Quellcode-Bibliothek stored.rs

use crate::deflate::buffer::{update_hash, LZ_HASH_SHIFT, LZ_HASH_SIZE};
use crate::deflate::core::{
    flush_block, CallbackOxide, CompressorOxide, TDEFLFlush, TDEFLStatus, LZ_DICT_SIZE,
    LZ_DICT_SIZE_MASK, MAX_MATCH_LEN, MIN_MATCH_LEN,
};
use core::cmp;

pub(crate) fn compress_stored(d: &mut CompressorOxide, callback: &mut CallbackOxide) -> bool {
    let in_buf = match callback.buf() {
        None => return true,
        Some(in_buf) => in_buf,
    };

    // Make sure this is cleared in case compression level is switched later.
    // TODO: It's possible we don't need this or could do this elsewhere later
    // but just do this here to avoid causing issues for now.
    d.params.saved_match_len = 0;
    let mut bytes_written = d.lz.total_bytes;
    let mut src_pos = d.params.src_pos;
    let mut lookahead_size = d.dict.lookahead_size;
    let mut lookahead_pos = d.dict.lookahead_pos;

    while src_pos < in_buf.len() || (d.params.flush != TDEFLFlush::None && lookahead_size != 0) {
        let src_buf_left = in_buf.len() - src_pos;
        let num_bytes_to_process = cmp::min(src_buf_left, MAX_MATCH_LEN - lookahead_size);

        if lookahead_size + d.dict.size >= usize::from(MIN_MATCH_LEN) - 1
            && num_bytes_to_process > 0
        {
            let dictb = &mut d.dict.b;

            let mut dst_pos = (lookahead_pos + lookahead_size) & LZ_DICT_SIZE_MASK;
            let mut ins_pos = lookahead_pos + lookahead_size - 2;
            // Start the hash value from the first two bytes
            let mut hash = update_hash(
                u16::from(dictb.dict[ins_pos & LZ_DICT_SIZE_MASK]),
                dictb.dict[(ins_pos + 1) & LZ_DICT_SIZE_MASK],
            );

            lookahead_size += num_bytes_to_process;

            for &c in &in_buf[src_pos..src_pos + num_bytes_to_process] {
                // Add byte to input buffer.
                dictb.dict[dst_pos] = c;
                if dst_pos < MAX_MATCH_LEN - 1 {
                    dictb.dict[LZ_DICT_SIZE + dst_pos] = c;
                }

                // Generate hash from the current byte,
                hash = update_hash(hash, c);
                dictb.next[ins_pos & LZ_DICT_SIZE_MASK] = dictb.hash[hash as usize];
                // and insert it into the hash chain.
                dictb.hash[hash as usize] = ins_pos as u16;
                dst_pos = (dst_pos + 1) & LZ_DICT_SIZE_MASK;
                ins_pos += 1;
            }
            src_pos += num_bytes_to_process;
        } else {
            let dictb = &mut d.dict.b;
            for &c in &in_buf[src_pos..src_pos + num_bytes_to_process] {
                let dst_pos = (lookahead_pos + lookahead_size) & LZ_DICT_SIZE_MASK;
                dictb.dict[dst_pos] = c;
                if dst_pos < MAX_MATCH_LEN - 1 {
                    dictb.dict[LZ_DICT_SIZE + dst_pos] = c;
                }

                lookahead_size += 1;
                if lookahead_size + d.dict.size >= MIN_MATCH_LEN.into() {
                    let ins_pos = lookahead_pos + lookahead_size - 3;
                    let hash = ((u32::from(dictb.dict[ins_pos & LZ_DICT_SIZE_MASK])
                        << (LZ_HASH_SHIFT * 2))
                        ^ ((u32::from(dictb.dict[(ins_pos + 1) & LZ_DICT_SIZE_MASK])
                            << LZ_HASH_SHIFT)
                            ^ u32::from(c)))
                        & (LZ_HASH_SIZE as u32 - 1);

                    dictb.next[ins_pos & LZ_DICT_SIZE_MASK] = dictb.hash[hash as usize];
                    dictb.hash[hash as usize] = ins_pos as u16;
                }
            }

            src_pos += num_bytes_to_process;
        }

        d.dict.size = cmp::min(LZ_DICT_SIZE - lookahead_size, d.dict.size);
        if d.params.flush == TDEFLFlush::None && lookahead_size < MAX_MATCH_LEN {
            break;
        }

        let len_to_move = 1;

        bytes_written += 1;

        lookahead_pos += len_to_move;
        assert!(lookahead_size >= len_to_move);
        lookahead_size -= len_to_move;
        d.dict.size = cmp::min(d.dict.size + len_to_move, LZ_DICT_SIZE);

        if bytes_written > 31 * 1024 {
            d.lz.total_bytes = bytes_written;

            d.params.src_pos = src_pos;
            // These values are used in flush_block, so we need to write them back here.
            d.dict.lookahead_size = lookahead_size;
            d.dict.lookahead_pos = lookahead_pos;

            let n = flush_block(d, callback, TDEFLFlush::None)
                .unwrap_or(TDEFLStatus::PutBufFailed as i32);
            if n != 0 {
                return n > 0;
            }
            bytes_written = d.lz.total_bytes;
        }
    }

    d.lz.total_bytes = bytes_written;
    d.params.src_pos = src_pos;
    d.dict.lookahead_size = lookahead_size;
    d.dict.lookahead_pos = lookahead_pos;
    true
}

/*
fn compress_rle(d: &mut CompressorOxide, callback: &mut CallbackOxide) -> bool {
    let mut src_pos = d.params.src_pos;
    let in_buf = match callback.in_buf {
        None => return true,
        Some(in_buf) => in_buf,
    };

    let mut lookahead_size = d.dict.lookahead_size;
    let mut lookahead_pos = d.dict.lookahead_pos;
    let mut saved_lit = d.params.saved_lit;
    let mut saved_match_dist = d.params.saved_match_dist;
    let mut saved_match_len = d.params.saved_match_len;

    while src_pos < in_buf.len() || (d.params.flush != TDEFLFlush::None && lookahead_size != 0) {
        let src_buf_left = in_buf.len() - src_pos;
        let num_bytes_to_process = cmp::min(src_buf_left, MAX_MATCH_LEN - lookahead_size);

        if lookahead_size + d.dict.size >= usize::from(MIN_MATCH_LEN) - 1
            && num_bytes_to_process > 0
        {
            let dictb = &mut d.dict.b;

            let mut dst_pos = (lookahead_pos + lookahead_size) & LZ_DICT_SIZE_MASK;
            let mut ins_pos = lookahead_pos + lookahead_size - 2;
            // Start the hash value from the first two bytes
            let mut hash = update_hash(
                u16::from(dictb.dict[ins_pos & LZ_DICT_SIZE_MASK]),
                dictb.dict[(ins_pos + 1) & LZ_DICT_SIZE_MASK],
            );

            lookahead_size += num_bytes_to_process;

            for &c in &in_buf[src_pos..src_pos + num_bytes_to_process] {
                // Add byte to input buffer.
                dictb.dict[dst_pos] = c;
                if dst_pos < MAX_MATCH_LEN - 1 {
                    dictb.dict[LZ_DICT_SIZE + dst_pos] = c;
                }

                // Generate hash from the current byte,
                hash = update_hash(hash, c);
                dictb.next[ins_pos & LZ_DICT_SIZE_MASK] = dictb.hash[hash as usize];
                // and insert it into the hash chain.
                dictb.hash[hash as usize] = ins_pos as u16;
                dst_pos = (dst_pos + 1) & LZ_DICT_SIZE_MASK;
                ins_pos += 1;
            }
            src_pos += num_bytes_to_process;
        } else {
            let dictb = &mut d.dict.b;
            for &c in &in_buf[src_pos..src_pos + num_bytes_to_process] {
                let dst_pos = (lookahead_pos + lookahead_size) & LZ_DICT_SIZE_MASK;
                dictb.dict[dst_pos] = c;
                if dst_pos < MAX_MATCH_LEN - 1 {
                    dictb.dict[LZ_DICT_SIZE + dst_pos] = c;
                }

                lookahead_size += 1;
                if lookahead_size + d.dict.size >= MIN_MATCH_LEN.into() {
                    let ins_pos = lookahead_pos + lookahead_size - 3;
                    let hash = ((u32::from(dictb.dict[ins_pos & LZ_DICT_SIZE_MASK])
                        << (LZ_HASH_SHIFT * 2))
                        ^ ((u32::from(dictb.dict[(ins_pos + 1) & LZ_DICT_SIZE_MASK])
                            << LZ_HASH_SHIFT)
                            ^ u32::from(c)))
                        & (LZ_HASH_SIZE as u32 - 1);

                    dictb.next[ins_pos & LZ_DICT_SIZE_MASK] = dictb.hash[hash as usize];
                    dictb.hash[hash as usize] = ins_pos as u16;
                }
            }

            src_pos += num_bytes_to_process;
        }

        d.dict.size = cmp::min(LZ_DICT_SIZE - lookahead_size, d.dict.size);
        if d.params.flush == TDEFLFlush::None && lookahead_size < MAX_MATCH_LEN {
            break;
        }

        let mut len_to_move = 1;
        let mut cur_match_dist = 0;
        let mut cur_match_len = if saved_match_len != 0 {
            saved_match_len
        } else {
            u32::from(MIN_MATCH_LEN) - 1
        };
        let cur_pos = lookahead_pos & LZ_DICT_SIZE_MASK;
                // If TDEFL_RLE_MATCHES is set, we only look for repeating sequences of the current byte.
        if d.dict.size != 0 && d.params.flags & TDEFL_FORCE_ALL_RAW_BLOCKS == 0 {
            let c = d.dict.b.dict[(cur_pos.wrapping_sub(1)) & LZ_DICT_SIZE_MASK];
                    cur_match_len = d.dict.b.dict[cur_pos..(cur_pos + lookahead_size)]
                        .iter()
                        .take_while(|&x| *x == c)
                        .count() as u32;
                    if cur_match_len < MIN_MATCH_LEN.into() {
                        cur_match_len = 0
                    } else {
                        cur_match_dist = 1
                    }
                }


        let far_and_small = cur_match_len == MIN_MATCH_LEN.into() && cur_match_dist >= 8 * 1024;
        let filter_small = d.params.flags & TDEFL_FILTER_MATCHES != 0 && cur_match_len <= 5;
        if far_and_small || filter_small || cur_pos == cur_match_dist as usize {
            cur_match_dist = 0;
            cur_match_len = 0;
        }

        if saved_match_len != 0 {
            if cur_match_len > saved_match_len {
                record_literal(&mut d.huff, &mut d.lz, saved_lit);
                if cur_match_len >= 128 {
                    record_match(&mut d.huff, &mut d.lz, cur_match_len, cur_match_dist);
                    saved_match_len = 0;
                    len_to_move = cur_match_len as usize;
                } else {
                    saved_lit = d.dict.b.dict[cur_pos];
                    saved_match_dist = cur_match_dist;
                    saved_match_len = cur_match_len;
                }
            } else {
                record_match(&mut d.huff, &mut d.lz, saved_match_len, saved_match_dist);
                len_to_move = (saved_match_len - 1) as usize;
                saved_match_len = 0;
            }
        } else if cur_match_dist == 0 {
            record_literal(
                &mut d.huff,
                &mut d.lz,
                d.dict.b.dict[cmp::min(cur_pos, d.dict.b.dict.len() - 1)],
            );
        } else if d.params.greedy_parsing
            || (d.params.flags & TDEFL_RLE_MATCHES != 0)
            || cur_match_len >= 128
        {
            // If we are using lazy matching, check for matches at the next byte if the current
            // match was shorter than 128 bytes.
            record_match(&mut d.huff, &mut d.lz, cur_match_len, cur_match_dist);
            len_to_move = cur_match_len as usize;
        } else {
            saved_lit = d.dict.b.dict[cmp::min(cur_pos, d.dict.b.dict.len() - 1)];
            saved_match_dist = cur_match_dist;
            saved_match_len = cur_match_len;
        }

        lookahead_pos += len_to_move;
        assert!(lookahead_size >= len_to_move);
        lookahead_size -= len_to_move;
        d.dict.size = cmp::min(d.dict.size + len_to_move, LZ_DICT_SIZE);

        let lz_buf_tight = d.lz.code_position > LZ_CODE_BUF_SIZE - 8;
        let raw = d.params.flags & TDEFL_FORCE_ALL_RAW_BLOCKS != 0;
        let fat = ((d.lz.code_position * 115) >> 7) >= d.lz.total_bytes as usize;
        let fat_or_raw = (d.lz.total_bytes > 31 * 1024) && (fat || raw);

        if lz_buf_tight || fat_or_raw {
            d.params.src_pos = src_pos;
            // These values are used in flush_block, so we need to write them back here.
            d.dict.lookahead_size = lookahead_size;
            d.dict.lookahead_pos = lookahead_pos;

            let n = flush_block(d, callback, TDEFLFlush::None)
                .unwrap_or(TDEFLStatus::PutBufFailed as i32);
            if n != 0 {
                d.params.saved_lit = saved_lit;
                d.params.saved_match_dist = saved_match_dist;
                d.params.saved_match_len = saved_match_len;
                return n > 0;
            }
        }
    }

    d.params.src_pos = src_pos;
    d.dict.lookahead_size = lookahead_size;
    d.dict.lookahead_pos = lookahead_pos;
    d.params.saved_lit = saved_lit;
    d.params.saved_match_dist = saved_match_dist;
    d.params.saved_match_len = saved_match_len;
    true
}*/