fix(parallel): keep normalized search fields UTF-16 safe

ly85206559 · ly85206559 · commit a6fe951e229f · 2026-07-07T21:02:47.000+08:00
diff --git a/extensions/parallel/src/parallel-search-normalize.ts b/extensions/parallel/src/parallel-search-normalize.ts
@@ -8,6 +8,7 @@ import {
   wrapWebContent,
 } from "openclaw/plugin-sdk/provider-web-search";
 import { normalizeOptionalString } from "openclaw/plugin-sdk/string-coerce-runtime";
+import { truncateUtf16Safe } from "openclaw/plugin-sdk/text-utility-runtime";
 
 // Internal-only bounds (the model-facing tool schema declares its own copies).
 const PARALLEL_MAX_SEARCH_COUNT = 40;
@@ -58,7 +59,7 @@ export function normalizeParallelObjective(value: string | undefined): string |
   }
   return trimmed.length <= PARALLEL_MAX_OBJECTIVE_CHARS
     ? trimmed
-    : trimmed.slice(0, PARALLEL_MAX_OBJECTIVE_CHARS);
+    : truncateUtf16Safe(trimmed, PARALLEL_MAX_OBJECTIVE_CHARS);
 }
 
 export function normalizeParallelClientModel(value: string | undefined): string | undefined {
@@ -68,7 +69,7 @@ export function normalizeParallelClientModel(value: string | undefined): string
   }
   return trimmed.length <= PARALLEL_CLIENT_MODEL_MAX_LENGTH
     ? trimmed
-    : trimmed.slice(0, PARALLEL_CLIENT_MODEL_MAX_LENGTH);
+    : truncateUtf16Safe(trimmed, PARALLEL_CLIENT_MODEL_MAX_LENGTH);
 }
 
 // Parallel's API caps each entry at 200 chars and accepts up to 5 queries. We
@@ -90,7 +91,7 @@ export function normalizeParallelSearchQueries(value: unknown): string[] {
     const capped =
       trimmed.length <= PARALLEL_MAX_SEARCH_QUERY_CHARS
         ? trimmed
-        : trimmed.slice(0, PARALLEL_MAX_SEARCH_QUERY_CHARS);
+        : truncateUtf16Safe(trimmed, PARALLEL_MAX_SEARCH_QUERY_CHARS);
     if (seen.has(capped)) {
       continue;
     }
diff --git a/extensions/parallel/src/parallel-web-search-provider.test.ts b/extensions/parallel/src/parallel-web-search-provider.test.ts
@@ -64,6 +64,8 @@ import { testing } from "../test-api.js";
 import { createParallelWebSearchProvider as createContractParallelWebSearchProvider } from "../web-search-contract-api.js";
 import { createParallelWebSearchProvider } from "./parallel-web-search-provider.js";
 
+const loneHighSurrogate = /[\uD800-\uDBFF](?![\uDC00-\uDFFF])/;
+
 describe("parallel web search provider", () => {
   beforeEach(() => {
     endpointMockState.calls = [];
@@ -254,6 +256,9 @@ describe("parallel web search provider", () => {
     expect(testing.normalizeParallelObjective(undefined)).toBeUndefined();
     expect(testing.normalizeParallelObjective("")).toBeUndefined();
     expect((testing.normalizeParallelObjective("x".repeat(6000)) ?? "").length).toBe(5000);
+    const boundary = testing.normalizeParallelObjective(`${"x".repeat(4999)}\u{1f63e}tail`);
+    expect(boundary).toBe("x".repeat(4999));
+    expect(loneHighSurrogate.test(boundary ?? "")).toBe(false);
   });
 
   it("normalizes search_queries: trim, drop blanks, dedupe, cap length, cap count", () => {
@@ -270,6 +275,14 @@ describe("parallel web search provider", () => {
     expect(testing.normalizeParallelSearchQueries(undefined)).toEqual([]);
     expect(testing.normalizeParallelSearchQueries("openclaw github")).toEqual([]);
     expect(testing.normalizeParallelSearchQueries(["x".repeat(250)])).toEqual(["x".repeat(200)]);
+    expect(testing.normalizeParallelSearchQueries([`${"x".repeat(199)}\u{1f63e}tail`])).toEqual([
+      "x".repeat(199),
+    ]);
+    expect(
+      loneHighSurrogate.test(
+        testing.normalizeParallelSearchQueries([`${"x".repeat(199)}\u{1f63e}tail`])[0] ?? "",
+      ),
+    ).toBe(false);
     const six = ["a", "b", "c", "d", "e", "f"];
     expect(testing.normalizeParallelSearchQueries(six)).toEqual(["a", "b", "c", "d", "e"]);
   });
@@ -289,6 +302,9 @@ describe("parallel web search provider", () => {
     expect(testing.normalizeParallelClientModel("  gpt-5.5  ")).toBe("gpt-5.5");
     expect(testing.normalizeParallelClientModel(undefined)).toBeUndefined();
     expect((testing.normalizeParallelClientModel("a".repeat(200)) ?? "").length).toBe(100);
+    const boundary = testing.normalizeParallelClientModel(`${"m".repeat(99)}\u{1f63e}tail`);
+    expect(boundary).toBe("m".repeat(99));
+    expect(loneHighSurrogate.test(boundary ?? "")).toBe(false);
   });
 
   it("normalizes the Parallel /v1/search response shape", () => {

Original file line number	Diff line number	Diff line change
`@@ -8,6 +8,7 @@ import {`
`8`	`8`	`wrapWebContent,`
`9`	`9`	`} from "openclaw/plugin-sdk/provider-web-search";`
`10`	`10`	`import { normalizeOptionalString } from "openclaw/plugin-sdk/string-coerce-runtime";`
	`11`	`+import { truncateUtf16Safe } from "openclaw/plugin-sdk/text-utility-runtime";`
`11`	`12`
`12`	`13`	`// Internal-only bounds (the model-facing tool schema declares its own copies).`
`13`	`14`	`const PARALLEL_MAX_SEARCH_COUNT = 40;`
`@@ -58,7 +59,7 @@ export function normalizeParallelObjective(value: string \| undefined): string \|`
`58`	`59`	`}`
`59`	`60`	`return trimmed.length <= PARALLEL_MAX_OBJECTIVE_CHARS`
`60`	`61`	`? trimmed`
`61`		`- : trimmed.slice(0, PARALLEL_MAX_OBJECTIVE_CHARS);`
	`62`	`+ : truncateUtf16Safe(trimmed, PARALLEL_MAX_OBJECTIVE_CHARS);`
`62`	`63`	`}`
`63`	`64`
`64`	`65`	`export function normalizeParallelClientModel(value: string \| undefined): string \| undefined {`
`@@ -68,7 +69,7 @@ export function normalizeParallelClientModel(value: string \| undefined): string`
`68`	`69`	`}`
`69`	`70`	`return trimmed.length <= PARALLEL_CLIENT_MODEL_MAX_LENGTH`
`70`	`71`	`? trimmed`
`71`		`- : trimmed.slice(0, PARALLEL_CLIENT_MODEL_MAX_LENGTH);`
	`72`	`+ : truncateUtf16Safe(trimmed, PARALLEL_CLIENT_MODEL_MAX_LENGTH);`
`72`	`73`	`}`
`73`	`74`
`74`	`75`	`// Parallel's API caps each entry at 200 chars and accepts up to 5 queries. We`
`@@ -90,7 +91,7 @@ export function normalizeParallelSearchQueries(value: unknown): string[] {`
`90`	`91`	`const capped =`
`91`	`92`	`trimmed.length <= PARALLEL_MAX_SEARCH_QUERY_CHARS`
`92`	`93`	`? trimmed`
`93`		`- : trimmed.slice(0, PARALLEL_MAX_SEARCH_QUERY_CHARS);`
	`94`	`+ : truncateUtf16Safe(trimmed, PARALLEL_MAX_SEARCH_QUERY_CHARS);`
`94`	`95`	`if (seen.has(capped)) {`
`95`	`96`	`continue;`
`96`	`97`	`}`