Merge branch 'master' into feature/nocopy

7 files changed, 177 insertions, 11 deletions

diff --git a/ChangeLog b/ChangeLog
index aa9e0f09..39d2abe9 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,23 @@
+2016-08-12         Arnold D. Robbins     <arnold@skeeve.com>
+
+	* dfa.c: Sync with GNU grep.
+
+	Unrelated:
+
+	* int_array.c: Minor text and formatting edits.
+
+2016-08-09         Andrew J. Schorr     <aschorr@telemetry-investments.com>
+
+	* awk.h: Add a comment explaining the NUMINT flag in more detail.
+	* int_array.c (standard_integer_string): New function to test whether
+	a string matches what would be produced by sprintf("%ld", <value>).
+	(is_integer): Fix bug -- if NUMBER was set, then the function was
+	accepting strnum values with nonstandard string representations. We
+	now call standard_integer_string to check that the string looks OK.
+	Also added ifdef'ed code to simplify the function by relying upon
+	force_number to parse the string, but this is disabled due to possible
+	negative performance impact.
+
 2016-08-03         Arnold D. Robbins     <arnold@skeeve.com>
 
 	Remove typed regexes until they can be done properly.
diff --git a/Checklist b/Checklist
index 6c26a4da..b33fac65 100644
--- a/Checklist
+++ b/Checklist
@@ -1,4 +1,4 @@
-Tue May  5 12:37:45 IDT 2015
+Fri Aug 12 11:39:43 IDT 2016
 ============================
 
 A checklist for making releases
@@ -39,7 +39,7 @@ Testing on
 	make maintainer-clean
 	make release
 
-	compile with tcc
+	compile with pcc
 	compile with clang
 
 	compile 32 bit tests - clang and gcc
diff --git a/awk.h b/awk.h
index c56332f5..15336c05 100644
--- a/awk.h
+++ b/awk.h
@@ -422,6 +422,14 @@ typedef struct exp_node {
 	 * and add WSTRCUR to the flags so that we don't have to do the
 	 * conversion more than once.
 	 *
+	 * The NUMINT flag may be used with a value of any type -- NUMBER,
+	 * STRING, or STRNUM. It indicates that the string representation
+	 * equals the result of sprintf("%ld", <numeric value>). So, for
+	 * example, NUMINT should NOT be set if it's a strnum or string value
+	 * where the string is " 1" or "01" or "+1" or "1.0" or "0.1E1". This
+	 * is a hint to indicate that an integer array optimization may be
+	 * used when this value appears as a subscript.
+	 *
 	 * We hope that the rest of the flags are self-explanatory. :-)
 	 */
 #		define	MALLOC	0x0001       /* stptr can be free'd, i.e. not a field node pointing into a shared buffer */
diff --git a/dfa.c b/dfa.c
index 9ff3d60f..9a557b36 100644
--- a/dfa.c
+++ b/dfa.c
@@ -3392,6 +3392,7 @@ dfaoptimize (struct dfa *d)
   free_mbdata (d);
   d->multibyte = false;
   d->dfaexec = dfaexec_sb;
+  d->fast = true;
 }
 
 static void
diff --git a/int_array.c b/int_array.c
index 9981fafe..e52bb40e 100644
--- a/int_array.c
+++ b/int_array.c
@@ -78,14 +78,50 @@ int_array_init(NODE *symbol, NODE *subs ATTRIBUTE_UNUSED)
 	return & success_node;
 }
 
+/*
+ * standard_integer_string -- check whether the string matches what
+ * sprintf("%ld", <value>) would produce. This is accomplished by accepting
+ * only strings that look like /^0$/ or /^-?[1-9][0-9]*$/. This should be
+ * faster than comparing vs. the results of actually calling sprintf.
+ */
+
+static bool
+standard_integer_string(const char *s, size_t len)
+{
+	const char *end;
+
+	if (len == 0)
+		return false;
+	if (*s == '0' && len == 1)
+		return true;
+	end = s + len;
+	/* ignore leading minus sign */
+	if (*s == '-' && ++s == end)
+		return false;
+	/* check first char is [1-9] */
+	if (*s < '1' || *s > '9')
+		return false;
+	while (++s < end) {
+		if (*s < '0' || *s > '9')
+			return false;
+	}
+	return true;
+}
+
 /* is_integer --- check if subscript is an integer */
 
 NODE **
 is_integer(NODE *symbol, NODE *subs)
 {
+#ifndef CHECK_INTEGER_USING_FORCE_NUMBER
 	long l;
+#endif
 	AWKNUM d;
 
+	if ((subs->flags & NUMINT) != 0)
+		/* quick exit */
+		return & success_node;
+
 	if (subs == Nnull_string || do_mpfr)
 		return NULL;
 
@@ -111,13 +147,40 @@ is_integer(NODE *symbol, NODE *subs)
 	if ((subs->flags & NUMINT) != 0)
 		return & success_node;
 
-	if ((subs->flags & NUMBER) != 0) {
+#ifdef CHECK_INTEGER_USING_FORCE_NUMBER
+	/*
+	 * This approach is much simpler, because we remove all of the strtol
+	 * logic below. But this may be slower in some usage cases.
+	 */
+	if ((subs->flags & NUMCUR) == 0) {
+		str2number(subs);
+
+		/* check again in case force_number set NUMINT */
+		if ((subs->flags & NUMINT) != 0)
+			return & success_node;
+	}
+#else /* CHECK_INTEGER_USING_FORCE_NUMBER */
+	if ((subs->flags & NUMCUR) != 0) {
+#endif /* CHECK_INTEGER_USING_FORCE_NUMBER */
 		d = subs->numbr;
 		if (d <= INT32_MAX && d >= INT32_MIN && d == (int32_t) d) {
-			subs->flags |= NUMINT;
-			return & success_node;
+			/*
+			 * The numeric value is an integer, but we must
+			 * protect against strings that cannot be generated
+			 * from sprintf("%ld", <subscript>). This can happen
+			 * with strnum or string values. We could skip this
+			 * check for pure NUMBER values, but unfortunately the
+			 * code does not currently distinguish between NUMBER
+			 * and strnum values.
+			 */
+			if (   (subs->flags & STRCUR) == 0
+			    || standard_integer_string(subs->stptr, subs->stlen)) {
+				subs->flags |= NUMINT;
+				return & success_node;
+			}
 		}
 		return NULL;
+#ifndef CHECK_INTEGER_USING_FORCE_NUMBER
 	}
 
 	/* a[3]=1; print "3" in a    -- true
@@ -174,6 +237,7 @@ is_integer(NODE *symbol, NODE *subs)
 	}
 
 	return NULL;
+#endif /* CHECK_INTEGER_USING_FORCE_NUMBER */
 }
 
 
diff --git a/vms/ChangeLog b/vms/ChangeLog
index 510f61cd..73b30a9a 100644
--- a/vms/ChangeLog
+++ b/vms/ChangeLog
@@ -1,8 +1,13 @@
+2016-08-08         John E. Malmberg      <wb8tyw@qsl.net>
+
+	* vmstest.com: New test (basic) ofmtstrnum
+	               New test (extra) ignrcas3 - skipped.
+	               Updated test (shlib) filefuncs, backsmalls1
+
 2016-06-16         John E. Malmberg      <wb8tyw@qsl.net>
 
 	* vmstest.com: New tests (basic) hex2, mixed1, subback
 
-
 2016-05-30         John E. Malmberg      <wb8tyw@qsl.net>
 
 	* vmstest.com: New tests (basic) fsnul1
diff --git a/vms/vmstest.com b/vms/vmstest.com
index 93e5a149..aa96ceba 100644
--- a/vms/vmstest.com
+++ b/vms/vmstest.com
@@ -50,8 +50,11 @@ $!
 $! 4.1.3.g: New tests
 $!      basic:  fsnul1, hex2, mixed1, subback
 $!	ext:    fpat4, symtab10
-$
-$
+$!
+$! 4.1.3.i: New tests
+$!	basic:  ofmtstrnum
+$!	extra:  ignrcas3
+$!
 $	echo	= "write sys$output"
 $	cmp	= "diff/Output=_NL:/Maximum=1"
 $	delsym  = "delete/symbol/local/nolog"
@@ -302,7 +305,7 @@ $		gosub list_of_tests
 $		return
 $
 $extra:		echo "extra..."
-$		list = "regtest inftest inet"
+$		list = "regtest inftest inet ignrcas3"
 $		gosub list_of_tests
 $		return
 $
@@ -563,6 +566,7 @@ $negrange:
 $nulinsrc:
 $nlstrina:
 $octsub:
+$ofmtstrnum:
 $paramtyp:
 $paramuninitglobal:
 $pcntplus:
@@ -1614,7 +1618,23 @@ $	gosub junit_report_fail_diff
 $   endif
 $   set On
 $   return
-$
+$!
+$ignrcas3:	echo "''test'"
+$   test_class = "extra"
+$   if f$search("sys$i18n_locale:el_gr_iso8859-7.locale") .nes. ""
+$   then
+$	define/user LC_ALL "el_gr_iso8859-7"
+$	define/user GAWKLOCALE "el_gr_iso8859-7"
+	AWKPATH_srcdir
+$!	goto common_without_test_in
+$	skip_reason = "VMS EL_GR_ISO8859-7 locale fails test"
+$	gosub junit_report_skip
+$   else
+$	skip_reason = "EL_GR_ISO8859-7 locale not installed"
+$	gosub junit_report_skip
+$   endif
+$   return
+$!
 $childin:	echo "''test'"
 $	test_class = "basic"
 $	cat = "type sys$input"
@@ -2712,6 +2732,10 @@ $	else
 $	    echo "''test'"
 $	endif
 $	gawk -f 'test'.awk 'test'.in >_'test'.tmp
+$	if f$search("sys$disk:[]_''test'.tmp;2") .nes. ""
+$	then
+$	    delete sys$disk:[]_'test'.tmp;2
+$	endif
 $	cmp 'test'.ok sys$disk:[]_'test'.tmp
 $	if $status
 $	then
@@ -3794,10 +3818,31 @@ $	endif
 $	return
 $!
 $filefuncs:
+$	echo "''test'"
+$	test_class = "shlib"
+$	filefunc_file = "[-]gawkapi.o"
+$	open/write gapi 'filefunc_file'
+$	close gapi
+$	set noOn
+$	AWKLIBPATH_dir
+$	gawk -v builddir="sys$disk:[-]" -
+	    -f 'test'.awk 'test'.in >_'test'.tmp 2>&1
+$	if .not. $status then call exit_code '$status' _'test'.tmp
+$	set On
+$	cmp 'test'.ok sys$disk:[]_'test'.tmp
+$	if $status
+$	then
+$	    rm _'test'.tmp;
+$	    gosub junit_report_pass
+$	else
+$	    gosub junit_report_fail_diff
+$	endif
+$	if f$search(filefunc_file) .nes. "" then rm 'filefunc_file';*
+$	return
+$!
 $fnmatch:
 $functab4:
 $ordchr:
-$ordchr2:
 $revout:
 $revtwoway:
 $time:
@@ -3822,6 +3867,29 @@ $	endif
 $	if f$search(filefunc_file) .nes. "" then rm 'filefunc_file';*
 $	return
 $!
+$ordchr2:
+$	echo "''test'"
+$	test_class = "shlib"
+$	filefunc_file = "[-]gawkapi.o"
+$	open/write gapi 'filefunc_file'
+$	close gapi
+$	set noOn
+$	AWKLIBPATH_dir
+$	gawk --load ordchr "BEGIN {print chr(ord(""z""))}" > _'test'.tmp 2>&1
+$	if .not. $status then call exit_code '$status' _'test'.tmp
+$	set On
+$	cmp 'test'.ok sys$disk:[]_'test'.tmp
+$	if $status
+$	then
+$	    rm _'test'.tmp;
+$	    gosub junit_report_pass
+$	else
+$	    gosub junit_report_fail_diff
+$	endif
+$	if f$search(filefunc_file) .nes. "" then rm 'filefunc_file';*
+$	return
+$
+$!
 $rwarray:
 $	echo "''test'"
 $	test_class = "shlib"