Export repository url

arkindex/documents/export/structure.sql

@@ -16,6 +16,7 @@ CREATE TABLE worker_version (
     slug VARCHAR(100) NOT NULL,
     type VARCHAR(50) NOT NULL,
     revision VARCHAR(50) NOT NULL,
+    repository_url TEXT NOT NULL,
     PRIMARY KEY (id)
 );
 

arkindex/documents/export/worker_version.sql

@@ -3,9 +3,10 @@
 -- fills up the RAM.  Adding DISTINCT to all the SELECT queries of the UNION
 -- slows this query down by ~20%.  Using multiple INs instead of a UNION makes
 -- this query twice as slow.
-SELECT version.id, worker.name, worker.slug, worker.type, revision.hash
+SELECT version.id, worker.name, worker.slug, worker.type, revision.hash, repository.url
 FROM dataimport_workerversion version
 INNER JOIN dataimport_worker worker ON (version.worker_id = worker.id)
+INNER JOIN dataimport_repository repository ON (worker.repository_id = repository.id)
 INNER JOIN dataimport_revision revision ON (version.revision_id = revision.id)
 WHERE version.id IN (
     SELECT worker_version_id FROM documents_element WHERE corpus_id = '{corpus_id}'::uuid

arkindex/documents/tests/tasks/test_export.py

@@ -137,21 +137,23 @@ class TestExport(FixtureTestCase):
         )
 
         self.assertCountEqual(
-            db.execute('SELECT id, name, slug, type, revision FROM worker_version').fetchall(),
+            db.execute('SELECT id, name, slug, type, revision, repository_url FROM worker_version').fetchall(),
             [
                 (
                     str(version.id),
                     version.worker.name,
                     version.worker.slug,
                     version.worker.type,
-                    version.revision.hash
+                    version.revision.hash,
+                    version.worker.repository.url
                 ),
                 (
                     str(metadata_version.id),
                     metadata_version.worker.name,
                     metadata_version.worker.slug,
                     metadata_version.worker.type,
-                    metadata_version.revision.hash
+                    metadata_version.revision.hash,
+                    metadata_version.worker.repository.url
                 )
             ]
         )