dsk/lzw.c

/*
 * LZW algorithm as used in the IBM .DSK floppy disk image format.
 * Uses a fixed 12bit code size and an LRU dictionary entry replacement policy.
 */

#include "lzw.h"
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/types.h>

void lzw_init(struct lzw_ctx* c)
{
    int i;

    c->coffset = c->doffset = c->codes = 0L;
    c->input_buffer = 0;
    c->dict[0].usecount = c->dict[0].length = 0;
    c->dict[0].prefix = c->dict[0].lru_prev = c->dict[0].lru_next = 0;
    for (i = 1; i <= 256; i++) {
        c->dict[i].length = 1;
        c->dict[i].prefix = 0;
        c->dict[i].last = i - 1;
        c->dict[i].first = i - 1;
        c->dict[i].usecount = 1; /* so that this code will never end up in the lru list */
        c->dict[i].lru_prev = c->dict[i].lru_next = 0;
    }
    for (i = 257; i < 4096; i++) {
        c->dict[i].length = 0;
        c->dict[i].prefix = 0;
        c->dict[i].last = 0;
        c->dict[i].first = 0;
        c->dict[i].usecount = 0;
        c->dict[i].lru_prev = (i == 257 ? 0 : i - 1);
        c->dict[i].lru_next = (i == 4095 ? 0 : i + 1);
    }
    c->lru_head = 257;
    c->lru_tail = 4095;
    c->last_emitted_code = 0;
}

/* Remove code from the lru list */
static void lzw_lru_unlink(struct lzw_ctx* c, uint16_t code)
{
    uint16_t next, prev;

    next = c->dict[code].lru_next;
    prev = c->dict[code].lru_prev;
    if (prev != 0)
        c->dict[prev].lru_next = next;
    else
        c->lru_head = next;
    if (next != 0)
        c->dict[next].lru_prev = prev;
    else
        c->lru_tail = prev;
}

/* Add the given code to the back of the lru list. The code must be removed from the list before calling this function. */
static void lzw_lru_append(struct lzw_ctx* c, uint16_t code)
{
    c->dict[code].lru_next = 0;
    c->dict[code].lru_prev = c->lru_tail;
    c->dict[c->lru_tail].lru_next = code;
    c->lru_tail = code;
}

#if 0
static void lzw_print_lru(struct lzw_ctx* c)
{
    printf("head=%03x tail=%03x\n", c->lru_head, c->lru_tail);
    for (int i = 0; i < 4096; i++)
        printf("[%03x] next=%03x prev=%03x\n", i, c->dict[i].lru_next,
            c->dict[i].lru_prev);
}

static void lzw_validate_lru(struct lzw_ctx* c)
{
    for (uint16_t code = c->lru_head; code != 0; code = c->dict[code].lru_next) {
        printf("%03x", code);
        if (c->dict[code].usecount != 0)
            putchar('!');
        if (c->dict[code].lru_next != 0 && c->dict[c->dict[code].lru_next].lru_prev != code)
            putchar('>');
        putchar(' ');
    }
    putchar('\n');
}
#endif

static int lzw_build_entry(struct lzw_ctx* c, uint16_t code)
{
    int pos;

    pos = c->lru_head;
    lzw_lru_unlink(c, pos);

    if (c->dict[pos].prefix != 0) {
        c->dict[c->dict[pos].prefix].usecount--;
        if (c->dict[c->dict[pos].prefix].usecount == 0)
            lzw_lru_append(c, c->dict[pos].prefix);
    }

    c->dict[pos].prefix = c->last_emitted_code;
    c->dict[pos].last = c->dict[(code == pos ? c->dict[pos].prefix : code)].first;
    c->dict[pos].first = c->dict[c->dict[pos].prefix].first;
    c->dict[pos].length = c->dict[c->dict[pos].prefix].length + 1;
    c->dict[pos].usecount = 0;

    if (c->dict[c->dict[pos].prefix].usecount == 0)
        lzw_lru_unlink(c, c->dict[pos].prefix);
    c->dict[c->dict[pos].prefix].usecount++;
    lzw_lru_append(c, pos);

    return pos;
}

static void lzw_emit(struct lzw_ctx* c, uint16_t code, FILE* f)
{
    int i;
    uint16_t t;

    for (i = c->dict[code].length, t = code; t != 0 && i > 0; i--, t = c->dict[t].prefix)
        c->output_buffer[i - 1] = c->dict[t].last;
    fwrite(c->output_buffer, c->dict[code].length, 1, f);

    c->doffset += c->dict[code].length;
    c->last_emitted_code = code;
}

static int lzw_get_next_code(struct lzw_ctx* c, FILE* f)
{
    uint16_t result;
    int next_byte;

    next_byte = fgetc(f);
    if (next_byte == EOF)
        return -1;
    c->coffset++;

    if ((c->codes & 1) == 0) {
        c->input_buffer = fgetc(f);
        if (c->input_buffer == EOF)
            return -1;
        c->coffset++;
        result = (next_byte << 4) | (c->input_buffer >> 4);
    } else {
        result = ((c->input_buffer & 0x0f) << 8) | next_byte;
    }

    c->codes++;
    return result & 0x0fff;
}

int lzw_decompress_file(FILE* infp, FILE* outfp)
{
    struct lzw_ctx* ctx;
    int code;

    ctx = (struct lzw_ctx*)malloc(sizeof(struct lzw_ctx));
    if (ctx == NULL)
        return -1;

    lzw_init(ctx);

    while ((code = lzw_get_next_code(ctx, infp)) != 0) {
        if (ctx->last_emitted_code != 0)
            lzw_build_entry(ctx, code);
        lzw_emit(ctx, code, outfp);
    }

    /*
    printf(
        "\nEnd of file after reading %ld bytes (%ld symbols) and writing %ld "
        "bytes\n",
        ctx.coffset, ctx.codes, ctx.doffset);
    */
    free(ctx);

    return 0;
}