From 432782b33e412bad6e7722a65e3fb1db34c6b474 Mon Sep 17 00:00:00 2001
From: Rebecca Cremona <rcremona@law.harvard.edu>
Date: Tue, 3 Dec 2024 17:56:25 -0500
Subject: [PATCH] Add handling for requests to download unavilable formats.

---
 perma_web/api/tests/test_link_resource.py | 12 ++++++++++++
 perma_web/perma/utils.py                  | 19 ++++++++++++-------
 2 files changed, 24 insertions(+), 7 deletions(-)

diff --git a/perma_web/api/tests/test_link_resource.py b/perma_web/api/tests/test_link_resource.py
index c08eb04be..e08313ec0 100644
--- a/perma_web/api/tests/test_link_resource.py
+++ b/perma_web/api/tests/test_link_resource.py
@@ -185,6 +185,18 @@ def test_public_download_wacz(self, get_wacz):
     def test_public_download_unsupported_format(self):
         self.rejected_get(f"{self.public_link_download_url}?file_format=asdf", expected_status_code=400)
 
+    @patch('perma.models.Link.get_warc', autospec=True)
+    def test_download_nonexistent_warc(self, get_warc):
+        get_warc.side_effect = RuntimeError
+        self.rejected_get(self.public_link_download_url, expected_status_code=404)
+        self.assertEqual(get_warc.call_count, 1)
+
+    @patch('perma.models.Link.get_wacz', autospec=True)
+    def test_download_nonexistent_wacz(self, get_wacz):
+        get_wacz.side_effect = RuntimeError
+        self.rejected_get(f"{self.public_link_download_url}?file_format=wacz", expected_status_code=404)
+        self.assertEqual(get_wacz.call_count, 1)
+
     def test_private_download_at_public_url(self):
         self.rejected_get(self.public_link_download_url_for_private_link, expected_status_code=404)
 
diff --git a/perma_web/perma/utils.py b/perma_web/perma/utils.py
index a52754dce..ba31a4255 100644
--- a/perma_web/perma/utils.py
+++ b/perma_web/perma/utils.py
@@ -632,13 +632,18 @@ def stream_archive(link, stream=True, file_format='warc'):
     if link.user_deleted or not link.can_play_back():
         raise Http404
 
-    match file_format:
-        case 'warc':
-            return get_warc_stream(link, stream)
-        case 'wacz':
-            return get_wacz_stream(link, stream)
-        case _:
-            raise NotImplementedError("Unsupported file format.")
+    try:
+        match file_format:
+            case 'warc':
+                return get_warc_stream(link, stream)
+            case 'wacz':
+                return get_wacz_stream(link, stream)
+            case _:
+                raise NotImplementedError("Unsupported file format.")
+    except RuntimeError:
+        # If the requested format is not available, return 404
+        # just like with deleted and failed Perma Links
+        raise Http404
 
 
 def stream_archive_if_permissible(link, user, stream=True, file_format='warc'):