From b5ad3c10281b804c936f903d516d2e8beab9982a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=D0=98=D0=BB=D1=8C=D1=8F=205f1b81b8-4f5d-11e8-9c2d-fa7ae01?=
 =?UTF-8?q?bbebc?= <gurevich1104@mail.ru>
Date: Sat, 27 Jun 2026 13:52:15 +0000
Subject: [PATCH] add: main.py

---
 main.py | 146 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 146 insertions(+)
 create mode 100644 main.py

diff --git a/main.py b/main.py
new file mode 100644
index 0000000..ca565b8
--- /dev/null
+++ b/main.py
@@ -0,0 +1,146 @@
+import os
+import asyncio
+import argparse
+from dotenv import load_dotenv
+
+from pydantic import BaseModel, Field
+from langchain_openai import ChatOpenAI
+from langchain_core.messages import HumanMessage
+from langchain.tools import tool
+from langchain_core.output_parsers import PydanticOutputParser
+from langchain_core.prompts import PromptTemplate
+
+from deepagents import create_deep_agent
+from deepagents.backends import FilesystemBackend, LocalShellBackend, CompositeBackend
+
+# Load API key from .env
+load_dotenv()
+
+# LLM configuration – OpenRouter
+llm = ChatOpenAI(
+    model="openai/gpt-oss-20b:free",
+    base_url="https://openrouter.ai/api/v1",
+    api_key=os.getenv("OPENAI_API_KEY"),
+    temperature=0.0,
+)
+
+# Pydantic models
+class PersonInfo(BaseModel):
+    name: str = Field(description="Full name of the person")
+    age: int | None = Field(description="Age of the person, optional", default=None)
+    profession: str = Field(description="Current profession or role")
+    skills: list[str] = Field(description="List of skills or technologies")
+
+class MeetingNotes(BaseModel):
+    date: str = Field(description="Date of the meeting (YYYY-MM-DD)")
+    participants: list[str] = Field(description="Names of participants")
+    topics: list[str] = Field(description="Main discussion topics")
+    decisions: list[str] = Field(description="Decisions made during the meeting")
+    next_steps: list[str] = Field(description="Action items for next steps")
+
+# Prompt templates
+person_prompt = PromptTemplate.from_template(
+    """
+Extract the following information from the text: name, age (optional), profession, skills.
+The text: {input_text}
+Output must be in JSON format according to the following schema:
+{format_instructions}
+"""
+)
+
+meeting_prompt = PromptTemplate.from_template(
+    """
+Extract the following information from the text: date, participants, topics, decisions, next_steps.
+The text: {input_text}
+Output must be in JSON format according to the following schema:
+{format_instructions}
+"""
+)
+
+# Output parsers
+person_parser = PydanticOutputParser(pydantic_object=PersonInfo)
+meeting_parser = PydanticOutputParser(pydantic_object=MeetingNotes)
+
+# Simple heuristic to decide which schema to use
+
+def determine_schema(text: str) -> str:
+    lower = text.lower()
+    meeting_keywords = ["meeting", "participants", "date", "topics", "decisions", "next steps", "next_step", "next_step"]
+    if any(k in lower for k in meeting_keywords):
+        return "meeting"
+    return "person"
+
+# Summary generation
+
+def create_summary(parsed: BaseModel, schema: str) -> str:
+    if schema == "meeting":
+        return (
+            f"Meeting on {parsed.date} with participants {', '.join(parsed.participants)}. "
+            f"Topics: {', '.join(parsed.topics)}. Decisions: {', '.join(parsed.decisions)}. "
+            f"Next steps: {', '.join(parsed.next_steps)}."
+        )
+    else:
+        age = parsed.age if parsed.age is not None else "N/A"
+        return (
+            f"{parsed.name}, age {age}, profession {parsed.profession}. "
+            f"Skills: {', '.join(parsed.skills)}."
+        )
+
+# Tool that performs extraction
+@tool
+def process_text(input_text: str) -> str:
+    schema = determine_schema(input_text)
+    if schema == "meeting":
+        prompt = meeting_prompt
+        parser = meeting_parser
+    else:
+        prompt = person_prompt
+        parser = person_parser
+    chain = prompt | llm | parser
+    parsed = chain.invoke({"input_text": input_text})
+    summary = create_summary(parsed, schema)
+    return parsed.model_dump_json(indent=2) + "\n\nSummary: " + summary
+
+# Backend for the agent
+backend = CompositeBackend([
+    LocalShellBackend(workspace_dir="./workspace"),
+    FilesystemBackend(),
+])
+
+# Create the deep agent
+agent = create_deep_agent(
+    model=llm,
+    tools=[process_text],
+    backend=backend,
+    system_prompt="You are a helpful agent that extracts structured data from text.",
+)
+
+# CLI entry point
+async def main():
+    parser = argparse.ArgumentParser(description="Extract structured data from text.")
+    parser.add_argument("--example", choices=["person", "meeting"], help="Run example")
+    parser.add_argument("--text", type=str, help="Input text")
+    args = parser.parse_args()
+
+    if args.example:
+        if args.example == "person":
+            input_text = "Анна, 28 лет, Python-разработчик. Навыки: FastAPI, Docker."
+        else:
+            input_text = (
+                "Дата: 2023-05-01. Участники: Иван, Мария. Темы: проект X, бюджет. "
+                "Решения: утвердить бюджет. Next steps: подготовить план."
+            )
+    elif args.text:
+        input_text = args.text
+    else:
+        input_text = input("Enter text: ")
+
+    result = await agent.ainvoke(
+        {"messages": [HumanMessage(content=input_text)]},
+        {"configurable": {"thread_id": "session-1"}},
+    )
+    output = result["messages"][-1].content
+    print(output)
+
+if __name__ == "__main__":
+    asyncio.run(main())