nomyo-ai/nomyo-router
2
2
Fork 0
Code Issues 1 Pull requests Projects Releases 5 Packages 1 Wiki Activity Actions

simplification in rechunk

Browse source
This commit is contained in:
Alpha Nerd 2025-09-13 12:38:13 +02:00
parent b7b3967296
commit 0a7fd8ca52
1 changed files with 17 additions and 32 deletions
Download patch file Download diff file Expand all files Collapse all files

49
router.py
View file

@ -270,37 +270,24 @@ def iso8601_ns():
class rechunk:
def openai_chat_completion2ollama(chunk: dict, stream: bool, start_ts: float):
rechunk = { "model": chunk.model,
"created_at": iso8601_ns() ,
"done_reason": chunk.choices[0].finish_reason,
"load_duration": None,
"prompt_eval_count": None,
"prompt_eval_duration": None,
"eval_count": None,
"eval_duration": None,
"eval_count": (chunk.usage.completion_tokens if chunk.usage is not None else None),
"prompt_eval_count": (chunk.usage.prompt_tokens if chunk.usage is not None else None),
"eval_duration": (int((time.perf_counter() - start_ts) * 1000) if chunk.usage is not None else None),
"response_token/s": (round(chunk.usage.total_tokens / (time.perf_counter() - start_ts), 2) if chunk.usage is not None else None)
}
if stream == True:
chunk = { "model": chunk.model,
"created_at": iso8601_ns() ,
"done_reason": chunk.choices[0].finish_reason,
"load_duration": None,
"prompt_eval_count": None,
"prompt_eval_duration": None,
"eval_count": None,
"eval_duration": None,
"message": {"role": chunk.choices[0].delta.role, "content": chunk.choices[0].delta.content, "thinking": None, "images": None, "tool_name": None, "tool_calls": None},
"eval_count": (chunk.usage.completion_tokens if chunk.usage is not None else None),
"prompt_eval_count": (chunk.usage.prompt_tokens if chunk.usage is not None else None),
"eval_duration": (int((time.perf_counter() - start_ts) * 1000) if chunk.usage is not None else None),
"response_token/s": (round(chunk.usage.total_tokens / (time.perf_counter() - start_ts), 2) if chunk.usage is not None else None)
}
rechunk["message"] = {"role": chunk.choices[0].delta.role, "content": chunk.choices[0].delta.content, "thinking": None, "images": None, "tool_name": None, "tool_calls": None}
else:
chunk = { "model": chunk.model,
"created_at": iso8601_ns() ,
"done_reason": chunk.choices[0].finish_reason,
"load_duration": None,
"prompt_eval_count": None,
"prompt_eval_duration": None,
"eval_count": None,
"eval_duration": None,
"message": {"role": chunk.choices[0].message.role, "content": chunk.choices[0].message.content, "thinking": None, "images": None, "tool_name": None, "tool_calls": None},
"eval_count": (chunk.usage.completion_tokens if chunk.usage is not None else None),
"prompt_eval_count": (chunk.usage.prompt_tokens if chunk.usage is not None else None),
"eval_duration": (int((time.perf_counter() - start_ts) * 1000) if chunk.usage is not None else None),
"response_token/s": (round(chunk.usage.total_tokens / (time.perf_counter() - start_ts), 2) if chunk.usage is not None else None)
}
return chunk
rechunk["message"] = {"role": chunk.choices[0].message.role, "content": chunk.choices[0].message.content, "thinking": None, "images": None, "tool_name": None, "tool_calls": None}
return rechunk
# ------------------------------------------------------------------
# SSE Helpser
@ -555,21 +542,19 @@ async def chat_proxy(request: Request):
if stream == True:
async for chunk in async_gen:
if is_openai_endpoint:
print(chunk)
chunk = rechunk.openai_chat_completion2ollama(chunk, stream, start_ts)
# `chunk` can be a dict or a pydantic model – dump to JSON safely
if hasattr(chunk, "model_dump_json"):
json_line = chunk.model_dump_json()
else:
json_line = json.dumps(chunk)
print(json_line)
yield json_line.encode("utf-8") + b"\n"
else:
if is_openai_endpoint:
response = rechunk.openai_chat_completion2ollama(async_gen, stream, start_ts)
response = json.dumps(response)
else:
repsonse = async_gen.model_dump_json()
response = async_gen.model_dump_json()
json_line = (
response
if hasattr(async_gen, "model_dump_json")