oimwiodev/ChatMock
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fixes #103: responses api max_output_tokens bug

Browse Source
This commit is contained in:
Game_Time
2026-03-23 23:13:58 +05:00
parent 8754203ec6
commit da0e3c3176
9 changed files with 132 additions and 331 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
chatmock/cli.py
View File

@@ -284,7 +284,7 @@ def cmd_serve(
default_web_search=default_web_search,
)
app.run(host=host, debug=False, use_reloader=False, port=port, threaded=True)
app.run(host=host, use_reloader=False, port=port, threaded=True)
return 0

1
chatmock/responses_api.py
View File

@@ -88,6 +88,7 @@ def normalize_responses_payload(
normalized = dict(payload)
normalized["model"] = normalized_model
normalized.pop("max_output_tokens", None)
if "input" in normalized:
normalized["input"] = canonicalize_responses_input(normalized.get("input"))

6
chatmock/routes_ollama.py
View File

@@ -250,7 +250,7 @@ def ollama_chat() -> Response:
input_items = convert_chat_messages_to_responses_input(messages)
model_reasoning = extract_reasoning_from_model_name(model)
normalized_model = normalize_model_name(model)
normalized_model = normalize_model_name(model, current_app.config.get("DEBUG_MODEL"))
service_tier_resolution = resolve_service_tier(
normalized_model,
request_fast_mode=payload.get("fast_mode"),
@@ -306,7 +306,7 @@ def ollama_chat() -> Response:
base_tools_only = convert_tools_chat_to_responses(normalize_ollama_tools(tools_req))
safe_choice = payload.get("tool_choice", "auto")
upstream2, err2 = start_upstream_request(
normalize_model_name(model),
normalize_model_name(model, current_app.config.get("DEBUG_MODEL")),
input_items,
instructions=BASE_INSTRUCTIONS,
tools=base_tools_only,
@@ -570,7 +570,7 @@ def ollama_chat() -> Response:
full_text = f"<think>{rtxt}</think>" + (full_text or "")
out_json = {
"model": normalize_model_name(model),
"model": normalize_model_name(model, current_app.config.get("DEBUG_MODEL")),
"created_at": created_at,
"message": {"role": "assistant", "content": full_text, **({"tool_calls": tool_calls} if tool_calls else {})},
"done": True,

6
chatmock/routes_openai.py
View File

@@ -109,7 +109,6 @@ def chat_completions() -> Response:
reasoning_effort = current_app.config.get("REASONING_EFFORT", "medium")
reasoning_summary = current_app.config.get("REASONING_SUMMARY", "auto")
reasoning_compat = current_app.config.get("REASONING_COMPAT", "think-tags")
debug_model = current_app.config.get("DEBUG_MODEL")
raw = request.get_data(cache=True, as_text=True) or ""
if verbose:
@@ -129,7 +128,7 @@ def chat_completions() -> Response:
return jsonify(err), 400
requested_model = payload.get("model")
model = normalize_model_name(requested_model, debug_model)
model = normalize_model_name(requested_model, current_app.config.get("DEBUG_MODEL"))
messages = payload.get("messages")
if messages is None and isinstance(payload.get("prompt"), str):
messages = [{"role": "user", "content": payload.get("prompt") or ""}]
@@ -413,7 +412,6 @@ def chat_completions() -> Response:
def completions() -> Response:
verbose = bool(current_app.config.get("VERBOSE"))
verbose_obfuscation = bool(current_app.config.get("VERBOSE_OBFUSCATION"))
debug_model = current_app.config.get("DEBUG_MODEL")
reasoning_effort = current_app.config.get("REASONING_EFFORT", "medium")
reasoning_summary = current_app.config.get("REASONING_SUMMARY", "auto")
@@ -432,7 +430,7 @@ def completions() -> Response:
return jsonify(err), 400
requested_model = payload.get("model")
model = normalize_model_name(requested_model, debug_model)
model = normalize_model_name(requested_model, current_app.config.get("DEBUG_MODEL"))
prompt = payload.get("prompt")
if isinstance(prompt, list):
prompt = "".join([p if isinstance(p, str) else "" for p in prompt])