From 48a3b2c28e0a4183e6420171ab2a47bbef13a366 Mon Sep 17 00:00:00 2001
From: Timothy Flynn <trflynn89@pm.me>
Date: Sat, 26 Jul 2025 09:59:47 -0400
Subject: [PATCH] AK: Implement a method to count instances of a needle in a
 UTF-16 string

---
 AK/Utf16StringBase.h       |  2 ++
 AK/Utf16View.h             | 14 ++++++++++++++
 Tests/AK/TestUtf16View.cpp | 24 ++++++++++++++++++++++++
 3 files changed, 40 insertions(+)

diff --git a/AK/Utf16StringBase.h b/AK/Utf16StringBase.h
index 120e056e48e..3242d725e8d 100644
--- a/AK/Utf16StringBase.h
+++ b/AK/Utf16StringBase.h
@@ -220,6 +220,8 @@ public:
     [[nodiscard]] ALWAYS_INLINE bool contains(char16_t needle) const { return find_code_unit_offset(needle).has_value(); }
     [[nodiscard]] ALWAYS_INLINE bool contains(Utf16View const& needle) const { return find_code_unit_offset(needle).has_value(); }
 
+    [[nodiscard]] ALWAYS_INLINE size_t count(Utf16View const& needle) const { return utf16_view().count(needle); }
+
     [[nodiscard]] ALWAYS_INLINE bool starts_with(Utf16View const& needle) const { return utf16_view().starts_with(needle); }
     [[nodiscard]] ALWAYS_INLINE bool ends_with(Utf16View const& needle) const { return utf16_view().ends_with(needle); }
 
diff --git a/AK/Utf16View.h b/AK/Utf16View.h
index fb8bc1ce643..df33da86d12 100644
--- a/AK/Utf16View.h
+++ b/AK/Utf16View.h
@@ -479,6 +479,20 @@ public:
     [[nodiscard]] constexpr bool contains(char16_t needle) const { return find_code_unit_offset(needle).has_value(); }
     [[nodiscard]] constexpr bool contains(Utf16View const& needle) const { return find_code_unit_offset(needle).has_value(); }
 
+    [[nodiscard]] constexpr size_t count(Utf16View const& needle) const
+    {
+        if (needle.is_empty())
+            return length_in_code_units();
+
+        size_t count = 0;
+        for (size_t i = 0; i < length_in_code_units() - needle.length_in_code_units() + 1; ++i) {
+            if (substring_view(i).starts_with(needle))
+                ++count;
+        }
+
+        return count;
+    }
+
     [[nodiscard]] constexpr bool starts_with(Utf16View const& needle) const
     {
         auto needle_length = needle.length_in_code_units();
diff --git a/Tests/AK/TestUtf16View.cpp b/Tests/AK/TestUtf16View.cpp
index 1313ad99f48..8c17e4f5961 100644
--- a/Tests/AK/TestUtf16View.cpp
+++ b/Tests/AK/TestUtf16View.cpp
@@ -558,6 +558,30 @@ TEST_CASE(contains)
     EXPECT(u"ab😀"sv.contains(u"😀"sv));
 }
 
+TEST_CASE(count)
+{
+    EXPECT_EQ(u""sv.count({}), 0uz);
+    EXPECT_EQ(u"abc"sv.count({}), 3uz);
+
+    EXPECT_EQ(u""sv.count(u"a"sv), 0uz);
+    EXPECT_EQ(u"abc"sv.count(u"a"sv), 1uz);
+    EXPECT_EQ(u"abc"sv.count(u"b"sv), 1uz);
+    EXPECT_EQ(u"abc"sv.count(u"c"sv), 1uz);
+    EXPECT_EQ(u"abc"sv.count(u"ab"sv), 1uz);
+    EXPECT_EQ(u"abc"sv.count(u"bc"sv), 1uz);
+    EXPECT_EQ(u"abc"sv.count(u"abc"sv), 1uz);
+    EXPECT_EQ(u"abc"sv.count(u"d"sv), 0uz);
+
+    EXPECT_EQ(u"aaaa"sv.count(u"aa"sv), 3uz);
+
+    EXPECT_EQ(u"😀"sv.count({}), 2uz);
+    EXPECT_EQ(u"😀"sv.count(u"\xd83d"sv), 1uz);
+    EXPECT_EQ(u"😀"sv.count(u"\xde00"sv), 1uz);
+    EXPECT_EQ(u"😀"sv.count(u"😀"sv), 1uz);
+    EXPECT_EQ(u"😀😀😀"sv.count(u"😀"sv), 3uz);
+    EXPECT_EQ(u"😀😀😀"sv.count(u"😀😀"sv), 2uz);
+}
+
 TEST_CASE(starts_with)
 {
     EXPECT(Utf16View {}.starts_with(u""sv));