From 2608c953eee654c08a70a8d2263da98684d85a65 Mon Sep 17 00:00:00 2001
From: mARTin-B78 <mail@martin-bierschenk.de>
Date: Fri, 4 Sep 2026 09:34:46 +0200
Subject: [PATCH] test: cover attribution and timeout safety

---
 CHANGELOG.md               |  2 ++
 TODO.md                    |  4 ++--
 VERSION                    |  2 +-
 tests/test_conversation.py | 22 ++++++++++++++++++++++
 4 files changed, 27 insertions(+), 3 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index dc853d7..7635957 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -48,6 +48,8 @@ Follows [Keep a Changelog](https://keepachangelog.com/en/1.0.0/) · versioned wi
   workflow submission, prompt injection, polling, seed derivation, and image retrieval.
 - **Casting Constraint Coverage.** Added deterministic regression coverage proving casting filters invented voice IDs
   and avoids unnecessary duplicate assignments when distinct catalogue voices are available.
+- **Attribution and Timeout Safety Coverage.** Added regressions preventing phantom narrator speakers, preserving source
+  prose after malformed LLM JSON, clamping caller timeouts, and bounding upstream error text.
 - **4 GiB Container Validation.** Verified the current image serves the application under a hard 4 GiB cgroup limit
   with no OOM kill; the probe recorded approximately 50.6 MiB idle memory and 0.09% CPU.
 - **Profile Completeness Coverage.** Added browser regression coverage for empty, partial, and fully populated
diff --git a/TODO.md b/TODO.md
index ab9b435..983ccff 100644
--- a/TODO.md
+++ b/TODO.md
@@ -165,13 +165,13 @@ the first block before investing in architecture or polish.
 
 ### Automated tests
 
-- [ ] Character parsing and phantom-speaker prevention.
+- [x] Character parsing and phantom-speaker prevention, including malformed-response prose preservation.
 - [x] Voice-matching constraints reject invented voice IDs and avoid unnecessary duplicate assignments.
 - [x] Profile-completeness counting for empty, partial, and fully populated sheets.
 - [x] Rating persistence.
 - [x] STT endpoint normalization and backend-response shape handling.
 - [x] ComfyUI workflow submission, prompt injection, polling, and image retrieval.
-- [ ] API failure and timeout behavior.
+- [x] API timeout clamping and bounded upstream error text; live-provider failure drills remain environment-specific.
 
 ### Playwright smoke suite
 
diff --git a/VERSION b/VERSION
index ca8ec41..d8c40e5 100644
--- a/VERSION
+++ b/VERSION
@@ -1 +1 @@
-1.23.5
+1.23.6
diff --git a/tests/test_conversation.py b/tests/test_conversation.py
index 81ea182..29402b9 100644
--- a/tests/test_conversation.py
+++ b/tests/test_conversation.py
@@ -21,6 +21,8 @@ from routes.conversation import (
     _comfyui_run,
     _CastJobRequest,
     match_characters_voices,
+    _request_timeout_seconds,
+    _response_error_text,
 )
 
 
@@ -43,6 +45,26 @@ class AttributionParserTests(unittest.TestCase):
         self.assertEqual([segment["type"] for segment in result["segments"]], ["narration", "dialogue", "narration"])
         self.assertIn("Ein langer Satz endet hier.", result["segments"][1]["text"])
 
+    def test_attribution_normalization_prevents_phantom_narrator_and_preserves_text_on_bad_json(self):
+        raw = json.dumps({"segments": [
+            {"speaker": "Invented Narrator", "type": "narration", "text": "A road disappeared.", "emotion": "dramatic"},
+            {"speaker": "Mira", "type": "dialogue", "text": "We keep walking.", "emotion": ""},
+        ]})
+        result = _attribution_parse(raw, "unused")
+        self.assertEqual(result["segments"][0]["speaker"], "Narrator")
+        self.assertEqual(result["characters"], ["Mira"])
+        fallback = _attribution_parse("not valid JSON", "Original prose must survive")
+        self.assertEqual(fallback["segments"], [{"speaker": "Narrator", "type": "narration", "text": "Original prose must survive", "emotion": ""}])
+
+    def test_api_timeout_and_error_text_are_bounded(self):
+        self.assertEqual(_request_timeout_seconds(-10), 5.0)
+        self.assertEqual(_request_timeout_seconds(99999), 600.0)
+        self.assertEqual(_request_timeout_seconds("not-a-number", default=12), 12.0)
+        class ResponseStub:
+            text = "x" * 5000
+            reason = "Bad Gateway"
+        self.assertEqual(len(_response_error_text(ResponseStub())), 2000)
+
     def test_repairs_epigraph_split_before_mixed_guillemet_close(self):
         raw = '{"segments":[' \
             '{"speaker":"Schwester Larissa","type":"dialogue","text":"»Von dem Windhager Felsgesteyn bis zur Kueste von Mendena ziehet es sich hin, von Ifirns Eisgebirg bis zu dem dampfenden Waldt der Mohas, und"},' \