SILVER/NODEDC_1C
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

АДРЕСНЫЙ РЕЖИМ - ллм декомпоз

Browse Source
This commit is contained in:
dctouch
2026-04-01 22:11:40 +03:00
parent 18e0f1364d
commit b2d32f869c
1540 changed files with 10464 additions and 1451 deletions
Download Patch File Download Diff File Expand all files Collapse all files
llm_normalizer/backend/tests/addressQueryRuntimeM23.test.ts
+91
View File
@@ -5,6 +5,7 @@ import { classifyAddressQueryShape } from "../src/services/addressQueryShapeClas
import { extractAddressFilters } from "../src/services/addressFilterExtractor";
import { AddressQueryService } from "../src/services/addressQueryService";
import { buildAddressRecipePlan, selectAddressRecipe } from "../src/services/addressRecipeCatalog";
import { runAddressDecomposeStage } from "../src/services/address_runtime/decomposeStage";
describe("address query shape classifier", () => {
it("classifies explain question as deep-shape", () => {
@@ -225,6 +226,56 @@ describe("address filter extraction for balance drilldown", () => {
expect(result.extracted_filters.counterparty).not.toBe("плс");
});
it("extracts short ordinal year period from noisy docs phrase", () => {
const result = extractAddressFilters(
"бля епт покажи доки по свк за 20-й",
"list_documents_by_counterparty"
);
expect(result.extracted_filters.counterparty).toBe("свк");
expect(result.extracted_filters.period_from).toBe("2020-01-01");
expect(result.extracted_filters.period_to).toBe("2020-12-31");
expect(result.warnings).toContain("period_derived_from_year_phrase");
});
it("does not use action verb as counterparty when phrase is 'Показать документы <counterparty>'", () => {
const result = extractAddressFilters(
"Показать документы СВК за 2020 год.",
"list_documents_by_counterparty"
);
expect(result.extracted_filters.counterparty).toBe("СВК");
expect(result.extracted_filters.counterparty).not.toBe("Показать");
expect(result.extracted_filters.period_from).toBe("2020-01-01");
expect(result.extracted_filters.period_to).toBe("2020-12-31");
});
it("extracts counterparty and short year from transliterated noisy phrase", () => {
const result = extractAddressFilters(
"svk doki za 20 god pokezh",
"list_documents_by_counterparty"
);
expect(result.extracted_filters.counterparty).toBe("svk");
expect(result.extracted_filters.period_from).toBe("2020-01-01");
expect(result.extracted_filters.period_to).toBe("2020-12-31");
expect(
result.warnings.some(
(warning) =>
warning === "counterparty_anchor_derived_from_free_text_heuristic" ||
warning === "counterparty_anchor_derived_from_implicit_phrase"
)
).toBe(true);
expect(result.warnings).toContain("period_derived_from_year_phrase");
});
it("repairs mojibake phrase before extracting counterparty filters", () => {
const result = extractAddressFilters(
"Показать документы СВК за 2020 год.",
"list_documents_by_counterparty"
);
expect(result.extracted_filters.counterparty).toBe("СВК");
expect(result.extracted_filters.period_from).toBe("2020-01-01");
expect(result.extracted_filters.period_to).toBe("2020-12-31");
});
it("extracts explicit year range period from phrase", () => {
const result = extractAddressFilters(
"Какие документы по СВК за 2000 - 2025 год?",
@@ -372,6 +423,46 @@ describe("address query limited taxonomy and stage diagnostics", () => {
});
});
describe("address decompose stage follow-up carryover", () => {
it("keeps short period follow-up in address lane and preserves previous counterparty anchor", () => {
const result = runAddressDecomposeStage("а теперь только за май 2020", {
previous_intent: "list_documents_by_counterparty",
previous_filters: {
counterparty: "свк",
period_from: "2020-01-01",
period_to: "2020-12-31"
},
previous_anchor_type: "counterparty",
previous_anchor_value: "Группа СВК"
});
expect(result).not.toBeNull();
expect(result?.intent.intent).toBe("list_documents_by_counterparty");
expect(result?.filters.extracted_filters.counterparty).toBe("свк");
expect(result?.filters.extracted_filters.period_from).toBe("2020-05-01");
expect(result?.filters.extracted_filters.period_to).toBe("2020-05-31");
expect(result?.baseReasons).toContain("address_followup_context_applied");
});
it("inherits as_of_date from previous period for same-date balance follow-up", () => {
const result = runAddressDecomposeStage("а по счету 60.01 на ту же дату", {
previous_intent: "list_documents_by_counterparty",
previous_filters: {
counterparty: "свк",
period_from: "2020-05-01",
period_to: "2020-05-31"
},
previous_anchor_type: "counterparty",
previous_anchor_value: "Группа СВК"
});
expect(result).not.toBeNull();
expect(result?.intent.intent).toBe("account_balance_snapshot");
expect(result?.filters.extracted_filters.account).toBe("60.01");
expect(result?.filters.extracted_filters.as_of_date).toBe("2020-05-31");
expect(result?.baseReasons).toContain("as_of_date_from_followup_context");
expect(result?.baseReasons).toContain("address_followup_context_applied");
});
});
describe("address recipe catalog counterparty filtering", () => {
it("boosts limit for all-time counterparty queries", () => {
const filters = extractAddressFilters(
llm_normalizer/backend/tests/assistantAddressFollowupContext.test.ts
+5 -7
View File
@@ -120,15 +120,13 @@ describe("assistant address follow-up carryover", () => {
expect(second.debug?.extracted_filters?.counterparty).toBe("свк");
expect(second.debug?.answer_grounding_check?.reasons).toContain("address_followup_context_applied");
expect(calls).toHaveLength(3);
expect(calls).toHaveLength(2);
expect(calls[0].message).toBe("какие есть доки по свк с 2020 по 2025 год");
expect(calls[1].message).toBe("а за все время?");
expect(calls[1].options?.followupContext).toBeUndefined();
expect(calls[2].message).toBe("а за все время?");
expect(calls[2].options?.followupContext?.previous_intent).toBe("list_documents_by_counterparty");
expect(calls[2].options?.followupContext?.previous_anchor_type).toBe("counterparty");
expect(calls[2].options?.followupContext?.previous_anchor_value).toBe("Группа СВК");
expect(calls[2].options?.followupContext?.previous_filters?.counterparty).toBe("свк");
expect(calls[1].options?.followupContext?.previous_intent).toBe("list_documents_by_counterparty");
expect(calls[1].options?.followupContext?.previous_anchor_type).toBe("counterparty");
expect(calls[1].options?.followupContext?.previous_anchor_value).toBe("Группа СВК");
expect(calls[1].options?.followupContext?.previous_filters?.counterparty).toBe("свк");
expect(normalizerService.normalize).not.toHaveBeenCalled();
});
});
llm_normalizer/backend/tests/assistantAddressLlmPredecompose.test.ts
+156
View File
@@ -0,0 +1,156 @@
import { describe, expect, it, vi } from "vitest";
import { AssistantService } from "../src/services/assistantService";
import { AssistantSessionStore } from "../src/services/assistantSessionStore";
function buildAddressLaneResult(message: string): any {
return {
handled: true,
reply_text: `handled: ${message}`,
reply_type: "factual",
response_type: "FACTUAL_LIST",
debug: {
detected_mode: "address_query",
detected_mode_confidence: "high",
query_shape: "DOCUMENT_LIST",
query_shape_confidence: "medium",
detected_intent: "list_documents_by_counterparty",
detected_intent_confidence: "medium",
extracted_filters: {
sort: "period_desc",
limit: 20,
counterparty: "svk",
period_from: "2020-01-01",
period_to: "2020-12-31"
},
missing_required_filters: [],
selected_recipe: "address_documents_by_counterparty_v1",
mcp_call_status_legacy: "matched_non_empty",
account_scope_mode: "preferred",
account_scope_fallback_applied: false,
anchor_type: "counterparty",
anchor_value_raw: "svk",
anchor_value_resolved: "Группа СВК",
resolver_confidence: "medium",
ambiguity_count: 0,
match_failure_stage: "none",
match_failure_reason: null,
mcp_call_status: "matched_non_empty",
rows_fetched: 20,
raw_rows_received: 20,
rows_after_account_scope: 5,
rows_after_recipe_filter: 3,
rows_materialized: 5,
rows_matched: 3,
raw_row_keys_sample: [],
materialization_drop_reason: "none",
account_token_raw: null,
account_token_normalized: null,
account_scope_fields_checked: ["account_dt", "account_kt", "registrator", "analytics"],
account_scope_match_strategy: "account_code_regex_plus_alias_map_v1",
account_scope_drop_reason: "not_applicable",
runtime_readiness: "LIVE_QUERYABLE_WITH_LIMITS",
limited_reason_category: null,
response_type: "FACTUAL_LIST",
limitations: [],
reasons: ["address_action_detected", "address_entity_detected", "document_list_signal_detected"]
}
};
}
describe("assistant address llm pre-decompose candidate preference", () => {
it("prefers raw fragment when normalized fragment loses counterparty 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-1",
ok: true,
normalized: {
schema_version: "normalized_query_v2_0_2",
user_message_raw: "бля svk doki za 20 god pokezh pls",
message_in_scope: true,
scope_confidence: "medium",
contains_multiple_tasks: false,
fragments: [
{
fragment_id: "F1",
raw_fragment_text: "svk doki za 20 god pokezh",
normalized_fragment_text: "Показать 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-${Date.now()}`,
user_message: "бля svk doki za 20 god pokezh pls",
llmProvider: "local",
useMock: false
} as any);
expect(response.ok).toBe(true);
expect(response.reply_type).toBe("factual");
expect(calls).toHaveLength(1);
expect(calls[0].message).toBe("svk doki za 20 god pokezh");
expect(response.debug?.llm_decomposition_attempted).toBe(true);
expect(response.debug?.llm_decomposition_applied).toBe(true);
expect(response.debug?.llm_decomposition_effective_message).toBe("svk doki za 20 god pokezh");
});
});
llm_normalizer/backend/tests/normalizerServiceCoercion.test.ts
+140
View File
@@ -0,0 +1,140 @@
import { describe, expect, it, vi } from "vitest";
import { NormalizerService } from "../src/services/normalizerService";
import type { OpenAIResponsesClient } from "../src/services/openaiResponsesClient";
import type { NormalizedQueryV2_0_2 } from "../src/types/normalizer";
function makeService(outputText: string) {
const normalizeMock = vi.fn().mockResolvedValue({
raw: { mode: "stubbed" },
outputText,
usage: {
input_tokens: 11,
output_tokens: 22,
total_tokens: 33
}
});
const openaiClient = {
normalize: normalizeMock
} as unknown as OpenAIResponsesClient;
return {
service: new NormalizerService(openaiClient),
normalizeMock
};
}
describe("NormalizerService coercion layer", () => {
it("coerces malformed v2.0.2 local payload into a usable routed fragment", async () => {
const malformed = {
schema_version: "normalized_query_v2_0_2",
global_notes: {
needs_clarification: true,
clarification_reason: "insufficient_specificity"
},
fragments: [
{
fragment_id: 1,
raw_fragment_text: "свк доки за 20год покеж",
normalized_fragment_text: "показать документы за 20 год",
domain_relevance: true,
business_scope: "document_review",
entity_hints: [],
account_hints: [],
document_hints: ["документы"],
register_hints: [],
time_scope: {
period_type: "year",
year: 2020,
month: null
},
flags: {},
candidate_labels: ["show_documents"],
confidence: 0.85,
execution_readiness: "needs_clarification",
clarification_reason: "insufficient_specificity",
soft_assumption_used: [],
route_status: "no_route",
no_route_reason: "insufficient_specificity"
}
]
};
const { service, normalizeMock } = makeService(JSON.stringify(malformed));
const response = await service.normalize({
llmProvider: "local",
model: "qwen2.5-14b-instruct-1m",
baseUrl: "http://127.0.0.1:1234/v1",
promptVersion: "normalizer_v2_0_2",
userQuestion: "свк доки за 20год покеж",
retryPolicy: "single-pass-strict"
});
expect(normalizeMock).toHaveBeenCalledTimes(1);
expect(response.ok).toBe(true);
expect(response.normalized?.schema_version).toBe("normalized_query_v2_0_2");
const normalized = response.normalized as NormalizedQueryV2_0_2;
expect(normalized.fragments).toHaveLength(1);
expect(normalized.fragments[0]).toMatchObject({
fragment_id: "F1",
domain_relevance: "in_scope",
business_scope: "company_specific_accounting",
confidence: "high",
route_status: "routed",
no_route_reason: null
});
expect(normalized.fragments[0].candidate_labels).toContain("simple_factual");
expect(normalized.fragments[0].time_scope).toMatchObject({
type: "explicit",
value: "2020"
});
});
it("coerces period_type month payload into explicit YYYY-MM time scope", async () => {
const malformed = {
schema_version: "normalized_query_v2_0_2",
fragments: [
{
fragment_id: "frag_main",
raw_fragment_text: "Какой остаток по счету 60 на 2020 май",
normalized_fragment_text: "остаток по счету 60 на май 2020",
domain_relevance: "in_scope",
business_scope: "company_specific_accounting",
entity_hints: ["счет"],
account_hints: ["60"],
document_hints: [],
register_hints: ["остаток"],
time_scope: {
period_type: "month",
year: 2020,
month: 5
},
flags: {
asks_for_period_summary: false
},
candidate_labels: ["simple_factual"],
confidence: 0.67
}
]
};
const { service } = makeService(JSON.stringify(malformed));
const response = await service.normalize({
llmProvider: "local",
model: "qwen2.5-14b-instruct-1m",
baseUrl: "http://127.0.0.1:1234/v1",
promptVersion: "normalizer_v2_0_2",
userQuestion: "Какой остаток по счету 60 на 2020 май",
retryPolicy: "single-pass-strict"
});
expect(response.ok).toBe(true);
const normalized = response.normalized as NormalizedQueryV2_0_2;
expect(normalized.fragments).toHaveLength(1);
expect(normalized.fragments[0].time_scope).toMatchObject({
type: "explicit",
value: "2020-05"
});
expect(normalized.fragments[0].route_status).toBe("routed");
});
});