AK: Add StringBuilder::append(Utf32View)

This encodes the incoming UTF-32 sequence as UTF-8.

2 files changed, 28 insertions, 0 deletions

diff --git a/AK/StringBuilder.cpp b/AK/StringBuilder.cpp
index ffae4e76da..f69077c6ee 100644
--- a/AK/StringBuilder.cpp
+++ b/AK/StringBuilder.cpp
@@ -30,6 +30,7 @@
 #include <AK/String.h>
 #include <AK/StringBuilder.h>
 #include <AK/StringView.h>
+#include <AK/Utf32View.h>
 
 namespace AK {
 
@@ -112,4 +113,30 @@ void StringBuilder::clear()
     m_length = 0;
 }
 
+void StringBuilder::append(const Utf32View& utf32_view)
+{
+    for (size_t i = 0; i < utf32_view.length(); ++i) {
+        auto codepoint = utf32_view.codepoints()[i];
+        if (codepoint <= 0x7f) {
+            append((char)codepoint);
+        } else if (codepoint <= 0x07ff) {
+            append((char)(((codepoint >> 6) & 0x1f) | 0xc0));
+            append((char)(((codepoint >> 0) & 0x3f) | 0x80));
+        } else if (codepoint <= 0xffff) {
+            append((char)(((codepoint >> 12) & 0x0f) | 0xe0));
+            append((char)(((codepoint >> 6) & 0x3f) | 0x80));
+            append((char)(((codepoint >> 0) & 0x3f) | 0x80));
+        } else if (codepoint <= 0x10ffff) {
+            append((char)(((codepoint >> 18) & 0x07) | 0xf0));
+            append((char)(((codepoint >> 12) & 0x3f) | 0x80));
+            append((char)(((codepoint >> 6) & 0x3f) | 0x80));
+            append((char)(((codepoint >> 0) & 0x3f) | 0x80));
+        } else {
+            append(0xef);
+            append(0xbf);
+            append(0xbd);
+        }
+    }
+}
+
 }
diff --git a/AK/StringBuilder.h b/AK/StringBuilder.h
index fc627b230b..53665b86ab 100644
--- a/AK/StringBuilder.h
+++ b/AK/StringBuilder.h
@@ -40,6 +40,7 @@ public:
     ~StringBuilder() {}
 
     void append(const StringView&);
+    void append(const Utf32View&);
     void append(char);
     void append(const char*, size_t);
     void appendf(const char*, ...);