fix(errors): detect HTML after HTTP reason phrases

Share one internal HTTP response-body parser across assistant error formatting and provider runtime failure classification so reason phrases do not hide complete HTML error pages. Fixes #122244.
2026-08-27 04:47:03 -06:00 · 2026-08-11 21:40:58 +00:00
parent f73b436f05
commit a816c4a3b4
7 changed files with 64 additions and 3 deletions
@@ -17,6 +17,12 @@ function expectNotFailoverSample(sample: string) {
 }

 describe("classifyProviderRuntimeFailureKind", () => {
+  it("classifies complete HTML after an HTTP reason phrase as upstream_html", () => {
+    const raw = "HTTP 502 Bad Gateway\n\n<!doctype html><html><body>down</body></html>";
+
+    expect(classifyProviderRuntimeFailureKind(raw)).toBe("upstream_html");
+  });
+
  it("classifies generic resource-exhausted codes as rate_limit", () => {
    expect(
      classifyProviderRuntimeFailureKind({
@@ -1,5 +1,6 @@
 import { normalizeLowercaseStringOrEmpty } from "@openclaw/normalization-core/string-coerce";
 import { extractLeadingHttpStatus } from "../../shared/assistant-error-format.js";
+import { extractHttpResponseBody } from "../../shared/http-error-response.js";
 import { classifyOAuthRefreshFailure } from "../auth-profiles/oauth-refresh-failure.js";
 import { formatExecDeniedUserMessage } from "../exec-approval-result.js";
 import {
@@ -70,7 +71,7 @@ function isHtmlErrorResponse(raw: string, status?: number): boolean {
  if (typeof inferred !== "number" || inferred < 400) {
    return false;
  }
-  const rest = extractLeadingHttpStatus(candidate)?.rest ?? candidate;
+  const rest = extractHttpResponseBody(extractLeadingHttpStatus(candidate))?.body ?? candidate;
  return HTML_BODY_RE.test(rest) && HTML_CLOSE_RE.test(rest);
 }
 function isCloudflareChallengeResponse(message: string): boolean {
@@ -17,6 +17,16 @@ describe("isCloudflareOrHtmlErrorPage", () => {
    expect(isCloudflareOrHtmlErrorPage(htmlError)).toBe(true);
  });

+  it("detects complete 5xx HTML pages after an HTTP reason phrase", () => {
+    const htmlError = "HTTP 502 Bad Gateway\n\n<!doctype html><html><body>down</body></html>";
+    expect(isCloudflareOrHtmlErrorPage(htmlError)).toBe(true);
+  });
+
+  it("does not flag partial HTML after an HTTP reason phrase", () => {
+    const partialHtml = "HTTP 502 Bad Gateway\n\n<!doctype html><html><body>down";
+    expect(isCloudflareOrHtmlErrorPage(partialHtml)).toBe(false);
+  });
+
  it("detects standalone Cloudflare challenge HTML pages", () => {
    // HTML challenge pages are provider transport failures, not model text.
    const htmlError = `<!DOCTYPE html>
@@ -81,6 +81,18 @@ describe("extractErrorHttpStatus", () => {
 });

 describe("HTTP status consumers", () => {
+  it("does not return raw HTML after an HTTP reason phrase", () => {
+    const raw = [
+      "HTTP 502 Bad Gateway",
+      "",
+      "<!doctype html><html><body><h1>502</h1></body></html>",
+    ].join("\n");
+
+    expect(formatRawAssistantErrorForUi(raw)).toBe(
+      "The AI service is temporarily unavailable (HTTP 502). Please try again in a moment.",
+    );
+  });
+
  it("formats only status lines inside the HTTP range", () => {
    expect(formatRawAssistantErrorForUi("100 Continue")).toBe("HTTP 100: Continue");
    expect(formatRawAssistantErrorForUi("599 Provider Error")).toBe("HTTP 599: Provider Error");
@@ -1,5 +1,6 @@
 // Assistant error formatting helpers normalize assistant-visible error payloads.
 import { truncateUtf16Safe } from "@openclaw/normalization-core/utf16-slice";
+import { extractHttpResponseBody } from "./http-error-response.js";
 const ERROR_PAYLOAD_PREFIX_RE =
  /^(?:error|(?:[a-z][\w-]*\s+)?api\s*error|apierror|openai\s*error|anthropic\s*error|gateway\s*error|codex\s*error)(?:\s+\d{3})?[:\s-]+/i;
 const HTTP_STATUS_DELIMITER_RE = /(?:\s*:\s*|\s+)/;
@@ -160,7 +161,7 @@ export function isCloudflareOrHtmlErrorPage(raw: string): boolean {
    return true;
  }

-  const status = extractLeadingHttpStatus(trimmed);
+  const status = extractHttpResponseBody(extractLeadingHttpStatus(trimmed));
  if (!status || status.code < 500) {
    return false;
  }
@@ -170,7 +171,7 @@ export function isCloudflareOrHtmlErrorPage(raw: string): boolean {
  }

  return (
-    status.code < 600 && HTML_ERROR_PREFIX_RE.test(status.rest) && HTML_CLOSE_RE.test(status.rest)
+    status.code < 600 && HTML_ERROR_PREFIX_RE.test(status.body) && HTML_CLOSE_RE.test(status.body)
  );
 }

@@ -0,0 +1,17 @@
+const HTML_ERROR_PREFIX_RE = /^\s*(?:<!doctype\s+html\b|<html\b)/i;
+
+export function extractHttpResponseBody(
+  status: { code: number; rest: string } | null,
+): { code: number; body: string } | null {
+  if (!status) {
+    return null;
+  }
+  if (HTML_ERROR_PREFIX_RE.test(status.rest)) {
+    return { code: status.code, body: status.rest };
+  }
+  const lineBreak = status.rest.indexOf("\n");
+  return {
+    code: status.code,
+    body: lineBreak === -1 ? status.rest : status.rest.slice(lineBreak + 1).trimStart(),
+  };
+}
@@ -12,11 +12,25 @@ import {
  isolateRtlRenderedLine,
  isTerminalSafeAutocompleteValue,
  isCommandMarkedMessage,
+  resolveFinalAssistantText,
  sanitizeMarkdownSource,
  sanitizeRenderableLine,
  sanitizeRenderableText,
 } from "./tui-formatters.js";

+describe("resolveFinalAssistantText", () => {
+  it("hides complete HTML error pages after an HTTP reason phrase", () => {
+    const raw = "HTTP 502 Bad Gateway\n\n<!doctype html><html><body>down</body></html>";
+
+    const rendered = resolveFinalAssistantText({ errorMessage: raw });
+
+    expect(rendered).toBe(
+      "The AI service is temporarily unavailable (HTTP 502). Please try again in a moment.",
+    );
+    expect(rendered).not.toContain("<html>");
+  });
+});
+
 describe("formatTuiFooter", () => {
  it("shows session modes and the process delivery mode in one compact summary", () => {
    expect(