# --- T2-COPYRIGHT-NOTE-BEGIN --- # This copyright note is auto-generated by ./scripts/Create-CopyPatch. # # T2 SDE: package/.../gzip/gzip.rsync.patch # Copyright (C) 2006 The T2 SDE Project # # More information can be found in the files COPYING and README. # # This patch file is dual-licensed. It is available under the license the # patched project is licensed under, as long as it is an OpenSource license # as defined at http://www.opensource.org/ (e.g. BSD, X11) or under the terms # of the GNU General Public License as published by the Free Software # Foundation; either version 2 of the License, or (at your option) any later # version. # --- T2-COPYRIGHT-NOTE-END --- In the past from: http://ozlabs.org/~rusty/gzip.rsync.patch but rediffed for 1.3.5 by: - Rene Rebe diff -r db58977bb81e -r 6759e961b519 deflate.c --- a/deflate.c Sun Dec 03 11:40:05 2006 +0100 +++ b/deflate.c Sun Dec 03 11:48:30 2006 +0100 @@ -135,6 +135,14 @@ static char rcsid[] = "$Id: deflate.c,v #endif /* Matches of length 3 are discarded if their distance exceeds TOO_FAR */ +#ifndef RSYNC_WIN +# define RSYNC_WIN 8192 +#endif +/* Size of rsync window, must be < MAX_DIST */ + +#define RSYNC_SUM_MATCH(sum) (((sum) & (RSYNC_WIN - 1)) == 0) +/* Whether window sum matches magic value */ + /* =========================================================================== * Local data used by the "longest match" routines. */ @@ -216,6 +224,8 @@ unsigned near good_match; unsigned near good_match; /* Use a faster search when the previous match is longer than this */ +local ulg rsync_sum; /* rolling sum of rsync window */ +local ulg rsync_chunk_end; /* next rsync sequence point */ /* Values for max_lazy_match, good_match and max_chain_length, depending on * the desired pack level (0..9). The values given below have been tuned to @@ -314,6 +324,10 @@ void lm_init (pack_level, flags) #endif /* prev will be initialized on the fly */ + /* rsync params */ + rsync_chunk_end = 0xFFFFFFFFUL; + rsync_sum = 0; + /* Set the default configuration parameters: */ max_lazy_match = configuration_table[pack_level].max_lazy; @@ -331,6 +345,7 @@ void lm_init (pack_level, flags) strstart = 0; block_start = 0L; + rsync_chunk_end = 0xFFFFFFFFUL; #ifdef ASMV match_init(); /* initialize the asm code */ #endif @@ -550,6 +565,8 @@ local void fill_window() memcpy((char*)window, (char*)window+WSIZE, (unsigned)WSIZE); match_start -= WSIZE; strstart -= WSIZE; /* we now have strstart >= MAX_DIST: */ + if (rsync_chunk_end != 0xFFFFFFFFUL) + rsync_chunk_end -= WSIZE; block_start -= (long) WSIZE; @@ -576,6 +593,39 @@ local void fill_window() } } } + +local void rsync_roll(start, num) + unsigned start; + unsigned num; +{ + unsigned i; + + if (start < RSYNC_WIN) { + /* before window fills. */ + for (i = start; i < RSYNC_WIN; i++) { + if (i == start + num) return; + rsync_sum += (ulg)window[i]; + } + num -= (RSYNC_WIN - start); + start = RSYNC_WIN; + } + + /* buffer after window full */ + for (i = start; i < start+num; i++) { + /* New character in */ + rsync_sum += (ulg)window[i]; + /* Old character out */ + rsync_sum -= (ulg)window[i - RSYNC_WIN]; + if (rsync_chunk_end == 0xFFFFFFFFUL && RSYNC_SUM_MATCH(rsync_sum)) + rsync_chunk_end = i; + } +} + +/* =========================================================================== + * Set rsync_chunk_end if window sum matches magic value. + */ +#define RSYNC_ROLL(s, n) \ + do { if (rsync) rsync_roll((s), (n)); } while(0) /* =========================================================================== * Flush the current block, with given end-of-file flag. @@ -623,6 +673,7 @@ local off_t deflate_fast() lookahead -= match_length; + RSYNC_ROLL(strstart, match_length); /* Insert new strings in the hash table only if the match length * is not too large. This saves time but degrades compression. */ @@ -651,8 +702,17 @@ local off_t deflate_fast() /* No match, output a literal byte */ Tracevv((stderr,"%c",window[strstart])); flush = ct_tally (0, window[strstart]); + RSYNC_ROLL(strstart, 1); lookahead--; strstart++; + } + if (rsync && strstart > rsync_chunk_end) { + ush attr = 0; /* ascii/binary flag */ + + flush = 1; + /* Reset huffman tree */ + ct_init(&attr, &method); + rsync_chunk_end = 0xFFFFFFFFUL; } if (flush) FLUSH_BLOCK(0), block_start = strstart; @@ -726,6 +786,7 @@ off_t deflate() */ lookahead -= prev_length-1; prev_length -= 2; + RSYNC_ROLL(strstart, prev_length+1); do { strstart++; INSERT_STRING(strstart, hash_head); @@ -738,24 +799,51 @@ off_t deflate() match_available = 0; match_length = MIN_MATCH-1; strstart++; + + if (rsync && strstart > rsync_chunk_end) { + ush attr = 0; /* ascii/binary flag */ + + /* Reset huffman tree */ + ct_init(&attr, &method); + rsync_chunk_end = 0xFFFFFFFFUL; + flush = 1; + } if (flush) FLUSH_BLOCK(0), block_start = strstart; - } else if (match_available) { /* If there was no match at the previous position, output a * single literal. If there was a match but the current match * is longer, truncate the previous match to a single literal. */ Tracevv((stderr,"%c",window[strstart-1])); - if (ct_tally (0, window[strstart-1])) { - FLUSH_BLOCK(0), block_start = strstart; - } + flush = ct_tally (0, window[strstart-1]); + if (rsync && strstart > rsync_chunk_end) { + ush attr = 0; /* ascii/binary flag */ + + /* Reset huffman tree */ + ct_init(&attr, &method); + rsync_chunk_end = 0xFFFFFFFFUL; + + flush = 1; + } + if (flush) FLUSH_BLOCK(0), block_start = strstart; + RSYNC_ROLL(strstart, 1); strstart++; lookahead--; } else { /* There is no previous match to compare with, wait for * the next step to decide. */ + if (rsync && strstart > rsync_chunk_end) { + ush attr = 0; /* ascii/binary flag */ + + /* Reset huffman tree */ + ct_init(&attr, &method); + rsync_chunk_end = 0xFFFFFFFFUL; + + FLUSH_BLOCK(0), block_start = strstart; + } match_available = 1; + RSYNC_ROLL(strstart, 1); strstart++; lookahead--; } diff -r db58977bb81e -r 6759e961b519 doc/gzip.texi --- a/doc/gzip.texi Sun Dec 03 11:40:05 2006 +0100 +++ b/doc/gzip.texi Sun Dec 03 11:48:30 2006 +0100 @@ -328,6 +328,14 @@ into the directory and compress all the into the directory and compress all the files it finds there (or decompress them in the case of @code{gunzip}). +@item --rsyncable +While compressing, synchronize the output occasionally based on the +input. This can reduce the compression slightly in some cases, but +means that the @code{rsync} program can take advantage of similarities +in the uncompressed input when syncronizing two files compressed with +this flag. @code{gunzip} cannot tell the difference between a +compressed file created with this option, and one created without it. + @item --suffix @var{suf} @itemx -S @var{suf} Use suffix @samp{@var{suf}} instead of @samp{.gz}. Any suffix can be diff -r db58977bb81e -r 6759e961b519 gzip.c --- a/gzip.c Sun Dec 03 11:40:05 2006 +0100 +++ b/gzip.c Sun Dec 03 11:48:30 2006 +0100 @@ -167,7 +167,7 @@ DECLARE(uch, inbuf, INBUFSIZ +INBUF_EXT DECLARE(uch, inbuf, INBUFSIZ +INBUF_EXTRA); DECLARE(uch, outbuf, OUTBUFSIZ+OUTBUF_EXTRA); DECLARE(ush, d_buf, DIST_BUFSIZE); -DECLARE(uch, window, 2L*WSIZE); +DECLARE(uch, window, 2L*WSIZE + 4096); #ifndef MAXSEG_64K DECLARE(ush, tab_prefix, 1L<