"""Convert an informal assignment description into a validated flat card."""

from __future__ import annotations

import argparse
import os

from dotenv import find_dotenv, load_dotenv
from langchain_core.output_parsers import PydanticOutputParser
from langchain_core.prompts import PromptTemplate
from langchain_openai import ChatOpenAI
from pydantic import BaseModel, Field


DEFAULT_DESCRIPTION = (
    "Сдайте к пятнице мини-отчёт по LangChain: 2 страницы, упор на агентов. "
    "Оценка: за полноту и за пример кода."
)


class AssignmentCard(BaseModel):
    """Flat data card extracted from one assignment description."""

    title: str = Field(..., description="Short title of the assignment")
    subject: str = Field(..., description="Main topic or subject")
    deadline_hint: str | None = Field(
        None,
        description="Deadline as a free-form phrase from the source text",
    )
    deliverable_type: str | None = Field(
        None,
        description="Expected deliverable: report, code, presentation, etc.",
    )
    grading_hints: list[str] = Field(
        default_factory=list,
        description="Assessment hints mentioned in the description",
    )


parser = PydanticOutputParser(pydantic_object=AssignmentCard)

prompt = PromptTemplate(
    template=(
        "You convert one informal study assignment description into a flat "
        "validated data card. Do not ask follow-up questions. Extract only "
        "information supported by the input. If a field is not mentioned, use "
        "null for optional strings or an empty list for grading_hints.\n\n"
        "{format_instructions}\n\n"
        "Assignment description:\n{description}"
    ),
    input_variables=["description"],
    partial_variables={"format_instructions": parser.get_format_instructions()},
)


def build_llm() -> ChatOpenAI:
    """Create an OpenAI-compatible chat model from environment variables."""

    load_dotenv(find_dotenv(usecwd=True))
    base_url = os.getenv("OPENAI_BASE_URL") or os.getenv("GPT2GIGA_BASE_URL")
    api_key = os.getenv("OPENAI_API_KEY") or os.getenv("GPT2GIGA_OPENAI_API_KEY")
    model = (
        os.getenv("OPENAI_MODEL")
        or os.getenv("GPT2GIGA_MODEL")
        or "openai/gpt-oss-20b"
    )
    if not base_url:
        raise RuntimeError("Set OPENAI_BASE_URL or GPT2GIGA_BASE_URL")
    if not api_key:
        raise RuntimeError("Set OPENAI_API_KEY or GPT2GIGA_OPENAI_API_KEY")

    return ChatOpenAI(
        model=model,
        base_url=base_url,
        api_key=api_key,
        temperature=0,
    )


def parse_assignment(description: str) -> AssignmentCard:
    """Run one prompt-model-parser chain and return a validated card."""

    chain = prompt | build_llm() | parser
    return chain.invoke({"description": description})


def format_summary(card: AssignmentCard) -> str:
    """Create a short human-readable summary from the validated card."""

    grading = ", ".join(card.grading_hints) if card.grading_hints else "не указано"
    return (
        f"{card.title}: {card.deliverable_type or 'тип сдачи не указан'} "
        f"по теме «{card.subject}». "
        f"Срок: {card.deadline_hint or 'не указан'}. "
        f"Критерии: {grading}."
    )


def print_card(card: AssignmentCard) -> None:
    """Print both machine-friendly and human-friendly output."""

    print("Validated object:")
    for key, value in card.model_dump().items():
        print(f"{key}: {value}")
    print("\nSummary:")
    print(format_summary(card))


def main(argv: list[str] | None = None) -> AssignmentCard:
    arg_parser = argparse.ArgumentParser(
        description="Parse one raw assignment description into a flat card."
    )
    arg_parser.add_argument(
        "description",
        nargs="*",
        help="One informal assignment description. Uses a sample if omitted.",
    )
    args = arg_parser.parse_args(argv)

    description = " ".join(args.description).strip() or DEFAULT_DESCRIPTION
    card = parse_assignment(description)
    print_card(card)
    return card


if __name__ == "__main__":
    main()