feat: webdav download via httpx mit basic-auth

app/ingest/webdav.py

@@ -0,0 +1,21 @@
+import httpx
+
+
+class WebDAVError(Exception):
+    pass
+
+
+async def download_file(base_url: str, user: str, password: str, file_path: str, *, timeout: float = 60.0) -> bytes:
+    """Fetch a file from Nextcloud WebDAV. Returns the raw bytes."""
+    base = base_url.rstrip("/")
+    rel = file_path.lstrip("/")
+    url = f"{base}/{rel}"
+
+    async with httpx.AsyncClient(auth=(user, password), timeout=timeout) as client:
+        response = await client.get(url)
+
+    if response.status_code != 200:
+        raise WebDAVError(
+            f"WebDAV GET {file_path} failed: status={response.status_code}"
+        )
+    return response.content

tests/test_webdav.py

@@ -0,0 +1,44 @@
+import pytest
+import httpx
+import respx
+
+from app.ingest.webdav import download_file, WebDAVError
+
+
+@pytest.mark.asyncio
+async def test_download_file_returns_bytes():
+    base = "https://nc.example.com/remote.php/dav/files/u"
+    file_path = "Documents/THB/Studium/2.Semester/Databases/x.pdf"
+    expected_url = f"{base}/{file_path}"
+
+    with respx.mock(base_url=base) as mock:
+        mock.get(expected_url).mock(return_value=httpx.Response(200, content=b"PDFBYTES"))
+        data = await download_file(base, "u", "pw", file_path)
+        assert data == b"PDFBYTES"
+
+
+@pytest.mark.asyncio
+async def test_download_file_404_raises():
+    base = "https://nc.example.com/remote.php/dav/files/u"
+    file_path = "missing.pdf"
+    expected_url = f"{base}/{file_path}"
+
+    with respx.mock(base_url=base) as mock:
+        mock.get(expected_url).mock(return_value=httpx.Response(404))
+        with pytest.raises(WebDAVError):
+            await download_file(base, "u", "pw", file_path)
+
+
+@pytest.mark.asyncio
+async def test_download_file_handles_url_encoding():
+    base = "https://nc.example.com/remote.php/dav/files/u"
+    file_path = "Documents/Folder With Space/file.pdf"
+
+    with respx.mock(base_url=base) as mock:
+        # httpx will percent-encode the spaces
+        route = mock.get(url__regex=r".*/Folder%20With%20Space/file\.pdf").mock(
+            return_value=httpx.Response(200, content=b"OK")
+        )
+        data = await download_file(base, "u", "pw", file_path)
+        assert data == b"OK"
+        assert route.called