From 31da2788a3379ad798d633645ea0af778eba3269 Mon Sep 17 00:00:00 2001
From: ajokar <ajokar91@yahoo.com>
Date: Tue, 12 May 2026 12:58:25 +0000
Subject: [PATCH] nahj get metadata

---
 nahj_get_metadata_oss.py | 124 ++++++++++++++++++++++++++++++++++++++-
 1 file changed, 123 insertions(+), 1 deletion(-)

diff --git a/nahj_get_metadata_oss.py b/nahj_get_metadata_oss.py
index 71195c9..e918ee2 100644
--- a/nahj_get_metadata_oss.py
+++ b/nahj_get_metadata_oss.py
@@ -11,6 +11,11 @@ import asyncio
 import traceback
 from openai import AsyncOpenAI 
 
+import copy, asyncio, traceback
+from openai import OpenAI, AsyncOpenAI, LengthFinishReasonError
+from typing import List, Union
+from pydantic import BaseModel
+
 today = f'{datetime.datetime.now().year}{datetime.datetime.now().month}{datetime.datetime.now().day}'
 
 SYSTEM_PROMPT = """
@@ -148,6 +153,99 @@ async def single_simple_async_proccess_item(
             traceback.print_exc()
             raise RuntimeError(f"⚠️ Error in API call: {str(e)}")
 
+class Result(BaseModel):
+    result : str 
+
+async def single_async_item(
+    api_url,
+    api_key,
+    item,
+    reasoning_effort,
+    temperature,
+    top_p,
+    semaphore_number,
+    model_name,
+    priority,
+    output_schema=None,
+    max_token=4096,
+    print_logs=False,
+    return_reason=False,
+    stop=None,
+    return_used_token=False,
+    timeout=300,
+    ):
+    try:
+        async with AsyncOpenAI(
+            base_url=api_url, api_key=api_key
+        ) as client:
+            semaphore = asyncio.Semaphore(semaphore_number)
+            async with semaphore:
+                messages = [{"role": "user", "content": item["user_prompt"]}]
+                if item.get("system_prompt"):
+                    messages.insert(
+                        0, {"role": "system", "content": item["system_prompt"]}
+                    )
+                if item.get("assistant_prompt"):
+                    messages.append(
+                        {"role": "assistant", "content": item["assistant_prompt"]}
+                    )
+
+                coro = client.chat.completions.parse(
+                    model=model_name,
+                    messages=messages,
+                    temperature=temperature,
+                    top_p=top_p,
+                    max_tokens=max_token,
+                    stop=stop,
+                    response_format=output_schema,
+                    reasoning_effort=reasoning_effort,
+                    extra_body={"priority": priority},
+                )
+                response = await asyncio.wait_for(coro, timeout=timeout)
+
+                if print_logs:
+                    print(f"parse response ----  {response}")
+
+                parsed_obj = response.choices[0].message.parsed 
+                # print(f'parsed_obj {parsed_obj}')
+                if parsed_obj is None:
+                    return {
+                        "error": "Failed to parse response",
+                        "raw": str(response),
+                    }
+                
+                parsed_obj = output_schema.model_validate(parsed_obj)
+                # Validate just in case (optional, چون .parse already does it)
+                if return_reason:
+                    reasoning_content = response.choices[
+                        0
+                    ].message.reasoning_content
+                    if return_used_token:
+                        _total_token = response.usage.total_tokens
+                        item["llm_output"] = (
+                            parsed_obj.model_dump(),
+                            str(reasoning_content),
+                            int(_total_token),
+                        )
+                        return item
+                    
+                    item["llm_output"] = (
+                        parsed_obj.model_dump(),
+                        str(reasoning_content)
+                    )
+                    return item
+
+                item["llm_output"] = parsed_obj.model_dump()
+                return item
+                
+    except asyncio.TimeoutError:
+        print(f"⏳ Timeout on item {item}")
+        return None
+
+    except Exception as e:
+        print(f"⚠️ Error __process_item {item}: {traceback.print_exc()}")
+        return None
+
 async def main():
     with open('./leader_data/khamenei_messages_4.json', 'r', encoding='utf-8') as file:
             data = json.load(file)
@@ -221,9 +319,33 @@ async def main():
     print(f'all_paragraphs: {all_paragraphs}')
     print('---------------------------------------------')
 
+async def oss_test():
+    item = {}
+    item['assistant_prompt'] = "تو یک دستیار خبره در زمینه تدوین متون علمی هستی"
+    item['system_prompt'] = "پاسخ ها فقط باید علمی باشند و سبک نگارش طنز، سرگرمی، ادبی،احساسی و ... قابل قبول نیست."
+    item['user_prompt'] = "ابعاد مختلف علوم اجتماعی محاسباتی کدام است؟"
+    response = await single_async_item(
+                api_url="http://2.188.15.102:8001/v1/",
+                api_key="EMPTY",
+                item=item,
+                reasoning_effort="medium",
+                temperature=0.1,
+                top_p=1,
+                semaphore_number=1,
+                model_name="gpt-oss-120b", 
+                priority=1,
+                output_schema=Result,
+                max_token=None,
+                return_reason=True,
+                return_used_token=True,
+                timeout=300
+            )
+    print(response['llm_output'])
+    pass
  
 
 if __name__ == "__main__":
-    asyncio.run(main())
+    # asyncio.run(main())
+    asyncio.run(oss_test())