dump_gp_file.cpp

/* dump Gobe Productive file structure
 * François Revol, 2007
 */
#include <stdio.h>
#include <string.h>
#include <unistd.h>
#include <fcntl.h>
#include <ctype.h>
#include <ByteOrder.h>

char buff[256];

#define READ(n) { int err; if ((err = read(fd, buff, (n))) < (n)) { printf("read(%d) = %d at line %s\n", (n), err, __LINE__); exit(1);} }

#define READTAG() { READ(4); printf("%stag:'%4.4s'\n", indent, buff); }

// read and compare
#define READTAGC(t) { READ(4); printf("%stag:'%4.4s'\n", indent, buff); if (strncmp(buff, t, 4)) { printf("Unmatched tag %4.4s != '%4.4s'\n", buff, t); exit(1); } }
#define READETAGC(t) { READ(4); printf("%send tag:'%4.4s'\n", indent, buff); if (strncmp(buff, t, 4)) { printf("Unmatched tag %4.4s != '%4.4s' level %d, tag %d left %d\n", buff, t, level, tags, count); exit(1); } }

#define READSTR() { int l; READ(2); l = B_BENDIAN_TO_HOST_INT16(*(uint16 *)buff)+1; \
if (l > 255) { printf("string too long (%d)\n", l); exit(1); } \
READ(l); printf("%sstring: '%s' (%d)\n", indent, buff, l); }

#define SKIP(n) { lseek(fd, n, SEEK_CUR); }

#define TELL() { off_t pos = lseek(fd, 0LL, SEEK_CUR); printf("%s @ %Ld 0x%Lx\n", indent, pos, pos); }

int dump(int level, int fd, int count)
{
    char indent[20] = "";
    int i;
    int tags = 0;
    int v;
    if (level > 19) {
        printf("level>19\n");
        exit(1);
    }
    memset(indent, '\0', 20);
    memset(indent, '\t', level);

    do {
        int len;
        TELL();
        READTAG();
        if (!isupper(buff[0]) && !isupper(buff[1])
         && !isupper(buff[2]) && !isupper(buff[3])) {
            printf("%s<<<<\n", indent);
            return 1;
        }
        if (level==0 && strncmp(buff, "DOC!", 4)) {
            printf("invalid!\n");
            exit(1);
        }
        printf("%sstart tag: %4.4s\n", indent, buff);
        switch (B_BENDIAN_TO_HOST_INT32(*(uint32 *)buff)) {
        case 'DOC!':
            SKIP(0x4e);
            dump(level+1, fd, -1);
            READETAGC("ENT!");
            break;
        case 'VERM':
            SKIP(0x20-4);
            break;
        case 'GRFX':
            READ(0x4); // number of strings ??
            v = B_BENDIAN_TO_HOST_INT32(*(uint32*)buff);
            while (v--)
                READSTR();
            break;
        case 'WORD':
        case 'PROT':
            READ(0x4);
            READSTR();
            break;
        case 'PAGE':
        case 'BROT':
            SKIP(0x4);
            READSTR();
            SKIP(0x20-4);
            break;
        case 'SETM':
            SKIP(0x10);
            break;
        case 'PENM':
            SKIP(0x10-2);
            break;
        case 'STYM':
            dump(level+1, fd, 3);
            READETAGC("StyM");
            break;
        case 'PRPM':
            READ(4);
            printf("%s%d properties following...\n", indent, B_BENDIAN_TO_HOST_INT32(*(uint32*)buff));
            dump(level+1, fd, B_BENDIAN_TO_HOST_INT32(*(uint32*)buff));
            READETAGC("PrpM"); // (Property Manager ?)
            break;
        case 'PROP':
            READ(4);
            printf("%s%4.4s\n", indent, buff);
            READ(4);
            SKIP(B_BENDIAN_TO_HOST_INT32(*(uint32*)buff));
            SKIP(4);
            break;
        case 'STYP':
            SKIP(4); // 0 ?
            READETAGC("StyP");
            break;
        case 'STLT': // StyLe LisT
            READ(4);
            dump(level+1, fd, B_BENDIAN_TO_HOST_INT32(*(uint32*)buff));
            READETAGC("StlT"); //
            break;
        case 'PTAB':
            READ(4);
            v = B_BENDIAN_TO_HOST_INT32(*(uint32*)buff) * 4;
            printf("%sskipping %d\n", indent, v);
            SKIP(v);
            READETAGC("PtaB");
            break;
        case 'REFT':
            READ(4);
            v = B_BENDIAN_TO_HOST_INT32(*(uint32*)buff) * 4;
            printf("%sskipping %d\n", indent, v);
            SKIP(v);
            READETAGC("RefT");
            break;
        case 'GRAP':
            READ(4);
            SKIP(16+3);
            READ(4);
            v = B_BENDIAN_TO_HOST_INT32(*(uint32*)buff);
            while (v--) {
                READSTR();
            }
            dump(level+1, fd, 2);
            SKIP(v);
            READETAGC("GraP");
            break;
        case 'STYL':
            READTAG();
            READ(4);
            SKIP(6);
            TELL();
            READTAG(); //??
            if (strncmp(buff, "Basi", 4) == 0) {
                READSTR();
                SKIP(16-4);
            } else if (strncmp(buff, "GRFX", 4) == 0) {
                //READSTR();
                SKIP(1+16+8);
            } else {
                printf("invalid tag\n");
                exit(1);
            }
            dump(level+1, fd, B_BENDIAN_TO_HOST_INT32(*(uint32*)buff));
            READETAGC("STYL");
            break;
        case 'lyr+':
            READ(4);
            v = B_BENDIAN_TO_HOST_INT32(*(uint32*)buff);
            while (v--) {
                READSTR();
            }
            READETAGC("lyr-");
            break;
        case 'vew+':
            READ(4);
            v = B_BENDIAN_TO_HOST_INT32(*(uint32*)buff);
            while (v--) {
                READSTR();
            }
            SKIP(4*3+1);
            READETAGC("vew-");
            break;
        default:
            printf("%sunknown tag level %d, tag %d left %d\n", indent, tags, count);
            exit(1);
        }
        tags++;
    } while (--count);
    return 0;
}

int main(int argc, char **argv)
{
    char *f = argv[1];
    int fd = open(f, O_RDONLY);
    READ(16);
    printf("%16.16s\n", buff);
    //READ(4);
/*  if (strncmp(buff, "DOC!", 4)) {
        printf("bad tag\n");
        exit(1);
    }*/
    dump(0, fd, -1);
}