Merging upstream version 1.18~pre1.

Signed-off-by: Daniel Baumann <mail@daniel-baumann.ch>

1 files changed, 165 insertions, 4 deletions

diff --git a/mtester.cc b/mtester.cc
index c23d51b..a9ac06e 100644
--- a/mtester.cc
+++ b/mtester.cc
@@ -32,6 +32,24 @@
 #include "mtester.h"
 
 
+namespace {
+
+const char * format_byte( const uint8_t byte )
+  {
+  enum { buffers = 8, bufsize = 16 };
+  static char buffer[buffers][bufsize];	// circle of static buffers for printf
+  static int current = 0;
+  char * const buf = buffer[current++]; current %= buffers;
+  if( ( byte >= 0x20 && byte <= 0x7E ) || byte >= 0xA0 )
+    snprintf( buf, bufsize, "'%c' (0x%02X)", byte, byte );
+  else
+    snprintf( buf, bufsize, "    (0x%02X)", byte );
+  return buf;
+  }
+
+} // end namespace
+
+
 void LZ_mtester::flush_data()
   {
   if( pos > stream_pos )
@@ -56,6 +74,19 @@ bool LZ_mtester::verify_trailer()
   }
 
 
+void LZ_mtester::print_block( const int len )
+  {
+  std::fputs( " \"", stdout );
+  for( int i = len - 1; i >= 0; --i )
+    {
+    uint8_t byte = peek( i );
+    if( byte < 0x20 || ( byte > 0x7E && byte < 0xA0 ) ) byte = '.';
+    std::fputc( byte, stdout );
+    }
+  std::fputs( "\"\n", stdout );
+  }
+
+
 void LZ_mtester::duplicate_buffer()
   {
   uint8_t * const tmp = new uint8_t[buffer_size];
@@ -80,7 +111,7 @@ int LZ_mtester::test_member( const long pos_limit )
     const int pos_state = data_position() & pos_state_mask;
     if( rdec.decode_bit( bm_match[state()][pos_state] ) == 0 )	// 1st bit
       {
-      const uint8_t prev_byte = get_prev_byte();
+      const uint8_t prev_byte = peek_prev();
       if( state.is_char() )
         {
         state.set_char1();
@@ -90,7 +121,7 @@ int LZ_mtester::test_member( const long pos_limit )
         {
         state.set_char2();
         put_byte( rdec.decode_matched( bm_literal[get_lit_state(prev_byte)],
-                                       get_byte( rep0 ) ) );
+                                       peek( rep0 ) ) );
         }
       }
     else
@@ -117,7 +148,7 @@ int LZ_mtester::test_member( const long pos_limit )
         else
           {
           if( rdec.decode_bit( bm_len[state()][pos_state] ) == 0 ) // 4th bit
-            { state.set_short_rep(); put_byte( get_byte( rep0 ) ); continue; }
+            { state.set_short_rep(); put_byte( peek( rep0 ) ); continue; }
           }
         state.set_rep();
         len = min_match_len + rdec.decode_len( rep_len_model, pos_state );
@@ -165,6 +196,136 @@ int LZ_mtester::test_member( const long pos_limit )
   }
 
 
+/* Return value: 0 = OK, 1 = decoder error, 2 = unexpected EOF,
+                 3 = trailer error, 4 = unknown marker found. */
+int LZ_mtester::debug_decode_member( const long long dpos, const long long mpos,
+                                     const bool show_packets )
+  {
+  rdec.load();
+  while( !rdec.finished() )
+    {
+    const unsigned long long dp = data_position() + dpos;
+    const unsigned long long mp = member_position() + mpos - 4;
+    const int pos_state = data_position() & pos_state_mask;
+    if( rdec.decode_bit( bm_match[state()][pos_state] ) == 0 )	// 1st bit
+      {
+      const uint8_t prev_byte = peek_prev();
+      if( state.is_char() )
+        {
+        state.set_char1();
+        const uint8_t cur_byte = rdec.decode_tree8( bm_literal[get_lit_state(prev_byte)] );
+        put_byte( cur_byte );
+        if( show_packets )
+          std::printf( "%6llu %6llu  literal %s\n",
+                       mp, dp, format_byte( cur_byte ) );
+        }
+      else
+        {
+        state.set_char2();
+        const uint8_t match_byte = peek( rep0 );
+        const uint8_t cur_byte =
+          rdec.decode_matched( bm_literal[get_lit_state(prev_byte)], match_byte );
+        put_byte( cur_byte );
+        if( show_packets )
+          std::printf( "%6llu %6llu  literal %s, match byte %6llu %s\n",
+                       mp, dp, format_byte( cur_byte ), dp - rep0 - 1,
+                       format_byte( match_byte ) );
+        }
+      }
+    else					/* match or repeated match */
+      {
+      int len;
+      if( rdec.decode_bit( bm_rep[state()] ) != 0 )		// 2nd bit
+        {
+        int rep = 0;
+        if( rdec.decode_bit( bm_rep0[state()] ) != 0 )		// 3rd bit
+          {
+          unsigned distance;
+          if( rdec.decode_bit( bm_rep1[state()] ) == 0 )	// 4th bit
+            { distance = rep1; rep = 1; }
+          else
+            {
+            if( rdec.decode_bit( bm_rep2[state()] ) == 0 )	// 5th bit
+              { distance = rep2; rep = 2; }
+            else
+              { distance = rep3; rep3 = rep2; rep = 3; }
+            rep2 = rep1;
+            }
+          rep1 = rep0;
+          rep0 = distance;
+          }
+        else
+          {
+          if( rdec.decode_bit( bm_len[state()][pos_state] ) == 0 ) // 4th bit
+            {
+            if( show_packets )
+              std::printf( "%6llu %6llu shortrep %s %6u (%6llu)\n",
+                           mp, dp, format_byte( peek( rep0 ) ),
+                           rep0 + 1, dp - rep0 - 1 );
+            state.set_short_rep(); put_byte( peek( rep0 ) ); continue;
+            }
+          }
+        state.set_rep();
+        len = min_match_len + rdec.decode_len( rep_len_model, pos_state );
+        if( show_packets )
+          std::printf( "%6llu %6llu  rep%c  %6u,%3d (%6llu)",
+                       mp, dp, rep + '0', rep0 + 1, len, dp - rep0 - 1 );
+        }
+      else					/* match */
+        {
+        const unsigned rep0_saved = rep0;
+        len = min_match_len + rdec.decode_len( match_len_model, pos_state );
+        const int dis_slot = rdec.decode_tree6( bm_dis_slot[get_len_state(len)] );
+        if( dis_slot < start_dis_model ) rep0 = dis_slot;
+        else
+          {
+          const int direct_bits = ( dis_slot >> 1 ) - 1;
+          rep0 = ( 2 | ( dis_slot & 1 ) ) << direct_bits;
+          if( dis_slot < end_dis_model )
+            rep0 += rdec.decode_tree_reversed( bm_dis + rep0 - dis_slot - 1,
+                                               direct_bits );
+          else
+            {
+            rep0 += rdec.decode( direct_bits - dis_align_bits ) << dis_align_bits;
+            rep0 += rdec.decode_tree_reversed4( bm_align );
+            if( rep0 == 0xFFFFFFFFU )		// marker found
+              {
+              rep0 = rep0_saved;
+              rdec.normalize();
+              flush_data();
+              if( show_packets )
+                std::printf( "%6llu %6llu  marker code '%d'\n", mp, dp, len );
+              if( len == min_match_len )	// End Of Stream marker
+                {
+                if( show_packets )
+                  std::printf( "%6llu %6llu  member trailer\n",
+                               mpos + member_position(), dpos + data_position() );
+                if( verify_trailer() ) return 0;
+                if( show_packets ) std::fputs( "trailer error\n", stdout );
+                return 3;
+                }
+              return 4;
+              }
+            }
+          }
+        rep3 = rep2; rep2 = rep1; rep1 = rep0_saved;
+        state.set_match();
+        if( show_packets )
+          std::printf( "%6llu %6llu  match %6u,%3d (%6lld)",
+                       mp, dp, rep0 + 1, len, dp - rep0 - 1 );
+        if( rep0 >= dictionary_size || rep0 >= data_position() )
+          { flush_data(); if( show_packets ) std::fputc( '\n', stdout );
+            return 1; }
+        }
+      copy_block( rep0, len );
+      if( show_packets ) print_block( len );
+      }
+    }
+  flush_data();
+  return 2;
+  }
+
+
 uint8_t * read_member( const int infd, const long long mpos,
                        const long long msize )
   {
@@ -184,7 +345,7 @@ const LZ_mtester * prepare_master( const uint8_t * const buffer,
                                    const long buffer_size,
                                    const long pos_limit )
   {
-  File_header & header = *(File_header *)buffer;
+  const File_header & header = *(File_header *)buffer;
   const unsigned dictionary_size = header.dictionary_size();
   if( header.verify_magic() && header.verify_version() &&
       dictionary_size >= min_dictionary_size &&