Merging upstream version 0.25.

Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>

1 files changed, 80 insertions, 39 deletions

diff --git a/decode.cc b/decode.cc
index a45a1fd..bcac4c8 100644
--- a/decode.cc
+++ b/decode.cc
@@ -1,5 +1,5 @@
 /* Tarlz - Archiver with multimember lzip compression
-   Copyright (C) 2013-2022 Antonio Diaz Diaz.
+   Copyright (C) 2013-2024 Antonio Diaz Diaz.
 
    This program is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
@@ -21,7 +21,7 @@
 #include <cctype>
 #include <cerrno>
 #include <cstdio>
-#include <cstdlib>
+#include <fcntl.h>
 #include <stdint.h>	// for lzlib.h
 #include <unistd.h>
 #include <utime.h>
@@ -29,6 +29,8 @@
 #if !defined __FreeBSD__ && !defined __OpenBSD__ && !defined __NetBSD__ && \
     !defined __DragonFly__ && !defined __APPLE__ && !defined __OS2__
 #include <sys/sysmacros.h>		// for major, minor, makedev
+#else
+#include <sys/types.h>			// for major, minor, makedev
 #endif
 #include <lzlib.h>
 
@@ -38,6 +40,9 @@
 #include "archive_reader.h"
 #include "decode.h"
 
+#ifndef O_DIRECTORY
+#define O_DIRECTORY 0
+#endif
 
 namespace {
 
@@ -124,7 +129,7 @@ int extract_member( const Cl_options & cl_opts, Archive_reader & ar,
   if( !show_member_name( extended, header, 1, grbuf ) ) return 1;
   // remove file (or empty dir) before extraction to prevent following links
   std::remove( filename );
-  if( !make_path( filename ) )
+  if( !make_dirs( filename ) )
     {
     show_file_error( filename, intdir_msg, errno );
     set_error_status( 1 );
@@ -192,7 +197,7 @@ int extract_member( const Cl_options & cl_opts, Archive_reader & ar,
         chown( filename, extended.get_uid(), extended.get_gid() ) != 0 ) )
     {
     if( outfd >= 0 ) mode &= ~( S_ISUID | S_ISGID | S_ISVTX );
-    // chown will in many cases return with EPERM, which can be safely ignored.
+    // chown in many cases returns with EPERM, which can be safely ignored.
     if( errno != EPERM && errno != EINVAL )
       { show_file_error( filename, chown_msg, errno ); set_error_status( 1 ); }
     }
@@ -246,9 +251,37 @@ void format_file_diff( std::string & ostr, const char * const filename,
   { if( verbosity >= 0 )
       { ostr += filename; ostr += ": "; ostr += msg; ostr += '\n'; } }
 
+
+bool option_C_present( const Arg_parser & parser )
+  {
+  for( int i = 0; i < parser.arguments(); ++i )
+    if( parser.code( i ) == 'C' ) return true;
+  return false;
+  }
+
+
+bool option_C_after_filename( const Arg_parser & parser )
+  {
+  for( int i = 0; i < parser.arguments(); ++i )
+    if( nonempty_arg( parser, i ) )
+      while( ++i < parser.arguments() )
+        if( parser.code( i ) == 'C' ) return true;
+  return false;
+  }
+
 } // end namespace
 
 
+mode_t get_umask()
+  {
+  static mode_t mask = 0;		// read once, cache the result
+  static bool first_call = true;
+  if( first_call ) { first_call = false; mask = umask( 0 ); umask( mask );
+                     mask &= S_IRWXU | S_IRWXG | S_IRWXO; }
+  return mask;
+  }
+
+
 bool compare_file_type( std::string & estr, std::string & ostr,
                         const Cl_options & cl_opts,
                         const Extended & extended, const Tar_header header )
@@ -258,7 +291,7 @@ bool compare_file_type( std::string & estr, std::string & ostr,
   struct stat st;
   bool diff = false, size_differs = false, type_differs = true;
   if( hstat( filename, &st, cl_opts.dereference ) != 0 )
-    format_file_error( estr, filename, "warning: Can't stat", errno );
+    format_file_error( estr, filename, "warning: can't stat", errno );
   else if( ( typeflag == tf_regular || typeflag == tf_hiperf ) &&
            !S_ISREG( st.st_mode ) )
     format_file_diff( ostr, filename, "Is not a regular file" );
@@ -275,14 +308,14 @@ bool compare_file_type( std::string & estr, std::string & ostr,
   else
     {
     type_differs = false;
-    if( typeflag != tf_symlink )
+    if( typeflag != tf_symlink && !cl_opts.ignore_metadata )
       {
       const mode_t mode = parse_octal( header + mode_o, mode_l );  // 12 bits
       if( mode != ( st.st_mode & ( S_ISUID | S_ISGID | S_ISVTX |
                                    S_IRWXU | S_IRWXG | S_IRWXO ) ) )
         { format_file_diff( ostr, filename, "Mode differs" ); diff = true; }
       }
-    if( !cl_opts.ignore_ids )
+    if( !cl_opts.ignore_ids && !cl_opts.ignore_metadata )
       {
       if( extended.get_uid() != (long long)st.st_uid )
         { format_file_diff( ostr, filename, "Uid differs" ); diff = true; }
@@ -291,7 +324,7 @@ bool compare_file_type( std::string & estr, std::string & ostr,
       }
     if( typeflag != tf_symlink )
       {
-      if( typeflag != tf_directory &&
+      if( typeflag != tf_directory && !cl_opts.ignore_metadata &&
           extended.mtime().sec() != (long long)st.st_mtime )
         {
         if( (time_t)extended.mtime().sec() == st.st_mtime )
@@ -380,32 +413,37 @@ int decode( const Cl_options & cl_opts )
   const Archive_descriptor ad( cl_opts.archive_name );
   if( ad.infd < 0 ) return 1;
 
-  // Execute -C options and mark filenames to be compared, extracted or listed.
-  // name_pending is of type char instead of bool to allow concurrent update.
-  std::vector< char > name_pending( cl_opts.parser.arguments(), false );
-  for( int i = 0; i < cl_opts.parser.arguments(); ++i )
-    {
-    const int code = cl_opts.parser.code( i );
-    if( code == 'C' && cl_opts.program_mode != m_list )
+  const bool c_present = option_C_present( cl_opts.parser ) &&
+                         cl_opts.program_mode != m_list;
+  const bool c_after_name = c_present &&
+                            option_C_after_filename( cl_opts.parser );
+  // save current working directory for sequential decoding
+  const int chdir_fd = c_after_name ? open( ".", O_RDONLY | O_DIRECTORY ) : -1;
+  if( c_after_name && chdir_fd < 0 )
+    { show_error( "Can't save current working directory", errno ); return 1; }
+  if( c_present && !c_after_name )		// execute all -C options
+    for( int i = 0; i < cl_opts.parser.arguments(); ++i )
       {
+      if( cl_opts.parser.code( i ) != 'C' ) continue;
       const char * const dir = cl_opts.parser.argument( i ).c_str();
       if( chdir( dir ) != 0 )
         { show_file_error( dir, chdir_msg, errno ); return 1; }
       }
-    if( !code && cl_opts.parser.argument( i ).size() &&
+  /* Mark filenames to be compared, extracted or listed.
+     name_pending is of type char instead of bool to allow concurrent update. */
+  std::vector< char > name_pending( cl_opts.parser.arguments(), false );
+  for( int i = 0; i < cl_opts.parser.arguments(); ++i )
+    if( nonempty_arg( cl_opts.parser, i ) &&	// skip opts, empty names
         !Exclude::excluded( cl_opts.parser.argument( i ).c_str() ) )
       name_pending[i] = true;
-    }
 
-  // multi-threaded --list is faster even with 1 thread and 1 file in archive
-  // but multi-threaded --diff and --extract probably need at least 2 of each
-  if( ( cl_opts.program_mode == m_diff || cl_opts.program_mode == m_list ||
-        cl_opts.program_mode == m_extract ) && cl_opts.num_workers > 0 &&
-      ad.indexed && ad.lzip_index.members() >= 2 )	// one file + EOA
-    {
-    // show_file_error( ad.namep, "Is compressed seekable" );
+  /* multi-threaded --list is faster even with 1 thread and 1 file in archive
+     but multi-threaded --diff and --extract probably need at least 2 of each.
+     CWD is not per-thread; multi-threaded decode can't be used if a
+     -C option appears after a file name in the command line. */
+  if( cl_opts.num_workers > 0 && !c_after_name && ad.indexed &&
+      ad.lzip_index.members() >= 2 )	// 2 lzip members may be 1 file + EOA
     return decode_lz( cl_opts, ad, name_pending );
-    }
 
   Archive_reader ar( ad );		// serial reader
   Extended extended;			// metadata from extended records
@@ -416,7 +454,7 @@ int decode( const Cl_options & cl_opts )
     Tar_header header;
     const int ret = ar.read( header, header_size );
     if( ret != 0 ) { read_error( ar ); if( ar.fatal() ) { retval = ret; break; } }
-    if( ret != 0 || !verify_ustar_chksum( header ) )	// error or EOA
+    if( ret != 0 || !check_ustar_chksum( header ) )	// error or EOA
       {
       if( ret == 0 && block_is_zero( header, header_size ) )	// EOA
         {
@@ -461,20 +499,23 @@ int decode( const Cl_options & cl_opts )
 
     extended.fill_from_ustar( header );		// copy metadata from header
 
-    // members without name are skipped except when listing
-    if( check_skip_filename( cl_opts, name_pending, extended.path().c_str() ) )
-      retval = skip_member( ar, extended, typeflag );
-    else
-      {
-      print_removed_prefix( extended.removed_prefix );
-      if( cl_opts.program_mode == m_list )
-        retval = list_member( ar, extended, header );
-      else if( extended.path().empty() )
-        retval = skip_member( ar, extended, typeflag );
-      else if( cl_opts.program_mode == m_diff )
-        retval = compare_member( cl_opts, ar, extended, header );
-      else retval = extract_member( cl_opts, ar, extended, header );
+    try {
+      // members without name are skipped except when listing
+      if( check_skip_filename( cl_opts, name_pending, extended.path().c_str(),
+          chdir_fd ) ) retval = skip_member( ar, extended, typeflag );
+      else
+        {
+        print_removed_prefix( extended.removed_prefix );
+        if( cl_opts.program_mode == m_list )
+          retval = list_member( ar, extended, header );
+        else if( extended.path().empty() )
+          retval = skip_member( ar, extended, typeflag );
+        else if( cl_opts.program_mode == m_diff )
+          retval = compare_member( cl_opts, ar, extended, header );
+        else retval = extract_member( cl_opts, ar, extended, header );
+        }
       }
+    catch( Chdir_error & ) { retval = 1; }
     extended.reset();
     if( retval )
       { show_error( "Error is not recoverable: exiting now." ); break; }