Починить восстановление кириллицы в автопрогонах

2026-05-10 08:38:52 +03:00
parent 9b02083493
commit 3be06b5f93
18 changed files with 618 additions and 74 deletions
@@ -0,0 +1,35 @@
+import { describe, expect, it } from "vitest";
+import { repairAddressMojibakeText } from "../src/services/addressTextRepair";
+
+describe("address text mojibake repair", () => {
+  it("preserves C1 bytes when repairing old autorun Russian text", () => {
+    const damagedTitle = String.fromCharCode(
+      0x420, 0x2018, 0x420, 0x45b, 0x420, 0x203a, 0x420, 0xac, 0x420, 0x401, 0x420, 0x45b, 0x420,
+      0x2122, 0x20, 0x420, 0x45b, 0x420, 0x2018, 0x420, 0xa9, 0x420, 0x98, 0x420, 0x2122, 0x20,
+      0x420, 0xa0, 0x421, 0x453, 0x421, 0x2021, 0x420, 0x405, 0x420, 0xb0, 0x421, 0x40f, 0x20,
+      0x421, 0x403, 0x420, 0xb5, 0x421, 0x403, 0x421, 0x403, 0x420, 0x451, 0x421, 0x40f, 0x20,
+      0x31, 0x36, 0x2e, 0x30, 0x34, 0x2e, 0x32, 0x30, 0x32, 0x36, 0x2c, 0x20, 0x32, 0x31,
+      0x3a, 0x32, 0x36, 0x3a, 0x30, 0x36
+    );
+    const damagedAlternative = String.fromCharCode(
+      0x420, 0x452, 0x420, 0x203a, 0x420, 0xac, 0x420, 0x45e, 0x420, 0x2022, 0x420, 0xa0, 0x420,
+      0x45c, 0x420, 0x452, 0x420, 0x45e, 0x420, 0x98, 0x420, 0x2019, 0x420, 0x452
+    );
+
+    expect(repairAddressMojibakeText(damagedTitle)).toBe(
+      "\u0411\u041e\u041b\u042c\u0428\u041e\u0419 \u041e\u0411\u0429\u0418\u0419 \u0420\u0443\u0447\u043d\u0430\u044f \u0441\u0435\u0441\u0441\u0438\u044f 16.04.2026, 21:26:06"
+    );
+    expect(repairAddressMojibakeText(damagedAlternative)).toBe(
+      "\u0410\u041b\u042c\u0422\u0415\u0420\u041d\u0410\u0422\u0418\u0412\u0410"
+    );
+  });
+
+  it("repairs already lossy known replacement fragments", () => {
+    expect(repairAddressMojibakeText("\u0410\u041b\u042c\u0422\u0415\u0420\u041d\u0410\u0422\uFFFD?\u0412\u0410")).toBe(
+      "\u0410\u041b\u042c\u0422\u0415\u0420\u041d\u0410\u0422\u0418\u0412\u0410"
+    );
+    expect(repairAddressMojibakeText("\u0411\u041e\u041b\u042c\u0428\u041e\u0419 \u041e\u0411\u0429\uFFFD\u0419")).toBe(
+      "\u0411\u041e\u041b\u042c\u0428\u041e\u0419 \u041e\u0411\u0429\u0418\u0419"
+    );
+  });
+});
@@ -2276,6 +2276,101 @@ describe("assistant address follow-up carryover", () => {
    expect(normalizerService.normalize).not.toHaveBeenCalled();
  });

+  it("continues the original inventory query after replacement-damaged organization clarification", async () => {
+    const calls: Array<{ message: string; options?: any }> = [];
+    const firstMessage = "кайф - что там на складе по остаткам?";
+    const secondMessage = "АЛЬТЕРНАТ\uFFFD?ВА";
+    const repairedSecondMessage = "\u0410\u041b\u042c\u0422\u0415\u0420\u041d\u0410\u0422\u0418\u0412\u0410";
+    const addressQueryService = {
+      tryHandle: vi.fn(async (message: string, options?: any) => {
+        calls.push({ message, options });
+        if (message === firstMessage) {
+          return buildAddressLimitedLaneResult("missing_anchor", {
+            reply_text: [
+              "Нужно уточнить организацию, чтобы не смешивать компании в одном ответе.",
+              "Сейчас в доступном контуре вижу такие организации:",
+              "- ООО Альтернатива Плюс",
+              "- ООО Лайсвуд",
+              "- РАЙМ"
+            ].join("\n"),
+            debug: {
+              ...buildAddressLimitedLaneResult("missing_anchor").debug,
+              detected_intent: "inventory_on_hand_as_of_date",
+              extracted_filters: {
+                as_of_date: "2026-04-15"
+              },
+              selected_recipe: null,
+              organization_candidates: ["ООО Альтернатива Плюс", "ООО Лайсвуд", "РАЙМ"],
+              reasons: ["organization_clarification_required", "multiple_known_organizations_detected"]
+            }
+          });
+        }
+        if (
+          (message === secondMessage || message === repairedSecondMessage) &&
+          options?.followupContext &&
+          options?.activeOrganization === "ООО Альтернатива Плюс"
+        ) {
+          return buildAddressLaneResult({
+            reply_text: "На 15.04.2026 по ООО Альтернатива Плюс подтвержден складской остаток по всем складам.",
+            debug: {
+              ...buildAddressLaneResult().debug,
+              detected_intent: "inventory_on_hand_as_of_date",
+              extracted_filters: {
+                as_of_date: "2026-04-15",
+                organization: "ООО Альтернатива Плюс"
+              },
+              reasons: ["address_followup_context_applied", "organization_grounded_from_scope_candidates"]
+            }
+          });
+        }
+        return null;
+      })
+    } as any;
+
+    const normalizerService = {
+      normalize: vi.fn(async () => ({
+        assistant_reply: "normalizer_fallback_should_not_be_used",
+        reply_type: "partial_coverage",
+        debug: {}
+      }))
+    } as any;
+
+    const sessions = new AssistantSessionStore();
+    const service = new AssistantService(
+      normalizerService,
+      sessions as any,
+      {} as any,
+      { persistSession: vi.fn() } as any,
+      addressQueryService
+    );
+
+    const sessionId = `asst-address-org-clarification-damaged-${Date.now()}`;
+    const first = await service.handleMessage({
+      session_id: sessionId,
+      user_message: firstMessage,
+      useMock: true
+    } as any);
+    expect(first.ok).toBe(true);
+    expect(first.reply_type).toBe("partial_coverage");
+
+    const second = await service.handleMessage({
+      session_id: sessionId,
+      user_message: secondMessage,
+      useMock: true
+    } as any);
+
+    expect(second.ok).toBe(true);
+    expect(second.reply_type).toBe("factual");
+    expect(calls).toHaveLength(2);
+    expect(calls[1].message).toBe(repairedSecondMessage);
+    expect(calls[1].options?.activeOrganization).toBe("ООО Альтернатива Плюс");
+    expect(calls[1].options?.knownOrganizations).toEqual(["ООО Альтернатива Плюс", "ООО Лайсвуд", "РАЙМ"]);
+    expect(calls[1].options?.followupContext?.previous_intent).toBe("inventory_on_hand_as_of_date");
+    expect(calls[1].options?.followupContext?.previous_filters?.organization).toBe("ООО Альтернатива Плюс");
+    expect(calls[1].options?.followupContext?.root_filters?.organization).toBe("ООО Альтернатива Плюс");
+    expect(normalizerService.normalize).not.toHaveBeenCalled();
+  });
+
  it("keeps historical inventory date follow-up alive after company clarification and a capability answer", async () => {
    const calls: Array<{ message: string; options?: any }> = [];
    const firstMessage = "покажи остатки по складу";
@@ -42,6 +42,26 @@ describe("assistant organization matcher", () => {
    expect(score).toBeGreaterThanOrEqual(90);
  });

+  it("matches replacement-damaged organization clarification when the live candidate is unique", () => {
+    const resolved = resolveOrganizationSelectionFromMessage("АЛЬТЕРНАТ\uFFFD?ВА", [
+      "ООО Альтернатива Плюс",
+      "ООО Лайсвуд",
+      "РАЙМ"
+    ]);
+
+    expect(resolved).toBe("ООО Альтернатива Плюс");
+  });
+
+  it("keeps replacement-damaged organization clarification ambiguous when candidates share the token", () => {
+    const resolved = resolveOrganizationSelectionFromMessage("АЛЬТЕРНАТ\uFFFD?ВА", [
+      "ООО Альтернатива Плюс",
+      "ООО Альтернатива Минус",
+      "ООО Лайсвуд"
+    ]);
+
+    expect(resolved).toBeNull();
+  });
+
  it("treats minor live label corruption as the same organization entity", () => {
    expect(organizationsLikelySameEntity("Альтернатива Плюс", 'ООО "Альтернати"а Плюс"')).toBe(true);
  });
@@ -31,4 +31,26 @@ describe("autoruns question extraction", () => {
    expect(parsed[0]).toMatch(/поставщик/i);
    expect(parsed[0]).toMatch(/коротко/i);
  });
+
+  it("repairs old autorun C1-control mojibake before exposing cards and questions", () => {
+    const damagedTitle = String.fromCharCode(
+      0x420, 0x2018, 0x420, 0x45b, 0x420, 0x203a, 0x420, 0xac, 0x420, 0x401, 0x420, 0x45b, 0x420,
+      0x2122, 0x20, 0x420, 0x45b, 0x420, 0x2018, 0x420, 0xa9, 0x420, 0x98, 0x420, 0x2122, 0x20,
+      0x420, 0xa0, 0x421, 0x453, 0x421, 0x2021, 0x420, 0x405, 0x420, 0xb0, 0x421, 0x40f, 0x20,
+      0x421, 0x403, 0x420, 0xb5, 0x421, 0x403, 0x421, 0x403, 0x420, 0x451, 0x421, 0x40f, 0x20,
+      0x31, 0x36, 0x2e, 0x30, 0x34, 0x2e, 0x32, 0x30, 0x32, 0x36, 0x2c, 0x20, 0x32, 0x31,
+      0x3a, 0x32, 0x36, 0x3a, 0x30, 0x36
+    );
+    const damagedAlternative = String.fromCharCode(
+      0x420, 0x452, 0x420, 0x203a, 0x420, 0xac, 0x420, 0x45e, 0x420, 0x2022, 0x420, 0xa0, 0x420,
+      0x45c, 0x420, 0x452, 0x420, 0x45e, 0x420, 0x98, 0x420, 0x2019, 0x420, 0x452
+    );
+
+    expect(__autoRunsQuestionTestUtils.repairAutogenMojibake(damagedTitle)).toBe(
+      "\u0411\u041e\u041b\u042c\u0428\u041e\u0419 \u041e\u0411\u0429\u0418\u0419 \u0420\u0443\u0447\u043d\u0430\u044f \u0441\u0435\u0441\u0441\u0438\u044f 16.04.2026, 21:26:06"
+    );
+    expect(__autoRunsQuestionTestUtils.repairAutogenMojibake(damagedAlternative)).toBe(
+      "\u0410\u041b\u042c\u0422\u0415\u0420\u041d\u0410\u0422\u0418\u0412\u0410"
+    );
+  });
 });
@@ -33,4 +33,37 @@ describe("eval runtime question splitting", () => {
    expect(parsed[0]).toMatch(/поставщик/i);
    expect(parsed[0]).toMatch(/коротко/i);
  });
+
+  it("repairs mojibake questions before runtime job materialization", () => {
+    const parsed = __evalRouteTestUtils.normalizeRuntimeQuestions([
+      "РєР°Р№С„ - С‡С‚Рѕ С‚Р°Рј РЅР° СЃРєР»Р°РґРµ РїРѕ РѕСЃС‚Р°С‚РєР°Рј?"
+    ]);
+
+    expect(parsed).toEqual(["кайф - что там на складе по остаткам?"]);
+  });
+
+  it("repairs damaged clarification as one scenario turn when splitting is disabled", () => {
+    const parsed = __evalRouteTestUtils.normalizeRuntimeQuestions(
+      ["\u0410\u041b\u042c\u0422\u0415\u0420\u041d\u0410\u0422\uFFFD?\u0412\u0410"],
+      {
+        dedupe: false,
+        splitCandidates: false
+      }
+    );
+
+    expect(parsed).toEqual(["\u0410\u041b\u042c\u0422\u0415\u0420\u041d\u0410\u0422\u0418\u0412\u0410"]);
+  });
+
+  it("repairs C1-control autorun clarification before runtime job materialization", () => {
+    const damagedAlternative = String.fromCharCode(
+      0x420, 0x452, 0x420, 0x203a, 0x420, 0xac, 0x420, 0x45e, 0x420, 0x2022, 0x420, 0xa0, 0x420,
+      0x45c, 0x420, 0x452, 0x420, 0x45e, 0x420, 0x98, 0x420, 0x2019, 0x420, 0x452
+    );
+    const parsed = __evalRouteTestUtils.normalizeRuntimeQuestions([damagedAlternative], {
+      dedupe: false,
+      splitCandidates: false
+    });
+
+    expect(parsed).toEqual(["\u0410\u041b\u042c\u0422\u0415\u0420\u041d\u0410\u0422\u0418\u0412\u0410"]);
+  });
 });