扫描1段数据是否为UTF8编码

tlwh163

用查表的思路写了一个


''扫描1段数据是否为UTF8编码,参数CharLength用于返回字符数
Public Function IS_UTF8(Utf8() As Byte, Optional ByRef CharLength As Long = 0) As Boolean
    If (UBound(Utf8) - LBound(Utf8) + 1) = 0 Then
        CharLength = 0: IS_UTF8 = False: Exit Function
    End If
    ''0-15的前导1数量: 比如 15有4个前导1, 9有1个前导1, 7以下的数没有前导1
    Dim PrecBit1(0 To 15) As Byte
    PrecBit1(8) = 1: PrecBit1(9) = 1: PrecBit1(10) = 1: PrecBit1(11) = 1
    PrecBit1(12) = 2: PrecBit1(13) = 2: PrecBit1(14) = 3: PrecBit1(15) = 4
    Dim i As Long, mode As Long, nCount As Long
    For i = LBound(Utf8) To UBound(Utf8)
        Dim c As Byte: c = Utf8(i)
        If c <= &H7F Then             'ANSI           0xxxxxxx
            If mode = 0 Then nCount = nCount + 1 Else Exit For
        ElseIf c <= &HBF Then         'UTF8后续字节   10xxxxxx
            mode = mode - 1: If mode < 0 Then Exit For
            If mode = 0 Then nCount = nCount + 1
        Else                          'UTF8首字节     110xxxxx - 1111110x
            If mode <> 0 Then Exit For
            Dim k As Byte: k = PrecBit1(c \ 16)
            If k = 4 Then k = k + PrecBit1(c And 15)
            mode = k - 1: If k > 6 Then Exit For    ' FE FF 不可能出现
        End If
    Next
    CharLength = nCount: IS_UTF8 = (mode = 0)
End Function