1 function string.utf8CharSize(c)
 2     if not c then return 0 end
 3     if c > 240 then return 4 end
 4     if c > 225 then return 3 end
 5     if c > 192 then return 2 end
 6     return 1
 7 end
 8 
 9 function string.utf8StringLen(str)
10     if nil == str or "" == str then return 0 end
11     assert("string" == type(str), "not string")
12     
13     local len = 0
14     local i = 1
15     while i <= #str do
16         local b = string.byte(str, i)
17         i = i + string.utf8CharSize(b)
18         len = len + 1
19     end
20     return len
21 end
22 
23 function string.utf8StringSub(str, index, count)
24     if nil == index then index = 1 end
25     if nil == count then count = string.utf8StringLen(str) end
26     
27     local byteCount = #str
28     local i = 1
29     --跳过起始的n个
30     while index > 1 and i <= byteCount do
31         local b = string.byte(str, i)
32         i = i + string.utf8CharSize(b)
33         index = index - 1
34     end
35     
36     local j = i
37     --子串结束位置
38     while count > 0 and j <= byteCount do
39         local b = string.byte(str, j)
40         j = j + string.utf8CharSize(b)
41         count = count - 1
42     end
43     
44     return str:sub(i, j-1)
45 end
46 
47 function string.utf8CharArray(str, index, count, dstTb, dstIndex)
48     if nil == index then index = 1 end
49     if nil == count then count = string.utf8StringLen(str) end
50     if nil == dstTb then dstTb = {} end
51     
52     local byteCount = #str
53     local i = 1
54     --跳过起始的n个
55     while index > 1 and i <= byteCount do
56         local b = string.byte(str, i)
57         i = i + string.utf8CharSize(b)
58         index = index - 1
59     end
60     
61     local j = i
62     local leftCount = count
63         --子串结束位置
64         if nil == dstIndex then
65             while leftCount > 0 and j <= byteCount do
66             local b = string.byte(str, j)
67             local size = string.utf8CharSize(b)
68             local utf8Char = str:sub(j, j+size-1)
69             table.insert(dstTb, utf8Char)
70             j = j + size
71             leftCount = leftCount - 1
72         end
73     else
74         while leftCount > 0 and j <= byteCount do
75             local b = string.byte(str, j)
76             local size = string.utf8CharSize(b)
77             local utf8Char = str:sub(j, j+size-1)
78             dstTb[dstIndex] = utf8Char
79             j = j + size
80             leftCount = leftCount - 1
81             dstIndex = dstIndex + 1
82         end
83     end
84     
85     return dstTb, count
86 end

 

测试代码：

 1 local function Test1()
 2     assert(6 == string.utf8StringLen("123abc"))
 3     assert(2 == string.utf8StringLen("中文"))
 4     assert(4 == string.utf8StringLen("中文1a"))
 5     assert(5 == string.utf8StringLen(".,;<>"))
 6     assert(5 == string.utf8StringLen("。，；《》"))
 7     assert(10 == string.utf8StringLen(".,;<>。，；《》"))
 8     
 9     assert("文1" == string.utf8StringSub("中文1a", 2, 2))
10     assert("文啊" == string.utf8StringSub("中文啊呀", 2, 2))
11     
12     local tb = string.utf8CharArray("中文啊呀", 2, 2)
13     assert("文" == tb[1])
14     assert("啊" == tb[2])
15 end
16 
17 Test1()