Improve the results_download API action to accept output_format #1091 (#1092)

Signed-off-by: tdruez <tdruez@nexb.com>

Author: tdruez

.github/workflows/ci.yml

@@ -38,7 +38,7 @@ jobs:
         uses: actions/checkout@v4
 
       - name: Set up Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v4
+        uses: actions/setup-python@v5
         with:
           python-version: ${{ matrix.python-version }}
 

.github/workflows/pypi-release.yml

@@ -15,7 +15,7 @@ jobs:
       - uses: actions/checkout@v4
 
       - name: Set up Python
-        uses: actions/setup-python@v4
+        uses: actions/setup-python@v5
         with:
           python-version: 3.11
 

CHANGELOG.rst

@@ -44,6 +44,10 @@ v33.2.0 (unreleased)
 - Add URL scheme validation with explicit error messages for input URLs.
   https://github.com/nexB/scancode.io/issues/1047
 
+- All supported `output_format` can now be downloaded using the results_download API
+  action providing a value for the new `output_format` parameter.
+  https://github.com/nexB/scancode.io/issues/1091
+
 - Update matchcode-toolkit to v3.0.0
 
 v33.1.0 (2024-02-02)

docs/rest-api.rst

@@ -427,12 +427,19 @@ Displays the results as JSON content compatible with ScanCode data format.
         ]
     }
 
-Results (download)
+Results (Download)
 ^^^^^^^^^^^^^^^^^^
 
-Finally, this action downloads the JSON results as an attachment.
+Finally, use this action to download the project results in the provided
+``output_format`` as an attachment file.
 
-``GET /api/projects/d4ed9405-5568-45ad-99f6-782a9b82d1d2/results_download/``
+Data:
+    - ``output_format``: ``json``, ``xlsx``, ``spdx``, ``cyclonedx``, ``attribution``
+
+``GET /api/projects/d4ed9405-5568-45ad-99f6-782a9b82d1d2/results_download/?output_format=cyclonedx``
+
+.. tip::
+  Refer to :ref:`output_files` to learn more about the available output formats.
 
 Run details
 -----------

scanpipe/api/views.py

@@ -47,18 +47,12 @@
 from scanpipe.models import Project
 from scanpipe.models import Run
 from scanpipe.models import RunInProgressError
+from scanpipe.pipes import output
 from scanpipe.views import project_results_json_response
 
 scanpipe_app = apps.get_app_config("scanpipe")
 
 
-class PassThroughRenderer(renderers.BaseRenderer):
-    media_type = ""
-
-    def render(self, data, **kwargs):
-        return data
-
-
 class ProjectFilterSet(django_filters.rest_framework.FilterSet):
     name = django_filters.CharFilter()
     name__contains = django_filters.CharFilter(
@@ -140,12 +134,32 @@ def results(self, request, *args, **kwargs):
         """
         return project_results_json_response(self.get_object())
 
-    @action(
-        detail=True, name="Results (download)", renderer_classes=[PassThroughRenderer]
-    )
+    @action(detail=True, name="Results (download)")
     def results_download(self, request, *args, **kwargs):
-        """Return the results as an attachment."""
-        return project_results_json_response(self.get_object(), as_attachment=True)
+        """Return the results in the provided `output_format` as an attachment."""
+        project = self.get_object()
+        format = request.query_params.get("output_format", "json")
+
+        if format == "json":
+            return project_results_json_response(project, as_attachment=True)
+        elif format == "xlsx":
+            output_file = output.to_xlsx(project)
+        elif format == "spdx":
+            output_file = output.to_spdx(project)
+        elif format == "cyclonedx":
+            output_file = output.to_cyclonedx(project)
+        elif format == "attribution":
+            output_file = output.to_attribution(project)
+        else:
+            message = {"status": f"Format {format} not supported."}
+            return Response(message, status=status.HTTP_400_BAD_REQUEST)
+
+        filename = output.safe_filename(f"scancodeio_{project.name}_{output_file.name}")
+        return FileResponse(
+            output_file.open("rb"),
+            filename=filename,
+            as_attachment=True,
+        )
 
     @action(detail=True)
     def summary(self, request, *args, **kwargs):

scanpipe/tests/test_api.py

@@ -53,6 +53,7 @@
 from scanpipe.pipes.input import copy_input
 from scanpipe.pipes.output import JSONResultsGenerator
 from scanpipe.tests import dependency_data1
+from scanpipe.tests import mocked_now
 from scanpipe.tests import package_data1
 
 
@@ -524,6 +525,29 @@ def test_scanpipe_api_project_action_results_download(self):
         expected = ["dependencies", "files", "headers", "packages", "relations"]
         self.assertEqual(expected, sorted(results.keys()))
 
+    @mock.patch("scanpipe.pipes.datetime", mocked_now)
+    def test_scanpipe_api_project_action_results_download_output_formats(self):
+        url = reverse("project-results-download", args=[self.project1.uuid])
+        data = {"output_format": "cyclonedx"}
+        response = self.csrf_client.get(url, data=data)
+
+        expected_filename = "scancodeio_analysis_results-2010-10-10-10-10-10.cdx.json"
+        expected = f'attachment; filename="{expected_filename}"'
+        self.assertEqual(expected, response["Content-Disposition"])
+        self.assertEqual("application/json", response["Content-Type"])
+
+        response_value = response.getvalue()
+        results = json.loads(response_value)
+        self.assertIn("$schema", sorted(results.keys()))
+
+        data = {"output_format": "xlsx"}
+        response = self.csrf_client.get(url, data=data)
+        expected = [
+            "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
+            "application/octet-stream",
+        ]
+        self.assertIn(response["Content-Type"], expected)
+
     def test_scanpipe_api_project_action_pipelines(self):
         url = reverse("project-pipelines")
         response = self.csrf_client.get(url)