better search sequence management

[hexdump.git] / hexdump.c

/* depend: */
/* cflags: */
/* linker: debug.o */

#include <assert.h>
#include <malloc.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>

#include "debug.h"

/* macros */

#define CEIL(x, y) (((x) + (y) - 1) / (y))
#define MIN(x, y) (((x) < (y)) ? (x) : (y))
#define MAX(x, y) (((x) > (y)) ? (x) : (y))

//#define BUFFERSIZE 4096
#define BUFFERSIZE 256
#define NBCOLS 8
#define NBDIGITS 6
#define SEQLEN 32

/* gobal variables */

int nbcols = NBCOLS;
int nbdigits = NBDIGITS;
int offset = 1;

char buffer[BUFFERSIZE] = {0};
FILE *fin = NULL;
int addrfile = 0;
FILE *fout = NULL;
char *progname = NULL;

/* type definitions */

typedef struct {
    char *sequence;
    char bytes[SEQLEN];
    int length;
} sequence_t;

/* help function */

int usage (int ret)
{
    FILE *fd = ret ? stderr : stdout;
    fprintf (fd, "usage: %s [-i file] [-h] [-n nbcols] [-o file] [-v]\n", progname);
    fprintf (fd, " -i: input file\n");
    fprintf (fd, " -h: help message\n");
    fprintf (fd, " -n: number of columns\n");
    fprintf (fd, " -e: commands\n");
    fprintf (fd, " -o: output file\n");
    fprintf (fd, " -v: verbose level (%d)\n", verbose);
    fprintf (fd, "\n");
    fprintf (fd, "commands: [/hstr/|0xaddr] [a hstr] [d nb|-] [i hstr] [p nb|-] [s/h1/h2/[g]]\n");
    fprintf (fd, " 0x: move to address addr\n");
    fprintf (fd, " //: move to hexa stringi hstr\n");
    fprintf (fd, " a : append hexa string hstr to current address\n");
    fprintf (fd, " d : delete nb bytes (- until end file)\n");
    fprintf (fd, " i : insert hexa string hstr to current address\n");
    fprintf (fd, " p : print nb bytes (- until end file)\n");
    fprintf (fd, " s : substitute h1 by h2 (g for globally)\n");

    return ret;
}

/* get number of digits */

int getnbdigits (long int l) {
    int n = 0;
    while (l) {
        n += 2;
        l /= 256;
    }
    return n;
}

/* print a line */

void printline (char *buffer, int nb, int addr) {
    int i;

    printf ("0x%0*x:", nbdigits, addr);
    for (i = 0; i < nb; i++) {
        printf (" %02x", buffer[i]);
    }
    for (i = nb; i < nbcols; i++) {
        printf ("   ");
    }
    printf ("  ");
    for (i = 0; i < nb; i++) {
        char c = buffer[i];
        printf ("%c", (c > 31) && (c < 127) ? c : '.');
    }
    printf ("\n");
}

/* write file function */

int writefile (char *pt, int nb) {
    if (fout) {
        fwrite (pt, 1, nb, fout);
    }
    return 1;
}

/* search sequence function */

int searchseq (sequence_t *seq) {
    char *pt = buffer;
    int nb = 0;
    int i, j;
    int valid = 0;

    VERBOSE (DEBUG, printf ("search sequence: %s\n", seq->sequence));

    while (!feof (fin)) {
        int nbread = fread (pt, 1, BUFFERSIZE - (pt - buffer), fin);
        nb += nbread;
        pt = buffer;
        for (i = 0; i < nb - seq->length; i++) {
            valid = 1;
            for (j = 0; (j < seq->length) && (valid); j++) {
                if (pt[i + j] != seq->bytes[j]) {
                    valid = 0;
                }
            }
            if (valid) {
                break;
            }
        }

        if (!valid) {
            writefile (buffer, nb - seq->length);
            offset = 0;
            addrfile += nb - seq->length;
            for (i = 0; i < seq->length; i++) {
                buffer[i] = buffer[nb - seq->length + i];
            }
            pt = buffer + seq->length;
            nb = seq->length;
        } else {
            writefile (buffer, i);
            offset = seq->length;
            addrfile += i;
            fseek (fin, i - nb, SEEK_CUR);
            VERBOSE (DEBUG, printf ("found sequence (%d)\n", i - nb));
            return 0;
        }
    }

    if (!valid) {
        writefile (buffer, nb);
        addrfile += seq->length;
    }

    return 1;
}

/* hexadecimal dump function */

int hexdump (int len) {
    char buffer[BUFFERSIZE] = {0};
    int i;

    char *pt = buffer;

    int nb = 0;
    while (!feof (fin)) {
        int nbtoread = BUFFERSIZE - (pt - buffer);
        if ((len > 0) && (nbtoread > len)) {
            nbtoread = len;
        }
        int nbread = fread (pt, 1, nbtoread, fin);
        if (len > 0) {
            len -= nbread;
        }
        nb += nbread;
        pt = buffer;

        /* print line */
        while ((nb - (int)(pt - buffer)) / nbcols > 0) {
            printline (pt, nbcols, addrfile);
            writefile (pt, nbcols);
            addrfile += nbcols;
            pt += nbcols;
        }

        /* copy end buffer */
        nb -= pt - buffer;
        for (i = 0; i < nb; i++) {
            buffer[i] = pt[i];
        }
        pt = buffer + nb;

        /* end partial reading */
        if (len == 0) {
            break;
        }
    }

    /* last line */
    if (nb > 0) {
        printline (buffer, nb, addrfile);
        writefile (buffer, nb);
        addrfile += nb;
    }

    return 0;
}

/* parse octal string */

long int octal (char *s, int n) {
    int i;
    long int l = 0;
    for (i = 0; i < n; i++) {
        if ((s[i] >= '0') && (s[i] <= '9')) {
            l = l * 8 + s[i] - '0';
        } else {
            return -1;
        }
    }
    return l;
}

/* parse hexa string */

long int hexa (char *s, int n) {
    int i;
    long int l = 0;
    for (i = 0; i < n; i++) {
        l *= 16;
        if ((s[i] >= '0') && (s[i] <= '9')) {
            l += s[i] - '0';
        } else if ((s[i] >= 'A') && (s[i] <= 'F')) {
            l += s[i] + 10 - 'A';
        } else if ((s[i] >= 'a') && (s[i] <= 'f')) {
            l += s[i] + 10 - 'a';
        } else {
            return -1;
        }
    }
    return l;
}

/* special character function */

int specialchar (char *s, char *b) {
    int i = 0, j = 0;
    while (s[i] != 0) {
        if (j == SEQLEN) {
            return 0;
        }
        if (s[i] != '\\') {
            b[j++] = s[i++];
            continue;
        }

        int l = -1;
        switch (s[i + 1]) {
        case 'a': l = 0x07; i += 2; break;
        case 'b': l = 0x08; i += 2; break;
        case 'e': l = 0x1b; i += 2; break;
        case 'f': l = 0x0c; i += 2; break;
        case 'n': l = 0x0a; i += 2; break;
        case 'r': l = 0x0d; i += 2; break;
        case 't': l = 0x09; i += 2; break;
        case 'v': l = 0x0b; i += 2; break;
        case '\\': l = '\\'; i += 2; break;
        case '\'': l = '\''; i += 2; break;
        case '"': l = '"'; i += 2; break;
        case '0':
        case '1':
        case '2':
        case '3':
            l = octal (s + i + 1, 3);
            if (l != -1) {
                i += 4;
            }
            break;
        case 'x':
            l = hexa (s + i + 2, 2);
            if (l != -1) {
                i += 4;
            }
            break;
        default:
        }
        if (l != -1) {
            VERBOSE (DEBUG, printf("l: 0x%02x '%c'\n", l, l));
        }
        b[j++] = (l != -1) ? l : s[i++];
    }

    return j;
}

/* main function */

int main (int argc, char *argv[])
{
    int rc = 0;
    char *input = NULL;
    char *output = NULL;
    char *commands = NULL;
    int printlen = -1;
    sequence_t seq = {0};
    char *addr = NULL;

    /* get basename */
    char *pt = progname = argv[0];
    while (*pt) {
        if ((*pt == '/') || (*pt == '\\')) {
           progname = pt + 1;
        }
        pt++;
    }

     while (argc-- > 1) {
        char *arg = *(++argv);
        if (arg[0] != '-') {
            VERBOSE (ERROR, fprintf (stderr, "%s: invalid option -- %s\n", progname, arg));
            return usage (1);
        }
        char c = arg[1];
        switch (c) {
        case 'e':
            arg = (arg[2]) ? arg + 2 : (--argc > 0) ? *(++argv) : NULL;
            if (arg) {
                if (commands == NULL) {
                    commands = arg;
                } else {
                    strcat (commands, " ");
                    strcat (commands, arg);
                }
            }
            break;
        case 'i':
            input = (arg[2]) ? arg + 2 : (--argc > 0 ) ? *(++argv) : NULL;
            break;
        case 'n':
            arg = (arg[2]) ? arg + 2 : (--argc > 0) ? *(++argv) : NULL;
            if (arg == NULL) {
                VERBOSE (ERROR, fprintf (stderr, "%s: missing number of columns\n", progname));
                return usage (1);
            }
            nbcols = atoi (arg);
            break;
        case 'o':
            output = (arg[2]) ? arg + 2 : (--argc > 0 ) ? *(++argv) : NULL;
            break;
        case 'v':
            arg = (arg[2]) ? arg + 2 : (--argc > 0) ? *(++argv) : NULL;
            if (arg == NULL) {
                VERBOSE (ERROR, fprintf (stderr, "%s: missing verbose level\n", progname));
                return usage (1);
            }
            verbose = atoi (arg);
            break;
        case 'h':
        default:
            return usage (c != 'h');
        }
    }

    /* check input */
    if (input) {
        fin = fopen (input, "rb");
        if (!fin) {
            VERBOSE (ERROR, fprintf (stderr, "error: can't open file '%s'\n", input));
            return 1;
        }
    } else {
        fin = stdin;
    }

    /* check output */
    if (output) {
        fout = fopen (output, "wb");
        if (!fout) {
            VERBOSE (ERROR, fprintf (stderr, "error: can't open file '%s'\n", output));
            fclose (fin);
            return 1;
        }
    } else {
        //fout = stdout;
    }

    /* get file size */
    if (fin != stdin) {
        fseek (fin, 0 , SEEK_END);
        long int filesize = ftell (fin);
        fseek (fin, 0 , SEEK_SET);
        nbdigits = getnbdigits (filesize);
    }

    if (commands == NULL) {
        hexdump (-1);
    } else {
        VERBOSE (DEBUG, printf ("commands: %s\n", commands));
        while ((*commands != '\0') && (rc == 0)) {
            switch (*commands++) {
            case ' ':
            case '\t':
                break;

            case '/': /* read patern */
                seq.sequence = commands;
                while (*commands) {
                    if (*commands == '/') {
                        *commands++ = 0;
                        break;
                    }
                    commands++;
                }
                seq.length = specialchar (seq.sequence, seq.bytes);
                if (seq.length != 0) {
                    rc = searchseq (&seq);
                } else {
                    VERBOSE (ERROR, fprintf (stderr, "incorrect sequence (%s)\n", seq.sequence));
                    rc = 1;
                }
                break;

            case '0': /* read address */
                break;

            case 'a': /* append mode */
                break;

            case 'd': /* delete mode */
                break;

            case 'i': /* insert mode */
                break;

            case 'p': /* print mode */
                printlen = -1;
                while (*commands != '\0') {
                    if ((*commands == ' ') || (*commands == '\t')) {
                        commands++;
                    } else if ((*commands >= '0') && (*commands <= '9')) {
                        printlen = strtol (commands, &commands, 10);
                        break;
                    } else if (*commands == '-') {
                        printlen = -1;
                        commands++;
                        break;
                    } else {
                        VERBOSE (ERROR, fprintf (stderr, "unkown print length (%s)\n", commands));
                        rc = 1;
                        break;
                    }
                }
                if (rc == 0) {
                    hexdump (printlen);
                }
                break;

            case 's': /* substitute mode */
                break;

            default:
                VERBOSE (ERROR, fprintf (stderr, "unknown command (%c)\n", commands[-1]));
                rc = 1;
            }
        }
    }

    /* end of file */
    if ((rc == 0) && (fout != NULL)) {
        while (!feof (fin)) {
            int nbread = fread (buffer, 1, BUFFERSIZE, fin);
            if (nbread) {
                fwrite (buffer, 1, nbread, fout);
            }
        }
    }

    /* close all */
    if (fin) fclose (fin);
    if (fout) fclose (fout);

    return rc;
}

// test: hexdump.exe -h | awk '/usage:/ { rc=1 } END { exit (1-rc) }'
// test: hexdump.exe -_ 2> /dev/null | awk 'END { if (NR == 0) { exit(0) } else exit (1) }'
// test: hexdump.exe -_ 2>&1 | awk '/usage:/ { rc=1 } END { exit (1-rc) }'
// test: hexdump.exe -i hexdump.c | grep -q '0x[0-9a-f]*: '
// test: hexdump.exe -i hexdump.c -n 3 | head -2 | tail -1 | grep -q '0x0003: 64 65 70  dep'
// test: hexdump.exe -i hexdump.c -o test.c -e 'p 200' | tail -1 | grep -q '0x00c0:'
// test: cmp hexdump.c test.c; x=$?; rm test.c; test x$x = x0
// test: hexdump.exe -i hexdump.c -e ' /cflags/ p 17 /debug/ p 8' | grep -q '0x0019: 2a 2f 0a 2f 2a 20 6c 69  \*/\./\* li'
// test: hexdump.exe -i hexdump.c -o test.c -e ' /cfl\x61gs/ p 16 /d\145bug/ p 8' | grep -q '0x0027: 64 65 62 75 67 2e 6f 20  debug.o'
// test: cmp hexdump.c test.c; x=$?; rm test.c; test x$x = x0
// test: hexdump.exe -i hexdump.c -e ' /\n/ p 8' | grep -q '0x000d: 0a 2f 2a 20 63 66 6c 61  \./\* cfla'
// test: hexdump.exe -i hexdump.c -o test.c -e ' /\a\b\e\f\r\t\v/ p 8'; x=$?; test x$x = x1
// test: cmp hexdump.c test.c; x=$?; rm test.c; test x$x = x0
// test: hexdump.exe -i hexdump.c -e ' /\"/' -e " /\\'/" -e ' /\\/'

/* vim: set ts=4 sw=4 et: */