fix: error handling when http error during LT download (jxmorris12#150)

Author: mdevolde

language_tool_python/download_lt.py

@@ -172,7 +172,8 @@ def http_get(
     :param proxies: Optional dictionary of proxies to use for the request.
     :type proxies: Optional[Dict[str, str]]
     :raises TimeoutError: If the request times out.
-    :raises PathError: If the file could not be found at the given URL (HTTP 404).
+    :raises PathError: If the file could not be found at the given URL (HTTP 404),
+                       if access is forbidden (HTTP 403), or for other HTTP errors.
     """
     logger.info("Starting download from %s", url)
     try:
@@ -185,6 +186,12 @@ def http_get(
     if req.status_code == 404:
         err = f"Could not find at URL {url}. The given version may not exist or is no longer available."
         raise PathError(err)
+    if req.status_code == 403:
+        err = f"Access forbidden to URL {url}. You may not have permission to access this resource. It may be related to network restrictions (e.g., firewall, proxy settings)."
+        raise PathError(err)
+    if req.status_code != 200:
+        err = f"Failed to download from {url}. HTTP status code: {req.status_code}."
+        raise PathError(err)
     version = (
         url.split("/")[-1].split("-")[1].replace("-snapshot", "").replace(".zip", "")
     )

tests/test_download.py

@@ -1,8 +1,12 @@
 """Tests for the download/language functionality of LanguageTool."""
 
+import io
+from unittest.mock import MagicMock, patch
+
 import pytest
 
-from language_tool_python.exceptions import LanguageToolError
+from language_tool_python.download_lt import http_get
+from language_tool_python.exceptions import LanguageToolError, PathError
 
 
 def test_install_inexistent_version() -> None:
@@ -31,3 +35,51 @@ def test_inexistent_language() -> None:
 
     with language_tool_python.LanguageTool("en-US") as tool, pytest.raises(ValueError):
         language_tool_python.LanguageTag("xx-XX", tool._get_languages())
+
+
+def test_http_get_403_forbidden() -> None:
+    """
+    Test that http_get raises PathError when receiving a 403 Forbidden status code.
+    This test verifies that the function correctly handles forbidden access errors
+    when attempting to download files.
+
+    :raises AssertionError: If PathError is not raised for a 403 status code.
+    """
+    mock_response = MagicMock()
+    mock_response.status_code = 403
+    mock_response.headers = {}
+
+    with (
+        patch(
+            "language_tool_python.download_lt.requests.get", return_value=mock_response
+        ),
+        pytest.raises(PathError, match="Access forbidden to URL"),
+    ):
+        out_file = io.BytesIO()
+        http_get("https://example.com/test.zip", out_file)
+
+
+def test_http_get_other_error_codes() -> None:
+    """
+    Test that http_get raises PathError for various HTTP error codes (other than 404 and 403).
+    This test verifies that the function correctly handles different HTTP error codes
+    like 500 (Internal Server Error), 503 (Service Unavailable), etc.
+
+    :raises AssertionError: If PathError is not raised for error status codes.
+    """
+    error_codes = [500, 502, 503, 504]
+
+    for error_code in error_codes:
+        mock_response = MagicMock()
+        mock_response.status_code = error_code
+        mock_response.headers = {}
+
+        with (
+            patch(
+                "language_tool_python.download_lt.requests.get",
+                return_value=mock_response,
+            ),
+            pytest.raises(PathError, match=f"Failed to download.*{error_code}"),
+        ):
+            out_file = io.BytesIO()
+            http_get("https://example.com/test.zip", out_file)