From acb58d4923e38cd3726b08c30890482dbd8d3ef4 Mon Sep 17 00:00:00 2001
From: ChAoS_UnItY <43753315+ChAoSUnItY@users.noreply.github.com>
Date: Sun, 11 Apr 2021 14:04:18 +0800
Subject: [PATCH] encoding.utf8: add support for indexing a utf8 str (#9670)

---
 vlib/encoding/utf8/utf8_util.v      | 21 +++++++++++++++++++++
 vlib/encoding/utf8/utf8_util_test.v | 17 +++++++++++++++++
 2 files changed, 38 insertions(+)

diff --git a/vlib/encoding/utf8/utf8_util.v b/vlib/encoding/utf8/utf8_util.v
index fe9e99a7a2..ca1d517a74 100644
--- a/vlib/encoding/utf8/utf8_util.v
+++ b/vlib/encoding/utf8/utf8_util.v
@@ -87,6 +87,27 @@ pub fn get_uchar(s string, index int) int {
 }
 
 
+// raw_index - get the raw chracter from the string by the given index value.
+// example: '我是V Lang'.raw_index(1) => '是'
+
+pub fn raw_index(s string, index int) string {
+	mut r := []rune{}
+
+	for i := 0; i < s.len; i++ {
+        b := s[i]
+        ch_len := ((0xe5000000>>((b>>3) & 0x1e)) & 3)
+        
+		r << if ch_len > 0 {
+            i += ch_len
+            rune(get_uchar(s,i-ch_len))
+        } else {
+            rune(b)
+        }
+    }
+
+	return r[index].str()
+}
+
 /*
 
 Conversion functions
diff --git a/vlib/encoding/utf8/utf8_util_test.v b/vlib/encoding/utf8/utf8_util_test.v
index 646bd69a7f..e26846c261 100644
--- a/vlib/encoding/utf8/utf8_util_test.v
+++ b/vlib/encoding/utf8/utf8_util_test.v
@@ -50,3 +50,20 @@ fn test_utf8_util() {
   	// test utility functions
   	assert utf8.get_uchar(b,0)==0x002E
 }
+
+fn test_raw_indexing() {
+	a := "我是V Lang!"
+
+	// test non ascii characters
+	assert utf8.raw_index(a, 0) == '我'
+	assert utf8.raw_index(a, 1) == '是'
+	
+	// test ascii characters
+	assert utf8.raw_index(a, 2) == 'V'
+	assert utf8.raw_index(a, 3) == ' '
+	assert utf8.raw_index(a, 4) == 'L'
+	assert utf8.raw_index(a, 5) == 'a'
+	assert utf8.raw_index(a, 6) == 'n'
+	assert utf8.raw_index(a, 7) == 'g'
+	assert utf8.raw_index(a, 8) == '!'
+}