From 3d40dd6c4b06f0245c34beefa4d0bbd3e3e201ad Mon Sep 17 00:00:00 2001
From: Carmine Sacco <fortekarma@gmail.com>
Date: Mon, 13 Oct 2025 22:21:46 +0200
Subject: [PATCH] fix: bug 4105 clean_bullets didn't remove en-dashes used as
 bullets

---
 CHANGELOG.md                            | 1 +
 test_unstructured/cleaners/test_core.py | 1 +
 unstructured/nlp/patterns.py            | 1 +
 3 files changed, 3 insertions(+)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index ed763c8ba6..91c261b322 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -6,6 +6,7 @@
 ### Features
 
 ### Fixes
+- Improved clean_bullets in order to remove en-dashes if they are used as bullets
 
 ## 0.18.15
 
diff --git a/test_unstructured/cleaners/test_core.py b/test_unstructured/cleaners/test_core.py
index eec8edd2b9..f4fb66ca8b 100644
--- a/test_unstructured/cleaners/test_core.py
+++ b/test_unstructured/cleaners/test_core.py
@@ -30,6 +30,7 @@ def test_clean_non_ascii_chars(text, expected):
         ("● An excellent point! ●●●", "An excellent point! ●●●"),
         ("An excellent point!", "An excellent point!"),
         ("Morse code! ●●●", "Morse code! ●●●"),
+        ("– An excellent point!", "An excellent point!")
     ],
 )
 def test_clean_bullets(text, expected):
diff --git a/unstructured/nlp/patterns.py b/unstructured/nlp/patterns.py
index fe73c6c0dc..f7f2084dbc 100644
--- a/unstructured/nlp/patterns.py
+++ b/unstructured/nlp/patterns.py
@@ -56,6 +56,7 @@
     r"\*",
     "\x95",
     "·",
+    "\u2013",
 ]
 BULLETS_PATTERN = "|".join(UNICODE_BULLETS)
 UNICODE_BULLETS_RE = re.compile(f"(?:{BULLETS_PATTERN})(?!{BULLETS_PATTERN})")