自己尝试着从头写了一个自动换行算法的测试工具,支持UTF-8。实现的重点是区分出UTF-8编码,并且识别它的全部字节。

代码如下:

#include <stdio.h>
#include <string.h>

#define LINE_CHARACTERS_COUNT 30

 

long GetFileSize(const char* filename);

 

int main (int argc, const char * argv[])
{
   int Width,i;
   unsigned char * pBuffer = NULL;
   unsigned char * pLine = NULL;
   int nLineWarp = LINE_CHARACTERS_COUNT;
   int nStart=0,nEnd =0;
   int nLastWordPos = 0;
   int bWriteNow = 0;
   int nUnicode = 0;
 

 ///!判断参数是否正确
 if(argc<2)
 {
  printf("Invalid parameters!/n");
  printf("Usage: linewrap InputFile [characters_each_line]/n");
  printf("Try it again!/n");
  return 1;
 }
 

///!取出相应的参数
 Width = GetFileSize(argv[1]);
 if(argc == 3)
  nLineWarp = atoi(argv[2]);
 
 ///! Allocate memory
 pBuffer = (unsigned char *)malloc(Width);
 pLine = (unsigned char *)malloc(nLineWarp+512);
 
 FILE * fp = fopen(argv[1], "r");

 

 ///!如果打开文件失败
 if(!fp)
 {
  //Failed to open source file
  free(pBuffer);
  free(pLine);
  printf("Cannot open %s!/n",argv[1]);
  return 2;
 }

 ///!Read data and close it soon
 Width = fread(pBuffer,1,Width,fp);
 fclose(fp);
 
 printf("=======OUTPUT=======/n");
 

///!开始进行换行处理并输出

 for(i=0;i<Width;i++)
 {
  switch(pBuffer[i])
  {
   case '/n':
   case '/r': //Line break
    bWriteNow = 1;
   case ' ':
   case ',':
   case ';':
   case '.':
   case '!': //Word splitters
    nEnd = i;
    nLastWordPos = i;
    if(nEnd-nStart>=nLineWarp)
     bWriteNow = 1;
    break;

   default: ///<Normal characters
    nEnd = i;
    ///!check UTF-8 decode
    nUnicode = pBuffer[i]>>5 & 0x07;
    switch(nUnicode)
    {
     case 0x07: ///<Three-bytes UTF-8
      i++;   ///<it will enter 0x06 then.
     case 0x06: ///<Two-bytes UTF-8
      i++;
      nEnd = i;
      nLastWordPos = i;
      break;
     default:
      ///!unsupported UTF-8
      break;
    }
    
    if(nEnd-nStart>=nLineWarp)
     bWriteNow = 1;
  }
  
  if(bWriteNow && nLastWordPos-nStart >0)
  {
   ///!If has output data
   strncpy(pLine,pBuffer+nStart,nLastWordPos-nStart+1);
   pLine[nLastWordPos-nStart+1] = '/0';
   printf("%s/n",pLine);
   nStart = nLastWordPos+1;
   bWriteNow = 0; ///<clear output flag here
  }
  else
   bWriteNow = 0; ///<clear output flag while has no any output string
 }
 
 if(nEnd>nStart)
 {
  printf("%s/n",pBuffer+nStart);
 }
 
 free(pLine);
 free(pBuffer);
 return 0;
}

///!Get the file size
long GetFileSize(const char* filename)
{

   ///! 不同系统下有不同的实现方法,只要返回指定文件的大小即可

   ///! Linux下可以使用stat函数取得
}