backend/core/generation_state.py

"""Generation state — hierarchia źródeł prawdy i helpery persystencji.

Problem (audyt): stan generacji był rozproszony w ``external_context`` JSON,
checkpointach LangGraph (sqlite) i GSD orchestratorze.

Hierarchia odczytu (od najbardziej autorytatywnego):
  1. **GSD orchestrator** — ``gsd/gsd_state.py`` + sqlite saver (aktywny run)
  2. **Hybrid checkpoint** — ``external_context['orchestrator_checkpoint_v5']``
  3. **Legacy LangGraph generator** — ``data/generator_checkpoints.db`` (DEPRECATED)
  4. **Pozostałe klucze w external_context** — cache / pochodne (nie źródło prawdy)

Zapis: używaj ``persist_generation_checkpoint`` aby synchronizować warstwy 1↔2.
"""
from __future__ import annotations

import logging
from copy import deepcopy
from typing import Any

logger = logging.getLogger(__name__)

CHECKPOINT_V5_KEY = "orchestrator_checkpoint_v5"
GSD_STATE_KEY = "gsd_state"
GENERATION_META_KEY = "generation_meta"

DERIVED_CONTEXT_KEYS = frozenset(
    {
        "holistic_review",
        "risk_score",
        "blocks_export_recommendation",
        "v5_grounding_certificate",
        "compliance_checklist",
        "regulation_section_titles",
        "credibility_flags",
    }
)


def get_authoritative_checkpoint(project) -> dict[str, Any] | None:
    """Zwraca hybrid checkpoint v5 z projektu lub None."""
    ext = getattr(project, "external_context", None) or {}
    ck = ext.get(CHECKPOINT_V5_KEY)
    return ck if isinstance(ck, dict) else None


def get_generation_meta(project) -> dict[str, Any]:
    """Metadane generacji (ścieżka orkiestracji, ostatnia faza)."""
    ext = getattr(project, "external_context", None) or {}
    meta = ext.get(GENERATION_META_KEY) or {}
    return meta if isinstance(meta, dict) else {}


def persist_generation_checkpoint(
    db,
    project,
    checkpoint: dict[str, Any],
    *,
    orchestration_path: str = "gsd",
    merge_meta: dict[str, Any] | None = None,
) -> None:
    """Zapisuje checkpoint do ``external_context`` (single durable source w Postgres/SQLite)."""
    from sqlalchemy.orm.attributes import flag_modified

    ext = deepcopy(getattr(project, "external_context", None) or {})
    ext[CHECKPOINT_V5_KEY] = checkpoint

    meta = dict(ext.get(GENERATION_META_KEY) or {})
    meta["orchestration_path"] = orchestration_path
    meta["last_checkpoint_stage"] = checkpoint.get("stage") or checkpoint.get("current_orchestrator_stage")
    if merge_meta:
        meta.update(merge_meta)
    ext[GENERATION_META_KEY] = meta

    project.external_context = ext
    flag_modified(project, "external_context")
    db.commit()
    logger.debug(
        "[generation_state] Persisted checkpoint stage=%s path=%s project=%s",
        meta.get("last_checkpoint_stage"),
        orchestration_path,
        getattr(project, "id", "?"),
    )


def is_legacy_langgraph_checkpoint(config: dict[str, Any] | None) -> bool:
    """Czy config wskazuje na legacy LangGraph thread (generator agent)."""
    if not config:
        return False
    tid = config.get("configurable", {}).get("thread_id", "")
    return isinstance(tid, str) and tid.startswith("gen-")