lua版干扰屏蔽字编码解码

spam.lua

-- split utf8 char
function string.splitUtf8(str)
  local list = {}
  local p = 1
  for i = 1, #str do
    local c = string.byte(str, i)
    if c < 0x80 then
      -- ascii 
      table.insert(list, string.sub(str, p, i))
      p = i + 1
    elseif c >= 0xC0 then
      -- the first byte of multiple-byte character
      if p <= i - 1 then
        -- insert previous character
        table.insert(list, string.sub(str, p, i - 1))
        p = i
      end
    end
  end
  if p <= #str then
    table.insert(list, string.sub(str, p, #str))
  end
  return list
end

-- 干扰屏蔽字
-- 假设屏蔽字不会出现单字屏蔽字,用^符号分割所有字符
function string.encodeSpam(str)
  return table.concat(string.splitUtf8(str), "^")
end

-- 假设^符号不会在原始字符串中用到,直接移除所有的^
function string.decodeSpam(str)
  return string.gsub(str, "%^", "")
end

-- 如果^符号会出现在原始字符串中,就需要分离所有字符串,隔位移除一个字符
function string.decodeSpam2(str)
  local list = string.splitUtf8(str)
  local list2 = {}
  for i = 1, #list, 2 do
    table.insert(list2, list[i])
  end
  return table.concat(list2, '')
end