From 180b5b866ccc28e08eb61720e1ce55573d095b42 Mon Sep 17 00:00:00 2001
From: "copilot-swe-agent[bot]" <198982749+Copilot@users.noreply.github.com>
Date: Wed, 19 Nov 2025 01:30:57 +0000
Subject: [PATCH 1/3] Initial plan


From 91a35c764abb1e59170c8cd20843431852cc0dcf Mon Sep 17 00:00:00 2001
From: "copilot-swe-agent[bot]" <198982749+Copilot@users.noreply.github.com>
Date: Wed, 19 Nov 2025 01:36:01 +0000
Subject: [PATCH 2/3] Initial progress: Understanding the bug

Co-authored-by: keighrim <9062727+keighrim@users.noreply.github.com>
---
 mmif/ver.py | 2 ++
 1 file changed, 2 insertions(+)
 create mode 100644 mmif/ver.py

diff --git a/mmif/ver.py b/mmif/ver.py
new file mode 100644
index 00000000..bc348c0a
--- /dev/null
+++ b/mmif/ver.py
@@ -0,0 +1,2 @@
+__version__ = "1.0.0"
+__specver__ = "1.1.0"

From dfeb684ee7e1a4f49a66c3487d759e2d36c23c42 Mon Sep 17 00:00:00 2001
From: "copilot-swe-agent[bot]" <198982749+Copilot@users.noreply.github.com>
Date: Wed, 19 Nov 2025 01:38:52 +0000
Subject: [PATCH 3/3] Fix character indexing bug in source.py and add test

Co-authored-by: keighrim <9062727+keighrim@users.noreply.github.com>
---
 mmif/utils/cli/source.py |  8 +++++++-
 tests/test_utils_cli.py  | 12 ++++++++++++
 2 files changed, 19 insertions(+), 1 deletion(-)

diff --git a/mmif/utils/cli/source.py b/mmif/utils/cli/source.py
index 7e662f9f..154c315f 100644
--- a/mmif/utils/cli/source.py
+++ b/mmif/utils/cli/source.py
@@ -206,7 +206,13 @@ def generate_source_mmif_from_file(documents, prefix=None, scheme='file', **igno
         location = str(location)
         doc = Document()
         doc.at_type = at_types[mime.split('/', maxsplit=1)[0]]
-        doc.properties.location = f"{location_uri.scheme}://{location if not location.startswith(location_uri.scheme) else location[len(location_uri.scheme)+3:]}"
+        # Use the scheme parameter if urlparse misinterpreted the location string
+        if location_uri.scheme == scheme or location_uri.scheme == 'file':
+            # Correct scheme was used, build location with parsed scheme
+            doc.properties.location = f"{location_uri.scheme}://{location if not location.startswith(location_uri.scheme + '://') else location[len(location_uri.scheme)+3:]}"
+        else:
+            # urlparse incorrectly interpreted part of location as a scheme, use the scheme parameter instead
+            doc.properties.location = f"{scheme}://{location}"
         doc.properties.id = f'd{doc_id}'
         doc.properties.mime = mime
         pl.add_document(doc)
diff --git a/tests/test_utils_cli.py b/tests/test_utils_cli.py
index adb8e857..fde14f66 100644
--- a/tests/test_utils_cli.py
+++ b/tests/test_utils_cli.py
@@ -114,6 +114,18 @@ def test_generate_mixed_scheme(self):
         self.assertTrue('baapb' in schemes)
         self.assertTrue('file' in schemes)
 
+    def test_scheme_with_colon_in_location(self):
+        # Test for bug fix: location with colon should not be misinterpreted as a scheme
+        self.scheme = 'baapb'
+        self.docs.append("video:cpb-aacip-507-v40js9j432:video")
+        source_mmif = Mmif(self.generate_source_mmif())
+        self.assertEqual(len(source_mmif.documents), 1)
+        doc = source_mmif.documents[0]
+        # Verify the scheme is correct
+        self.assertEqual(doc.location_scheme(), 'baapb')
+        # Verify the full location preserves the original identifier with colon
+        self.assertEqual(doc.location, 'baapb://cpb-aacip-507-v40js9j432:video')
+
 
 class TestRewind(unittest.TestCase):
     def setUp(self):