From 5483bd300421f56c62047197e8f543b0ab2ceb34 Mon Sep 17 00:00:00 2001
From: Shubham Padkonde <shubhampadkonde12@gmail.com>
Date: Mon, 26 Oct 2026 16:02:20 +0530
Subject: [PATCH 1/2] fix: resolve relative download redirect URLs

---
 databusclient/api/download.py    |  4 +--
 tests/test_download_redirects.py | 49 ++++++++++++++++++++++++++++++++
 2 files changed, 51 insertions(+), 2 deletions(-)
 create mode 100644 tests/test_download_redirects.py

diff --git a/databusclient/api/download.py b/databusclient/api/download.py
index 356a2dd..5b8abfc 100644
--- a/databusclient/api/download.py
+++ b/databusclient/api/download.py
@@ -7,7 +7,7 @@
 import re
 import shutil
 import tempfile
-from urllib.parse import urlparse
+from urllib.parse import urljoin, urlparse
 
 import requests
 from SPARQLWrapper import JSON, SPARQLWrapper
@@ -467,7 +467,7 @@ def _download_file(
         307,
         308,
     ]:
-        url = response.headers.get("Location")
+        url = urljoin(url, response.headers["Location"])
         print("Redirects url: ", url)
         # Re-do HEAD request on redirect URL
         response = requests.head(url, timeout=30)
diff --git a/tests/test_download_redirects.py b/tests/test_download_redirects.py
new file mode 100644
index 0000000..07f91f8
--- /dev/null
+++ b/tests/test_download_redirects.py
@@ -0,0 +1,49 @@
+"""Exercise download redirects against a local HTTP server."""
+
+from http.server import BaseHTTPRequestHandler, ThreadingHTTPServer
+from threading import Thread
+
+import pytest
+
+from databusclient.api.download import _download_file
+
+
+@pytest.mark.parametrize(
+    "location",
+    ["/data.ttl", "../data.ttl", "data.ttl", "{base}/data.ttl", "//{host}/data.ttl"],
+)
+def test_download_resolves_redirect_location(tmp_path, location):
+    content = b"<urn:s> <urn:p> <urn:o> .\n"
+
+    class Handler(BaseHTTPRequestHandler):
+        def do_HEAD(self):
+            if self.path == "/entry/file.ttl":
+                self.send_response(302)
+                self.send_header("Location", redirect_location)
+            else:
+                self.send_response(200)
+            self.end_headers()
+
+        def do_GET(self):
+            assert self.path in ("/data.ttl", "/entry/data.ttl")
+            self.send_response(200)
+            self.send_header("Content-Length", str(len(content)))
+            self.end_headers()
+            self.wfile.write(content)
+
+        def log_message(self, format, *args):
+            pass
+
+    with ThreadingHTTPServer(("127.0.0.1", 0), Handler) as server:
+        host = f"127.0.0.1:{server.server_port}"
+        base = f"http://{host}"
+        redirect_location = location.format(base=base, host=host)
+        thread = Thread(target=server.serve_forever, kwargs={"poll_interval": 0.01})
+        thread.start()
+        try:
+            _download_file(f"{base}/entry/file.ttl", localDir=str(tmp_path))
+        finally:
+            server.shutdown()
+            thread.join()
+
+    assert (tmp_path / "file.ttl").read_bytes() == content

From 1715222c27ef313069021b4a5d6a9efdd9563a51 Mon Sep 17 00:00:00 2001
From: Shubham Padkonde <shubhampadkonde12@gmail.com>
Date: Mon, 26 Oct 2026 17:14:05 +0530
Subject: [PATCH 2/2] test: assert the exact path for each download redirect

---
 tests/test_download_redirects.py | 16 ++++++++++++----
 1 file changed, 12 insertions(+), 4 deletions(-)

diff --git a/tests/test_download_redirects.py b/tests/test_download_redirects.py
index 07f91f8..8a5b456 100644
--- a/tests/test_download_redirects.py
+++ b/tests/test_download_redirects.py
@@ -9,11 +9,18 @@
 
 
 @pytest.mark.parametrize(
-    "location",
-    ["/data.ttl", "../data.ttl", "data.ttl", "{base}/data.ttl", "//{host}/data.ttl"],
+    ("location", "expected_path"),
+    [
+        ("/data.ttl", "/data.ttl"),
+        ("../data.ttl", "/data.ttl"),
+        ("data.ttl", "/entry/data.ttl"),
+        ("{base}/data.ttl", "/data.ttl"),
+        ("//{host}/data.ttl", "/data.ttl"),
+    ],
 )
-def test_download_resolves_redirect_location(tmp_path, location):
+def test_download_resolves_redirect_location(tmp_path, location, expected_path):
     content = b"<urn:s> <urn:p> <urn:o> .\n"
+    requested_paths = []
 
     class Handler(BaseHTTPRequestHandler):
         def do_HEAD(self):
@@ -25,7 +32,7 @@ def do_HEAD(self):
             self.end_headers()
 
         def do_GET(self):
-            assert self.path in ("/data.ttl", "/entry/data.ttl")
+            requested_paths.append(self.path)
             self.send_response(200)
             self.send_header("Content-Length", str(len(content)))
             self.end_headers()
@@ -46,4 +53,5 @@ def log_message(self, format, *args):
             server.shutdown()
             thread.join()
 
+    assert requested_paths == [expected_path]
     assert (tmp_path / "file.ttl").read_bytes() == content