commit bash-20180427 snapshot

CWRU/CWRU.chlog

@@ -14019,7 +14019,7 @@ variables.c
 	- make_env_array_from_var_list: if a dynamic variable has the
 	  regenerate attribute set, call the dynamic value function to generate
 	  an updated value before placing it in the environment.  From a report
-	  about exporting LINENO from by Robert Elz <kre@bmunnari.OZ.AU>
+	  about exporting LINENO from Robert Elz <kre@bmunnari.OZ.AU>
 
 				    6/4
 				    ---
@@ -15269,3 +15269,52 @@ execute_cmd.c
 	  when trying to open a FIFO with no process having it open for
 	  reading
 
+				   4/26
+				   ----
+parse.y
+	- read_token_word: if returning REDIR_WORD for a {id}>foo construct,
+	  for example, make sure to assign the_word to yylval.word before
+	  returning, in case a recursive call to the parser overwrites it
+	  (e.g., when evaluating array indexes). From a message to
+	  austin-group-l from Stephane Chazelas <stephane.chazelas@gmail.com>
+
+lib/glob/sm_loop.c
+	- BRACKMATCH: if we have an invalid character class in an otherwise
+	  well-formed bracket expression, don't try to match each character
+	  of the (invalid) class individually; just skip over the class and
+	  move on.  From a message on the austin-group list from
+	  Stephane Chazelas <stephane.chazelas@gmail.com>
+
+				   4/27
+				   ----
+variables.c
+	- push_exported_var,push_func_var,push_temp_var: make sure to set the
+	  context correctly in the variable we bind in the previous (non-temp)
+	  scope. Report from Martijn Dekker <martijn@inlv.org>
+
+pathexp.c
+	- unquoted_glob_pattern_p: a pattern that contains a backslash can
+	  have it removed by the matching engine (since backslash is special
+	  in pattern matching), so if the pattern contains a backslash, and
+	  does not end in a backslash, we need to return true. Fixes bug
+	  reported by Robert Elz <kre@bmunnari.OZ.AU>
+
+lib/glob/gm_loop.c
+	- INTERNAL_GLOB_PATTERN_P: same change to return TRUE for a backslash
+	  that doesn't end the pattern
+
+lib/sh/timeval.c
+	- print_timeval: use locale_decpoint() instead of fixed `.' to print
+	  decimal point. Bug report in austin-group email from Joerg Schilling
+	  <Joerg.Schilling@fokus.fraunhofer.de>
+
+lib/sh/clock.c
+	- print_clock_t: use locale_depoint() in the same way as print_timeval
+
+				   4/29
+				   ----
+subst.c
+	- expand_cond_node: if special != 0, make sure to add QGLOB_CTLESC
+	  to the flags passed to quote_string_for_globbing. Same issue as the
+	  one with `case' fixed on 4/7, report from Martijn Dekker
+	  <martijn@inlv.org>

MANIFEST

@@ -537,6 +537,8 @@ po/nb.gmo		f
 po/nl.po		f
 po/pl.gmo		f
 po/pl.po		f
+po/pt.gmo		f
+po/pt.po		f
 po/pt_BR.gmo		f
 po/pt_BR.po		f
 po/ro.gmo		f
@@ -1050,6 +1052,7 @@ tests/getopts10.sub	f
 tests/glob.tests	f
 tests/glob1.sub		f
 tests/glob2.sub		f
+tests/glob3.sub		f
 tests/glob.right	f
 tests/globstar.tests	f
 tests/globstar.right	f
@@ -1360,6 +1363,7 @@ tests/varenv8.sub	f
 tests/varenv9.sub	f
 tests/varenv10.sub	f
 tests/varenv11.sub	f
+tests/varenv12.sub	f
 tests/version		f
 tests/version.mini	f
 tests/vredir.tests	f

doc/bashref.texi

@@ -8411,7 +8411,7 @@ compiled and linked, rather than changing run-time features.
 
 @table @code
 @item --enable-largefile
-Enable support for @uref{http://www.sas.com/standards/large_file/x_open.20Mar96.html,
+Enable support for @uref{http://www.unix.org/version2/whatsnew/lfs20mar.html,
 large files} if the operating system requires special compiler options
 to build programs which can access large files.  This is enabled by
 default, if the operating system provides large file support.

execute_cmd.c

@@ -776,6 +776,7 @@ execute_command_internal (command, asynchronous, pipe_in, pipe_out,
       return (last_command_exit_value = EXECUTION_FAILURE);
     }
 
+#if 0
   if (redirection_undo_list)
     {
       /* XXX - why copy here? */
@@ -784,7 +785,12 @@ execute_command_internal (command, asynchronous, pipe_in, pipe_out,
     }
   else
     my_undo_list = (REDIRECT *)NULL;
+#else
+  my_undo_list = redirection_undo_list;
+  redirection_undo_list = (REDIRECT *)NULL;
+#endif
 
+#if 0
   if (exec_redirection_undo_list)
     {
       /* XXX - why copy here? */
@@ -793,6 +799,10 @@ execute_command_internal (command, asynchronous, pipe_in, pipe_out,
     }
   else
     exec_undo_list = (REDIRECT *)NULL;
+#else
+  exec_undo_list = exec_redirection_undo_list;
+  exec_redirection_undo_list = (REDIRECT *)NULL;
+#endif
 
   if (my_undo_list || exec_undo_list)
     begin_unwind_frame ("loop_redirections");

general.h

@@ -92,15 +92,27 @@ extern char *strcpy __P((char *, const char *));
 /* Nonzero if the integer type T is signed.  */
 #define TYPE_SIGNED(t) (! ((t) 0 < (t) -1))
 
+/* The width in bits of the integer type or expression T.
+   Padding bits are not supported; this is checked at compile-time below.  */
+#define TYPE_WIDTH(t) (sizeof (t) * CHAR_BIT)
+
+/* Bound on length of the string representing an unsigned integer
+   value representable in B bits.  log10 (2.0) < 146/485.  The
+      smallest value of B where this bound is not tight is 2621.  */
+#define INT_BITS_STRLEN_BOUND(b) (((b) * 146 + 484) / 485)
+
 /* Bound on length of the string representing an integer value of type T.
    Subtract one for the sign bit if T is signed;
    302 / 1000 is log10 (2) rounded up;
    add one for integer division truncation;
    add one more for a minus sign if t is signed.  */
 #define INT_STRLEN_BOUND(t) \
-  ((sizeof (t) * CHAR_BIT - TYPE_SIGNED (t)) * 302 / 1000 \
+  ((TYPE_WIDTH (t) - TYPE_SIGNED (t)) * 302 / 1000 \
    + 1 + TYPE_SIGNED (t))
 
+/* Bound on buffer size needed to represent an integer type or expression T,
+   including the terminating null.  */
+#define INT_BUFSIZE_BOUND(t) (INT_STRLEN_BOUND (t) + 1)
 
 /* Define exactly what a legal shell identifier consists of. */
 #define legal_variable_starter(c) (ISALPHA(c) || (c == '_'))

lib/glob/glob_loop.c

@@ -54,8 +54,11 @@ INTERNAL_GLOB_PATTERN_P (pattern)
 	continue;
 
       case L('\\'):
-	if (*p++ == L('\0'))
-	  return 0;
+	/* Don't let the pattern end in a backslash (GMATCH returns no match
+	   if the pattern ends in a backslash anyway), but otherwise return 1,
+	   since the matching engine uses backslash as an escape character
+	   and it can be removed. */
+	return (*p != L('\0'));
       }
 
   return 0;

lib/glob/sm_loop.c

@@ -458,7 +458,18 @@ BRACKMATCH (p, test, flags)
 		  pc = IS_CCLASS (orig_test, (XCHAR *)ccname);
 		}
 	      if (pc == -1)
-		pc = 0;
+		{
+		  /* CCNAME is not a valid character class in the current
+		     locale. In addition to noting no match (pc = 0), we have
+		     a choice about what to do with the invalid charclass.
+		     Posix leaves the behavior unspecified, but we're going
+		     to skip over the charclass and keep going instead of
+		     testing ORIG_TEST against each character in the class
+		     string. If we don't want to do that, take out the update
+		     of P. */
+		  pc = 0;
+		  p = close + 2;
+		}
 	      else
 		p = close + 2;		/* move past the closing `]' */
 

lib/sh/clock.c

@@ -32,7 +32,13 @@
 #include <stdio.h>
 #include <stdc.h>
 
-extern long get_clk_tck __P((void));
+#include <bashintl.h>
+
+#ifndef locale_decpoint
+extern int locale_decpoint PARAMS((void));
+#endif
+
+extern long get_clk_tck PARAMS((void));
 
 void
 clock_t_to_secs (t, sp, sfp)
@@ -76,6 +82,6 @@ print_clock_t (fp, t)
   minutes = timestamp / 60;
   seconds = timestamp % 60;
 
-  fprintf (fp, "%ldm%d.%03ds",  minutes, seconds, seconds_fraction);
+  fprintf (fp, "%ldm%d%c%03ds",  minutes, seconds, locale_decpoint(), seconds_fraction);
 }
 #endif /* HAVE_TIMES */

lib/sh/timeval.c

@@ -25,6 +25,13 @@
 #include <sys/types.h>
 #include <posixtime.h>
 
+#include <bashintl.h>
+#include <stdc.h>
+
+#ifndef locale_decpoint
+extern int locale_decpoint PARAMS((void));
+#endif
+
 #include <stdio.h>
 
 struct timeval *
@@ -140,6 +147,6 @@ print_timeval (fp, tvp)
   minutes = timestamp / 60;
   seconds = timestamp % 60;
 
-  fprintf (fp, "%ldm%d.%03ds",  minutes, seconds, seconds_fraction);
+  fprintf (fp, "%ldm%d%c%03ds",  minutes, seconds, locale_decpoint (), seconds_fraction);
 }
 #endif /* HAVE_TIMEVAL */

parse.y

@@ -5295,6 +5295,7 @@ got_token:
 
   yylval.word = the_word;
 
+  /* should we check that quoted == 0 as well? */
   if (token[0] == '{' && token[token_index-1] == '}' &&
       (character == '<' || character == '>'))
     {
@@ -5307,7 +5308,8 @@ got_token:
 #endif
 	{
 	  strcpy (the_word->word, token+1);
-/*itrace("read_token_word: returning REDIR_WORD for %s", the_word->word);*/
+/* itrace("read_token_word: returning REDIR_WORD for %s", the_word->word); */
+	  yylval.word = the_word;	/* accommodate recursive call */
 	  return (REDIR_WORD);
 	}
       else

pathexp.c

@@ -96,8 +96,13 @@ unquoted_glob_pattern_p (string)
 	    return (1);
 	  continue;
 
-	case CTLESC:
+	/* A pattern can't end with a backslash, but a backslash in the pattern
+	   can be removed by the matching engine, so we have to run it through
+	   globbing. */
 	case '\\':
+	  return (*string != 0);
+	 	  
+	case CTLESC:
 	  if (*string++ == '\0')
 	    return (0);
 	}
@@ -172,10 +177,11 @@ glob_char_p (s)
    is performed, (flags & QGLOB_CVTNULL) should be 0; if called when quote
    removal has not been done (for example, before attempting to match a
    pattern while executing a case statement), flags should include
-   QGLOB_CVTNULL.  If flags includes QGLOB_FILENAME, appropriate quoting
-   to match a filename should be performed.  QGLOB_REGEXP means we're
-   quoting for a Posix ERE (for [[ string =~ pat ]]) and that requires
-   some special handling. */
+   QGLOB_CVTNULL.  If flags includes QGLOB_CTLESC, we need to remove CTLESC
+   quoting CTLESC or CTLNUL (as if dequote_string were called).  If flags
+   includes QGLOB_FILENAME, appropriate quoting to match a filename should be
+   performed.  QGLOB_REGEXP means we're quoting for a Posix ERE (for
+   [[ string =~ pat ]]) and that requires some special handling. */
 char *
 quote_string_for_globbing (pathname, qflags)
      const char *pathname;

po/LINGUAS

@@ -1,2 +1,2 @@
 # Set of available languages.
-en@quot en@boldquot af bg ca cs da de el eo es et fi fr ga gl hr hu id it ja lt nb nl pl pt_BR ro ru sk sl sr sv tr uk vi zh_CN zh_TW
+en@quot en@boldquot af bg ca cs da de el eo es et fi fr ga gl hr hu id it ja lt nb nl pl pt pt_BR ro ru sk sl sr sv tr uk vi zh_CN zh_TW

print_cmd.c

@@ -1475,7 +1475,7 @@ cprintf (control, va_alist)
 #endif
 {
   register const char *s;
-  char char_arg[2], *argp, intbuf[INT_STRLEN_BOUND (int) + 1];
+  char char_arg[2], *argp, intbuf[INT_STRLEN_BOUND (unsigned int) + 1];
   int digit_arg, arg_len, c;
   va_list args;
 
@@ -1520,7 +1520,7 @@ cprintf (control, va_alist)
 	      digit_arg = va_arg (args, int);
 	      if (digit_arg < 0)
 		{
-		  sprintf (intbuf, "%u", (unsigned)-1);
+		  sprintf (intbuf, "%u", (unsigned int)-1);
 		  argp = intbuf;
 		}
 	      else

subst.c

@@ -3632,7 +3632,7 @@ cond_expand_word (w, special)
 	  /* Need to figure out whether or not we should call dequote_escapes
 	     or a new dequote_ctlnul function here, and under what
 	     circumstances. */
-	  qflags = QGLOB_CVTNULL;
+	  qflags = QGLOB_CVTNULL|QGLOB_CTLESC;
 	  if (special == 2)
 	    qflags |= QGLOB_REGEXP;
 	  word_list_remove_quoted_nulls (l);

tests/RUN-ONE-TEST

@@ -1,4 +1,4 @@
-BUILD_DIR=/usr/local/build/chet/bash/bash-current
+BUILD_DIR=/usr/local/build/bash/bash-current
 THIS_SH=$BUILD_DIR/bash
 PATH=$PATH:$BUILD_DIR
 

tests/case3.sub

@@ -35,3 +35,6 @@ for c in $'\1' $'\2' $'\177'; do
 	testmatch "x${c}" "\\x\\${c}"
 	testmatch "x${c}" "x\\${c}"
 done
+
+match() { case $1 in ( $2 ) ;; ( * ) return 1 ;; esac; } 
+match $'? *x\1y\177z' $'??\\*\\x\\\1\\y\\\177\\z' || echo bad 6

tests/cond.tests

@@ -180,6 +180,9 @@ echo ${BASH_REMATCH[@]}
 if [[ "123abc" == *?(a)bc ]]; then echo ok 42; else echo bad 42; fi
 if [[ "123abc" == *?(a)bc ]]; then echo ok 43; else echo bad 43; fi
 
+match() { [[ $1 == $2 ]]; }
+match $'? *x\1y\177z' $'??\\*\\x\\\1\\y\\\177\\z' || echo bad 44
+
 foo=""
 [[ bar == *"${foo,,}"* ]] && echo ok 1
 [[ bar == *${foo,,}* ]] && echo ok 2

tests/glob.right

@@ -13,6 +13,49 @@ argv[1] = <a
 0000004
 ok 6
 ok 7
+invalid bracket expression
+== LANG=C ==
+[[:alpha:]
+ok 1
+[a
+[[:alpha:]
+ok 2
+ok 3
+== LANG=en_US.UTF-8 ==
+[[:alpha:]
+ok 1
+[a
+[[:alpha:]
+ok 2
+ok 3
+invalid character class
+== LANG=C ==
+ok 1
+ok 2
+ok 3
+ok 4
+ok 5
+ok 6
+== LANG=en_US.UTF-8 ==
+ok 1
+ok 2
+ok 3
+ok 4
+ok 5
+ok 6
+invalid collating symbols
+== LANG=C ==
+ok 1
+ok 2
+ok 3
+ok 4
+ok 5
+== LANG=en_US.UTF-8 ==
+ok 1
+ok 2
+ok 3
+ok 4
+ok 5
 argv[1] = <a>
 argv[2] = <abc>
 argv[3] = <abd>
@@ -27,7 +70,7 @@ argv[2] = <abc>
 argv[3] = <abd>
 argv[4] = <abe>
 tmp/l1 tmp/l2 tmp/*4 tmp/l3
-./glob.tests: line 45: no match: tmp/*4
+./glob.tests: line 46: no match: tmp/*4
 argv[1] = <bdir/>
 argv[1] = <*>
 argv[1] = <a*>

tests/glob.tests

@@ -10,6 +10,7 @@ expect()
 # First, a test that bash-2.01.1 fails
 ${THIS_SH} ./glob1.sub
 ${THIS_SH} ./glob2.sub
+${THIS_SH} ./glob3.sub
 
 MYDIR=$PWD	# save where we are
 

tests/glob3.sub

@@ -0,0 +1,123 @@
+: ${TMPDIR:=/var/tmp}
+cd $TMPDIR
+
+matchfunc()
+{
+	echo == LANG=$LANG ==
+
+	touch a p
+	echo [[:alpha:]
+	rm a p
+
+	case l in
+	[[:alpha:])	echo bad 1;;
+	*)		echo ok 1;;
+	esac
+
+	touch '[a' '[x'
+	echo [[:alpha:]
+	rm '[a'
+	echo [[:alpha:]
+	rm '[x'
+
+	case [a in
+	[[:alpha:])	echo ok 2;;
+	*)		echo bad 2;;
+	esac
+
+	case [x in
+	[[:alpha:])	echo bad 3;;
+	*)		echo ok 3;;
+	esac
+}
+
+echo invalid bracket expression
+export LANG=C
+matchfunc
+
+export LANG=en_US.UTF-8
+matchfunc
+
+unset -f matchfunc
+
+matchfunc()
+{
+	echo == LANG=$LANG ==
+
+	case a] in
+	[[:aleph:]])	echo bad 1;;
+	*)		echo ok 1;;
+	esac
+
+	case a in
+	[[:aleph:]])	echo bad 2;;
+	*)		echo ok 2;;
+	esac
+
+	case a] in
+	[[:"alpha":]])	echo bad 3;;
+	*)		echo ok 3;;
+	esac
+
+	case a in
+	[[:"alpha":]])	echo bad 4;;
+	*)		echo ok 4;;
+	esac
+
+	case a in
+	[abc[:foo:]])	echo ok 5;;
+	*)		echo bad 5 ;;
+	esac
+
+	case a in
+	[[:foo:]abc])	echo ok 6;;
+	*)		echo bad 6 ;;
+	esac
+}
+
+echo invalid character class
+export LANG=C
+matchfunc
+
+export LANG=en_US.UTF-8
+matchfunc
+
+unset -f matchfunc
+
+matchfunc()
+{
+	echo == LANG=$LANG ==
+
+	case h in
+	[[.hyphen.])	echo bad 1;;
+	*)		echo ok 1;;
+	esac
+
+	case - in
+	[[.hyphen.]])	echo ok 2;;
+	*)		echo bad 2;;
+	esac
+
+	case slash in
+	[[.slash.]])	echo bad 3;;
+	*)		echo ok 3;;
+	esac
+
+	case a in
+	[abc[.nonsense.]])	echo ok 4;;
+	*)		echo bad 4 ;;
+	esac
+
+	case a in
+	[[.nonsense.]abc])	echo ok 5;;
+	*)		echo bad 5 ;;
+	esac
+}
+
+echo invalid collating symbols
+
+export LANG=C
+matchfunc
+
+export LANG=en_US.UTF-8
+matchfunc

tests/posixpat.tests

@@ -20,13 +20,15 @@ case 9 in
 [![:alpha:]])	echo ok 4;;
 esac
 
-# invalid character class expressions are just characters to be matched
 case a in
 [:al:])		echo ok 5;;
 esac
 
+# invalid character class expressions are no longer just characters to be
+# matched
 case a in
-[[:al:])	echo ok 6;;
+[[:al:])	echo bad 6;;
+*)		echo ok 6;;
 esac
 
 case '!' in
@@ -230,4 +232,3 @@ case a in
 [[=b=])		echo oops;;	# an incomplete equiv class is just a string
 *)		echo ok 3;;
 esac
-

tests/varenv.sh

@@ -231,5 +231,8 @@ ${THIS_SH} ./varenv10.sub
 # tests of compound assignments in function scope
 ${THIS_SH} ./varenv11.sub
 
+# temporary environment variable propagation in posix mode
+${THIS_SH} ./varenv12.sub
+
 # make sure variable scoping is done right
 tt() { typeset a=b;echo a=$a; };a=z;echo a=$a;tt;echo a=$a

tests/varenv12.sub

@@ -0,0 +1,4 @@
+set -o posix
+fn() { foo=abc : ; typeset +x foo; printenv|grep ^foo=; }
+
+fn