@librechat/agents 3.9.1 → 3.9.3

package/src/utils/toolContent.ts

@@ -3,6 +3,7 @@ import { ToolMessage, type BaseMessage } from '@langchain/core/messages';
 import {
   HARD_MAX_TOTAL_TOOL_OUTPUT_SIZE,
   truncateToolResultContent,
+  sliceWithoutSplittingSurrogates,
 } from './truncation';
 
 type ToolContent = BaseMessage['content'];
@@ -421,7 +422,7 @@ type SegmentedStringBuffer = {
 export type BoundedStructuredSerialization = {
   /** Provider-facing head/tail preview, bounded by `maxChars`. */
   content: string;
-  /** Exact serialized prefix up to the defensive traversal-work ceiling. */
+  /** Code-point-aligned serialized prefix up to the traversal-work ceiling. */
   prefix: string;
   /** Exact length, or `Number.MAX_SAFE_INTEGER` when traversal was capped. */
   originalChars: number;
@@ -505,6 +506,10 @@ function materializeSegmentedStringBuffer(
   return segments.join('');
 }
 
+/**
+ * Retains exact code-unit prefixes internally. Trimming surrogate edges here
+ * would let later chunks fill the gap and corrupt the prefix; trim on exposure.
+ */
 function appendSerializedChunk(
   collector: StructuredSerializationCollector,
   chunk: string
@@ -953,7 +958,7 @@ function formatBoundedStructuredContent(
   suffix: string
 ): string {
   if (collector.totalChars <= maxChars) {
-    return prefix.slice(0, collector.totalChars);
+    return sliceWithoutSplittingSurrogates(prefix, 0, collector.totalChars);
   }
 
   const indicator = collector.work.exceeded
@@ -962,10 +967,13 @@ function formatBoundedStructuredContent(
       `${maxChars} limit] …\n\n`;
   const available = maxChars - indicator.length;
   if (available <= 0) {
-    return prefix.slice(0, maxChars);
+    return sliceWithoutSplittingSurrogates(prefix, 0, maxChars);
   }
   if (available < 200) {
-    return prefix.slice(0, available) + indicator.trimEnd();
+    return (
+      sliceWithoutSplittingSurrogates(prefix, 0, available) +
+      indicator.trimEnd()
+    );
   }
 
   const headSize = Math.ceil(available * 0.7);
@@ -982,7 +990,11 @@ function formatBoundedStructuredContent(
     tailStart = tailNewline + 1;
   }
 
-  return prefix.slice(0, headEnd) + indicator + suffix.slice(tailStart);
+  return (
+    sliceWithoutSplittingSurrogates(prefix, 0, headEnd) +
+    indicator +
+    sliceWithoutSplittingSurrogates(suffix, tailStart)
+  );
 }
 
 /**
@@ -1036,7 +1048,7 @@ export function serializeStructuredValueBounded(
       prefix,
       suffix
     ),
-    prefix: prefix.slice(0, normalizedPrefixChars),
+    prefix: sliceWithoutSplittingSurrogates(prefix, 0, normalizedPrefixChars),
     originalChars,
     truncated:
       collector.work.exceeded || collector.totalChars > normalizedMaxChars,
@@ -1931,7 +1943,11 @@ function serializeDenseTextBlocksWithinLimit(
     }
     preview += textBlocks[i].text.slice(0, available - preview.length);
   }
-  return (preview + indicator).slice(0, normalizedMaxChars);
+  return sliceWithoutSplittingSurrogates(
+    sliceWithoutSplittingSurrogates(preview, 0) + indicator,
+    0,
+    normalizedMaxChars
+  );
 }
 
 export type CompactToolContentResult = {

package/src/utils/truncation.ts

@@ -5,6 +5,22 @@
  * consuming the entire context window.
  */
 
+/** Slices by UTF-16 code units, dropping orphaned surrogate halves at either edge. */
+export function sliceWithoutSplittingSurrogates(
+  value: string,
+  start: number,
+  end?: number
+): string {
+  const sliced = value.slice(start, end);
+  const first = sliced.charCodeAt(0);
+  const last = sliced.charCodeAt(sliced.length - 1);
+  const trimStart = first >= 0xdc00 && first <= 0xdfff;
+  const trimEnd = last >= 0xd800 && last <= 0xdbff;
+  return trimStart || trimEnd
+    ? sliced.slice(trimStart ? 1 : 0, sliced.length - (trimEnd ? 1 : 0))
+    : sliced;
+}
+
 /**
  * Absolute hard cap on tool result length (characters).
  * Even if the model has a 1M-token context, a single tool result
@@ -85,7 +101,9 @@ export function truncateToolInput(
 
   if (available < 100) {
     return {
-      _truncated: serialized.slice(0, maxChars) + indicator.trimEnd(),
+      _truncated:
+        sliceWithoutSplittingSurrogates(serialized, 0, maxChars) +
+        indicator.trimEnd(),
       _originalChars: serialized.length,
     };
   }
@@ -95,9 +113,9 @@ export function truncateToolInput(
 
   return {
     _truncated:
-      serialized.slice(0, headSize) +
+      sliceWithoutSplittingSurrogates(serialized, 0, headSize) +
       indicator +
-      serialized.slice(serialized.length - tailSize),
+      sliceWithoutSplittingSurrogates(serialized, serialized.length - tailSize),
     _originalChars: serialized.length,
   };
 }
@@ -126,12 +144,15 @@ export function truncateToolResultContent(
   const indicator = `\n\n… [truncated: ${content.length} chars exceeded ${maxChars} limit] …\n\n`;
   const available = maxChars - indicator.length;
   if (available <= 0) {
-    return content.slice(0, maxChars);
+    return sliceWithoutSplittingSurrogates(content, 0, maxChars);
   }
 
   // When budget is too small for a meaningful tail, fall back to head-only
   if (available < 200) {
-    return content.slice(0, available) + indicator.trimEnd();
+    return (
+      sliceWithoutSplittingSurrogates(content, 0, available) +
+      indicator.trimEnd()
+    );
   }
 
   const headSize = Math.ceil(available * 0.7);
@@ -150,7 +171,11 @@ export function truncateToolResultContent(
     tailStart = tailNewline + 1;
   }
 
-  return content.slice(0, headEnd) + indicator + content.slice(tailStart);
+  return (
+    sliceWithoutSplittingSurrogates(content, 0, headEnd) +
+    indicator +
+    sliceWithoutSplittingSurrogates(content, tailStart)
+  );
 }
 
 /** Absolute hard cap on a single tool-call input (characters). */