From f02c75616ae0fc2ac0b7382f1b431a888e49c65f Mon Sep 17 00:00:00 2001
From: ccl-core <stanzelc@google.com>
Date: Fri, 24 Jan 2025 14:20:51 +0000
Subject: [PATCH] Handle dash in the regex2glob function

---
 python/mlcroissant/mlcroissant/_src/core/regex.py      | 2 ++
 python/mlcroissant/mlcroissant/_src/core/regex_test.py | 6 ++++++
 2 files changed, 8 insertions(+)

diff --git a/python/mlcroissant/mlcroissant/_src/core/regex.py b/python/mlcroissant/mlcroissant/_src/core/regex.py
index 468bfb6e..c7ffb3bb 100644
--- a/python/mlcroissant/mlcroissant/_src/core/regex.py
+++ b/python/mlcroissant/mlcroissant/_src/core/regex.py
@@ -65,6 +65,8 @@ def _regex_to_glob_for_str(regex: str) -> Iterable[str]:
     regex = re.sub(r"\.\*", "*", regex)
     # Interpret .+ as *
     regex = re.sub(r"\.\+", "*", regex)
+    # Interpret \\- as -
+    regex = re.sub(r"\\-", "-", regex)
     return [regex]
 
 
diff --git a/python/mlcroissant/mlcroissant/_src/core/regex_test.py b/python/mlcroissant/mlcroissant/_src/core/regex_test.py
index 3a153f01..c9a46bee 100644
--- a/python/mlcroissant/mlcroissant/_src/core/regex_test.py
+++ b/python/mlcroissant/mlcroissant/_src/core/regex_test.py
@@ -27,6 +27,12 @@
                 "*/train/*.parquet",  # ...to a valid glob pattern.
             ],
         ],
+        [
+            "^.+/my\\-train/.*\.parquet$",  # From a valid regex...
+            [
+                "*/my-train/*.parquet",  # ...to a valid glob pattern.
+            ],
+        ],
     ],
 )
 def test_regex_to_glob(regex: str, output: list[str]):