# ============================================================
# SPID — extraction interface & record schema
# ============================================================
from __future__ import annotations

import hashlib
from dataclasses import dataclass, field, asdict
from typing import Protocol


@dataclass
class ProjectMention:
    """A single project signal extracted from one filing section.

    Mentions are the atomic unit; `resolve.py` merges many mentions
    (across filings & years) into a single Project with a timeline.
    """
    mention_id: str
    accession_number: str
    cik: str
    section_id: str
    section_name: str

    project_name: str
    project_type: str
    description: str
    objective: str = ""

    amount_usd: float | None = None
    amount_raw: str | None = None
    locations: list[str] = field(default_factory=list)
    technologies: list[str] = field(default_factory=list)
    partners: list[str] = field(default_factory=list)
    suppliers: list[str] = field(default_factory=list)
    benefits: list[str] = field(default_factory=list)
    risks: list[str] = field(default_factory=list)
    years: list[int] = field(default_factory=list)
    status: str = "mentioned"

    confidence: float = 0.0
    backend: str = "rules"

    def to_row(self) -> dict:
        d = asdict(self)
        # store list fields as JSON-friendly via duckdb list[str]
        return d

    @staticmethod
    def make_id(accession: str, section_id: str, ptype: str, anchor: str) -> str:
        h = hashlib.sha1(f"{accession}|{section_id}|{ptype}|{anchor}".encode()).hexdigest()
        return h[:20]


class Extractor(Protocol):
    name: str

    def extract(self, *, text: str, accession_number: str, cik: str,
                section_id: str, section_name: str) -> list[ProjectMention]:
        ...


def get_extractor(backend: str):
    """Factory. `backend` in {'rules','llm'}."""
    backend = (backend or "rules").lower()
    if backend == "llm":
        from spid.extract.llm import LLMExtractor
        return LLMExtractor()
    from spid.extract.rules import RuleExtractor
    return RuleExtractor()