fix(backend): 修复二进制产物误判文本导致 PDF 返回异常

backend/app/gateway/routers/artifacts.py

@@ -176,6 +176,11 @@ async def get_artifact(thread_id: str, path: str, request: Request, download: bo
         return PlainTextResponse(content=actual_path.read_text(encoding="utf-8"), media_type=mime_type)
 
     if is_text_file_by_content(actual_path):
-        return PlainTextResponse(content=actual_path.read_text(encoding="utf-8"), media_type=mime_type)
+        try:
+            return PlainTextResponse(content=actual_path.read_text(encoding="utf-8"), media_type=mime_type)
+        except UnicodeDecodeError:
+            # Some binary formats (e.g. certain PDFs) may not contain NUL bytes in
+            # the sampled chunk and be misclassified as text. Fall back to binary.
+            logger.debug("Artifact looked like text but is not valid UTF-8: %s", actual_path, exc_info=True)
 
     return Response(content=actual_path.read_bytes(), media_type=mime_type, headers={"Content-Disposition": _build_content_disposition("inline", actual_path.name)})

backend/tests/test_artifacts_router.py

@@ -102,3 +102,18 @@ def test_get_artifact_download_true_forces_attachment_for_skill_archive(tmp_path
     assert response.status_code == 200
     assert response.text == "hello"
     assert response.headers.get("content-disposition", "").startswith("attachment;")
+
+
+def test_get_artifact_pdf_with_no_null_bytes_and_non_utf8_content_is_served_inline(tmp_path, monkeypatch) -> None:
+    artifact_path = tmp_path / "slides.pdf"
+    # No NUL bytes, but invalid UTF-8 to simulate binary content misdetected as text.
+    binary_content = b"%PDF-1.7\n\xff\xfe\xfa\n%%EOF"
+    artifact_path.write_bytes(binary_content)
+
+    monkeypatch.setattr(artifacts_router, "resolve_thread_virtual_path", lambda _thread_id, _path: artifact_path)
+
+    response = asyncio.run(artifacts_router.get_artifact("thread-1", "mnt/user-data/outputs/slides.pdf", _make_request()))
+
+    assert bytes(response.body) == binary_content
+    assert response.media_type == "application/pdf"
+    assert response.headers.get("content-disposition", "").startswith("inline;")