6 BEGIN { unshift @INC, '.' }
8 require './regen/regen_lib.pl';
9 require './regen/charset_translations.pl';
11 # Generates the EBCDIC translation tables that were formerly hard-coded into
14 my $out_fh = open_new('ebcdic_tables.h', '>',
15 {style => '*', by => $0, });
17 sub get_column_headers ($$;$) {
18 my ($row_hdr_len, $field_width, $dfa_columns) = @_;
20 my $final_column_format;
23 if (defined $dfa_columns) {
24 $num_columns = $dfa_columns;
26 # Trailing blank to correspond with commas in the rows below
27 $format = "%${field_width}d ";
29 else { # Is a regular table
32 # Use blanks to separate the fields
33 $format = " " x ( $field_width
35 $format .= "_%X "; # Again, trailing blank over the commas below
38 my $header = "/*" . " " x ($row_hdr_len - length "/*");
40 # All but the final column
41 $header .= sprintf($format, $_) for 0 .. $num_columns - 2;
43 # Get rid of trailing blank, so that the final column takes up one less
44 # space so that the "*/" doesn't extend past the commas in the rows below
46 $header .= sprintf $format, $num_columns - 1;
48 # Again, remove trailing blank
51 return $header . "*/\n";
54 sub output_table_start($$$;$) {
55 my ($out_fh, $TYPE, $name, $size) = @_;
57 $size = "" unless defined $size;
58 my $declaration = "EXTCONST $TYPE $name\[$size\]";
67 sub output_table_end($) {
68 print $out_fh "};\n# endif\n\n";
71 sub output_table ($$;$) {
72 my $table_ref = shift;
75 # 0 => print in decimal
76 # 1 => print in hex (translates code point to code point)
77 # >= 2 => is a dfa table, like https://bjoern.hoehrmann.de/utf-8/decoder/dfa/
78 # The number is how many columns in the part after the code point
81 # code point tables in hex areasier to debug, but don't fit into 80
83 my $type = shift // 1;
85 my $print_in_hex = $type == 1;
86 my $is_dfa = ($type >= 2) ? $type : 0;
87 my $columns_after_256 = 16;
89 die "Requres 256 entries in table $name, got @$table_ref"
90 if ! $is_dfa && @$table_ref != 256;
92 die "Requres 256 entries in table $name, got @$table_ref"
93 if @$table_ref != 256;
96 $columns_after_256 = $is_dfa;
98 print $out_fh <<'EOF';
100 /* The table below is adapted from
101 * https://bjoern.hoehrmann.de/utf-8/decoder/dfa/
102 * See copyright notice at the beginning of this file.
108 # Highest number in the table
110 $max_entry = map { $_ > $max_entry ? $_ : $max_entry } @$table_ref;
112 # We assume that every table has at least one two digit entry, and none
113 # are more than three digit.
114 my $field_width = ($print_in_hex)
116 : (($max_entry) > 99 ? 3 : 2);
119 my $node_number_field_width;
120 my $node_value_field_width;
122 # dfa tables have a special header for the rows in the transitions part of
123 # the table. It is longer than the regular one.
125 my $max_node_number = ($max_entry - 256) / $columns_after_256 - 1;
126 $node_number_field_width = ($max_node_number > 9) ? 2 : 1;
127 $node_value_field_width = ($max_node_number * $columns_after_256 > 99)
129 # The header starts with this template, and adds in the number of
130 # digits needed to represent the maximum node number and its value
131 $row_hdr_length = length("/*N=*/")
132 + $node_number_field_width
133 + $node_value_field_width;
136 $row_hdr_length = length "/*_X*/"; # Template for what the header
140 # The table may not be representable in 8 bits.
142 $TYPE = 'U16' if grep { $_ > 255 } @$table_ref;
144 output_table_start $out_fh, $TYPE, $name;
146 # First the headers for the columns
147 print $out_fh get_column_headers($row_hdr_length, $field_width);
150 my $count = @$table_ref;
153 # Print each element individually, arranged in rows of columns
154 for my $i (0 .. $count - 1) {
156 # Node number for here is -1 until get into the dfa state transitions
157 my $node = ($i < 256) ? -1 : ($i - 256) / $columns_after_256;
159 # Print row header at beginning of each row
162 printf $out_fh "/*N%-*d=%*d*/", $node_number_field_width, $node,
163 $node_value_field_width, $i - 256;
165 else { # Otherwise is regular row; print its number
166 printf $out_fh "/*%X_", $i / 16;
168 # These rows in a dfa table require extra space so columns
169 # will align vertically (because the Ndd=ddd requires extra
172 print $out_fh " " x ( $node_number_field_width
173 + $node_value_field_width);
180 printf $out_fh "0x%02X", $table_ref->[$i];
183 printf $out_fh "%${field_width}d", $table_ref->[$i];
186 print $out_fh ",", if $i < $count -1; # No comma on final entry
188 # Add \n if at end of row, which is 16 columns until we get to the
190 if ( ($node < 0 && $i % 16 == 15)
191 || ($node >= 0 && ($i -256) % $columns_after_256
192 == $columns_after_256 - 1))
202 # Print column footer
203 print $out_fh get_column_headers($row_hdr_length, $field_width,
204 ($is_dfa) ? $columns_after_256 : undef);
206 output_table_end($out_fh);
209 print $out_fh <<'END';
211 #ifndef PERL_EBCDIC_TABLES_H_ /* Guard against nested #includes */
212 #define PERL_EBCDIC_TABLES_H_ 1
214 /* This file contains definitions for various tables used in EBCDIC handling.
215 * More info is in utfebcdic.h
217 * Some of the tables are adapted from
218 * https://bjoern.hoehrmann.de/utf-8/decoder/dfa/
219 * which requires this copyright notice:
221 Copyright (c) 2008-2009 Bjoern Hoehrmann <bjoern@hoehrmann.de>
223 Permission is hereby granted, free of charge, to any person obtaining a copy of
224 this software and associated documentation files (the "Software"), to deal in
225 the Software without restriction, including without limitation the rights to
226 use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies
227 of the Software, and to permit persons to whom the Software is furnished to do
228 so, subject to the following conditions:
230 The above copyright notice and this permission notice shall be included in all
231 copies or substantial portions of the Software.
233 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
234 IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
235 FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
236 AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
237 LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
238 OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
244 my @charsets = get_supported_code_pages();
245 shift @charsets; # ASCII is the 0th, and we don't deal with that here.
246 foreach my $charset (@charsets) {
247 # we process the whole array several times, make a copy
248 my @a2e = @{get_a2n($charset)};
251 print $out_fh "\n" . get_conditional_compile_line_start($charset);
254 print $out_fh "/* Index is ASCII platform code point; value is $charset equivalent */\n";
255 output_table(\@a2e, "PL_a2e");
257 { # Construct the inverse
258 for my $i (0 .. 255) {
261 print $out_fh "/* Index is $charset code point; value is ASCII platform equivalent */\n";
262 output_table(\@e2a, "PL_e2a");
265 my @i82utf = @{get_I8_2_utf($charset)};
267 /* (Confusingly named) Index is $charset I8 byte; value is
268 * $charset UTF-EBCDIC equivalent */
270 output_table(\@i82utf, "PL_utf2e");
272 { #Construct the inverse
274 for my $i (0 .. 255) {
275 $utf2i8[$i82utf[$i]] = $i;
278 /* (Confusingly named) Index is $charset UTF-EBCDIC byte; value is
279 * $charset I8 equivalent */
281 output_table(\@utf2i8, "PL_e2utf");
287 # These are invariants or continuation bytes.
288 for my $i (0 .. 0xBF) {
289 $utf8skip[$i82utf[$i]] = 1;
292 # These are start bytes; The skip is the number of consecutive highest
293 # order 1-bits (up to 7)
294 for my $i (0xC0 .. 255) {
296 if ($i == 0b11111111) {
298 $count = $CHARSET_TRANSLATIONS::UTF_EBCDIC_MAXBYTES;
300 elsif (($i & 0b11111110) == 0b11111110) {
303 elsif (($i & 0b11111100) == 0b11111100) {
306 elsif (($i & 0b11111000) == 0b11111000) {
309 elsif (($i & 0b11110000) == 0b11110000) {
312 elsif (($i & 0b11100000) == 0b11100000) {
315 elsif (($i & 0b11000000) == 0b11000000) {
319 die "Something wrong for UTF8SKIP calculation for $i";
321 $utf8skip[$i82utf[$i]] = $count;
325 /* Index is $charset UTF-EBCDIC byte; value is UTF8SKIP for start bytes
326 * (including for overlongs); 1 for continuation. Adapted from the shadow
327 * flags table in tr16. The entries marked 9 in tr16 are continuation bytes
328 * and are marked as length 1 here so that we can recover. */
330 output_table(\@utf8skip, "PL_utf8skip", 0); # The 0 means don't print
334 use feature 'unicode_strings';
338 for my $i (0 .. 255) {
339 $lc[$a2e[$i]] = $a2e[ord lc chr $i];
342 "/* Index is $charset code point; value is its lowercase equivalent */\n";
343 output_table(\@lc, "PL_latin1_lc");
348 for my $i (0 .. 255) {
350 if (length $uc > 1 || ord $uc > 255) {
351 $uc = "\N{LATIN SMALL LETTER Y WITH DIAERESIS}";
353 $uc[$a2e[$i]] = $a2e[ord $uc];
356 /* Index is $charset code point; value is its uppercase equivalent.
357 * The 'mod' in the name means that codepoints whose uppercase is above 255 or
358 * longer than 1 character map to LATIN SMALL LETTER Y WITH DIARESIS */
360 output_table(\@uc, "PL_mod_latin1_uc");
365 for my $i (0 .. 255) { # Initialise to identity map
366 $ascii_fold[$i] = $i;
369 # Overwrite the entries that aren't identity
370 for my $chr ('A' .. 'Z') {
371 $ascii_fold[$a2e[ord $chr]] = $a2e[ord lc $chr];
373 for my $chr ('a' .. 'z') {
374 $ascii_fold[$a2e[ord $chr]] = $a2e[ord uc $chr];
377 /* Index is $charset code point; For A-Z, value is a-z; for a-z, value
378 * is A-Z; all other code points map to themselves */
380 output_table(\@ascii_fold, "PL_fold");
385 for my $i (0 .. 255) {
389 # lc and uc adequately proxy for fold-case pairs in this 0-255
392 $uc = $char if length $uc > 1 || ord $uc > 255;
394 $latin1_fold[$a2e[$i]] = $a2e[ord $lc];
396 elsif ($uc ne $char) {
397 $latin1_fold[$a2e[$i]] = $a2e[ord $uc];
400 $latin1_fold[$a2e[$i]] = $a2e[$i];
404 /* Index is $charset code point; value is its other fold-pair equivalent
405 * (A => a; a => A, etc) in the 0-255 range. If no such equivalent, value is
406 * the code point itself */
408 output_table(\@latin1_fold, "PL_fold_latin1");
412 # This generates the dfa table for perl extended UTF-8, which accepts
413 # surrogates, non-characters, and accepts start bytes up through FE
414 # (start byte FF has to be handled outside this dfa). The class numbers
415 # for start bytes are constrained so that they can be used as a shift
416 # count for masking off the leading one bits
420 # A0-A1 7 Not legal immediately after start bytes F0 F8 FC
422 # A2-A3 8 Not legal immediately after start bytes F0 F8 FC
423 # A4-A7 9 Not legal immediately after start bytes F0 F8
424 # A8-AF 10 Not legal immediately after start bytes F0
439 # Here's the I8 for the code points before which overlongs occur:
440 # U+4000: \xF0\xB0\xA0\xA0
441 # U+40000: \xF8\xA8\xA0\xA0\xA0
442 # U+400000: \xFC\xA4\xA0\xA0\xA0\xA0
443 # U+4000000: \xFE\xA2\xA0\xA0\xA0\xA0\xA0
445 # The first part of the table maps bytes to character classes to reduce
446 # the size of the transition table and create bitmasks.
448 # The second part is a transition table that maps a combination of a
449 # state of the automaton and a character class to a new state. The
450 # numbering of the original nodes is retained, but some have been split
451 # so that there are new nodes. They mean:
452 # N0 The initial state, and final accepting one.
453 # N1 One continuation byte (A0-BF) left. This is transitioned to
454 # immediately when the start byte indicates a two-byte sequence
455 # N2 Two continuation bytes left.
456 # N3 Three continuation bytes left.
457 # N4 Four continuation bytes left.
458 # N5 Five continuation bytes left.
459 # N6 Start byte is F0. Continuation bytes A[0-F] are illegal
460 # (overlong); the other continuations transition to N2
461 # N7 Start byte is F8. Continuation bytes A[0-7] are illegal
462 # (overlong); the other continuations transition to N3
463 # N8 Start byte is FC. Continuation bytes A[0-3] are illegal
464 # (overlong); the other continuations transition to N4
465 # N9 Start byte is FE. Continuation bytes A[01] are illegal
466 # (overlong); the other continuations transition to N5
467 # 1 Reject. All transitions not mentioned above (except the single
468 # byte ones (as they are always legal) are to this state.
470 my $NUM_CLASSES = 16;
472 my $N1 = $N0 + $NUM_CLASSES;
473 my $N2 = $N1 + $NUM_CLASSES;
474 my $N3 = $N2 + $NUM_CLASSES;
475 my $N4 = $N3 + $NUM_CLASSES;
476 my $N5 = $N4 + $NUM_CLASSES;
477 my $N6 = $N5 + $NUM_CLASSES;
478 my $N7 = $N6 + $NUM_CLASSES;
479 my $N8 = $N7 + $NUM_CLASSES;
480 my $N9 = $N8 + $NUM_CLASSES;
481 my $N10 = $N9 + $NUM_CLASSES;
483 my @perl_extended_utf8_dfa;
485 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 00-0F
486 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 10-1F
487 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 20-2F
488 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 30-3F
489 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 40-4F
490 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 50-5F
491 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 60-6F
492 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 70-7F
493 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 80-8F
494 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 90-9F
495 7, 7, 8, 8, 9, 9, 9, 9,10,10,10,10,10,10,10,10, # A0-AF
496 11,11,11,11,11,11,11,11,11,11,11,11,11,11,11,11, # B0-BF
497 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, # C0-CF
498 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, # D0-DF
499 1, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, # E0-EF
500 12, 4, 4, 4, 4, 4, 4, 4,13, 5, 5, 5,14, 6,15, 1, # F0-FF
502 $perl_extended_utf8_dfa[$i82utf[$_]] = $i8[$_] for (0 .. 255);
503 push @perl_extended_utf8_dfa, (
505 # 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
506 0, 1,$N1,$N2,$N3,$N4,$N5, 1, 1, 1, 1, 1,$N6,$N7,$N8,$N9, # N0
507 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 1, 1, 1, 1, # N1
508 1, 1, 1, 1, 1, 1, 1,$N1,$N1,$N1,$N1,$N1, 1, 1, 1, 1, # N2
509 1, 1, 1, 1, 1, 1, 1,$N2,$N2,$N2,$N2,$N2, 1, 1, 1, 1, # N3
510 1, 1, 1, 1, 1, 1, 1,$N3,$N3,$N3,$N3,$N3, 1, 1, 1, 1, # N4
511 1, 1, 1, 1, 1, 1, 1,$N4,$N4,$N4,$N4,$N4, 1, 1, 1, 1, # N5
513 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,$N2, 1, 1, 1, 1, # N6
514 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,$N3,$N3, 1, 1, 1, 1, # N7
515 1, 1, 1, 1, 1, 1, 1, 1, 1,$N4,$N4,$N4, 1, 1, 1, 1, # N8
516 1, 1, 1, 1, 1, 1, 1, 1,$N5,$N5,$N5,$N5, 1, 1, 1, 1, # N9
518 output_table(\@perl_extended_utf8_dfa, "PL_extended_utf8_dfa_tab",
523 # This generates the dfa table for strict UTF-8, which rejects
524 # surrogates, non-characters, and above Unicode.
527 # 00-9F 0 Always legal at start
528 # A0 10 Not legal immediately after start bytes F0 F8
529 # A1 11 Not legal immediately after start bytes F0 F8,
530 # A2-A7 12 Not legal immediately after start bytes F0 F8 F9
531 # A8,AA,AC 13 Not legal immediately after start bytes F0 F9
532 # A9,AB,AD 14 Not legal immediately after start byte F0
533 # AE 15 Not legal immediately after start byte F0
534 # AF 16 Not legal immediately after start bytes F0
535 # B[0248AC] 17 Not legal immediately after start byte F9
536 # B[1359D] 18 Not legal immediately after start byte F9
537 # B6 19 Not legal immediately after start byte F9
538 # B7 20 Not legal immediately after start byte F9
539 # BE 21 Not legal immediately after start byte F9
540 # BF 22 Not legal immediately after start byte F9
541 # C0-C4 1 (reject, all are overlong)
542 # C5-DF 2 Accepts any legal continuation
543 # E0 1 (reject, all are overlong)
544 # E1-EF 3 Accepts any legal continuation
545 # F0 8 (has overlongs)
546 # F1 6 (has surrogates, non-chars)
547 # F2,F4,F6 4 Accepts any legal continuation
548 # F3,F5,F7 5 (has non-chars)
549 # F8 9 (has overlongs, non-chars)
550 # F9 7 (has non-chars, non-Unicode)
551 # FA-FF 1 (reject, all are non-Unicode)
553 # Here's the I8 for enough code points so that you can figure out what's
556 # U+D800: \xF1\xB6\xA0\xA0
557 # U+DFFF: \xF1\xB7\xBF\xBF
558 # U+FDD0: \xF1\xBF\xAE\xB0
559 # U+FDEF: \xF1\xBF\xAF\xAF
560 # U+FFFE: \xF1\xBF\xBF\xBE
561 # U+1FFFE: \xF3\xBF\xBF\xBE
562 # U+2FFFE: \xF5\xBF\xBF\xBE
563 # U+3FFFE: \xF7\xBF\xBF\xBE
564 # U+4FFFE: \xF8\xA9\xBF\xBF\xBE
565 # U+5FFFE: \xF8\xAB\xBF\xBF\xBE
566 # U+6FFFE: \xF8\xAD\xBF\xBF\xBE
567 # U+7FFFE: \xF8\xAF\xBF\xBF\xBE
568 # U+8FFFE: \xF8\xB1\xBF\xBF\xBE
569 # U+9FFFE: \xF8\xB3\xBF\xBF\xBE
570 # U+AFFFE: \xF8\xB5\xBF\xBF\xBE
571 # U+BFFFE: \xF8\xB7\xBF\xBF\xBE
572 # U+CFFFE: \xF8\xB9\xBF\xBF\xBE
573 # U+DFFFE: \xF8\xBB\xBF\xBF\xBE
574 # U+EFFFE: \xF8\xBD\xBF\xBF\xBE
575 # U+FFFFE: \xF8\xBF\xBF\xBF\xBE
576 # U+10FFFE: \xF9\xA1\xBF\xBF\xBE
578 # The first part of the table maps bytes to character classes to reduce
579 # the size of the transition table and create bitmasks.
581 # The second part is a transition table that maps a combination of a
582 # state of the automaton and a character class to a new state. The
583 # numbering of the original nodes is retained, but some have been split
584 # so that there are new nodes. They mean:
585 # N0 The initial state, and final accepting one.
586 # N1 One continuation byte (A0-BF) left. This is transitioned to
587 # immediately when the start byte indicates a two-byte sequence
588 # N2 Two continuation bytes left.
589 # N3 Three continuation bytes left.
590 # N4 Start byte is F0. Continuation bytes A[0-F] are illegal
591 # (overlong); the other continuations transition to N2
592 # N5 Start byte is F1. Continuation bytes B6 and B7 are illegal
593 # (surrogates); BF transitions to N9; the other continuations to
595 # N6 Start byte is F[357]. Continuation byte BF transitions to N12;
596 # other continuations to N2
597 # N5 Start byte is F8. Continuation bytes A[0-7] are illegal
598 # (overlong); continuations A[9BDF] and B[13579BDF] transition to
599 # N14; the other continuations to N3
600 # N8 Start byte is F9. Continuation byte A0 transitions to N3; A1
601 # to N14; the other continuation bytes are illegal.
602 # N9 Initial sequence is F1 BF. Continuation byte AE transitions to
603 # state N10; AF to N11; BF to N13; the other continuations to N1.
604 # N10 Initial sequence is F1 BF AE. Continuation bytes B0-BF are
605 # illegal (non-chars); the other continuations are legal
606 # N11 Initial sequence is F1 BF AF. Continuation bytes A0-AF are
607 # illegal (non-chars); the other continuations are legal
608 # N12 Initial sequence is F[357] BF. Continuation bytes BF
609 # transitions to N13; the other continuations to N1
610 # N13 Initial sequence is F[1357] BF BF or F8 x BF (where x is
611 # something that can lead to a non-char. Continuation bytes BE
612 # and BF are illegal (non-chars); the other continuations are
614 # N14 Initial sequence is F8 A[9BDF]; or F8 B[13579BDF]; or F9 A1.
615 # Continuation byte BF transitions to N13; the other
616 # continuations to N2
617 # 1 Reject. All transitions not mentioned above (except the single
618 # byte ones (as they are always legal) are to this state.
620 my $NUM_CLASSES = 23;
622 my $N1 = $N0 + $NUM_CLASSES;
623 my $N2 = $N1 + $NUM_CLASSES;
624 my $N3 = $N2 + $NUM_CLASSES;
625 my $N4 = $N3 + $NUM_CLASSES;
626 my $N5 = $N4 + $NUM_CLASSES;
627 my $N6 = $N5 + $NUM_CLASSES;
628 my $N7 = $N6 + $NUM_CLASSES;
629 my $N8 = $N7 + $NUM_CLASSES;
630 my $N9 = $N8 + $NUM_CLASSES;
631 my $N10 = $N9 + $NUM_CLASSES;
632 my $N11 = $N10 + $NUM_CLASSES;
633 my $N12 = $N11 + $NUM_CLASSES;
634 my $N13 = $N12 + $NUM_CLASSES;
635 my $N14 = $N13 + $NUM_CLASSES;
639 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 00-0F
640 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 10-1F
641 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 20-2F
642 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 30-3F
643 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 40-4F
644 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 50-5F
645 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 60-6F
646 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 70-7F
647 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 80-8F
648 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 90-9F
649 10,11,12,12,12,12,12,12,13,14,13,14,13,14,15,16, # A0-AF
650 17,18,17,18,17,18,19,20,17,18,17,18,17,18,21,22, # B0-BF
651 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, # C0-CF
652 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, # D0-DF
653 1, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, # E0-EF
654 8, 6, 4, 5, 4, 5, 4, 5, 9, 7, 1, 1, 1, 1, 1, 1, # F0-FF
656 $strict_utf8_dfa[$i82utf[$_]] = $i8[$_] for (0 .. 255);
657 push @strict_utf8_dfa, (
659 # 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
660 0,1,$N1,$N2,$N3,$N6,$N5,$N8,$N4,$N7, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, # N0
661 1,1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # N1
662 1,1, 1, 1, 1, 1, 1, 1, 1, 1, $N1, $N1, $N1, $N1, $N1, $N1, $N1, $N1, $N1, $N1, $N1, $N1, $N1, # N2
663 1,1, 1, 1, 1, 1, 1, 1, 1, 1, $N2, $N2, $N2, $N2, $N2, $N2, $N2, $N2, $N2, $N2, $N2, $N2, $N2, # N3
665 1,1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, $N2, $N2, $N2, $N2, $N2, $N2, # N4
666 1,1, 1, 1, 1, 1, 1, 1, 1, 1, $N2, $N2, $N2, $N2, $N2, $N2, $N2, $N2, $N2, 1, 1, $N2, $N9, # N5
667 1,1, 1, 1, 1, 1, 1, 1, 1, 1, $N2, $N2, $N2, $N2, $N2, $N2, $N2, $N2, $N2, $N2, $N2, $N2,$N12, # N6
668 1,1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, $N3,$N14, $N3,$N14, $N3,$N14, $N3,$N14, $N3,$N14, # N7
669 1,1, 1, 1, 1, 1, 1, 1, 1, 1, $N3,$N14, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, # N8
670 1,1, 1, 1, 1, 1, 1, 1, 1, 1, $N1, $N1, $N1, $N1, $N1,$N10,$N11, $N1, $N1, $N1, $N1, $N1,$N13, # N9
671 1,1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, # N10
672 1,1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, # N11
673 1,1, 1, 1, 1, 1, 1, 1, 1, 1, $N1, $N1, $N1, $N1, $N1, $N1, $N1, $N1, $N1, $N1, $N1, $N1,$N13, # N12
674 1,1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, # N13
675 1,1, 1, 1, 1, 1, 1, 1, 1, 1, $N2, $N2, $N2, $N2, $N2, $N2, $N2, $N2, $N2, $N2, $N2, $N2,$N13, # N14
677 output_table(\@strict_utf8_dfa, "PL_strict_utf8_dfa_tab", $NUM_CLASSES);
681 # This generates the dfa table for C9 strict UTF-8, which rejects
682 # surrogates and above Unicode, but allows non-characters,.
685 # 00-9F 0 Always legal at start
686 # A0-A1 9 Not legal immediately after start bytes F0 F8
687 # A2-A7 10 Not legal immediately after start bytes F0 F8 F9
688 # A8-AF 11 Not legal immediately after start bytes F0 F9
689 # B0-B5,B8-BF 12 Not legal immediately after start byte F9
691 # C0-C4 1 (reject, all are overlong)
692 # C5-DF 2 Accepts any legal continuation
693 # E0 1 (reject, all are overlong)
694 # E1-EF 3 Accepts any legal continuation
695 # F0 6 (has overlongs)
696 # F1 5 (has surrogates)
697 # F2-F7 4 Accepts any legal continuation
698 # F8 8 (has overlongs)
699 # F9 7 (has non-Unicode)
700 # FA-FF 1 (reject, all are non-Unicode)
702 # The first part of the table maps bytes to character classes to reduce
703 # the size of the transition table and create bitmasks.
705 # The second part is a transition table that maps a combination of a
706 # state of the automaton and a character class to a new state. The
707 # numbering of the original nodes is retained, but some have been split
708 # so that there are new nodes. They mean:
709 # N0 The initial state, and final accepting one.
710 # N1 One continuation byte (A0-BF) left. This is transitioned to
711 # immediately when the start byte indicates a two-byte sequence
712 # N2 Two continuation bytes left.
713 # N3 Three continuation bytes left.
714 # N4 Start byte is F0. Continuation bytes A[0-F] are illegal
715 # (overlong); the other continuations transition to N2
716 # N5 Start byte is F1. B6 and B7 are illegal (surrogates); the
717 # other continuations transition to N2
718 # N6 Start byte is F8. Continuation bytes A[0-7] are illegal
719 # (overlong); the other continuations transition to N3
720 # N7 Start byte is F9. Continuation bytes A0 and A1 transition to
721 # N3; the other continuation bytes are illegal (non-Unicode)
722 # 1 Reject. All transitions not mentioned above (except the single
723 # byte ones (as they are always legal) are to this state.
725 my $NUM_CLASSES = 14;
727 my $N1 = $N0 + $NUM_CLASSES;
728 my $N2 = $N1 + $NUM_CLASSES;
729 my $N3 = $N2 + $NUM_CLASSES;
730 my $N4 = $N3 + $NUM_CLASSES;
731 my $N5 = $N4 + $NUM_CLASSES;
732 my $N6 = $N5 + $NUM_CLASSES;
733 my $N7 = $N6 + $NUM_CLASSES;
737 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 00-0F
738 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 10-1F
739 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 20-2F
740 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 30-3F
741 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 40-4F
742 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 50-5F
743 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 60-6F
744 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 70-7F
745 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 80-8F
746 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, # 90-9F
747 9, 9,10,10,10,10,10,10,11,11,11,11,11,11,11,11, # A0-AF
748 12,12,12,12,12,12,13,13,12,12,12,12,12,12,12,12, # B0-BF
749 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, # C0-CF
750 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, # D0-DF
751 1, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, # E0-EF
752 6, 5, 4, 4, 4, 4, 4, 4, 8, 7, 1, 1, 1, 1, 1, 1, # F0-FF
754 $C9_utf8_dfa[$i82utf[$_]] = $i8[$_] for (0 .. 255);
757 # 0 1 2 3 4 5 6 7 8 9 10 11 12 13
758 0,1,$N1,$N2,$N3,$N5,$N4,$N7,$N6, 1, 1, 1, 1, 1, # N0
759 1,1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, # N1
760 1,1, 1, 1, 1, 1, 1, 1, 1,$N1, $N1, $N1, $N1, $N1, # N2
761 1,1, 1, 1, 1, 1, 1, 1, 1,$N2, $N2, $N2, $N2, $N2, # N3
763 1,1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, $N2, $N2, # N4
764 1,1, 1, 1, 1, 1, 1, 1, 1,$N2, $N2, $N2, $N2, 1, # N5
765 1,1, 1, 1, 1, 1, 1, 1, 1, 1, 1, $N3, $N3, $N3, # N6
766 1,1, 1, 1, 1, 1, 1, 1, 1,$N3, 1, 1, 1, 1, # N7
768 output_table(\@C9_utf8_dfa, "PL_c9_utf8_dfa_tab", $NUM_CLASSES);
773 /* This table partitions all the code points of the platform into ranges which
774 * have the property that all the code points in each range have the same
775 * number of bytes in their UTF-EBCDIC representations, and the adjacent
776 * ranges have a different number of bytes.
778 * Each number in the table begins such a range, which extends up to just
779 * before the following table entry, except the final entry is understood to
780 * extend to the platform's infinity
783 # The lengths of the characters between 0 and 255 are either 1 or 2,
784 # with those whose ASCII platform equivalents below 160 being 1, and
788 my $pushed_range_is_length_1 = 1;
790 for my $i (1 .. 0xFF) {
791 my $this_code_point_is_length_1 = ($e2a[$i] < 160);
792 if ($pushed_range_is_length_1 != $this_code_point_is_length_1) {
794 $pushed_range_is_length_1 = $this_code_point_is_length_1;
798 # Starting at 256, the length is 2.
799 push @list, 0x100 if $pushed_range_is_length_1;
801 # These are based on the fundamental properties of UTF-EBCDIC. Each
802 # continuation byte has 5 bits of information. Comments in utf8.h
804 my $UTF_ACCUMULATION_SHIFT = 5;
805 push @list, (32 * (1 << ( $UTF_ACCUMULATION_SHIFT)));
806 push @list, (16 * (1 << (2 * $UTF_ACCUMULATION_SHIFT)));
807 push @list, ( 8 * (1 << (3 * $UTF_ACCUMULATION_SHIFT)));
808 push @list, ( 4 * (1 << (4 * $UTF_ACCUMULATION_SHIFT)));
809 push @list, ( 2 * (1 << (5 * $UTF_ACCUMULATION_SHIFT)));
810 push @list, ( (1 << (6 * $UTF_ACCUMULATION_SHIFT)));
812 output_table_start($out_fh, "UV", "PL_partition_by_byte_length", scalar @list);
815 print $out_fh join ",\n\t", map { sprintf "0x%02x", $_ } @list;
818 output_table_end($out_fh);
821 print $out_fh get_conditional_compile_line_end();
824 print $out_fh "\n#endif /* PERL_EBCDIC_TABLES_H_ */\n";
826 read_only_bottom_close_and_rename($out_fh);