zip.cpp

这是CODEPROJECT上的zip压缩解压代码

     */
    do {
        pqremove(tree, n);   /* n = node of least frequency */
        m = state.ts.heap[SMALLEST];  /* m = node of next least frequency */

        state.ts.heap[--state.ts.heap_max] = n; /* keep the nodes sorted by frequency */
        state.ts.heap[--state.ts.heap_max] = m;

        /* Create a new node father of n and m */
        tree[node].fc.freq = (ush)(tree[n].fc.freq + tree[m].fc.freq);
        state.ts.depth[node] = (uch) (Max(state.ts.depth[n], state.ts.depth[m]) + 1);
        tree[n].dl.dad = tree[m].dl.dad = (ush)node;
        /* and insert the new node in the heap */
        state.ts.heap[SMALLEST] = node++;
        pqdownheap(state,tree, SMALLEST);

    } while (state.ts.heap_len >= 2);

    state.ts.heap[--state.ts.heap_max] = state.ts.heap[SMALLEST];

    /* At this point, the fields freq and dad are set. We can now
     * generate the bit lengths.
     */
    gen_bitlen(state,(tree_desc *)desc);

    /* The field len is now set, we can generate the bit codes */
    gen_codes (state,(ct_data *)tree, max_code);
}

/* ===========================================================================
 * Scan a literal or distance tree to determine the frequencies of the codes
 * in the bit length tree. Updates opt_len to take into account the repeat
 * counts. (The contribution of the bit length codes will be added later
 * during the construction of bl_tree.)
 */
void scan_tree (TState &state,ct_data *tree, int max_code)
{
    int n;                     /* iterates over all tree elements */
    int prevlen = -1;          /* last emitted length */
    int curlen;                /* length of current code */
    int nextlen = tree[0].dl.len; /* length of next code */
    int count = 0;             /* repeat count of the current code */
    int max_count = 7;         /* max repeat count */
    int min_count = 4;         /* min repeat count */

    if (nextlen == 0) max_count = 138, min_count = 3;
    tree[max_code+1].dl.len = (ush)-1; /* guard */

    for (n = 0; n <= max_code; n++) {
        curlen = nextlen; nextlen = tree[n+1].dl.len;
        if (++count < max_count && curlen == nextlen) {
            continue;
        } else if (count < min_count) {
            state.ts.bl_tree[curlen].fc.freq = (ush)(state.ts.bl_tree[curlen].fc.freq + count);
        } else if (curlen != 0) {
            if (curlen != prevlen) state.ts.bl_tree[curlen].fc.freq++;
            state.ts.bl_tree[REP_3_6].fc.freq++;
        } else if (count <= 10) {
            state.ts.bl_tree[REPZ_3_10].fc.freq++;
        } else {
            state.ts.bl_tree[REPZ_11_138].fc.freq++;
        }
        count = 0; prevlen = curlen;
        if (nextlen == 0) {
            max_count = 138, min_count = 3;
        } else if (curlen == nextlen) {
            max_count = 6, min_count = 3;
        } else {
            max_count = 7, min_count = 4;
        }
    }
}

/* ===========================================================================
 * Send a literal or distance tree in compressed form, using the codes in
 * bl_tree.
 */
void send_tree (TState &state, ct_data *tree, int max_code)
{
    int n;                     /* iterates over all tree elements */
    int prevlen = -1;          /* last emitted length */
    int curlen;                /* length of current code */
    int nextlen = tree[0].dl.len; /* length of next code */
    int count = 0;             /* repeat count of the current code */
    int max_count = 7;         /* max repeat count */
    int min_count = 4;         /* min repeat count */

    /* tree[max_code+1].dl.len = -1; */  /* guard already set */
    if (nextlen == 0) max_count = 138, min_count = 3;

    for (n = 0; n <= max_code; n++) {
        curlen = nextlen; nextlen = tree[n+1].dl.len;
        if (++count < max_count && curlen == nextlen) {
            continue;
        } else if (count < min_count) {
            do { send_code(state, curlen, state.ts.bl_tree); } while (--count != 0);

        } else if (curlen != 0) {
            if (curlen != prevlen) {
                send_code(state, curlen, state.ts.bl_tree); count--;
            }
            Assert(state,count >= 3 && count <= 6, " 3_6?");
            send_code(state,REP_3_6, state.ts.bl_tree); send_bits(state,count-3, 2);

        } else if (count <= 10) {
            send_code(state,REPZ_3_10, state.ts.bl_tree); send_bits(state,count-3, 3);

        } else {
            send_code(state,REPZ_11_138, state.ts.bl_tree); send_bits(state,count-11, 7);
        }
        count = 0; prevlen = curlen;
        if (nextlen == 0) {
            max_count = 138, min_count = 3;
        } else if (curlen == nextlen) {
            max_count = 6, min_count = 3;
        } else {
            max_count = 7, min_count = 4;
        }
    }
}

/* ===========================================================================
 * Construct the Huffman tree for the bit lengths and return the index in
 * bl_order of the last bit length code to send.
 */
int build_bl_tree(TState &state)
{
    int max_blindex;  /* index of last bit length code of non zero freq */

    /* Determine the bit length frequencies for literal and distance trees */
    scan_tree(state,(ct_data *)state.ts.dyn_ltree, state.ts.l_desc.max_code);
    scan_tree(state,(ct_data *)state.ts.dyn_dtree, state.ts.d_desc.max_code);

    /* Build the bit length tree: */
    build_tree(state,(tree_desc *)(&state.ts.bl_desc));
    /* opt_len now includes the length of the tree representations, except
     * the lengths of the bit lengths codes and the 5+5+4 bits for the counts.
     */

    /* Determine the number of bit length codes to send. The pkzip format
     * requires that at least 4 bit length codes be sent. (appnote.txt says
     * 3 but the actual value used is 4.)
     */
    for (max_blindex = BL_CODES-1; max_blindex >= 3; max_blindex--) {
        if (state.ts.bl_tree[bl_order[max_blindex]].dl.len != 0) break;
    }
    /* Update opt_len to include the bit length tree and counts */
    state.ts.opt_len += 3*(max_blindex+1) + 5+5+4;
    Trace("\ndyn trees: dyn %ld, stat %ld", state.ts.opt_len, state.ts.static_len);

    return max_blindex;
}

/* ===========================================================================
 * Send the header for a block using dynamic Huffman trees: the counts, the
 * lengths of the bit length codes, the literal tree and the distance tree.
 * IN assertion: lcodes >= 257, dcodes >= 1, blcodes >= 4.
 */
void send_all_trees(TState &state,int lcodes, int dcodes, int blcodes)
{
    int rank;                    /* index in bl_order */

    Assert(state,lcodes >= 257 && dcodes >= 1 && blcodes >= 4, "not enough codes");
    Assert(state,lcodes <= L_CODES && dcodes <= D_CODES && blcodes <= BL_CODES,
            "too many codes");
    Trace("\nbl counts: ");
    send_bits(state,lcodes-257, 5);
    /* not +255 as stated in appnote.txt 1.93a or -256 in 2.04c */
    send_bits(state,dcodes-1,   5);
    send_bits(state,blcodes-4,  4); /* not -3 as stated in appnote.txt */
    for (rank = 0; rank < blcodes; rank++) {
        Trace("\nbl code %2d ", bl_order[rank]);
        send_bits(state,state.ts.bl_tree[bl_order[rank]].dl.len, 3);
    }    
    Trace("\nbl tree: sent %ld", state.bs.bits_sent);

    send_tree(state,(ct_data *)state.ts.dyn_ltree, lcodes-1); /* send the literal tree */
    Trace("\nlit tree: sent %ld", state.bs.bits_sent);

    send_tree(state,(ct_data *)state.ts.dyn_dtree, dcodes-1); /* send the distance tree */
    Trace("\ndist tree: sent %ld", state.bs.bits_sent);
}

/* ===========================================================================
 * Determine the best encoding for the current block: dynamic trees, static
 * trees or store, and output the encoded block to the zip file. This function
 * returns the total compressed length (in bytes) for the file so far.
 */
ulg flush_block(TState &state,char *buf, ulg stored_len, int eof)
{
    ulg opt_lenb, static_lenb; /* opt_len and static_len in bytes */
    int max_blindex;  /* index of last bit length code of non zero freq */

    state.ts.flag_buf[state.ts.last_flags] = state.ts.flags; /* Save the flags for the last 8 items */

     /* Check if the file is ascii or binary */
    if (*state.ts.file_type == (ush)UNKNOWN) set_file_type(state);

    /* Construct the literal and distance trees */
    build_tree(state,(tree_desc *)(&state.ts.l_desc));
    Trace("\nlit data: dyn %ld, stat %ld", state.ts.opt_len, state.ts.static_len);

    build_tree(state,(tree_desc *)(&state.ts.d_desc));
    Trace("\ndist data: dyn %ld, stat %ld", state.ts.opt_len, state.ts.static_len);
    /* At this point, opt_len and static_len are the total bit lengths of
     * the compressed block data, excluding the tree representations.
     */

    /* Build the bit length tree for the above two trees, and get the index
     * in bl_order of the last bit length code to send.
     */
    max_blindex = build_bl_tree(state);

    /* Determine the best encoding. Compute first the block length in bytes */
    opt_lenb = (state.ts.opt_len+3+7)>>3;
    static_lenb = (state.ts.static_len+3+7)>>3;
    state.ts.input_len += stored_len; /* for debugging only */

    Trace("\nopt %lu(%lu) stat %lu(%lu) stored %lu lit %u dist %u ",
            opt_lenb, state.ts.opt_len, static_lenb, state.ts.static_len, stored_len,
            state.ts.last_lit, state.ts.last_dist);

    if (static_lenb <= opt_lenb) opt_lenb = static_lenb;

    // Originally, zip allowed the file to be transformed from a compressed
    // into a stored file in the case where compression failed, there
    // was only one block, and it was allowed to change. I've removed this
    // possibility since the code's cleaner if no changes are allowed.
    //if (stored_len <= opt_lenb && eof && state.ts.cmpr_bytelen == 0L
    //   && state.ts.cmpr_len_bits == 0L && state.seekable)
    //{   // && state.ts.file_method != NULL
    //    // Since LIT_BUFSIZE <= 2*WSIZE, the input data must be there:
    //    Assert(state,buf!=NULL,"block vanished");
    //    copy_block(state,buf, (unsigned)stored_len, 0); // without header
    //    state.ts.cmpr_bytelen = stored_len;
    //    Assert(state,false,"unimplemented *state.ts.file_method = STORE;");
    //    //*state.ts.file_method = STORE;
    //}
    //else
    if (stored_len+4 <= opt_lenb && buf != (char*)NULL) {
                       /* 4: two words for the lengths */
        /* The test buf != NULL is only necessary if LIT_BUFSIZE > WSIZE.
         * Otherwise we can't have processed more than WSIZE input bytes since
         * the last block flush, because compression would have been
         * successful. If LIT_BUFSIZE <= WSIZE, it is never too late to
         * transform a block into a stored block.
         */
        send_bits(state,(STORED_BLOCK<<1)+eof, 3);  /* send block type */
        state.ts.cmpr_bytelen += ((state.ts.cmpr_len_bits + 3 + 7) >> 3) + stored_len + 4;
        state.ts.cmpr_len_bits = 0L;

        copy_block(state,buf, (unsigned)stored_len, 1); /* with header */
    }
    else if (static_lenb == opt_lenb) {
        send_bits(state,(STATIC_TREES<<1)+eof, 3);
        compress_block(state,(ct_data *)state.ts.static_ltree, (ct_data *)state.ts.static_dtree);
        state.ts.cmpr_len_bits += 3 + state.ts.static_len;
        state.ts.cmpr_bytelen += state.ts.cmpr_len_bits >> 3;
        state.ts.cmpr_len_bits &= 7L;
    }
    else {
        send_bits(state,(DYN_TREES<<1)+eof, 3);
        send_all_trees(state,state.ts.l_desc.max_code+1, state.ts.d_desc.max_code+1, max_blindex+1);
        compress_block(state,(ct_data *)state.ts.dyn_ltree, (ct_data *)state.ts.dyn_dtree);
        state.ts.cmpr_len_bits += 3 + state.ts.opt_len;
        state.ts.cmpr_bytelen += state.ts.cmpr_len_bits >> 3;
        state.ts.cmpr_len_bits &= 7L;
    }
    Assert(state,((state.ts.cmpr_bytelen << 3) + state.ts.cmpr_len_bits) == state.bs.bits_sent, "bad compressed size");
    init_block(state);

    if (eof) {
        // Assert(state,input_len == isize, "bad input size");
        bi_windup(state);
        state.ts.cmpr_len_bits += 7;  /* align on byte boundary */
    }
    Trace("\n");

    return state.ts.cmpr_bytelen + (state.ts.cmpr_len_bits >> 3);
}

/* ===========================================================================
 * Save the match info and tally the frequency counts. Return true if
 * the current block must be flushed.
 */
int ct_tally (TState &state,int dist, int lc)
{
    state.ts.l_buf[state.ts.last_lit++] = (uch)lc;
    if (dist == 0) {
        /* lc is the unmatched char */
        state.ts.dyn_ltree[lc].fc.freq++;
    } else {
        /* Here, lc is the match length - MIN_MATCH */
        dist--;             /* dist = match distance - 1 */
        Assert(state,(ush)dist < (ush)MAX_DIST &&
               (ush)lc <= (ush)(MAX_MATCH-MIN_MATCH) &&
               (ush)d_code(dist) < (ush)D_CODES,  "ct_tally: bad match");

        state.ts.dyn_ltree[state.ts.length_code[lc]+LITERALS+1].fc.freq++;
        state.ts.dyn_dtree[d_code(dist)].fc.freq++;

        state.ts.d_buf[state.ts.last_dist++] = (ush)dist;
        state.ts.flags |= state.ts.flag_bit;
    }
    state.ts.flag_bit <<= 1;

    /* Output the flags if they fill a byte: */
    if ((state.ts.last_lit & 7) == 0) {
        state.ts.flag_buf[state.ts.last_flags++] = state.ts.flags;
        state.ts.flags = 0, state.ts.flag_bit = 1;
    }
    /* Try to guess if it is profitable to stop the current block here */
    if (state.level > 2 && (state.ts.last_lit & 0xfff) == 0) {
        /* Compute an upper bound for the compressed length */
        ulg out_length = (ulg)state.ts.last_lit*8L;
        ulg in_length = (ulg)state.ds.strstart-state.ds.block_start;
        int dcode;
        for (dcode = 0; dcode < D_CODES; dcode++) {
            out_length += (ulg)state.ts.dyn_dtree[dcode].fc.freq*(5L+extra_dbits[dcode]);
        }
        out_length >>= 3;
        Trace("\nlast_lit %u, last_dist %u, in %ld, out ~%ld(%ld%%) ",
               state.ts.last_lit, state.ts.last_dist, in_length, out_length,