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

llm_normalizer/backend/src/services/address_runtime/decomposeStage.ts

@@ -62,7 +62,7 @@ function hasOpenItemsHint(text: string): boolean {
 }
 
 function hasDocumentSignal(text: string): boolean {
-  return /(?:док(?:и|умент|ументы|ументов|ументами)|docs?|documents?)/iu.test(String(text ?? ""));
+  return /(?:док(?:и|умент|ументы|ументов|ументами)|docs?|documents?|doki|docy|doci)/iu.test(String(text ?? ""));
 }
 
 function hasBankSignal(text: string): boolean {
@@ -134,6 +134,122 @@ function hasExplicitLooseByAnchorToken(text: string): boolean {
   return !pronounTokens.has(token) && !genericTokens.has(token);
 }
 
+const FOLLOWUP_LOW_QUALITY_COUNTERPARTY_TOKENS = new Set([
+  "есть",
+  "же",
+  "что",
+  "все",
+  "всё",
+  "год",
+  "года",
+  "году",
+  "контрагентам",
+  "предоставьте",
+  "получить",
+  "сверка",
+  "теперь",
+  "сейчас",
+  "этому",
+  "этомуже",
+  "тому",
+  "томуже",
+  "нему",
+  "ней",
+  "ним",
+  "неуказанному",
+  "неуказанный",
+  "неуказанная",
+  "неуказанное",
+  "указанному",
+  "указанный",
+  "указанная",
+  "указанное",
+  "объект",
+  "объекту",
+  "период",
+  "периоду",
+  "указанныйпериод",
+  "сводные",
+  "сводный",
+  "сводная",
+  "сводную",
+  "сводном",
+  "сводного",
+  "сводному"
+]);
+
+const FOLLOWUP_LOW_QUALITY_CONTRACT_TOKENS = new Set([
+  "за",
+  "же",
+  "это",
+  "указанный",
+  "указанному",
+  "период",
+  "периоду",
+  "тот",
+  "тотже",
+  "этот",
+  "этому",
+  "этомуже",
+  "договор",
+  "договору",
+  "номер"
+]);
+
+function normalizeAnchorToken(value: string): string {
+  return String(value ?? "").trim().toLowerCase().replace(/ё/g, "е");
+}
+
+function hasContractLikeToken(value: string): boolean {
+  return /\b[a-zа-я0-9]{1,20}[\/_-][a-zа-я0-9]{1,20}(?:[\/_-][a-zа-я0-9]{1,20})?\b/iu.test(String(value ?? ""));
+}
+
+function isLowQualityCounterpartyAnchor(value: string | null): boolean {
+  const normalized = normalizeAnchorToken(value ?? "");
+  if (!normalized) {
+    return true;
+  }
+  const tokens = normalized
+    .split(/[^a-zа-я0-9]+/iu)
+    .map((token) => token.trim())
+    .filter(Boolean);
+  if (tokens.length === 0) {
+    return true;
+  }
+  const meaningful = tokens.filter((token) => {
+    if (token.length < 2) {
+      return false;
+    }
+    if (/^(?:19|20)\d{2}$/.test(token)) {
+      return false;
+    }
+    return !FOLLOWUP_LOW_QUALITY_COUNTERPARTY_TOKENS.has(token);
+  });
+  return meaningful.length === 0;
+}
+
+function isLowQualityContractAnchor(value: string | null): boolean {
+  const normalized = normalizeAnchorToken(value ?? "");
+  if (!normalized) {
+    return true;
+  }
+  if (hasContractLikeToken(normalized)) {
+    return false;
+  }
+  if (!/\d/.test(normalized)) {
+    return true;
+  }
+  const tokens = normalized
+    .split(/[^a-zа-я0-9]+/iu)
+    .map((token) => token.trim())
+    .filter(Boolean);
+  if (tokens.length === 0) {
+    return true;
+  }
+  const meaningful = tokens.filter((token) => !FOLLOWUP_LOW_QUALITY_CONTRACT_TOKENS.has(token));
+  return meaningful.length === 0;
+}
+
 function mapCounterpartyIntentToContractIntent(intent: AddressIntent): AddressIntent | null {
   if (intent === "list_documents_by_counterparty") {
     return "list_documents_by_contract";
@@ -199,24 +315,30 @@ function mergeFollowupFilters(
     intent === "bank_operations_by_counterparty" ||
     intent === "list_contracts_by_counterparty"
   ) {
-    if (!toNonEmptyString(merged.counterparty)) {
-      const inheritedCounterparty =
-        previousCounterparty ??
-        (followupContext.previous_anchor_type === "counterparty" ? previousAnchorValue : null);
-      if (inheritedCounterparty) {
-        merged.counterparty = inheritedCounterparty;
-        reasons.push("counterparty_from_followup_context");
-      }
+    const inheritedCounterparty =
+      previousCounterparty ??
+      (followupContext.previous_anchor_type === "counterparty" ? previousAnchorValue : null);
+    const currentCounterparty = toNonEmptyString(merged.counterparty);
+    const shouldInheritCounterparty =
+      !currentCounterparty ||
+      (Boolean(inheritedCounterparty) &&
+        isLowQualityCounterpartyAnchor(currentCounterparty) &&
+        !isLowQualityCounterpartyAnchor(inheritedCounterparty));
+    if (inheritedCounterparty && shouldInheritCounterparty) {
+      merged.counterparty = inheritedCounterparty;
+      reasons.push(currentCounterparty ? "counterparty_replaced_from_followup_context" : "counterparty_from_followup_context");
     }
   }
 
   if (intent === "list_documents_by_contract" || intent === "bank_operations_by_contract") {
-    if (!toNonEmptyString(merged.contract)) {
-      const inheritedContract = previousContract ?? (followupContext.previous_anchor_type === "contract" ? previousAnchorValue : null);
-      if (inheritedContract) {
-        merged.contract = inheritedContract;
-        reasons.push("contract_from_followup_context");
-      }
+    const inheritedContract = previousContract ?? (followupContext.previous_anchor_type === "contract" ? previousAnchorValue : null);
+    const currentContract = toNonEmptyString(merged.contract);
+    const shouldInheritContract =
+      !currentContract ||
+      (Boolean(inheritedContract) && isLowQualityContractAnchor(currentContract) && !isLowQualityContractAnchor(inheritedContract));
+    if (inheritedContract && shouldInheritContract) {
+      merged.contract = inheritedContract;
+      reasons.push(currentContract ? "contract_replaced_from_followup_context" : "contract_from_followup_context");
     }
   }
 
@@ -238,21 +360,27 @@ function mergeFollowupFilters(
   }
 
   if (intent === "open_items_by_counterparty_or_contract" || intent === "list_open_contracts") {
-    if (!toNonEmptyString(merged.contract)) {
-      const inheritedContract = previousContract ?? (followupContext.previous_anchor_type === "contract" ? previousAnchorValue : null);
-      if (inheritedContract) {
-        merged.contract = inheritedContract;
-        reasons.push("contract_from_followup_context");
-      }
+    const inheritedContract = previousContract ?? (followupContext.previous_anchor_type === "contract" ? previousAnchorValue : null);
+    const currentContract = toNonEmptyString(merged.contract);
+    const shouldInheritContract =
+      !currentContract ||
+      (Boolean(inheritedContract) && isLowQualityContractAnchor(currentContract) && !isLowQualityContractAnchor(inheritedContract));
+    if (inheritedContract && shouldInheritContract) {
+      merged.contract = inheritedContract;
+      reasons.push(currentContract ? "contract_replaced_from_followup_context" : "contract_from_followup_context");
     }
-    if (!toNonEmptyString(merged.counterparty)) {
-      const inheritedCounterparty =
-        previousCounterparty ??
-        (followupContext.previous_anchor_type === "counterparty" ? previousAnchorValue : null);
-      if (inheritedCounterparty) {
-        merged.counterparty = inheritedCounterparty;
-        reasons.push("counterparty_from_followup_context");
-      }
+    const inheritedCounterparty =
+      previousCounterparty ??
+      (followupContext.previous_anchor_type === "counterparty" ? previousAnchorValue : null);
+    const currentCounterparty = toNonEmptyString(merged.counterparty);
+    const shouldInheritCounterparty =
+      !currentCounterparty ||
+      (Boolean(inheritedCounterparty) &&
+        isLowQualityCounterpartyAnchor(currentCounterparty) &&
+        !isLowQualityCounterpartyAnchor(inheritedCounterparty));
+    if (inheritedCounterparty && shouldInheritCounterparty) {
+      merged.counterparty = inheritedCounterparty;
+      reasons.push(currentCounterparty ? "counterparty_replaced_from_followup_context" : "counterparty_from_followup_context");
     }
   }
 
@@ -342,7 +470,7 @@ function deriveIntentWithFollowupContext(
   ) {
     const preferDocumentsForming =
       hasDocumentSignal(normalizedMessage) &&
-      /(?:раскрой|раскры|формир|документами|по\s+документ)/iu.test(normalizedMessage);
+      /(?:раскрой|раскры|формир|документами)/iu.test(normalizedMessage);
     return {
       intent: preferDocumentsForming ? "documents_forming_balance" : "account_balance_snapshot",
       confidence: "low",
@@ -351,6 +479,23 @@ function deriveIntentWithFollowupContext(
   }
 
   if (hasPreviousContract) {
+    if (detectedIntent.intent === "list_contracts_by_counterparty") {
+      if (hasBankSignal(normalizedMessage)) {
+        return {
+          intent: "bank_operations_by_contract",
+          confidence: "low",
+          reasons: [...detectedIntent.reasons, "intent_adjusted_to_contract_followup_context"]
+        };
+      }
+      if (hasDocumentSignal(normalizedMessage)) {
+        return {
+          intent: "list_documents_by_contract",
+          confidence: "low",
+          reasons: [...detectedIntent.reasons, "intent_adjusted_to_contract_followup_context"]
+        };
+      }
+    }
+
     if (detectedIntent.intent === "unknown") {
       if (hasBankSignal(normalizedMessage)) {
         return {