From 6c73dff120ad9fd7a37d81f5900cd74e84ebe0e9 Mon Sep 17 00:00:00 2001
From: Timothy Flynn <trflynn89@pm.me>
Date: Thu, 24 Jul 2025 11:44:19 -0400
Subject: [PATCH] AK: Implement a UTF-16 method to check if a string is ASCII
 whitespace

---
 AK/Utf16StringBase.h       |  1 +
 AK/Utf16View.h             |  8 ++++++++
 Tests/AK/TestUtf16View.cpp | 16 ++++++++++++++++
 3 files changed, 25 insertions(+)

diff --git a/AK/Utf16StringBase.h b/AK/Utf16StringBase.h
index f245c99ecf3..120e056e48e 100644
--- a/AK/Utf16StringBase.h
+++ b/AK/Utf16StringBase.h
@@ -156,6 +156,7 @@ public:
 
     [[nodiscard]] ALWAYS_INLINE bool is_empty() const { return length_in_code_units() == 0uz; }
     [[nodiscard]] ALWAYS_INLINE bool is_ascii() const { return utf16_view().is_ascii(); }
+    [[nodiscard]] ALWAYS_INLINE bool is_ascii_whitespace() const { return utf16_view().is_ascii_whitespace(); }
 
     [[nodiscard]] ALWAYS_INLINE size_t length_in_code_units() const
     {
diff --git a/AK/Utf16View.h b/AK/Utf16View.h
index 22d9c0a0dea..566821b6128 100644
--- a/AK/Utf16View.h
+++ b/AK/Utf16View.h
@@ -268,8 +268,16 @@ public:
     }
 
     [[nodiscard]] constexpr bool is_empty() const { return length_in_code_units() == 0; }
+
     [[nodiscard]] bool is_ascii() const;
 
+    [[nodiscard]] constexpr bool is_ascii_whitespace() const
+    {
+        if (has_ascii_storage())
+            return all_of(ascii_span(), AK::is_ascii_space);
+        return all_of(utf16_span(), AK::is_ascii_space);
+    }
+
     [[nodiscard]] ALWAYS_INLINE bool validate(AllowLonelySurrogates allow_lonely_surrogates = AllowLonelySurrogates::Yes) const
     {
         size_t valid_code_units = 0;
diff --git a/Tests/AK/TestUtf16View.cpp b/Tests/AK/TestUtf16View.cpp
index 3a3bd759f9c..1313ad99f48 100644
--- a/Tests/AK/TestUtf16View.cpp
+++ b/Tests/AK/TestUtf16View.cpp
@@ -341,6 +341,22 @@ TEST_CASE(is_ascii)
     EXPECT(!u"The quick (“brown”) fox can’t jump 32.3 feet, right?"sv.is_ascii());
 }
 
+TEST_CASE(is_ascii_whitespace)
+{
+    EXPECT(Utf16View {}.is_ascii_whitespace());
+    EXPECT(u" "sv.is_ascii_whitespace());
+    EXPECT(u"\t"sv.is_ascii_whitespace());
+    EXPECT(u"\r"sv.is_ascii_whitespace());
+    EXPECT(u"\n"sv.is_ascii_whitespace());
+    EXPECT(u" \t\r\n\v "sv.is_ascii_whitespace());
+
+    EXPECT(!u"a"sv.is_ascii_whitespace());
+    EXPECT(!u"😀"sv.is_ascii_whitespace());
+    EXPECT(!u"\u00a0"sv.is_ascii_whitespace());
+    EXPECT(!u"\ufeff"sv.is_ascii_whitespace());
+    EXPECT(!u"  \t \u00a0 \ufeff  "sv.is_ascii_whitespace());
+}
+
 TEST_CASE(to_ascii_lowercase)
 {
     EXPECT_EQ(u""sv.to_ascii_lowercase(), u""sv);