test: 补充知识处理、检索与 Agent 回归用例

覆盖切块、向量检索、工具调用和历史预算；模型相关用例使用桩。

backend/tests/knowledge/test_pdf_download.py

@@ -0,0 +1,122 @@
+"""
+Copyright 2024-2026 ReplyPier
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+"""
+
+from unittest.mock import patch
+
+import httpx
+import pytest
+
+from app.knowledge import pdf_download, url_safety
+from app.knowledge.pdf_download import PdfTooLargeError, download_pdf
+from app.knowledge.url_safety import BlockedHostError
+
+PDF = b"%PDF-1.4 test body"
+PUBLIC_IP = [(2, 1, 6, "", ("93.184.216.34", 0))]
+
+
+@pytest.fixture
+def served(monkeypatch):
+    """Route download_pdf's httpx client to an in-process handler and record
+    every URL it was asked to open."""
+    requested = []
+    routes = {}
+
+    def handler(request: httpx.Request) -> httpx.Response:
+        url = str(request.url)
+        requested.append(url)
+        return routes[url]()
+
+    real_client = httpx.Client
+    monkeypatch.setattr(
+        pdf_download.httpx, "Client",
+        lambda **kwargs: real_client(transport=httpx.MockTransport(handler), **kwargs),
+    )
+    return routes, requested
+
+
+@pytest.fixture
+def public_dns():
+    with patch.object(url_safety.socket, "getaddrinfo", return_value=PUBLIC_IP):
+        yield
+
+
+@pytest.mark.parametrize("url", [
+    "http://127.0.0.1/internal.pdf",
+    "http://localhost:8000/api/v1/secret.pdf",
+    "http://169.254.169.254/latest/meta-data/",
+])
+def test_internal_url_is_refused_before_any_request(served, tmp_path, url):
+    _, requested = served
+    with pytest.raises(BlockedHostError):
+        download_pdf(url, str(tmp_path / "out.pdf"))
+    assert requested == []
+
+
+def test_hostname_resolving_to_internal_address_is_refused(served, tmp_path):
+    _, requested = served
+    with patch.object(url_safety.socket, "getaddrinfo",
+                      return_value=[(2, 1, 6, "", ("10.0.0.7", 0))]):
+        with pytest.raises(BlockedHostError):
+            download_pdf("https://pdfs.attacker.example/a.pdf", str(tmp_path / "out.pdf"))
+    assert requested == []
+
+
+def test_public_url_redirecting_to_private_address_is_refused(served, public_dns, tmp_path):
+    routes, requested = served
+    routes["https://files.example.com/doc.pdf"] = lambda: httpx.Response(
+        302, headers={"location": "http://169.254.169.254/latest/meta-data/"})
+    with pytest.raises(BlockedHostError):
+        download_pdf("https://files.example.com/doc.pdf", str(tmp_path / "out.pdf"))
+    # The internal hop was never opened.
+    assert requested == ["https://files.example.com/doc.pdf"]
+
+
+def test_public_url_is_streamed_to_disk_following_safe_redirects(served, public_dns, tmp_path):
+    routes, _ = served
+    routes["https://files.example.com/doc.pdf"] = lambda: httpx.Response(
+        301, headers={"location": "/v2/doc.pdf"})
+    routes["https://files.example.com/v2/doc.pdf"] = lambda: httpx.Response(200, content=PDF)
+    dest = tmp_path / "out.pdf"
+
+    assert download_pdf("https://files.example.com/doc.pdf", str(dest)) == str(dest)
+    assert dest.read_bytes() == PDF
+
+
+def test_declared_oversize_is_refused(served, public_dns, tmp_path):
+    routes, _ = served
+    routes["https://files.example.com/big.pdf"] = lambda: httpx.Response(
+        200, content=PDF, headers={"content-length": str(10 ** 9)})
+    with pytest.raises(PdfTooLargeError):
+        download_pdf("https://files.example.com/big.pdf", str(tmp_path / "out.pdf"))
+
+
+def test_body_past_the_cap_is_refused_even_without_content_length(served, public_dns, tmp_path):
+    routes, _ = served
+
+    def chunked():
+        return httpx.Response(200, content=iter([PDF] * 10))  # streamed: no length
+
+    routes["https://files.example.com/big.pdf"] = chunked
+    with pytest.raises(PdfTooLargeError):
+        download_pdf("https://files.example.com/big.pdf", str(tmp_path / "out.pdf"),
+                     max_bytes=len(PDF) * 2)
+
+
+def test_http_error_is_raised(served, public_dns, tmp_path):
+    routes, _ = served
+    routes["https://files.example.com/gone.pdf"] = lambda: httpx.Response(404)
+    with pytest.raises(httpx.HTTPStatusError):
+        download_pdf("https://files.example.com/gone.pdf", str(tmp_path / "out.pdf"))