From 2a43219d346ac77a9edbe7b7f9bdb19073671a17 Mon Sep 17 00:00:00 2001
From: Florian Spreckelsen <f.spreckelsen@indiscale.com>
Date: Mon, 26 Aug 2024 10:18:42 +0200
Subject: [PATCH] FIX: Respect context-root and make links platform independent
---
src/caoscrawler/utils.py | 9 +++++++--
unittests/test_utilities.py | 4 ++++
2 files changed, 11 insertions(+), 2 deletions(-)
diff --git a/src/caoscrawler/utils.py b/src/caoscrawler/utils.py
index 65a1bb7c..d9a5af83 100644
--- a/src/caoscrawler/utils.py
+++ b/src/caoscrawler/utils.py
@@ -25,9 +25,9 @@
# Some utility functions, e.g. for extending pylib.
-import os
import sys
+from posixpath import join as posixjoin
from typing import Optional
from urllib.parse import urljoin
@@ -81,4 +81,9 @@ def get_shared_resource_link(host_url, filename):
"""
- return urljoin(host_url, os.path.join("/Shared/", filename))
+ if not host_url.endswith('/'):
+ # Fill with trailing '/' s. that urljoin doesn't remove the context root.
+ host_url += '/'
+ # Use posixjoin to always have '/' in links, even when running on
+ # Windows systems.
+ return urljoin(host_url, posixjoin("Shared/", filename))
diff --git a/unittests/test_utilities.py b/unittests/test_utilities.py
index 89b0b40c..15e84a60 100644
--- a/unittests/test_utilities.py
+++ b/unittests/test_utilities.py
@@ -76,3 +76,7 @@ def test_shared_resource_link():
"https://example.com", "file.txt") == "https://example.com/Shared/file.txt"
assert get_shared_resource_link(
"https://example.com", "path/to/file.txt") == "https://example.com/Shared/path/to/file.txt"
+ assert get_shared_resource_link(
+ "https://example.com/context-root", "path/to/file.txt") == "https://example.com/context-root/Shared/path/to/file.txt"
+ assert get_shared_resource_link(
+ "https://example.com/context-root/", "path/to/file.txt") == "https://example.com/context-root/Shared/path/to/file.txt"
--
GitLab