• 无视编码都统一转成unicode 然后截断 例如 。“发发发发发发” 操作之后显示为 “发发发发...”


    -- local function checkPlayName( str )

    -- str = Utils.utf8_to_unicode(str)
    -- local retStr = ""
    -- local num = 0
    -- local lenInByte = #str
    -- local x = 1

    -- for i=1,lenInByte do
    -- i = x
    -- local curByte = string.byte(str, x)
    -- local byteCount = 1;
    -- if curByte>0 and curByte<=127 then
    -- byteCount = 1
    -- elseif curByte>127 and curByte<240 then
    -- byteCount = 3
    -- elseif curByte>=240 and curByte<=247 then
    -- byteCount = 4
    -- end
    -- local curStr = string.sub(str, i, i+byteCount-1)
    -- retStr = retStr .. curStr
    -- x = x + byteCount
    -- if x >= lenInByte then
    -- local finalRetStr = Utils.unicode_to_utf8(retStr)
    -- return finalRetStr
    -- end
    -- num = num + 1
    -- if num >= 24 then
    -- retStr = retStr.."..."
    -- local finalRetStr = Utils.unicode_to_utf8(retStr)
    -- return finalRetStr
    -- end
    -- end
    -- local finalRetStr = Utils.unicode_to_utf8(retStr)
    -- return finalRetStr
    -- end
    -- Utils.checkPlayName = checkPlayName


    -- local bit = require("bit")

    -- local function unicode_to_utf8(convertStr)

    -- if type(convertStr)~="string" then
    -- return convertStr
    -- end

    -- local resultStr=""
    -- local i=1
    -- while true do

    -- local num1=string.byte(convertStr,i)
    -- local unicode

    -- if num1~=nil and string.sub(convertStr,i,i+1)=="\u" then
    -- unicode=tonumber("0x"..string.sub(convertStr,i+2,i+5))
    -- i=i+6
    -- elseif num1~=nil then
    -- unicode=num1
    -- i=i+1
    -- else
    -- break
    -- end

    -- print(unicode)

    -- if unicode <= 0x007f then

    -- resultStr=resultStr..string.char(bit.band(unicode,0x7f))

    -- elseif unicode >= 0x0080 and unicode <= 0x07ff then

    -- resultStr=resultStr..string.char(bit.bor(0xc0,bit.band(bit.rshift(unicode,6),0x1f)))

    -- resultStr=resultStr..string.char(bit.bor(0x80,bit.band(unicode,0x3f)))

    -- elseif unicode >= 0x0800 and unicode <= 0xffff then

    -- resultStr=resultStr..string.char(bit.bor(0xe0,bit.band(bit.rshift(unicode,12),0x0f)))

    -- resultStr=resultStr..string.char(bit.bor(0x80,bit.band(bit.rshift(unicode,6),0x3f)))

    -- resultStr=resultStr..string.char(bit.bor(0x80,bit.band(unicode,0x3f)))

    -- end

    -- end

    -- resultStr=resultStr..''

    -- print(resultStr)

    -- return resultStr

    -- end
    -- Utils.unicode_to_utf8 = unicode_to_utf8

    -- local function utf8_to_unicode(convertStr)

    -- if type(convertStr)~="string" then
    -- return convertStr
    -- end

    -- local resultStr=""
    -- local i=1
    -- local num1=string.byte(convertStr,i)

    -- while num1~=nil do

    -- print(num1)

    -- local tempVar1,tempVar2

    -- if num1 >= 0x00 and num1 <= 0x7f then

    -- tempVar1=num1

    -- tempVar2=0

    -- elseif bit.band(num1,0xe0)== 0xc0 then

    -- local t1 = 0
    -- local t2 = 0

    -- t1 = bit.band(num1,bit.rshift(0xff,3))
    -- i=i+1
    -- num1=string.byte(convertStr,i)

    -- t2 = bit.band(num1,bit.rshift(0xff,2))


    -- tempVar1=bit.bor(t2,bit.lshift(bit.band(t1,bit.rshift(0xff,6)),6))

    -- tempVar2=bit.rshift(t1,2)

    -- elseif bit.band(num1,0xf0)== 0xe0 then

    -- local t1 = 0
    -- local t2 = 0
    -- local t3 = 0

    -- t1 = bit.band(num1,bit.rshift(0xff,3))
    -- i=i+1
    -- num1=string.byte(convertStr,i)
    -- t2 = bit.band(num1,bit.rshift(0xff,2))
    -- i=i+1
    -- num1=string.byte(convertStr,i)
    -- t3 = bit.band(num1,bit.rshift(0xff,2))

    -- tempVar1=bit.bor(bit.lshift(bit.band(t2,bit.rshift(0xff,6)),6),t3)
    -- tempVar2=bit.bor(bit.lshift(t1,4),bit.rshift(t2,2))

    -- end

    -- resultStr=resultStr..string.format("\u%02x%02x",tempVar2,tempVar1)
    -- print(resultStr)

    -- i=i+1
    -- num1=string.byte(convertStr,i)
    -- end

    -- print(resultStr)

    -- return resultStr

    -- end
    -- Utils.utf8_to_unicode = utf8_to_unicode

  • 相关阅读:
    理解MapReduce计算构架
    熟悉HBase基本操作
    爬虫大作业
    第三章 熟悉常用的HDFS操作
    数据结构化与保存
    使用正则表达式,取得点击次数,函数抽离
    爬取校园新闻首页的新闻
    网络爬虫基础练习
    Hadoop综合大作业
    hive基本操作与应用
  • 原文地址:https://www.cnblogs.com/rexzhao/p/6553499.html
Copyright © 2020-2023  润新知