diff -u -r gzip-1.5/deflate.c /home/bdale/debian/gzip/deflate.c --- gzip-1.5/deflate.c 2012-04-24 10:25:28.000000000 -0600 +++ /home/bdale/debian/gzip/deflate.c 2012-06-19 14:01:41.885498670 -0600 @@ -131,6 +131,14 @@ #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. */ @@ -212,6 +220,8 @@ unsigned 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 @@ #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 @@ strstart = 0; block_start = 0L; + rsync_chunk_end = 0xFFFFFFFFUL; #ifdef ASMV match_init(); /* initialize the asm code */ #endif @@ -550,6 +565,8 @@ 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; @@ -579,6 +596,39 @@ } } +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. * IN assertion: strstart is set to the end of the current match. @@ -626,7 +676,8 @@ lookahead -= match_length; - /* Insert new strings in the hash table only if the 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. */ if (match_length <= max_insert_length) { @@ -654,9 +705,18 @@ /* 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; /* Make sure that we always have enough lookahead, except @@ -730,6 +790,7 @@ */ lookahead -= prev_length-1; prev_length -= 2; + RSYNC_ROLL(strstart, prev_length+1); do { strstart++; INSERT_STRING(strstart, hash_head); @@ -742,24 +803,51 @@ match_available = 0; match_length = MIN_MATCH-1; strstart++; - if (flush) FLUSH_BLOCK(0), block_start = 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 -u -r gzip-1.5/doc/gzip.texi /home/bdale/debian/gzip/doc/gzip.texi --- gzip-1.5/doc/gzip.texi 2012-01-01 01:53:58.000000000 -0700 +++ /home/bdale/debian/gzip/doc/gzip.texi 2012-06-19 14:09:36.041559612 -0600 @@ -207,6 +207,7 @@ -V, --version display version number -1, --fast compress faster -9, --best compress better + --rsyncable Make rsync-friendly archive With no FILE, or when FILE is -, read standard input. @@ -353,6 +354,30 @@ into the directory and compress all the files it finds there (or decompress them in the case of @command{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 --rsyncable +While compressing, synchronize the output occasionally based on +the input. This increases size by less than 1 percent most +cases, but means that the @command{rsync} program can much more efficiently +synchronize files compressed with this flag. @command{gunzip} +cannot tell the difference between a compressed file created +with this option, and one created without it. + +@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 @var{suf} instead of @samp{.gz}. Any suffix can be --- gzip-1.5/gzip.c 2012-04-24 10:25:28.000000000 -0600 +++ /home/bdale/debian/gzip/gzip.c 2012-06-19 14:00:43.283873180 -0600 @@ -213,6 +213,7 @@ unsigned insize; /* valid bytes in inbuf */ unsigned inptr; /* index of next byte to be processed in inbuf */ unsigned outcnt; /* bytes in output buffer */ +int rsync = 0; /* make ryncable chunks */ static int handled_sig[] = { @@ -270,7 +271,7 @@ {"best", 0, 0, '9'}, /* compress better */ {"lzw", 0, 0, 'Z'}, /* make output compatible with old compress */ {"bits", 1, 0, 'b'}, /* max number of bits per code (implies -Z) */ - + {"rsyncable", 0, 0, 'R'}, /* make rsync-friendly archive */ { 0, 0, 0, 0 } }; @@ -353,6 +354,7 @@ " -Z, --lzw produce output compatible with old compress", " -b, --bits=BITS max number of bits per code (implies -Z)", #endif + " --rsyncable Make rsync-friendly archive", "", "With no FILE, or when FILE is -, read standard input.", "", @@ -481,8 +483,11 @@ #else recursive = 1; #endif - break; - case 'S': + break; + case 'R': + rsync = 1; break; + + case 'S': #ifdef NO_MULTIPLE_DOTS if (*optarg == '.') optarg++; #endif diff -u -r gzip-1.5/gzip.h /home/bdale/debian/gzip/gzip.h --- gzip-1.5/gzip.h 2012-01-01 01:53:58.000000000 -0700 +++ /home/bdale/debian/gzip/gzip.h 2012-06-19 13:57:36.833883093 -0600 @@ -140,6 +140,7 @@ extern unsigned insize; /* valid bytes in inbuf */ extern unsigned inptr; /* index of next byte to be processed in inbuf */ extern unsigned outcnt; /* bytes in output buffer */ +extern int rsync; /* deflate into rsyncable chunks */ extern off_t bytes_in; /* number of input bytes */ extern off_t bytes_out; /* number of output bytes */