Re: [Bug-tar] GNUTYPE_NAMES dangerous

[Sergey Poznyakoff]

Paul Eggert <address@hidden> wrote:

> To implement Sergey's suggestion to remove mangle.c, I installed the
> following patch against the main CVS branch of GNU tar.

Thank you.

> Sergey, if you see anything wrong with this please let me know.

It's ok, except that you've forgotten to remove mangle.c from
POTFILES.in.  I have fixed this in repository.

> Also, I'd be inclined to generate a new official release soon.  Shall we
> just increment the version number (to 1.16.1, say?) and ship CVS head?

Yes, these are minor changes so 1.16.1 is OK.  I will need a week or so
to flush my local changes.  In the meantime, I have installed the
following patch:

2006-11-30  Sergey Poznyakoff  <address@hidden>

        * po/POTFILES.in: Remove src/mangle.c
        * bootstrap: Implement --update-po option and .bootstrap
        local config file.

        * src/create.c (dump_dir0): Implement --exclude-tag option
        * src/tar.c: Likewise
        * doc/tar.texi (exclude): Document --exclude-tag

Index: bootstrap
===================================================================
RCS file: /cvsroot/tar/tar/bootstrap,v
retrieving revision 1.49
diff -p -u -r1.49 bootstrap
--- bootstrap   30 Nov 2006 00:14:10 -0000      1.49
+++ bootstrap   30 Nov 2006 09:19:47 -0000
@@ -49,12 +49,18 @@ Options:
  --force                  Attempt to bootstrap even if the sources seem
                           not to have been checked out.
  --skip-po                Do not download po files.
+ --update-po[=LANG]       Update po file(s) and exit.
  --cvs-user=USERNAME      Set the CVS username to be used when accessing
                           the gnulib repository.
 
-If the file .bootstrap.conf exists in the current working directory, its
+If the file bootstrap.conf exists in the current working directory, its
 contents are read as shell variables to configure the bootstrap.
 
+Local defaults can be provided by placing the file \`.bootstrap' in the
+current working directory.  The file is read after bootstrap.conf, comments
+and empty lines are removed, shell variables expanded and the result is
+prepended to the command line options. 
+
 Running without arguments will suffice in most cases.
 "
 }
@@ -159,6 +165,12 @@ copy=false
 # Override the default configuration, if necessary.
 test -r bootstrap.conf && . ./bootstrap.conf
 
+# Read local configuration file
+if [ -r .bootstrap ]; then
+  echo "$0: Reading configuration file .bootstrap"
+  eval set -- "`sed 's/#.*$//;/^$/d' .bootstrap | tr '\n' ' '` $*"
+fi
+
 # Translate configuration into internal form.
 
 # Parse options.
@@ -176,7 +188,11 @@ do
   --cvs-user=*)
     CVS_USER=`expr "$option" : '--cvs-user=\(.*\)'`;;
   --skip-po | --no-po) # --no-po is for compatibility with 'tar' tradition.
-    SKIP_PO=t;;
+    DOWNLOAD_PO=skip;;
+  --update-po=*)
+    DOWNLOAD_PO=`expr "$option" : '--update-po=\(.*\)'`;;
+  --update-po)
+    DOWNLOAD_PO=only;;
   --force)
     CVS_only_file=;;
   --copy)
@@ -194,6 +210,80 @@ fi
 
 echo "$0: Bootstrapping CVS $package..."
 
+# Get translations.
+
+get_translations() {
+  subdir=$1
+  domain=$2
+  po_file=$3
+
+  case $WGET_COMMAND in
+  '')
+    echo "$0: wget not available; skipping translations";;
+  ?*)
+    echo "$0: getting ${po_file:-translations} into $subdir for $domain..."  &&
+    case $po_file in
+    '') (cd $subdir && rm -f dummy `ls | sed -n '/\.gmo$/p; /\.po/p'`);;
+    esac &&
+  
+    $WGET_COMMAND -O "$subdir/$domain.html" "$TP_URL$domain" &&
+
+    sed -n 
's|.*"http://[^"]*/translation/teams/PO/\([^/"]*\)/'"$domain"'-\([^/"]*\)\.[^."]*\.po".*|\1.\2|p'
 <"$subdir/$domain.html" |
+    sort -k 1,1 -k 2,2n -k2,2 -k3,3n -k3,3 -k4,4n -k4,4 -k5,5n -k5.5 |
+    awk -F. '
+      { if (lang && $1 != lang) print lang, ver }
+      { lang = $1; ver = substr($0, index($0, ".") + 1) }
+      END { if (lang) print lang, ver }
+    ' | awk -v domain="$domain" -v po_file="$po_file" -v subdir="$subdir" '
+      {
+       lang = $1
+        if (po_file && po_file != (lang ".po")) next
+       
+       ver = $2
+       urlfmt = ""
+       printf "{ $WGET_COMMAND -O %s/%s.po '\'"$TP_PO_URL"'/%s/%s-%s.%s.po'\'' 
&&\n", subdir, lang, lang, domain, ver, lang
+       printf "  msgfmt -c -o /dev/null %s/%s.po || {\n", subdir, lang
+       printf "    echo >&2 '\'"$0"': omitting translation for %s'\''\n", lang
+       printf "    rm -f %s/%s.po; }; } &&\n", subdir, lang
+      }
+      END { print ":" }
+    ' | WGET_COMMAND="$WGET_COMMAND" sh;;
+  esac &&
+  ls "$subdir"/*.po 2>/dev/null |
+    sed 's|.*/||; s|\.po$||' >"$subdir/LINGUAS" &&
+  rm -f "$subdir/$domain.html"
+}
+
+case `wget --help` in
+*'--no-cache'*)
+    WGET_COMMAND='wget -nv --no-cache';;
+*'--cache=on/off'*)
+    WGET_COMMAND='wget -nv --cache=off';;
+*'--non-verbose'*)
+    WGET_COMMAND='wget -nv';;
+*)
+    WGET_COMMAND='';;
+esac
+
+case $DOWNLOAD_PO in
+'skip')
+  ;;
+'')
+  get_translations po $package || exit
+  ;;
+'only')
+  get_translations po $package
+  exit
+  ;;
+*.po)
+  get_translations po $package "$DOWNLOAD_PO"
+  exit
+  ;;
+*)
+  get_translations po $package "${DOWNLOAD_PO}.po"
+  exit
+esac
+
 # Get paxutils files.
 
 case ${PAXUTILS_SRCDIR--} in
@@ -259,65 +349,6 @@ esac
 gnulib_tool=$GNULIB_SRCDIR/gnulib-tool
 <$gnulib_tool || exit
 
-# Get translations.
-
-get_translations() {
-  subdir=$1
-  domain=$2
-
-  case $WGET_COMMAND in
-  '')
-    echo "$0: wget not available; skipping translations";;
-  ?*)
-    echo "$0: getting translations into $subdir for $domain..." &&
-
-    (cd $subdir && rm -f dummy `ls | sed -n '/\.gmo$/p; /\.po/p'`) &&
-    $WGET_COMMAND -O "$subdir/$domain.html" "$TP_URL$domain" &&
-
-    sed -n 
's|.*"http://[^"]*/translation/teams/PO/\([^/"]*\)/'"$domain"'-\([^/"]*\)\.[^."]*\.po".*|\1.\2|p'
 <"$subdir/$domain.html" |
-    sort -k 1,1 -k 2,2n -k2,2 -k3,3n -k3,3 -k4,4n -k4,4 -k5,5n -k5.5 |
-    awk -F. '
-      { if (lang && $1 != lang) print lang, ver }
-      { lang = $1; ver = substr($0, index($0, ".") + 1) }
-      END { if (lang) print lang, ver }
-    ' | awk -v domain="$domain" -v subdir="$subdir" '
-      {
-       lang = $1
-       ver = $2
-       urlfmt = ""
-       printf "{ $WGET_COMMAND -O %s/%s.po '\'"$TP_PO_URL"'/%s/%s-%s.%s.po'\'' 
&&\n", subdir, lang, lang, domain, ver, lang
-       printf "  msgfmt -c -o /dev/null %s/%s.po || {\n", subdir, lang
-       printf "    echo >&2 '\'"$0"': omitting translation for %s'\''\n", lang
-       printf "    rm -f %s/%s.po; }; } &&\n", subdir, lang
-      }
-      END { print ":" }
-    ' | WGET_COMMAND="$WGET_COMMAND" sh;;
-  esac &&
-  ls "$subdir"/*.po 2>/dev/null |
-    sed 's|.*/||; s|\.po$||' >"$subdir/LINGUAS" &&
-  rm -f "$subdir/$domain.html"
-}
-
-case $SKIP_PO in
-'')
-  case `wget --help` in
-  *'--no-cache'*)
-    WGET_COMMAND='wget -nv --no-cache';;
-  *'--cache=on/off'*)
-    WGET_COMMAND='wget -nv --cache=off';;
-  *'--non-verbose'*)
-    WGET_COMMAND='wget -nv';;
-  *)
-    WGET_COMMAND='';;
-  esac
-
-  get_translations po $package || exit
-
-  if test -d runtime-po; then
-    get_translations runtime-po $package-runtime || exit
-  fi;;
-esac
-
 symlink_to_gnulib()
 {
   src=$GNULIB_SRCDIR/$1
Index: doc/tar.texi
===================================================================
RCS file: /cvsroot/tar/tar/doc/tar.texi,v
retrieving revision 1.124
diff -p -u -r1.124 tar.texi
--- doc/tar.texi        30 Nov 2006 06:39:29 -0000      1.124
+++ doc/tar.texi        30 Nov 2006 09:20:58 -0000
@@ -2505,6 +2505,11 @@ patterns in the file @var{file}.  @xref{
 Automatically excludes all directories
 containing a cache directory tag.  @xref{exclude}.
 
address@hidden
address@hidden address@hidden
+
+Exclude all directories, containing file named @var{file}.  @xref{exclude}.
+
 @opsummary{file}
 @item address@hidden
 @itemx -f @var{archive}
@@ -6481,6 +6486,43 @@ Various applications write cache directo
 use to hold regenerable, non-precious data, so that such data can be
 more easily excluded from backups.
 
address@hidden exclude-tag
+Another option, @option{--exclude-tag}, provides a generalization of
+this concept.  It takes a single argument, a file name to look for.
+Any directory that contains this file will be excluded from the dump.
+
address@hidden @option
address@hidden exclude-tag
address@hidden address@hidden
+Causes @command{tar} to ignore directories containing @var{file}.
address@hidden table
+
+You can specify as many exclusion tags as you like.
+
+Notice, that when speaking about ``excluding the directory'' we mean
+that the files and subdirectories it contains will not get into the
+archive, while the entry for the directory itself will, e.g.:
+
address@hidden
+$ @kbd{find dir}
+dir
+dir/blues
+dir/jazz
+dir/folk
+dir/folk/tagfile
+$ @kbd{tar -cf archive.tar --extract-tag=tagfile -v}
+dir/
+dir/blues
+dir/jazz
+dir/folk/
+./tar: dir/folk/: contains a cache directory tag tagfile; not dumped
+$ @kbd{tar -tf archive.tar}
+dir/
+dir/blues
+dir/jazz
+dir/folk/
address@hidden smallexample
+
 @menu
 * problems with exclude::
 @end menu
Index: lib/.cvsignore
===================================================================
RCS file: /cvsroot/tar/tar/lib/.cvsignore,v
retrieving revision 1.41
diff -p -u -r1.41 .cvsignore
--- lib/.cvsignore      30 Nov 2006 06:27:24 -0000      1.41
+++ lib/.cvsignore      30 Nov 2006 09:20:58 -0000
@@ -135,6 +135,7 @@ paxnames.c
 pipe-safer.c
 printf-args.c
 printf-args.h
+printf-parse.c
 printf-parse.h
 quote.c
 quote.h
@@ -215,6 +216,7 @@ unsetenv.c
 utime.c
 utimens.c
 utimens.h
+vasnprintf.c
 vasnprintf.h
 verify.h
 version-etc-fsf.c
Index: m4/.cvsignore
===================================================================
RCS file: /cvsroot/tar/tar/m4/.cvsignore,v
retrieving revision 1.17
diff -p -u -r1.17 .cvsignore
--- m4/.cvsignore       1 Feb 2005 05:51:24 -0000       1.17
+++ m4/.cvsignore       30 Nov 2006 09:20:58 -0000
@@ -1,3 +1,135 @@
+*.m4
 Makefile
 Makefile.in
-*.m4
+absolute-header.m4
+alloca.m4
+allocsa.m4
+argmatch.m4
+argp.m4
+backupfile.m4
+bison.m4
+chdir-long.m4
+chown.m4
+clock_time.m4
+close-stream.m4
+closeout.m4
+codeset.m4
+d-ino.m4
+dirname.m4
+dos.m4
+double-slash-root.m4
+eealloc.m4
+eoverflow.m4
+error.m4
+exclude.m4
+exitfail.m4
+extensions.m4
+fcntl-safer.m4
+fcntl_h.m4
+fileblocks.m4
+fnmatch.m4
+fpending.m4
+ftruncate.m4
+getcwd-abort-bug.m4
+getcwd-path-max.m4
+getcwd.m4
+getdate.m4
+getdelim.m4
+getline.m4
+getopt.m4
+getpagesize.m4
+gettext.m4
+gettime.m4
+gettimeofday.m4
+glibc21.m4
+gnulib-cache.m4
+gnulib-comp.m4
+gnulib-tool.m4
+hash.m4
+human.m4
+iconv.m4
+inline.m4
+intl.m4
+intldir.m4
+intmax.m4
+intmax_t.m4
+inttostr.m4
+inttypes-pri.m4
+inttypes.m4
+inttypes_h.m4
+lchown.m4
+lib-ld.m4
+lib-link.m4
+lib-prefix.m4
+localcharset.m4
+longdouble.m4
+longlong.m4
+lstat.m4
+mbchar.m4
+mbiter.m4
+mbrtowc.m4
+mbstate_t.m4
+memchr.m4
+mempcpy.m4
+memrchr.m4
+memset.m4
+minmax.m4
+mkdtemp.m4
+mktime.m4
+modechange.m4
+nls.m4
+openat.m4
+po.m4
+progtest.m4
+quote.m4
+quotearg.m4
+regex.m4
+rmdir.m4
+rpmatch.m4
+safe-read.m4
+safe-write.m4
+save-cwd.m4
+savedir.m4
+setenv.m4
+ssize_t.m4
+stat-time.m4
+stdarg.m4
+stdbool.m4
+stdint.m4
+stdint_h.m4
+stpcpy.m4
+strcase.m4
+strchrnul.m4
+strdup.m4
+strndup.m4
+strnlen.m4
+strtoimax.m4
+strtol.m4
+strtoll.m4
+strtoul.m4
+strtoull.m4
+strtoumax.m4
+sys_stat_h.m4
+sysexits.m4
+tempname.m4
+time_r.m4
+timespec.m4
+tm_gmtoff.m4
+unistd-safer.m4
+unistd_h.m4
+unlinkdir.m4
+unlocked-io.m4
+utimbuf.m4
+utime.m4
+utimens.m4
+utimes-null.m4
+utimes.m4
+vasnprintf.m4
+vsnprintf.m4
+wchar_t.m4
+wcwidth.m4
+wint_t.m4
+xalloc.m4
+xgetcwd.m4
+xstrndup.m4
+xstrtol.m4
Index: src/create.c
===================================================================
RCS file: /cvsroot/tar/tar/src/create.c,v
retrieving revision 1.121
diff -p -u -r1.121 create.c
--- src/create.c        13 Nov 2006 10:39:51 -0000      1.121
+++ src/create.c        30 Nov 2006 09:21:06 -0000
@@ -33,6 +33,65 @@ struct link
     size_t nlink;
     char name[1];
   };
+
+struct exclude_tag
+{
+  const char *name;
+  size_t length;
+  struct exclude_tag *next;
+};
+
+static struct exclude_tag *exclude_tags;
+
+void
+add_exclude_tag (const char *name)
+{
+  struct exclude_tag *tag = xmalloc (sizeof tag[0]);
+  tag->next = exclude_tags;
+  tag->name = name;
+  tag->length = strlen (name);
+  exclude_tags = tag;
+}
+
+static bool
+check_exclude_tags (char *dirname)
+{
+  static char *tagname;
+  static size_t tagsize;
+  struct exclude_tag *tag;
+  size_t dlen = strlen (dirname);
+  char *nptr = NULL;
+  char *ret = NULL;
+  
+  for (tag = exclude_tags; tag; tag = tag->next)
+    {
+      size_t size = dlen + tag->length + 1;
+      if (size > tagsize)
+       {
+         tagsize = size;
+         tagname = xrealloc (tagname, tagsize);
+       }
+
+      if (!nptr)
+       {
+         strcpy (tagname, dirname);
+         nptr = tagname + dlen;
+       }
+      strcpy (nptr, tag->name);
+      if (access (tagname, F_OK) == 0)
+       {
+         if (verbose_option)
+           WARN ((0, 0,
+                  _("%s: contains a cache directory tag %s; not dumped"),
+                  quotearg_colon (dirname),
+                  quotearg_n (1, tag->name)));
+         return true;
+       }
+    }
+
+  return false;
+}
+
 
 /* The maximum uintmax_t value that can be represented with DIGITS digits,
    assuming that each digit is BITS_PER_DIGIT wide.  */
@@ -983,6 +1042,7 @@ dump_regular_file (int fd, struct tar_st
   return dump_status_ok;
 }
 
+
 /* Look in directory DIRNAME for a cache directory tag file
    with the magic name "CACHEDIR.TAG" and a standard header,
    as described at:
@@ -1000,7 +1060,7 @@ check_cache_directory (char *dirname)
   static char tagname[] = "CACHEDIR.TAG";
   char *tagpath;
   int fd;
-  int tag_present = false;
+  bool tag_present = false;
 
   tagpath = xmalloc (strlen (dirname) + strlen (tagname) + 1);
   strcpy (tagpath, dirname);
@@ -1124,6 +1184,9 @@ dump_dir0 (char *directory,
       return;
     }
 
+  if (check_exclude_tags (st->orig_file_name))
+    return;
+
   {
     char const *entry;
     size_t entry_len;
Index: src/tar.c
===================================================================
RCS file: /cvsroot/tar/tar/src/tar.c,v
retrieving revision 1.162
diff -p -u -r1.162 tar.c
--- src/tar.c   30 Nov 2006 00:14:11 -0000      1.162
+++ src/tar.c   30 Nov 2006 09:21:19 -0000
@@ -255,6 +255,7 @@ enum
   DELETE_OPTION,
   EXCLUDE_CACHES_OPTION,
   EXCLUDE_OPTION,
+  EXCLUDE_TAG_OPTION,
   FORCE_LOCAL_OPTION,
   GROUP_OPTION,
   HANG_OPTION,
@@ -604,6 +605,8 @@ static struct argp_option options[] = {
    N_("exclude patterns listed in FILE"), GRID+1 },
   {"exclude-caches", EXCLUDE_CACHES_OPTION, 0, 0,
    N_("exclude directories containing a cache tag"), GRID+1 },
+  {"exclude-tag", EXCLUDE_TAG_OPTION, N_("FILE"), 0,
+   N_("exclude directories containing FILE"), GRID+1 }, 
   {"no-recursion", NO_RECURSION_OPTION, 0, 0,
    N_("avoid descending automatically in directories"), GRID+1 },
   {"one-file-system", ONE_FILE_SYSTEM_OPTION, 0, 0,
@@ -1507,6 +1510,10 @@ parse_opt (int key, char *arg, struct ar
       exclude_caches_option = true;
       break;
 
+    case EXCLUDE_TAG_OPTION:
+      add_exclude_tag (arg);
+      break;
+      
     case FORCE_LOCAL_OPTION:
       force_local_option = true;
       break;


Regards,
Sergey