tukubai-comma.c

/******************************************************************************
Usage   : comma [+<n>h][-4] <f1> <f2> ... <file>
        : comma -d[4] <string>
Version : Thu Dec 15 13:46:38 JST 2011
******************************************************************************/

#include <unistd.h>
#include <stdbool.h>
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <assert.h>

// for debug
#ifdef DEBUG
#define dbg(...) (printf("%s %u @%s:",__FILE__,__LINE__,__func__), printf(" "__VA_ARGS__))
#else
#define dbg(...)
#endif

typedef enum _opt_type
{
    NO_OPT = 0b0000,
    H_OPT = 0b0001,
    D_OPT = 0b0010,
    FOUR_OPT = 0b0100,
} opt_type_t;

typedef struct _line_t
{
    int num;
    char chars[];
} line_t;

typedef struct _context_t
{
    int keta_step;
    int cols_len;
    int cols[];
} ctx_t;

/**
 * Print usage of comma program
 * !! This function calls exit with EXIT_FAILURE. !!
 */
static void usage(void);

/**
 * Setup context.
 * @param {ctx_t *} context
 * @param {int}     arg array size
 * @param {char *}  arg chars array
 */
static void setup_context(ctx_t *, int, char *[]);

/**
 * Print given file with comma.
 * @param {FILE *} file pointer
 * @param const {ctx_t *} context
 */
static void print_comma_file(FILE *, const ctx_t *const);

/**
 * Print given line chars.
 * @param const {char *} line chars
 * @param const {ctx_t *} context
 */
static void print_comma_line(const char *const, const ctx_t *const);

/**
 * Convert given chars to comma separated chars.
 * E.g.
 * input chars  : 12345678
 * output chars : 1234,5678
 * @param  in_chars {char *} target characters
 * @param  ctx      {ctx_t *} context
 * @return          {char *} comma separated characters
 */
static char *to_comma_word(const char *const, const ctx_t *const);

int main(int argc, char *argv[])
{
    int opt;

    extern char *optarg;
    extern int optind, opterr;

    unsigned int opts = NO_OPT;

    while ((opt = getopt(argc, argv, "h4d")) != -1)
    {
        switch (opt)
        {
        case 'h':
            opts |= H_OPT;
            break;
        case 'd':
            opts |= D_OPT;
            break;
        case '4':
            opts |= FOUR_OPT;
            break;
        default:
            usage();
            break;
        }
    }

    // subtract number of options
    argc -= optind;

    // forward pointer for number of options
    argv += optind;

    // check remain args, or if there are no args the program will exit
    if (argc < 1)
    {
        usage();
    }

    dbg("opts = %u\n", opts);

    char *comma_word;
    int cols_len = 10;
    ctx_t *ctx = (ctx_t *) malloc(sizeof(ctx_t) + sizeof(int) * cols_len);
    ctx->keta_step = 3;
    ctx->cols_len = cols_len;

    switch (opts)
    {
    case (H_OPT | FOUR_OPT):
        dbg("### 4 MODE\n");
        ctx->keta_step = 4;
    case (H_OPT):
        dbg("### H MODE\n");
        char *filename = argv[argc - 1];
        dbg("filename = %s\n", filename);
        FILE *f = fopen(filename, "r");
        setup_context(ctx, argc, argv);
        print_comma_file(f, ctx);
        break;

    case (D_OPT | FOUR_OPT):
        dbg("### 4 MODE\n");
        ctx->keta_step = 4;
    case (D_OPT):
        dbg("### D MODE\n");
        comma_word = to_comma_word(argv[0], ctx);
        fprintf(stdout, "%s\n", comma_word);
        free(comma_word);
        break;

    case (FOUR_OPT):
        dbg("### 4 MODE");
        ctx->keta_step = 4;
        break;

    default:
        dbg("### No match  mode\n");
        usage();
    }

    exit(EXIT_SUCCESS);
}

static void usage(void)
{
    fprintf(stderr, "Usage   : comma [+<n>h][-4] <f1> <f2> ... <file>\n");
    fprintf(stderr, "        : comma -d[4] <string>\n");
    fprintf(stderr, "Version : Thu Dec 15 13:46:38 JST 2011\n");
    exit(EXIT_FAILURE);
}

static void setup_context(ctx_t *ctx, int argc, char *argv[])
{
    int cols_len = argc - 1;
    if (cols_len > ctx->cols_len)
    {
        if (!realloc(ctx, sizeof(ctx_t) + sizeof(int) * cols_len))
        {
            exit(EXIT_FAILURE);
        }
    }
    ctx->cols_len = cols_len;

    for (int i = 0; i < cols_len; i++)
    {
        ctx->cols[i] = atoi(argv[i]);
        dbg("ctx->cols[%d] = %d\n", i, ctx->cols[i]);
    }
}

static void print_comma_file(FILE *f, const ctx_t *const ctx)
{
    int line_num = 1;
    char buf[512];
    while (fgets(buf, 512, f))
    {
        dbg("buf = %s\n", buf);
        print_comma_line(buf, ctx);
        line_num++;
    }
}

static void print_comma_line(const char *const _in_line,
                             const ctx_t *const ctx)
{
    int col_num = 1;
    char *in_line = (char *) calloc(strlen(_in_line), sizeof(char));
    strcpy(in_line, _in_line);
    char *tok = strtok(in_line, " ");
    while (tok)
    {
        dbg("line[%d] = %s\n", col_num, tok);
        char *comma_tok = to_comma_word(tok, ctx);
        fprintf(stdout, "%s ", comma_tok);
        tok = strtok(NULL, " ");
        col_num++;
    }
}

static char *to_comma_word(const char *const in_chars,
                           const ctx_t *const ctx)
{
    // input chars length
    size_t in_chars_length = strlen(in_chars);

    // inout chars each keta steps.
    // input chars : XYZXYZXYZ   : input chars's keta step is 3.
    // output chars: XYZ,XYZ,XYZ : output chars's keta step is 4.
    int in_chars_keta_step = ctx->keta_step;
    int out_chars_keta_step = in_chars_keta_step + 1;

    // number of comma in output chars.
    // if output chars length is multiple number of the input keta step,
    // we must decrement comma count. Because most top comma is not necessary.
    // input chars  : 1234
    // output chars : ,1234
    //                ^
    //               This comma is garbage!!
    int comma_count = in_chars_length / in_chars_keta_step;
    size_t out_chars_length = in_chars_length +
                              ((comma_count * in_chars_keta_step == in_chars_length) ? comma_count - 1 : comma_count);

    dbg("in chars keta step  = %d\n", in_chars_keta_step);
    dbg("out chars keta step = %d\n", out_chars_keta_step);
    dbg("input chars length  = %ld\n", in_chars_length);
    dbg("output chars length = %ld\n", out_chars_length);

    char *out_chars = calloc(out_chars_length, sizeof(char));
    strncpy(out_chars, in_chars, in_chars_length);

    int ii = in_chars_length - 1;
    int oi = out_chars_length - 1;
    for ( ; ii >= in_chars_keta_step - 1 & oi >= out_chars_keta_step - 1; ii -= in_chars_keta_step, oi -= out_chars_keta_step)
    {

        dbg("ii = %d\n", ii);
        dbg("oi = %d\n", oi);

        strncpy(&out_chars[oi - in_chars_keta_step + 1], &in_chars[ii - in_chars_keta_step + 1], in_chars_keta_step);

        int comma_index = oi - out_chars_keta_step + 1;
        if (comma_index != 0)
            out_chars[comma_index] = ',';
    }

    dbg("in_chars  = %s\n", in_chars);
    dbg("out_chars = %s\n", out_chars);

    return out_chars;
}

サンプルデータファイル

20060201 296030 6710000
20060202 1300100 3130000
20060203 309500 20100
20060204 16300 300100
20060205 41000 210000
20060206 771100 400000