#!perl -w
# This is a base file to be used by various .t's in its directory
-# It tests various code points that are "problematic", and verifies that the
-# correct warnings/flags etc are generated when using them. It also takes the
-# UTF-8 for some of them and perturbs it to be malformed in various ways, and
-# tests that this gets appropriately detected.
+# It tests various malformed UTF-8 sequences and some code points that are
+# "problematic", and verifies that the correct warnings/flags etc are
+# generated when using them. For the code points, it also takes the UTF-8 and
+# perturbs it to be malformed in various ways, and tests that this gets
+# appropriately detected.
use strict;
use Test::More;
$|=1;
-no warnings 'deprecated'; # Some of the below are above IV_MAX on 32 bit
- # machines, and that is tested elsewhere
use XS::APItest;
my @warnings_gotten;
push @warnings_gotten, map { chomp; $_ } @copy;
};
+my $highest_non_extended_utf8_cp = (isASCII) ? 0x7FFFFFFF : 0x3FFFFFFF;
my $native_lowest_continuation_chr = I8_to_native(chr $::lowest_continuation);
+# C5 is chosen as it is valid for both ASCII and EBCDIC platforms
+my $known_start_byte = I8_to_native("\xC5");
+
sub requires_extended_utf8($) {
# Returns a boolean as to whether or not the code point parameter fits
# into 31 bits, subject to the convention that a negative code point
# stands for one that overflows the word size, so won't fit in 31 bits.
- my $cp = shift;
- return $cp > 0x7FFFFFFF;
+ return shift > $highest_non_extended_utf8_cp;
+}
+
+sub is_extended_utf8($) {
+
+ # Returns a boolean as to whether or not the input UTF-8 sequence uses
+ # Perl extended UTF-8.
+
+ my $byte = substr(shift, 0, 1);
+ return ord $byte >= 0xFE if isASCII;
+ return $byte == I8_to_native("\xFF");
+}
+
+sub overflow_discern_len($) {
+
+ # Returns how many bytes are needed to tell if a non-overlong UTF-8
+ # sequence is for a code point that won't fit in the platform's word size.
+ # Only the length of the sequence representing a single code point is
+ # needed.
+
+ if (isASCII) {
+ return ($::is64bit) ? 3 : 1;
+
+ # Below is needed for code points above IV_MAX
+ #return ($::is64bit) ? 3 : ((shift == $::max_bytes)
+ # ? 1
+ # : 2);
+ }
+
+ return ($::is64bit) ? 2 : 8;
+}
+
+sub overlong_discern_len($) {
+
+ # Returns how many bytes are needed to tell if the input UTF-8 sequence
+ # for a code point is overlong
+
+ my $string = shift;
+ my $length = length $string;
+ my $byte = ord native_to_I8(substr($string, 0, 1));
+ if (isASCII) {
+ return ($byte >= 0xFE)
+ ? ((! $::is64bit)
+ ? 1
+ : ($byte == 0xFF) ? 7 : 2)
+ : (($length == 2) ? 1 : 2);
+ # Below is needed for code points above IV_MAX
+ #return ($length == $::max_bytes)
+ # # This is constrained to 1 on 32-bit machines, as it
+ # # overflows there
+ # ? (($::is64bit) ? 7 : 1)
+ # : (($length == 2) ? 1 : 2);
+ }
+
+ return ($length == $::max_bytes) ? 8 : (($length <= 3) ? 1 : 2);
}
my @tests;
# like being a surrogate; 0 indicates we need
# the whole string. Some categories have a
# default that is used if this is omitted.
+ [ "orphan continuation byte malformation",
+ I8_to_native("$::I8c"),
+ 0xFFFD,
+ 1,
+ ],
+ [ "overlong malformation, lowest 2-byte",
+ (isASCII) ? "\xc0\x80" : I8_to_native("\xc0\xa0"),
+ 0, # NUL
+ ],
+ [ "overlong malformation, highest 2-byte",
+ (isASCII) ? "\xc1\xbf" : I8_to_native("\xc4\xbf"),
+ (isASCII) ? 0x7F : 0xFF,
+ ],
+ [ "overlong malformation, lowest 3-byte",
+ (isASCII) ? "\xe0\x80\x80" : I8_to_native("\xe0\xa0\xa0"),
+ 0, # NUL
+ ],
+ [ "overlong malformation, highest 3-byte",
+ (isASCII) ? "\xe0\x9f\xbf" : I8_to_native("\xe0\xbf\xbf"),
+ (isASCII) ? 0x7FF : 0x3FF,
+ ],
[ "lowest surrogate",
(isASCII) ? "\xed\xa0\x80" : I8_to_native("\xf1\xb6\xa0\xa0"),
0xD800,
(isASCII) ? "\xef\xbf\xbf" : I8_to_native("\xf1\xbf\xbf\xbf"),
0xFFFF,
],
+ [ "overlong malformation, lowest 4-byte",
+ (isASCII) ? "\xf0\x80\x80\x80" : I8_to_native("\xf0\xa0\xa0\xa0"),
+ 0, # NUL
+ ],
+ [ "overlong malformation, highest 4-byte",
+ (isASCII) ? "\xf0\x8F\xbf\xbf" : I8_to_native("\xf0\xaf\xbf\xbf"),
+ (isASCII) ? 0xFFFF : 0x3FFF,
+ ],
[ "non-character code point U+1FFFE",
(isASCII) ? "\xf0\x9f\xbf\xbe" : I8_to_native("\xf3\xbf\xbf\xbe"),
0x1FFFE,
(isASCII) ? 0x140000 : 0x200000,
1,
],
- [ "lowest 32 bit code point",
+ [ "overlong malformation, lowest 5-byte",
(isASCII)
- ? "\xfe\x82\x80\x80\x80\x80\x80"
- : I8_to_native(
- "\xff\xa0\xa0\xa0\xa0\xa0\xa0\xa2\xa0\xa0\xa0\xa0\xa0\xa0"),
- 0x80000000,
- (isASCII) ? 1 : 8,
+ ? "\xf8\x80\x80\x80\x80"
+ : I8_to_native("\xf8\xa0\xa0\xa0\xa0"),
+ 0, # NUL
+ ],
+ [ "overlong malformation, highest 5-byte",
+ (isASCII)
+ ? "\xf8\x87\xbf\xbf\xbf"
+ : I8_to_native("\xf8\xa7\xbf\xbf\xbf"),
+ (isASCII) ? 0x1FFFFF : 0x3FFFF,
+ ],
+ [ "overlong malformation, lowest 6-byte",
+ (isASCII)
+ ? "\xfc\x80\x80\x80\x80\x80"
+ : I8_to_native("\xfc\xa0\xa0\xa0\xa0\xa0"),
+ 0, # NUL
+ ],
+ [ "overlong malformation, highest 6-byte",
+ (isASCII)
+ ? "\xfc\x83\xbf\xbf\xbf\xbf"
+ : I8_to_native("\xfc\xa3\xbf\xbf\xbf\xbf"),
+ (isASCII) ? 0x3FFFFFF : 0x3FFFFF,
+ ],
+ [ "overlong malformation, lowest 7-byte",
+ (isASCII)
+ ? "\xfe\x80\x80\x80\x80\x80\x80"
+ : I8_to_native("\xfe\xa0\xa0\xa0\xa0\xa0\xa0"),
+ 0, # NUL
],
- [ "highest 32 bit code point",
+ [ "overlong malformation, highest 7-byte",
(isASCII)
- ? "\xfe\x83\xbf\xbf\xbf\xbf\xbf"
+ ? "\xfe\x81\xbf\xbf\xbf\xbf\xbf"
+ : I8_to_native("\xfe\xa1\xbf\xbf\xbf\xbf\xbf"),
+ (isASCII) ? 0x7FFFFFFF : 0x3FFFFFF,
+ ],
+ [ "highest 31 bit code point",
+ (isASCII)
+ ? "\xfd\xbf\xbf\xbf\xbf\xbf"
: I8_to_native(
- "\xff\xa0\xa0\xa0\xa0\xa0\xa0\xa3\xbf\xbf\xbf\xbf\xbf\xbf"),
- 0xFFFFFFFF,
- (isASCII) ? 1 : 8,
+ "\xff\xa0\xa0\xa0\xa0\xa0\xa0\xa1\xbf\xbf\xbf\xbf\xbf\xbf"),
+ 0x7FFFFFFF,
+ 1,
],
- [ "Lowest 33 bit code point",
+ [ "lowest 32 bit code point",
(isASCII)
- ? "\xfe\x84\x80\x80\x80\x80\x80"
+ ? "\xfe\x82\x80\x80\x80\x80\x80"
: I8_to_native(
- "\xff\xa0\xa0\xa0\xa0\xa0\xa0\xa4\xa0\xa0\xa0\xa0\xa0\xa0"),
- ($::is64bit) ? 0x100000000 : -1, # Overflows on 32-bit systems
- (isASCII && ! $::is64bit) ? 2 : 1,
+ "\xff\xa0\xa0\xa0\xa0\xa0\xa0\xa2\xa0\xa0\xa0\xa0\xa0\xa0"),
+ ($::is64bit) ? 0x80000000 : -1, # Overflows on 32-bit systems
+ 1,
],
+ # Used when UV_MAX is allowed as a code point
+ #[ "highest 32 bit code point",
+ # (isASCII)
+ # ? "\xfe\x83\xbf\xbf\xbf\xbf\xbf"
+ # : I8_to_native(
+ # "\xff\xa0\xa0\xa0\xa0\xa0\xa0\xa3\xbf\xbf\xbf\xbf\xbf\xbf"),
+ # 0xFFFFFFFF,
+ #],
+ #[ "Lowest 33 bit code point",
+ # (isASCII)
+ # ? "\xfe\x84\x80\x80\x80\x80\x80"
+ # : I8_to_native(
+ # "\xff\xa0\xa0\xa0\xa0\xa0\xa0\xa4\xa0\xa0\xa0\xa0\xa0\xa0"),
+ # ($::is64bit) ? 0x100000000 : 0x0, # Overflows on 32-bit systems
+ #],
);
if (! $::is64bit) {
if (isASCII) {
push @tests,
+ [ "overlong malformation, but naively looks like overflow",
+ "\xff\x80\x80\x80\x80\x80\x80\x81\xbf\xbf\xbf\xbf\xbf",
+ 0x7FFFFFFF,
+ ],
+ # Used when above IV_MAX are allowed.
+ #[ "overlong malformation, but naively looks like overflow",
+ # "\xff\x80\x80\x80\x80\x80\x80\x83\xbf\xbf\xbf\xbf\xbf",
+ # 0xFFFFFFFF,
+ #],
[ "overflow that old algorithm failed to detect",
"\xfe\x86\x80\x80\x80\x80\x80",
-1,
- 2,
];
}
}
+ push @tests,
+ [ "overlong malformation, lowest max-byte",
+ (isASCII)
+ ? "\xff\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80"
+ : I8_to_native(
+ "\xff\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0"),
+ 0, # NUL
+ ],
+ [ "overlong malformation, highest max-byte",
+ (isASCII) # 2**36-1 on ASCII; 2**30-1 on EBCDIC
+ ? "\xff\x80\x80\x80\x80\x80\x80\xbf\xbf\xbf\xbf\xbf\xbf"
+ : I8_to_native(
+ "\xff\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xbf\xbf\xbf\xbf\xbf\xbf"),
+ (isASCII) ? (($::is64bit) ? 0xFFFFFFFFF : -1) : 0x3FFFFFFF,
+ ];
+
+ if (isASCII) {
+ push @tests,
+ [ "Lowest code point requiring 13 bytes to represent", # 2**36
+ "\xff\x80\x80\x80\x80\x80\x81\x80\x80\x80\x80\x80\x80",
+ ($::is64bit) ? 0x1000000000 : -1, # overflows on 32bit
+ ],
+ };
+
if ($::is64bit) {
push @tests,
- [ "highest 64 bit code point",
+ [ "highest 63 bit code point",
(isASCII)
- ? "\xff\x80\x8f\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf"
+ ? "\xff\x80\x87\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf"
: I8_to_native(
- "\xff\xaf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf"),
- 0xFFFFFFFFFFFFFFFF,
+ "\xff\xa7\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf"),
+ 0x7FFFFFFFFFFFFFFF,
(isASCII) ? 1 : 2,
],
- [ "first 65 bit code point",
+ [ "first 64 bit code point",
(isASCII)
- ? "\xff\x80\x9f\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80"
+ ? "\xff\x80\x88\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80"
: I8_to_native(
- "\xff\xb0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0"),
+ "\xff\xa8\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0"),
-1,
- (isASCII) ? 3 : 2,
];
+ # Used when UV_MAX is allowed as a code point
+ #[ "highest 64 bit code point",
+ # (isASCII)
+ # ? "\xff\x80\x8f\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf"
+ # : I8_to_native(
+ # "\xff\xaf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf"),
+ # 0xFFFFFFFFFFFFFFFF,
+ # (isASCII) ? 1 : 2,
+ #],
+ #[ "first 65 bit code point",
+ # (isASCII)
+ # ? "\xff\x80\x9f\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80"
+ # : I8_to_native(
+ # "\xff\xb0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0"),
+ # 0,
+ #];
if (isASCII) {
push @tests,
- [ "Lowest code point requiring 13 bytes to represent",
- "\xff\x80\x80\x80\x80\x80\x81\x80\x80\x80\x80\x80\x80",
- 0x1000000000,
- 1,
- ],
[ "overflow that old algorithm failed to detect",
"\xff\x80\x90\x90\x90\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf",
-1,
- 3,
];
}
else {
I8_to_native(
"\xff\xa0\xa0\xa0\xa0\xa0\xa1\xa0\xa0\xa0\xa0\xa0\xa0\xa0"),
0x800000000,
- 7,
+ 40000000
],
[ "requires at least 32 bits",
I8_to_native(
"\xff\xa0\xa0\xa0\xa0\xa1\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0"),
0x10000000000,
- 6,
],
[ "requires at least 32 bits",
I8_to_native(
"\xff\xa0\xa0\xa0\xa1\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0"),
0x200000000000,
- 5,
],
[ "requires at least 32 bits",
I8_to_native(
"\xff\xa0\xa0\xa1\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0"),
0x4000000000000,
- 4,
],
[ "requires at least 32 bits",
I8_to_native(
"\xff\xa0\xa1\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0"),
0x80000000000000,
- 3,
],
[ "requires at least 32 bits",
I8_to_native(
"\xff\xa1\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0"),
0x1000000000000000,
- 2,
];
}
}
W_NONCHAR
D_SUPER
W_SUPER
- D_ABOVE_31_BIT
- W_ABOVE_31_BIT
+ D_PERL_EXTENDED
+ W_PERL_EXTENDED
CHECK_ONLY
NO_CONFIDENCE_IN_CURLEN_
) );
. ')';
}
+my @uvchr_flags_to_text = ( qw(
+ W_SURROGATE
+ W_NONCHAR
+ W_SUPER
+ W_PERL_EXTENDED
+ D_SURROGATE
+ D_NONCHAR
+ D_SUPER
+ D_PERL_EXTENDED
+) );
+
sub uvchr_display_call($)
{
# Converts an eval string that calls test_uvchr_to_utf8 into a more human
# test_uvchr_to_utf8n_flags($uv, $flags)
#diag $_[0];
- my @flags_to_text = ( qw(
- W_SURROGATE
- W_NONCHAR
- W_SUPER
- W_ABOVE_31_BIT
- D_SURROGATE
- D_NONCHAR
- D_SUPER
- D_ABOVE_31_BIT
- ) );
$_[0] =~ / ^ ( [^(]* \( ) ( \d+ ) , \s* ( \d+ ) \) $ /x;
my $text = $1;
my $cp = sprintf "%X", $2;
my $flags = $3;
- return "${text}0x$cp, " . flags_to_text($flags, \@flags_to_text) . ')';
+ return "${text}0x$cp, " . flags_to_text($flags, \@uvchr_flags_to_text) . ')';
}
sub do_warnings_test(@)
my $num_test_files = $ENV{TEST_JOBS} || 1;
$num_test_files = 10 if $num_test_files > 10;
+# We only really need to test utf8n_to_uvchr_msgs() once with this flag.
+my $tested_CHECK_ONLY = 0;
+
my $test_count = -1;
foreach my $test (@tests) {
- $test_count++;
- next if $test_count % $num_test_files != $::TEST_CHUNK;
-
- my ($testname, $bytes, $allowed_uv, $needed_to_discern_len) = @$test;
-
- my $length = length $bytes;
- my $will_overflow = $allowed_uv < 0;
-
- my $uv_string = sprintf(($allowed_uv < 0x100) ? "%02X" : "%04X", $allowed_uv);
- my $display_bytes = display_bytes($bytes);
-
- my $controlling_warning_category;
- my $utf8n_flag_to_warn;
- my $utf8n_flag_to_disallow;
- my $uvchr_flag_to_warn;
- my $uvchr_flag_to_disallow;
-
- # We want to test that the independent flags are actually independent.
- # For example, that a surrogate doesn't trigger a non-character warning,
- # and conversely, turning off an above-Unicode flag doesn't suppress a
- # surrogate warning. Earlier versions of this file used nested loops to
- # test all possible combinations. But that creates lots of tests, making
- # this run too long. What is now done instead is to use the complement of
- # the category we are testing to greatly reduce the combinatorial
- # explosion. For example, if we have a surrogate and we aren't expecting
- # a warning about it, we set all the flags for non-surrogates to raise
- # warnings. If one shows up, it indicates the flags aren't independent.
- my $utf8n_flag_to_warn_complement;
- my $utf8n_flag_to_disallow_complement;
- my $uvchr_flag_to_warn_complement;
- my $uvchr_flag_to_disallow_complement;
-
- # Many of the code points being tested are middling in that if code point
- # edge cases work, these are very likely to as well. Because this test
- # file takes a while to execute, we skip testing the edge effects of code
- # points deemed middling, while testing their basics and continuing to
- # fully test the non-middling code points.
- my $skip_most_tests = 0;
-
- my $cp_message_qr; # Pattern that matches the message raised when
- # that message contains the problematic code
- # point. The message is the same (currently) both
- # when going from/to utf8.
- my $non_cp_trailing_text; # The suffix text when the message doesn't
- # contain a code point. (This is a result of
- # some sort of malformation that means we
- # can't get an exact code poin
-
- # Is this test malformed from the beginning? If so, we know to generally
- # expect that the tests will show it isn't valid.
- my $initially_malformed = 0;
-
- if ($will_overflow || $allowed_uv > 0x10FFFF) {
-
- # Set the SUPER flags; later, we test for ABOVE_31_BIT as well.
- $utf8n_flag_to_warn = $::UTF8_WARN_SUPER;
- $utf8n_flag_to_disallow = $::UTF8_DISALLOW_SUPER;
- $uvchr_flag_to_warn = $::UNICODE_WARN_SUPER;
- $uvchr_flag_to_disallow = $::UNICODE_DISALLOW_SUPER;;
-
- # Below, we add the flags for non-above-31 bit to the code points that
- # don't fit that category. Special tests are done for this category
- # in the inner loop.
- $utf8n_flag_to_warn_complement = $::UTF8_WARN_NONCHAR
- |$::UTF8_WARN_SURROGATE;
- $utf8n_flag_to_disallow_complement = $::UTF8_DISALLOW_NONCHAR
- |$::UTF8_DISALLOW_SURROGATE;
- $uvchr_flag_to_warn_complement = $::UNICODE_WARN_NONCHAR
- |$::UNICODE_WARN_SURROGATE;
- $uvchr_flag_to_disallow_complement = $::UNICODE_DISALLOW_NONCHAR
- |$::UNICODE_DISALLOW_SURROGATE;
- $controlling_warning_category = 'non_unicode';
-
- if ($will_overflow) { # This is realy a malformation
- $non_cp_trailing_text = "if you see this, there is an error";
- $cp_message_qr = qr/\Q$non_cp_trailing_text\E/;
- $initially_malformed = 1;
- }
- elsif (requires_extended_utf8($allowed_uv)) {
- $cp_message_qr = qr/\QCode point 0x$uv_string is not Unicode,\E
- \Q and not portable\E/x;
- $non_cp_trailing_text = "is for a non-Unicode code point, and is not portable";
- }
- else {
- $cp_message_qr = qr/\QCode point 0x$uv_string is not Unicode,\E
- \Q may not be portable\E/x;
- $non_cp_trailing_text = "is for a non-Unicode code point, may not"
- . " be portable";
- $utf8n_flag_to_warn_complement |= $::UTF8_WARN_ABOVE_31_BIT;
- $utf8n_flag_to_disallow_complement |= $::UTF8_DISALLOW_ABOVE_31_BIT;
- $uvchr_flag_to_warn_complement |= $::UNICODE_WARN_ABOVE_31_BIT;
- $uvchr_flag_to_disallow_complement
- |= $::UNICODE_DISALLOW_ABOVE_31_BIT;
- }
- }
- elsif ($allowed_uv >= 0xD800 && $allowed_uv <= 0xDFFF) {
- $cp_message_qr = qr/UTF-16 surrogate U\+$uv_string/;
- $non_cp_trailing_text = "is for a surrogate";
- $needed_to_discern_len = 2 unless defined $needed_to_discern_len;
- $skip_most_tests = 1 if $allowed_uv > 0xD800 && $allowed_uv < 0xDFFF;
-
- $utf8n_flag_to_warn = $::UTF8_WARN_SURROGATE;
- $utf8n_flag_to_disallow = $::UTF8_DISALLOW_SURROGATE;
- $uvchr_flag_to_warn = $::UNICODE_WARN_SURROGATE;
- $uvchr_flag_to_disallow = $::UNICODE_DISALLOW_SURROGATE;;
-
- $utf8n_flag_to_warn_complement = $::UTF8_WARN_NONCHAR
- |$::UTF8_WARN_SUPER
- |$::UTF8_WARN_ABOVE_31_BIT;
- $utf8n_flag_to_disallow_complement = $::UTF8_DISALLOW_NONCHAR
- |$::UTF8_DISALLOW_SUPER
- |$::UTF8_DISALLOW_ABOVE_31_BIT;
- $uvchr_flag_to_warn_complement = $::UNICODE_WARN_NONCHAR
- |$::UNICODE_WARN_SUPER
- |$::UNICODE_WARN_ABOVE_31_BIT;
- $uvchr_flag_to_disallow_complement = $::UNICODE_DISALLOW_NONCHAR
- |$::UNICODE_DISALLOW_SUPER
- |$::UNICODE_DISALLOW_ABOVE_31_BIT;
- $controlling_warning_category = 'surrogate';
- }
- elsif ( ($allowed_uv >= 0xFDD0 && $allowed_uv <= 0xFDEF)
- || ($allowed_uv & 0xFFFE) == 0xFFFE)
- {
- $cp_message_qr = qr/\QUnicode non-character U+$uv_string\E
- \Q is not recommended for open interchange\E/x;
- $non_cp_trailing_text = "if you see this, there is an error";
- $needed_to_discern_len = $length unless defined $needed_to_discern_len;
- if ( ($allowed_uv > 0xFDD0 && $allowed_uv < 0xFDEF)
- || ($allowed_uv > 0xFFFF && $allowed_uv < 0x10FFFE))
- {
- $skip_most_tests = 1;
- }
-
- $utf8n_flag_to_warn = $::UTF8_WARN_NONCHAR;
- $utf8n_flag_to_disallow = $::UTF8_DISALLOW_NONCHAR;
- $uvchr_flag_to_warn = $::UNICODE_WARN_NONCHAR;
- $uvchr_flag_to_disallow = $::UNICODE_DISALLOW_NONCHAR;;
-
- $utf8n_flag_to_warn_complement = $::UTF8_WARN_SURROGATE
- |$::UTF8_WARN_SUPER
- |$::UTF8_WARN_ABOVE_31_BIT;
- $utf8n_flag_to_disallow_complement = $::UTF8_DISALLOW_SURROGATE
- |$::UTF8_DISALLOW_SUPER
- |$::UTF8_DISALLOW_ABOVE_31_BIT;
- $uvchr_flag_to_warn_complement = $::UNICODE_WARN_SURROGATE
- |$::UNICODE_WARN_SUPER
- |$::UNICODE_WARN_ABOVE_31_BIT;
- $uvchr_flag_to_disallow_complement = $::UNICODE_DISALLOW_SURROGATE
- |$::UNICODE_DISALLOW_SUPER
- |$::UNICODE_DISALLOW_ABOVE_31_BIT;
-
- $controlling_warning_category = 'nonchar';
- }
- else {
- die "Can't figure out what type of warning to test for $testname"
- }
-
- die 'Didn\'t set $needed_to_discern_len for ' . $testname
- unless defined $needed_to_discern_len;
-
- { # First test the isFOO calls
- use warnings; no warnings 'deprecated'; # Make sure these don't raise warnings
- undef @warnings_gotten;
-
- my $ret = test_isUTF8_CHAR($bytes, $length);
- my $ret_flags = test_isUTF8_CHAR_flags($bytes, $length, 0);
- if ($initially_malformed) {
- is($ret, 0, "For $testname: isUTF8_CHAR() returns 0");
- is($ret_flags, 0, " And isUTF8_CHAR_flags() returns 0");
- }
- else {
- is($ret, $length,
- "For $testname: isUTF8_CHAR() returns expected length: $length");
- is($ret_flags, $length, " And isUTF8_CHAR_flags(...,0)"
- . " returns expected length: $length");
- }
- is(scalar @warnings_gotten, 0,
- " And neither isUTF8_CHAR() nor isUTF8_CHAR()_flags generated"
- . " any warnings")
- or output_warnings(@warnings_gotten);
-
- undef @warnings_gotten;
- $ret = test_isSTRICT_UTF8_CHAR($bytes, $length);
- if ($initially_malformed) {
- is($ret, 0, " And isSTRICT_UTF8_CHAR() returns 0");
- }
- else {
- my $expected_ret = ( $testname =~ /surrogate|non-character/
- || $allowed_uv > 0x10FFFF)
- ? 0
- : $length;
- is($ret, $expected_ret, " And isSTRICT_UTF8_CHAR() returns"
- . " expected length: $expected_ret");
- $ret = test_isUTF8_CHAR_flags($bytes, $length,
- $::UTF8_DISALLOW_ILLEGAL_INTERCHANGE);
- is($ret, $expected_ret,
- " And isUTF8_CHAR_flags('DISALLOW_ILLEGAL_INTERCHANGE')"
- . " acts like isSTRICT_UTF8_CHAR");
- }
- is(scalar @warnings_gotten, 0,
- " And neither isSTRICT_UTF8_CHAR() nor isUTF8_CHAR_flags"
- . " generated any warnings")
- or output_warnings(@warnings_gotten);
-
- undef @warnings_gotten;
- $ret = test_isC9_STRICT_UTF8_CHAR($bytes, $length);
- if ($initially_malformed) {
- is($ret, 0, " And isC9_STRICT_UTF8_CHAR() returns 0");
- }
- else {
- my $expected_ret = ( $testname =~ /surrogate/
- || $allowed_uv > 0x10FFFF)
- ? 0
- : $length;
- is($ret, $expected_ret, " And isC9_STRICT_UTF8_CHAR()"
- ." returns expected length: $expected_ret");
- $ret = test_isUTF8_CHAR_flags($bytes, $length,
- $::UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE);
- is($ret, $expected_ret,
- " And isUTF8_CHAR_flags('DISALLOW_ILLEGAL_C9_INTERCHANGE')"
- . " acts like isC9_STRICT_UTF8_CHAR");
- }
- is(scalar @warnings_gotten, 0,
- " And neither isC9_STRICT_UTF8_CHAR() nor isUTF8_CHAR_flags"
- . " generated any warnings")
- or output_warnings(@warnings_gotten);
+ $test_count++;
+ next if $test_count % $num_test_files != $::TEST_CHUNK;
+
+ my ($testname, $bytes, $allowed_uv, $needed_to_discern_len) = @$test;
+
+ my $length = length $bytes;
+ my $initially_overlong = $testname =~ /overlong/;
+ my $initially_orphan = $testname =~ /orphan/;
+ my $will_overflow = $allowed_uv < 0;
+
+ my $uv_string = sprintf(($allowed_uv < 0x100) ? "%02X" : "%04X", $allowed_uv);
+ my $display_bytes = display_bytes($bytes);
+
+ my $controlling_warning_category;
+ my $utf8n_flag_to_warn;
+ my $utf8n_flag_to_disallow;
+ my $uvchr_flag_to_warn;
+ my $uvchr_flag_to_disallow;
+
+ # We want to test that the independent flags are actually independent.
+ # For example, that a surrogate doesn't trigger a non-character warning,
+ # and conversely, turning off an above-Unicode flag doesn't suppress a
+ # surrogate warning. Earlier versions of this file used nested loops to
+ # test all possible combinations. But that creates lots of tests, making
+ # this run too long. What is now done instead is to use the complement of
+ # the category we are testing to greatly reduce the combinatorial
+ # explosion. For example, if we have a surrogate and we aren't expecting
+ # a warning about it, we set all the flags for non-surrogates to raise
+ # warnings. If one shows up, it indicates the flags aren't independent.
+ my $utf8n_flag_to_warn_complement;
+ my $utf8n_flag_to_disallow_complement;
+ my $uvchr_flag_to_warn_complement;
+ my $uvchr_flag_to_disallow_complement;
+
+ # Many of the code points being tested are middling in that if code point
+ # edge cases work, these are very likely to as well. Because this test
+ # file takes a while to execute, we skip testing the edge effects of code
+ # points deemed middling, while testing their basics and continuing to
+ # fully test the non-middling code points.
+ my $skip_most_tests = 0;
+
+ my $cp_message_qr; # Pattern that matches the message raised when
+ # that message contains the problematic code
+ # point. The message is the same (currently) both
+ # when going from/to utf8.
+ my $non_cp_trailing_text; # The suffix text when the message doesn't
+ # contain a code point. (This is a result of
+ # some sort of malformation that means we
+ # can't get an exact code poin
+ my $extended_cp_message_qr = qr/\QCode point 0x$uv_string is not Unicode,\E
+ \Q requires a Perl extension, and so is not\E
+ \Q portable\E/x;
+ my $extended_non_cp_trailing_text
+ = "is a Perl extension, and so is not portable";
+
+ # What bytes should have been used to specify a code point that has been
+ # specified as an overlong.
+ my $correct_bytes_for_overlong;
+
+ # Is this test malformed from the beginning? If so, we know to generally
+ # expect that the tests will show it isn't valid.
+ my $initially_malformed = 0;
+
+ if ($initially_overlong || $initially_orphan) {
+ $non_cp_trailing_text = "if you see this, there is an error";
+ $cp_message_qr = qr/\Q$non_cp_trailing_text\E/;
+ $initially_malformed = 1;
+ $utf8n_flag_to_warn = 0;
+ $utf8n_flag_to_disallow = 0;
+
+ $utf8n_flag_to_warn_complement = $::UTF8_WARN_SURROGATE;
+ $utf8n_flag_to_disallow_complement = $::UTF8_DISALLOW_SURROGATE;
+ if (! $will_overflow && $allowed_uv <= 0x10FFFF) {
+ $utf8n_flag_to_warn_complement |= $::UTF8_WARN_SUPER;
+ $utf8n_flag_to_disallow_complement |= $::UTF8_DISALLOW_SUPER;
+ if (($allowed_uv & 0xFFFF) != 0xFFFF) {
+ $utf8n_flag_to_warn_complement |= $::UTF8_WARN_NONCHAR;
+ $utf8n_flag_to_disallow_complement |= $::UTF8_DISALLOW_NONCHAR;
+ }
+ }
+ if (! is_extended_utf8($bytes)) {
+ $utf8n_flag_to_warn_complement |= $::UTF8_WARN_PERL_EXTENDED;
+ $utf8n_flag_to_disallow_complement |= $::UTF8_DISALLOW_PERL_EXTENDED;
+ }
- foreach my $disallow_type (0..2) {
- # 0 is don't disallow this type of code point
- # 1 is do disallow
- # 2 is do disallow, but only for above 31 bit
+ $controlling_warning_category = 'utf8';
- my $disallow_flags;
- my $expected_ret;
+ if ($initially_overlong) {
+ if (! defined $needed_to_discern_len) {
+ $needed_to_discern_len = overlong_discern_len($bytes);
+ }
+ $correct_bytes_for_overlong = display_bytes_no_quotes(chr $allowed_uv);
+ }
+ }
+ elsif($will_overflow || $allowed_uv > 0x10FFFF) {
+
+ # Set the SUPER flags; later, we test for PERL_EXTENDED as well.
+ $utf8n_flag_to_warn = $::UTF8_WARN_SUPER;
+ $utf8n_flag_to_disallow = $::UTF8_DISALLOW_SUPER;
+ $uvchr_flag_to_warn = $::UNICODE_WARN_SUPER;
+ $uvchr_flag_to_disallow = $::UNICODE_DISALLOW_SUPER;;
+
+ # Below, we add the flags for non-perl_extended to the code points
+ # that don't fit that category. Special tests are done for this
+ # category in the inner loop.
+ $utf8n_flag_to_warn_complement = $::UTF8_WARN_NONCHAR
+ |$::UTF8_WARN_SURROGATE;
+ $utf8n_flag_to_disallow_complement = $::UTF8_DISALLOW_NONCHAR
+ |$::UTF8_DISALLOW_SURROGATE;
+ $uvchr_flag_to_warn_complement = $::UNICODE_WARN_NONCHAR
+ |$::UNICODE_WARN_SURROGATE;
+ $uvchr_flag_to_disallow_complement = $::UNICODE_DISALLOW_NONCHAR
+ |$::UNICODE_DISALLOW_SURROGATE;
+ $controlling_warning_category = 'non_unicode';
+
+ if ($will_overflow) { # This is realy a malformation
+ $non_cp_trailing_text = "if you see this, there is an error";
+ $cp_message_qr = qr/\Q$non_cp_trailing_text\E/;
+ $initially_malformed = 1;
+ if (! defined $needed_to_discern_len) {
+ $needed_to_discern_len = overflow_discern_len($length);
+ }
+ }
+ elsif (requires_extended_utf8($allowed_uv)) {
+ $cp_message_qr = $extended_cp_message_qr;
+ $non_cp_trailing_text = $extended_non_cp_trailing_text;
+ $needed_to_discern_len = 1 unless defined $needed_to_discern_len;
+ }
+ else {
+ $cp_message_qr = qr/\QCode point 0x$uv_string is not Unicode,\E
+ \Q may not be portable\E/x;
+ $non_cp_trailing_text = "is for a non-Unicode code point, may not"
+ . " be portable";
+ $utf8n_flag_to_warn_complement |= $::UTF8_WARN_PERL_EXTENDED;
+ $utf8n_flag_to_disallow_complement
+ |= $::UTF8_DISALLOW_PERL_EXTENDED;
+ $uvchr_flag_to_warn_complement |= $::UNICODE_WARN_PERL_EXTENDED;
+ $uvchr_flag_to_disallow_complement
+ |= $::UNICODE_DISALLOW_PERL_EXTENDED;
+ }
+ }
+ elsif ($allowed_uv >= 0xD800 && $allowed_uv <= 0xDFFF) {
+ $cp_message_qr = qr/UTF-16 surrogate U\+$uv_string/;
+ $non_cp_trailing_text = "is for a surrogate";
+ $needed_to_discern_len = 2 unless defined $needed_to_discern_len;
+ $skip_most_tests = 1 if $allowed_uv > 0xD800 && $allowed_uv < 0xDFFF;
+
+ $utf8n_flag_to_warn = $::UTF8_WARN_SURROGATE;
+ $utf8n_flag_to_disallow = $::UTF8_DISALLOW_SURROGATE;
+ $uvchr_flag_to_warn = $::UNICODE_WARN_SURROGATE;
+ $uvchr_flag_to_disallow = $::UNICODE_DISALLOW_SURROGATE;;
+
+ $utf8n_flag_to_warn_complement = $::UTF8_WARN_NONCHAR
+ |$::UTF8_WARN_SUPER
+ |$::UTF8_WARN_PERL_EXTENDED;
+ $utf8n_flag_to_disallow_complement = $::UTF8_DISALLOW_NONCHAR
+ |$::UTF8_DISALLOW_SUPER
+ |$::UTF8_DISALLOW_PERL_EXTENDED;
+ $uvchr_flag_to_warn_complement = $::UNICODE_WARN_NONCHAR
+ |$::UNICODE_WARN_SUPER
+ |$::UNICODE_WARN_PERL_EXTENDED;
+ $uvchr_flag_to_disallow_complement = $::UNICODE_DISALLOW_NONCHAR
+ |$::UNICODE_DISALLOW_SUPER
+ |$::UNICODE_DISALLOW_PERL_EXTENDED;
+ $controlling_warning_category = 'surrogate';
+ }
+ elsif ( ($allowed_uv >= 0xFDD0 && $allowed_uv <= 0xFDEF)
+ || ($allowed_uv & 0xFFFE) == 0xFFFE)
+ {
+ $cp_message_qr = qr/\QUnicode non-character U+$uv_string\E
+ \Q is not recommended for open interchange\E/x;
+ $non_cp_trailing_text = "if you see this, there is an error";
+ $needed_to_discern_len = $length unless defined $needed_to_discern_len;
+ if ( ($allowed_uv > 0xFDD0 && $allowed_uv < 0xFDEF)
+ || ($allowed_uv > 0xFFFF && $allowed_uv < 0x10FFFE))
+ {
+ $skip_most_tests = 1;
+ }
- if ($initially_malformed) {
+ $utf8n_flag_to_warn = $::UTF8_WARN_NONCHAR;
+ $utf8n_flag_to_disallow = $::UTF8_DISALLOW_NONCHAR;
+ $uvchr_flag_to_warn = $::UNICODE_WARN_NONCHAR;
+ $uvchr_flag_to_disallow = $::UNICODE_DISALLOW_NONCHAR;;
+
+ $utf8n_flag_to_warn_complement = $::UTF8_WARN_SURROGATE
+ |$::UTF8_WARN_SUPER
+ |$::UTF8_WARN_PERL_EXTENDED;
+ $utf8n_flag_to_disallow_complement = $::UTF8_DISALLOW_SURROGATE
+ |$::UTF8_DISALLOW_SUPER
+ |$::UTF8_DISALLOW_PERL_EXTENDED;
+ $uvchr_flag_to_warn_complement = $::UNICODE_WARN_SURROGATE
+ |$::UNICODE_WARN_SUPER
+ |$::UNICODE_WARN_PERL_EXTENDED;
+ $uvchr_flag_to_disallow_complement = $::UNICODE_DISALLOW_SURROGATE
+ |$::UNICODE_DISALLOW_SUPER
+ |$::UNICODE_DISALLOW_PERL_EXTENDED;
+
+ $controlling_warning_category = 'nonchar';
+ }
+ else {
+ die "Can't figure out what type of warning to test for $testname"
+ }
+
+ die 'Didn\'t set $needed_to_discern_len for ' . $testname
+ unless defined $needed_to_discern_len;
+
+ # We try various combinations of malformations that can occur
+ foreach my $short (0, 1) {
+ next if $skip_most_tests && $short;
+ foreach my $unexpected_noncont (0, 1) {
+ next if $skip_most_tests && $unexpected_noncont;
+ foreach my $overlong (0, 1) {
+ next if $overlong && $skip_most_tests;
+ next if $initially_overlong && ! $overlong;
+
+ # If we're creating an overlong, it can't be longer than the
+ # maximum length, so skip if we're already at that length.
+ next if (! $initially_overlong && $overlong)
+ && $length >= $::max_bytes;
+
+ my $this_cp_message_qr = $cp_message_qr;
+ my $this_non_cp_trailing_text = $non_cp_trailing_text;
+
+ foreach my $malformed_allow_type (0..2) {
+ # 0 don't allow this malformation; ignored if no malformation
+ # 1 allow, with REPLACEMENT CHARACTER returned
+ # 2 allow, with intended code point returned. All malformations
+ # other than overlong can't determine the intended code point,
+ # so this isn't valid for them.
+ next if $malformed_allow_type == 2
+ && ($will_overflow || $short || $unexpected_noncont);
+ next if $skip_most_tests && $malformed_allow_type;
+
+ # Here we are in the innermost loop for malformations. So we
+ # know which ones are in effect. Can now change the input to be
+ # appropriately malformed. We also can set up certain other
+ # things now, like whether we expect a return flag from this
+ # malformation, and which flag.
+
+ my $this_bytes = $bytes;
+ my $this_length = $length;
+ my $this_expected_len = $length;
+ my $this_needed_to_discern_len = $needed_to_discern_len;
+
+ my @malformation_names;
+ my @expected_malformation_warnings;
+ my @expected_malformation_return_flags;
+
+ # Contains the flags for any allowed malformations. Currently no
+ # combinations of on/off are tested for. It's either all are
+ # allowed, or none are.
+ my $allow_flags = 0;
+ my $overlong_is_in_perl_extended_utf8 = 0;
+ my $dont_use_overlong_cp = 0;
+
+ if ($initially_orphan) {
+ next if $overlong || $short || $unexpected_noncont;
+ }
- # Malformations are by default disallowed, so testing with
- # $disallow_type equal to 0 is sufficicient.
- next if $disallow_type;
+ if ($overlong) {
+ if (! $initially_overlong) {
+ my $new_expected_len;
+
+ # To force this malformation, we convert the original start
+ # byte into a continuation byte with the same data bits as
+ # originally. ...
+ my $start_byte = substr($this_bytes, 0, 1);
+ my $converted_to_continuation_byte
+ = start_byte_to_cont($start_byte);
+
+ # ... Then we prepend it with a known overlong sequence.
+ # This should evaluate to the exact same code point as the
+ # original. We try to avoid an overlong using Perl
+ # extended UTF-8. The code points are the highest
+ # representable as overlongs on the respective platform
+ # without using extended UTF-8.
+ if (native_to_I8($start_byte) lt "\xFC") {
+ $start_byte = I8_to_native("\xFC");
+ $new_expected_len = 6;
+ }
+ elsif (! isASCII && native_to_I8($start_byte) lt "\xFE") {
+
+ # FE is not extended UTF-8 on EBCDIC
+ $start_byte = I8_to_native("\xFE");
+ $new_expected_len = 7;
+ }
+ else { # Must use extended UTF-8. On ASCII platforms, we
+ # could express some overlongs here starting with
+ # \xFE, but there's no real reason to do so.
+ $overlong_is_in_perl_extended_utf8 = 1;
+ $start_byte = I8_to_native("\xFF");
+ $new_expected_len = $::max_bytes;
+ $this_cp_message_qr = $extended_cp_message_qr;
+
+ # The warning that gets raised doesn't include the
+ # code point in the message if the code point can be
+ # expressed without using extended UTF-8, but the
+ # particular overlong sequence used is in extended
+ # UTF-8. To do otherwise would be confusing to the
+ # user, as it would claim the code point requires
+ # extended, when it doesn't.
+ $dont_use_overlong_cp = 1
+ unless requires_extended_utf8($allowed_uv);
+ $this_non_cp_trailing_text
+ = $extended_non_cp_trailing_text;
+ }
+
+ # Splice in the revise continuation byte, preceded by the
+ # start byte and the proper number of the lowest
+ # continuation bytes.
+ $this_bytes = $start_byte
+ . ($native_lowest_continuation_chr
+ x ( $new_expected_len
+ - 1
+ - length($this_bytes)))
+ . $converted_to_continuation_byte
+ . substr($this_bytes, 1);
+ $this_length = length($this_bytes);
+ $this_needed_to_discern_len = $new_expected_len
+ - ( $this_expected_len
+ - $this_needed_to_discern_len);
+ $this_expected_len = $new_expected_len;
+ }
+ }
- $disallow_flags = 0;
- $expected_ret = 0;
- }
- elsif ($disallow_type == 1) {
- $disallow_flags = $utf8n_flag_to_disallow;
- $expected_ret = 0;
- }
- elsif ($disallow_type == 2) {
- next if ! requires_extended_utf8($allowed_uv);
- $disallow_flags = $::UTF8_DISALLOW_ABOVE_31_BIT;
- $expected_ret = 0;
- }
- else { # type is 0
- $disallow_flags = $utf8n_flag_to_disallow_complement;
- $expected_ret = $length;
- }
+ if ($short) {
- $ret = test_isUTF8_CHAR_flags($bytes, $length, $disallow_flags);
- is($ret, $expected_ret, " And isUTF8_CHAR_flags("
- . "$display_bytes, $disallow_flags) returns "
- . $expected_ret)
- or diag "The flags mean "
- . flags_to_text($disallow_flags, \@utf8n_flags_to_text);
-
- is(scalar @warnings_gotten, 0,
- " And isUTF8_CHAR_flags(...) generated no warnings")
- or output_warnings(@warnings_gotten);
-
- # Test partial character handling, for each byte not a full character
- my $did_test_partial = 0;
- for (my $j = 1; $j < $length - 1; $j++) {
- $did_test_partial = 1;
- my $partial = substr($bytes, 0, $j);
- my $ret_should_be;
- my $comment;
- if ($disallow_type || $initially_malformed) {
- $ret_should_be = 0;
- $comment = "disallowed";
- if ($j < $needed_to_discern_len) {
- $ret_should_be = 1;
- $comment .= ", but need $needed_to_discern_len"
- . " bytes to discern:";
- }
- }
- else {
- $ret_should_be = 1;
- $comment = "allowed";
- }
+ # To force this malformation, just tell the test to not look
+ # as far as it should into the input.
+ $this_length--;
+ $this_expected_len--;
- undef @warnings_gotten;
+ $allow_flags |= $::UTF8_ALLOW_SHORT if $malformed_allow_type;
+ }
- $ret = test_is_utf8_valid_partial_char_flags($partial, $j,
- $disallow_flags);
- is($ret, $ret_should_be,
- " And is_utf8_valid_partial_char_flags("
- . display_bytes($partial)
- . ", $disallow_flags), $comment: returns $ret_should_be")
- or diag "The flags mean "
- . flags_to_text($disallow_flags, \@utf8n_flags_to_text);
- }
+ if ($unexpected_noncont) {
- if ($did_test_partial) {
- is(scalar @warnings_gotten, 0,
- " And is_utf8_valid_partial_char_flags()"
- . " generated no warnings for any of the lengths")
- or output_warnings(@warnings_gotten);
- }
- }
- }
+ # To force this malformation, change the final continuation
+ # byte into a start byte.
+ my $pos = ($short) ? -2 : -1;
+ substr($this_bytes, $pos, 1) = $known_start_byte;
+ $this_expected_len--;
+ }
- # Now test the to/from UTF-8 calls
- # This is more complicated than the malformations tested in other files in
- # this directory, as there are several orthogonal variables involved. We
- # test most possible combinations
-
- # We try various combinations of malformations that can occur
- foreach my $short (0, 1) {
- next if $skip_most_tests && $short;
- foreach my $unexpected_noncont (0, 1) {
- next if $skip_most_tests && $unexpected_noncont;
- foreach my $overlong (0, 1) {
- next if $overlong && $skip_most_tests;
-
- # If we're creating an overlong, it can't be longer than the
- # maximum length, so skip if we're already at that length.
- next if $overlong && $length >= $::max_bytes;
-
- foreach my $malformed_allow_type (0..2) {
- # 0 don't allow this malformation; ignored if no malformation
- # 1 allow, with REPLACEMENT CHARACTER returned
- # 2 allow, with intended code point returned. All malformations
- # other than overlong can't determine the intended code point,
- # so this isn't valid for them.
- next if $malformed_allow_type == 2
- && ($will_overflow || $short || $unexpected_noncont);
- next if $skip_most_tests && $malformed_allow_type;
-
- # Here we are in the innermost loop for malformations. So we
- # know which ones are in effect. Can now change the input to be
- # appropriately malformed. We also can set up certain other
- # things now, like whether we expect a return flag from this
- # malformation, and which flag.
-
- my $this_bytes = $bytes;
- my $this_length = $length;
- my $this_expected_len = $length;
- my $this_needed_to_discern_len = $needed_to_discern_len;
-
- my @malformation_names;
- my @expected_malformation_warnings;
- my @expected_malformation_return_flags;
-
- # Contains the flags for any allowed malformations. Currently no
- # combinations of on/off are tested for. It's either all are
- # allowed, or none are.
- my $allow_flags = 0;
-
- if ($overlong) {
- my $new_expected_len;
-
- # To force this malformation, we convert the original start
- # byte into a continuation byte with the same data bits as
- # originally. ...
- my $start_byte = substr($this_bytes, 0, 1);
- my $converted_to_continuation_byte
- = start_byte_to_cont($start_byte);
-
- # ... Then we prepend it with a known overlong sequence. This
- # should evaluate to the exact same code point as the
- # original. We try to avoid an overlong using Perl extended
- # UTF-8. The code points are the highest representable as
- # overlongs on the respective platform without using extended
- # UTF-8.
- if (native_to_I8($start_byte) lt "\xFC") {
- $start_byte = I8_to_native("\xFC");
- $new_expected_len = 6;
- }
- elsif (! isASCII && native_to_I8($start_byte) lt "\xFE") {
+ # The whole point of a test that is malformed from the beginning
+ # is to test for that malformation. If we've modified things so
+ # much that we don't have enough information to detect that
+ # malformation, there's no point in testing.
+ next if $initially_malformed
+ && $this_expected_len < $this_needed_to_discern_len;
+
+ # Here, we've transformed the input with all of the desired
+ # non-overflow malformations. We are now in a position to
+ # construct any potential warnings for those malformations. But
+ # it's a pain to get the detailed messages exactly right, so for
+ # now XXX, only do so for those that return an explicit code
+ # point.
+
+ if ($initially_orphan) {
+ push @malformation_names, "orphan continuation";
+ push @expected_malformation_return_flags,
+ $::UTF8_GOT_CONTINUATION;
+ $allow_flags |= $::UTF8_ALLOW_CONTINUATION
+ if $malformed_allow_type;
+ push @expected_malformation_warnings, qr/unexpected continuation/;
+ }
- # FE is not extended UTF-8 on EBCDIC
- $start_byte = I8_to_native("\xFE");
- $new_expected_len = 7;
- }
- else { # Must use extended UTF-8. On ASCII platforms, we
- # could express some overlongs here starting with
- # \xFE, but there's no real reason to do so.
- $start_byte = I8_to_native("\xFF");
- $new_expected_len = $::max_bytes;
- }
+ if ($overlong) {
+ push @malformation_names, 'overlong';
+ push @expected_malformation_return_flags, $::UTF8_GOT_LONG;
- # Splice in the revise continuation byte, preceded by the
- # start byte and the proper number of the lowest continuation
- # bytes.
- $this_bytes = $start_byte
- . ($native_lowest_continuation_chr
- x ( $new_expected_len - 1 - length($this_bytes)))
- . $converted_to_continuation_byte
- . substr($this_bytes, 1);
- $this_length = length($this_bytes);
- $this_needed_to_discern_len = $new_expected_len
- - ( $this_expected_len
- - $this_needed_to_discern_len);
- $this_expected_len = $new_expected_len;
- push @expected_malformation_return_flags, $::UTF8_GOT_LONG;
- push @malformation_names, 'overlong';
-
- if ($malformed_allow_type == 2) {
- $allow_flags |= $::UTF8_ALLOW_LONG_AND_ITS_VALUE;
- }
- elsif ($malformed_allow_type) {
- $allow_flags |= $::UTF8_ALLOW_LONG;
- }
- }
+ # If one of the other malformation types is also in effect, we
+ # don't know what the intended code point was.
+ if ($short || $unexpected_noncont || $will_overflow) {
+ push @expected_malformation_warnings, qr/overlong/;
+ }
+ else {
+ my $wrong_bytes = display_bytes_no_quotes(
+ substr($this_bytes, 0, $this_length));
+ if (! defined $correct_bytes_for_overlong) {
+ $correct_bytes_for_overlong
+ = display_bytes_no_quotes($bytes);
+ }
+ my $prefix = ( $allowed_uv > 0x10FFFF
+ || ! isASCII && $allowed_uv < 256)
+ ? "0x"
+ : "U+";
+ push @expected_malformation_warnings,
+ qr/\QMalformed UTF-8 character: $wrong_bytes\E
+ \Q (overlong; instead use\E
+ \Q $correct_bytes_for_overlong to\E
+ \Q represent $prefix$uv_string)/x;
+ }
- if ($short) {
- push @malformation_names, 'short';
+ if ($malformed_allow_type == 2) {
+ $allow_flags |= $::UTF8_ALLOW_LONG_AND_ITS_VALUE;
+ }
+ elsif ($malformed_allow_type) {
+ $allow_flags |= $::UTF8_ALLOW_LONG;
+ }
+ }
+ if ($short) {
+ push @malformation_names, 'short';
+ push @expected_malformation_return_flags, $::UTF8_GOT_SHORT;
+ push @expected_malformation_warnings, qr/too short/;
+ }
+ if ($unexpected_noncont) {
+ push @malformation_names, 'unexpected non-continuation';
+ push @expected_malformation_return_flags,
+ $::UTF8_GOT_NON_CONTINUATION;
+ $allow_flags |= $::UTF8_ALLOW_NON_CONTINUATION
+ if $malformed_allow_type;
+ push @expected_malformation_warnings,
+ qr/unexpected non-continuation byte/;
+ }
- # To force this malformation, just tell the test to not look
- # as far as it should into the input.
- $this_length--;
- $this_expected_len--;
- push @expected_malformation_return_flags, $::UTF8_GOT_SHORT;
+ # The overflow malformation is done differently than other
+ # malformations. It comes from manually typed tests in the test
+ # array. We now make it be treated like one of the other
+ # malformations. But some has to be deferred until the inner loop
+ my $overflow_msg_pattern;
+ if ($will_overflow) {
+ push @malformation_names, 'overflow';
+
+ $overflow_msg_pattern = display_bytes_no_quotes(
+ substr($this_bytes, 0, $this_expected_len));
+ $overflow_msg_pattern = qr/\QMalformed UTF-8 character:\E
+ \Q $overflow_msg_pattern\E
+ \Q (overflows)\E/x;
+ push @expected_malformation_return_flags, $::UTF8_GOT_OVERFLOW;
+ $allow_flags |= $::UTF8_ALLOW_OVERFLOW if $malformed_allow_type;
+ }
- $allow_flags |= $::UTF8_ALLOW_SHORT if $malformed_allow_type;
- }
+ # And we can create the malformation-related text for the the test
+ # names we eventually will generate.
+ my $malformations_name = "";
+ if (@malformation_names) {
+ $malformations_name .= "dis" unless $malformed_allow_type;
+ $malformations_name .= "allowed ";
+ $malformations_name .= "malformation";
+ $malformations_name .= "s" if @malformation_names > 1;
+ $malformations_name .= ": ";
+ $malformations_name .= join "/", @malformation_names;
+ $malformations_name = " ($malformations_name)";
+ }
- if ($unexpected_noncont) {
- push @malformation_names, 'unexpected non-continuation';
-
- # To force this malformation, change the final continuation
- # byte into a non continuation.
- my $pos = ($short) ? -2 : -1;
- substr($this_bytes, $pos, 1) = '?';
- $this_expected_len--;
- push @expected_malformation_return_flags,
- $::UTF8_GOT_NON_CONTINUATION;
- $allow_flags |= $::UTF8_ALLOW_NON_CONTINUATION
- if $malformed_allow_type;
- }
+ # Done setting up the malformation related stuff
- # Here, we've transformed the input with all of the desired
- # non-overflow malformations. We are now in a position to
- # construct any potential warnings for those malformations. But
- # it's a pain to get the detailed messages exactly right, so for
- # now XXX, only do so for those that return an explicit code
- # point.
+ { # First test the isFOO calls
+ use warnings; # XXX no warnings 'deprecated'; # Make sure these don't raise warnings
+ undef @warnings_gotten;
- if ($overlong) {
+ my $ret = test_isUTF8_CHAR($this_bytes, $this_length);
+ my $ret_flags
+ = test_isUTF8_CHAR_flags($this_bytes, $this_length, 0);
+ if ($malformations_name) {
+ is($ret, 0, "For $testname$malformations_name: isUTF8_CHAR() returns 0");
+ is($ret_flags, 0, " And isUTF8_CHAR_flags() returns 0");
+ }
+ else {
+ is($ret, $this_length, "For $testname: isUTF8_CHAR() returns"
+ . " expected length: $this_length");
+ is($ret_flags, $this_length,
+ " And isUTF8_CHAR_flags(...,0) returns expected"
+ . " length: $this_length");
+ }
+ is(scalar @warnings_gotten, 0,
+ " And neither isUTF8_CHAR() nor isUTF8_CHAR()_flags"
+ . " generated any warnings")
+ or output_warnings(@warnings_gotten);
+
+ undef @warnings_gotten;
+ $ret = test_isSTRICT_UTF8_CHAR($this_bytes, $this_length);
+ if ($malformations_name) {
+ is($ret, 0, " And isSTRICT_UTF8_CHAR() returns 0");
+ }
+ else {
+ my $expected_ret
+ = ( $testname =~ /surrogate|non-character/
+ || $allowed_uv > 0x10FFFF)
+ ? 0
+ : $this_length;
+ is($ret, $expected_ret,
+ " And isSTRICT_UTF8_CHAR() returns expected"
+ . " length: $expected_ret");
+ $ret = test_isUTF8_CHAR_flags($this_bytes, $this_length,
+ $::UTF8_DISALLOW_ILLEGAL_INTERCHANGE);
+ is($ret, $expected_ret,
+ " And isUTF8_CHAR_flags('"
+ . "DISALLOW_ILLEGAL_INTERCHANGE') acts like"
+ . " isSTRICT_UTF8_CHAR");
+ }
+ is(scalar @warnings_gotten, 0,
+ " And neither isSTRICT_UTF8_CHAR() nor"
+ . " isUTF8_CHAR_flags generated any warnings")
+ or output_warnings(@warnings_gotten);
+
+ undef @warnings_gotten;
+ $ret = test_isC9_STRICT_UTF8_CHAR($this_bytes, $this_length);
+ if ($malformations_name) {
+ is($ret, 0, " And isC9_STRICT_UTF8_CHAR() returns 0");
+ }
+ else {
+ my $expected_ret = ( $testname =~ /surrogate/
+ || $allowed_uv > 0x10FFFF)
+ ? 0
+ : $this_expected_len;
+ is($ret, $expected_ret, " And isC9_STRICT_UTF8_CHAR()"
+ . " returns expected length:"
+ . " $expected_ret");
+ $ret = test_isUTF8_CHAR_flags($this_bytes, $this_length,
+ $::UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE);
+ is($ret, $expected_ret,
+ " And isUTF8_CHAR_flags('"
+ . "DISALLOW_ILLEGAL_C9_INTERCHANGE') acts like"
+ . " isC9_STRICT_UTF8_CHAR");
+ }
+ is(scalar @warnings_gotten, 0,
+ " And neither isC9_STRICT_UTF8_CHAR() nor"
+ . " isUTF8_CHAR_flags generated any warnings")
+ or output_warnings(@warnings_gotten);
+
+ foreach my $disallow_type (0..2) {
+ # 0 is don't disallow this type of code point
+ # 1 is do disallow
+ # 2 is do disallow, but only code points requiring
+ # perl-extended-UTF8
+
+ my $disallow_flags;
+ my $expected_ret;
+
+ if ($malformations_name) {
+
+ # Malformations are by default disallowed, so testing
+ # with $disallow_type equal to 0 is sufficicient.
+ next if $disallow_type;
+
+ $disallow_flags = 0;
+ $expected_ret = 0;
+ }
+ elsif ($disallow_type == 1) {
+ $disallow_flags = $utf8n_flag_to_disallow;
+ $expected_ret = 0;
+ }
+ elsif ($disallow_type == 2) {
+ next if ! requires_extended_utf8($allowed_uv);
+ $disallow_flags = $::UTF8_DISALLOW_PERL_EXTENDED;
+ $expected_ret = 0;
+ }
+ else { # type is 0
+ $disallow_flags = $utf8n_flag_to_disallow_complement;
+ $expected_ret = $this_length;
+ }
+
+ $ret = test_isUTF8_CHAR_flags($this_bytes, $this_length,
+ $disallow_flags);
+ is($ret, $expected_ret,
+ " And isUTF8_CHAR_flags($display_bytes,"
+ . " $disallow_flags) returns $expected_ret")
+ or diag "The flags mean "
+ . flags_to_text($disallow_flags,
+ \@utf8n_flags_to_text);
+ is(scalar @warnings_gotten, 0,
+ " And isUTF8_CHAR_flags(...) generated"
+ . " no warnings")
+ or output_warnings(@warnings_gotten);
- # If one of the other malformation types is also in effect, we
- # don't know what the intended code point was.
- if ($short || $unexpected_noncont || $will_overflow) {
- push @expected_malformation_warnings, qr/overlong/;
- }
- else {
- my $wrong_bytes = display_bytes_no_quotes(
- substr($this_bytes, 0, $this_length));
- my $correct_bytes = display_bytes_no_quotes($bytes);
- my $prefix = ($allowed_uv > 0x10FFFF) ? "0x" : "U+";
- push @expected_malformation_warnings,
- qr/\QMalformed UTF-8 character: $wrong_bytes\E
- \Q (overlong; instead use\E
- \Q $correct_bytes to\E
- \Q represent $prefix$uv_string)/x;
- }
- }
- if ($short) {
- push @expected_malformation_warnings, qr/too short/;
- }
- if ($unexpected_noncont) {
- push @expected_malformation_warnings,
- qr/unexpected non-continuation byte/;
- }
+ # Test partial character handling, for each byte not a
+ # full character
+ my $did_test_partial = 0;
+ for (my $j = 1; $j < $this_length - 1; $j++) {
+ $did_test_partial = 1;
+ my $partial = substr($this_bytes, 0, $j);
+ my $ret_should_be;
+ my $comment;
+ if ($disallow_type || $malformations_name) {
+ $ret_should_be = 0;
+ $comment = "disallowed";
+
+ # The number of bytes required to tell if a
+ # sequence has something wrong is the smallest of
+ # all the things wrong with it. We start with the
+ # number for this type of code point, if that is
+ # disallowed; or the whole length if not. The
+ # latter is what a couple of the malformations
+ # require.
+ my $needed_to_tell = ($disallow_type)
+ ? $this_needed_to_discern_len
+ : $this_expected_len;
+
+ # Then we see if the malformations that are
+ # detectable early in the string are present.
+ if ($overlong) {
+ my $dl = overlong_discern_len($this_bytes);
+ $needed_to_tell = $dl if $dl < $needed_to_tell;
+ }
+ if ($will_overflow) {
+ my $dl = overflow_discern_len($length);
+ $needed_to_tell = $dl if $dl < $needed_to_tell;
+ }
+
+ if ($j < $needed_to_tell) {
+ $ret_should_be = 1;
+ $comment .= ", but need $needed_to_tell"
+ . " bytes to discern:";
+ }
+ }
+ else {
+ $ret_should_be = 1;
+ $comment = "allowed";
+ }
+
+ undef @warnings_gotten;
+
+ $ret = test_is_utf8_valid_partial_char_flags($partial,
+ $j, $disallow_flags);
+ is($ret, $ret_should_be,
+ " And is_utf8_valid_partial_char_flags("
+ . display_bytes($partial)
+ . ", $disallow_flags), $comment: returns"
+ . " $ret_should_be")
+ or diag "The flags mean "
+ . flags_to_text($disallow_flags, \@utf8n_flags_to_text);
+ }
+
+ if ($did_test_partial) {
+ is(scalar @warnings_gotten, 0,
+ " And is_utf8_valid_partial_char_flags()"
+ . " generated no warnings for any of the lengths")
+ or output_warnings(@warnings_gotten);
+ }
+ }
+ }
- # The overflow malformation is done differently than other
- # malformations. It comes from manually typed tests in the test
- # array. We now make it be treated like one of the other
- # malformations. But some has to be deferred until the inner loop
- my $overflow_msg_pattern;
- if ($will_overflow) {
- push @malformation_names, 'overflow';
+ # Now test the to/from UTF-8 calls. There are several orthogonal
+ # variables involved. We test most possible combinations
- $overflow_msg_pattern = display_bytes_no_quotes(
- substr($this_bytes, 0, $this_expected_len));
- $overflow_msg_pattern = qr/\QMalformed UTF-8 character:\E
- \Q $overflow_msg_pattern\E
- \Q (overflows)\E/x;
- push @expected_malformation_return_flags, $::UTF8_GOT_OVERFLOW;
- $allow_flags |= $::UTF8_ALLOW_OVERFLOW if $malformed_allow_type;
+ foreach my $do_disallow (0, 1) {
+ if ($do_disallow) {
+ next if $initially_overlong || $initially_orphan;
}
-
- # And we can create the malformation-related text for the the test
- # names we eventually will generate.
- my $malformations_name = "";
- if (@malformation_names) {
- $malformations_name .= "dis" unless $malformed_allow_type;
- $malformations_name .= "allowed ";
- $malformations_name .= "malformation";
- $malformations_name .= "s" if @malformation_names > 1;
- $malformations_name .= ": ";
- $malformations_name .= join "/", @malformation_names;
- $malformations_name = " ($malformations_name)";
+ else {
+ next if $skip_most_tests;
}
- # Done setting up the malformation related stuff
-
- foreach my $do_disallow (0, 1) {
- next if $skip_most_tests && ! $do_disallow;
+ # This tests three functions. utf8n_to_uvchr_error,
+ # utf8n_to_uvchr_msgs, and uvchr_to_utf8_flags. But only the
+ # first two are variants of each other. We use a loop
+ # 'which_func' to determine which of these. uvchr_to_utf8_flags
+ # is done separately at the end of each iteration, only when
+ # which_func is 0. which_func is numeric in part so we don't
+ # have to type in the function name and risk misspelling it
+ # somewhere, and also it sets whether we are expecting warnings
+ # or not in certain places. The _msgs() version of the function
+ # expects warnings even if lexical ones are turned off, so by
+ # making its which_func == 1, we can say we want warnings;
+ # whereas the other one with the value 0, doesn't get them.
+ for my $which_func (0, 1) {
+ my $func = ($which_func)
+ ? 'utf8n_to_uvchr_msgs'
+ : 'utf8n_to_uvchr_error';
# We classify the warnings into certain "interesting" types,
# described later
foreach my $warning_type (0..4) {
next if $skip_most_tests && $warning_type != 1;
foreach my $use_warn_flag (0, 1) {
- next if $skip_most_tests && ! $use_warn_flag;
+ if ($use_warn_flag) {
+ next if $initially_overlong || $initially_orphan;
+
+ # Since utf8n_to_uvchr_msgs() expects warnings even
+ # when lexical ones are turned off, we can skip
+ # testing it when they are turned on, with little
+ # likelihood of missing an error case.
+ next if $which_func;
+ }
+ else {
+ next if $skip_most_tests;
+ }
# Finally, here is the inner loop
my $expect_warnings_for_overflow;
if ($warning_type == 0) {
- $eval_warn = "use warnings; no warnings 'deprecated'";
+ $eval_warn = "use warnings";
$expect_regular_warnings = $use_warn_flag;
# We ordinarily expect overflow warnings here. But it
}
elsif ($warning_type == 1) {
$eval_warn = "no warnings";
- $expect_regular_warnings = 0;
- $expect_warnings_for_overflow = 0;
- $expect_warnings_for_malformed = 0;
+ $expect_regular_warnings = $which_func;
+ $expect_warnings_for_overflow = $which_func;
+ $expect_warnings_for_malformed = $which_func;
}
elsif ($warning_type == 2) {
$eval_warn = "no warnings; use warnings 'utf8'";
$expect_regular_warnings = $use_warn_flag;
$expect_warnings_for_overflow
= $controlling_warning_category eq 'non_unicode';
- $expect_warnings_for_malformed = 0;
+ $expect_warnings_for_malformed = $which_func;
}
elsif ($warning_type == 4) { # Like type 3, but uses the
- # above-31-bit flags
+ # PERL_EXTENDED flags
# The complement flags were set up so that the
- # above-31-bit flags have been tested that they don't
+ # PERL_EXTENDED flags have been tested that they don't
# trigger wrongly for too small code points. And the
# flags have been set up so that those small code
# points are tested for being above Unicode. What's
# left to test is that the large code points do
- # trigger the above-31-bit flags.
+ # trigger the PERL_EXTENDED flags.
next if ! requires_extended_utf8($allowed_uv);
next if $controlling_warning_category ne 'non_unicode';
$eval_warn = "no warnings; use warnings 'non_unicode'";
$expect_regular_warnings = 1;
$expect_warnings_for_overflow = 1;
$expect_warnings_for_malformed = 0;
- $this_utf8n_flag_to_warn = $::UTF8_WARN_ABOVE_31_BIT;
+ $this_utf8n_flag_to_warn = $::UTF8_WARN_PERL_EXTENDED;
$this_utf8n_flag_to_disallow
- = $::UTF8_DISALLOW_ABOVE_31_BIT;
- $this_uvchr_flag_to_warn = $::UNICODE_WARN_ABOVE_31_BIT;
+ = $::UTF8_DISALLOW_PERL_EXTENDED;
+ $this_uvchr_flag_to_warn
+ = $::UNICODE_WARN_PERL_EXTENDED;
$this_uvchr_flag_to_disallow
- = $::UNICODE_DISALLOW_ABOVE_31_BIT;
+ = $::UNICODE_DISALLOW_PERL_EXTENDED;
}
else {
die "Unexpected warning type '$warning_type'";
# on all the other flags. That makes sure that they all
# are independent of this flag, and so we don't need to
# test them individually.
- my $this_warning_flags = ($use_warn_flag)
- ? $this_utf8n_flag_to_warn
- : $utf8n_flag_to_warn_complement;
- my $this_disallow_flags = ($do_disallow)
- ? $this_utf8n_flag_to_disallow
- : $utf8n_flag_to_disallow_complement;
+ my $this_warning_flags
+ = ($use_warn_flag)
+ ? $this_utf8n_flag_to_warn
+ : ($overlong_is_in_perl_extended_utf8
+ ? ($utf8n_flag_to_warn_complement
+ & ~$::UTF8_WARN_PERL_EXTENDED)
+ : $utf8n_flag_to_warn_complement);
+ my $this_disallow_flags
+ = ($do_disallow)
+ ? $this_utf8n_flag_to_disallow
+ : ($overlong_is_in_perl_extended_utf8
+ ? ($utf8n_flag_to_disallow_complement
+ & ~$::UTF8_DISALLOW_PERL_EXTENDED)
+ : $utf8n_flag_to_disallow_complement);
my $expected_uv = $allowed_uv;
my $this_uv_string = $uv_string;
# should emit a message or not. It's tentative
# because, even if we ordinarily would output it, we
# don't if malformations are allowed -- except an
- # overflow is also a SUPER and ABOVE_31_BIT, and if
+ # overflow is also a SUPER and PERL_EXTENDED, and if
# warnings for those are enabled, the overflow
# warning does get raised.
if ( $expect_warnings_for_overflow
&& ( $malformed_allow_type == 0
|| ( $this_warning_flags
& ($::UTF8_WARN_SUPER
- |$::UTF8_WARN_ABOVE_31_BIT))))
+ |$::UTF8_WARN_PERL_EXTENDED))))
{
push @expected_warnings, $overflow_msg_pattern;
}
# So far the array contains warnings generated by
# malformations. Add the expected regular one.
- unshift @expected_warnings, $cp_message_qr;
+ unshift @expected_warnings, $this_cp_message_qr;
# But it may need to be modified, because either of
# these malformations means we can't determine the
# expected code point.
- if ($short || $unexpected_noncont) {
+ if ( $short || $unexpected_noncont
+ || $dont_use_overlong_cp)
+ {
my $first_byte = substr($this_bytes, 0, 1);
$expected_warnings[0] = display_bytes(
substr($this_bytes, 0, $this_expected_len));
$expected_warnings[0]
= qr/[Aa]\Qny UTF-8 sequence that starts with\E
\Q $expected_warnings[0]\E
- \Q $non_cp_trailing_text\E/x;
+ \Q $this_non_cp_trailing_text\E/x;
}
}
}
}
- my $this_name = "utf8n_to_uvchr_error() $testname: "
- . (($disallowed)
- ? 'disallowed'
- : 'allowed');
- $this_name .= ", $eval_warn";
+ my $this_name = "$func() $testname: ";
+ my @scratch_expected_return_flags = @expected_return_flags;
+ if (! $initially_malformed) {
+ $this_name .= ($disallowed)
+ ? 'disallowed, '
+ : 'allowed, ';
+ }
+ $this_name .= "$eval_warn";
$this_name .= ", " . (( $this_warning_flags
& $this_utf8n_flag_to_warn)
? 'with flag for raising warnings'
my $this_flags
= $allow_flags|$this_warning_flags|$this_disallow_flags;
my $eval_text = "$eval_warn; \$ret_ref"
- . " = test_utf8n_to_uvchr_error("
+ . " = test_$func("
. "'$this_bytes', $this_length, $this_flags)";
eval "$eval_text";
- if (! ok ("$@ eq ''", "$this_name: eval succeeded"))
+ if (! ok ($@ eq "", "$this_name: eval succeeded"))
{
diag "\$@='$@'; call was: "
. utf8n_display_call($eval_text);
next;
}
+
if ($disallowed) {
is($ret_ref->[0], 0, " And returns 0")
or diag "Call was: " . utf8n_display_call($eval_text);
my $returned_flags = $ret_ref->[2];
- for (my $i = @expected_return_flags - 1; $i >= 0; $i--) {
- if ($expected_return_flags[$i] & $returned_flags) {
- if ($expected_return_flags[$i]
- == $::UTF8_DISALLOW_ABOVE_31_BIT)
- {
- pass(" Expected and got return flag for"
- . " above_31_bit");
- }
- # The first entries in this are
- # malformations
- elsif ($i > @malformation_names - 1) {
- pass(" Expected and got return flag"
- . " for " . $controlling_warning_category);
- }
- else {
- pass(" Expected and got return flag for "
- . $malformation_names[$i]
- . " malformation");
- }
- $returned_flags &= ~$expected_return_flags[$i];
- splice @expected_return_flags, $i, 1;
- }
+ for (my $i = @scratch_expected_return_flags - 1;
+ $i >= 0;
+ $i--)
+ {
+ if ($scratch_expected_return_flags[$i] & $returned_flags)
+ {
+ if ($scratch_expected_return_flags[$i]
+ == $::UTF8_GOT_PERL_EXTENDED)
+ {
+ pass(" Expected and got return flag for"
+ . " PERL_EXTENDED");
+ }
+ # The first entries in this are
+ # malformations
+ elsif ($i > @malformation_names - 1) {
+ pass(" Expected and got return flag"
+ . " for " . $controlling_warning_category);
+ }
+ else {
+ pass(" Expected and got return flag for "
+ . $malformation_names[$i]
+ . " malformation");
+ }
+ $returned_flags
+ &= ~$scratch_expected_return_flags[$i];
+ splice @scratch_expected_return_flags, $i, 1;
+ }
}
- is($returned_flags, 0,
- " Got no unexpected return flags")
- or diag "The unexpected flags gotten were: "
+ if (! is($returned_flags, 0,
+ " Got no unexpected return flags"))
+ {
+ diag "The unexpected flags gotten were: "
. (flags_to_text($returned_flags,
\@utf8n_flags_to_text)
# We strip off any prefixes from the flag
# names
=~ s/ \b [A-Z] _ //xgr);
- is (scalar @expected_return_flags, 0,
- " Got all expected return flags")
- or diag "The expected flags not gotten were: "
+ diag "Call was: " . utf8n_display_call($eval_text);
+ }
+
+ if (! is (scalar @scratch_expected_return_flags, 0,
+ " Got all expected return flags"))
+ {
+ diag "The expected flags not gotten were: "
. (flags_to_text(eval join("|",
- @expected_return_flags),
+ @scratch_expected_return_flags),
\@utf8n_flags_to_text)
# We strip off any prefixes from the flag
# names
=~ s/ \b [A-Z] _ //xgr);
+ diag "Call was: " . utf8n_display_call($eval_text);
+ }
+
+ if ($which_func) {
+ my @returned_warnings;
+ for my $element_ref (@{$ret_ref->[3]}) {
+ push @returned_warnings, $element_ref->{'text'};
+ my $text = $element_ref->{'text'};
+ my $flag = $element_ref->{'flag_bit'};
+ my $category = $element_ref->{'warning_category'};
+
+ if (! ok(($flag & ($flag-1)) == 0,
+ "flag for returned msg is a single bit"))
+ {
+ diag sprintf("flags are %x; msg=%s", $flag, $text);
+ }
+ else {
+ if (grep { $_ == $flag } @expected_return_flags) {
+ pass("flag for returned msg is expected");
+ }
+ else {
+ fail("flag for returned msg is expected: "
+ . flags_to_text($flag, \@utf8n_flags_to_text));
+ }
+ }
+
+ # In perl space, don't know the category numbers
+ isnt($category, 0,
+ "returned category for msg isn't 0");
+ }
+
+ ok(@warnings_gotten == 0, "$func raised no warnings;"
+ . " the next tests are for ones in the returned"
+ . " variable")
+ or diag join "\n", "The unexpected warnings were:",
+ @warnings_gotten;
+ @warnings_gotten = @returned_warnings;
+ }
do_warnings_test(@expected_warnings)
or diag "Call was: " . utf8n_display_call($eval_text);
# Check CHECK_ONLY results when the input is
# disallowed. Do this when actually disallowed,
- # not just when the $this_disallow_flags is set
- if ($disallowed) {
+ # not just when the $this_disallow_flags is set. We only
+ # test once utf8n_to_uvchr_msgs() with this.
+ if ( $disallowed
+ && ($which_func == 0 || ! $tested_CHECK_ONLY))
+ {
+ $tested_CHECK_ONLY = 1;
my $this_flags = $this_disallow_flags|$::UTF8_CHECK_ONLY;
- my $eval_text = "use warnings; no warnings 'deprecated'; \$ret_ref ="
- . " test_utf8n_to_uvchr_error('"
+ my $eval_text = "use warnings; \$ret_ref ="
+ . " test_$func('"
. "$this_bytes', $this_length,"
. " $this_flags)";
eval $eval_text;
- if (! ok ("$@ eq ''",
+ if (! ok ($@ eq "",
" And eval succeeded with CHECK_ONLY"))
{
diag "\$@='$@'; Call was: "
# existing code point, it hasn't overflowed, and isn't
# malformed.
next if @malformation_names;
+ next if $which_func;
$this_warning_flags = ($use_warn_flag)
? $this_uvchr_flag_to_warn
. " test_uvchr_to_utf8_flags("
. "$allowed_uv, $this_flags)";
eval "$eval_text";
- if (! ok ("$@ eq ''", "$this_name: eval succeeded"))
+ if (! ok ($@ eq "", "$this_name: eval succeeded"))
{
diag "\$@='$@'; call was: "
. uvchr_display_call($eval_text);
}
}
}
+ }
}
done_testing;