From fb50596d22b101031b7ff6d84c818310c8862456 Mon Sep 17 00:00:00 2001
From: Luke Swanson <lwswanson2@wisc.edu>
Date: Fri, 9 Jan 2026 13:17:36 -0600
Subject: [PATCH] Add custom urljoin for more-nicely working with paths to WSGI
 apps.

---
 src/ogd/apis/utils/APIUtils.py | 41 ++++++++++++++++++++++++++++++++++
 1 file changed, 41 insertions(+)

diff --git a/src/ogd/apis/utils/APIUtils.py b/src/ogd/apis/utils/APIUtils.py
index 1177f2b..2e716aa 100644
--- a/src/ogd/apis/utils/APIUtils.py
+++ b/src/ogd/apis/utils/APIUtils.py
@@ -11,6 +11,7 @@
 from json.decoder import JSONDecodeError
 from logging import Logger
 from typing import Any, List, Optional
+from urllib import parse
 
 # import OGD libraries
 from ogd.common.storage.interfaces.Interface import Interface
@@ -42,6 +43,46 @@ def parse_list(list_str:str, logger:Optional[Logger]=None) -> Optional[List[Any]
             ret_val = None
     return ret_val
 
+def urljoin(base:str, url:str, ignore_base_file:bool=False, allow_fragments:bool=True):
+    """Custom variation of the `urllib.parse.urljoin` function provided by Python.
+
+    By default, this version allows filenames in the base path to remain in the joined path.
+
+    This is useful for working with Flask apps, particularly on Apache.
+    Specifically, unless you alias things in Apache, you'll have `app.py` or `app.wsgi` in the URL.
+    For a Flask API, then, you'll likely be joining a base URL like `"https://host.of.app/path/to/app.wsgi"` with an endpoint, call it `"endpoint"`.
+    Under `urllib.parse.urljoin`, you'll get `"https://host.of.app/path/to/endpoint"`.
+    With _this_ function, setting `ignore_base_file=False`, you'll get `"http://host.of.app/path/to/app.wsgi/endpoint"` as desired.
+
+    When `ignore_base_file=True`, this function directly falls back to use `urllib.parse.urljoin`.
+
+    :param base: The base URL, onto which the `url` parameter is joined.
+    :type base: str
+    :param url: The URL to be joined onto the given base URL.
+    :type url: str
+    :param ignore_base_file: Whether to ignore filenames in the base URL.  
+        When True, such filenames are removed from the joined URL.
+        For example, joining `https://host.of.app/path/to/app.wsgi` with `endpoint` would yield `https://host.of.app/path/to/endpoint`.  
+        When False, such filenames are included in the joined URL.
+    :type ignore_base_file: bool
+    :param allow_fragments: Whether to allow fragment parts in the URLs.  
+        This is only used when `ignore_base_file=True`, in which case this function falls back on `urllib.parse.urljoin` and `allow_fragments` is passed to that function call.
+    :type ignore_base_file: bool
+    """
+    if ignore_base_file:
+        return parse.urljoin(base=base, url=url, allow_fragments=allow_fragments)
+    else:
+        # Make sure we have a scheme
+        if not (base.startswith("http://") or base.startswith("https://")):
+            base = f"https://{base}"
+        # If base ends with a /, remove it so we don't double-up when joining
+        if base.endswith("/"):
+            base = base[:-1]
+        # If url starts with a /, remove it so we don't double-up when joining
+        if url.startswith("/"):
+            url = url[1:]
+        return f"{base}/{url}"
+
 # def gen_interface(game_id, core_config:ConfigSchema, logger:Optional[Logger]=None) -> Optional[Interface]:
 #     """Utility to set up an Interface object for use by the API, given a game_id.