tabstospace.cpp

#include <stdio.h>  // for fopen, fread, fwrite and fclose
#include <string.h> // for memset and strcmp

int main(int argc, char* argv[])
{
    if (argc != 3)
    {
        fprintf(stderr, "Usage: %s <src_file> <dest_file>\n", argv[0]);
        return 1;
    }
    // TODO make robust check for same file
    /*
     * #include <sys/stat.h>
     *
     */
    if (0 == strcmp(argv[1], argv[2]))
    {
        fprintf(stderr, "src and dest must not be the file\n");
        return 1;
    }

    FILE* r_fd = fopen(argv[1], "rb");
    FILE* w_fd = fopen(argv[2], "wb");

#define READ_BUFFER_SIZE 1024
#define WRITE_BUFFER_SIZE (READ_BUFFER_SIZE * 4)
    char read_buffer[READ_BUFFER_SIZE];
    char write_buffer[WRITE_BUFFER_SIZE];

    unsigned int write_pos = 0;
    unsigned int bytes_read = 0;
    bytes_read = fread(read_buffer, 1, 1024, r_fd);

    int lastNonWhiteSpace = 0;
    int line = 1;
    int col = 0;

    while (bytes_read)
    {
        for(unsigned int i = 0; i < bytes_read; i++)
        {
            col++;
            char c = read_buffer[i];
            if (c == '\t')
            {
                memset(write_buffer + write_pos, ' ', 4);
                write_pos += 4;
                col += 4;
            }
            else
            {
                if (c == '\n')
                {
                    // strip trailing whitespace
                    int trailing_whitespace = ((col - 1) - lastNonWhiteSpace);
                    if (trailing_whitespace)
                    {
//                        printf("%d bytes of trailing whitespace detected at: {%d, %d}\n", trailing_whitespace, line, lastNonWhiteSpace);
                    }
                    write_pos -= trailing_whitespace;
                    line++;
                    col = 0;
                    lastNonWhiteSpace = 0;
                }
                else if (c != ' ')
                {
                    lastNonWhiteSpace = col;
                }
                write_buffer[write_pos++] = c;
            }
        }
        fwrite(write_buffer, 1, write_pos, w_fd);
        bytes_read = fread(read_buffer, 1, 1024, r_fd);
        write_pos = 0;
    }

    fclose(w_fd);
    fclose(r_fd);
}