This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
APItest/t/utf8.t: Fix EBCDIC test
[perl5.git] / ext / XS-APItest / t / utf8.t
1 #!perl -w
2
3 use strict;
4 use Test::More;
5 $|=1;
6
7 no warnings 'deprecated'; # Some of the below are above IV_MAX on 32 bit
8                           # machines, and that is tested elsewhere
9
10 use XS::APItest;
11
12 my $pound_sign = chr utf8::unicode_to_native(163);
13
14 sub isASCII { ord "A" == 65 }
15
16 sub display_bytes {
17     use bytes;
18     my $string = shift;
19     return   '"'
20            . join("", map { sprintf("\\x%02x", ord $_) } split "", $string)
21            . '"';
22 }
23
24 # This  test file can't use byte_utf8a_to_utf8n() from t/charset_tools.pl
25 # because that uses the same functions we are testing here.  So UTF-EBCDIC
26 # strings are hard-coded as I8 strings in this file instead, and we use array
27 # lookup to translate into the appropriate code page.
28
29 my @i8_to_native = (    # Only code page 1047 so far.
30 # _0   _1   _2   _3   _4   _5   _6   _7   _8   _9   _A   _B   _C   _D   _E   _F
31 0x00,0x01,0x02,0x03,0x37,0x2D,0x2E,0x2F,0x16,0x05,0x15,0x0B,0x0C,0x0D,0x0E,0x0F,
32 0x10,0x11,0x12,0x13,0x3C,0x3D,0x32,0x26,0x18,0x19,0x3F,0x27,0x1C,0x1D,0x1E,0x1F,
33 0x40,0x5A,0x7F,0x7B,0x5B,0x6C,0x50,0x7D,0x4D,0x5D,0x5C,0x4E,0x6B,0x60,0x4B,0x61,
34 0xF0,0xF1,0xF2,0xF3,0xF4,0xF5,0xF6,0xF7,0xF8,0xF9,0x7A,0x5E,0x4C,0x7E,0x6E,0x6F,
35 0x7C,0xC1,0xC2,0xC3,0xC4,0xC5,0xC6,0xC7,0xC8,0xC9,0xD1,0xD2,0xD3,0xD4,0xD5,0xD6,
36 0xD7,0xD8,0xD9,0xE2,0xE3,0xE4,0xE5,0xE6,0xE7,0xE8,0xE9,0xAD,0xE0,0xBD,0x5F,0x6D,
37 0x79,0x81,0x82,0x83,0x84,0x85,0x86,0x87,0x88,0x89,0x91,0x92,0x93,0x94,0x95,0x96,
38 0x97,0x98,0x99,0xA2,0xA3,0xA4,0xA5,0xA6,0xA7,0xA8,0xA9,0xC0,0x4F,0xD0,0xA1,0x07,
39 0x20,0x21,0x22,0x23,0x24,0x25,0x06,0x17,0x28,0x29,0x2A,0x2B,0x2C,0x09,0x0A,0x1B,
40 0x30,0x31,0x1A,0x33,0x34,0x35,0x36,0x08,0x38,0x39,0x3A,0x3B,0x04,0x14,0x3E,0xFF,
41 0x41,0x42,0x43,0x44,0x45,0x46,0x47,0x48,0x49,0x4A,0x51,0x52,0x53,0x54,0x55,0x56,
42 0x57,0x58,0x59,0x62,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0x6A,0x70,0x71,0x72,0x73,
43 0x74,0x75,0x76,0x77,0x78,0x80,0x8A,0x8B,0x8C,0x8D,0x8E,0x8F,0x90,0x9A,0x9B,0x9C,
44 0x9D,0x9E,0x9F,0xA0,0xAA,0xAB,0xAC,0xAE,0xAF,0xB0,0xB1,0xB2,0xB3,0xB4,0xB5,0xB6,
45 0xB7,0xB8,0xB9,0xBA,0xBB,0xBC,0xBE,0xBF,0xCA,0xCB,0xCC,0xCD,0xCE,0xCF,0xDA,0xDB,
46 0xDC,0xDD,0xDE,0xDF,0xE1,0xEA,0xEB,0xEC,0xED,0xEE,0xEF,0xFA,0xFB,0xFC,0xFD,0xFE,
47 );
48
49 my @native_to_i8;
50 for (my $i = 0; $i < 256; $i++) {
51     $native_to_i8[$i8_to_native[$i]] = $i;
52 }
53
54 *I8_to_native = (isASCII)
55                     ? sub { return shift }
56                     : sub { return join "", map { chr $i8_to_native[ord $_] }
57                                             split "", shift };
58 *native_to_I8 = (isASCII)
59                     ? sub { return shift }
60                     : sub { return join "", map { chr $native_to_i8[ord $_] }
61                                             split "", shift };
62
63 my $is64bit = length sprintf("%x", ~0) > 8;
64
65
66 # Test utf8n_to_uvchr().  These provide essentially complete code coverage.
67 # Copied from utf8.h
68 my $UTF8_ALLOW_EMPTY            = 0x0001;
69 my $UTF8_ALLOW_CONTINUATION     = 0x0002;
70 my $UTF8_ALLOW_NON_CONTINUATION = 0x0004;
71 my $UTF8_ALLOW_SHORT            = 0x0008;
72 my $UTF8_ALLOW_LONG             = 0x0010;
73 my $UTF8_DISALLOW_SURROGATE     = 0x0020;
74 my $UTF8_WARN_SURROGATE         = 0x0040;
75 my $UTF8_DISALLOW_NONCHAR       = 0x0080;
76 my $UTF8_WARN_NONCHAR           = 0x0100;
77 my $UTF8_DISALLOW_SUPER         = 0x0200;
78 my $UTF8_WARN_SUPER             = 0x0400;
79 my $UTF8_DISALLOW_ABOVE_31_BIT  = 0x0800;
80 my $UTF8_WARN_ABOVE_31_BIT      = 0x1000;
81 my $UTF8_CHECK_ONLY             = 0x2000;
82 my $UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE
83                              = $UTF8_DISALLOW_SUPER|$UTF8_DISALLOW_SURROGATE;
84 my $UTF8_DISALLOW_ILLEGAL_INTERCHANGE
85               = $UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE|$UTF8_DISALLOW_NONCHAR;
86
87 # Test uvchr_to_utf8().
88 my $UNICODE_WARN_SURROGATE        = 0x0001;
89 my $UNICODE_WARN_NONCHAR          = 0x0002;
90 my $UNICODE_WARN_SUPER            = 0x0004;
91 my $UNICODE_WARN_ABOVE_31_BIT     = 0x0008;
92 my $UNICODE_DISALLOW_SURROGATE    = 0x0010;
93 my $UNICODE_DISALLOW_NONCHAR      = 0x0020;
94 my $UNICODE_DISALLOW_SUPER        = 0x0040;
95 my $UNICODE_DISALLOW_ABOVE_31_BIT = 0x0080;
96
97 my $look_for_everything_utf8n_to
98                         = $UTF8_DISALLOW_SURROGATE
99                         | $UTF8_WARN_SURROGATE
100                         | $UTF8_DISALLOW_NONCHAR
101                         | $UTF8_WARN_NONCHAR
102                         | $UTF8_DISALLOW_SUPER
103                         | $UTF8_WARN_SUPER
104                         | $UTF8_DISALLOW_ABOVE_31_BIT
105                         | $UTF8_WARN_ABOVE_31_BIT;
106 my $look_for_everything_uvchr_to
107                         = $UNICODE_DISALLOW_SURROGATE
108                         | $UNICODE_WARN_SURROGATE
109                         | $UNICODE_DISALLOW_NONCHAR
110                         | $UNICODE_WARN_NONCHAR
111                         | $UNICODE_DISALLOW_SUPER
112                         | $UNICODE_WARN_SUPER
113                         | $UNICODE_DISALLOW_ABOVE_31_BIT
114                         | $UNICODE_WARN_ABOVE_31_BIT;
115
116 foreach ([0, '', '', 'empty'],
117          [0, 'N', 'N', '1 char'],
118          [1, 'NN', 'N', '1 char substring'],
119          [-2, 'Perl', 'Rules', 'different'],
120          [0, $pound_sign, $pound_sign, 'pound sign'],
121          [1, $pound_sign . 10, $pound_sign . 1, '10 pounds is more than 1 pound'],
122          [1, $pound_sign . $pound_sign, $pound_sign, '2 pound signs are more than 1'],
123          [-2, ' $!', " \x{1F42B}!", 'Camels are worth more than 1 dollar'],
124          [-1, '!', "!\x{1F42A}", 'Initial substrings match'],
125         ) {
126     my ($expect, $left, $right, $desc) = @$_;
127     my $copy = $right;
128     utf8::encode($copy);
129     is(bytes_cmp_utf8($left, $copy), $expect, $desc);
130     next if $right =~ tr/\0-\377//c;
131     utf8::encode($left);
132     is(bytes_cmp_utf8($right, $left), -$expect, "$desc reversed");
133 }
134
135 # The keys to this hash are Unicode code points, their values are the native
136 # UTF-8 representations of them.  The code points are chosen because they are
137 # "interesting" on either or both ASCII and EBCDIC platforms.  First we add
138 # boundaries where the number of bytes required to represent them increase, or
139 # are adjacent to problematic code points, so we want to make sure they aren't
140 # considered problematic.
141 my %code_points = (
142     0x0100     => (isASCII) ? "\xc4\x80" : I8_to_native("\xc8\xa0"),
143     0x0400 - 1 => (isASCII) ? "\xcf\xbf" : I8_to_native("\xdf\xbf"),
144     0x0400     => (isASCII) ? "\xd0\x80" : I8_to_native("\xe1\xa0\xa0"),
145     0x0800 - 1 => (isASCII) ? "\xdf\xbf"     : I8_to_native("\xe1\xbf\xbf"),
146     0x0800     => (isASCII) ? "\xe0\xa0\x80" : I8_to_native("\xe2\xa0\xa0"),
147     0x4000 - 1 => (isASCII) ? "\xe3\xbf\xbf" : I8_to_native("\xef\xbf\xbf"),
148     0x4000     => (isASCII) ? "\xe4\x80\x80" : I8_to_native("\xf0\xb0\xa0\xa0"),
149     0x8000 - 1 => (isASCII) ? "\xe7\xbf\xbf" : I8_to_native("\xf0\xbf\xbf\xbf"),
150
151     # First code point that the implementation of isUTF8_POSSIBLY_PROBLEMATIC,
152     # as of this writing, considers potentially problematic on EBCDIC
153     0x8000     => (isASCII) ? "\xe8\x80\x80" : I8_to_native("\xf1\xa0\xa0\xa0"),
154
155     0xD000 - 1 => (isASCII) ? "\xec\xbf\xbf" : I8_to_native("\xf1\xb3\xbf\xbf"),
156
157     # First code point that the implementation of isUTF8_POSSIBLY_PROBLEMATIC,
158     # as of this writing, considers potentially problematic on ASCII
159     0xD000     => (isASCII) ? "\xed\x80\x80" : I8_to_native("\xf1\xb4\xa0\xa0"),
160
161     # Bracket the surrogates, and include several surrogates
162     0xD7FF      => (isASCII) ? "\xed\x9f\xbf" : I8_to_native("\xf1\xb5\xbf\xbf"),
163     0xD800      => (isASCII) ? "\xed\xa0\x80" : I8_to_native("\xf1\xb6\xa0\xa0"),
164     0xDC00      => (isASCII) ? "\xed\xb0\x80" : I8_to_native("\xf1\xb7\xa0\xa0"),
165     0xDFFF      => (isASCII) ? "\xee\x80\x80" : I8_to_native("\xf1\xb8\xa0\xa0"),
166     0xDFFF      => (isASCII) ? "\xed\xbf\xbf" : I8_to_native("\xf1\xb7\xbf\xbf"),
167     0xE000      => (isASCII) ? "\xee\x80\x80" : I8_to_native("\xf1\xb8\xa0\xa0"),
168
169     # Include the 32 contiguous non characters, and surrounding code points
170     0xFDCF      => (isASCII) ? "\xef\xb7\x8f" : I8_to_native("\xf1\xbf\xae\xaf"),
171     0xFDD0      => (isASCII) ? "\xef\xb7\x90" : I8_to_native("\xf1\xbf\xae\xb0"),
172     0xFDD1      => (isASCII) ? "\xef\xb7\x91" : I8_to_native("\xf1\xbf\xae\xb1"),
173     0xFDD2      => (isASCII) ? "\xef\xb7\x92" : I8_to_native("\xf1\xbf\xae\xb2"),
174     0xFDD3      => (isASCII) ? "\xef\xb7\x93" : I8_to_native("\xf1\xbf\xae\xb3"),
175     0xFDD4      => (isASCII) ? "\xef\xb7\x94" : I8_to_native("\xf1\xbf\xae\xb4"),
176     0xFDD5      => (isASCII) ? "\xef\xb7\x95" : I8_to_native("\xf1\xbf\xae\xb5"),
177     0xFDD6      => (isASCII) ? "\xef\xb7\x96" : I8_to_native("\xf1\xbf\xae\xb6"),
178     0xFDD7      => (isASCII) ? "\xef\xb7\x97" : I8_to_native("\xf1\xbf\xae\xb7"),
179     0xFDD8      => (isASCII) ? "\xef\xb7\x98" : I8_to_native("\xf1\xbf\xae\xb8"),
180     0xFDD9      => (isASCII) ? "\xef\xb7\x99" : I8_to_native("\xf1\xbf\xae\xb9"),
181     0xFDDA      => (isASCII) ? "\xef\xb7\x9a" : I8_to_native("\xf1\xbf\xae\xba"),
182     0xFDDB      => (isASCII) ? "\xef\xb7\x9b" : I8_to_native("\xf1\xbf\xae\xbb"),
183     0xFDDC      => (isASCII) ? "\xef\xb7\x9c" : I8_to_native("\xf1\xbf\xae\xbc"),
184     0xFDDD      => (isASCII) ? "\xef\xb7\x9d" : I8_to_native("\xf1\xbf\xae\xbd"),
185     0xFDDE      => (isASCII) ? "\xef\xb7\x9e" : I8_to_native("\xf1\xbf\xae\xbe"),
186     0xFDDF      => (isASCII) ? "\xef\xb7\x9f" : I8_to_native("\xf1\xbf\xae\xbf"),
187     0xFDE0      => (isASCII) ? "\xef\xb7\xa0" : I8_to_native("\xf1\xbf\xaf\xa0"),
188     0xFDE1      => (isASCII) ? "\xef\xb7\xa1" : I8_to_native("\xf1\xbf\xaf\xa1"),
189     0xFDE2      => (isASCII) ? "\xef\xb7\xa2" : I8_to_native("\xf1\xbf\xaf\xa2"),
190     0xFDE3      => (isASCII) ? "\xef\xb7\xa3" : I8_to_native("\xf1\xbf\xaf\xa3"),
191     0xFDE4      => (isASCII) ? "\xef\xb7\xa4" : I8_to_native("\xf1\xbf\xaf\xa4"),
192     0xFDE5      => (isASCII) ? "\xef\xb7\xa5" : I8_to_native("\xf1\xbf\xaf\xa5"),
193     0xFDE6      => (isASCII) ? "\xef\xb7\xa6" : I8_to_native("\xf1\xbf\xaf\xa6"),
194     0xFDE7      => (isASCII) ? "\xef\xb7\xa7" : I8_to_native("\xf1\xbf\xaf\xa7"),
195     0xFDE8      => (isASCII) ? "\xef\xb7\xa8" : I8_to_native("\xf1\xbf\xaf\xa8"),
196     0xFDEa      => (isASCII) ? "\xef\xb7\x99" : I8_to_native("\xf1\xbf\xaf\xa9"),
197     0xFDEA      => (isASCII) ? "\xef\xb7\xaa" : I8_to_native("\xf1\xbf\xaf\xaa"),
198     0xFDEB      => (isASCII) ? "\xef\xb7\xab" : I8_to_native("\xf1\xbf\xaf\xab"),
199     0xFDEC      => (isASCII) ? "\xef\xb7\xac" : I8_to_native("\xf1\xbf\xaf\xac"),
200     0xFDED      => (isASCII) ? "\xef\xb7\xad" : I8_to_native("\xf1\xbf\xaf\xad"),
201     0xFDEE      => (isASCII) ? "\xef\xb7\xae" : I8_to_native("\xf1\xbf\xaf\xae"),
202     0xFDEF      => (isASCII) ? "\xef\xb7\xaf" : I8_to_native("\xf1\xbf\xaf\xaf"),
203     0xFDF0      => (isASCII) ? "\xef\xb7\xb0" : I8_to_native("\xf1\xbf\xaf\xb0"),
204
205     # Mostly around non-characters, but some are transitions to longer strings
206     0xFFFD      => (isASCII) ? "\xef\xbf\xbd" : I8_to_native("\xf1\xbf\xbf\xbd"),
207     0x10000 - 1 => (isASCII) ? "\xef\xbf\xbf" : I8_to_native("\xf1\xbf\xbf\xbf"),
208     0x10000     => (isASCII) ? "\xf0\x90\x80\x80" : I8_to_native("\xf2\xa0\xa0\xa0"),
209     0x1FFFD     => (isASCII) ? "\xf0\x9f\xbf\xbd" : I8_to_native("\xf3\xbf\xbf\xbd"),
210     0x1FFFE     => (isASCII) ? "\xf0\x9f\xbf\xbe" : I8_to_native("\xf3\xbf\xbf\xbe"),
211     0x1FFFF     => (isASCII) ? "\xf0\x9f\xbf\xbf" : I8_to_native("\xf3\xbf\xbf\xbf"),
212     0x20000     => (isASCII) ? "\xf0\xa0\x80\x80" : I8_to_native("\xf4\xa0\xa0\xa0"),
213     0x2FFFD     => (isASCII) ? "\xf0\xaf\xbf\xbd" : I8_to_native("\xf5\xbf\xbf\xbd"),
214     0x2FFFE     => (isASCII) ? "\xf0\xaf\xbf\xbe" : I8_to_native("\xf5\xbf\xbf\xbe"),
215     0x2FFFF     => (isASCII) ? "\xf0\xaf\xbf\xbf" : I8_to_native("\xf5\xbf\xbf\xbf"),
216     0x30000     => (isASCII) ? "\xf0\xb0\x80\x80" : I8_to_native("\xf6\xa0\xa0\xa0"),
217     0x3FFFD     => (isASCII) ? "\xf0\xbf\xbf\xbd" : I8_to_native("\xf7\xbf\xbf\xbd"),
218     0x3FFFE     => (isASCII) ? "\xf0\xbf\xbf\xbe" : I8_to_native("\xf7\xbf\xbf\xbe"),
219     0x40000 - 1 => (isASCII) ? "\xf0\xbf\xbf\xbf" : I8_to_native("\xf7\xbf\xbf\xbf"),
220     0x40000     => (isASCII) ? "\xf1\x80\x80\x80" : I8_to_native("\xf8\xa8\xa0\xa0\xa0"),
221     0x4FFFD     => (isASCII) ? "\xf1\x8f\xbf\xbd" : I8_to_native("\xf8\xa9\xbf\xbf\xbd"),
222     0x4FFFE     => (isASCII) ? "\xf1\x8f\xbf\xbe" : I8_to_native("\xf8\xa9\xbf\xbf\xbe"),
223     0x4FFFF     => (isASCII) ? "\xf1\x8f\xbf\xbf" : I8_to_native("\xf8\xa9\xbf\xbf\xbf"),
224     0x50000     => (isASCII) ? "\xf1\x90\x80\x80" : I8_to_native("\xf8\xaa\xa0\xa0\xa0"),
225     0x5FFFD     => (isASCII) ? "\xf1\x9f\xbf\xbd" : I8_to_native("\xf8\xab\xbf\xbf\xbd"),
226     0x5FFFE     => (isASCII) ? "\xf1\x9f\xbf\xbe" : I8_to_native("\xf8\xab\xbf\xbf\xbe"),
227     0x5FFFF     => (isASCII) ? "\xf1\x9f\xbf\xbf" : I8_to_native("\xf8\xab\xbf\xbf\xbf"),
228     0x60000     => (isASCII) ? "\xf1\xa0\x80\x80" : I8_to_native("\xf8\xac\xa0\xa0\xa0"),
229     0x6FFFD     => (isASCII) ? "\xf1\xaf\xbf\xbd" : I8_to_native("\xf8\xad\xbf\xbf\xbd"),
230     0x6FFFE     => (isASCII) ? "\xf1\xaf\xbf\xbe" : I8_to_native("\xf8\xad\xbf\xbf\xbe"),
231     0x6FFFF     => (isASCII) ? "\xf1\xaf\xbf\xbf" : I8_to_native("\xf8\xad\xbf\xbf\xbf"),
232     0x70000     => (isASCII) ? "\xf1\xb0\x80\x80" : I8_to_native("\xf8\xae\xa0\xa0\xa0"),
233     0x7FFFD     => (isASCII) ? "\xf1\xbf\xbf\xbd" : I8_to_native("\xf8\xaf\xbf\xbf\xbd"),
234     0x7FFFE     => (isASCII) ? "\xf1\xbf\xbf\xbe" : I8_to_native("\xf8\xaf\xbf\xbf\xbe"),
235     0x7FFFF     => (isASCII) ? "\xf1\xbf\xbf\xbf" : I8_to_native("\xf8\xaf\xbf\xbf\xbf"),
236     0x80000     => (isASCII) ? "\xf2\x80\x80\x80" : I8_to_native("\xf8\xb0\xa0\xa0\xa0"),
237     0x8FFFD     => (isASCII) ? "\xf2\x8f\xbf\xbd" : I8_to_native("\xf8\xb1\xbf\xbf\xbd"),
238     0x8FFFE     => (isASCII) ? "\xf2\x8f\xbf\xbe" : I8_to_native("\xf8\xb1\xbf\xbf\xbe"),
239     0x8FFFF     => (isASCII) ? "\xf2\x8f\xbf\xbf" : I8_to_native("\xf8\xb1\xbf\xbf\xbf"),
240     0x90000     => (isASCII) ? "\xf2\x90\x80\x80" : I8_to_native("\xf8\xb2\xa0\xa0\xa0"),
241     0x9FFFD     => (isASCII) ? "\xf2\x9f\xbf\xbd" : I8_to_native("\xf8\xb3\xbf\xbf\xbd"),
242     0x9FFFE     => (isASCII) ? "\xf2\x9f\xbf\xbe" : I8_to_native("\xf8\xb3\xbf\xbf\xbe"),
243     0x9FFFF     => (isASCII) ? "\xf2\x9f\xbf\xbf" : I8_to_native("\xf8\xb3\xbf\xbf\xbf"),
244     0xA0000     => (isASCII) ? "\xf2\xa0\x80\x80" : I8_to_native("\xf8\xb4\xa0\xa0\xa0"),
245     0xAFFFD     => (isASCII) ? "\xf2\xaf\xbf\xbd" : I8_to_native("\xf8\xb5\xbf\xbf\xbd"),
246     0xAFFFE     => (isASCII) ? "\xf2\xaf\xbf\xbe" : I8_to_native("\xf8\xb5\xbf\xbf\xbe"),
247     0xAFFFF     => (isASCII) ? "\xf2\xaf\xbf\xbf" : I8_to_native("\xf8\xb5\xbf\xbf\xbf"),
248     0xB0000     => (isASCII) ? "\xf2\xb0\x80\x80" : I8_to_native("\xf8\xb6\xa0\xa0\xa0"),
249     0xBFFFD     => (isASCII) ? "\xf2\xbf\xbf\xbd" : I8_to_native("\xf8\xb7\xbf\xbf\xbd"),
250     0xBFFFE     => (isASCII) ? "\xf2\xbf\xbf\xbe" : I8_to_native("\xf8\xb7\xbf\xbf\xbe"),
251     0xBFFFF     => (isASCII) ? "\xf2\xbf\xbf\xbf" : I8_to_native("\xf8\xb7\xbf\xbf\xbf"),
252     0xC0000     => (isASCII) ? "\xf3\x80\x80\x80" : I8_to_native("\xf8\xb8\xa0\xa0\xa0"),
253     0xCFFFD     => (isASCII) ? "\xf3\x8f\xbf\xbd" : I8_to_native("\xf8\xb9\xbf\xbf\xbd"),
254     0xCFFFE     => (isASCII) ? "\xf3\x8f\xbf\xbe" : I8_to_native("\xf8\xb9\xbf\xbf\xbe"),
255     0xCFFFF     => (isASCII) ? "\xf3\x8f\xbf\xbf" : I8_to_native("\xf8\xb9\xbf\xbf\xbf"),
256     0xD0000     => (isASCII) ? "\xf3\x90\x80\x80" : I8_to_native("\xf8\xba\xa0\xa0\xa0"),
257     0xDFFFD     => (isASCII) ? "\xf3\x9f\xbf\xbd" : I8_to_native("\xf8\xbb\xbf\xbf\xbd"),
258     0xDFFFE     => (isASCII) ? "\xf3\x9f\xbf\xbe" : I8_to_native("\xf8\xbb\xbf\xbf\xbe"),
259     0xDFFFF     => (isASCII) ? "\xf3\x9f\xbf\xbf" : I8_to_native("\xf8\xbb\xbf\xbf\xbf"),
260     0xE0000     => (isASCII) ? "\xf3\xa0\x80\x80" : I8_to_native("\xf8\xbc\xa0\xa0\xa0"),
261     0xEFFFD     => (isASCII) ? "\xf3\xaf\xbf\xbd" : I8_to_native("\xf8\xbd\xbf\xbf\xbd"),
262     0xEFFFE     => (isASCII) ? "\xf3\xaf\xbf\xbe" : I8_to_native("\xf8\xbd\xbf\xbf\xbe"),
263     0xEFFFF     => (isASCII) ? "\xf3\xaf\xbf\xbf" : I8_to_native("\xf8\xbd\xbf\xbf\xbf"),
264     0xF0000     => (isASCII) ? "\xf3\xb0\x80\x80" : I8_to_native("\xf8\xbe\xa0\xa0\xa0"),
265     0xFFFFD     => (isASCII) ? "\xf3\xbf\xbf\xbd" : I8_to_native("\xf8\xbf\xbf\xbf\xbd"),
266     0xFFFFE     => (isASCII) ? "\xf3\xbf\xbf\xbe" : I8_to_native("\xf8\xbf\xbf\xbf\xbe"),
267     0xFFFFF     => (isASCII) ? "\xf3\xbf\xbf\xbf" : I8_to_native("\xf8\xbf\xbf\xbf\xbf"),
268     0x100000    => (isASCII) ? "\xf4\x80\x80\x80" : I8_to_native("\xf9\xa0\xa0\xa0\xa0"),
269     0x10FFFD    => (isASCII) ? "\xf4\x8f\xbf\xbd" : I8_to_native("\xf9\xa1\xbf\xbf\xbd"),
270     0x10FFFE    => (isASCII) ? "\xf4\x8f\xbf\xbe" : I8_to_native("\xf9\xa1\xbf\xbf\xbe"),
271     0x10FFFF    => (isASCII) ? "\xf4\x8f\xbf\xbf" : I8_to_native("\xf9\xa1\xbf\xbf\xbf"),
272     0x110000    => (isASCII) ? "\xf4\x90\x80\x80" : I8_to_native("\xf9\xa2\xa0\xa0\xa0"),
273
274     # Things that would be noncharacters if they were in Unicode, and might be
275     # mistaken, if the C code is bad, to be nonchars
276     0x11FFFE    => (isASCII) ? "\xf4\x9f\xbf\xbe" : I8_to_native("\xf9\xa3\xbf\xbf\xbe"),
277     0x11FFFF    => (isASCII) ? "\xf4\x9f\xbf\xbf" : I8_to_native("\xf9\xa3\xbf\xbf\xbf"),
278     0x20FFFE    => (isASCII) ? "\xf8\x88\x8f\xbf\xbe" : I8_to_native("\xfa\xa1\xbf\xbf\xbe"),
279     0x20FFFF    => (isASCII) ? "\xf8\x88\x8f\xbf\xbf" : I8_to_native("\xfa\xa1\xbf\xbf\xbf"),
280
281     0x200000 - 1 => (isASCII) ? "\xf7\xbf\xbf\xbf" : I8_to_native("\xf9\xbf\xbf\xbf\xbf"),
282     0x200000     => (isASCII) ? "\xf8\x88\x80\x80\x80" : I8_to_native("\xfa\xa0\xa0\xa0\xa0"),
283     0x400000 - 1 => (isASCII) ? "\xf8\x8f\xbf\xbf\xbf" : I8_to_native("\xfb\xbf\xbf\xbf\xbf"),
284     0x400000     => (isASCII) ? "\xf8\x90\x80\x80\x80" : I8_to_native("\xfc\xa4\xa0\xa0\xa0\xa0"),
285     0x4000000 - 1 => (isASCII) ? "\xfb\xbf\xbf\xbf\xbf" : I8_to_native("\xfd\xbf\xbf\xbf\xbf\xbf"),
286     0x4000000     => (isASCII) ? "\xfc\x84\x80\x80\x80\x80" : I8_to_native("\xfe\xa2\xa0\xa0\xa0\xa0\xa0"),
287     0x4000000 - 1 => (isASCII) ? "\xfb\xbf\xbf\xbf\xbf" : I8_to_native("\xfd\xbf\xbf\xbf\xbf\xbf"),
288     0x4000000     => (isASCII) ? "\xfc\x84\x80\x80\x80\x80" : I8_to_native("\xfe\xa2\xa0\xa0\xa0\xa0\xa0"),
289     0x40000000 - 1 => (isASCII) ? "\xfc\xbf\xbf\xbf\xbf\xbf" : I8_to_native("\xfe\xbf\xbf\xbf\xbf\xbf\xbf"),
290     0x40000000     => (isASCII) ? "\xfd\x80\x80\x80\x80\x80" : I8_to_native("\xff\xa0\xa0\xa0\xa0\xa0\xa0\xa1\xa0\xa0\xa0\xa0\xa0\xa0"),
291     0x80000000 - 1 => (isASCII) ? "\xfd\xbf\xbf\xbf\xbf\xbf" : I8_to_native("\xff\xa0\xa0\xa0\xa0\xa0\xa0\xa1\xbf\xbf\xbf\xbf\xbf\xbf"),
292     0x80000000     => (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"),
293     0xFFFFFFFF     => (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"),
294 );
295
296 if ($is64bit) {
297     no warnings qw(overflow portable);
298     $code_points{0x100000000}        = (isASCII)
299                                         ?              "\xfe\x84\x80\x80\x80\x80\x80"
300                                         : I8_to_native("\xff\xa0\xa0\xa0\xa0\xa0\xa0\xa4\xa0\xa0\xa0\xa0\xa0\xa0");
301     $code_points{0x1000000000 - 1}   = (isASCII)
302                                         ?              "\xfe\xbf\xbf\xbf\xbf\xbf\xbf"
303                                         : I8_to_native("\xff\xa0\xa0\xa0\xa0\xa0\xa1\xbf\xbf\xbf\xbf\xbf\xbf\xbf");
304     $code_points{0x1000000000}       = (isASCII)
305                                         ?              "\xff\x80\x80\x80\x80\x80\x81\x80\x80\x80\x80\x80\x80"
306                                         : I8_to_native("\xff\xa0\xa0\xa0\xa0\xa0\xa2\xa0\xa0\xa0\xa0\xa0\xa0\xa0");
307     $code_points{0xFFFFFFFFFFFFFFFF} = (isASCII)
308                                         ?              "\xff\x80\x8f\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf"
309                                         : I8_to_native("\xff\xaf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf");
310     if (isASCII) {  # These could falsely show as overlongs in a naive implementation
311         $code_points{0x40000000000}  = "\xff\x80\x80\x80\x80\x81\x80\x80\x80\x80\x80\x80\x80";
312         $code_points{0x1000000000000} = "\xff\x80\x80\x80\x81\x80\x80\x80\x80\x80\x80\x80\x80";
313         $code_points{0x40000000000000} = "\xff\x80\x80\x81\x80\x80\x80\x80\x80\x80\x80\x80\x80";
314         $code_points{0x1000000000000000} = "\xff\x80\x81\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80";
315         # overflows
316         #$code_points{0xfoo}     = "\xff\x81\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80";
317     }
318 }
319 elsif (! isASCII) { # 32-bit EBCDIC.  64-bit is clearer to handle, so doesn't need this test case
320     no warnings qw(overflow portable);
321     $code_points{0x40000000} = I8_to_native("\xff\xa0\xa0\xa0\xa0\xa0\xa0\xa1\xa0\xa0\xa0\xa0\xa0\xa0");
322 }
323
324 # Now add in entries for each of code points 0-255, which require special
325 # handling on EBCDIC.  Remember the keys are Unicode values, and the values
326 # are the native UTF-8.  For invariants, the bytes are just the native chr.
327
328 my $cp = 0;
329 while ($cp < ((isASCII) ? 128 : 160)) {   # This is from the definition of
330                                         # invariant
331     $code_points{$cp} = chr utf8::unicode_to_native($cp);
332     $cp++;
333 }
334
335 # Done with the invariants.  Now do the variants.  All in this range are 2
336 # byte.  Again, we can't use the internal functions to generate UTF-8, as
337 # those are what we are trying to test.  In the loop, we know what range the
338 # continuation bytes can be in, and what the lowest start byte can be.  So we
339 # cycle through them.
340
341 my $first_continuation = (isASCII) ? 0x80 : 0xA0;
342 my $final_continuation = 0xBF;
343 my $start = (isASCII) ? 0xC2 : 0xC5;
344
345 my $continuation = $first_continuation - 1;
346
347 while ($cp < 255) {
348     if (++$continuation > $final_continuation) {
349
350         # Wrap to the next start byte when we reach the final continuation
351         # byte possible
352         $continuation = $first_continuation;
353         $start++;
354     }
355     $code_points{$cp} = I8_to_native(chr($start) . chr($continuation));
356
357     $cp++;
358 }
359
360 my @warnings;
361
362 use warnings 'utf8';
363 local $SIG{__WARN__} = sub { push @warnings, @_ };
364
365 my %restriction_types;
366
367 $restriction_types{""}{'valid_strings'} = "";
368 $restriction_types{"c9strict"}{'valid_strings'} = "";
369 $restriction_types{"strict"}{'valid_strings'} = "";
370 $restriction_types{"fits_in_31_bits"}{'valid_strings'} = "";
371
372 # This set of tests looks for basic sanity, and lastly tests various routines
373 # for the given code point.  If the earlier tests for that code point fail,
374 # the later ones probably will too.  Malformations are tested in later
375 # segments of code.
376 for my $u (sort { utf8::unicode_to_native($a) <=> utf8::unicode_to_native($b) }
377           keys %code_points)
378 {
379     my $hex_u = sprintf("0x%02X", $u);
380     my $n = utf8::unicode_to_native($u);
381     my $hex_n = sprintf("0x%02X", $n);
382     my $bytes = $code_points{$u};
383
384     my $offskip_should_be;
385     {
386         no warnings qw(overflow portable);
387         $offskip_should_be = (isASCII)
388             ? ( $u < 0x80           ? 1 :
389                 $u < 0x800          ? 2 :
390                 $u < 0x10000        ? 3 :
391                 $u < 0x200000       ? 4 :
392                 $u < 0x4000000      ? 5 :
393                 $u < 0x80000000     ? 6 : (($is64bit)
394                                         ? ($u < 0x1000000000 ? 7 : 13)
395                                         : 7)
396               )
397             : ($u < 0xA0        ? 1 :
398                $u < 0x400       ? 2 :
399                $u < 0x4000      ? 3 :
400                $u < 0x40000     ? 4 :
401                $u < 0x400000    ? 5 :
402                $u < 0x4000000   ? 6 :
403                $u < 0x40000000  ? 7 : 14 );
404     }
405
406     # If this test fails, subsequent ones are meaningless.
407     next unless is(test_OFFUNISKIP($u), $offskip_should_be,
408                    "Verify OFFUNISKIP($hex_u) is $offskip_should_be");
409     my $invariant = $offskip_should_be == 1;
410     my $display_invariant = $invariant || 0;
411     is(test_OFFUNI_IS_INVARIANT($u), $invariant,
412        "Verify OFFUNI_IS_INVARIANT($hex_u) is $display_invariant");
413
414     my $uvchr_skip_should_be = $offskip_should_be;
415     next unless is(test_UVCHR_SKIP($n), $uvchr_skip_should_be,
416                    "Verify UVCHR_SKIP($hex_n) is $uvchr_skip_should_be");
417     is(test_UVCHR_IS_INVARIANT($n), $offskip_should_be == 1,
418        "Verify UVCHR_IS_INVARIANT($hex_n) is $display_invariant");
419
420     my $n_chr = chr $n;
421     utf8::upgrade $n_chr;
422
423     is(test_UTF8_SKIP($n_chr), $uvchr_skip_should_be,
424         "Verify UTF8_SKIP(chr $hex_n) is $uvchr_skip_should_be");
425
426     use bytes;
427     my $byte_length = length $n_chr;
428     for (my $j = 0; $j < $byte_length; $j++) {
429         undef @warnings;
430
431         if ($j == $byte_length - 1) {
432             my $ret = test_is_utf8_valid_partial_char_flags($n_chr, $byte_length, 0);
433             is($ret, 0, "   Verify is_utf8_valid_partial_char_flags(" . display_bytes($n_chr) . ") returns 0 for full character");
434         }
435         else {
436             my $bytes_so_far = substr($n_chr, 0, $j + 1);
437             my $ret = test_is_utf8_valid_partial_char_flags($bytes_so_far, $j + 1, 0);
438             is($ret, 1, "   Verify is_utf8_valid_partial_char_flags(" . display_bytes($bytes_so_far) . ") returns 1");
439         }
440
441         unless (is(scalar @warnings, 0,
442                 "   Verify is_utf8_valid_partial_char_flags generated no warnings"))
443         {
444             diag "The warnings were: " . join(", ", @warnings);
445         }
446
447         my $b = substr($n_chr, $j, 1);
448         my $hex_b = sprintf("\"\\x%02x\"", ord $b);
449
450         my $byte_invariant = $j == 0 && $uvchr_skip_should_be == 1;
451         my $display_byte_invariant = $byte_invariant || 0;
452         next unless is(test_UTF8_IS_INVARIANT($b), $byte_invariant,
453                        "   Verify UTF8_IS_INVARIANT($hex_b) for byte $j "
454                      . "is $display_byte_invariant");
455
456         my $is_start = $j == 0 && $uvchr_skip_should_be > 1;
457         my $display_is_start = $is_start || 0;
458         next unless is(test_UTF8_IS_START($b), $is_start,
459                     "      Verify UTF8_IS_START($hex_b) is $display_is_start");
460
461         my $is_continuation = $j != 0 && $uvchr_skip_should_be > 1;
462         my $display_is_continuation = $is_continuation || 0;
463         next unless is(test_UTF8_IS_CONTINUATION($b), $is_continuation,
464                        "      Verify UTF8_IS_CONTINUATION($hex_b) is "
465                      . "$display_is_continuation");
466
467         my $is_continued = $uvchr_skip_should_be > 1;
468         my $display_is_continued = $is_continued || 0;
469         next unless is(test_UTF8_IS_CONTINUED($b), $is_continued,
470                        "      Verify UTF8_IS_CONTINUED($hex_b) is "
471                      . "$display_is_continued");
472
473         my $is_downgradeable_start =    $n < 256
474                                      && $uvchr_skip_should_be > 1
475                                      && $j == 0;
476         my $display_is_downgradeable_start = $is_downgradeable_start || 0;
477         next unless is(test_UTF8_IS_DOWNGRADEABLE_START($b),
478                        $is_downgradeable_start,
479                        "      Verify UTF8_IS_DOWNGRADEABLE_START($hex_b) is "
480                      . "$display_is_downgradeable_start");
481
482         my $is_above_latin1 =  $n > 255 && $j == 0;
483         my $display_is_above_latin1 = $is_above_latin1 || 0;
484         next unless is(test_UTF8_IS_ABOVE_LATIN1($b),
485                        $is_above_latin1,
486                        "      Verify UTF8_IS_ABOVE_LATIN1($hex_b) is "
487                      . "$display_is_above_latin1");
488
489         my $is_possibly_problematic =  $j == 0
490                                     && $n >= ((isASCII)
491                                               ? 0xD000
492                                               : 0x8000);
493         my $display_is_possibly_problematic = $is_possibly_problematic || 0;
494         next unless is(test_isUTF8_POSSIBLY_PROBLEMATIC($b),
495                        $is_possibly_problematic,
496                        "      Verify isUTF8_POSSIBLY_PROBLEMATIC($hex_b) is "
497                      . "$display_is_above_latin1");
498     }
499
500     # We are not trying to look for warnings, etc, so if they should occur, it
501     # is an error.  But some of the code points here do cause warnings, so we
502     # check here and turn off the ones that apply to such code points.  A
503     # later section of the code tests for these kinds of things.
504     my $this_utf8_flags = $look_for_everything_utf8n_to;
505     my $len = length $bytes;
506
507     my $valid_under_strict = 1;
508     my $valid_under_c9strict = 1;
509     my $valid_for_fits_in_31_bits = 1;
510     if ($n > 0x10FFFF) {
511         $this_utf8_flags &= ~($UTF8_DISALLOW_SUPER|$UTF8_WARN_SUPER);
512         $valid_under_strict = 0;
513         $valid_under_c9strict = 0;
514         if ($n > 2 ** 31 - 1) {
515             $this_utf8_flags &=
516                             ~($UTF8_DISALLOW_ABOVE_31_BIT|$UTF8_WARN_ABOVE_31_BIT);
517             $valid_for_fits_in_31_bits = 0;
518         }
519     }
520     elsif (($n >= 0xFDD0 && $n <= 0xFDEF) || ($n & 0xFFFE) == 0xFFFE) {
521         $this_utf8_flags &= ~($UTF8_DISALLOW_NONCHAR|$UTF8_WARN_NONCHAR);
522         $valid_under_strict = 0;
523     }
524     elsif ($n >= 0xD800 && $n <= 0xDFFF) {
525         $this_utf8_flags &= ~($UTF8_DISALLOW_SURROGATE|$UTF8_WARN_SURROGATE);
526         $valid_under_c9strict = 0;
527         $valid_under_strict = 0;
528     }
529
530     undef @warnings;
531
532     my $display_flags = sprintf "0x%x", $this_utf8_flags;
533     my $display_bytes = display_bytes($bytes);
534     my $ret_ref = test_utf8n_to_uvchr($bytes, $len, $this_utf8_flags);
535
536     # Rest of tests likely meaningless if it gets the wrong code point.
537     next unless is($ret_ref->[0], $n,
538        "Verify utf8n_to_uvchr($display_bytes, $display_flags) returns $hex_n");
539     is($ret_ref->[1], $len,
540        "Verify utf8n_to_uvchr() for $hex_n returns expected length: $len");
541
542     unless (is(scalar @warnings, 0,
543                "Verify utf8n_to_uvchr() for $hex_n generated no warnings"))
544     {
545         diag "The warnings were: " . join(", ", @warnings);
546     }
547
548     undef @warnings;
549
550     my $ret = test_isUTF8_CHAR($bytes, $len);
551     is($ret, $len, "Verify isUTF8_CHAR($display_bytes) returns expected length: $len");
552
553     unless (is(scalar @warnings, 0,
554                "Verify isUTF8_CHAR() for $hex_n generated no warnings"))
555     {
556         diag "The warnings were: " . join(", ", @warnings);
557     }
558
559     undef @warnings;
560
561     $ret = test_isUTF8_CHAR($bytes, $len - 1);
562     is($ret, 0, "Verify isUTF8_CHAR() with too short length parameter returns 0");
563
564     unless (is(scalar @warnings, 0,
565                "Verify isUTF8_CHAR() generated no warnings"))
566     {
567         diag "The warnings were: " . join(", ", @warnings);
568     }
569
570     undef @warnings;
571
572     $ret = test_isUTF8_CHAR_flags($bytes, $len, 0);
573     is($ret, $len, "Verify isUTF8_CHAR_flags($display_bytes, 0) returns expected length: $len");
574
575     unless (is(scalar @warnings, 0,
576                "Verify isUTF8_CHAR_flags() for $hex_n generated no warnings"))
577     {
578         diag "The warnings were: " . join(", ", @warnings);
579     }
580
581     undef @warnings;
582
583     $ret = test_isUTF8_CHAR_flags($bytes, $len - 1, 0);
584     is($ret, 0, "Verify isUTF8_CHAR_flags() with too short length parameter returns 0");
585
586     unless (is(scalar @warnings, 0,
587                "Verify isUTF8_CHAR_flags() generated no warnings"))
588     {
589         diag "The warnings were: " . join(", ", @warnings);
590     }
591
592     undef @warnings;
593
594     $ret = test_isSTRICT_UTF8_CHAR($bytes, $len);
595     my $expected_len = ($valid_under_strict) ? $len : 0;
596     is($ret, $expected_len, "Verify isSTRICT_UTF8_CHAR($display_bytes) returns expected length: $expected_len");
597
598     unless (is(scalar @warnings, 0,
599                "Verify isSTRICT_UTF8_CHAR() for $hex_n generated no warnings"))
600     {
601         diag "The warnings were: " . join(", ", @warnings);
602     }
603
604     undef @warnings;
605
606     $ret = test_isSTRICT_UTF8_CHAR($bytes, $len - 1);
607     is($ret, 0, "Verify isSTRICT_UTF8_CHAR() with too short length parameter returns 0");
608
609     unless (is(scalar @warnings, 0,
610                "Verify isSTRICT_UTF8_CHAR() generated no warnings"))
611     {
612         diag "The warnings were: " . join(", ", @warnings);
613     }
614
615     undef @warnings;
616
617     $ret = test_isUTF8_CHAR_flags($bytes, $len, $UTF8_DISALLOW_ILLEGAL_INTERCHANGE);
618     is($ret, $expected_len, "Verify isUTF8_CHAR_flags('DISALLOW_ILLEGAL_INTERCHANGE') acts like isSTRICT_UTF8_CHAR");
619
620     unless (is(scalar @warnings, 0,
621                "Verify isUTF8_CHAR() for $hex_n generated no warnings"))
622     {
623         diag "The warnings were: " . join(", ", @warnings);
624     }
625
626     undef @warnings;
627
628     $ret = test_isC9_STRICT_UTF8_CHAR($bytes, $len);
629     $expected_len = ($valid_under_c9strict) ? $len : 0;
630     is($ret, $expected_len, "Verify isC9_STRICT_UTF8_CHAR($display_bytes) returns expected length: $len");
631
632     unless (is(scalar @warnings, 0,
633                "Verify isC9_STRICT_UTF8_CHAR() for $hex_n generated no warnings"))
634     {
635         diag "The warnings were: " . join(", ", @warnings);
636     }
637
638     undef @warnings;
639
640     $ret = test_isC9_STRICT_UTF8_CHAR($bytes, $len - 1);
641     is($ret, 0, "Verify isC9_STRICT_UTF8_CHAR() with too short length parameter returns 0");
642
643     unless (is(scalar @warnings, 0,
644                "Verify isC9_STRICT_UTF8_CHAR() generated no warnings"))
645     {
646         diag "The warnings were: " . join(", ", @warnings);
647     }
648
649     undef @warnings;
650
651     $ret = test_isUTF8_CHAR_flags($bytes, $len, $UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE);
652     is($ret, $expected_len, "Verify isUTF8_CHAR_flags('DISALLOW_ILLEGAL_C9_INTERCHANGE') acts like isC9_STRICT_UTF8_CHAR");
653
654     unless (is(scalar @warnings, 0,
655                "Verify isUTF8_CHAR() for $hex_n generated no warnings"))
656     {
657         diag "The warnings were: " . join(", ", @warnings);
658     }
659
660     undef @warnings;
661
662     $ret_ref = test_valid_utf8_to_uvchr($bytes);
663     is($ret_ref->[0], $n, "Verify valid_utf8_to_uvchr($display_bytes) returns $hex_n");
664     is($ret_ref->[1], $len, "Verify valid_utf8_to_uvchr() for $hex_n returns expected length: $len");
665
666     unless (is(scalar @warnings, 0,
667                "Verify valid_utf8_to_uvchr() for $hex_n generated no warnings"))
668     {
669         diag "The warnings were: " . join(", ", @warnings);
670     }
671
672     # Similarly for uvchr_to_utf8
673     my $this_uvchr_flags = $look_for_everything_uvchr_to;
674     if ($n > 2 ** 31 - 1) {
675         $this_uvchr_flags &=
676                 ~($UNICODE_DISALLOW_ABOVE_31_BIT|$UNICODE_WARN_ABOVE_31_BIT);
677     }
678     if ($n > 0x10FFFF) {
679         $this_uvchr_flags &= ~($UNICODE_DISALLOW_SUPER|$UNICODE_WARN_SUPER);
680     }
681     elsif (($n >= 0xFDD0 && $n <= 0xFDEF) || ($n & 0xFFFE) == 0xFFFE) {
682         $this_uvchr_flags &= ~($UNICODE_DISALLOW_NONCHAR|$UNICODE_WARN_NONCHAR);
683     }
684     elsif ($n >= 0xD800 && $n <= 0xDFFF) {
685         $this_uvchr_flags &= ~($UNICODE_DISALLOW_SURROGATE|$UNICODE_WARN_SURROGATE);
686     }
687     $display_flags = sprintf "0x%x", $this_uvchr_flags;
688
689     undef @warnings;
690
691     $ret = test_uvchr_to_utf8_flags($n, $this_uvchr_flags);
692     ok(defined $ret, "Verify uvchr_to_utf8_flags($hex_n, $display_flags) returned success");
693     is($ret, $bytes, "Verify uvchr_to_utf8_flags($hex_n, $display_flags) returns correct bytes");
694
695     unless (is(scalar @warnings, 0,
696         "Verify uvchr_to_utf8_flags($hex_n, $display_flags) for $hex_n generated no warnings"))
697     {
698         diag "The warnings were: " . join(", ", @warnings);
699     }
700
701     # Now append this code point to a string that we will test various
702     # versions of is_foo_utf8_string_bar on, and keep a count of how many code
703     # points are in it.  All the code points in this loop are valid in Perl's
704     # extended UTF-8, but some are not valid under various restrictions.  A
705     # string and count is kept separately that is entirely valid for each
706     # restriction.  And, for each restriction, we note the first occurrence in
707     # the unrestricted string where we find something not in the restricted
708     # string.
709     $restriction_types{""}{'valid_strings'} .= $bytes;
710     $restriction_types{""}{'valid_counts'}++;
711
712     if ($valid_under_c9strict) {
713         $restriction_types{"c9strict"}{'valid_strings'} .= $bytes;
714         $restriction_types{"c9strict"}{'valid_counts'}++;
715     }
716     elsif (! exists $restriction_types{"c9strict"}{'first_invalid_offset'}) {
717         $restriction_types{"c9strict"}{'first_invalid_offset'}
718                     = length $restriction_types{"c9strict"}{'valid_strings'};
719         $restriction_types{"c9strict"}{'first_invalid_count'}
720                             = $restriction_types{"c9strict"}{'valid_counts'};
721     }
722
723     if ($valid_under_strict) {
724         $restriction_types{"strict"}{'valid_strings'} .= $bytes;
725         $restriction_types{"strict"}{'valid_counts'}++;
726     }
727     elsif (! exists $restriction_types{"strict"}{'first_invalid_offset'}) {
728         $restriction_types{"strict"}{'first_invalid_offset'}
729                         = length $restriction_types{"strict"}{'valid_strings'};
730         $restriction_types{"strict"}{'first_invalid_count'}
731                                 = $restriction_types{"strict"}{'valid_counts'};
732     }
733
734     if ($valid_for_fits_in_31_bits) {
735         $restriction_types{"fits_in_31_bits"}{'valid_strings'} .= $bytes;
736         $restriction_types{"fits_in_31_bits"}{'valid_counts'}++;
737     }
738     elsif (! exists
739                 $restriction_types{"fits_in_31_bits"}{'first_invalid_offset'})
740     {
741         $restriction_types{"fits_in_31_bits"}{'first_invalid_offset'}
742                 = length $restriction_types{"fits_in_31_bits"}{'valid_strings'};
743         $restriction_types{"fits_in_31_bits"}{'first_invalid_count'}
744                         = $restriction_types{"fits_in_31_bits"}{'valid_counts'};
745     }
746 }
747
748 my $I8c = (isASCII) ? "\x80" : "\xa0";    # A continuation byte
749 my $cont_byte = I8_to_native($I8c);
750 my $p = (isASCII) ? "\xe1\x80" : I8_to_native("\xE4\xA0");  # partial
751
752 # The loop above tested the single or partial character functions/macros,
753 # while building up strings to test the string functions, which we do now.
754
755 for my $restriction (sort keys %restriction_types) {
756     use bytes;
757
758     for my $use_flags ("", "_flags") {
759
760         # For each restriction, we test it in both the is_foo_flags functions
761         # and the specially named foo function.  But not if there isn't such a
762         # specially named function.  Currently, this is the only tested
763         # restriction that doesn't have a specially named function
764         next if $use_flags eq "" && $restriction eq "fits_in_31_bits";
765
766         # Start building up the name of the function we will test.
767         my $base_name = "is_";
768
769         if (! $use_flags  && $restriction ne "") {
770             $base_name .= $restriction . "_";
771         }
772
773         # We test both "is_utf8_string_foo" and "is_fixed_width_buf" functions
774         foreach my $operand ('string', 'fixed_width_buf') {
775
776             # Currently, the only fixed_width_buf functions have the '_flags'
777             # suffix.
778             next if $operand eq 'fixed_width_buf' && $use_flags eq "";
779
780             my $name = "${base_name}utf8_$operand";
781
782             # We test each version of the function
783             for my $function ("_loclen", "_loc", "") {
784
785                 # We test each function against
786                 #   a) valid input
787                 #   b) invalid input created by appending an out-of-place
788                 #      continuation character to the valid string
789                 #   c) input created by appending a partial character.  This
790                 #      is valid in the 'fixed_width' functions, but invalid in
791                 #   the 'string' ones
792                 #   d) invalid input created by calling a function that is
793                 #      expecting a restricted form of the input using the string
794                 #      that's valid when unrestricted
795                 for my $error_type (0, $cont_byte, $p, $restriction) {
796                     #diag "restriction=$restriction, use_flags=$use_flags, function=$function, error_type=" . display_bytes($error_type);
797
798                     # If there is no restriction, the error type will be "",
799                     # which is redundant with 0.
800                     next if $error_type eq "";
801
802                     my $this_name = "$name$function$use_flags";
803                     my $bytes
804                             = $restriction_types{$restriction}{'valid_strings'};
805                     my $expected_offset = length $bytes;
806                     my $expected_count
807                             = $restriction_types{$restriction}{'valid_counts'};
808                     my $test_name_suffix = "";
809
810                     my $this_error_type = $error_type;
811                     if ($this_error_type) {
812
813                         # Appending a bare continuation byte or a partial
814                         # character doesn't change the character count or
815                         # offset.  But in the other cases, we have saved where
816                         # the failures should occur, so use those.  Appending
817                         # a continuation byte makes it invalid; appending a
818                         # partial character makes the 'string' form invalid,
819                         # but not the 'fixed_width_buf' form.
820                         if ($this_error_type eq $cont_byte || $this_error_type eq $p) {
821                             $bytes .= $this_error_type;
822                             if ($this_error_type eq $cont_byte) {
823                                 $test_name_suffix
824                                             = " for an unexpected continuation";
825                             }
826                             else {
827                                 $test_name_suffix
828                                         = " if ends with a partial character";
829                                 $this_error_type
830                                         = 0 if $operand eq "fixed_width_buf";
831                             }
832                         }
833                         else {
834                             $test_name_suffix
835                                         = " if contains forbidden code points";
836                             if ($this_error_type eq "c9strict") {
837                                 $bytes = $restriction_types{""}{'valid_strings'};
838                                 $expected_offset
839                                  = $restriction_types{"c9strict"}
840                                                      {'first_invalid_offset'};
841                                 $expected_count
842                                   = $restriction_types{"c9strict"}
843                                                       {'first_invalid_count'};
844                             }
845                             elsif ($this_error_type eq "strict") {
846                                 $bytes = $restriction_types{""}{'valid_strings'};
847                                 $expected_offset
848                                   = $restriction_types{"strict"}
849                                                       {'first_invalid_offset'};
850                                 $expected_count
851                                   = $restriction_types{"strict"}
852                                                       {'first_invalid_count'};
853
854                             }
855                             elsif ($this_error_type eq "fits_in_31_bits") {
856                                 $bytes = $restriction_types{""}{'valid_strings'};
857                                 $expected_offset
858                                   = $restriction_types{"fits_in_31_bits"}
859                                                       {'first_invalid_offset'};
860                                 $expected_count
861                                     = $restriction_types{"fits_in_31_bits"}
862                                                         {'first_invalid_count'};
863                             }
864                             else {
865                                 fail("Internal test error: Unknown error type "
866                                 . "'$this_error_type'");
867                                 next;
868                             }
869                         }
870                     }
871
872                     my $length = length $bytes;
873                     my $ret_ref;
874
875                     my $test = "\$ret_ref = test_$this_name(\$bytes, $length";
876
877                     # If using the _flags functions, we have to figure out what
878                     # flags to pass.  This is done to match the restriction.
879                     if ($use_flags eq "_flags") {
880                         if (! $restriction) {
881                             $test .= ", 0";     # The flag
882
883                             # Indicate the kind of flag in the test name.
884                             $this_name .= "(0)";
885                         }
886                         else {
887                             $this_name .= "($restriction)";
888                             if ($restriction eq "c9strict") {
889                                 $test
890                                   .= ", $UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE";
891                             }
892                             elsif ($restriction eq "strict") {
893                                 $test .= ", $UTF8_DISALLOW_ILLEGAL_INTERCHANGE";
894                             }
895                             elsif ($restriction eq "fits_in_31_bits") {
896                                 $test .= ", $UTF8_DISALLOW_ABOVE_31_BIT";
897                             }
898                             else {
899                                 fail("Internal test error: Unknown restriction "
900                                 . "'$restriction'");
901                                 next;
902                             }
903                         }
904                     }
905                     $test .= ")";
906
907                     # Actually run the test
908                     eval $test;
909                     if ($@) {
910                         fail($test);
911                         diag $@;
912                         next;
913                     }
914
915                     my $ret;
916                     my $error_offset;
917                     my $cp_count;
918
919                     if ($function eq "") {
920                         $ret = $ret_ref;    # For plain function, there's only a
921                                             # single return value
922                     }
923                     else {  # Otherwise, the multiple values come in an array.
924                         $ret = shift @$ret_ref ;
925                         $error_offset = shift @$ret_ref;
926                         $cp_count = shift@$ret_ref if $function eq "_loclen";
927                     }
928
929                     if ($this_error_type) {
930                         is($ret, 0,
931                            "Verify $this_name is FALSE$test_name_suffix");
932                     }
933                     else {
934                         unless(is($ret, 1,
935                                   "Verify $this_name is TRUE for valid input"
936                                 . "$test_name_suffix"))
937                         {
938                             diag("The bytes starting at offset"
939                                . " $error_offset are"
940                                . display_bytes(substr(
941                                           $restriction_types{$restriction}
942                                                             {'valid_strings'},
943                                           $error_offset)));
944                             next;
945                         }
946                     }
947
948                     if ($function ne "") {
949                         unless (is($error_offset, $expected_offset,
950                                    "\tAnd returns the correct offset"))
951                         {
952                             my $min = ($error_offset < $expected_offset)
953                                     ? $error_offset
954                                     : $expected_offset;
955                             diag display_bytes(substr($bytes, $min));
956                         }
957
958                         if ($function eq '_loclen') {
959                             is($cp_count, $expected_count,
960                                "\tAnd returns the correct character count");
961                         }
962                     }
963                 }
964             }
965         }
966     }
967 }
968
969 my $REPLACEMENT = 0xFFFD;
970
971 # Now test the malformations.  All these raise category utf8 warnings.
972 my @malformations = (
973     [ "zero length string malformation", "", 0,
974         $UTF8_ALLOW_EMPTY, 0, 0,
975         qr/empty string/
976     ],
977     [ "orphan continuation byte malformation", I8_to_native("${I8c}a"),
978         2,
979         $UTF8_ALLOW_CONTINUATION, $REPLACEMENT, 1,
980         qr/unexpected continuation byte/
981     ],
982     [ "premature next character malformation (immediate)",
983         (isASCII) ? "\xc2\xc2\x80" : I8_to_native("\xc5\xc5\xa0"),
984         3,
985         $UTF8_ALLOW_NON_CONTINUATION, $REPLACEMENT, 1,
986         qr/unexpected non-continuation byte.*immediately after start byte/
987     ],
988     [ "premature next character malformation (non-immediate)",
989         I8_to_native("\xf0${I8c}a"),
990         3,
991         $UTF8_ALLOW_NON_CONTINUATION, $REPLACEMENT, 2,
992         qr/unexpected non-continuation byte .* 2 bytes after start byte/
993     ],
994     [ "too short malformation", I8_to_native("\xf0${I8c}a"), 2,
995         # Having the 'a' after this, but saying there are only 2 bytes also
996         # tests that we pay attention to the passed in length
997         $UTF8_ALLOW_SHORT, $REPLACEMENT, 2,
998         qr/2 bytes, need 4/
999     ],
1000     [ "overlong malformation, lowest 2-byte",
1001         (isASCII) ? "\xc0\x80" : I8_to_native("\xc0\xa0"),
1002         2,
1003         $UTF8_ALLOW_LONG,
1004         0,   # NUL
1005         2,
1006         qr/2 bytes, need 1/
1007     ],
1008     [ "overlong malformation, highest 2-byte",
1009         (isASCII) ? "\xc1\xbf" : I8_to_native("\xc4\xbf"),
1010         2,
1011         $UTF8_ALLOW_LONG,
1012         (isASCII) ? 0x7F : utf8::unicode_to_native(0xBF),
1013         2,
1014         qr/2 bytes, need 1/
1015     ],
1016     [ "overlong malformation, lowest 3-byte",
1017         (isASCII) ? "\xe0\x80\x80" : I8_to_native("\xe0\xa0\xa0"),
1018         3,
1019         $UTF8_ALLOW_LONG,
1020         0,   # NUL
1021         3,
1022         qr/3 bytes, need 1/
1023     ],
1024     [ "overlong malformation, highest 3-byte",
1025         (isASCII) ? "\xe0\x9f\xbf" : I8_to_native("\xe0\xbf\xbf"),
1026         3,
1027         $UTF8_ALLOW_LONG,
1028         (isASCII) ? 0x7FF : 0x3FF,
1029         3,
1030         qr/3 bytes, need 2/
1031     ],
1032     [ "overlong malformation, lowest 4-byte",
1033         (isASCII) ? "\xf0\x80\x80\x80" : I8_to_native("\xf0\xa0\xa0\xa0"),
1034         4,
1035         $UTF8_ALLOW_LONG,
1036         0,   # NUL
1037         4,
1038         qr/4 bytes, need 1/
1039     ],
1040     [ "overlong malformation, highest 4-byte",
1041         (isASCII) ? "\xf0\x8F\xbf\xbf" : I8_to_native("\xf0\xaf\xbf\xbf"),
1042         4,
1043         $UTF8_ALLOW_LONG,
1044         (isASCII) ? 0xFFFF : 0x3FFF,
1045         4,
1046         qr/4 bytes, need 3/
1047     ],
1048     [ "overlong malformation, lowest 5-byte",
1049         (isASCII)
1050          ?              "\xf8\x80\x80\x80\x80"
1051          : I8_to_native("\xf8\xa0\xa0\xa0\xa0"),
1052         5,
1053         $UTF8_ALLOW_LONG,
1054         0,   # NUL
1055         5,
1056         qr/5 bytes, need 1/
1057     ],
1058     [ "overlong malformation, highest 5-byte",
1059         (isASCII)
1060          ?              "\xf8\x87\xbf\xbf\xbf"
1061          : I8_to_native("\xf8\xa7\xbf\xbf\xbf"),
1062         5,
1063         $UTF8_ALLOW_LONG,
1064         (isASCII) ? 0x1FFFFF : 0x3FFFF,
1065         5,
1066         qr/5 bytes, need 4/
1067     ],
1068     [ "overlong malformation, lowest 6-byte",
1069         (isASCII)
1070          ?              "\xfc\x80\x80\x80\x80\x80"
1071          : I8_to_native("\xfc\xa0\xa0\xa0\xa0\xa0"),
1072         6,
1073         $UTF8_ALLOW_LONG,
1074         0,   # NUL
1075         6,
1076         qr/6 bytes, need 1/
1077     ],
1078     [ "overlong malformation, highest 6-byte",
1079         (isASCII)
1080          ?              "\xfc\x83\xbf\xbf\xbf\xbf"
1081          : I8_to_native("\xfc\xa3\xbf\xbf\xbf\xbf"),
1082         6,
1083         $UTF8_ALLOW_LONG,
1084         (isASCII) ? 0x3FFFFFF : 0x3FFFFF,
1085         6,
1086         qr/6 bytes, need 5/
1087     ],
1088     [ "overlong malformation, lowest 7-byte",
1089         (isASCII)
1090          ?              "\xfe\x80\x80\x80\x80\x80\x80"
1091          : I8_to_native("\xfe\xa0\xa0\xa0\xa0\xa0\xa0"),
1092         7,
1093         $UTF8_ALLOW_LONG,
1094         0,   # NUL
1095         7,
1096         qr/7 bytes, need 1/
1097     ],
1098     [ "overlong malformation, highest 7-byte",
1099         (isASCII)
1100          ?              "\xfe\x81\xbf\xbf\xbf\xbf\xbf"
1101          : I8_to_native("\xfe\xa1\xbf\xbf\xbf\xbf\xbf"),
1102         7,
1103         $UTF8_ALLOW_LONG,
1104         (isASCII) ? 0x7FFFFFFF : 0x3FFFFFF,
1105         7,
1106         qr/7 bytes, need 6/
1107     ],
1108 );
1109
1110 if (isASCII && ! $is64bit) {    # 32-bit ASCII platform
1111     no warnings 'portable';
1112     push @malformations,
1113         [ "overflow malformation",
1114             "\xfe\x84\x80\x80\x80\x80\x80",  # Represents 2**32
1115             7,
1116             0,  # There is no way to allow this malformation
1117             $REPLACEMENT,
1118             7,
1119             qr/overflow/
1120         ],
1121         [ "overflow malformation, can tell on first byte",
1122             "\xff\x80\x80\x80\x80\x80\x81\x80\x80\x80\x80\x80\x80",
1123             13,
1124             0,  # There is no way to allow this malformation
1125             $REPLACEMENT,
1126             13,
1127             qr/overflow/
1128         ];
1129 }
1130 else {
1131     # On EBCDIC platforms, another overlong test is needed even on 32-bit
1132     # systems, whereas it doesn't happen on ASCII except on 64-bit ones.
1133
1134     no warnings 'portable';
1135     no warnings 'overflow'; # Doesn't run on 32-bit systems, but compiles
1136     push @malformations,
1137         [ "overlong malformation, lowest max-byte",
1138             (isASCII)
1139              ?              "\xff\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80"
1140              : I8_to_native("\xff\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0"),
1141             (isASCII) ? 13 : 14,
1142             $UTF8_ALLOW_LONG,
1143             0,   # NUL
1144             (isASCII) ? 13 : 14,
1145             qr/1[34] bytes, need 1/,    # 1[34] to work on either ASCII or EBCDIC
1146         ],
1147         [ "overlong malformation, highest max-byte",
1148             (isASCII)    # 2**36-1 on ASCII; 2**30-1 on EBCDIC
1149              ?              "\xff\x80\x80\x80\x80\x80\x80\xbf\xbf\xbf\xbf\xbf\xbf"
1150              : I8_to_native("\xff\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xbf\xbf\xbf\xbf\xbf\xbf"),
1151             (isASCII) ? 13 : 14,
1152             $UTF8_ALLOW_LONG,
1153             (isASCII) ? 0xFFFFFFFFF : 0x3FFFFFFF,
1154             (isASCII) ? 13 : 14,
1155             qr/1[34] bytes, need 7/,
1156         ];
1157
1158     if (! $is64bit) {   # 32-bit EBCDIC
1159         push @malformations,
1160         [ "overflow malformation",
1161             I8_to_native("\xff\xa0\xa0\xa0\xa0\xa0\xa0\xa4\xa0\xa0\xa0\xa0\xa0\xa0"),
1162             14,
1163             0,  # There is no way to allow this malformation
1164             $REPLACEMENT,
1165             14,
1166             qr/overflow/
1167         ];
1168     }
1169     else {  # 64-bit
1170         push @malformations,
1171             [ "overflow malformation",
1172                (isASCII)
1173                 ?              "\xff\x80\x90\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0"
1174                 : I8_to_native("\xff\xb0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0"),
1175                 (isASCII) ? 13 : 14,
1176                 0,  # There is no way to allow this malformation
1177                 $REPLACEMENT,
1178                 (isASCII) ? 13 : 14,
1179                 qr/overflow/
1180             ];
1181     }
1182 }
1183
1184 foreach my $test (@malformations) {
1185     my ($testname, $bytes, $length, $allow_flags, $allowed_uv, $expected_len, $message ) = @$test;
1186
1187     next if ! ok(length($bytes) >= $length, "$testname: Make sure won't read beyond buffer: " . length($bytes) . " >= $length");
1188
1189     undef @warnings;
1190
1191     my $ret = test_isUTF8_CHAR($bytes, $length);
1192     is($ret, 0, "$testname: isUTF8_CHAR returns 0");
1193     unless (is(scalar @warnings, 0,
1194                "$testname: isUTF8_CHAR() generated no warnings"))
1195     {
1196         diag "The warnings were: " . join(", ", @warnings);
1197     }
1198
1199     undef @warnings;
1200
1201     $ret = test_isUTF8_CHAR_flags($bytes, $length, 0);
1202     is($ret, 0, "$testname: isUTF8_CHAR_flags returns 0");
1203     unless (is(scalar @warnings, 0,
1204                "$testname: isUTF8_CHAR() generated no warnings"))
1205     {
1206         diag "The warnings were: " . join(", ", @warnings);
1207     }
1208
1209     $ret = test_isSTRICT_UTF8_CHAR($bytes, $length);
1210     is($ret, 0, "$testname: isSTRICT_UTF8_CHAR returns 0");
1211     unless (is(scalar @warnings, 0,
1212                "$testname: isSTRICT_UTF8_CHAR() generated no warnings"))
1213     {
1214         diag "The warnings were: " . join(", ", @warnings);
1215     }
1216
1217     $ret = test_isC9_STRICT_UTF8_CHAR($bytes, $length);
1218     is($ret, 0, "$testname: isC9_STRICT_UTF8_CHAR returns 0");
1219     unless (is(scalar @warnings, 0,
1220                "$testname: isC9_STRICT_UTF8_CHAR() generated no warnings"))
1221     {
1222         diag "The warnings were: " . join(", ", @warnings);
1223     }
1224
1225     for my $j (1 .. $length - 1) {
1226         my $partial = substr($bytes, 0, $j);
1227
1228         undef @warnings;
1229
1230         $ret = test_is_utf8_valid_partial_char_flags($bytes, $j, 0);
1231         my $ret_should_be = 0;
1232         my $comment = "";
1233         if ($testname =~ /premature|short/ && $j < 2) {
1234             $ret_should_be = 1;
1235             $comment = ", but need 2 bytes to discern:";
1236         }
1237         elsif ($testname =~ /overlong/ && $length > 2) {
1238             if ($length <= 7 && $j < 2) {
1239                 $ret_should_be = 1;
1240                 $comment = ", but need 2 bytes to discern:";
1241             }
1242             elsif ($length > 7 && $j < 7) {
1243                 $ret_should_be = 1;
1244                 $comment = ", but need 7 bytes to discern:";
1245             }
1246         }
1247         elsif ($testname =~ /overflow/ && $testname !~ /first byte/) {
1248             if (isASCII) {
1249                 if ($j < (($is64bit) ? 3 : 2)) {
1250                     $comment = ", but need $j bytes to discern:";
1251                     $ret_should_be = 1;
1252                 }
1253             }
1254             else {
1255                 if ($j < (($is64bit) ? 2 : 8)) {
1256                     $comment = ", but need $j bytes to discern:";
1257                     $ret_should_be = 1;
1258                 }
1259             }
1260         }
1261         is($ret, $ret_should_be, "$testname: is_utf8_valid_partial_char_flags("
1262                                 . display_bytes($partial)
1263                                 . ")$comment returns $ret_should_be");
1264         unless (is(scalar @warnings, 0,
1265                 "$testname: is_utf8_valid_partial_char_flags() generated no warnings"))
1266         {
1267             diag "The warnings were: " . join(", ", @warnings);
1268         }
1269     }
1270
1271
1272     # Test what happens when this malformation is not allowed
1273     undef @warnings;
1274     my $ret_ref = test_utf8n_to_uvchr($bytes, $length, 0);
1275     is($ret_ref->[0], 0, "$testname: disallowed: Returns 0");
1276     is($ret_ref->[1], $expected_len, "$testname: utf8n_to_uvchr(), disallowed: Returns expected length: $expected_len");
1277     if (is(scalar @warnings, 1, "$testname: disallowed: Got a single warning ")) {
1278         like($warnings[0], $message, "$testname: disallowed: Got expected warning");
1279     }
1280     else {
1281         if (scalar @warnings) {
1282             diag "The warnings were: " . join(", ", @warnings);
1283         }
1284     }
1285
1286     {   # Next test when disallowed, and warnings are off.
1287         undef @warnings;
1288         no warnings 'utf8';
1289         my $ret_ref = test_utf8n_to_uvchr($bytes, $length, 0);
1290         is($ret_ref->[0], 0, "$testname: utf8n_to_uvchr(), disallowed: no warnings 'utf8': Returns 0");
1291         is($ret_ref->[1], $expected_len, "$testname: utf8n_to_uvchr(), disallowed: no warnings 'utf8': Returns expected length: $expected_len");
1292         if (!is(scalar @warnings, 0, "$testname: utf8n_to_uvchr(), disallowed: no warnings 'utf8': no warnings generated")) {
1293             diag "The warnings were: " . join(", ", @warnings);
1294         }
1295     }
1296
1297     # Test with CHECK_ONLY
1298     undef @warnings;
1299     $ret_ref = test_utf8n_to_uvchr($bytes, $length, $UTF8_CHECK_ONLY);
1300     is($ret_ref->[0], 0, "$testname: CHECK_ONLY: Returns 0");
1301     is($ret_ref->[1], -1, "$testname: CHECK_ONLY: returns -1 for length");
1302     if (! is(scalar @warnings, 0, "$testname: CHECK_ONLY: no warnings generated")) {
1303         diag "The warnings were: " . join(", ", @warnings);
1304     }
1305
1306     next if $allow_flags == 0;    # Skip if can't allow this malformation
1307
1308     # Test when the malformation is allowed
1309     undef @warnings;
1310     $ret_ref = test_utf8n_to_uvchr($bytes, $length, $allow_flags);
1311     is($ret_ref->[0], $allowed_uv, "$testname: utf8n_to_uvchr(), allowed: Returns expected uv: " . sprintf("0x%04X", $allowed_uv));
1312     is($ret_ref->[1], $expected_len, "$testname: utf8n_to_uvchr(), allowed: Returns expected length: $expected_len");
1313     if (!is(scalar @warnings, 0, "$testname: utf8n_to_uvchr(), allowed: no warnings generated"))
1314     {
1315         diag "The warnings were: " . join(", ", @warnings);
1316     }
1317 }
1318
1319 # Now test the cases where a legal code point is generated, but may or may not
1320 # be allowed/warned on.
1321 my @tests = (
1322     [ "lowest surrogate",
1323         (isASCII) ? "\xed\xa0\x80" : I8_to_native("\xf1\xb6\xa0\xa0"),
1324         $UTF8_WARN_SURROGATE, $UTF8_DISALLOW_SURROGATE,
1325         'surrogate', 0xD800,
1326         (isASCII) ? 3 : 4,
1327         qr/surrogate/
1328     ],
1329     [ "a middle surrogate",
1330         (isASCII) ? "\xed\xa4\x8d" : I8_to_native("\xf1\xb6\xa8\xad"),
1331         $UTF8_WARN_SURROGATE, $UTF8_DISALLOW_SURROGATE,
1332         'surrogate', 0xD90D,
1333         (isASCII) ? 3 : 4,
1334         qr/surrogate/
1335     ],
1336     [ "highest surrogate",
1337         (isASCII) ? "\xed\xbf\xbf" : I8_to_native("\xf1\xb7\xbf\xbf"),
1338         $UTF8_WARN_SURROGATE, $UTF8_DISALLOW_SURROGATE,
1339         'surrogate', 0xDFFF,
1340         (isASCII) ? 3 : 4,
1341         qr/surrogate/
1342     ],
1343     [ "first non_unicode",
1344         (isASCII) ? "\xf4\x90\x80\x80" : I8_to_native("\xf9\xa2\xa0\xa0\xa0"),
1345         $UTF8_WARN_SUPER, $UTF8_DISALLOW_SUPER,
1346         'non_unicode', 0x110000,
1347         (isASCII) ? 4 : 5,
1348         qr/not Unicode.* may not be portable/
1349     ],
1350     [ "non_unicode whose first byte tells that",
1351         (isASCII) ? "\xf5\x80\x80\x80" : I8_to_native("\xfa\xa0\xa0\xa0\xa0"),
1352         $UTF8_WARN_SUPER, $UTF8_DISALLOW_SUPER,
1353         'non_unicode',
1354         (isASCII) ? 0x140000 : 0x200000,
1355         (isASCII) ? 4 : 5,
1356         qr/not Unicode.* may not be portable/
1357     ],
1358     [ "first of 32 consecutive non-character code points",
1359         (isASCII) ? "\xef\xb7\x90" : I8_to_native("\xf1\xbf\xae\xb0"),
1360         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1361         'nonchar', 0xFDD0,
1362         (isASCII) ? 3 : 4,
1363         qr/Unicode non-character.*is not recommended for open interchange/
1364     ],
1365     [ "a mid non-character code point of the 32 consecutive ones",
1366         (isASCII) ? "\xef\xb7\xa0" : I8_to_native("\xf1\xbf\xaf\xa0"),
1367         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1368         'nonchar', 0xFDE0,
1369         (isASCII) ? 3 : 4,
1370         qr/Unicode non-character.*is not recommended for open interchange/
1371     ],
1372     [ "final of 32 consecutive non-character code points",
1373         (isASCII) ? "\xef\xb7\xaf" : I8_to_native("\xf1\xbf\xaf\xaf"),
1374         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1375         'nonchar', 0xFDEF,
1376         (isASCII) ? 3 : 4,
1377         qr/Unicode non-character.*is not recommended for open interchange/
1378     ],
1379     [ "non-character code point U+FFFE",
1380         (isASCII) ? "\xef\xbf\xbe" : I8_to_native("\xf1\xbf\xbf\xbe"),
1381         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1382         'nonchar', 0xFFFE,
1383         (isASCII) ? 3 : 4,
1384         qr/Unicode non-character.*is not recommended for open interchange/
1385     ],
1386     [ "non-character code point U+FFFF",
1387         (isASCII) ? "\xef\xbf\xbf" : I8_to_native("\xf1\xbf\xbf\xbf"),
1388         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1389         'nonchar', 0xFFFF,
1390         (isASCII) ? 3 : 4,
1391         qr/Unicode non-character.*is not recommended for open interchange/
1392     ],
1393     [ "non-character code point U+1FFFE",
1394         (isASCII) ? "\xf0\x9f\xbf\xbe" : I8_to_native("\xf3\xbf\xbf\xbe"),
1395         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1396         'nonchar', 0x1FFFE, 4,
1397         qr/Unicode non-character.*is not recommended for open interchange/
1398     ],
1399     [ "non-character code point U+1FFFF",
1400         (isASCII) ? "\xf0\x9f\xbf\xbf" : I8_to_native("\xf3\xbf\xbf\xbf"),
1401         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1402         'nonchar', 0x1FFFF, 4,
1403         qr/Unicode non-character.*is not recommended for open interchange/
1404     ],
1405     [ "non-character code point U+2FFFE",
1406         (isASCII) ? "\xf0\xaf\xbf\xbe" : I8_to_native("\xf5\xbf\xbf\xbe"),
1407         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1408         'nonchar', 0x2FFFE, 4,
1409         qr/Unicode non-character.*is not recommended for open interchange/
1410     ],
1411     [ "non-character code point U+2FFFF",
1412         (isASCII) ? "\xf0\xaf\xbf\xbf" : I8_to_native("\xf5\xbf\xbf\xbf"),
1413         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1414         'nonchar', 0x2FFFF, 4,
1415         qr/Unicode non-character.*is not recommended for open interchange/
1416     ],
1417     [ "non-character code point U+3FFFE",
1418         (isASCII) ? "\xf0\xbf\xbf\xbe" : I8_to_native("\xf7\xbf\xbf\xbe"),
1419         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1420         'nonchar', 0x3FFFE, 4,
1421         qr/Unicode non-character.*is not recommended for open interchange/
1422     ],
1423     [ "non-character code point U+3FFFF",
1424         (isASCII) ? "\xf0\xbf\xbf\xbf" : I8_to_native("\xf7\xbf\xbf\xbf"),
1425         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1426         'nonchar', 0x3FFFF, 4,
1427         qr/Unicode non-character.*is not recommended for open interchange/
1428     ],
1429     [ "non-character code point U+4FFFE",
1430         (isASCII) ? "\xf1\x8f\xbf\xbe" : I8_to_native("\xf8\xa9\xbf\xbf\xbe"),
1431         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1432         'nonchar', 0x4FFFE,
1433         (isASCII) ? 4 : 5,
1434         qr/Unicode non-character.*is not recommended for open interchange/
1435     ],
1436     [ "non-character code point U+4FFFF",
1437         (isASCII) ? "\xf1\x8f\xbf\xbf" : I8_to_native("\xf8\xa9\xbf\xbf\xbf"),
1438         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1439         'nonchar', 0x4FFFF,
1440         (isASCII) ? 4 : 5,
1441         qr/Unicode non-character.*is not recommended for open interchange/
1442     ],
1443     [ "non-character code point U+5FFFE",
1444         (isASCII) ? "\xf1\x9f\xbf\xbe" : I8_to_native("\xf8\xab\xbf\xbf\xbe"),
1445         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1446         'nonchar', 0x5FFFE,
1447         (isASCII) ? 4 : 5,
1448         qr/Unicode non-character.*is not recommended for open interchange/
1449     ],
1450     [ "non-character code point U+5FFFF",
1451         (isASCII) ? "\xf1\x9f\xbf\xbf" : I8_to_native("\xf8\xab\xbf\xbf\xbf"),
1452         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1453         'nonchar', 0x5FFFF,
1454         (isASCII) ? 4 : 5,
1455         qr/Unicode non-character.*is not recommended for open interchange/
1456     ],
1457     [ "non-character code point U+6FFFE",
1458         (isASCII) ? "\xf1\xaf\xbf\xbe" : I8_to_native("\xf8\xad\xbf\xbf\xbe"),
1459         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1460         'nonchar', 0x6FFFE,
1461         (isASCII) ? 4 : 5,
1462         qr/Unicode non-character.*is not recommended for open interchange/
1463     ],
1464     [ "non-character code point U+6FFFF",
1465         (isASCII) ? "\xf1\xaf\xbf\xbf" : I8_to_native("\xf8\xad\xbf\xbf\xbf"),
1466         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1467         'nonchar', 0x6FFFF,
1468         (isASCII) ? 4 : 5,
1469         qr/Unicode non-character.*is not recommended for open interchange/
1470     ],
1471     [ "non-character code point U+7FFFE",
1472         (isASCII) ? "\xf1\xbf\xbf\xbe" : I8_to_native("\xf8\xaf\xbf\xbf\xbe"),
1473         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1474         'nonchar', 0x7FFFE,
1475         (isASCII) ? 4 : 5,
1476         qr/Unicode non-character.*is not recommended for open interchange/
1477     ],
1478     [ "non-character code point U+7FFFF",
1479         (isASCII) ? "\xf1\xbf\xbf\xbf" : I8_to_native("\xf8\xaf\xbf\xbf\xbf"),
1480         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1481         'nonchar', 0x7FFFF,
1482         (isASCII) ? 4 : 5,
1483         qr/Unicode non-character.*is not recommended for open interchange/
1484     ],
1485     [ "non-character code point U+8FFFE",
1486         (isASCII) ? "\xf2\x8f\xbf\xbe" : I8_to_native("\xf8\xb1\xbf\xbf\xbe"),
1487         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1488         'nonchar', 0x8FFFE,
1489         (isASCII) ? 4 : 5,
1490         qr/Unicode non-character.*is not recommended for open interchange/
1491     ],
1492     [ "non-character code point U+8FFFF",
1493         (isASCII) ? "\xf2\x8f\xbf\xbf" : I8_to_native("\xf8\xb1\xbf\xbf\xbf"),
1494         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1495         'nonchar', 0x8FFFF,
1496         (isASCII) ? 4 : 5,
1497         qr/Unicode non-character.*is not recommended for open interchange/
1498     ],
1499     [ "non-character code point U+9FFFE",
1500         (isASCII) ? "\xf2\x9f\xbf\xbe" : I8_to_native("\xf8\xb3\xbf\xbf\xbe"),
1501         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1502         'nonchar', 0x9FFFE,
1503         (isASCII) ? 4 : 5,
1504         qr/Unicode non-character.*is not recommended for open interchange/
1505     ],
1506     [ "non-character code point U+9FFFF",
1507         (isASCII) ? "\xf2\x9f\xbf\xbf" : I8_to_native("\xf8\xb3\xbf\xbf\xbf"),
1508         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1509         'nonchar', 0x9FFFF,
1510         (isASCII) ? 4 : 5,
1511         qr/Unicode non-character.*is not recommended for open interchange/
1512     ],
1513     [ "non-character code point U+AFFFE",
1514         (isASCII) ? "\xf2\xaf\xbf\xbe" : I8_to_native("\xf8\xb5\xbf\xbf\xbe"),
1515         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1516         'nonchar', 0xAFFFE,
1517         (isASCII) ? 4 : 5,
1518         qr/Unicode non-character.*is not recommended for open interchange/
1519     ],
1520     [ "non-character code point U+AFFFF",
1521         (isASCII) ? "\xf2\xaf\xbf\xbf" : I8_to_native("\xf8\xb5\xbf\xbf\xbf"),
1522         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1523         'nonchar', 0xAFFFF,
1524         (isASCII) ? 4 : 5,
1525         qr/Unicode non-character.*is not recommended for open interchange/
1526     ],
1527     [ "non-character code point U+BFFFE",
1528         (isASCII) ? "\xf2\xbf\xbf\xbe" : I8_to_native("\xf8\xb7\xbf\xbf\xbe"),
1529         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1530         'nonchar', 0xBFFFE,
1531         (isASCII) ? 4 : 5,
1532         qr/Unicode non-character.*is not recommended for open interchange/
1533     ],
1534     [ "non-character code point U+BFFFF",
1535         (isASCII) ? "\xf2\xbf\xbf\xbf" : I8_to_native("\xf8\xb7\xbf\xbf\xbf"),
1536         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1537         'nonchar', 0xBFFFF,
1538         (isASCII) ? 4 : 5,
1539         qr/Unicode non-character.*is not recommended for open interchange/
1540     ],
1541     [ "non-character code point U+CFFFE",
1542         (isASCII) ? "\xf3\x8f\xbf\xbe" : I8_to_native("\xf8\xb9\xbf\xbf\xbe"),
1543         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1544         'nonchar', 0xCFFFE,
1545         (isASCII) ? 4 : 5,
1546         qr/Unicode non-character.*is not recommended for open interchange/
1547     ],
1548     [ "non-character code point U+CFFFF",
1549         (isASCII) ? "\xf3\x8f\xbf\xbf" : I8_to_native("\xf8\xb9\xbf\xbf\xbf"),
1550         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1551         'nonchar', 0xCFFFF,
1552         (isASCII) ? 4 : 5,
1553         qr/Unicode non-character.*is not recommended for open interchange/
1554     ],
1555     [ "non-character code point U+DFFFE",
1556         (isASCII) ? "\xf3\x9f\xbf\xbe" : I8_to_native("\xf8\xbb\xbf\xbf\xbe"),
1557         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1558         'nonchar', 0xDFFFE,
1559         (isASCII) ? 4 : 5,
1560         qr/Unicode non-character.*is not recommended for open interchange/
1561     ],
1562     [ "non-character code point U+DFFFF",
1563         (isASCII) ? "\xf3\x9f\xbf\xbf" : I8_to_native("\xf8\xbb\xbf\xbf\xbf"),
1564         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1565         'nonchar', 0xDFFFF,
1566         (isASCII) ? 4 : 5,
1567         qr/Unicode non-character.*is not recommended for open interchange/
1568     ],
1569     [ "non-character code point U+EFFFE",
1570         (isASCII) ? "\xf3\xaf\xbf\xbe" : I8_to_native("\xf8\xbd\xbf\xbf\xbe"),
1571         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1572         'nonchar', 0xEFFFE,
1573         (isASCII) ? 4 : 5,
1574         qr/Unicode non-character.*is not recommended for open interchange/
1575     ],
1576     [ "non-character code point U+EFFFF",
1577         (isASCII) ? "\xf3\xaf\xbf\xbf" : I8_to_native("\xf8\xbd\xbf\xbf\xbf"),
1578         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1579         'nonchar', 0xEFFFF,
1580         (isASCII) ? 4 : 5,
1581         qr/Unicode non-character.*is not recommended for open interchange/
1582     ],
1583     [ "non-character code point U+FFFFE",
1584         (isASCII) ? "\xf3\xbf\xbf\xbe" : I8_to_native("\xf8\xbf\xbf\xbf\xbe"),
1585         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1586         'nonchar', 0xFFFFE,
1587         (isASCII) ? 4 : 5,
1588         qr/Unicode non-character.*is not recommended for open interchange/
1589     ],
1590     [ "non-character code point U+FFFFF",
1591         (isASCII) ? "\xf3\xbf\xbf\xbf" : I8_to_native("\xf8\xbf\xbf\xbf\xbf"),
1592         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1593         'nonchar', 0xFFFFF,
1594         (isASCII) ? 4 : 5,
1595         qr/Unicode non-character.*is not recommended for open interchange/
1596     ],
1597     [ "non-character code point U+10FFFE",
1598         (isASCII) ? "\xf4\x8f\xbf\xbe" : I8_to_native("\xf9\xa1\xbf\xbf\xbe"),
1599         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1600         'nonchar', 0x10FFFE,
1601         (isASCII) ? 4 : 5,
1602         qr/Unicode non-character.*is not recommended for open interchange/
1603     ],
1604     [ "non-character code point U+10FFFF",
1605         (isASCII) ? "\xf4\x8f\xbf\xbf" : I8_to_native("\xf9\xa1\xbf\xbf\xbf"),
1606         $UTF8_WARN_NONCHAR, $UTF8_DISALLOW_NONCHAR,
1607         'nonchar', 0x10FFFF,
1608         (isASCII) ? 4 : 5,
1609         qr/Unicode non-character.*is not recommended for open interchange/
1610     ],
1611     [ "requires at least 32 bits",
1612         (isASCII)
1613          ? "\xfe\x82\x80\x80\x80\x80\x80"
1614          : I8_to_native("\xff\xa0\xa0\xa0\xa0\xa0\xa0\xa2\xa0\xa0\xa0\xa0\xa0\xa0"),
1615         # This code point is chosen so that it is representable in a UV on
1616         # 32-bit machines
1617         $UTF8_WARN_ABOVE_31_BIT, $UTF8_DISALLOW_ABOVE_31_BIT,
1618         'utf8', 0x80000000, (isASCII) ? 7 :14,
1619         qr/Code point 0x80000000 is not Unicode, and not portable/
1620     ],
1621     [ "requires at least 32 bits, and use SUPER-type flags, instead of ABOVE_31_BIT",
1622         (isASCII)
1623          ? "\xfe\x82\x80\x80\x80\x80\x80"
1624          : I8_to_native("\xff\xa0\xa0\xa0\xa0\xa0\xa0\xa2\xa0\xa0\xa0\xa0\xa0\xa0"),
1625         $UTF8_WARN_SUPER, $UTF8_DISALLOW_SUPER,
1626         'utf8', 0x80000000, (isASCII) ? 7 :14,
1627         qr/Code point 0x80000000 is not Unicode, and not portable/
1628     ],
1629     [ "overflow with warnings/disallow for more than 31 bits",
1630         # This tests the interaction of WARN_ABOVE_31_BIT/DISALLOW_ABOVE_31_BIT
1631         # with overflow.  The overflow malformation is never allowed, so
1632         # preventing it takes precedence if the ABOVE_31_BIT options would
1633         # otherwise allow in an overflowing value.  The ASCII code points (1
1634         # for 32-bits; 1 for 64) were chosen because the old overflow
1635         # detection algorithm did not catch them; this means this test also
1636         # checks for that fix.  The EBCDIC are arbitrary overflowing ones
1637         # since we have no reports of failures with it.
1638        (($is64bit)
1639         ? ((isASCII)
1640            ?              "\xff\x80\x90\x90\x90\xbf\xbf\xbf\xbf\xbf\xbf\xbf\xbf"
1641            : I8_to_native("\xff\xB0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0"))
1642         : ((isASCII)
1643            ?              "\xfe\x86\x80\x80\x80\x80\x80"
1644            : I8_to_native("\xff\xa0\xa0\xa0\xa0\xa0\xa0\xa4\xa0\xa0\xa0\xa0\xa0\xa0"))),
1645
1646         # We include both warning categories to make sure the ABOVE_31_BIT one
1647         # has precedence
1648         "$UTF8_WARN_ABOVE_31_BIT|$UTF8_WARN_SUPER",
1649         "$UTF8_DISALLOW_ABOVE_31_BIT",
1650         'utf8', 0,
1651         (! isASCII) ? 14 : ($is64bit) ? 13 : 7,
1652         qr/overflow at byte .*, after start byte 0xf/
1653     ],
1654 );
1655
1656 if ($is64bit) {
1657     no warnings qw{portable overflow};
1658     push @tests,
1659         [ "More than 32 bits",
1660             (isASCII)
1661             ?              "\xff\x80\x80\x80\x80\x80\x81\x80\x80\x80\x80\x80\x80"
1662             : I8_to_native("\xff\xa0\xa0\xa0\xa0\xa0\xa2\xa0\xa0\xa0\xa0\xa0\xa0\xa0"),
1663             $UTF8_WARN_ABOVE_31_BIT, $UTF8_DISALLOW_ABOVE_31_BIT,
1664             'utf8', 0x1000000000, (isASCII) ? 13 : 14,
1665             qr/Code point 0x.* is not Unicode, and not portable/
1666         ];
1667     if (! isASCII) {
1668         push @tests,   # These could falsely show wrongly in a naive implementation
1669             [ "requires at least 32 bits",
1670                 I8_to_native("\xff\xa0\xa0\xa0\xa0\xa0\xa1\xa0\xa0\xa0\xa0\xa0\xa0\xa0"),
1671                 $UTF8_WARN_ABOVE_31_BIT,$UTF8_DISALLOW_ABOVE_31_BIT,
1672                 'utf8', 0x800000000, 14,
1673                 qr/Code point 0x800000000 is not Unicode, and not portable/
1674             ],
1675             [ "requires at least 32 bits",
1676                 I8_to_native("\xff\xa0\xa0\xa0\xa0\xa1\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0"),
1677                 $UTF8_WARN_ABOVE_31_BIT,$UTF8_DISALLOW_ABOVE_31_BIT,
1678                 'utf8', 0x10000000000, 14,
1679                 qr/Code point 0x10000000000 is not Unicode, and not portable/
1680             ],
1681             [ "requires at least 32 bits",
1682                 I8_to_native("\xff\xa0\xa0\xa0\xa1\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0"),
1683                 $UTF8_WARN_ABOVE_31_BIT,$UTF8_DISALLOW_ABOVE_31_BIT,
1684                 'utf8', 0x200000000000, 14,
1685                 qr/Code point 0x200000000000 is not Unicode, and not portable/
1686             ],
1687             [ "requires at least 32 bits",
1688                 I8_to_native("\xff\xa0\xa0\xa1\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0"),
1689                 $UTF8_WARN_ABOVE_31_BIT,$UTF8_DISALLOW_ABOVE_31_BIT,
1690                 'utf8', 0x4000000000000, 14,
1691                 qr/Code point 0x4000000000000 is not Unicode, and not portable/
1692             ],
1693             [ "requires at least 32 bits",
1694                 I8_to_native("\xff\xa0\xa1\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0"),
1695                 $UTF8_WARN_ABOVE_31_BIT,$UTF8_DISALLOW_ABOVE_31_BIT,
1696                 'utf8', 0x80000000000000, 14,
1697                 qr/Code point 0x80000000000000 is not Unicode, and not portable/
1698             ],
1699             [ "requires at least 32 bits",
1700                 I8_to_native("\xff\xa1\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0\xa0"),
1701                 $UTF8_WARN_ABOVE_31_BIT,$UTF8_DISALLOW_ABOVE_31_BIT,
1702                 'utf8', 0x1000000000000000, 14,
1703                 qr/Code point 0x1000000000000000 is not Unicode, and not portable/
1704             ];
1705     }
1706 }
1707
1708 foreach my $test (@tests) {
1709     my ($testname, $bytes, $warn_flags, $disallow_flags, $category, $allowed_uv, $expected_len, $message ) = @$test;
1710
1711     my $length = length $bytes;
1712     my $will_overflow = $testname =~ /overflow/;
1713
1714     {
1715         use warnings;
1716         undef @warnings;
1717         my $ret = test_isUTF8_CHAR($bytes, $length);
1718         my $ret_flags = test_isUTF8_CHAR_flags($bytes, $length, 0);
1719         if ($will_overflow) {
1720             is($ret, 0, "isUTF8_CHAR() $testname: returns 0");
1721             is($ret_flags, 0, "isUTF8_CHAR_flags() $testname: returns 0");
1722         }
1723         else {
1724             is($ret, $length,
1725                "isUTF8_CHAR() $testname: returns expected length: $length");
1726             is($ret_flags, $length,
1727                "isUTF8_CHAR_flags(...,0) $testname: returns expected length: $length");
1728         }
1729         unless (is(scalar @warnings, 0,
1730                 "isUTF8_CHAR() and isUTF8_CHAR()_flags $testname: generated no warnings"))
1731         {
1732             diag "The warnings were: " . join(", ", @warnings);
1733         }
1734
1735         undef @warnings;
1736         $ret = test_isSTRICT_UTF8_CHAR($bytes, $length);
1737         if ($will_overflow) {
1738             is($ret, 0, "isSTRICT_UTF8_CHAR() $testname: returns 0");
1739         }
1740         else {
1741             my $expected_ret = (   $testname =~ /surrogate|non-character/
1742                                 || $allowed_uv > 0x10FFFF)
1743                                ? 0
1744                                : $length;
1745             is($ret, $expected_ret,
1746                "isSTRICT_UTF8_CHAR() $testname: returns expected length: $expected_ret");
1747             $ret = test_isUTF8_CHAR_flags($bytes, $length,
1748                                           $UTF8_DISALLOW_ILLEGAL_INTERCHANGE);
1749             is($ret, $expected_ret,
1750                "isUTF8_CHAR_flags('DISALLOW_ILLEGAL_INTERCHANGE') acts like isSTRICT_UTF8_CHAR");
1751         }
1752         unless (is(scalar @warnings, 0,
1753                 "isSTRICT_UTF8_CHAR() and isUTF8_CHAR_flags $testname: generated no warnings"))
1754         {
1755             diag "The warnings were: " . join(", ", @warnings);
1756         }
1757
1758         undef @warnings;
1759         $ret = test_isC9_STRICT_UTF8_CHAR($bytes, $length);
1760         if ($will_overflow) {
1761             is($ret, 0, "isC9_STRICT_UTF8_CHAR() $testname: returns 0");
1762         }
1763         else {
1764             my $expected_ret = (   $testname =~ /surrogate/
1765                                 || $allowed_uv > 0x10FFFF)
1766                                ? 0
1767                                : $length;
1768             is($ret, $expected_ret,
1769                "isC9_STRICT_UTF8_CHAR() $testname: returns expected length: $expected_ret");
1770             $ret = test_isUTF8_CHAR_flags($bytes, $length,
1771                                           $UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE);
1772             is($ret, $expected_ret,
1773                "isUTF8_CHAR_flags('DISALLOW_ILLEGAL_C9_INTERCHANGE') acts like isC9_STRICT_UTF8_CHAR");
1774         }
1775         unless (is(scalar @warnings, 0,
1776                 "isC9_STRICT_UTF8_CHAR() and isUTF8_CHAR_flags $testname: generated no warnings"))
1777         {
1778             diag "The warnings were: " . join(", ", @warnings);
1779         }
1780
1781         # Test partial character handling, for each byte not a full character
1782         for my $j (1.. $length - 1) {
1783
1784             # Skip the test for the interaction between overflow and above-31
1785             # bit.  It is really testing other things than the partial
1786             # character tests, for which other tests in this file are
1787             # sufficient
1788             last if $testname =~ /overflow/;
1789
1790             foreach my $disallow_flag (0, $disallow_flags) {
1791                 my $partial = substr($bytes, 0, $j);
1792                 my $ret_should_be;
1793                 my $comment;
1794                 if ($disallow_flag) {
1795                     $ret_should_be = 0;
1796                     $comment = "disallowed";
1797                 }
1798                 else {
1799                     $ret_should_be = 1;
1800                     $comment = "allowed";
1801                 }
1802
1803                 if ($disallow_flag) {
1804                     if ($testname =~ /non-character/) {
1805                         $ret_should_be = 1;
1806                         $comment .= ", but but need full char to discern";
1807                     }
1808                     elsif ($testname =~ /surrogate/) {
1809                         if ($j < 2) {
1810                             $ret_should_be = 1;
1811                             $comment .= ", but need 2 bytes to discern";
1812                         }
1813                     }
1814                     elsif ($testname =~ /first non_unicode/ && $j < 2) {
1815                         $ret_should_be = 1;
1816                         $comment .= ", but need 2 bytes to discern";
1817                     }
1818                     elsif (   ! isASCII
1819                            && $testname =~ /requires at least 32 bits/)
1820                     {
1821                         # On EBCDIC, the boundary between 31 and 32 bits is
1822                         # more complicated.
1823                         $ret_should_be = 1 if native_to_I8($partial) le
1824                      "\xFF\xA0\xA0\xA0\xA0\xA0\xA0\xA1\xBF\xBF\xBF\xBF\xBF\xBF";
1825                     }
1826                 }
1827
1828                 undef @warnings;
1829
1830                 $ret = test_is_utf8_valid_partial_char_flags($partial, $j, $disallow_flag);
1831                 is($ret, $ret_should_be, "$testname: is_utf8_valid_partial_char_flags("
1832                                         . display_bytes($partial)
1833                                         . "), $comment: returns $ret_should_be");
1834                 unless (is(scalar @warnings, 0,
1835                         "$testname: is_utf8_valid_partial_char_flags() generated no warnings"))
1836                 {
1837                     diag "The warnings were: " . join(", ", @warnings);
1838                 }
1839             }
1840         }
1841     }
1842
1843     # This is more complicated than the malformations tested earlier, as there
1844     # are several orthogonal variables involved.  We test all the subclasses
1845     # of utf8 warnings to verify they work with and without the utf8 class,
1846     # and don't have effects on other sublass warnings
1847     foreach my $warning ('utf8', 'surrogate', 'nonchar', 'non_unicode') {
1848         foreach my $warn_flag (0, $warn_flags) {
1849             foreach my $disallow_flag (0, $disallow_flags) {
1850                 foreach my $do_warning (0, 1) {
1851
1852                     my $eval_warn = $do_warning
1853                                   ? "use warnings '$warning'"
1854                                   : $warning eq "utf8"
1855                                     ? "no warnings 'utf8'"
1856                                     : "use warnings 'utf8'; no warnings '$warning'";
1857
1858                     # is effectively disallowed if will overflow, even if the
1859                     # flag indicates it is allowed, fix up test name to
1860                     # indicate this as well
1861                     my $disallowed = $disallow_flag || $will_overflow;
1862
1863                     my $this_name = "utf8n_to_uvchr() $testname: " . (($disallow_flag)
1864                                                     ? 'disallowed'
1865                                                     : ($disallowed)
1866                                                         ? 'ABOVE_31_BIT allowed'
1867                                                         : 'allowed');
1868                     $this_name .= ", $eval_warn";
1869                     $this_name .= ", " . (($warn_flag)
1870                                           ? 'with warning flag'
1871                                           : 'no warning flag');
1872
1873                     undef @warnings;
1874                     my $ret_ref;
1875                     my $display_bytes = display_bytes($bytes);
1876                     my $call = "Call was: $eval_warn; \$ret_ref = test_utf8n_to_uvchr('$display_bytes', $length, $warn_flag|$disallow_flag)";
1877                     my $eval_text =      "$eval_warn; \$ret_ref = test_utf8n_to_uvchr('$bytes', $length, $warn_flag|$disallow_flag)";
1878                     eval "$eval_text";
1879                     if (! ok ("$@ eq ''", "$this_name: eval succeeded")) {
1880                         diag "\$!='$!'; eval'd=\"$call\"";
1881                         next;
1882                     }
1883                     if ($disallowed) {
1884                         unless (is($ret_ref->[0], 0, "$this_name: Returns 0"))
1885                         {
1886                             diag $call;
1887                         }
1888                     }
1889                     else {
1890                         unless (is($ret_ref->[0], $allowed_uv,
1891                                    "$this_name: Returns expected uv: "
1892                                  . sprintf("0x%04X", $allowed_uv)))
1893                         {
1894                             diag $call;
1895                         }
1896                     }
1897                     unless (is($ret_ref->[1], $expected_len,
1898                         "$this_name: Returns expected length: $expected_len"))
1899                     {
1900                         diag $call;
1901                     }
1902
1903                     if (! $do_warning
1904                         && ($warning eq 'utf8' || $warning eq $category))
1905                     {
1906                         if (!is(scalar @warnings, 0,
1907                                             "$this_name: No warnings generated"))
1908                         {
1909                             diag $call;
1910                             diag "The warnings were: " . join(", ", @warnings);
1911                         }
1912                     }
1913                     elsif ($will_overflow
1914                            && ! $disallow_flag
1915                            && $warning eq 'utf8')
1916                     {
1917
1918                         # Will get the overflow message instead of the expected
1919                         # message under these circumstances, as they would
1920                         # otherwise accept an overflowed value, which the code
1921                         # should not allow, so falls back to overflow.
1922                         if (is(scalar @warnings, 1,
1923                                "$this_name: Got a single warning "))
1924                         {
1925                             unless (like($warnings[0], qr/overflow/,
1926                                         "$this_name: Got overflow warning"))
1927                             {
1928                                 diag $call;
1929                             }
1930                         }
1931                         else {
1932                             diag $call;
1933                             if (scalar @warnings) {
1934                                 diag "The warnings were: "
1935                                                         . join(", ", @warnings);
1936                             }
1937                         }
1938                     }
1939                     elsif ($warn_flag
1940                            && ($warning eq 'utf8' || $warning eq $category))
1941                     {
1942                         if (is(scalar @warnings, 1,
1943                                "$this_name: Got a single warning "))
1944                         {
1945                             unless (like($warnings[0], $message,
1946                                         "$this_name: Got expected warning"))
1947                             {
1948                                 diag $call;
1949                             }
1950                         }
1951                         else {
1952                             diag $call;
1953                             if (scalar @warnings) {
1954                                 diag "The warnings were: "
1955                                                         . join(", ", @warnings);
1956                             }
1957                         }
1958                     }
1959
1960                     # Check CHECK_ONLY results when the input is disallowed.  Do
1961                     # this when actually disallowed, not just when the
1962                     # $disallow_flag is set
1963                     if ($disallowed) {
1964                         undef @warnings;
1965                         $ret_ref = test_utf8n_to_uvchr($bytes, $length,
1966                                                 $disallow_flag|$UTF8_CHECK_ONLY);
1967                         unless (is($ret_ref->[0], 0, "$this_name, CHECK_ONLY: Returns 0")) {
1968                             diag $call;
1969                         }
1970                         unless (is($ret_ref->[1], -1,
1971                             "$this_name: CHECK_ONLY: returns -1 for length"))
1972                         {
1973                             diag $call;
1974                         }
1975                         if (! is(scalar @warnings, 0,
1976                             "$this_name, CHECK_ONLY: no warnings generated"))
1977                         {
1978                             diag $call;
1979                             diag "The warnings were: " . join(", ", @warnings);
1980                         }
1981                     }
1982
1983                     # Now repeat some of the above, but for
1984                     # uvchr_to_utf8_flags().  Since this comes from an
1985                     # existing code point, it hasn't overflowed.
1986                     next if $will_overflow;
1987
1988                     # The warning and disallow flags passed in are for
1989                     # utf8n_to_uvchr().  Convert them for
1990                     # uvchr_to_utf8_flags().
1991                     my $uvchr_warn_flag = 0;
1992                     my $uvchr_disallow_flag = 0;
1993                     if ($warn_flag) {
1994                         if ($warn_flag == $UTF8_WARN_SURROGATE) {
1995                             $uvchr_warn_flag = $UNICODE_WARN_SURROGATE
1996                         }
1997                         elsif ($warn_flag == $UTF8_WARN_NONCHAR) {
1998                             $uvchr_warn_flag = $UNICODE_WARN_NONCHAR
1999                         }
2000                         elsif ($warn_flag == $UTF8_WARN_SUPER) {
2001                             $uvchr_warn_flag = $UNICODE_WARN_SUPER
2002                         }
2003                         elsif ($warn_flag == $UTF8_WARN_ABOVE_31_BIT) {
2004                             $uvchr_warn_flag = $UNICODE_WARN_ABOVE_31_BIT;
2005                         }
2006                         else {
2007                             fail(sprintf "Unexpected warn flag: %x",
2008                                  $warn_flag);
2009                             next;
2010                         }
2011                     }
2012                     if ($disallow_flag) {
2013                         if ($disallow_flag == $UTF8_DISALLOW_SURROGATE) {
2014                             $uvchr_disallow_flag = $UNICODE_DISALLOW_SURROGATE
2015                         }
2016                         elsif ($disallow_flag == $UTF8_DISALLOW_NONCHAR) {
2017                             $uvchr_disallow_flag = $UNICODE_DISALLOW_NONCHAR
2018                         }
2019                         elsif ($disallow_flag == $UTF8_DISALLOW_SUPER) {
2020                             $uvchr_disallow_flag = $UNICODE_DISALLOW_SUPER
2021                         }
2022                         elsif ($disallow_flag == $UTF8_DISALLOW_ABOVE_31_BIT) {
2023                             $uvchr_disallow_flag =
2024                             $UNICODE_DISALLOW_ABOVE_31_BIT;
2025                         }
2026                         else {
2027                             fail(sprintf "Unexpected disallow flag: %x",
2028                                  $disallow_flag);
2029                             next;
2030                         }
2031                     }
2032
2033                     $disallowed = $uvchr_disallow_flag;
2034
2035                     $this_name = "uvchr_to_utf8_flags() $testname: "
2036                                                   . (($uvchr_disallow_flag)
2037                                                     ? 'disallowed'
2038                                                     : ($disallowed)
2039                                                       ? 'ABOVE_31_BIT allowed'
2040                                                       : 'allowed');
2041                     $this_name .= ", $eval_warn";
2042                     $this_name .= ", " . (($uvchr_warn_flag)
2043                                           ? 'with warning flag'
2044                                           : 'no warning flag');
2045
2046                     undef @warnings;
2047                     my $ret;
2048                     my $warn_flag = sprintf "0x%x", $uvchr_warn_flag;
2049                     my $disallow_flag = sprintf "0x%x", $uvchr_disallow_flag;
2050                     $call = sprintf "call was: $eval_warn; \$ret = test_uvchr_to_utf8_flags(0x%x, $warn_flag|$disallow_flag)", $allowed_uv;
2051                     $eval_text = "$eval_warn; \$ret = test_uvchr_to_utf8_flags($allowed_uv, $warn_flag|$disallow_flag)";
2052                     eval "$eval_text";
2053                     if (! ok ("$@ eq ''", "$this_name: eval succeeded")) {
2054                         diag "\$!='$!'; eval'd=\"$eval_text\"";
2055                         next;
2056                     }
2057                     if ($disallowed) {
2058                         unless (is($ret, undef, "$this_name: Returns undef")) {
2059                             diag $call;
2060                         }
2061                     }
2062                     else {
2063                         unless (is($ret, $bytes, "$this_name: Returns expected string")) {
2064                             diag $call;
2065                         }
2066                     }
2067                     if (! $do_warning
2068                         && ($warning eq 'utf8' || $warning eq $category))
2069                     {
2070                         if (!is(scalar @warnings, 0,
2071                                             "$this_name: No warnings generated"))
2072                         {
2073                             diag $call;
2074                             diag "The warnings were: " . join(", ", @warnings);
2075                         }
2076                     }
2077                     elsif ($uvchr_warn_flag
2078                            && ($warning eq 'utf8' || $warning eq $category))
2079                     {
2080                         if (is(scalar @warnings, 1,
2081                                "$this_name: Got a single warning "))
2082                         {
2083                             unless (like($warnings[0], $message,
2084                                             "$this_name: Got expected warning"))
2085                             {
2086                                 diag $call;
2087                             }
2088                         }
2089                         else {
2090                             diag $call;
2091                             if (scalar @warnings) {
2092                                 diag "The warnings were: "
2093                                                         . join(", ", @warnings);
2094                             }
2095                         }
2096                     }
2097                 }
2098             }
2099         }
2100     }
2101 }
2102
2103 done_testing;