# # from langchain.agents import create_tool_calling_agent, AgentExecutor
# # from langchain.prompts import ChatPromptTemplate
# # from modules import model, llm_g
# # import json
# # import re
# # import pandas as pd


# # def units_langchain_agent(topic: str, units_number: int):
# #     # الخطوة 1: نطلب من llm_g إنشاء Markdown plan
# #     result = llm_g.call(
# #         f"Create a structured {units_number}-unit study plan for the topic [{topic}]"
# #     )

# #     # الخطوة 2: الـ System Prompt الخاص بالاستخراج
# #     system_prompt = """
# #     You will receive Markdown text representing a study plan.
# #     Extract structured data and return it strictly in this JSON format:

# #     {{
# #         "topic": "<topic name>",
# #         "General_objectives_of_the_plan": ["string", "string", ...],
# #         "total_units": <total_units_number>,
# #         "units": [
# #             {{
# #                 "unit_number": 1,
# #                 "unit_title": "string",
# #                 "description": "string",
# #                 "unit_objectives": ["string", "string", ...],
# #                 "unit_main_topics": ["string", "string", ...]
# #             }},
# #             ...
# #         ]
# #     }}

# #     Output **only valid JSON**, no explanations, markdown, or extra text.
# #     """

# #     # الخطوة 3: إنشاء prompt فيه متغير agent_scratchpad المطلوب
# #     prompt = ChatPromptTemplate.from_messages(
# #         [
# #             ("system", system_prompt),
# #             ("user", "{input}"),
# #             ("assistant", "{agent_scratchpad}"),  # << الإضافة المطلوبة
# #         ]
# #     )

# #     # الخطوة 4: إنشاء الـ Agent
# #     tools = []  # مفيش Tools هنا
# #     agent = create_tool_calling_agent(model, tools, prompt=prompt)

# #     # الخطوة 5: تنفيذ الـ Agent
# #     agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)
# #     response = agent_executor.invoke({"input": result, "agent_scratchpad": ""})

# #     # الخطوة 6: استخراج النتيجة
# #     raw_output = response.get("output", "").strip()

# #     # الخطوة 7: تنظيف وتحويل JSON
# #     try:
# #         structured_data = json.loads(raw_output)
# #     except json.JSONDecodeError:
# #         cleaned = re.sub(r"```json|```", "", raw_output).strip()
# #         structured_data = json.loads(cleaned)

# #     return structured_data


# from modules import model, llm_g
# import json
# import re
# from langchain.agents import initialize_agent, AgentType

# # from langchain.agents import create_agent


# def units_langchain1_agent(topic: str, units_number: int):
#     llm = llm_g()
#     # الخطوة 1: نطلب من llm_g إنشاء Markdown plan
#     result = llm.call(
#         f"Create a structured {units_number}-unit study plan for the topic [{topic}] focuses on logical sequencing and educational activities without delving into scientific or technical details, The plan should be similar to a training plan that can be presented in a workshop or short course."
#     )

#     # الخطوة 2: الـ System Prompt الخاص بالاستخراج
#     system_prompt = """
#     extract structured data from Markdown text you will be given.

#     The JSON should have this exact structure:
#     {
#         "topic": "<topic name>",
#         "General_objectives_of_the_plan": ["string", "string", ...]
#         "total_units": <total_units_number>,
#         "units": [
#             {
#                 "unit_number": 1,
#                 "unit_title": "string",
#                 "description": "string",
#                 "unit_objectives": ["string", "string", ...],
#                 "unit_main_topics": ["string", "string", ...]
#             },
#             ...
#         ]
#     }

#     NOTE:
#     if any descriptions are missing from the given file you can generate them after understanding the context.

#     Output **only valid JSON**, no explanations, markdown, or extra text.
#     """

#     # agent = create_agent(
#     #     model=model,
#     #     system_prompt=system_prompt,
#     # )
#     model_open = model()
#     agent = initialize_agent(
#         system_prompt=system_prompt,
#         llm=model_open,
#         agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION,
#         verbose=True,
#     )

#     response = agent.invoke({"messages": [{"role": "user", "content": result}]})

#     #######################
#     # الخطوة 6: استخراج النتيجة
#     raw_output = response["messages"][1].content.strip()

#     # الخطوة 7: تنظيف وتحويل JSON
#     try:
#         structured_data = json.loads(raw_output)
#     except json.JSONDecodeError:
#         cleaned = re.sub(r"```json|```", "", raw_output).strip()
#         structured_data = json.loads(cleaned)

#     return structured_data