Include analysis inputs in batch downloads

api/download.py

@@ -33,7 +33,9 @@ def _filter_check(property_filter, property_values):
 class Download(base.RequestHandler):
 
     def _append_targets(self, targets, cont_name, container, prefix, total_size, total_cnt, data_path, filters):
-        for f in container.get('files', []):
+        inputs = [('input', f) for f in container.get('inputs', [])]
+        outputs = [('output', f) for f in container.get('files', [])]
+        for file_group, f in inputs + outputs:
             if filters:
                 filtered = True
                 for filter_ in filters:
@@ -49,10 +51,9 @@ class Download(base.RequestHandler):
             filepath = os.path.join(data_path, util.path_from_hash(f['hash']))
             if os.path.exists(filepath): # silently skip missing files
                 if cont_name == 'analyses':
-                    # TODO implement analysis input downloads via generic download endpoint, too
-                    targets.append((filepath, prefix + '/output/' + f['name'], cont_name, str(container.get('_id')),f['size']))
+                    targets.append((filepath, '/'.join([prefix, file_group, f['name']]), cont_name, str(container.get('_id')), f['size']))
                 else:
-                    targets.append((filepath, prefix + '/' + f['name'], cont_name, str(container.get('_id')),f['size']))
+                    targets.append((filepath, prefix + '/' + f['name'], cont_name, str(container.get('_id')), f['size']))
                 total_size += f['size']
                 total_cnt += 1
             else:
@@ -214,7 +215,7 @@ class Download(base.RequestHandler):
                 total_size, file_cnt = self._append_targets(targets, 'acquisitions', acq, prefix, total_size, file_cnt, data_path, req_spec.get('filters'))
 
             elif item['level'] == 'analysis':
-                analysis = config.db.analyses.find_one(base_query, ['parent', 'label', 'files', 'uid', 'timestamp'])
+                analysis = config.db.analyses.find_one(base_query, ['parent', 'label', 'inputs', 'files', 'uid', 'timestamp'])
                 if not analysis:
                     # silently(while logging it) skip missing objects/objects user does not have access to
                     log.warn("Expected anaylysis {} to exist but it is missing. Node will be skipped".format(item_id))

tests/integration_tests/python/test_analyses.py

@@ -84,6 +84,29 @@ def test_legacy_analysis(data_builder, as_admin, file_form, api_db):
     api_db.analyses.delete_one({'_id': bson.ObjectId(analysis)})
 
 
+def test_analysis_download(data_builder, as_admin, file_form, api_db):
+    session = data_builder.create_session()
+
+    r = as_admin.post('/sessions/' + session + '/analyses', files=file_form('input.csv', 'output.csv', meta={
+        'label': 'legacy',
+        'inputs': [{'name': 'input.csv', 'info': {'foo': 'foo'}}],
+        'outputs': [{'name': 'output.csv', 'info': {'bar': 'bar'}}],
+    }))
+    assert r.ok
+    analysis = r.json()['_id']
+
+    # Get download ticket for analysis via /download
+    r = as_admin.get('/download', params={'ticket': ''}, json={'optional':True, 'nodes': [{'level':'analysis','_id': analysis}]})
+    assert r.ok
+    ticket = r.json()['ticket']
+
+    # Verify both inputs and outputs are present
+    r = as_admin.get('/download', params={'ticket': ticket})
+    assert r.ok
+    with tarfile.open(mode='r', fileobj=cStringIO.StringIO(r.content)) as tar:
+        assert set(m.name for m in tar.getmembers()) == set(['legacy/input/input.csv', 'legacy/output/output.csv'])
+
+
 def check_files(as_admin, analysis_id, filegroup, *filenames):
     # Verify that filegroup has all files, inflated
     r = as_admin.get('/analyses/' + analysis_id)

tests/integration_tests/python/test_download.py

@@ -451,9 +451,9 @@ def test_analysis_download(data_builder, file_form, as_admin, as_drone, default_
     r = as_admin.get('/download', params={'ticket': ticket})
     assert r.ok
 
-    # Check to make sure outputs are in tar
+    # Check to make sure inputs and outputs are in tar
     with tarfile.open(mode='r', fileobj=cStringIO.StringIO(r.content)) as tar:
-        assert [m.name for m in tar.getmembers()] == ['test/output/two.zip']
+        assert set([m.name for m in tar.getmembers()]) == set(['test/input/one.csv', 'test/output/two.zip'])
 
     # try to get download ticket for non-existent analysis file
     r = as_admin.get(analysis_inputs + '/non-existent.csv')