import { listPinAssets, type PinAsset } from "./memory-assets.js";
import * as envConfig from "./env-config.js";
import { buildStableContextSections } from "./context-composer-stable.js";
import {
  buildAssociativeNestEntityFallbackQuery,
  buildScopedEntityFallbackQuery,
  buildStableQuery,
  buildStylePreferenceFallbackQuery,
  buildTaskQuery,
  formatLatestCheckpointHeadline,
} from "./context-composer-queries.js";
import { cleanText, dedupeText } from "./context-composer-text.js";
import {
  buildTaskResultSections,
} from "./context-composer-task-results.js";
import type { RetrievalContext, RetrievalResult } from "./retriever.js";
import type { EssentialContext, ResumeContextResponse, SessionCheckpointRecord } from "./session-schema.js";
import { ResumeContextRequestSchema, ResumeContextResponseSchema } from "./session-schema.js";
import { formatCheckpointRecallSummary } from "./session-output.js";
import {
  STRONG_WORKFLOW_CUE_TERMS,
  looksLikeContinuityTask,
  looksLikeRecallOnlyTask,
  looksLikeStyleTask,
} from "./term-registry.js";
import { synthesizeSection } from "./result-synthesizer.js";
import type { LLMClient } from "./llm-client.js";
import { collapseResults, estimateTokens, type CollapseInput } from "./context-collapse-renderer.js";
import { filterByRelevance } from "./post-retrieval-filter.js";
import { reconstruct as runReconstruction } from "./context-reconstructor.js";
import { parseNarrative, isNarrativeModeEnabled } from "./narrative-schema.js";
type ResumeCategory = "profile" | "preferences" | "entities" | "patterns" | "cases";

interface ResumeRetriever {
  retrieve(context: RetrievalContext): Promise<RetrievalResult[]>;
  /** Optional: 影子记录 resume reconstruction 引用的记忆（injection-vs-use）。未实现则 no-op。 */
  recordReconstructionUsage?(citedIds: string[]): Promise<void>;
}

interface CheckpointLookup {
  getLatest(query?: { sessionId?: string; scope?: string }): Promise<SessionCheckpointRecord | null>;
}

export interface ResumeContextDeps {
  retriever: ResumeRetriever;
  checkpointStore: CheckpointLookup;
  listPins?: (limit?: number) => Array<PinAsset & { path: string }>;
  /** Optional LLM client for Tier 3.5 result synthesis. */
  llm?: LLMClient | null;
}

async function retrieveCandidates(
  retriever: ResumeRetriever,
  params: {
    category?: ResumeCategory;
    query: string;
    limit: number;
    scope?: string;
  },
): Promise<RetrievalResult[]> {
  const { category, query, limit, scope } = params;

  if (!scope) {
    return retriever.retrieve({
      query,
      limit,
      ...(category ? { category } : {}),
      source: "auto-recall",
    });
  }

  const [scoped, global] = await Promise.all([
    retriever.retrieve({
      query,
      limit,
      ...(category ? { category } : {}),
      scopeFilter: [scope],
      source: "auto-recall",
    }),
    retriever.retrieve({
      query,
      limit: Math.min(10, limit * 2),
      ...(category ? { category } : {}),
      source: "auto-recall",
    }),
  ]);

  const seen = new Set<string>();
  const merged: RetrievalResult[] = [];
  for (const result of [...scoped, ...global]) {
    if (seen.has(result.entry.id)) continue;
    seen.add(result.entry.id);
    merged.push(result);
    if (merged.length >= Math.min(10, limit * 2)) break;
  }
  return merged;
}

function mergeRetrievalResults(resultSets: RetrievalResult[][], limit: number): RetrievalResult[] {
  const seen = new Set<string>();
  const merged: RetrievalResult[] = [];
  for (const set of resultSets) {
    for (const result of set) {
      if (seen.has(result.entry.id)) continue;
      seen.add(result.entry.id);
      merged.push(result);
      if (merged.length >= limit) return merged;
    }
  }
  return merged;
}

async function resolveLatestCheckpoint(
  checkpointStore: CheckpointLookup,
  params: {
    includeLatestCheckpoint: boolean;
    sessionId?: string;
    scope?: string;
  },
): Promise<SessionCheckpointRecord | null> {
  if (!params.includeLatestCheckpoint) return null;

  if (params.sessionId) {
    const bySession = await checkpointStore.getLatest({ sessionId: params.sessionId });
    if (bySession) return bySession;
  }

  if (params.scope) {
    return checkpointStore.getLatest({ scope: params.scope });
  }

  return null;
}

function buildSummary(params: {
  stableContext: string[];
  relevantPatterns: string[];
  recentCases: string[];
  latestCheckpoint: SessionCheckpointRecord | null;
}): string {
  const parts: string[] = [];

  if (params.latestCheckpoint) {
    parts.push(
      formatLatestCheckpointHeadline(
        params.latestCheckpoint.sessionId,
        params.latestCheckpoint.updatedAt,
        formatCheckpointRecallSummary(params.latestCheckpoint),
      ),
    );
  }

  if (params.stableContext.length > 0) {
    parts.push(`Stable context: ${params.stableContext.slice(0, 2).map((item) => cleanText(item, 120)).join(" | ")}`);
  }

  parts.push(
    `Loaded ${params.stableContext.length} stable context item(s), ${params.relevantPatterns.length} pattern(s), and ${params.recentCases.length} case(s).`,
  );

  return cleanText(parts.join(" "), 800);
}

// ---------------------------------------------------------------------------
// CC-8: Post-Compact Reconstruction — assemble essential context
// ---------------------------------------------------------------------------

const ESSENTIAL_CONTEXT_TOKEN_BUDGET = 2000;

function buildEssentialContext(params: {
  pinAssets: Array<PinAsset & { path: string }>;
  patternResults: RetrievalResult[];
  latestCheckpoint: SessionCheckpointRecord | null;
}): EssentialContext | undefined {
  let tokensUsed = 0;

  // 1. Pinned memories: take up to 3 most recent pins' summaries
  const pinnedMemories: string[] = [];
  for (const pin of params.pinAssets.slice(0, 3)) {
    const text = pin.summary || pin.title;
    const cost = estimateTokens(text);
    if (tokensUsed + cost > ESSENTIAL_CONTEXT_TOKEN_BUDGET) break;
    pinnedMemories.push(text);
    tokensUsed += cost;
  }

  // 2. Active patterns: take top 1-2 by score
  const activePatterns: string[] = [];
  const sortedPatterns = [...params.patternResults].sort((a, b) => b.score - a.score);
  for (const pattern of sortedPatterns.slice(0, 2)) {
    const text = cleanText(pattern.entry.text, 200);
    const cost = estimateTokens(text);
    if (tokensUsed + cost > ESSENTIAL_CONTEXT_TOKEN_BUDGET) break;
    activePatterns.push(text);
    tokensUsed += cost;
  }

  // 3. Open loops from latest checkpoint
  const openLoops: string[] = [];
  if (params.latestCheckpoint?.openLoops) {
    for (const loop of params.latestCheckpoint.openLoops.slice(0, 3)) {
      const cost = estimateTokens(loop);
      if (tokensUsed + cost > ESSENTIAL_CONTEXT_TOKEN_BUDGET) break;
      openLoops.push(loop);
      tokensUsed += cost;
    }
  }

  // Return undefined if nothing was collected
  if (pinnedMemories.length === 0 && activePatterns.length === 0 && openLoops.length === 0) {
    return undefined;
  }

  return {
    ...(pinnedMemories.length > 0 ? { pinnedMemories } : {}),
    ...(activePatterns.length > 0 ? { activePatterns } : {}),
    ...(openLoops.length > 0 ? { openLoops } : {}),
  };
}

// ---------------------------------------------------------------------------
// MP-3: Ultra-Light Wake-up — <300 token resume for low-budget terminals
// ---------------------------------------------------------------------------

const LIGHT_TOKEN_BUDGET = 300;
const LIGHT_MEMORY_CHAR_LIMIT = 200; // ~50 tokens

export interface LightResumeResult {
  text: string;
  resolvedScope?: string;
  generatedAt: string;
}

export async function composeLightResumeContext(
  deps: ResumeContextDeps,
  rawInput: unknown,
): Promise<LightResumeResult> {
  const input = ResumeContextRequestSchema.parse(rawInput);
  const latestCheckpoint = await resolveLatestCheckpoint(deps.checkpointStore, {
    includeLatestCheckpoint: true,
    sessionId: input.sessionId,
    scope: input.scope,
  });

  const resolvedScope = input.scope || latestCheckpoint?.resolvedScope;

  // Retrieve top 3 stable memories (profile + preferences + entities combined)
  const stableResults = await retrieveCandidates(deps.retriever, {
    query: input.task || "identity key facts and preferences",
    limit: 3,
    scope: resolvedScope,
  });

  // Pin assets provide high-signal stable context
  const pinAssets = (deps.listPins || listPinAssets)(3);

  const parts: string[] = [];
  let tokensUsed = 0;

  // 1. Checkpoint summary (1 sentence, ~30 tokens)
  if (latestCheckpoint) {
    const cpLine = `Last session (${latestCheckpoint.sessionId}): ${cleanText(latestCheckpoint.summary, 120)}`;
    parts.push(cpLine);
    tokensUsed += estimateTokens(cpLine);
  }

  // 2. Top pinned memories (most reliable stable context)
  if (pinAssets.length > 0) {
    for (const pin of pinAssets) {
      if (tokensUsed >= LIGHT_TOKEN_BUDGET - 40) break;
      const line = `- ${cleanText(pin.summary || pin.title, LIGHT_MEMORY_CHAR_LIMIT)}`;
      parts.push(line);
      tokensUsed += estimateTokens(line);
    }
  }

  // 3. Top stable memories (fill remaining budget)
  if (stableResults.length > 0 && tokensUsed < LIGHT_TOKEN_BUDGET - 40) {
    const pinTexts = new Set(pinAssets.map(p => (p.summary || p.title).toLowerCase()));
    for (const r of stableResults) {
      if (tokensUsed >= LIGHT_TOKEN_BUDGET - 40) break;
      const textLower = r.entry.text.toLowerCase().slice(0, 100);
      if (pinTexts.has(textLower)) continue;
      const line = `- ${cleanText(r.entry.text, LIGHT_MEMORY_CHAR_LIMIT)}`;
      parts.push(line);
      tokensUsed += estimateTokens(line);
    }
  }

  // 4. Upgrade hint
  parts.push("\nFor complete context, call resume_context(mode='full').");

  return {
    text: parts.join("\n"),
    resolvedScope,
    generatedAt: new Date().toISOString(),
  };
}

export async function composeResumeContext(
  deps: ResumeContextDeps,
  rawInput: unknown,
): Promise<ResumeContextResponse> {
  const input = ResumeContextRequestSchema.parse(rawInput);
  const latestCheckpoint = await resolveLatestCheckpoint(deps.checkpointStore, {
    includeLatestCheckpoint: input.includeLatestCheckpoint,
    sessionId: input.sessionId,
    scope: input.scope,
  });

  const resolvedScope = input.scope || latestCheckpoint?.resolvedScope;
  const taskSeed = input.task || latestCheckpoint?.task || latestCheckpoint?.summary;
  const stableLimit = input.limitPerSection;
  const taskLimit = input.limitPerSection;
  const styleFocusedTask = looksLikeStyleTask(taskSeed) || input.profile === "writing";
  const recallOnlyTask = looksLikeRecallOnlyTask(taskSeed) && styleFocusedTask;

  const preferenceQueries = dedupeText([
    buildStableQuery("preferences", taskSeed),
    ...(styleFocusedTask ? [buildStylePreferenceFallbackQuery(taskSeed)] : []),
  ], 2);
  const entityQueries = dedupeText([
    buildStableQuery("entities", taskSeed),
    ...(resolvedScope ? [buildScopedEntityFallbackQuery(resolvedScope, taskSeed)] : []),
    ...(!resolvedScope ? [buildAssociativeNestEntityFallbackQuery(taskSeed)] : []),
  ], 3);

  const [profileResults, preferenceResultSets, entityResultSets, patternResults, caseResults] = await Promise.all([
    retrieveCandidates(deps.retriever, {
      category: "profile",
      query: buildStableQuery("profile", taskSeed),
      limit: Math.max(2, stableLimit),
      scope: resolvedScope,
    }),
    Promise.all(preferenceQueries.map((query) =>
      retrieveCandidates(deps.retriever, {
        category: "preferences",
        query,
        limit: Math.max(2, stableLimit),
          scope: resolvedScope,
        })
    )),
    Promise.all(entityQueries.map((query) =>
      retrieveCandidates(deps.retriever, {
        category: "entities",
        query,
        limit: Math.max(2, stableLimit),
        scope: resolvedScope,
      })
    )),
    retrieveCandidates(deps.retriever, {
      category: "patterns",
      query: buildTaskQuery("patterns", taskSeed),
      limit: taskLimit,
      scope: resolvedScope,
    }),
    retrieveCandidates(deps.retriever, {
      category: "cases",
      query: buildTaskQuery("cases", taskSeed),
      limit: taskLimit,
      scope: resolvedScope,
    }),
  ]);
  const preferenceResults = mergeRetrievalResults(preferenceResultSets, Math.max(4, stableLimit * 3));
  const entityResults = mergeRetrievalResults(entityResultSets, Math.max(4, stableLimit * 3));

  // E-2: Post-retrieval LLM relevance filter for pattern/case results.
  // Only fires when LLM is available; stable context (profile/preferences/entities) skipped
  // because those are identity-level and already curated by the selection pipeline.
  const llm = deps.llm ?? null;
  const [filteredPatterns, filteredCases] = llm && taskSeed
    ? await Promise.all([
        filterByRelevance(patternResults, taskSeed, llm),
        filterByRelevance(caseResults, taskSeed, llm),
      ])
    : [patternResults, caseResults];

  const continuityTask = looksLikeContinuityTask(taskSeed);
  const pinAssets = (deps.listPins || listPinAssets)(Math.max(4, stableLimit * 2));
  const {
    preferenceContext,
    stableContext,
  } = buildStableContextSections({
    profileResults,
    preferenceResults,
    entityResults,
    pinAssets,
    latestCheckpoint,
    taskSeed,
    scope: resolvedScope,
    stableLimit,
    styleFocusedTask,
  });

  const { relevantPatterns, recentCases } = await buildTaskResultSections({
    retrieveCandidates: ({ category, query, limit, scope }) =>
      retrieveCandidates(deps.retriever, {
        ...(category ? { category } : {}),
        query,
        limit,
        scope,
      }),
    patternResults: filteredPatterns,
    caseResults: filteredCases,
    continuityTask,
    hasLatestCheckpoint: Boolean(latestCheckpoint),
    taskLimit,
    taskSeed,
    scope: resolvedScope,
    strongWorkflowCueTerms: STRONG_WORKFLOW_CUE_TERMS,
  });

  // Tier 3.5: Optionally synthesize sections into coherent narratives via LLM.
  // Only activates when RECALLNEST_SYNTHESIZE=true and llm is provided.
  const queryHint = taskSeed || resolvedScope || "general";
  const [synthStable, synthPatterns, synthCases] = await Promise.all([
    synthesizeSection(stableContext, queryHint, llm),
    synthesizeSection(relevantPatterns, queryHint, llm),
    synthesizeSection(recentCases, queryHint, llm),
  ]);

  // HP-narrative: Group recalled items by life period for narrative context
  let narrativeGroups: Array<{ period: string; items: string[] }> | undefined;
  if (isNarrativeModeEnabled()) {
    const allNarrativeResults = [...profileResults, ...preferenceResults, ...entityResults, ...filteredPatterns, ...filteredCases];
    const periodMap = new Map<string, string[]>();
    for (const r of allNarrativeResults) {
      const narrative = parseNarrative(r.entry.metadata);
      if (!narrative) continue;
      const items = periodMap.get(narrative.lifePeriodLabel) ?? [];
      items.push(cleanText(r.entry.text, 120));
      periodMap.set(narrative.lifePeriodLabel, items);
    }
    if (periodMap.size > 0) {
      narrativeGroups = [...periodMap.entries()]
        .map(([period, items]) => ({ period, items: items.slice(0, 5) }));
    }
  }

  // CC-7: Collapse rendering — build mixed-granularity view of all recalled items.
  // Gathers all retrieval results, deduplicates, and renders at L0/L1/L2 based on score.
  const allResults: RetrievalResult[] = [];
  const seenIds = new Set<string>();
  for (const r of [...profileResults, ...preferenceResults, ...entityResults, ...filteredPatterns, ...filteredCases]) {
    if (!seenIds.has(r.entry.id)) {
      seenIds.add(r.entry.id);
      allResults.push(r);
    }
  }
  const collapseInput: CollapseInput[] = allResults.map(r => ({
    entryId: r.entry.id,
    text: r.entry.text,
    metadata: r.entry.metadata,
    score: r.score,
    timestamp: r.entry.timestamp,
  }));
  const collapsedItems = collapseInput.length > 0
    ? collapseResults(collapseInput)
    : undefined;

  // CC-8: Build essential context from pinned memories, top patterns, and open loops.
  const essentialContext = buildEssentialContext({
    pinAssets,
    patternResults: filteredPatterns,
    latestCheckpoint,
  });

  // Phase 4: Constructive retrieval reconstruction for resume context
  // Pass checkpoint openLoops/nextActions/scope into reconstruction prompt
  let reconstructedContext: string | undefined;
  let reconstructionConfidence: number | undefined;
  let reconstructionContradictions: Array<{ memoryIds: [string, string]; description: string }> | undefined;
  const constructiveFlag = envConfig.constructiveRetrieval();
  if (constructiveFlag && deps.llm?.isAvailable?.()) {
    const allReconResults = [...profileResults, ...preferenceResults, ...entityResults, ...filteredPatterns, ...filteredCases];
    if (allReconResults.length >= 3) {
      try {
        const taskQuery = latestCheckpoint?.summary ?? taskSeed ?? "general context";
        const checkpointContext = latestCheckpoint ? {
          openLoops: latestCheckpoint.openLoops,
          nextActions: latestCheckpoint.nextActions,
          scope: resolvedScope,
        } : resolvedScope ? { scope: resolvedScope } : undefined;
        const recon = await runReconstruction(
          { query: taskQuery, results: allReconResults, mode: "resume", maxTokens: 600, checkpointContext },
          deps.llm,
        );
        if (recon.reconstructed) {
          reconstructedContext = recon.reconstructed;
          reconstructionConfidence = recon.confidence;
        }
        if (recon.contradictions.length > 0) {
          reconstructionContradictions = recon.contradictions;
        }
        // Shadow usage：resume reconstruction 真正引用的记忆记 use 计数（不动 ranking/forget）。
        if (recon.sources.length > 0) {
          await deps.retriever.recordReconstructionUsage?.(recon.sources.map(s => s.id));
        }
      } catch { /* silent fallback */ }
    }
  }

  const response = {
    summary: buildSummary({
      stableContext: synthStable,
      relevantPatterns: synthPatterns,
      recentCases: synthCases,
      latestCheckpoint,
    }),
    resolvedScope,
    stableContext: synthStable,
    relevantPatterns: synthPatterns,
    recentCases: synthCases,
    collapsedItems: collapsedItems && collapsedItems.length > 0 ? collapsedItems : undefined,
    essentialContext,
    latestCheckpoint: latestCheckpoint
      ? {
        sessionId: latestCheckpoint.sessionId,
        resolvedScope: latestCheckpoint.resolvedScope,
        summary: formatCheckpointRecallSummary(latestCheckpoint),
        updatedAt: latestCheckpoint.updatedAt,
      }
      : undefined,
    injectionHint: "user_attachment" as const,
    ephemeral: true,
    responseMode: recallOnlyTask ? "recall-only" as const : "default" as const,
    responseGuidance: recallOnlyTask
      ? (
          stableContext.length <= 1
            ? "Recall-only mode: answer from the recalled stable context item only. Restate it briefly and do not expand into extra rules, examples, or local writing docs unless the user explicitly asks."
            : "Recall-only mode: answer only from the recalled stable context items. Keep the reply brief and do not expand into extra rules, examples, or local writing docs unless the user explicitly asks."
        )
      : undefined,
    reconstructedContext,
    reconstructionConfidence,
    reconstructionContradictions,
    narrativeGroups,
    generatedAt: new Date().toISOString(),
  };

  return ResumeContextResponseSchema.parse(response);
}