NODEDC_1C/llm_normalizer/backend/tests/assistantWave18ManualCommentsRegression.test.ts

import fs from "node:fs";
import path from "node:path";
import { describe, expect, it } from "vitest";
import { AddressQueryService } from "../src/services/addressQueryService";
import { resolveAssistantOrchestrationDecision } from "../src/services/assistantService";
import { resolveAddressIntent } from "../src/services/addressIntentResolver";

type ManualCaseDecision =
  | "candidate_for_implementation"
  | "needs_dialog_policy_fix"
  | "needs_routing_extension"
  | "bad_test_case";

interface AnnotationContext {
  question_text?: string;
}

interface AnnotationRecord {
  run_id: string;
  case_id: string;
  manual_case_decision: ManualCaseDecision;
  resolved: boolean;
  context?: AnnotationContext | null;
}

interface ManualCase {
  runId: string;
  caseId: string;
  decision: ManualCaseDecision;
  question: string;
  resolved: boolean;
}

const MANUAL_CASE_KEYS = [
  "assistant-stage1-UMKkFYfg2L::AUTO-003",
  "assistant-stage1-UMKkFYfg2L::AUTO-007",
  "assistant-stage1-UMKkFYfg2L::AUTO-009",
  "assistant-stage1-UMKkFYfg2L::AUTO-012",
  "assistant-stage1-UMKkFYfg2L::AUTO-015",
  "assistant-stage1-UMKkFYfg2L::AUTO-017",
  "assistant-stage1-ywEyJgFkC4::AUTO-002",
  "assistant-stage1-ywEyJgFkC4::AUTO-004",
  "assistant-stage1-ywEyJgFkC4::AUTO-005",
  "assistant-stage1-ywEyJgFkC4::AUTO-006",
  "assistant-stage1-ywEyJgFkC4::AUTO-009",
  "assistant-stage1-ywEyJgFkC4::AUTO-013",
  "assistant-stage1-ywEyJgFkC4::AUTO-014",
  "assistant-stage1-ywEyJgFkC4::AUTO-015",
  "assistant-stage1-ZL97weIIRG::AUTO-005",
  "assistant-stage1-ZL97weIIRG::AUTO-008",
  "assistant-stage1-ZL97weIIRG::AUTO-009",
  "assistant-stage1-ZL97weIIRG::AUTO-010"
] as const;

const FORMER_UNKNOWN_INTENTS = new Set([
  "assistant-stage1-ywEyJgFkC4::AUTO-002",
  "assistant-stage1-ywEyJgFkC4::AUTO-009",
  "assistant-stage1-ywEyJgFkC4::AUTO-013",
  "assistant-stage1-ywEyJgFkC4::AUTO-015",
  "assistant-stage1-ZL97weIIRG::AUTO-009",
  "assistant-stage1-ZL97weIIRG::AUTO-010"
]);

function textMojibakeScore(value: string): number {
  const lower = value.toLowerCase();
  let score = 0;
  const badFragments = ["рџ", "р°", "сѓ", "с‚", "рµ", "рё", "с€", "с‡", "сЏ", "сЊ", "с‹", "с“", "вђ", "в€"];
  for (const fragment of badFragments) {
    if (lower.includes(fragment)) {
      score -= 4;
    }
  }
  const cyrillic = value.match(/[А-Яа-яЁё]/g)?.length ?? 0;
  score += cyrillic;
  const replacementCount = value.match(/<2F>/g)?.length ?? 0;
  score -= replacementCount * 3;
  return score;
}

function decodeUtf8FromWin1251Mojibake(value: string): string {
  try {
    const bytes = Uint8Array.from(Array.from(value).map((char) => char.charCodeAt(0) & 0xff));
    const decoded = Buffer.from(bytes).toString("utf8");
    return textMojibakeScore(decoded) > textMojibakeScore(value) ? decoded : value;
  } catch {
    return value;
  }
}

function decodeUtf8FromLatin1Mojibake(value: string): string {
  try {
    const decoded = Buffer.from(value, "latin1").toString("utf8");
    return textMojibakeScore(decoded) > textMojibakeScore(value) ? decoded : value;
  } catch {
    return value;
  }
}

function repairTextMojibake(value: string): string {
  const fromWin1251 = decodeUtf8FromWin1251Mojibake(value);
  return decodeUtf8FromLatin1Mojibake(fromWin1251);
}

function buildManualCasesFromAnnotations(): ManualCase[] {
  const filePath = path.resolve(__dirname, "../../data/autorun_annotations/annotations.json");
  const rows = JSON.parse(fs.readFileSync(filePath, "utf8")) as AnnotationRecord[];
  const byKey = new Map<string, AnnotationRecord>();

  for (const row of rows) {
    const key = `${row.run_id}::${row.case_id}`;
    if (MANUAL_CASE_KEYS.includes(key as (typeof MANUAL_CASE_KEYS)[number])) {
      byKey.set(key, row);
    }
  }

  const result: ManualCase[] = [];
  for (const key of MANUAL_CASE_KEYS) {
    const row = byKey.get(key);
    if (!row) {
      throw new Error(`Missing annotation for ${key}`);
    }
    const rawQuestion = String(row.context?.question_text ?? "").trim();
    if (!rawQuestion) {
      throw new Error(`Missing question_text for ${key}`);
    }
    result.push({
      runId: row.run_id,
      caseId: row.case_id,
      decision: row.manual_case_decision,
      question: repairTextMojibake(rawQuestion),
      resolved: row.resolved === true
    });
  }
  return result;
}

const MANUAL_WAVE18_CASES = buildManualCasesFromAnnotations();
const MANUAL_LIVE_ASSERT_CASES = MANUAL_WAVE18_CASES.filter((entry) => !entry.resolved);

describe("wave18 manual comments regressions", { timeout: 120000 }, () => {
  it("keeps manual-comment prompts in address lane (no capability/data-scope drift)", () => {
    for (const entry of MANUAL_WAVE18_CASES) {
      const decision = resolveAssistantOrchestrationDecision({
        rawUserMessage: entry.question,
        effectiveAddressUserMessage: entry.question,
        followupContext: null,
        llmPreDecomposeMeta: null,
        useMock: false
      });

      expect(decision.runAddressLane, `${entry.runId} ${entry.caseId}`).toBe(true);
      expect(decision.livingMode, `${entry.runId} ${entry.caseId}`).toBe("address_data");
      expect(String(decision.toolGateReason), `${entry.runId} ${entry.caseId}`).not.toBe("assistant_capability_query_detected");
      expect(String(decision.toolGateReason), `${entry.runId} ${entry.caseId}`).not.toBe("assistant_data_scope_query_detected");
    }
  });

  it("resolves previously-unknown manual prompts to supported intents", () => {
    for (const entry of MANUAL_WAVE18_CASES) {
      const key = `${entry.runId}::${entry.caseId}`;
      if (!FORMER_UNKNOWN_INTENTS.has(key)) {
        continue;
      }
      const intent = resolveAddressIntent(entry.question);
      expect(intent.intent, key).not.toBe("unknown");
    }
  });

  it(
    "returns handled address responses for manual-comment prompts without legacy rigid unsupported template",
    async () => {
      const service = new AddressQueryService();

      for (const entry of MANUAL_LIVE_ASSERT_CASES) {
        const result = await service.tryHandle(entry.question);
        const reply = String(result?.reply_text ?? "");

        expect(result, `${entry.runId} ${entry.caseId}`).not.toBeNull();
        expect(result?.handled, `${entry.runId} ${entry.caseId}`).toBe(true);
        expect(reply, `${entry.runId} ${entry.caseId}`).not.toMatch(
          /Сейчас этот тип вопроса вне поддерживаемого контура адресного режима/iu
        );
      }
    },
    120_000
  );
});