From 3f10a5701d9634e47111203b837283bdfc2d8b18 Mon Sep 17 00:00:00 2001
From: Sam Atkins <sam@ladybird.org>
Date: Fri, 15 Nov 2024 15:43:59 +0000
Subject: [PATCH] AK: Add Utf8View::for_each_split_view() method

Returns one Utf8View at a time, using a callback function to identify
code points to split on.
---
 AK/Utf8View.h         | 47 +++++++++++++++++++++++++++++++++++++++++++
 Tests/AK/TestUtf8.cpp | 23 +++++++++++++++++++++
 2 files changed, 70 insertions(+)

diff --git a/AK/Utf8View.h b/AK/Utf8View.h
index 79a26594d8f..9158bf669c1 100644
--- a/AK/Utf8View.h
+++ b/AK/Utf8View.h
@@ -10,6 +10,7 @@
 #include <AK/ByteString.h>
 #include <AK/Debug.h>
 #include <AK/Format.h>
+#include <AK/Function.h>
 #include <AK/StringView.h>
 #include <AK/Types.h>
 
@@ -139,6 +140,52 @@ public:
 
     bool validate(size_t& valid_bytes, AllowSurrogates allow_surrogates = AllowSurrogates::Yes) const;
 
+    template<typename Callback>
+    auto for_each_split_view(Function<bool(u32)> splitter, SplitBehavior split_behavior, Callback callback) const
+    {
+        bool keep_empty = has_flag(split_behavior, SplitBehavior::KeepEmpty);
+        bool keep_trailing_separator = has_flag(split_behavior, SplitBehavior::KeepTrailingSeparator);
+
+        auto start_offset = 0u;
+        auto offset = 0u;
+
+        auto run_callback = [&]() {
+            auto length = offset - start_offset;
+
+            if (length == 0 && !keep_empty)
+                return;
+
+            auto substring = unicode_substring_view(start_offset, length);
+
+            // Reject splitter-only entries if we're not keeping empty results
+            if (keep_trailing_separator && !keep_empty && length == 1 && splitter(*substring.begin()))
+                return;
+
+            callback(substring);
+        };
+
+        auto iterator = begin();
+        while (iterator != end()) {
+            if (splitter(*iterator)) {
+                if (keep_trailing_separator)
+                    ++offset;
+
+                run_callback();
+
+                if (!keep_trailing_separator)
+                    ++offset;
+
+                start_offset = offset;
+                ++iterator;
+                continue;
+            }
+
+            ++offset;
+            ++iterator;
+        }
+        run_callback();
+    }
+
 private:
     friend class Utf8CodePointIterator;
 
diff --git a/Tests/AK/TestUtf8.cpp b/Tests/AK/TestUtf8.cpp
index 988707c41d9..c007cdd7e6c 100644
--- a/Tests/AK/TestUtf8.cpp
+++ b/Tests/AK/TestUtf8.cpp
@@ -312,3 +312,26 @@ TEST_CASE(trim)
         EXPECT_EQ(view.trim(whitespace, TrimMode::Right).as_string(), "\u180E");
     }
 }
+
+static bool is_period(u32 code_point) { return code_point == '.'; }
+
+TEST_CASE(for_each_split_view)
+{
+    Utf8View view { "...Well..hello.friends!..."sv };
+    auto gather = [&](auto split_behavior) {
+        Vector<StringView> results;
+        view.for_each_split_view(is_period, split_behavior, [&](auto part) {
+            results.append(part.as_string());
+        });
+        return results;
+    };
+
+    EXPECT_EQ(gather(SplitBehavior::Nothing),
+        Vector({ "Well"sv, "hello"sv, "friends!"sv }));
+    EXPECT_EQ(gather(SplitBehavior::KeepEmpty),
+        Vector({ ""sv, ""sv, ""sv, "Well"sv, ""sv, "hello"sv, "friends!"sv, ""sv, ""sv, ""sv }));
+    EXPECT_EQ(gather(SplitBehavior::KeepTrailingSeparator),
+        Vector({ "Well."sv, "hello."sv, "friends!."sv }));
+    EXPECT_EQ(gather(SplitBehavior::KeepEmpty | SplitBehavior::KeepTrailingSeparator),
+        Vector({ "."sv, "."sv, "."sv, "Well."sv, "."sv, "hello."sv, "friends!."sv, "."sv, "."sv, ""sv }));
+}