@@ -46,13 +46,13 @@ class RobustnessTestCase(unittest.TestCase):
4646 "\ufffe " , # BOM reversed
4747 "\x00 " , # Null character
4848 "\u0000 " , # Null unicode
49- "\u200c \u200d " , # Zero-width non-joiner (ZWNJ) , zero-width joiner (ZWJ)
50- "\u200c \u200d " , # ZWNJ, ZWJ (duplicate for testing)
49+ " " , # Zero-width non-joiner, zero-width joiner
50+ "\u200c \u200d " , # ZWNJ, ZWJ
5151 "\u200e \u200f " , # LTR mark, RTL mark
52- "\u00ad " , # Soft hyphen
53- "\u00ad " , # Soft hyphen unicode (duplicate)
54- "\u2026 " , # Ellipsis (\u2026)
55- "\u2014 \u2013 -" , # Em dash, en dash, hyphen-minus
52+ " " , # Soft hyphen
53+ "\u00ad " , # Soft hyphen unicode
54+ "… " , # Ellipsis
55+ "—– -" , # Different dashes
5656 "\u201c \u201d " , # Smart double quotes (curly quotes)
5757 "\u2018 \u2019 " , # Smart single quotes (curly quotes)
5858 ]
@@ -69,13 +69,13 @@ class RobustnessTestCase(unittest.TestCase):
6969 # Category: Emoji and Modern Unicode
7070 # Emoji, emoji sequences, and variations
7171 EMOJI_CASES = [
72- "\U0001f600 " , # Basic emoji (grinning face)
73- "\U0001f468 \u200d \U0001f469 \u200d \U0001f467 \u200d \U0001f466 " , # Family emoji (ZWJ sequence)
74- "\U0001f44d \U0001f3fb " , # Emoji with skin tone modifier (thumbs up, light skin tone)
75- "\U0001f1f9 \U0001f1ed " , # Flag (TH - regional indicator symbols)
76- "\U0001f600 \U0001f603 \U0001f604 " , # Multiple emoji
77- "สวัสดี\U0001f600 ครับ " , # Thai text with emoji
78- "\U0001f3f4 \U000e0067 \U000e0062 \U000e0065 \U000e006e \U000e0067 \U000e007f " , # Flag with tag sequences (England)
72+ "😀 " , # Basic emoji
73+ "👨👩👧👦 " , # Family emoji (ZWJ sequence)
74+ "👍🏻 " , # Emoji with skin tone modifier
75+ "🇹🇭 " , # Flag (regional indicator symbols)
76+ "😀😃😄 " , # Multiple emoji
77+ "สวัสดี😀ครับ " , # Thai text with emoji
78+ "🏴 " , # Flag with tag sequences
7979 ]
8080
8181 # Category: Control and Hidden Characters
0 commit comments