life-echo/api/app/agents/memoir/orchestrator.py

"""
MemoirOrchestrator：按 segment 编排流水线，调用各 Specialist Agent。
负责：遍历 segments、按 category 聚合、调用 Specialist、更新 state；
持久化与章节生成由 process_category 回调完成。
"""

from __future__ import annotations

import time
from dataclasses import dataclass
from typing import Any, Callable, Dict, List, Set, Tuple

from app.agents.memoir.batch_phase1_prep import (
    STAGE_ALLOWED_SLOTS,
    run_batch_phase1_prep,
)
from app.agents.memoir.classification_agent import (
    ClassificationAgent,
)
from app.agents.memoir.classification_agent import (
    _detect_stage as detect_stage_from_keywords,
)
from app.agents.memoir.classification_agent import _looks_like_fragment_only
from app.agents.memoir.extraction_agent import ExtractionAgent, ExtractionResult
from app.agents.stage_constants import normalize_chapter_category, normalize_chat_stage
from app.agents.state_schema import MemoirStateSchema
from app.core.agent_logging import agent_span, agent_summary_enabled, log_agent_detail
from app.core.config import settings
from app.core.logging import get_logger
from app.features.conversation.models import Segment

logger = get_logger(__name__)


@dataclass
class PreparedMemoirBatches:
    """Explicit batching result: updated state + segments grouped by chapter category."""

    state: MemoirStateSchema
    category_to_segments: Dict[str, List[Segment]]
    #: segment id 在「LLM 判 none 且 extraction slots 为空」时加入；batch 级短路见 memoir_tasks
    segment_skip_story_ids: Set[str]
    #: 每个 segment → Phase 1 分类 chapter_category（持久化到 Segment.topic_category）
    segment_chapter_category: Dict[str, str]


class MemoirOrchestrator:
    """
    回忆录生成编排器。
    遍历 segments → ExtractionAgent → ClassificationAgent → 按 category 聚合 →
    调用 process_category 生成叙事并持久化。
    """

    def __init__(self) -> None:
        self.extraction_agent = ExtractionAgent()
        self.classification_agent = ClassificationAgent()

    def prepare_batches(
        self,
        *,
        segments: List[Segment],
        llm: Any,
        get_or_create_state: Callable[[], MemoirStateSchema],
        update_slot: Callable[[str, str, str, List[str]], MemoirStateSchema],
        llm_fast: Any | None = None,
    ) -> PreparedMemoirBatches:
        """
        遍历 segments：Extraction → slot 更新 → Classification → 按 category 分桶。
        不含锁与写章节/故事（由调用方显式执行）。

        ``llm_fast``：分类与抽取专用；未传时与 ``llm`` 相同（叙事/路由仍用 ``llm``）。
        """
        state = get_or_create_state()
        category_to_segments: Dict[str, List[Segment]] = {}
        segment_skip_story_ids: Set[str] = set()
        segment_chapter_category: Dict[str, str] = {}
        classify_extract_llm = llm_fast if llm_fast is not None else llm

        # 仅 MEMOIR_PHASE1_BATCH_LLM_ENABLED=true 时走批处理；关则与旧版一致逐段（含多段一批）
        use_batch = (
            bool(segments)
            and classify_extract_llm is not None
            and settings.memoir_phase1_batch_llm_enabled
        )
        if use_batch:
            try:
                return self._prepare_batches_via_batch_llm(
                    segments=segments,
                    state=state,
                    classify_extract_llm=classify_extract_llm,
                    update_slot=update_slot,
                )
            except Exception as e:
                logger.warning(
                    "MemoirOrchestrator.prepare_batches batch LLM 失败，回退逐段: {}",
                    e,
                )

        for segment in segments:
            text = segment.user_input_text or ""
            seg_t0 = time.perf_counter()
            initial_stage = detect_stage_from_keywords(
                text, state.current_stage or "childhood"
            )
            stage_slots_raw = state.slots.get(initial_stage, {}) or {}

            with agent_span(
                logger,
                "MemoirOrchestrator.ExtractionAgent.extract",
                segment_id=segment.id,
            ):
                result: ExtractionResult = self.extraction_agent.extract(
                    user_message=text,
                    current_stage=state.current_stage or "childhood",
                    stage_slots=stage_slots_raw,
                    llm=classify_extract_llm,
                )
            detected_stage = result.detected_stage
            for slot_name, snippet in result.slots.items():
                state = update_slot(detected_stage, slot_name, snippet, [segment.id])

            with agent_span(
                logger,
                "MemoirOrchestrator.ClassificationAgent.classify",
                segment_id=segment.id,
            ):
                classify_result = self.classification_agent.classify(
                    text=text,
                    fallback_stage=detected_stage,
                    llm=classify_extract_llm,
                    segment_id=segment.id,
                )
            chapter_category = classify_result.category
            if (not result.slots) and classify_result.llm_said_none:
                segment_skip_story_ids.add(str(segment.id))
            segment_chapter_category[str(segment.id)] = chapter_category

            if agent_summary_enabled():
                logger.info(
                    "MemoirOrchestrator.segment segment_id={} text_len={} "
                    "detected_stage={} category={} segment_total_ms={:.2f}",
                    segment.id,
                    len(text),
                    detected_stage,
                    chapter_category,
                    (time.perf_counter() - seg_t0) * 1000,
                )
            log_agent_detail(
                logger,
                "MemoirOrchestrator.segment_done segment_id={} slots={}",
                segment.id,
                list((result.slots or {}).keys()),
            )
            category_to_segments.setdefault(chapter_category, []).append(segment)

        return PreparedMemoirBatches(
            state=state,
            category_to_segments=category_to_segments,
            segment_skip_story_ids=segment_skip_story_ids,
            segment_chapter_category=segment_chapter_category,
        )

    def _prepare_batches_via_batch_llm(
        self,
        *,
        segments: List[Segment],
        state: MemoirStateSchema,
        classify_extract_llm: Any,
        update_slot: Callable[[str, str, str, List[str]], MemoirStateSchema],
    ) -> PreparedMemoirBatches:
        category_to_segments: Dict[str, List[Segment]] = {}
        segment_skip_story_ids: Set[str] = set()
        segment_chapter_category: Dict[str, str] = {}

        by_id = run_batch_phase1_prep(segments, state, classify_extract_llm)

        for segment in segments:
            text = segment.user_input_text or ""
            seg_t0 = time.perf_counter()
            row = by_id[str(segment.id)]

            result_slots = dict(row.slots)
            fb = state.current_stage or "childhood"

            if not result_slots:
                detected_stage = normalize_chat_stage(fb, fb)
            else:
                detected_stage = normalize_chat_stage(row.detected_stage, fb)

            allowed = STAGE_ALLOWED_SLOTS.get(detected_stage, frozenset())
            result_slots = {k: v for k, v in result_slots.items() if k in allowed}
            if not result_slots:
                detected_stage = normalize_chat_stage(fb, fb)

            with agent_span(
                logger,
                "MemoirOrchestrator.BatchPhase1Prep.apply",
                segment_id=segment.id,
            ):
                for slot_name, snippet in result_slots.items():
                    state = update_slot(
                        detected_stage, slot_name, snippet, [segment.id]
                    )

            if _looks_like_fragment_only(text):
                chapter_category = "summary"
                llm_said_none = False
            else:
                raw_cat = (row.chapter_category_raw or "").strip().lower()
                if raw_cat == "none":
                    chapter_category = "summary"
                    llm_said_none = True
                else:
                    chapter_category = normalize_chapter_category(
                        row.chapter_category_raw,
                        "summary",
                    )
                    llm_said_none = False

            if (not result_slots) and llm_said_none:
                segment_skip_story_ids.add(str(segment.id))
            segment_chapter_category[str(segment.id)] = chapter_category

            if agent_summary_enabled():
                logger.info(
                    "MemoirOrchestrator.segment(batch) segment_id={} text_len={} "
                    "detected_stage={} category={} segment_total_ms={:.2f}",
                    segment.id,
                    len(text),
                    detected_stage,
                    chapter_category,
                    (time.perf_counter() - seg_t0) * 1000,
                )
            log_agent_detail(
                logger,
                "MemoirOrchestrator.segment_done(batch) segment_id={} slots={}",
                segment.id,
                list(result_slots.keys()),
            )
            category_to_segments.setdefault(chapter_category, []).append(segment)

        return PreparedMemoirBatches(
            state=state,
            category_to_segments=category_to_segments,
            segment_skip_story_ids=segment_skip_story_ids,
            segment_chapter_category=segment_chapter_category,
        )

    def run(
        self,
        *,
        segments: List[Segment],
        llm: Any,
        user_profile: str = "",
        user_birth_year: Any = None,
        get_or_create_state: Callable[[], MemoirStateSchema],
        update_slot: Callable[[str, str, str, List[str]], MemoirStateSchema],
        acquire_lock: Callable[[str], bool],
        release_lock: Callable[[str], None],
        process_category: Callable[
            [
                str,
                List[Segment],
                MemoirStateSchema,
                str,
                Any,
                Any,
            ],
            Tuple[Any, bool],
        ],
        raise_retry: Callable[[], None],
        llm_fast: Any | None = None,
    ) -> Tuple[Set[str], int]:
        """
        执行回忆录流水线。
        process_category(category, segments, state, user_profile, user_birth_year, llm)
        返回 (chapter, has_images_to_generate)。
        返回 (chapters_to_enqueue, processed_count)。
        raise_retry 用于锁竞争时抛出 Celery retry。
        """
        prepared = self.prepare_batches(
            segments=segments,
            llm=llm,
            llm_fast=llm_fast,
            get_or_create_state=get_or_create_state,
            update_slot=update_slot,
        )
        state = prepared.state
        chapters_to_enqueue: Set[str] = set()
        category_to_segments = prepared.category_to_segments

        # 按 category 调用 process_category：叙事生成、持久化、封面入队标记
        for chapter_category, category_segments in category_to_segments.items():
            if not acquire_lock(chapter_category):
                logger.warning(
                    "章节锁竞争: category={}, 延迟重试",
                    chapter_category,
                )
                raise_retry()

            try:
                chapter, has_images = process_category(
                    chapter_category,
                    category_segments,
                    state,
                    user_profile,
                    user_birth_year,
                    llm,
                )
                if chapter and has_images:
                    chapters_to_enqueue.add(chapter.id)
            finally:
                release_lock(chapter_category)

        return chapters_to_enqueue, len(segments)