From edecf8f6a375b04e60e7c4ddcc10b40f350a5d77 Mon Sep 17 00:00:00 2001
From: Idan Horowitz <idan.horowitz@gmail.com>
Date: Sun, 21 Mar 2021 22:31:15 +0200
Subject: [PATCH] AK: Add starts_with to Utf8View

Unlike String/StringView::starts_with this compares utf8 code points
instead of "characters" (bytes), which is important when handling
aribtary utf-8 input that could include overlong characters.
---
 AK/Utf8View.cpp | 18 ++++++++++++++++++
 AK/Utf8View.h   |  2 ++
 2 files changed, 20 insertions(+)

diff --git a/AK/Utf8View.cpp b/AK/Utf8View.cpp
index d28be5e425..a6fba541f2 100644
--- a/AK/Utf8View.cpp
+++ b/AK/Utf8View.cpp
@@ -144,6 +144,24 @@ size_t Utf8View::calculate_length() const
     return length;
 }
 
+bool Utf8View::starts_with(const Utf8View& start) const
+{
+    if (start.is_empty())
+        return true;
+    if (is_empty())
+        return false;
+    if (start.length() > length())
+        return false;
+    if (begin_ptr() == start.begin_ptr())
+        return true;
+
+    for (auto k = begin(), l = start.begin(); l != start.end(); ++k, ++l) {
+        if (*k != *l)
+            return false;
+    }
+    return true;
+}
+
 Utf8CodepointIterator::Utf8CodepointIterator(const unsigned char* ptr, size_t length)
     : m_ptr(ptr)
     , m_length(length)
diff --git a/AK/Utf8View.h b/AK/Utf8View.h
index af21c6792c..fe3b84029c 100644
--- a/AK/Utf8View.h
+++ b/AK/Utf8View.h
@@ -80,6 +80,8 @@ public:
     Utf8View substring_view(int byte_offset, int byte_length) const;
     bool is_empty() const { return m_string.is_empty(); }
 
+    bool starts_with(const Utf8View&) const;
+
     size_t iterator_offset(const Utf8CodepointIterator& it) const
     {
         return byte_offset_of(it);