АДРЕСНЫЙ РЕЖИМ - fix(address): закрыт S019 predecompose-anchor drift; nightly 102/102+25/25 PASS; docs synced

2026-04-08 11:15:33 +03:00
parent a5dee965c2
commit eec5ef4fc0
210 changed files with 325486 additions and 100 deletions
@@ -85,6 +85,16 @@ describe("address query shape classifier", () => {
    expect(result.mode).toBe("address_query");
  });

+  it("keeps noisy management phrase about years alive in address lane", () => {
+    const result = detectAddressQuestionMode("за какие года база ваще живая?");
+    expect(result.mode).toBe("address_query");
+  });
+
+  it("keeps noisy month-peak phrase in address lane", () => {
+    const result = detectAddressQuestionMode("а теперь месяц-пик по операциям");
+    expect(result.mode).toBe("address_query");
+  });
+
  it("keeps management contract usage overview question in address lane", () => {
    const result = detectAddressQuestionMode("Сколько всего договоров заведено и сколько из них реально использовались?");
    expect(result.mode).toBe("address_query");
@@ -1151,6 +1161,11 @@ describe("address intent resolver expansion (M2.3a)", () => {
    expect(result.intent).toBe("list_documents_by_counterparty");
  });

+  it("resolves transliterated docy slang as documents by counterparty intent", () => {
+    const result = resolveAddressIntent("svk poka docy za 2020");
+    expect(result.intent).toBe("list_documents_by_counterparty");
+  });
+
  it("resolves bank operations by supplier phrase", () => {
    const result = resolveAddressIntent("Покажи платежи по поставщику Альфа за июль 2020");
    expect(result.intent).toBe("bank_operations_by_counterparty");
@@ -1273,6 +1288,16 @@ describe("address intent resolver expansion (M2.3a)", () => {
    expect(result.intent).toBe("period_coverage_profile");
  });

+  it("resolves period coverage profile for noisy active-year-by-docs phrase", () => {
+    const result = resolveAddressIntent("какой год тут самый движовый по докам");
+    expect(result.intent).toBe("period_coverage_profile");
+  });
+
+  it("resolves period coverage profile for month-peak follow-up phrase", () => {
+    const result = resolveAddressIntent("а теперь месяц-пик по операциям");
+    expect(result.intent).toBe("period_coverage_profile");
+  });
+
  it("resolves document+section profile for document type usage question", () => {
    const result = resolveAddressIntent("Какие типы документов используются чаще всего в базе?");
    expect(result.intent).toBe("document_type_and_account_section_profile");
@@ -1293,6 +1318,16 @@ describe("address intent resolver expansion (M2.3a)", () => {
    expect(result.intent).toBe("document_type_and_account_section_profile");
  });

+  it("resolves document+section profile for noisy docs usage phrase", () => {
+    const result = resolveAddressIntent("каких доков у нас больше всего крутится?");
+    expect(result.intent).toBe("document_type_and_account_section_profile");
+  });
+
+  it("resolves document+section profile for summary by doc types and share phrase", () => {
+    const result = resolveAddressIntent("Сформируй сводку по типам документов и их доле в общем объеме.");
+    expect(result.intent).toBe("document_type_and_account_section_profile");
+  });
+
  it("resolves counterparty population intent for total unique counterparties question", () => {
    const result = resolveAddressIntent("Сколько всего уникальных контрагентов в базе?");
    expect(result.intent).toBe("counterparty_population_and_roles");
@@ -1363,6 +1398,11 @@ describe("address intent resolver expansion (M2.3a)", () => {
    expect(result.intent).toBe("customer_revenue_and_payments");
  });

+  it("resolves customer revenue intent from small deals by budget slang wording", () => {
+    const result = resolveAddressIntent("покажи топ-20 самых маленьких сделок по бюджету");
+    expect(result.intent).toBe("customer_revenue_and_payments");
+  });
+
  it("resolves customer revenue intent from typo highest-check wording", () => {
    const result = resolveAddressIntent("с каких кликентов самый высокий чек");
    expect(result.intent).toBe("customer_revenue_and_payments");
@@ -1382,6 +1422,11 @@ describe("address intent resolver expansion (M2.3a)", () => {
    const result = resolveAddressIntent("покажи договора все по жуковке 51");
    expect(result.intent).toBe("list_contracts_by_counterparty");
  });
+
+  it("prefers documents-by-contract intent for explicit document follow-up wording", () => {
+    const result = resolveAddressIntent("покажи документы по этому же договору");
+    expect(result.intent).toBe("list_documents_by_contract");
+  });
 });

 describe("address filter extraction for balance drilldown", () => {
@@ -1630,6 +1675,14 @@ describe("address filter extraction for balance drilldown", () => {
    expect(result.warnings).toContain("period_derived_from_year_phrase");
  });

+  it("does not use filler token 'есть' as counterparty when explicit shorthand anchor exists", () => {
+    const result = extractAddressFilters("какие у свк есть доки за 2020?", "list_documents_by_counterparty");
+    expect(result.extracted_filters.counterparty).toBe("свк");
+    expect(result.extracted_filters.counterparty).not.toBe("есть");
+    expect(result.extracted_filters.period_from).toBe("2020-01-01");
+    expect(result.extracted_filters.period_to).toBe("2020-12-31");
+  });
+
  it("extracts free-text counterparty and relaxed short-year period from noisy phrase", () => {
    const result = extractAddressFilters(
      "свк 20 год - покажи доки плс",
@@ -1690,6 +1743,17 @@ describe("address filter extraction for balance drilldown", () => {
    expect(result.warnings).toContain("period_derived_from_year_phrase");
  });

+  it("does not treat transliterated filler verb as counterparty in docy phrase", () => {
+    const result = extractAddressFilters(
+      "svk poka docy za 2020",
+      "list_documents_by_counterparty"
+    );
+    expect(result.extracted_filters.counterparty).toBe("svk");
+    expect(result.extracted_filters.counterparty).not.toBe("poka");
+    expect(result.extracted_filters.period_from).toBe("2020-01-01");
+    expect(result.extracted_filters.period_to).toBe("2020-12-31");
+  });
+
  it("repairs mojibake phrase before extracting counterparty filters", () => {
    const result = extractAddressFilters(
      "РџРѕРєР°Р·Р°С‚СЊ РґРѕРєСѓРјРµРЅС‚С‹ РЎР’Рљ Р·Р° 2020 РіРѕРґ.",
@@ -1785,6 +1849,14 @@ describe("address filter extraction for balance drilldown", () => {
    expect(result.extracted_filters.period_to).toBe("2020-12-31");
  });

+  it("trims explanatory tail after contract token", () => {
+    const result = extractAddressFilters(
+      "документы по договору 19/15 выведите связанные документы",
+      "list_documents_by_contract"
+    );
+    expect(result.extracted_filters.contract).toBe("19/15");
+  });
+
  it("extracts multiline year range period from phrase", () => {
    const result = extractAddressFilters(
      "Какие документы по СВК за 2000 - 2025\n год?",
@@ -2158,6 +2230,46 @@ describe("address decompose stage follow-up carryover", () => {
    expect(result?.baseReasons).toContain("intent_adjusted_to_contract_followup_context");
  });

+  it("replaces noisy follow-up contract anchor with previous contract from context", () => {
+    const result = runAddressDecomposeStage("а документы по этому же договору за тот же период", {
+      previous_intent: "bank_operations_by_contract",
+      previous_filters: {
+        contract: "19/15",
+        period_from: "2020-01-01",
+        period_to: "2020-12-31"
+      },
+      previous_anchor_type: "contract",
+      previous_anchor_value: "19/15"
+    });
+    expect(result).not.toBeNull();
+    expect(result?.intent.intent).toBe("list_documents_by_contract");
+    expect(result?.filters.extracted_filters.contract).toBe("19/15");
+    expect(
+      result?.baseReasons?.includes("contract_replaced_from_followup_context") ||
+        result?.baseReasons?.includes("contract_from_followup_context")
+    ).toBe(true);
+  });
+
+  it("replaces noisy referential counterparty anchor with previous counterparty from context", () => {
+    const result = runAddressDecomposeStage("а теперь документы по нему", {
+      previous_intent: "bank_operations_by_counterparty",
+      previous_filters: {
+        counterparty: "свк",
+        period_from: "2020-11-01",
+        period_to: "2020-11-30"
+      },
+      previous_anchor_type: "counterparty",
+      previous_anchor_value: "свк"
+    });
+    expect(result).not.toBeNull();
+    expect(result?.intent.intent).toBe("list_documents_by_counterparty");
+    expect(result?.filters.extracted_filters.counterparty).toBe("свк");
+    expect(
+      result?.baseReasons?.includes("counterparty_replaced_from_followup_context") ||
+        result?.baseReasons?.includes("counterparty_from_followup_context")
+    ).toBe(true);
+  });
+
  it("promotes open-items intent from follow-up wording with inherited contract anchor", () => {
    const result = runAddressDecomposeStage("а теперь открытые позиции по нему", {
      previous_intent: "bank_operations_by_contract",
@@ -159,6 +159,291 @@ describe("assistant address llm pre-decompose candidate preference", () => {
    expect(["unknown", "list_documents_by_counterparty"]).toContain(response.debug?.llm_predecompose_contract?.intent);
  });

+  it("keeps counterparty anchor for docy typo when llm fragment drops anchor", async () => {
+    const calls: Array<{ message: string }> = [];
+    const addressQueryService = {
+      tryHandle: vi.fn(async (message: string) => {
+        calls.push({ message });
+        return buildAddressLaneResult(message);
+      })
+    } as any;
+
+    const normalizerService = {
+      normalize: vi.fn(async () => ({
+        trace_id: "norm-predecompose-docy",
+        ok: true,
+        normalized: {
+          schema_version: "normalized_query_v2_0_2",
+          user_message_raw: "svk poka docy za 2020",
+          message_in_scope: true,
+          scope_confidence: "medium",
+          contains_multiple_tasks: false,
+          fragments: [
+            {
+              fragment_id: "F1",
+              raw_fragment_text: "svk poka docy za 2020",
+              normalized_fragment_text: "Покажи документы за 2020 год",
+              domain_relevance: "in_scope",
+              business_scope: "company_specific_accounting",
+              entity_hints: [],
+              account_hints: [],
+              document_hints: ["документы"],
+              register_hints: [],
+              time_scope: {
+                type: "explicit",
+                value: "2020",
+                confidence: "high"
+              },
+              flags: {
+                has_multi_entity_scope: false,
+                asks_for_chain_explanation: false,
+                asks_for_ranking_or_top: false,
+                asks_for_period_summary: false,
+                asks_for_rule_check: false,
+                asks_for_anomaly_scan: false,
+                asks_for_exact_object_trace: false,
+                asks_for_evidence: false,
+                mentions_period_close_context: false
+              },
+              candidate_labels: ["simple_factual"],
+              confidence: "medium",
+              execution_readiness: "executable",
+              clarification_reason: null,
+              soft_assumption_used: [],
+              route_status: "routed",
+              no_route_reason: null
+            }
+          ],
+          discarded_fragments: [],
+          global_notes: {
+            needs_clarification: false,
+            clarification_reason: null
+          }
+        },
+        raw_model_output: null,
+        validation: { passed: true, errors: [] },
+        usage: { input_tokens: 1, output_tokens: 1, total_tokens: 2 },
+        latency_ms: 10,
+        prompt_version: "normalizer_v2_0_2",
+        schema_version: "v2_0_2",
+        request_count_for_case: 1
+      }))
+    } as any;
+
+    const sessions = new AssistantSessionStore();
+    const service = new AssistantService(
+      normalizerService,
+      sessions as any,
+      {} as any,
+      { persistSession: vi.fn() } as any,
+      addressQueryService
+    );
+
+    const response = await service.handleMessage({
+      session_id: `asst-predecompose-docy-${Date.now()}`,
+      user_message: "svk poka docy za 2020",
+      llmProvider: "local",
+      useMock: false
+    } as any);
+
+    expect(response.ok).toBe(true);
+    expect(calls).toHaveLength(1);
+    expect(calls[0].message.toLowerCase()).toContain("svk");
+    expect(calls[0].message).not.toBe("Покажи документы за 2020 год");
+    expect(String(response.debug?.llm_decomposition_effective_message ?? "").toLowerCase()).toContain("svk");
+    expect(response.debug?.llm_decomposition_reason).not.toBe("normalized_fragment_applied");
+  });
+
+  it("does not treat service verb as counterparty anchor when llm rewrites noisy bank phrase", async () => {
+    const calls: Array<{ message: string }> = [];
+    const addressQueryService = {
+      tryHandle: vi.fn(async (message: string) => {
+        calls.push({ message });
+        return buildAddressLaneResult(message);
+      })
+    } as any;
+
+    const normalizerService = {
+      normalize: vi.fn(async () => ({
+        trace_id: "norm-predecompose-skazhi-anchor",
+        ok: true,
+        normalized: {
+          schema_version: "normalized_query_v2_0_2",
+          user_message_raw: "свк списания/поступления за 2020",
+          message_in_scope: true,
+          scope_confidence: "medium",
+          contains_multiple_tasks: false,
+          fragments: [
+            {
+              fragment_id: "F1",
+              raw_fragment_text: "свк списания/поступления за 2020",
+              normalized_fragment_text: "скажи списания/поступления за 2020",
+              domain_relevance: "in_scope",
+              business_scope: "company_specific_accounting",
+              entity_hints: [],
+              account_hints: [],
+              document_hints: ["списания", "поступления"],
+              register_hints: [],
+              time_scope: {
+                type: "explicit",
+                value: "2020",
+                confidence: "high"
+              },
+              flags: {
+                has_multi_entity_scope: false,
+                asks_for_chain_explanation: false,
+                asks_for_ranking_or_top: false,
+                asks_for_period_summary: false,
+                asks_for_rule_check: false,
+                asks_for_anomaly_scan: false,
+                asks_for_exact_object_trace: false,
+                asks_for_evidence: false,
+                mentions_period_close_context: false
+              },
+              candidate_labels: ["simple_factual"],
+              confidence: "medium",
+              execution_readiness: "executable",
+              clarification_reason: null,
+              soft_assumption_used: [],
+              route_status: "routed",
+              no_route_reason: null
+            }
+          ],
+          discarded_fragments: [],
+          global_notes: {
+            needs_clarification: false,
+            clarification_reason: null
+          }
+        },
+        raw_model_output: null,
+        validation: { passed: true, errors: [] },
+        usage: { input_tokens: 1, output_tokens: 1, total_tokens: 2 },
+        latency_ms: 10,
+        prompt_version: "normalizer_v2_0_2",
+        schema_version: "v2_0_2",
+        request_count_for_case: 1
+      }))
+    } as any;
+
+    const sessions = new AssistantSessionStore();
+    const service = new AssistantService(
+      normalizerService,
+      sessions as any,
+      {} as any,
+      { persistSession: vi.fn() } as any,
+      addressQueryService
+    );
+
+    const response = await service.handleMessage({
+      session_id: `asst-predecompose-skazhi-anchor-${Date.now()}`,
+      user_message: "свк списания/поступления за 2020",
+      llmProvider: "local",
+      useMock: false
+    } as any);
+
+    expect(response.ok).toBe(true);
+    expect(calls).toHaveLength(1);
+    expect(calls[0].message.toLowerCase()).toContain("свк");
+    expect(calls[0].message.toLowerCase()).not.toContain("скажи");
+    expect(response.debug?.llm_decomposition_reason).not.toBe("normalized_fragment_applied");
+  });
+
+  it("rejects follow-up intent injection when llm adds documents to same-date account prompt", async () => {
+    const calls: Array<{ message: string }> = [];
+    const addressQueryService = {
+      tryHandle: vi.fn(async (message: string) => {
+        calls.push({ message });
+        return buildAddressLaneResult(message);
+      })
+    } as any;
+
+    const normalizerService = {
+      normalize: vi.fn(async () => ({
+        trace_id: "norm-predecompose-followup-injection",
+        ok: true,
+        normalized: {
+          schema_version: "normalized_query_v2_0_2",
+          user_message_raw: "а на ту же дату по 62",
+          message_in_scope: true,
+          scope_confidence: "medium",
+          contains_multiple_tasks: false,
+          fragments: [
+            {
+              fragment_id: "F1",
+              raw_fragment_text: "а на ту же дату по 62",
+              normalized_fragment_text: "документы или проводки по счету 62 на ту же дату",
+              domain_relevance: "in_scope",
+              business_scope: "company_specific_accounting",
+              entity_hints: [],
+              account_hints: ["62"],
+              document_hints: ["документы"],
+              register_hints: [],
+              time_scope: {
+                type: "explicit",
+                value: "same-date",
+                confidence: "medium"
+              },
+              flags: {
+                has_multi_entity_scope: false,
+                asks_for_chain_explanation: false,
+                asks_for_ranking_or_top: false,
+                asks_for_period_summary: false,
+                asks_for_rule_check: false,
+                asks_for_anomaly_scan: false,
+                asks_for_exact_object_trace: false,
+                asks_for_evidence: false,
+                mentions_period_close_context: false
+              },
+              candidate_labels: ["simple_factual"],
+              confidence: "medium",
+              execution_readiness: "executable",
+              clarification_reason: null,
+              soft_assumption_used: [],
+              route_status: "routed",
+              no_route_reason: null
+            }
+          ],
+          discarded_fragments: [],
+          global_notes: {
+            needs_clarification: false,
+            clarification_reason: null
+          }
+        },
+        raw_model_output: null,
+        validation: { passed: true, errors: [] },
+        usage: { input_tokens: 1, output_tokens: 1, total_tokens: 2 },
+        latency_ms: 10,
+        prompt_version: "normalizer_v2_0_2",
+        schema_version: "v2_0_2",
+        request_count_for_case: 1
+      }))
+    } as any;
+
+    const sessions = new AssistantSessionStore();
+    const service = new AssistantService(
+      normalizerService,
+      sessions as any,
+      {} as any,
+      { persistSession: vi.fn() } as any,
+      addressQueryService
+    );
+
+    const sourceMessage = "а на ту же дату по 62";
+    const response = await service.handleMessage({
+      session_id: `asst-predecompose-followup-injection-${Date.now()}`,
+      user_message: sourceMessage,
+      llmProvider: "local",
+      useMock: false
+    } as any);
+
+    expect(response.ok).toBe(true);
+    expect(calls).toHaveLength(1);
+    expect(calls[0].message).toBe(sourceMessage);
+    expect(response.debug?.llm_decomposition_applied).toBe(false);
+    expect(response.debug?.llm_decomposition_reason).toBe("normalized_fragment_rejected_followup_intent_injection");
+    expect(response.debug?.llm_decomposition_effective_message).toBe(sourceMessage);
+  });
+
  it("applies deterministic fallback rule when llm fragment is unusable", async () => {
    const calls: Array<{ message: string }> = [];
    const addressQueryService = {