Merging upstream version 1.6~pre3.

Signed-off-by: Daniel Baumann <daniel@debian.org>

ChangeLog

@@ -1,3 +1,9 @@
+2014-03-30  Antonio Diaz Diaz  <antonio@gnu.org>
+
+	* Version 1.6-pre3 released.
+	* Compression ratio has been slightly increased.
+	* configure: Added new option '--disable-static'.
+
 2014-01-30  Antonio Diaz Diaz  <antonio@gnu.org>
 
 	* Version 1.6-pre2 released.

Makefile.in

@@ -15,7 +15,7 @@ objs = carg_parser.o main.o
          install-as-lzip uninstall uninstall-bin uninstall-info uninstall-man \
          doc info man check dist clean distclean
 
-all : $(progname) $(progname_shared)
+all : $(progname_static) $(progname_shared)
 
 lib$(libname).a : lzlib.o
 	$(AR) -rcs $@ $<
@@ -73,7 +73,7 @@ $(VPATH)/doc/$(progname).1 : $(progname)
 Makefile : $(VPATH)/configure $(VPATH)/Makefile.in
 	./config.status
 
-check : all bbexample lzcheck
+check : $(progname) bbexample lzcheck
 	@$(VPATH)/testsuite/check.sh $(VPATH)/testsuite $(pkgversion)
 
 install : install-bin install-info
@@ -82,17 +82,17 @@ install-bin : all
 	if [ ! -d "$(DESTDIR)$(includedir)" ] ; then $(INSTALL_DIR) "$(DESTDIR)$(includedir)" ; fi
 	if [ ! -d "$(DESTDIR)$(libdir)" ] ; then $(INSTALL_DIR) "$(DESTDIR)$(libdir)" ; fi
 	$(INSTALL_DATA) $(VPATH)/$(libname)lib.h "$(DESTDIR)$(includedir)/$(libname)lib.h"
-	$(INSTALL_DATA) ./lib$(libname).a "$(DESTDIR)$(libdir)/lib$(libname).a"
+	if [ -n "$(progname_static)" ] ; then \
+	  $(INSTALL_DATA) ./lib$(libname).a "$(DESTDIR)$(libdir)/lib$(libname).a" ; \
+	fi
 	if [ -n "$(progname_shared)" ] ; then \
 	  $(INSTALL_PROGRAM) ./lib$(libname).so.$(pkgversion) "$(DESTDIR)$(libdir)/lib$(libname).so.$(pkgversion)" ; \
 	  if [ -e "$(DESTDIR)$(libdir)/lib$(libname).so.$(soversion)" ] ; then \
-	    rm -f "$(DESTDIR)$(libdir)/lib$(libname).so.$(soversion)" ; \
 	    run_ldconfig=no ; \
 	  else run_ldconfig=yes ; \
 	  fi ; \
-	  if [ -e "$(DESTDIR)$(libdir)/lib$(libname).so" ] ; then \
+	  rm -f "$(DESTDIR)$(libdir)/lib$(libname).so" ; \
-	    rm -f "$(DESTDIR)$(libdir)/lib$(libname).so" ; \
+	  rm -f "$(DESTDIR)$(libdir)/lib$(libname).so.$(soversion)" ; \
-	  fi ; \
 	  cd "$(DESTDIR)$(libdir)" && ln -s lib$(libname).so.$(pkgversion) lib$(libname).so ; \
 	  cd "$(DESTDIR)$(libdir)" && ln -s lib$(libname).so.$(pkgversion) lib$(libname).so.$(soversion) ; \
 	  if [ "${disable_ldconfig}" != yes ] && [ $${run_ldconfig} = yes ] && \
@@ -113,7 +113,7 @@ install-strip : all
 
 install-as-lzip : install install-man
 	if [ ! -d "$(DESTDIR)$(bindir)" ] ; then $(INSTALL_DIR) "$(DESTDIR)$(bindir)" ; fi
-	$(INSTALL_PROGRAM) ./$(progname) "$(DESTDIR)$(bindir)/$(progname)"
+	$(INSTALL_PROGRAM) ./$(progname_lzip) "$(DESTDIR)$(bindir)/$(progname)"
 	-rm -f "$(DESTDIR)$(bindir)/lzip"
 	cd "$(DESTDIR)$(bindir)" && ln -s $(progname) lzip
 

NEWS

@@ -1,5 +1,9 @@
 Changes in version 1.6:
 
+Compression ratio has been slightly increased.
+
+The configure script now accepts the option "--disable-static".
+
 Improved portability to BSD systems:
 
     The configure script now accepts the option "--disable-ldconfig".

README

@@ -5,9 +5,24 @@ and decompression functions, including integrity checking of the
 decompressed data. The compressed data format used by the library is the
 lzip format. Lzlib is written in C.
 
-The lzip file format is designed for long-term data archiving. It is
+The lzip file format is designed for long-term data archiving, taking
-clean, provides very safe 4 factor integrity checking, and is backed by
+into account both data integrity and decoder availability:
-the recovery capabilities of lziprecover.
+
+   * The lzip format provides very safe integrity checking and some data
+     recovery means. The lziprecover program can repair bit-flip errors
+     (one of the most common forms of data corruption) in lzip files,
+     and provides data recovery capabilities, including error-checked
+     merging of damaged copies of a file.
+
+   * The lzip format is as simple as possible (but not simpler). The
+     lzip manual provides the code of a simple decompressor along with a
+     detailed explanation of how it works, so that with the only help of
+     the lzip manual it would be possible for a digital archaeologist to
+     extract the data from a lzip file long after quantum computers
+     eventually render LZMA obsolete.
+
+   * Additionally lzip is copylefted, which guarantees that it will
+     remain free forever.
 
 The functions and variables forming the interface of the compression
 library are declared in the file 'lzlib.h'. Usage examples of the
@@ -34,13 +49,23 @@ All the library functions are thread safe. The library does not install
 any signal handler. The decoder checks the consistency of the compressed
 data, so the library should never crash even in case of corrupted input.
 
-Lzlib implements a simplified version of the LZMA (Lempel-Ziv-Markov
+There is no such thing as a "LZMA algorithm"; it is more like a "LZMA
-chain-Algorithm) algorithm. The high compression of LZMA comes from
+coding scheme". For example, the option '-0' of lzip uses the scheme in
-combining two basic, well-proven compression ideas: sliding dictionaries
+almost the simplest way possible; issuing the longest match it can find,
-(LZ77/78) and markov models (the thing used by every compression
+or a literal byte if it can't find a match. Conversely, a much more
-algorithm that uses a range encoder or similar order-0 entropy coder as
+elaborated way of finding coding sequences of minimum price than the one
-its last stage) with segregation of contexts according to what the bits
+currently used by lzip could be developed, and the resulting sequence
-are used for.
+could also be coded using the LZMA coding scheme.
+
+Lzip currently implements two variants of the LZMA algorithm; fast (used
+by option -0) and normal (used by all other compression levels). Lzlib
+just implements the "normal" variant.
+
+The high compression of LZMA comes from combining two basic, well-proven
+compression ideas: sliding dictionaries (LZ77/78) and markov models (the
+thing used by every compression algorithm that uses a range encoder or
+similar order-0 entropy coder as its last stage) with segregation of
+contexts according to what the bits are used for.
 
 The ideas embodied in lzlib are due to (at least) the following people:
 Abraham Lempel and Jacob Ziv (for the LZ algorithm), Andrey Markov (for

bbexample.c

@@ -34,7 +34,7 @@ uint8_t * bbcompress( const uint8_t * const data, const int size,
   struct LZ_Encoder * encoder;
   uint8_t * new_data;
   const int match_len_limit = 36;
-  const unsigned long long member_size = INT64_MAX;
+  const unsigned long long member_size = 0x7FFFFFFFFFFFFFFFULL;	/* INT64_MAX */
   int delta_size, new_data_size;
   int new_pos = 0;
   int written = 0;

configure

@@ -6,10 +6,12 @@
 # to copy, distribute and modify it.
 
 pkgname=lzlib
-pkgversion=1.6-pre2
+pkgversion=1.6-pre3
 soversion=1
 progname=minilzip
+progname_static=${progname}
 progname_shared=
+progname_lzip=${progname}
 disable_ldconfig=
 libname=lz
 srctrigger=doc/${pkgname}.texi
@@ -72,6 +74,8 @@ while [ $# != 0 ] ; do
 		echo "  --infodir=DIR         info files directory [${infodir}]"
 		echo "  --libdir=DIR          object code libraries [${libdir}]"
 		echo "  --mandir=DIR          man pages directory [${mandir}]"
+		echo "  --disable-static      do not build a static library [enable]"
+		echo "                        (implies --enable-shared)"
 		echo "  --enable-shared       build also a shared library [disable]"
 		echo "  --disable-ldconfig    do not run ldconfig after install"
 		echo "  CC=COMPILER           C compiler to use [gcc]"
@@ -103,6 +107,10 @@ while [ $# != 0 ] ; do
 	--libdir=*)            libdir=${optarg} ;;
 	--mandir=*)            mandir=${optarg} ;;
 	--no-create)              no_create=yes ;;
+	--disable-static)
+		progname_static=
+		progname_shared=${progname}_shared
+		progname_lzip=${progname}_shared ;;
 	--enable-shared) progname_shared=${progname}_shared ;;
 	--disable-ldconfig) disable_ldconfig=yes ;;
 
@@ -193,7 +201,9 @@ pkgname = ${pkgname}
 pkgversion = ${pkgversion}
 soversion = ${soversion}
 progname = ${progname}
+progname_static = ${progname_static}
 progname_shared = ${progname_shared}
+progname_lzip = ${progname_lzip}
 disable_ldconfig = ${disable_ldconfig}
 libname = ${libname}
 VPATH = ${srcdir}

doc/lzlib.info

@@ -11,7 +11,7 @@ File: lzlib.info,  Node: Top,  Next: Introduction,  Up: (dir)
 Lzlib Manual
 ************
 
-This manual is for Lzlib (version 1.6-pre2, 30 January 2014).
+This manual is for Lzlib (version 1.6-pre3, 30 March 2014).
 
 * Menu:
 
@@ -45,9 +45,24 @@ and decompression functions, including integrity checking of the
 decompressed data. The compressed data format used by the library is the
 lzip format. Lzlib is written in C.
 
-   The lzip file format is designed for long-term data archiving. It is
+   The lzip file format is designed for long-term data archiving, taking
-clean, provides very safe 4 factor integrity checking, and is backed by
+into account both data integrity and decoder availability:
-the recovery capabilities of lziprecover.
+
+   * The lzip format provides very safe integrity checking and some data
+     recovery means. The lziprecover program can repair bit-flip errors
+     (one of the most common forms of data corruption) in lzip files,
+     and provides data recovery capabilities, including error-checked
+     merging of damaged copies of a file.
+
+   * The lzip format is as simple as possible (but not simpler). The
+     lzip manual provides the code of a simple decompressor along with
+     a detailed explanation of how it works, so that with the only help
+     of the lzip manual it would be possible for a digital
+     archaeologist to extract the data from a lzip file long after
+     quantum computers eventually render LZMA obsolete.
+
+   * Additionally lzip is copylefted, which guarantees that it will
+     remain free forever.
 
    The functions and variables forming the interface of the compression
 library are declared in the file 'lzlib.h'. Usage examples of the
@@ -75,13 +90,23 @@ install any signal handler. The decoder checks the consistency of the
 compressed data, so the library should never crash even in case of
 corrupted input.
 
-   Lzlib implements a simplified version of the LZMA (Lempel-Ziv-Markov
+   There is no such thing as a "LZMA algorithm"; it is more like a "LZMA
-chain-Algorithm) algorithm. The high compression of LZMA comes from
+coding scheme". For example, the option '-0' of lzip uses the scheme in
-combining two basic, well-proven compression ideas: sliding dictionaries
+almost the simplest way possible; issuing the longest match it can find,
-(LZ77/78) and markov models (the thing used by every compression
+or a literal byte if it can't find a match. Conversely, a much more
-algorithm that uses a range encoder or similar order-0 entropy coder as
+elaborated way of finding coding sequences of minimum price than the one
-its last stage) with segregation of contexts according to what the bits
+currently used by lzip could be developed, and the resulting sequence
-are used for.
+could also be coded using the LZMA coding scheme.
+
+   Lzip currently implements two variants of the LZMA algorithm; fast
+(used by option -0) and normal (used by all other compression levels).
+Lzlib just implements the "normal" variant.
+
+   The high compression of LZMA comes from combining two basic,
+well-proven compression ideas: sliding dictionaries (LZ77/78) and
+markov models (the thing used by every compression algorithm that uses
+a range encoder or similar order-0 entropy coder as its last stage)
+with segregation of contexts according to what the bits are used for.
 
    The ideas embodied in lzlib are due to (at least) the following
 people: Abraham Lempel and Jacob Ziv (for the LZ algorithm), Andrey
@@ -736,18 +761,18 @@ Concept index
 
 Tag Table:
 Node: Top220
-Node: Introduction1335
+Node: Introduction1333
-Node: Library version3916
+Node: Library version5247
-Node: Buffering4561
+Node: Buffering5892
-Node: Parameter limits5684
+Node: Parameter limits7015
-Node: Compression functions6643
+Node: Compression functions7974
-Node: Decompression functions12856
+Node: Decompression functions14187
-Node: Error codes19017
+Node: Error codes20348
-Node: Error messages20956
+Node: Error messages22287
-Node: Data format21535
+Node: Data format22866
-Node: Examples24184
+Node: Examples25515
-Node: Problems28267
+Node: Problems29598
-Node: Concept index28839
+Node: Concept index30170
 
 End Tag Table
 

doc/lzlib.texi

@@ -6,8 +6,8 @@
 @finalout
 @c %**end of header
 
-@set UPDATED 30 January 2014
+@set UPDATED 30 March 2014
-@set VERSION 1.6-pre2
+@set VERSION 1.6-pre3
 
 @dircategory Data Compression
 @direntry
@@ -66,9 +66,29 @@ and decompression functions, including integrity checking of the
 decompressed data. The compressed data format used by the library is the
 lzip format. Lzlib is written in C.
 
-The lzip file format is designed for long-term data archiving. It is
+The lzip file format is designed for long-term data archiving, taking
-clean, provides very safe 4 factor integrity checking, and is backed by
+into account both data integrity and decoder availability:
-the recovery capabilities of lziprecover.
+
+@itemize @bullet
+@item
+The lzip format provides very safe integrity checking and some data
+recovery means. The lziprecover program can repair bit-flip errors (one
+of the most common forms of data corruption) in lzip files, and provides
+data recovery capabilities, including error-checked merging of damaged
+copies of a file.
+
+@item
+The lzip format is as simple as possible (but not simpler). The lzip
+manual provides the code of a simple decompressor along with a detailed
+explanation of how it works, so that with the only help of the lzip
+manual it would be possible for a digital archaeologist to extract the
+data from a lzip file long after quantum computers eventually render
+LZMA obsolete.
+
+@item
+Additionally lzip is copylefted, which guarantees that it will remain
+free forever.
+@end itemize
 
 The functions and variables forming the interface of the compression
 library are declared in the file @samp{lzlib.h}. Usage examples of the
@@ -95,13 +115,23 @@ All the library functions are thread safe. The library does not install
 any signal handler. The decoder checks the consistency of the compressed
 data, so the library should never crash even in case of corrupted input.
 
-Lzlib implements a simplified version of the LZMA (Lempel-Ziv-Markov
+There is no such thing as a "LZMA algorithm"; it is more like a "LZMA
-chain-Algorithm) algorithm. The high compression of LZMA comes from
+coding scheme". For example, the option '-0' of lzip uses the scheme in
-combining two basic, well-proven compression ideas: sliding dictionaries
+almost the simplest way possible; issuing the longest match it can find,
-(LZ77/78) and markov models (the thing used by every compression
+or a literal byte if it can't find a match. Conversely, a much more
-algorithm that uses a range encoder or similar order-0 entropy coder as
+elaborated way of finding coding sequences of minimum price than the one
-its last stage) with segregation of contexts according to what the bits
+currently used by lzip could be developed, and the resulting sequence
-are used for.
+could also be coded using the LZMA coding scheme.
+
+Lzip currently implements two variants of the LZMA algorithm; fast (used
+by option -0) and normal (used by all other compression levels). Lzlib
+just implements the "normal" variant.
+
+The high compression of LZMA comes from combining two basic, well-proven
+compression ideas: sliding dictionaries (LZ77/78) and markov models (the
+thing used by every compression algorithm that uses a range encoder or
+similar order-0 entropy coder as its last stage) with segregation of
+contexts according to what the bits are used for.
 
 The ideas embodied in lzlib are due to (at least) the following people:
 Abraham Lempel and Jacob Ziv (for the LZ algorithm), Andrey Markov (for

doc/minilzip.1

@@ -1,7 +1,7 @@
 .\" DO NOT MODIFY THIS FILE!  It was generated by help2man 1.37.1.
-.TH MINILZIP "1" "January 2014" "Minilzip 1.6-pre2" "User Commands"
+.TH MINILZIP "1" "March 2014" "minilzip 1.6-pre3" "User Commands"
 .SH NAME
-Minilzip \- reduces the size of files
+minilzip \- reduces the size of files
 .SH SYNOPSIS
 .B minilzip
 [\fIoptions\fR] [\fIfiles\fR]
@@ -83,7 +83,7 @@ Report bugs to lzip\-bug@nongnu.org
 Lzlib home page: http://www.nongnu.org/lzip/lzlib.html
 .SH COPYRIGHT
 Copyright \(co 2014 Antonio Diaz Diaz.
-Using Lzlib 1.6\-pre2
+Using lzlib 1.6\-pre3
 License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
 .br
 This is free software: you are free to change and redistribute it.

encoder.c

@@ -231,36 +231,6 @@ static int Mf_get_match_pairs( struct Matchfinder * const mf, struct Pair * pair
   }
 
 
-static void Lee_encode( struct Len_encoder * const le,
-                        struct Range_encoder * const renc,
-                        int symbol, const int pos_state )
-  {
-  symbol -= min_match_len;
-  if( symbol < len_low_symbols )
-    {
-    Re_encode_bit( renc, &le->lm.choice1, 0 );
-    Re_encode_tree( renc, le->lm.bm_low[pos_state], symbol, len_low_bits );
-    }
-  else
-    {
-    Re_encode_bit( renc, &le->lm.choice1, 1 );
-    if( symbol < len_low_symbols + len_mid_symbols )
-      {
-      Re_encode_bit( renc, &le->lm.choice2, 0 );
-      Re_encode_tree( renc, le->lm.bm_mid[pos_state],
-                      symbol - len_low_symbols, len_mid_bits );
-      }
-    else
-      {
-      Re_encode_bit( renc, &le->lm.choice2, 1 );
-      Re_encode_tree( renc, le->lm.bm_high,
-                      symbol - len_low_symbols - len_mid_symbols, len_high_bits );
-      }
-    }
-  if( --le->counters[pos_state] <= 0 ) Lee_update_prices( le, pos_state );
-  }
-
-
      /* End Of Stream mark => (dis == 0xFFFFFFFFU, len == min_match_len) */
 static bool LZe_full_flush( struct LZ_encoder * const e, const State state )
   {
@@ -298,16 +268,7 @@ static bool LZe_sync_flush( struct LZ_encoder * const e )
   }
 
 
-static void LZe_fill_align_prices( struct LZ_encoder * const e )
+static void LZe_update_distance_prices( struct LZ_encoder * const e )
-  {
-  int i;
-  for( i = 0; i < dis_align_size; ++i )
-    e->align_prices[i] = price_symbol_reversed( e->bm_align, i, dis_align_bits );
-  e->align_price_count = dis_align_size;
-  }
-
-
-static void LZe_fill_distance_prices( struct LZ_encoder * const e )
   {
   int dis, len_state;
   for( dis = start_dis_model; dis < modeled_distances; ++dis )
@@ -364,12 +325,15 @@ static bool LZe_init( struct LZ_encoder * const e,
 
   e->matchfinder = mf;
   if( !Re_init( &e->renc ) ) return false;
-  Lee_init( &e->match_len_encoder, mf->match_len_limit );
+  Lm_init( &e->match_len_model );
-  Lee_init( &e->rep_len_encoder, mf->match_len_limit );
+  Lm_init( &e->rep_len_model );
+  Lp_init( &e->match_len_prices, &e->match_len_model, mf->match_len_limit );
+  Lp_init( &e->rep_len_prices, &e->rep_len_model, mf->match_len_limit );
   for( i = 0; i < num_rep_distances; ++i ) e->reps[i] = 0;
-  e->align_price_count = 0;
   e->num_dis_slots = 2 * real_bits( mf->dictionary_size - 1 );
-  e->fill_counter = 0;
+  e->price_counter = 0;
+  e->dis_price_counter = 0;
+  e->align_price_counter = 0;
   e->state = 0;
   e->member_finished = false;
 
@@ -382,6 +346,7 @@ static bool LZe_init( struct LZ_encoder * const e,
 /* Return value == number of bytes advanced (ahead).
    trials[0]..trials[ahead-1] contain the steps to encode.
    ( trials[0].dis == -1 ) means literal.
+   A match/rep longer or equal than match_len_limit finishes the sequence.
 */
 static int LZe_sequence_optimizer( struct LZ_encoder * const e,
                                    const int reps[num_rep_distances],
@@ -468,7 +433,7 @@ static int LZe_sequence_optimizer( struct LZ_encoder * const e,
 
     for( len = min_match_len; len <= replens[rep]; ++len )
       Tr_update( &e->trials[len], price +
-                 Lee_price( &e->rep_len_encoder, len, pos_state ), rep, 0 );
+                 Lp_price( &e->rep_len_prices, len, pos_state ), rep, 0 );
     }
 
   if( main_len > replens[0] )
@@ -487,8 +452,6 @@ static int LZe_sequence_optimizer( struct LZ_encoder * const e,
     }
   }
 
-  if( !Mf_move_pos( e->matchfinder ) ) return 0;
-
   while( true )				/* price optimization loop */
     {
     struct Trial *cur_trial, *next_trial;
@@ -498,6 +461,7 @@ static int LZe_sequence_optimizer( struct LZ_encoder * const e,
     State cur_state;
     uint8_t prev_byte, cur_byte, match_byte;
 
+    if( !Mf_move_pos( e->matchfinder ) ) return 0;
     if( ++cur >= num_trials )		/* no more initialized trials */
       {
       LZe_backward( e, cur );
@@ -557,7 +521,6 @@ static int LZe_sequence_optimizer( struct LZ_encoder * const e,
     prev_byte = Mf_peek( e->matchfinder, 1 );
     cur_byte = Mf_peek( e->matchfinder, 0 );
     match_byte = Mf_peek( e->matchfinder, cur_trial->reps[0] + 1 );
-    if( !Mf_move_pos( e->matchfinder ) ) return 0;
 
     next_price = cur_trial->price +
                  price0( e->bm_match[cur_state][pos_state] );
@@ -587,7 +550,7 @@ static int LZe_sequence_optimizer( struct LZ_encoder * const e,
         }
       }
 
-    available_bytes = min( Mf_available_bytes( e->matchfinder ) + 1,
+    available_bytes = min( Mf_available_bytes( e->matchfinder ),
                            max_num_trials - 1 - cur );
     if( available_bytes < min_match_len ) continue;
 
@@ -596,7 +559,7 @@ static int LZe_sequence_optimizer( struct LZ_encoder * const e,
     /* try literal + rep0 */
     if( match_byte != cur_byte && next_trial->prev_index != cur )
       {
-      const uint8_t * const data = Mf_ptr_to_current_pos( e->matchfinder ) - 1;
+      const uint8_t * const data = Mf_ptr_to_current_pos( e->matchfinder );
       const int dis = cur_trial->reps[0] + 1;
       const int limit = min( e->matchfinder->match_len_limit + 1,
                              available_bytes );
@@ -619,7 +582,7 @@ static int LZe_sequence_optimizer( struct LZ_encoder * const e,
     /* try rep distances */
     for( rep = 0; rep < num_rep_distances; ++rep )
       {
-      const uint8_t * const data = Mf_ptr_to_current_pos( e->matchfinder ) - 1;
+      const uint8_t * const data = Mf_ptr_to_current_pos( e->matchfinder );
       int price;
       const int dis = cur_trial->reps[rep] + 1;
 
@@ -631,7 +594,7 @@ static int LZe_sequence_optimizer( struct LZ_encoder * const e,
       price = rep_match_price + LZe_price_rep( e, rep, cur_state, pos_state );
       for( i = min_match_len; i <= len; ++i )
         Tr_update( &e->trials[cur+i], price +
-                   Lee_price( &e->rep_len_encoder, i, pos_state ), rep, cur );
+                   Lp_price( &e->rep_len_prices, i, pos_state ), rep, cur );
 
       if( rep == 0 ) start_len = len + 1;	/* discard shorter matches */
 
@@ -647,7 +610,7 @@ static int LZe_sequence_optimizer( struct LZ_encoder * const e,
 
       pos_state2 = ( pos_state + len ) & pos_state_mask;
       state2 = St_set_rep( cur_state );
-      price += Lee_price( &e->rep_len_encoder, len, pos_state ) +
+      price += Lp_price( &e->rep_len_prices, len, pos_state ) +
                price0( e->bm_match[state2][pos_state2] ) +
                LZe_price_matched( e, data[len-1], data[len], data[len-dis] );
       pos_state2 = ( pos_state2 + 1 ) & pos_state_mask;
@@ -683,7 +646,7 @@ static int LZe_sequence_optimizer( struct LZ_encoder * const e,
         /* try match + literal + rep0 */
         if( len == e->pairs[i].len )
           {
-          const uint8_t * const data = Mf_ptr_to_current_pos( e->matchfinder ) - 1;
+          const uint8_t * const data = Mf_ptr_to_current_pos( e->matchfinder );
           const int dis2 = dis + 1;
           int len2 = len + 1;
           const int limit = min( e->matchfinder->match_len_limit + len2,
@@ -718,7 +681,10 @@ static int LZe_sequence_optimizer( struct LZ_encoder * const e,
 
 static bool LZe_encode_member( struct LZ_encoder * const e )
   {
-  const int fill_count = ( e->matchfinder->match_len_limit > 12 ) ? 128 : 512;
+  const bool best = ( e->matchfinder->match_len_limit > 12 );
+  const int dis_price_count = best ? 1 : 512;
+  const int align_price_count = best ? 1 : dis_align_size;
+  const int price_count = ( e->matchfinder->match_len_limit > 36 ) ? 1024 : 4096;
   int ahead, i;
   State * const state = &e->state;
 
@@ -733,12 +699,13 @@ static bool LZe_encode_member( struct LZ_encoder * const e )
       !Mf_finished( e->matchfinder ) )		/* encode first byte */
     {
     const uint8_t prev_byte = 0;
-    const uint8_t cur_byte = Mf_peek( e->matchfinder, 0 );
+    uint8_t cur_byte;
     if( !Mf_enough_available_bytes( e->matchfinder ) ||
         !Re_enough_free_bytes( &e->renc ) ) return true;
-    CRC32_update_byte( &e->crc, cur_byte );
+    cur_byte = Mf_peek( e->matchfinder, 0 );
     Re_encode_bit( &e->renc, &e->bm_match[*state][0], 0 );
     LZe_encode_literal( e, prev_byte, cur_byte );
+    CRC32_update_byte( &e->crc, cur_byte );
     Mf_get_match_pairs( e->matchfinder, 0 );
     if( !Mf_move_pos( e->matchfinder ) ) return false;
     }
@@ -747,15 +714,24 @@ static bool LZe_encode_member( struct LZ_encoder * const e )
     {
     if( !Mf_enough_available_bytes( e->matchfinder ) ||
         !Re_enough_free_bytes( &e->renc ) ) return true;
-    if( e->pending_num_pairs == 0 )
+    if( e->price_counter <= 0 && e->pending_num_pairs == 0 )
       {
-      if( e->fill_counter <= 0 )
+      e->price_counter = price_count;	/* recalculate prices every these bytes */
-        { LZe_fill_distance_prices( e ); e->fill_counter = fill_count; }
+      if( e->dis_price_counter <= 0 )
-      if( e->align_price_count <= 0 ) LZe_fill_align_prices( e );
+        { e->dis_price_counter = dis_price_count; LZe_update_distance_prices( e ); }
+      if( e->align_price_counter <= 0 )
+        {
+        e->align_price_counter = align_price_count;
+        for( i = 0; i < dis_align_size; ++i )
+          e->align_prices[i] = price_symbol_reversed( e->bm_align, i, dis_align_bits );
+        }
+      Lp_update_prices( &e->match_len_prices );
+      Lp_update_prices( &e->rep_len_prices );
       }
 
     ahead = LZe_sequence_optimizer( e, e->reps, *state );
     if( ahead <= 0 ) return false;		/* can't happen */
+    e->price_counter -= ahead;
 
     for( i = 0; ahead > 0; )
       {
@@ -802,14 +778,18 @@ static bool LZe_encode_member( struct LZ_encoder * const e )
           if( len == 1 ) *state = St_set_short_rep( *state );
           else
             {
-            Lee_encode( &e->rep_len_encoder, &e->renc, len, pos_state );
+            Re_encode_len( &e->renc, &e->rep_len_model, len, pos_state );
+            Lp_decrement_counter( &e->rep_len_prices, pos_state );
             *state = St_set_rep( *state );
             }
           }
         else					/* match */
           {
           LZe_encode_pair( e, dis - num_rep_distances, len, pos_state );
-          --e->fill_counter;
+          if( get_slot( dis - num_rep_distances ) >= end_dis_model )
+            --e->align_price_counter;
+          --e->dis_price_counter;
+          Lp_decrement_counter( &e->match_len_prices, pos_state );
           *state = St_set_match( *state );
           }
         }

encoder.h

@@ -476,52 +476,93 @@ static inline void Re_encode_matched( struct Range_encoder * const renc,
   while( symbol < 0x10000 );
   }
 
-
+static inline void Re_encode_len( struct Range_encoder * const renc,
-struct Len_encoder
+                                  struct Len_model * const lm,
+                                  int symbol, const int pos_state )
   {
-  struct Len_model lm;
+  bool bit = ( ( symbol -= min_match_len ) >= len_low_symbols );
-  int prices[pos_states][max_len_symbols];
+  Re_encode_bit( renc, &lm->choice1, bit );
+  if( !bit )
+    Re_encode_tree( renc, lm->bm_low[pos_state], symbol, len_low_bits );
+  else
+    {
+    bit = ( symbol >= len_low_symbols + len_mid_symbols );
+    Re_encode_bit( renc, &lm->choice2, bit );
+    if( !bit )
+      Re_encode_tree( renc, lm->bm_mid[pos_state],
+                      symbol - len_low_symbols, len_mid_bits );
+    else
+      Re_encode_tree( renc, lm->bm_high,
+                      symbol - len_low_symbols - len_mid_symbols, len_high_bits );
+    }
+  }
+
+
+struct Len_prices
+  {
+  const struct Len_model * lm;
   int len_symbols;
+  int count;
+  int prices[pos_states][max_len_symbols];
   int counters[pos_states];
   };
 
-static void Lee_update_prices( struct Len_encoder * const le,
+static inline void Lp_update_low_mid_prices( struct Len_prices * const lp,
-                               const int pos_state )
+                                             const int pos_state )
   {
-  int * const pps = le->prices[pos_state];
+  int * const pps = lp->prices[pos_state];
-  int tmp = price0( le->lm.choice1 );
+  int tmp = price0( lp->lm->choice1 );
   int len = 0;
-  for( ; len < len_low_symbols && len < le->len_symbols; ++len )
+  lp->counters[pos_state] = lp->count;
-    pps[len] = tmp + price_symbol( le->lm.bm_low[pos_state], len, len_low_bits );
+  for( ; len < len_low_symbols && len < lp->len_symbols; ++len )
-  tmp = price1( le->lm.choice1 );
+    pps[len] = tmp + price_symbol( lp->lm->bm_low[pos_state], len, len_low_bits );
-  for( ; len < len_low_symbols + len_mid_symbols && len < le->len_symbols; ++len )
+  if( len >= lp->len_symbols ) return;
-    pps[len] = tmp + price0( le->lm.choice2 ) +
+  tmp = price1( lp->lm->choice1 ) + price0( lp->lm->choice2 );
-               price_symbol( le->lm.bm_mid[pos_state], len - len_low_symbols, len_mid_bits );
+  for( ; len < len_low_symbols + len_mid_symbols && len < lp->len_symbols; ++len )
-  for( ; len < le->len_symbols; ++len )
+    pps[len] = tmp +
-    /* using 4 slots per value makes "Lee_price" faster */
+               price_symbol( lp->lm->bm_mid[pos_state], len - len_low_symbols, len_mid_bits );
-    le->prices[3][len] = le->prices[2][len] =
+    }
-    le->prices[1][len] = le->prices[0][len] =
+
-      tmp + price1( le->lm.choice2 ) +
+static inline void Lp_update_high_prices( struct Len_prices * const lp )
-      price_symbol( le->lm.bm_high, len - len_low_symbols - len_mid_symbols, len_high_bits );
+  {
-  le->counters[pos_state] = le->len_symbols;
+  const int tmp = price1( lp->lm->choice1 ) + price1( lp->lm->choice2 );
+  int len;
+  for( len = len_low_symbols + len_mid_symbols; len < lp->len_symbols; ++len )
+    /* using 4 slots per value makes "Lp_price" faster */
+    lp->prices[3][len] = lp->prices[2][len] =
+    lp->prices[1][len] = lp->prices[0][len] = tmp +
+      price_symbol( lp->lm->bm_high, len - len_low_symbols - len_mid_symbols, len_high_bits );
   }
 
-static void Lee_init( struct Len_encoder * const le,
+static inline void Lp_init( struct Len_prices * const lp,
-                      const int match_len_limit )
+                            const struct Len_model * const lm,
+                            const int match_len_limit )
   {
   int i;
-  Lm_init( &le->lm );
+  lp->lm = lm;
-  le->len_symbols = match_len_limit + 1 - min_match_len;
+  lp->len_symbols = match_len_limit + 1 - min_match_len;
-  for( i = 0; i < pos_states; ++i ) Lee_update_prices( le, i );
+  lp->count = ( match_len_limit > 12 ) ? 1 : lp->len_symbols;
+  for( i = 0; i < pos_states; ++i ) lp->counters[i] = 0;
   }
 
-static void Lee_encode( struct Len_encoder * const le,
+static inline void Lp_decrement_counter( struct Len_prices * const lp,
-                        struct Range_encoder * const renc,
+                                         const int pos_state )
-                        int symbol, const int pos_state );
+  { --lp->counters[pos_state]; }
 
-static inline int Lee_price( const struct Len_encoder * const le,
+static inline void Lp_update_prices( struct Len_prices * const lp )
-                             const int symbol, const int pos_state )
+  {
-  { return le->prices[pos_state][symbol - min_match_len]; }
+  int pos_state;
+  bool high_pending = false;
+  for( pos_state = 0; pos_state < pos_states; ++pos_state )
+    if( lp->counters[pos_state] <= 0 )
+      { Lp_update_low_mid_prices( lp, pos_state ); high_pending = true; }
+  if( high_pending && lp->len_symbols > len_low_symbols + len_mid_symbols )
+    Lp_update_high_prices( lp );
+  }
+
+static inline int Lp_price( const struct Len_prices * const lp,
+                            const int symbol, const int pos_state )
+  { return lp->prices[pos_state][symbol - min_match_len]; }
 
 
 enum { infinite_price = 0x0FFFFFFF,
@@ -593,8 +634,10 @@ struct LZ_encoder
 
   struct Matchfinder * matchfinder;
   struct Range_encoder renc;
-  struct Len_encoder match_len_encoder;
+  struct Len_model match_len_model;
-  struct Len_encoder rep_len_encoder;
+  struct Len_model rep_len_model;
+  struct Len_prices match_len_prices;
+  struct Len_prices rep_len_prices;
 
   struct Pair pairs[max_match_len+1];
   struct Trial trials[max_num_trials];
@@ -603,9 +646,10 @@ struct LZ_encoder
   int dis_slot_prices[len_states][2*max_dictionary_bits];
   int dis_prices[len_states][modeled_distances];
   int align_prices[dis_align_size];
-  int align_price_count;
   int num_dis_slots;
-  int fill_counter;
+  int price_counter;
+  int dis_price_counter;
+  int align_price_counter;
   State state;
   bool member_finished;
   };
@@ -669,14 +713,14 @@ static inline int LZe_price_rep0_len( const struct LZ_encoder * const e,
                                       const State state, const int pos_state )
   {
   return LZe_price_rep( e, 0, state, pos_state ) +
-         Lee_price( &e->rep_len_encoder, len, pos_state );
+         Lp_price( &e->rep_len_prices, len, pos_state );
   }
 
 static inline int LZe_price_pair( const struct LZ_encoder * const e,
                                   const int dis, const int len,
                                   const int pos_state )
   {
-  const int price = Lee_price( &e->match_len_encoder, len, pos_state );
+  const int price = Lp_price( &e->match_len_prices, len, pos_state );
   const int len_state = get_len_state( len );
   if( dis < modeled_distances )
     return price + e->dis_prices[len_state][dis];
@@ -711,7 +755,7 @@ static inline void LZe_encode_pair( struct LZ_encoder * const e,
                                     const int pos_state )
   {
   const int dis_slot = get_slot( dis );
-  Lee_encode( &e->match_len_encoder, &e->renc, len, pos_state );
+  Re_encode_len( &e->renc, &e->match_len_model, len, pos_state );
   Re_encode_tree( &e->renc, e->bm_dis_slot[get_len_state(len)], dis_slot,
                   dis_slot_bits );
 
@@ -729,7 +773,6 @@ static inline void LZe_encode_pair( struct LZ_encoder * const e,
       Re_encode( &e->renc, direct_dis >> dis_align_bits,
                  direct_bits - dis_align_bits );
       Re_encode_tree_reversed( &e->renc, e->bm_align, direct_dis, dis_align_bits );
-      --e->align_price_count;
       }
     }
   }

lzcheck.c

@@ -38,7 +38,7 @@ int main( const int argc, const char * const argv[] )
   {
   const int dictionary_size = 1 << 20;
   const int match_len_limit = 16;
-  const unsigned long long member_size = INT64_MAX;
+  const unsigned long long member_size = 0x7FFFFFFFFFFFFFFFULL;	/* INT64_MAX */
   struct LZ_Encoder * encoder;
   struct LZ_Decoder * decoder;
   FILE * file;

lzlib.h

@@ -29,7 +29,7 @@
 extern "C" {
 #endif
 
-static const char * const LZ_version_string = "1.6-pre2";
+static const char * const LZ_version_string = "1.6-pre3";
 
 enum LZ_Errno { LZ_ok = 0,         LZ_bad_argument, LZ_mem_error,
                 LZ_sequence_error, LZ_header_error, LZ_unexpected_eof,

main.c

@@ -35,7 +35,7 @@
 #include <unistd.h>
 #include <utime.h>
 #include <sys/stat.h>
-#if defined(__MSVCRT__)
+#if defined(__MSVCRT__) || defined(_MSC_VER)
 #include <io.h>
 #define fchmod(x,y) 0
 #define fchown(x,y,z) 0
@@ -202,9 +202,9 @@ static void show_help( void )
 
 static void show_version( void )
   {
-  printf( "%s %s\n", Program_name, PROGVERSION );
+  printf( "%s %s\n", program_name, PROGVERSION );
   printf( "Copyright (C) %s Antonio Diaz Diaz.\n", program_year );
-  printf( "Using Lzlib %s\n", LZ_version() );
+  printf( "Using lzlib %s\n", LZ_version() );
   printf( "License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>\n"
           "This is free software: you are free to change and redistribute it.\n"
           "There is NO WARRANTY, to the extent permitted by law.\n" );
@@ -322,8 +322,7 @@ static int open_instream( const char * const name, struct stat * const in_statsp
     }
   else
     {
-    do infd = open( name, O_RDONLY | O_BINARY );
+    infd = open( name, O_RDONLY | O_BINARY );
-    while( infd < 0 && errno == EINTR );
     if( infd < 0 )
       {
       if( verbosity >= 0 )
@@ -407,8 +406,7 @@ static bool open_outstream( const bool force )
   int flags = O_CREAT | O_WRONLY | O_BINARY;
   if( force ) flags |= O_TRUNC; else flags |= O_EXCL;
 
-  do outfd = open( output_filename, flags, outfd_mode );
+  outfd = open( output_filename, flags, outfd_mode );
-  while( outfd < 0 && errno == EINTR );
   if( outfd < 0 && verbosity >= 0 )
     {
     if( errno == EEXIST )
@@ -497,7 +495,7 @@ static int readblock( const int fd, uint8_t * const buf, const int size )
     const int n = read( fd, buf + size - rest, rest );
     if( n > 0 ) rest -= n;
     else if( n == 0 ) break;				/* EOF */
-    else if( errno != EINTR && errno != EAGAIN ) break;
+    else if( errno != EINTR ) break;
     errno = 0;
     }
   return size - rest;
@@ -515,7 +513,7 @@ static int writeblock( const int fd, const uint8_t * const buf, const int size )
     {
     const int n = write( fd, buf + size - rest, rest );
     if( n > 0 ) rest -= n;
-    else if( n < 0 && errno != EINTR && errno != EAGAIN ) break;
+    else if( n < 0 && errno != EINTR ) break;
     errno = 0;
     }
   return size - rest;
@@ -931,7 +929,7 @@ int main( const int argc, const char * const argv[] )
       }
     } /* end process options */
 
-#if defined(__MSVCRT__) || defined(__OS2__)
+#if defined(__MSVCRT__) || defined(__OS2__) || defined(_MSC_VER)
   setmode( STDIN_FILENO, O_BINARY );
   setmode( STDOUT_FILENO, O_BINARY );
 #endif

testsuite/check.sh

@@ -14,7 +14,7 @@ BBEXAMPLE="${objdir}"/bbexample
 LZCHECK="${objdir}"/lzcheck
 framework_failure() { echo "failure in testing framework" ; exit 1 ; }
 
-if [ ! -x "${LZIP}" ] ; then
+if [ ! -f "${LZIP}" ] || [ ! -x "${LZIP}" ] ; then
 	echo "${LZIP}: cannot execute"
 	exit 1
 fi
@@ -66,13 +66,13 @@ printf .
 
 "${LZIP}" -cfq "${in_lz}" > out
 if [ $? = 1 ] ; then printf . ; else fail=1 ; printf - ; fi
-"${LZIP}" -cF "${in_lz}" > out || fail=1
+"${LZIP}" -cF -s16 "${in_lz}" > out || fail=1
 "${LZIP}" -cd out | "${LZIP}" -d > copy || fail=1
 cmp in copy || fail=1
 printf .
 
-for i in s4Ki 0 1 2 3 4 5 6 7s16 8s16 9s16 ; do
+for i in s4Ki 0 1 2 3 4 5 6 7 8 9 ; do
-	"${LZIP}" -k -$i in || fail=1
+	"${LZIP}" -k -$i -s16 in || fail=1
 	mv -f in.lz copy.lz || fail=1
 	printf "garbage" >> copy.lz || fail=1
 	"${LZIP}" -df copy.lz || fail=1
@@ -80,35 +80,35 @@ for i in s4Ki 0 1 2 3 4 5 6 7s16 8s16 9s16 ; do
 done
 printf .
 
-for i in s4Ki 0 1 2 3 4 5 6 7s16 8s16 9s16 ; do
+for i in s4Ki 0 1 2 3 4 5 6 7 8 9 ; do
-	"${LZIP}" -c -$i in > out || fail=1
+	"${LZIP}" -c -$i -s16 in > out || fail=1
 	printf "g" >> out || fail=1
 	"${LZIP}" -cd out > copy || fail=1
 	cmp in copy || fail=1
 done
 printf .
 
-for i in s4Ki 0 1 2 3 4 5 6 7s16 8s16 9s16 ; do
+for i in s4Ki 0 1 2 3 4 5 6 7 8 9 ; do
-	"${LZIP}" -$i < in > out || fail=1
+	"${LZIP}" -$i -s16 < in > out || fail=1
 	"${LZIP}" -d < out > copy || fail=1
 	cmp in copy || fail=1
 done
 printf .
 
-for i in s4Ki 0 1 2 3 4 5 6 7s16 8s16 9s16 ; do
+for i in s4Ki 0 1 2 3 4 5 6 7 8 9 ; do
-	"${LZIP}" -f -$i -o out < in || fail=1
+	"${LZIP}" -f -$i -s16 -o out < in || fail=1
 	"${LZIP}" -df -o copy < out.lz || fail=1
 	cmp in copy || fail=1
 done
 printf .
 
-"${LZIP}" < in > anyothername || fail=1
+"${LZIP}" -s16 < in > anyothername || fail=1
 "${LZIP}" -d anyothername || fail=1
 cmp in anyothername.out || fail=1
 printf .
 
 cat in in > in2 || framework_failure
-"${LZIP}" -o copy2 < in2 || fail=1
+"${LZIP}" -s16 -o copy2 < in2 || fail=1
 "${LZIP}" -t copy2.lz || fail=1
 printf .
 "${LZIP}" -cd copy2.lz > copy2 || fail=1