fix: recover character sheet requests from gateway 405

CHANGELOG.md

@@ -9,6 +9,16 @@ Follows [Keep a Changelog](https://keepachangelog.com/en/1.0.0/) · versioned wi
 
 ---
 
+## [1.26.2] — 2026-09-09
+
+### Fixed
+- **Character-Sheet Generation Could Stop on HTTP 405.** If an OpenAI-compatible LLM gateway rejects the configured
+  version-prefixed chat route (`/v1/chat/completions`) with 405, character-sheet extraction now retries the gateway's
+  root chat route (`/chat/completions`). This applies to both the live stream and its non-streaming fallback, so an
+  otherwise healthy gateway no longer drops the current passage.
+
+---
+
 ## [1.26.1] — 2026-09-08
 
 ### Added

VERSION

@@ -1 +1 @@
-1.26.1
+1.26.2

routes/conversation.py

@@ -395,11 +395,23 @@ def _response_error_text(resp) -> str:
 
 
 def _llm_chat_completion_urls(llm_url: str) -> list[str]:
+    """Return compatible chat-completions routes for an LLM base URL.
+
+    Most OpenAI-compatible servers use a ``/v1`` prefix, but a few routing
+    gateways expose the same API at their root.  In particular, gateways can
+    reply 405 (rather than 404) when the version-prefixed route is disabled.
+    Trying the root route after a versioned route prevents a single passage
+    from being lost just because the configured gateway uses that variant.
+    """
     base = (llm_url or "http://localhost:11434/v1").rstrip("/")
+    versioned = bool(re.search(r"/v\d+(?:\.\d+)?$", base))
+    root = re.sub(r"/v\d+(?:\.\d+)?$", "", base) if versioned else base
     urls: list[str] = []
-    if not re.search(r"/v\d+(?:\.\d+)?$", base):
+    if not versioned:
         urls.append(f"{base}/v1/chat/completions")
     urls.append(f"{base}/chat/completions")
+    if versioned:
+        urls.append(f"{root}/chat/completions")
     return list(dict.fromkeys(urls))
 
 

tests/test_conversation.py

@@ -28,6 +28,7 @@ from routes.conversation import (
     match_characters_voices,
     _request_timeout_seconds,
     _response_error_text,
+    _llm_chat_completion_urls,
     _persist_consistent_image_artifacts,
     _data_image_digest,
 )
@@ -112,6 +113,25 @@ class AttributionParserTests(unittest.TestCase):
             reason = "Bad Gateway"
         self.assertEqual(len(_response_error_text(ResponseStub())), 2000)
 
+    def test_chat_completion_urls_fall_back_from_versioned_gateway_route(self):
+        # Some OpenAI-compatible routing gateways return 405 for /v1 but
+        # accept their chat endpoint at the root.  Keep the configured route
+        # first, then try that compatible alternative.
+        self.assertEqual(
+            _llm_chat_completion_urls("http://gateway:28080/v1"),
+            [
+                "http://gateway:28080/v1/chat/completions",
+                "http://gateway:28080/chat/completions",
+            ],
+        )
+        self.assertEqual(
+            _llm_chat_completion_urls("http://gateway:28080"),
+            [
+                "http://gateway:28080/v1/chat/completions",
+                "http://gateway:28080/chat/completions",
+            ],
+        )
+
     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"},' \