content-find-agent-new/content_agent/integrations/policy_json.py

from __future__ import annotations

from pathlib import Path
from typing import Any

from content_agent.constants import DEFAULT_POLICY_BUNDLE_ID, EVIDENCE_BUNDLE_SCHEMA_VERSION, RUNTIME_SCHEMA_VERSION
from content_agent.errors import ContentAgentError, ErrorCode
from content_agent.integrations import config_store
from content_agent.integrations.walk_strategy_json import WalkStrategyStore


class JsonPolicyBundleStore:
    def __init__(self, root_dir: Path | str = Path(".")) -> None:
        self.root_dir = Path(root_dir)

    def load_policy_bundle(self, strategy_version: str) -> dict[str, Any]:
        rule_pack_path = self.root_dir / "product_documents/规则包/douyin_rule_packs.v1.json"
        if not rule_pack_path.exists():
            raise FileNotFoundError(rule_pack_path)

        rule_package, rule_pack_text = config_store.load_json(rule_pack_path)
        rule_pack_hash = config_store.sha256_text(rule_pack_text)

        actual_strategy_version = _strategy_version(rule_package)
        if strategy_version != actual_strategy_version:
            raise ValueError(f"unknown strategy_version: {strategy_version}")

        dispatch = _select_dispatch(rule_package, actual_strategy_version)
        rule_pack = _find_rule_pack_by_dispatch(rule_package, dispatch)
        rule_pack_by_entity = _build_rule_pack_by_entity(rule_package, actual_strategy_version)
        walk_strategy = WalkStrategyStore(self.root_dir).load_walk_strategy()
        # V4 判定/游走门槛:独立可手维护配置(不走 Excel→规则包管线),按平台解耦。
        score_thresholds_path = self.root_dir / "tech_documents/数据接口与来源/score_thresholds.json"
        score_thresholds: dict[str, Any] = {}
        if score_thresholds_path.exists():
            score_thresholds, _ = config_store.load_json(score_thresholds_path)
        bundle = {
            "policy_bundle_id": DEFAULT_POLICY_BUNDLE_ID,
            "strategy_version": actual_strategy_version,
            "rule_package_id": rule_package.get("package_id"),
            "rule_package_name": rule_package.get("package_name"),
            "rule_pack": rule_pack,
            "rule_pack_id": rule_pack["rule_pack_id"],
            "rule_pack_version": rule_pack["version"],
            "dispatch": dispatch,
            "dispatch_id": dispatch["dispatch_id"],
            "rule_pack_by_entity": rule_pack_by_entity,
            "runtime_stage": dispatch["runtime_stage"],
            "target_entity": dispatch["target_entity"],
            "content_format": dispatch["content_format"],
            "evidence_bundle_schema_version": EVIDENCE_BUNDLE_SCHEMA_VERSION,
            "runtime_record_schema_version": RUNTIME_SCHEMA_VERSION,
            "shared_contracts": rule_package.get("shared_contracts", {}),
            "source_evidence_policy": rule_package.get("source_evidence_policy", {}),
            "effect_status_mapping": rule_package.get("effect_status_mapping", []),
            "query_effect_aggregation": rule_package.get("query_effect_aggregation", []),
            "runtime_status_contract": rule_package.get("runtime_status_contract", {}),
            "decision_reason_codes": rule_package.get("decision_reason_codes", []),
            "strategy_id": (rule_package.get("strategy_binding") or {}).get(
                "strategy_id", "douyin_content_find_v1"
            ),
            "walk_strategy_id": walk_strategy.get("strategy_id"),
            "walk_strategy_version": walk_strategy.get("walk_strategy_version"),
            "walk_strategy_source_ref": walk_strategy.get("walk_strategy_source_ref"),
            "strategy_source_ref": {
                "file": str(rule_pack_path),
                "updated_at": rule_package.get("updated_at"),
                "content_sha256": rule_pack_hash,
                "generated_from": "p5_rule_pack_only",
            },
            "rule_pack_source_ref": {
                "file": str(rule_pack_path),
                "updated_at": rule_package.get("updated_at"),
                "content_sha256": rule_pack_hash,
                "generated_from": "local_product_json",
            },
            "policy_bundle_hash": rule_pack_hash,
            "score_thresholds": score_thresholds,
        }
        return bundle


def _strategy_version(rule_package: dict[str, Any]) -> str:
    binding_version = (rule_package.get("strategy_binding") or {}).get("strategy_version")
    return binding_version or "V1"


def _select_dispatch(
    rule_package: dict[str, Any],
    strategy_version: str,
    *,
    target_entity: str = "Content",
    content_format: str = "video",
    runtime_stage: str = "V1.0",
    platform: str = "douyin",
) -> dict[str, Any]:
    matches = _enabled_dispatches(
        rule_package,
        strategy_version,
        target_entity=target_entity,
        content_format=content_format,
        runtime_stage=runtime_stage,
        platform=platform,
    )
    return _assert_single_enabled_dispatch(matches, target_entity=target_entity, content_format=content_format)


def _enabled_dispatches(
    rule_package: dict[str, Any],
    strategy_version: str,
    *,
    target_entity: str,
    content_format: str,
    runtime_stage: str,
    platform: str,
) -> list[dict[str, Any]]:
    return [
        dispatch
        for dispatch in rule_package.get("rule_pack_dispatch", [])
        if dispatch.get("dispatch_enabled")
        and dispatch.get("platform") == platform
        and dispatch.get("runtime_stage") == runtime_stage
        and dispatch.get("strategy_version") == strategy_version
        and dispatch.get("target_entity") == target_entity
        and dispatch.get("content_format") == content_format
    ]


def _assert_single_enabled_dispatch(
    matches: list[dict[str, Any]],
    *,
    target_entity: str,
    content_format: str,
) -> dict[str, Any]:
    if len(matches) == 1:
        return matches[0]
    if not matches:
        raise ValueError(f"dispatch not found for {target_entity}/{content_format}")
    conflict_rule_pack_ids = [dispatch.get("rule_pack_id") for dispatch in matches]
    raise ContentAgentError(
        ErrorCode.CONFIG_RULE_PACK_DISPATCH_CONFLICT,
        f"CONFIG_RULE_PACK_DISPATCH_CONFLICT: multiple enabled dispatches for "
        f"{target_entity}/{content_format}: {conflict_rule_pack_ids}",
        {"target_entity": target_entity, "content_format": content_format,
         "conflict_rule_pack_ids": conflict_rule_pack_ids},
    )


def _build_rule_pack_by_entity(rule_package: dict[str, Any], strategy_version: str) -> dict[str, Any]:
    by_entity: dict[str, Any] = {}
    for dispatch in rule_package.get("rule_pack_dispatch", []):
        if not (
            dispatch.get("dispatch_enabled")
            and dispatch.get("platform") == "douyin"
            and dispatch.get("runtime_stage") == "V1.0"
            and dispatch.get("strategy_version") == strategy_version
        ):
            continue
        by_entity[dispatch["target_entity"]] = {
            "dispatch": dispatch,
            "rule_pack": _find_rule_pack_by_dispatch(rule_package, dispatch),
        }
    return by_entity


def _find_rule_pack_by_dispatch(rule_package: dict[str, Any], dispatch: dict[str, Any]) -> dict[str, Any]:
    matches = [
        rule_pack
        for rule_pack in rule_package.get("rule_packs", [])
        if rule_pack.get("enabled")
        and rule_pack.get("rule_pack_id") == dispatch.get("rule_pack_id")
        and rule_pack.get("version") == dispatch.get("rule_pack_version")
    ]
    if len(matches) != 1:
        raise ValueError(
            f"dispatch {dispatch.get('dispatch_id')} matched {len(matches)} enabled rule packs"
        )
    return matches[0]