From 745e61438d3742e9a11fed3f8987e9a22f67bded Mon Sep 17 00:00:00 2001
From: Doug Halley <douglascomet@gmail.com>
Date: Fri, 13 Oct 2023 09:48:53 -0400
Subject: [PATCH] Adds support to `filepath_from_url` to support non-encoded
 URLs with Windows drive letters

* Add support to `filepath_from_url` to support non-encoded URLs with Windows drive letters
* Added additional unit tests

Signed-off-by: Doug Halley <douglascomet@gmail.com>
---
 .../opentimelineio/url_utils.py               | 36 ++++++++++++++++---
 tests/test_url_conversions.py                 |  9 +++++
 2 files changed, 41 insertions(+), 4 deletions(-)

diff --git a/src/py-opentimelineio/opentimelineio/url_utils.py b/src/py-opentimelineio/opentimelineio/url_utils.py
index 9e1c02520..61e44be58 100644
--- a/src/py-opentimelineio/opentimelineio/url_utils.py
+++ b/src/py-opentimelineio/opentimelineio/url_utils.py
@@ -9,7 +9,10 @@
     parse as urlparse,
     request
 )
-import pathlib
+from pathlib import (
+    Path,
+    PureWindowsPath
+)
 
 
 def url_from_filepath(fpath):
@@ -18,7 +21,7 @@ def url_from_filepath(fpath):
     try:
         # appears to handle absolute windows paths better, which are absolute
         # and start with a drive letter.
-        return urlparse.unquote(pathlib.Path(fpath).as_uri())
+        return urlparse.unquote(Path(fpath).as_uri())
     except ValueError:
         # scheme is "file" for absolute paths, else ""
         scheme = "file" if os.path.isabs(fpath) else ""
@@ -37,7 +40,32 @@ def url_from_filepath(fpath):
 
 
 def filepath_from_url(urlstr):
-    """ Take a url and return a filepath """
+    """
+    Take an url and return a filepath.
+
+    URLs can either be encoded according to the `RFC 3986`_ standard or not.
+    Additionally, Windows mapped paths need to be accounted for when processing a
+    URL; however, there are `ongoing discussions`_ about how to best handle this within
+    Python. This function is meant to cover all of these scenarios in the interim.
 
+    .. _RFC 3986: https://tools.ietf.org/html/rfc3986#section-2.1
+    .. _ongoing discussions: https://discuss.python.org/t/file-uris-in-python/15600
+    """
+
+    # Parse provided URL
     parsed_result = urlparse.urlparse(urlstr)
-    return request.url2pathname(parsed_result.path)
+
+    # Convert the parsed URL to a path
+    filepath = Path(request.url2pathname(parsed_result.path))
+
+    # If the network location is a window drive, reassemble the path
+    if PureWindowsPath(parsed_result.netloc).drive:
+        filepath = Path(parsed_result.netloc + parsed_result.path)
+
+    # Otherwise check if the specified index is a windows drive, then offset the path
+    elif PureWindowsPath(filepath.parts[1]).drive:
+        # Remove leading "/" if/when `request.url2pathname` yields "/S:/path/file.ext"
+        filepath = filepath.relative_to(filepath.root)
+
+    # Convert "\" to "/" if needed
+    return filepath.as_posix()
diff --git a/tests/test_url_conversions.py b/tests/test_url_conversions.py
index 5e994148b..6a2e8d63c 100644
--- a/tests/test_url_conversions.py
+++ b/tests/test_url_conversions.py
@@ -32,6 +32,10 @@
     MEDIA_EXAMPLE_PATH_ABS
 )
 
+ENCODED_WINDOWS_URL = "file://localhost/S%3a/path/file.ext"
+WINDOWS_URL = "file://S:/path/file.ext"
+CORRECTED_WINDOWS_PATH = "S:/path/file.ext"
+
 
 class TestConversions(unittest.TestCase):
     def test_roundtrip_abs(self):
@@ -51,6 +55,11 @@ def test_roundtrip_rel(self):
         # should have reconstructed it by this point
         self.assertEqual(os.path.normpath(result), MEDIA_EXAMPLE_PATH_REL)
 
+    def test_windows_urls(self):
+        for url in (ENCODED_WINDOWS_URL, WINDOWS_URL):
+            processed_url = otio.url_utils.filepath_from_url(url)
+            self.assertEqual(processed_url, CORRECTED_WINDOWS_PATH)
+
 
 if __name__ == "__main__":
     unittest.main()