace-host',
+            type=str,
+            help='Replace https://envipath.org/ with this host, e.g. http://localhost:8000/',
+        )
+
+    @transaction.atomic
+    def handle(self, *args, **options):
+        with open(options['data']) as fh:
+            reader = DictReader(fh)
+            for row in reader:
+                clz = self.STR_TO_MODEL[row['model']]
+
+                url = row['url']
+                if options['replace_host']:
+                    url = url.replace('https://envipath.org/', options['replace_host'])
+
+                instance = clz.objects.get(url=url)
+                db = self.STR_TO_DATABASE[row['identifier_type']]
+
+                ExternalIdentifier.objects.create(
+                    content_object=instance,
+                    database=db,
+                    identifier_value=row['identifier_value'],
+                    url=db.url_pattern.format(id=row['identifier_value']),
+                    is_primary=False
+                )