修改代码以复制/粘贴.zip和.tar.gzip文件吗？

Question

Introduction

I'm writing my own cp program. With the code I currently have I'm able to copy and paste files.

Code

char *buf;
int fd;
int ret;
struct stat sb;
FILE *stream;

/*opening and getting size of file to copy*/
fd = open(argv[1],O_RDONLY);
if(fd == -1)
{
   perror("open");
   return 1;
}
/*obtaining size of file*/
ret = fstat(fd,&sb);
if(ret)
{
    perror("stat");
    return 1;
}

/*opening a stream for reading/writing file*/
stream fdopen(fd,"rb");
if(!stream)
{
     perror("fdopen");
     return 1;
}

/*allocating space for reading binary file*/
buf = malloc(sb.st_size);

/*reading data*/
if(!fread(buf,sb.st_size,1,stream))
{
    perror("fread");
    return 1;
}

/*writing file to a duplicate*/
fclose(stream);
stream = fopen("duplicate","wb");

if(!fwrite(buf,sb.st_size,1,stream))
{
    perror("fwrite");
    return 1;
}

 fclose(stream);
 close(fd);
 free(buf);
 return 0;

The problem

I'm unable to copy and paste .zip files and .tar.gz files. If i alter the code and give an extension such as 'duplicate.zip' (assuming im copying a zip file) such as .zip and then try and copy a .zip file everything is copied, however the new duplicated file does not act like a zip file and when i use cat it outputs nothing and this error when i attempt to unzip it anyway:

End-of-central-directory signature not found. Either this file is not a zipfile, or it constitutes one disk of a multi-part archive. In the latter case the central directory and zipfile comment will be found on the last disk(s) of this archive.

So how do i go about copying zip and pasting zip files and also .tar.gz files. Any pointers will be helpful, thanks in advance.

Answer 1

• You are using malloc() incorrectly. You want to allocate sb.st_size bytes.

   malloc(sb.st_size * sizeof buf) 

  should be

   malloc(sb.st_size) 

• The use of fread() is dubious and you are throwing away the result of fread() . Instead of

   if(!fread(buf,sb.st_size,1,stream)) 

  you should have

   size_t num_bytes_read = fread (buf, 1, sb.st_size, stream); if (num_bytes_read < sb.st_size) 

• You are using strlen() incorrectly. The content of buf is not guaranteed to be a string; and anyway you already know how many bytes you have in buf : sb.st_size . (Because if fread() returned a smaller number of bytes read you got angry and terminated the process.) So instead of

   fwrite(buf,strlen(buf),1,stream) 

you should have

    fwrite (buf, 1, sb.st_size, stream)

Answer 2

In addition to AlexP's notes ...

/*obtaining size of file*/
ret = fstat(fd,&sb);
if(ret)
{
    perror("stat");
    return 1;
}

// ...some code...

/*allocating space for reading binary file*/
buf = malloc(sb.st_size);

/*reading data*/
if(!fread(buf,sb.st_size,1,stream))
{
    perror("fread");
    return 1;
}

You have a race condition here. If the file size changes between your fstat call and malloc or fread you will read too much or too little of the file.

Fixing this leads us to the next issue, you're slurping the entire file into memory. While this might work for small files, it is extremely inefficient with your memory on large ones. For very large files it might be too large for a single malloc , and you're not checking if your malloc succeeds.

Instead, read and write the file a piece at a time. And read until there isn't any more to read.

uint8_t *buffer[4096]; // 4K buffer
size_t num_read;
while( (num_read = fread(buffer, sizeof(uint8_t), sizeof(buffer), in)) != 0 ) {
    if( fwrite( buffer, sizeof(uint8_t), num_read, out ) == 0 ) {
        perror("fwrite");
    }
}

This avoids the race condition by not having to call fstat in the first place. And it avoids allocating a potentially enormous hunk of memory. Instead it can all be done on the stack .

I've used uint8_t to get a hunk of bytes. It's a standard fixed width integer type from stdint.h . You can also use unsigned char to read bytes, and that's probably what uint8_t really is, but uint8_t makes it explicit.