sub TEST {
my $string = shift;
+ my $name = shift;
my $t = eval $string;
++$TNUM;
$t =~ s/([A-Z]+)\(0x[0-9a-f]+\)/$1(0xdeadbeef)/g
$t = join("\n",sort(split(/\n/,$t)));
$t =~ s/\,$//mg;
}
- print( ($t eq $WANT and not $@) ? "ok $TNUM\n"
- : "not ok $TNUM\n--Expected--\n$WANT\n--Got--\n$@$t\n");
+ $name = $name ? " - $name" : '';
+ print( ($t eq $WANT and not $@) ? "ok $TNUM$name\n"
+ : "not ok $TNUM$name\n--Expected--\n$WANT\n--Got--\n$@$t\n");
++$TNUM;
eval "$t";
: "not ok $TNUM\n--Expected--\n$WANT\n--Got--\n$@$t\n");
}
+# Force Data::Dumper::Dump to use perl. We test Dumpxs explicitly by calling
+# it direct. Out here it lets us knobble the next if to test that the perl
+# only tests do work (and count correctly)
+$Data::Dumper::Useperl = 1;
if (defined &Data::Dumper::Dumpxs) {
print "### XS extension loaded, will run XS tests\n";
- $TMAX = 186; $XS = 1;
+ $TMAX = 339; $XS = 1;
}
else {
print "### XS extensions not loaded, will NOT run XS tests\n";
- $TMAX = 93; $XS = 0;
+ $TMAX = 171; $XS = 0;
}
print "1..$TMAX\n";
+#XXXif (0) {
#############
#############
TEST q(Data::Dumper->new([$b],['b'])->Purity(1)->Dumpxs;)
if $XS;
}
+
+{
+ $a = "\x{09c10}";
+############# 187
+## XS code was adding an extra \0
+ $WANT = <<'EOT';
+#$a = "\x{9c10}";
+EOT
+
+ TEST q(Data::Dumper->Dump([$a], ['a'])), "\\x{9c10}";
+ TEST q(Data::Dumper->Dumpxs([$a], ['a'])), "XS \\x{9c10}"
+ if $XS;
+
+}
+
+{
+ $i = 0;
+ $a = { map { ("$_$_$_", ++$i) } 'I'..'Q' };
+ local $Data::Dumper::Sortkeys = 1;
+
+############# 193
+##
+ $WANT = <<'EOT';
+#$VAR1 = {
+# III => 1,
+# JJJ => 2,
+# KKK => 3,
+# LLL => 4,
+# MMM => 5,
+# NNN => 6,
+# OOO => 7,
+# PPP => 8,
+# QQQ => 9
+#};
+EOT
+
+TEST q(Data::Dumper->new([$a])->Dump;);
+TEST q(Data::Dumper->new([$a])->Dumpxs;)
+ if $XS;
+}
+
+{
+ $i = 5;
+ $c = { map { (++$i, "$_$_$_") } 'I'..'Q' };
+ local $Data::Dumper::Sortkeys = \&sort199;
+ sub sort199 {
+ my $hash = shift;
+ return [ sort { $b <=> $a } keys %$hash ];
+ }
+
+############# 199
+##
+ $WANT = <<'EOT';
+#$VAR1 = {
+# 14 => 'QQQ',
+# 13 => 'PPP',
+# 12 => 'OOO',
+# 11 => 'NNN',
+# 10 => 'MMM',
+# 9 => 'LLL',
+# 8 => 'KKK',
+# 7 => 'JJJ',
+# 6 => 'III'
+#};
+EOT
+
+# perl code does keys and values as numbers if possible
+TEST q(Data::Dumper->new([$c])->Dump;);
+# XS code always does them as strings
+$WANT =~ s/ (\d+)/ '$1'/gs;
+TEST q(Data::Dumper->new([$c])->Dumpxs;)
+ if $XS;
+}
+
+{
+ $i = 5;
+ $c = { map { (++$i, "$_$_$_") } 'I'..'Q' };
+ $d = { reverse %$c };
+ local $Data::Dumper::Sortkeys = \&sort205;
+ sub sort205 {
+ my $hash = shift;
+ return [
+ $hash eq $c ? (sort { $a <=> $b } keys %$hash)
+ : (reverse sort keys %$hash)
+ ];
+ }
+
+############# 205
+##
+ $WANT = <<'EOT';
+#$VAR1 = [
+# {
+# 6 => 'III',
+# 7 => 'JJJ',
+# 8 => 'KKK',
+# 9 => 'LLL',
+# 10 => 'MMM',
+# 11 => 'NNN',
+# 12 => 'OOO',
+# 13 => 'PPP',
+# 14 => 'QQQ'
+# },
+# {
+# QQQ => 14,
+# PPP => 13,
+# OOO => 12,
+# NNN => 11,
+# MMM => 10,
+# LLL => 9,
+# KKK => 8,
+# JJJ => 7,
+# III => 6
+# }
+#];
+EOT
+
+TEST q(Data::Dumper->new([[$c, $d]])->Dump;);
+$WANT =~ s/ (\d+)/ '$1'/gs;
+TEST q(Data::Dumper->new([[$c, $d]])->Dumpxs;)
+ if $XS;
+}
+
+{
+ local $Data::Dumper::Deparse = 1;
+ local $Data::Dumper::Indent = 2;
+
+############# 211
+##
+ $WANT = <<'EOT';
+#$VAR1 = {
+# foo => sub {
+# print 'foo';
+# }
+# };
+EOT
+
+ TEST q(Data::Dumper->new([{ foo => sub { print "foo"; } }])->Dump);
+}
+
+############# 214
+##
+
+# This is messy.
+# The controls (bare numbers) are stored either as integers or floating point.
+# [depending on whether the tokeniser sees things like ".".
+# The peephole optimiser only runs for constant folding, not single constants,
+# so I already have some NVs, some IVs
+# The string versions are not. They are all PV
+
+# This is arguably all far too chummy with the implementation, but I really
+# want to ensure that we don't go wrong when flags on scalars get as side
+# effects of reading them.
+
+# These tests are actually testing the precise output of the current
+# implementation, so will most likely fail if the implementation changes,
+# even if the new implementation produces different but correct results.
+# It would be nice to test for wrong answers, but I can't see how to do that,
+# so instead I'm checking for unexpected answers. (ie -2 becoming "-2" is not
+# wrong, but I can't see an easy, reliable way to code that knowledge)
+
+# Numbers (seen by the tokeniser as numbers, stored as numbers.
+ @numbers =
+ (
+ 0, +1, -2, 3.0, +4.0, -5.0, 6.5, +7.5, -8.5,
+ 9, +10, -11, 12.0, +13.0, -14.0, 15.5, +16.25, -17.75,
+ );
+# Strings
+ @strings =
+ (
+ "0", "+1", "-2", "3.0", "+4.0", "-5.0", "6.5", "+7.5", "-8.5", " 9",
+ " +10", " -11", " 12.0", " +13.0", " -14.0", " 15.5", " +16.25", " -17.75",
+ );
+
+# The perl code always does things the same way for numbers.
+ $WANT_PL_N = <<'EOT';
+#$VAR1 = 0;
+#$VAR2 = 1;
+#$VAR3 = -2;
+#$VAR4 = 3;
+#$VAR5 = 4;
+#$VAR6 = -5;
+#$VAR7 = '6.5';
+#$VAR8 = '7.5';
+#$VAR9 = '-8.5';
+#$VAR10 = 9;
+#$VAR11 = 10;
+#$VAR12 = -11;
+#$VAR13 = 12;
+#$VAR14 = 13;
+#$VAR15 = -14;
+#$VAR16 = '15.5';
+#$VAR17 = '16.25';
+#$VAR18 = '-17.75';
+EOT
+# The perl code knows that 0 and -2 stringify exactly back to the strings,
+# so it dumps them as numbers, not strings.
+ $WANT_PL_S = <<'EOT';
+#$VAR1 = 0;
+#$VAR2 = '+1';
+#$VAR3 = -2;
+#$VAR4 = '3.0';
+#$VAR5 = '+4.0';
+#$VAR6 = '-5.0';
+#$VAR7 = '6.5';
+#$VAR8 = '+7.5';
+#$VAR9 = '-8.5';
+#$VAR10 = ' 9';
+#$VAR11 = ' +10';
+#$VAR12 = ' -11';
+#$VAR13 = ' 12.0';
+#$VAR14 = ' +13.0';
+#$VAR15 = ' -14.0';
+#$VAR16 = ' 15.5';
+#$VAR17 = ' +16.25';
+#$VAR18 = ' -17.75';
+EOT
+
+# The XS code differs.
+# These are the numbers as seen by the tokeniser. Constants aren't folded
+# (which makes IVs where possible) so values the tokeniser thought were
+# floating point are stored as NVs. The XS code outputs these as strings,
+# but as it has converted them from NVs, leading + signs will not be there.
+ $WANT_XS_N = <<'EOT';
+#$VAR1 = 0;
+#$VAR2 = 1;
+#$VAR3 = -2;
+#$VAR4 = '3';
+#$VAR5 = '4';
+#$VAR6 = '-5';
+#$VAR7 = '6.5';
+#$VAR8 = '7.5';
+#$VAR9 = '-8.5';
+#$VAR10 = 9;
+#$VAR11 = 10;
+#$VAR12 = -11;
+#$VAR13 = '12';
+#$VAR14 = '13';
+#$VAR15 = '-14';
+#$VAR16 = '15.5';
+#$VAR17 = '16.25';
+#$VAR18 = '-17.75';
+EOT
+
+# These are the strings as seen by the tokeniser. The XS code will output
+# these for all cases except where the scalar has been used in integer context
+ $WANT_XS_S = <<'EOT';
+#$VAR1 = '0';
+#$VAR2 = '+1';
+#$VAR3 = '-2';
+#$VAR4 = '3.0';
+#$VAR5 = '+4.0';
+#$VAR6 = '-5.0';
+#$VAR7 = '6.5';
+#$VAR8 = '+7.5';
+#$VAR9 = '-8.5';
+#$VAR10 = ' 9';
+#$VAR11 = ' +10';
+#$VAR12 = ' -11';
+#$VAR13 = ' 12.0';
+#$VAR14 = ' +13.0';
+#$VAR15 = ' -14.0';
+#$VAR16 = ' 15.5';
+#$VAR17 = ' +16.25';
+#$VAR18 = ' -17.75';
+EOT
+
+# These are the numbers as IV-ized by &
+# These will differ from WANT_XS_N because now IV flags will be set on all
+# values that were actually integer, and the XS code will then output these
+# as numbers not strings.
+ $WANT_XS_I = <<'EOT';
+#$VAR1 = 0;
+#$VAR2 = 1;
+#$VAR3 = -2;
+#$VAR4 = 3;
+#$VAR5 = 4;
+#$VAR6 = -5;
+#$VAR7 = '6.5';
+#$VAR8 = '7.5';
+#$VAR9 = '-8.5';
+#$VAR10 = 9;
+#$VAR11 = 10;
+#$VAR12 = -11;
+#$VAR13 = 12;
+#$VAR14 = 13;
+#$VAR15 = -14;
+#$VAR16 = '15.5';
+#$VAR17 = '16.25';
+#$VAR18 = '-17.75';
+EOT
+
+# Some of these tests will be redundant.
+@numbers_s = @numbers_i = @numbers_is = @numbers_n = @numbers_ns = @numbers_ni
+ = @numbers_nis = @numbers;
+@strings_s = @strings_i = @strings_is = @strings_n = @strings_ns = @strings_ni
+ = @strings_nis = @strings;
+# Use them in an integer context
+foreach (@numbers_i, @numbers_ni, @numbers_nis, @numbers_is,
+ @strings_i, @strings_ni, @strings_nis, @strings_is) {
+ my $b = sprintf "%d", $_;
+}
+# Use them in a floating point context
+foreach (@numbers_n, @numbers_ni, @numbers_nis, @numbers_ns,
+ @strings_n, @strings_ni, @strings_nis, @strings_ns) {
+ my $b = sprintf "%e", $_;
+}
+# Use them in a string context
+foreach (@numbers_s, @numbers_is, @numbers_nis, @numbers_ns,
+ @strings_s, @strings_is, @strings_nis, @strings_ns) {
+ my $b = sprintf "%s", $_;
+}
+
+# use Devel::Peek; Dump ($_) foreach @vanilla_c;
+
+$WANT=$WANT_PL_N;
+TEST q(Data::Dumper->new(\@numbers)->Dump), 'Numbers';
+TEST q(Data::Dumper->new(\@numbers_s)->Dump), 'Numbers PV';
+TEST q(Data::Dumper->new(\@numbers_i)->Dump), 'Numbers IV';
+TEST q(Data::Dumper->new(\@numbers_is)->Dump), 'Numbers IV,PV';
+TEST q(Data::Dumper->new(\@numbers_n)->Dump), 'Numbers NV';
+TEST q(Data::Dumper->new(\@numbers_ns)->Dump), 'Numbers NV,PV';
+TEST q(Data::Dumper->new(\@numbers_ni)->Dump), 'Numbers NV,IV';
+TEST q(Data::Dumper->new(\@numbers_nis)->Dump), 'Numbers NV,IV,PV';
+$WANT=$WANT_PL_S;
+TEST q(Data::Dumper->new(\@strings)->Dump), 'Strings';
+TEST q(Data::Dumper->new(\@strings_s)->Dump), 'Strings PV';
+TEST q(Data::Dumper->new(\@strings_i)->Dump), 'Strings IV';
+TEST q(Data::Dumper->new(\@strings_is)->Dump), 'Strings IV,PV';
+TEST q(Data::Dumper->new(\@strings_n)->Dump), 'Strings NV';
+TEST q(Data::Dumper->new(\@strings_ns)->Dump), 'Strings NV,PV';
+TEST q(Data::Dumper->new(\@strings_ni)->Dump), 'Strings NV,IV';
+TEST q(Data::Dumper->new(\@strings_nis)->Dump), 'Strings NV,IV,PV';
+if ($XS) {
+ $WANT=$WANT_XS_N;
+ TEST q(Data::Dumper->new(\@numbers)->Dumpxs), 'XS Numbers';
+ TEST q(Data::Dumper->new(\@numbers_s)->Dumpxs), 'XS Numbers PV';
+ $WANT=$WANT_XS_I;
+ TEST q(Data::Dumper->new(\@numbers_i)->Dumpxs), 'XS Numbers IV';
+ TEST q(Data::Dumper->new(\@numbers_is)->Dumpxs), 'XS Numbers IV,PV';
+ $WANT=$WANT_XS_N;
+ TEST q(Data::Dumper->new(\@numbers_n)->Dumpxs), 'XS Numbers NV';
+ TEST q(Data::Dumper->new(\@numbers_ns)->Dumpxs), 'XS Numbers NV,PV';
+ $WANT=$WANT_XS_I;
+ TEST q(Data::Dumper->new(\@numbers_ni)->Dumpxs), 'XS Numbers NV,IV';
+ TEST q(Data::Dumper->new(\@numbers_nis)->Dumpxs), 'XS Numbers NV,IV,PV';
+
+ $WANT=$WANT_XS_S;
+ TEST q(Data::Dumper->new(\@strings)->Dumpxs), 'XS Strings';
+ TEST q(Data::Dumper->new(\@strings_s)->Dumpxs), 'XS Strings PV';
+ # This one used to really mess up. New code actually emulates the .pm code
+ $WANT=$WANT_PL_S;
+ TEST q(Data::Dumper->new(\@strings_i)->Dumpxs), 'XS Strings IV';
+ TEST q(Data::Dumper->new(\@strings_is)->Dumpxs), 'XS Strings IV,PV';
+ $WANT=$WANT_XS_S;
+ TEST q(Data::Dumper->new(\@strings_n)->Dumpxs), 'XS Strings NV';
+ TEST q(Data::Dumper->new(\@strings_ns)->Dumpxs), 'XS Strings NV,PV';
+ # This one used to really mess up. New code actually emulates the .pm code
+ $WANT=$WANT_PL_S;
+ TEST q(Data::Dumper->new(\@strings_ni)->Dumpxs), 'XS Strings NV,IV';
+ TEST q(Data::Dumper->new(\@strings_nis)->Dumpxs), 'XS Strings NV,IV,PV';
+}
+
+{
+ $a = "1\n";
+############# 310
+## Perl code was using /...$/ and hence missing the \n.
+ $WANT = <<'EOT';
+my $VAR1 = '42
+';
+EOT
+
+ # Can't pad with # as the output has an embedded newline.
+ local $Data::Dumper::Pad = "my ";
+ TEST q(Data::Dumper->Dump(["42\n"])), "number with trailing newline";
+ TEST q(Data::Dumper->Dumpxs(["42\n"])), "XS number with trailing newline"
+ if $XS;
+}
+
+{
+ @a = (
+ 999999999,
+ 1000000000,
+ 9999999999,
+ 10000000000,
+ -999999999,
+ -1000000000,
+ -9999999999,
+ -10000000000,
+ 4294967295,
+ 4294967296,
+ -2147483648,
+ -2147483649,
+ );
+############# 316
+## Perl code flips over at 10 digits.
+ $WANT = <<'EOT';
+#$VAR1 = 999999999;
+#$VAR2 = '1000000000';
+#$VAR3 = '9999999999';
+#$VAR4 = '10000000000';
+#$VAR5 = -999999999;
+#$VAR6 = '-1000000000';
+#$VAR7 = '-9999999999';
+#$VAR8 = '-10000000000';
+#$VAR9 = '4294967295';
+#$VAR10 = '4294967296';
+#$VAR11 = '-2147483648';
+#$VAR12 = '-2147483649';
+EOT
+
+ TEST q(Data::Dumper->Dump(\@a)), "long integers";
+
+ if ($XS) {
+## XS code flips over at 11 characters ("-" is a char) or larger than int.
+ if (~0 == 0xFFFFFFFF) {
+ # 32 bit system
+ $WANT = <<'EOT';
+#$VAR1 = 999999999;
+#$VAR2 = 1000000000;
+#$VAR3 = '9999999999';
+#$VAR4 = '10000000000';
+#$VAR5 = -999999999;
+#$VAR6 = '-1000000000';
+#$VAR7 = '-9999999999';
+#$VAR8 = '-10000000000';
+#$VAR9 = 4294967295;
+#$VAR10 = '4294967296';
+#$VAR11 = '-2147483648';
+#$VAR12 = '-2147483649';
+EOT
+ } else {
+ $WANT = <<'EOT';
+#$VAR1 = 999999999;
+#$VAR2 = 1000000000;
+#$VAR3 = 9999999999;
+#$VAR4 = '10000000000';
+#$VAR5 = -999999999;
+#$VAR6 = '-1000000000';
+#$VAR7 = '-9999999999';
+#$VAR8 = '-10000000000';
+#$VAR9 = 4294967295;
+#$VAR10 = 4294967296;
+#$VAR11 = '-2147483648';
+#$VAR12 = '-2147483649';
+EOT
+ }
+ TEST q(Data::Dumper->Dumpxs(\@a)), "XS long integers";
+ }
+}
+
+#XXX}
+{
+ $b = "Bad. XS didn't escape dollar sign";
+############# 322
+ $WANT = <<"EOT"; # Careful. This is '' string written inside '' here doc
+#\$VAR1 = '\$b\"\@\\\\\xA3';
+EOT
+
+ $a = "\$b\"\@\\\xA3\x{100}";
+ chop $a;
+ TEST q(Data::Dumper->Dump([$a])), "utf8 flag with \" and \$";
+ if ($XS) {
+ $WANT = <<'EOT'; # While this is "" string written inside "" here doc
+#$VAR1 = "\$b\"\@\\\x{a3}";
+EOT
+ TEST q(Data::Dumper->Dumpxs([$a])), "XS utf8 flag with \" and \$";
+ }
+ # XS used to produce "$b\"' which is 4 chars, not 3. [ie wrongly qq(\$b\\\")]
+############# 328
+ $WANT = <<'EOT';
+#$VAR1 = '$b"';
+EOT
+
+ $a = "\$b\"\x{100}";
+ chop $a;
+ TEST q(Data::Dumper->Dump([$a])), "utf8 flag with \" and \$";
+ if ($XS) {
+ TEST q(Data::Dumper->Dumpxs([$a])), "XS utf8 flag with \" and \$";
+ }
+
+
+ # XS used to produce 'D'oh!' which is well, D'oh!
+ # Andreas found this one, which in turn discovered the previous two.
+############# 334
+ $WANT = <<'EOT';
+#$VAR1 = 'D\'oh!';
+EOT
+
+ $a = "D'oh!\x{100}";
+ chop $a;
+ TEST q(Data::Dumper->Dump([$a])), "utf8 flag with '";
+ if ($XS) {
+ TEST q(Data::Dumper->Dumpxs([$a])), "XS utf8 flag with '";
+ }
+}