perf: switch to HTTP request for GSHHG SAD

src/fcollections/sad/_gshhg.py

@@ -5,7 +5,8 @@
 import logging
 import tarfile
 import typing as tp
-from ftplib import FTP
+
+import requests
 
 from ._interface import IAuxiliaryDataFetcher
 
@@ -33,7 +34,7 @@ class GSHHG(IAuxiliaryDataFetcher):
         the user home (~/.config/sad)
     """
 
-    FTP_URL = "ftp.soest.hawaii.edu"
+    HTTP_URL = "http://www.soest.hawaii.edu/pwessel"
     FILE = "gshhg/gshhg-gmt-2.3.7.tar.gz"
 
     @property
@@ -43,38 +44,40 @@ def keys(self) -> set[str]:
         return {f"{s}_{r}" for s, r in itertools.product(subset, resolutions)}
 
     def _download(self, remote_file: str, target_folder: Path):
-        fetch_ftp_file(self.FTP_URL, self.FILE, target_folder)
+        fetch_http_file(self.HTTP_URL, self.FILE, target_folder)
         return target_folder / remote_file
 
     def _file_name(self, key: str):
         return f"binned_{key}.nc"
 
 
-def fetch_ftp_file(url: str, filename: str, target_folder: Path):
+def fetch_http_file(url: str, filename: str, target_folder: Path):
 
-    logger.debug("Connecting as anonymous to %s", url)
-    ftp = FTP(url)
-    ftp.login()
+    full_url = url + "/" + filename
 
-    # Download in-memory. This should be limited to a few MB
-    logger.info("Downloading %s...", filename)
+    logger.info("Downloading %s...", full_url)
     tar_data = io.BytesIO()
-    ftp.retrbinary(f"RETR {filename}", tar_data.write)
-    ftp.quit()
-    logger.info("Downloading %s... Done", filename)
+    response = requests.get(full_url, timeout=60)
+    try:
+        response.raise_for_status()
+    except requests.exceptions.RequestException as e:
+        raise RuntimeError(f"Failed to download file from {full_url}") from e
 
     # Filter out non-netcdf and flatten the tar gz structure
     def tar_info_filter(tar_info: tarfile.TarInfo, _) -> tarfile.TarInfo | None:
         if ".nc" not in tar_info.name:
             logger.debug("Not an netcdf, skipping extraction")
             return None
 
-        tar_info.name = tar_info.name.split("/")[-1]
+        logger.debug("Extracting %s", tar_info.name)
+        tar_info.path = tar_info.path.split("/")[-1]
         return tar_info
 
     # Extract in-memory buffer
+    tar_data.write(response.content)
     tar_data.seek(0)
     with tarfile.open(fileobj=tar_data, mode="r") as tar:
         for member in tar.getmembers():
-            logger.debug("Extracting %s", member.name)
             tar.extract(member, path=target_folder, filter=tar_info_filter)
+
+    logger.info("Downloading %s... Done", full_url)

tests/sad/test_aux.py

@@ -66,23 +66,41 @@ def gshhg_tar_gz() -> bytes:
 def test_gshhg(gshhg_tar_gz: bytes, tmp_path_factory: pytest.TempPathFactory):
     path = tmp_path_factory.mktemp("sad")
 
-    def retrbinary_side_effect(_, callback):
-        callback(gshhg_tar_gz)
+    buf = io.BytesIO()
+    with tarfile.open(fileobj=buf, mode="w") as tar:
+        data = b"Hello World"
+        info = tarfile.TarInfo(name="binned_border_i.nc")
+        info.size = len(data)
+        tar.addfile(info, io.BytesIO(data))
 
-    ftp_mock = Mock()
-    ftp_mock.retrbinary.side_effect = retrbinary_side_effect
+    mock_resp = Mock()
+    mock_resp.content = buf.getvalue()
+    mock_resp.raise_for_status.return_value = None
 
     aux = GSHHG()
     with (
-        patch("fcollections.sad._gshhg.FTP", return_value=ftp_mock),
+        patch("requests.get", return_value=mock_resp) as get,
         patch("fcollections.sad.GSHHG.lookup_folders", return_value=[path]),
     ):
         fetched_file = aux["border_i"]
-        with open(fetched_file, "rb") as f:
-            assert f.read() == b"hello"
 
-        fetched_file = aux["GSHHS_h"]
-        with open(fetched_file, "rb") as f:
-            assert f.read() == b"world"
+    with open(fetched_file) as f:
+        assert f.read() == "Hello World"
 
-    ftp_mock.retrbinary.assert_called_once()
+    get.assert_called_once_with(
+        aux.HTTP_URL + "/gshhg/gshhg-gmt-2.3.7.tar.gz", timeout=60
+    )
+
+
+def test_gshhg_http_error(tmp_path_factory: pytest.TempPathFactory):
+    path = tmp_path_factory.mktemp("sad")
+    mock_resp = Mock()
+    mock_resp.raise_for_status.side_effect = requests.HTTPError("404")
+
+    aux = GSHHG()
+    with (
+        patch("fcollections.sad._gshhg.requests.get", return_value=mock_resp),
+        patch("fcollections.sad.GSHHG.lookup_folders", return_value=[path]),
+    ):
+        with pytest.raises(RuntimeError):
+            aux["border_i"]