Spaces:
Sleeping
Sleeping
Download src/cli.py from andreamourao/MeuAgente: direct link, hf CLI and curl.
- Browser
- Download file 5.79 kB
-
https://huggingface.co/spaces/andreamourao/MeuAgente/resolve/main/src/cli.py
- Command line
-
hf download hf://spaces/andreamourao/MeuAgente/src/cli.py
-
curl -L -o cli.py https://huggingface.co/spaces/andreamourao/MeuAgente/resolve/main/src/cli.py
5.79 kB
| """CLI principal da aula. Wrapper Typer sobre todos os entry points. | |
| Comandos: | |
| smoke → testa OpenAI key + imports | |
| demo → roda um exemplo simples do agente single | |
| ask → faz uma pergunta ao agente single | |
| multi → roda o agente multi LangGraph | |
| serve → inicia FastAPI | |
| mcp → inicia servidor MCP local | |
| eval → roda avaliação BFCL local | |
| """ | |
| from __future__ import annotations | |
| import typer | |
| from rich.console import Console | |
| from rich.panel import Panel | |
| from rich.table import Table | |
| console = Console() | |
| app = typer.Typer(help="Aula 07 · CLI dos agentes") | |
| def smoke() -> None: | |
| """Smoke test: imports + chave OpenAI + chamada nano de 1 token.""" | |
| console.print(Panel.fit("Smoke test", style="cyan")) | |
| # 1. imports | |
| try: | |
| import langgraph # noqa: F401 | |
| import openai # noqa: F401 | |
| from fastapi import FastAPI # noqa: F401 | |
| console.print(" [green]✓[/green] imports OK") | |
| except ImportError as exc: | |
| console.print(f" [red]✗[/red] import falhou: {exc}") | |
| raise typer.Exit(1) from exc | |
| # 2. settings | |
| try: | |
| from src.config import get_product_config, get_settings | |
| settings = get_settings() | |
| product = get_product_config() | |
| console.print(f" [green]✓[/green] produto ativo: {product['name']}") | |
| console.print(f" [green]✓[/green] modelo: {settings.openai_model}") | |
| except Exception as exc: # noqa: BLE001 | |
| console.print(f" [red]✗[/red] config falhou: {exc}") | |
| raise typer.Exit(1) from exc | |
| # 3. chamada nano | |
| try: | |
| from src.llm import chat_completion | |
| msg = chat_completion( | |
| messages=[{"role": "user", "content": "Responda só com a palavra: pong"}], | |
| model="gpt-4.1-nano", | |
| temperature=0, | |
| ) | |
| content = (msg.get("content") or "").strip() | |
| if "pong" in content.lower(): | |
| console.print(f" [green]✓[/green] OpenAI respondeu: '{content}'") | |
| else: | |
| console.print(f" [yellow]?[/yellow] OpenAI respondeu (inesperado): '{content}'") | |
| except Exception as exc: # noqa: BLE001 | |
| console.print(f" [red]✗[/red] chamada OpenAI falhou: {exc}") | |
| raise typer.Exit(1) from exc | |
| console.print(Panel.fit("[bold green]Smoke test OK[/bold green]")) | |
| def demo() -> None: | |
| """Roda um exemplo simples (primeira pergunta do products.yaml).""" | |
| from src.agents.single_agent import SingleAgent | |
| from src.config import get_product_config | |
| config = get_product_config() | |
| query = config["example_prompts"][0] | |
| console.print(Panel.fit(f"Demo · {config['name']}", style="cyan")) | |
| console.print(f"[bold]Pergunta:[/bold] {query}\n") | |
| agent = SingleAgent(product_config=config) | |
| result = agent.run(query, verbose=False) | |
| _print_run_table(result) | |
| def ask(query: str = typer.Argument(..., help="Pergunta para o agente")) -> None: | |
| """Faz uma pergunta ao agente single ReAct.""" | |
| from src.agents.single_agent import SingleAgent | |
| console.print(Panel.fit(f"Pergunta: {query}", style="cyan")) | |
| agent = SingleAgent() | |
| result = agent.run(query, verbose=False) | |
| _print_run_table(result) | |
| def multi(query: str = typer.Argument(..., help="Pergunta para o agente multi")) -> None: | |
| """Roda o agente multi (LangGraph: planner → researcher → writer → critic).""" | |
| from src.agents.multi_agent import run_multi_agent | |
| console.print(Panel.fit(f"Multi-agente · Pergunta: {query}", style="cyan")) | |
| state = run_multi_agent(query) | |
| console.print(Panel(state.get("plan", ""), title="1. PLAN", style="blue")) | |
| console.print( | |
| Panel((state.get("context", "")[:800] + "..."), title="2. CONTEXT", style="magenta") | |
| ) | |
| console.print(Panel(state.get("draft_answer", ""), title="3. DRAFT", style="yellow")) | |
| console.print(Panel(state.get("critique", ""), title="4. CRITIQUE", style="red")) | |
| console.print( | |
| Panel.fit(state.get("final_answer", state.get("draft_answer", "")), | |
| title="RESPOSTA FINAL", style="green") | |
| ) | |
| def serve() -> None: | |
| """Inicia o FastAPI em 0.0.0.0:8000.""" | |
| import uvicorn | |
| from src.config import get_settings | |
| settings = get_settings() | |
| console.print(f"[cyan]Iniciando API em http://{settings.api_host}:{settings.api_port}[/cyan]") | |
| uvicorn.run("src.api.app:app", host=settings.api_host, port=settings.api_port, reload=False) | |
| def mcp() -> None: | |
| """Inicia o servidor MCP local mock.""" | |
| import uvicorn | |
| from src.config import get_settings | |
| settings = get_settings() | |
| console.print(f"[cyan]Iniciando MCP em http://{settings.mcp_host}:{settings.mcp_port}[/cyan]") | |
| uvicorn.run("src.mcp.server:app", host=settings.mcp_host, port=settings.mcp_port, reload=False) | |
| def eval() -> None: | |
| """Roda avaliação BFCL local.""" | |
| from evals.run_bfcl import run as run_bfcl | |
| run_bfcl() | |
| def _print_run_table(result) -> None: | |
| """Pretty print do AgentRun.""" | |
| table = Table(title="Trace do agente") | |
| table.add_column("Passo", style="cyan", width=6) | |
| table.add_column("Tipo", style="magenta", width=10) | |
| table.add_column("Detalhe", style="white") | |
| for step in result.steps: | |
| if step.is_final: | |
| table.add_row(str(step.step_number), "FINAL", step.final_answer[:300] or "") | |
| else: | |
| detail = f"tool={step.tool_name} args={step.tool_args}\n result={(step.tool_result or '')[:200]}" | |
| table.add_row(str(step.step_number), "TOOL", detail) | |
| console.print(table) | |
| console.print(Panel.fit(result.final_answer, title="Resposta final", style="green")) | |
| if __name__ == "__main__": | |
| app() | |