| 1 | #!perl -w |
| 2 | use 5.015; |
| 3 | use strict; |
| 4 | use warnings; |
| 5 | use Unicode::UCD qw(prop_aliases |
| 6 | prop_values |
| 7 | prop_value_aliases |
| 8 | prop_invlist |
| 9 | prop_invmap search_invlist |
| 10 | charprop |
| 11 | num |
| 12 | charblock |
| 13 | ); |
| 14 | use constant DEBUG => $ENV{DEBUG} // 0; |
| 15 | require './regen/regen_lib.pl'; |
| 16 | require './regen/charset_translations.pl'; |
| 17 | require './lib/unicore/UCD.pl'; |
| 18 | require './regen/mph.pl'; |
| 19 | use re "/aa"; |
| 20 | |
| 21 | print "Starting...\n" if DEBUG; |
| 22 | |
| 23 | # This program outputs charclass_invlists.h, which contains various inversion |
| 24 | # lists in the form of C arrays that are to be used as-is for inversion lists. |
| 25 | # Thus, the lists it contains are essentially pre-compiled, and need only a |
| 26 | # light-weight fast wrapper to make them usable at run-time. |
| 27 | |
| 28 | # As such, this code knows about the internal structure of these lists, and |
| 29 | # any change made to that has to be done here as well. A random number stored |
| 30 | # in the headers is used to minimize the possibility of things getting |
| 31 | # out-of-sync, or the wrong data structure being passed. Currently that |
| 32 | # random number is: |
| 33 | |
| 34 | my $VERSION_DATA_STRUCTURE_TYPE = 148565664; |
| 35 | |
| 36 | # charclass_invlists.h now also contains inversion maps and enum definitions |
| 37 | # for those maps that have a finite number of possible values |
| 38 | |
| 39 | # integer or float (no exponent) |
| 40 | my $integer_or_float_re = qr/ ^ -? \d+ (:? \. \d+ )? $ /x; |
| 41 | |
| 42 | # Also includes rationals |
| 43 | my $numeric_re = qr! $integer_or_float_re | ^ -? \d+ / \d+ $ !x; |
| 44 | |
| 45 | # More than one code point may have the same code point as their fold. This |
| 46 | # gives the maximum number in the current Unicode release. (The folded-to |
| 47 | # code point is not included in this count.) Most folds are pairs of code |
| 48 | # points, like 'B' and 'b', so this number is at least one. |
| 49 | my $max_fold_froms = 1; |
| 50 | |
| 51 | my %keywords; |
| 52 | my $table_name_prefix = "UNI_"; |
| 53 | |
| 54 | # Matches valid C language enum names: begins with ASCII alphabetic, then any |
| 55 | # ASCII \w |
| 56 | my $enum_name_re = qr / ^ [[:alpha:]] \w* $ /ax; |
| 57 | |
| 58 | my $out_fh = open_new('charclass_invlists.h', '>', |
| 59 | {style => '*', by => 'regen/mk_invlists.pl', |
| 60 | from => "Unicode::UCD"}); |
| 61 | |
| 62 | my $in_file_pound_if = ""; |
| 63 | |
| 64 | my $max_hdr_len = 3; # In headings, how wide a name is allowed? |
| 65 | |
| 66 | print $out_fh "/* See the generating file for comments */\n\n"; |
| 67 | |
| 68 | print $out_fh <<'EOF'; |
| 69 | /* This gives the number of code points that can be in the bitmap of an ANYOF |
| 70 | * node. The shift number must currently be one of: 8..12. It can't be less |
| 71 | * than 8 (256) because some code relies on it being at least that. Above 12 |
| 72 | * (4096), and you start running into warnings that some data structure widths |
| 73 | * have been exceeded, though the test suite as of this writing still passes |
| 74 | * for up through 16, which is as high as anyone would ever want to go, |
| 75 | * encompassing all of the Unicode BMP, and thus including all the economically |
| 76 | * important world scripts. At 12 most of them are: including Arabic, |
| 77 | * Cyrillic, Greek, Hebrew, Indian subcontinent, Latin, and Thai; but not Han, |
| 78 | * Japanese, nor Korean. The regnode sizing data structure in regnodes.h currently |
| 79 | * uses a U8, and the trie types TRIEC and AHOCORASICKC are larger than U8 for |
| 80 | * shift values above 12.) Be sure to benchmark before changing, as larger sizes |
| 81 | * do significantly slow down the test suite. */ |
| 82 | |
| 83 | EOF |
| 84 | |
| 85 | my $num_anyof_code_points = '(1 << 8)'; |
| 86 | |
| 87 | print $out_fh "#define NUM_ANYOF_CODE_POINTS $num_anyof_code_points\n\n"; |
| 88 | |
| 89 | $num_anyof_code_points = eval $num_anyof_code_points; |
| 90 | |
| 91 | no warnings 'once'; |
| 92 | print $out_fh <<"EOF"; |
| 93 | /* The precision to use in "%.*e" formats */ |
| 94 | #define PL_E_FORMAT_PRECISION $Unicode::UCD::e_precision |
| 95 | EOF |
| 96 | |
| 97 | # enums that should be made public |
| 98 | my %public_enums = ( |
| 99 | _Perl_SCX => 1 |
| 100 | ); |
| 101 | |
| 102 | # The symbols generated by this program are all currently defined only in a |
| 103 | # single dot c each. The code knows where most of them go, but this hash |
| 104 | # gives overrides for the exceptions to the typical place |
| 105 | my %exceptions_to_where_to_define = |
| 106 | ( |
| 107 | #_Perl_IVCF => 'PERL_IN_REGCOMP_C', |
| 108 | ); |
| 109 | |
| 110 | my %where_to_define_enums = (); |
| 111 | |
| 112 | my $applies_to_all_charsets_text = "all charsets"; |
| 113 | |
| 114 | my %gcb_enums; |
| 115 | my @gcb_short_enums; |
| 116 | my %gcb_abbreviations; |
| 117 | my %lb_enums; |
| 118 | my @lb_short_enums; |
| 119 | my %lb_abbreviations; |
| 120 | my %wb_enums; |
| 121 | my @wb_short_enums; |
| 122 | my %wb_abbreviations; |
| 123 | |
| 124 | my @a2n; |
| 125 | |
| 126 | my %prop_name_aliases; |
| 127 | # Invert this hash so that for each canonical name, we get a list of things |
| 128 | # that map to it (excluding itself) |
| 129 | foreach my $name (sort keys %Unicode::UCD::loose_property_name_of) { |
| 130 | my $canonical = $Unicode::UCD::loose_property_name_of{$name}; |
| 131 | push @{$prop_name_aliases{$canonical}}, $name if $canonical ne $name; |
| 132 | } |
| 133 | |
| 134 | # Output these tables in the same vicinity as each other, so that will get |
| 135 | # paged in at about the same time. These are also assumed to be the exact |
| 136 | # same list as those properties used internally by perl. |
| 137 | my %keep_together = ( |
| 138 | assigned => 1, |
| 139 | ascii => 1, |
| 140 | upper => 1, |
| 141 | lower => 1, |
| 142 | title => 1, |
| 143 | cased => 1, |
| 144 | uppercaseletter => 1, |
| 145 | lowercaseletter => 1, |
| 146 | titlecaseletter => 1, |
| 147 | casedletter => 1, |
| 148 | vertspace => 1, |
| 149 | xposixalnum => 1, |
| 150 | xposixalpha => 1, |
| 151 | xposixblank => 1, |
| 152 | xposixcntrl => 1, |
| 153 | xposixdigit => 1, |
| 154 | xposixgraph => 1, |
| 155 | xposixlower => 1, |
| 156 | xposixprint => 1, |
| 157 | xposixpunct => 1, |
| 158 | xposixspace => 1, |
| 159 | xposixupper => 1, |
| 160 | xposixword => 1, |
| 161 | xposixxdigit => 1, |
| 162 | posixalnum => 1, |
| 163 | posixalpha => 1, |
| 164 | posixblank => 1, |
| 165 | posixcntrl => 1, |
| 166 | posixdigit => 1, |
| 167 | posixgraph => 1, |
| 168 | posixlower => 1, |
| 169 | posixprint => 1, |
| 170 | posixpunct => 1, |
| 171 | posixspace => 1, |
| 172 | posixupper => 1, |
| 173 | posixword => 1, |
| 174 | posixxdigit => 1, |
| 175 | _perl_any_folds => 1, |
| 176 | _perl_folds_to_multi_char => 1, |
| 177 | _perl_is_in_multi_char_fold => 1, |
| 178 | _perl_non_final_folds => 1, |
| 179 | _perl_idstart => 1, |
| 180 | _perl_idcont => 1, |
| 181 | _perl_charname_begin => 1, |
| 182 | _perl_charname_continue => 1, |
| 183 | _perl_problematic_locale_foldeds_start => 1, |
| 184 | _perl_problematic_locale_folds => 1, |
| 185 | _perl_quotemeta => 1, |
| 186 | ); |
| 187 | my %perl_tags; # So can find synonyms of the above properties |
| 188 | |
| 189 | my $unused_table_hdr = 'u'; # Heading for row or column for unused values |
| 190 | |
| 191 | sub uniques { |
| 192 | # Returns non-duplicated input values. From "Perl Best Practices: |
| 193 | # Encapsulated Cleverness". p. 455 in first edition. |
| 194 | |
| 195 | my %seen; |
| 196 | return grep { ! $seen{$_}++ } @_; |
| 197 | } |
| 198 | |
| 199 | sub caselessly { lc $a cmp lc $b } |
| 200 | |
| 201 | sub a2n($) { |
| 202 | my $cp = shift; |
| 203 | |
| 204 | # Returns the input Unicode code point translated to native. |
| 205 | |
| 206 | return $cp if $cp !~ $integer_or_float_re || $cp > 255; |
| 207 | return $a2n[$cp]; |
| 208 | } |
| 209 | |
| 210 | sub end_file_pound_if { |
| 211 | if ($in_file_pound_if) { |
| 212 | print $out_fh "\n#endif\t/* $in_file_pound_if */\n"; |
| 213 | $in_file_pound_if = ""; |
| 214 | } |
| 215 | } |
| 216 | |
| 217 | sub end_charset_pound_if { |
| 218 | print $out_fh "\n" . get_conditional_compile_line_end(); |
| 219 | } |
| 220 | |
| 221 | sub switch_pound_if ($$;$) { |
| 222 | my $name = shift; |
| 223 | my $new_pound_if = shift; |
| 224 | my $charset = shift; |
| 225 | |
| 226 | my @new_pound_if = ref ($new_pound_if) |
| 227 | ? sort @$new_pound_if |
| 228 | : $new_pound_if; |
| 229 | |
| 230 | # Switch to new #if given by the 2nd argument. If there is an override |
| 231 | # for this, it instead switches to that. The 1st argument is the |
| 232 | # static's name, used only to check if there is an override for this |
| 233 | # |
| 234 | # The 'charset' parmameter, if present, is used to first end the charset |
| 235 | # #if if we actually do a switch, and then restart it afterwards. This |
| 236 | # code, then assumes that the charset #if's are enclosed in the file ones. |
| 237 | |
| 238 | if (exists $exceptions_to_where_to_define{$name}) { |
| 239 | @new_pound_if = $exceptions_to_where_to_define{$name}; |
| 240 | } |
| 241 | |
| 242 | foreach my $element (@new_pound_if) { |
| 243 | |
| 244 | # regcomp.c is arranged so that the tables are not compiled in |
| 245 | # re_comp.c, but general enums and defines (which take no space) are |
| 246 | # compiled */ |
| 247 | my $no_xsub = 1 if $name !~ /enum|define/ |
| 248 | && $element =~ / PERL_IN_ (?: REGCOMP ) _C /x; |
| 249 | $element = "defined($element)"; |
| 250 | $element = "($element && ! defined(PERL_IN_XSUB_RE))" if $no_xsub; |
| 251 | } |
| 252 | $new_pound_if = join " || ", @new_pound_if; |
| 253 | |
| 254 | # Change to the new one if different from old |
| 255 | if ($in_file_pound_if ne $new_pound_if) { |
| 256 | |
| 257 | end_charset_pound_if() if defined $charset; |
| 258 | |
| 259 | # Exit any current #if |
| 260 | if ($in_file_pound_if) { |
| 261 | end_file_pound_if; |
| 262 | } |
| 263 | |
| 264 | $in_file_pound_if = $new_pound_if; |
| 265 | print $out_fh "\n#if $in_file_pound_if\n"; |
| 266 | |
| 267 | start_charset_pound_if ($charset, 1) if defined $charset; |
| 268 | } |
| 269 | } |
| 270 | |
| 271 | sub start_charset_pound_if ($;$) { |
| 272 | print $out_fh "\n" . get_conditional_compile_line_start(shift, shift); |
| 273 | } |
| 274 | |
| 275 | { # Closure |
| 276 | my $fh; |
| 277 | my $in_doinit = 0; |
| 278 | |
| 279 | sub output_table_header($$$;$@) { |
| 280 | |
| 281 | # Output to $fh the heading for a table given by the other inputs |
| 282 | |
| 283 | $fh = shift; |
| 284 | my ($type, # typedef of table, like UV, UV* |
| 285 | $name, # name of table |
| 286 | $comment, # Optional comment to put on header line |
| 287 | @sizes # Optional sizes of each array index. If omitted, |
| 288 | # there is a single index whose size is computed by |
| 289 | # the C compiler. |
| 290 | ) = @_; |
| 291 | |
| 292 | $type =~ s/ \s+ $ //x; |
| 293 | |
| 294 | # If a the typedef is a ptr, add in an extra const |
| 295 | $type .= " const" if $type =~ / \* $ /x; |
| 296 | |
| 297 | $comment = "" unless defined $comment; |
| 298 | $comment = " /* $comment */" if $comment; |
| 299 | |
| 300 | my $array_declaration; |
| 301 | if (@sizes) { |
| 302 | $array_declaration = ""; |
| 303 | $array_declaration .= "[$_]" for @sizes; |
| 304 | } |
| 305 | else { |
| 306 | $array_declaration = '[]'; |
| 307 | } |
| 308 | |
| 309 | my $declaration = "$type ${name}$array_declaration"; |
| 310 | |
| 311 | # Things not matching this are static. Otherwise, it is an external |
| 312 | # constant, initialized only under DOINIT. |
| 313 | # |
| 314 | # (Currently everything is static) |
| 315 | if ($in_file_pound_if !~ / PERL_IN_ (?: ) _C /x) { |
| 316 | $in_doinit = 0; |
| 317 | print $fh "\nstatic const $declaration = {$comment\n"; |
| 318 | } |
| 319 | else { |
| 320 | $in_doinit = 1; |
| 321 | print $fh <<EOF; |
| 322 | |
| 323 | # ifndef DOINIT |
| 324 | |
| 325 | EXTCONST $declaration; |
| 326 | |
| 327 | # else |
| 328 | |
| 329 | EXTCONST $declaration = {$comment |
| 330 | EOF |
| 331 | } |
| 332 | } |
| 333 | |
| 334 | sub output_table_trailer() { |
| 335 | |
| 336 | # Close out a table started by output_table_header() |
| 337 | |
| 338 | print $fh "};\n"; |
| 339 | if ($in_doinit) { |
| 340 | print $fh "\n# endif /* DOINIT */\n\n"; |
| 341 | $in_doinit = 0; |
| 342 | } |
| 343 | } |
| 344 | } # End closure |
| 345 | |
| 346 | |
| 347 | sub output_invlist ($$;$) { |
| 348 | my $name = shift; |
| 349 | my $invlist = shift; # Reference to inversion list array |
| 350 | my $charset = shift // ""; # name of character set for comment |
| 351 | |
| 352 | print " output_invlist($name) $charset\n" if DEBUG; |
| 353 | |
| 354 | die "No inversion list for $name" unless defined $invlist |
| 355 | && ref $invlist eq 'ARRAY'; |
| 356 | |
| 357 | # Output the inversion list $invlist using the name $name for it. |
| 358 | # It is output in the exact internal form for inversion lists. |
| 359 | |
| 360 | # Is the last element of the header 0, or 1 ? |
| 361 | my $zero_or_one = 0; |
| 362 | if (@$invlist && $invlist->[0] != 0) { |
| 363 | unshift @$invlist, 0; |
| 364 | $zero_or_one = 1; |
| 365 | } |
| 366 | |
| 367 | $charset = "for $charset" if $charset; |
| 368 | output_table_header($out_fh, "UV", "${name}_invlist", $charset); |
| 369 | |
| 370 | my $count = @$invlist; |
| 371 | print $out_fh <<EOF; |
| 372 | \t$count,\t/* Number of elements */ |
| 373 | \t$VERSION_DATA_STRUCTURE_TYPE, /* Version and data structure type */ |
| 374 | \t$zero_or_one,\t/* 0 if the list starts at 0; |
| 375 | \t\t 1 if it starts at the element beyond 0 */ |
| 376 | EOF |
| 377 | |
| 378 | # The main body are the UVs passed in to this routine. Do the final |
| 379 | # element separately |
| 380 | for my $i (0 .. @$invlist - 1) { |
| 381 | printf $out_fh "\t0x%X", $invlist->[$i]; |
| 382 | print $out_fh "," if $i < @$invlist - 1; |
| 383 | print $out_fh "\n"; |
| 384 | } |
| 385 | |
| 386 | output_table_trailer(); |
| 387 | } |
| 388 | |
| 389 | sub output_invmap ($$$$$$$) { |
| 390 | my $name = shift; |
| 391 | my $invmap = shift; # Reference to inversion map array |
| 392 | my $prop_name = shift; |
| 393 | my $input_format = shift; # The inversion map's format |
| 394 | my $default = shift; # The property value for code points who |
| 395 | # otherwise don't have a value specified. |
| 396 | my $extra_enums = shift; # comma-separated list of our additions to the |
| 397 | # property's standard possible values |
| 398 | my $charset = shift // ""; # name of character set for comment |
| 399 | |
| 400 | print " output_invmap($name,$prop_name) $charset\n" if DEBUG; |
| 401 | |
| 402 | # Output the inversion map $invmap for property $prop_name, but use $name |
| 403 | # as the actual data structure's name. |
| 404 | |
| 405 | my $count = @$invmap; |
| 406 | |
| 407 | my $output_format; |
| 408 | my $invmap_declaration_type; |
| 409 | my $enum_declaration_type; |
| 410 | my $aux_declaration_type; |
| 411 | my %enums; |
| 412 | my $name_prefix; |
| 413 | |
| 414 | if ($input_format =~ / ^ [as] l? $ /x) { |
| 415 | $prop_name = (prop_aliases($prop_name))[1] |
| 416 | // $prop_name =~ s/^_Perl_//r; # Get full name |
| 417 | my $short_name = (prop_aliases($prop_name))[0] // $prop_name; |
| 418 | my @input_enums; |
| 419 | |
| 420 | # Find all the possible input values. These become the enum names |
| 421 | # that comprise the inversion map. For inputs that don't have sub |
| 422 | # lists, we can just get the unique values. Otherwise, we have to |
| 423 | # expand the sublists first. |
| 424 | if ($input_format !~ / ^ a /x) { |
| 425 | if ($input_format ne 'sl') { |
| 426 | @input_enums = sort caselessly uniques(@$invmap); |
| 427 | } |
| 428 | else { |
| 429 | foreach my $element (@$invmap) { |
| 430 | if (ref $element) { |
| 431 | push @input_enums, @$element; |
| 432 | } |
| 433 | else { |
| 434 | push @input_enums, $element; |
| 435 | } |
| 436 | } |
| 437 | @input_enums = sort caselessly uniques(@input_enums); |
| 438 | } |
| 439 | } |
| 440 | |
| 441 | # The internal enums come last, and in the order specified. |
| 442 | # |
| 443 | # The internal one named EDGE is also used a marker. Any ones that |
| 444 | # come after it are used in the algorithms below, and so must be |
| 445 | # defined, even if the release of Unicode this is being compiled for |
| 446 | # doesn't use them. But since no code points are assigned to them in |
| 447 | # such a release, those values will never be accessed. We collapse |
| 448 | # all of them into a single placholder row and a column. The |
| 449 | # algorithms below will fill in those cells with essentially garbage, |
| 450 | # but they are never read, so it doesn't matter. This allows the |
| 451 | # algorithm to remain the same from release to release. |
| 452 | # |
| 453 | # In one case, regexec.c also uses a placeholder which must be defined |
| 454 | # here, and we put it in the unused row and column as its value is |
| 455 | # never read. |
| 456 | # |
| 457 | my @enums = @input_enums; |
| 458 | my @extras; |
| 459 | my @unused_enums; |
| 460 | my $unused_enum_value = @enums; |
| 461 | if ($extra_enums ne "") { |
| 462 | @extras = split /,/, $extra_enums; |
| 463 | my $seen_EDGE = 0; |
| 464 | |
| 465 | # Don't add if already there. |
| 466 | foreach my $this_extra (@extras) { |
| 467 | next if grep { $_ eq $this_extra } @enums; |
| 468 | if ($this_extra eq 'EDGE') { |
| 469 | push @enums, $this_extra; |
| 470 | $seen_EDGE = 1; |
| 471 | } |
| 472 | elsif ($seen_EDGE) { |
| 473 | push @unused_enums, $this_extra; |
| 474 | } |
| 475 | else { |
| 476 | push @enums, $this_extra; |
| 477 | } |
| 478 | } |
| 479 | |
| 480 | @unused_enums = sort caselessly @unused_enums; |
| 481 | $unused_enum_value = @enums; # All unused have the same value, |
| 482 | # one beyond the final used one |
| 483 | } |
| 484 | |
| 485 | # These properties have extra tables written out for them that we want |
| 486 | # to make as compact and legible as possible. So we find short names |
| 487 | # for their property values. For non-official ones we will need to |
| 488 | # add a legend at the top of the table to say what the abbreviation |
| 489 | # stands for. |
| 490 | my $property_needs_table_re = qr/ ^ _Perl_ (?: GCB | LB | WB ) $ /x; |
| 491 | |
| 492 | my %short_enum_name; |
| 493 | my %need_explanation; # For non-official abbreviations, we will need |
| 494 | # to explain what the one we come up with |
| 495 | # stands for |
| 496 | my $type = lc $prop_name; |
| 497 | if ($name =~ $property_needs_table_re) { |
| 498 | my @short_names; # List of already used abbreviations, so we |
| 499 | # don't duplicate |
| 500 | for my $enum (@enums) { |
| 501 | my $short_enum; |
| 502 | my $is_official_name = 0; |
| 503 | |
| 504 | # Special case this wb property value to make the |
| 505 | # name more clear |
| 506 | if ($enum eq 'Perl_Tailored_HSpace') { |
| 507 | $short_enum = 'hs'; |
| 508 | } |
| 509 | else { |
| 510 | |
| 511 | # Use the official short name, if found. |
| 512 | ($short_enum) = prop_value_aliases($type, $enum); |
| 513 | if ( defined $short_enum) { |
| 514 | $is_official_name = 1; |
| 515 | } |
| 516 | else { |
| 517 | # But if there is no official name, use the name that |
| 518 | # came from the data (if any). Otherwise, the name |
| 519 | # had to come from the extras list. There are two |
| 520 | # types of values in that list. |
| 521 | # |
| 522 | # First are those enums that are not part of the |
| 523 | # property, but are defined by the code in this file. |
| 524 | # By convention these have all-caps names. We use the |
| 525 | # lowercased name for these. |
| 526 | # |
| 527 | # Second are enums that are needed to get the |
| 528 | # algorithms below to work and/or to get regexec.c to |
| 529 | # compile, but don't exist in all Unicode releases. |
| 530 | # These are handled outside this loop as |
| 531 | # 'unused_enums' (as they are unused they all get |
| 532 | # collapsed into a single column, and their names |
| 533 | # don't matter) |
| 534 | if (grep { $_ eq $enum } @input_enums) { |
| 535 | $short_enum = $enum |
| 536 | } |
| 537 | else { |
| 538 | $short_enum = lc $enum; |
| 539 | } |
| 540 | } |
| 541 | |
| 542 | # If our short name is too long, or we already know that |
| 543 | # the name is an abbreviation, truncate to make sure it's |
| 544 | # short enough, and remember that we did this so we can |
| 545 | # later add a comment in the generated file |
| 546 | if (length $short_enum > $max_hdr_len) { |
| 547 | # First try using just the uppercase letters of the name; |
| 548 | # if it is something like FooBar, FB is a better |
| 549 | # abbreviation than Foo. That's not the case if it is |
| 550 | # entirely lowercase. |
| 551 | my $uc = $short_enum; |
| 552 | $uc =~ s/[[:^upper:]]//g; |
| 553 | $short_enum = $uc if length $uc > 1 |
| 554 | && length $uc < length $short_enum; |
| 555 | |
| 556 | $short_enum = substr($short_enum, 0, $max_hdr_len); |
| 557 | $is_official_name = 0; |
| 558 | } |
| 559 | } |
| 560 | |
| 561 | # If the name we are to display conflicts, try another. |
| 562 | if (grep { $_ eq $short_enum } @short_names) { |
| 563 | $is_official_name = 0; |
| 564 | do { # The increment operator on strings doesn't work on |
| 565 | # those containing an '_', so get rid of any final |
| 566 | # portion. |
| 567 | $short_enum =~ s/_//g; |
| 568 | $short_enum++; |
| 569 | } while grep { $_ eq $short_enum } @short_names; |
| 570 | } |
| 571 | |
| 572 | push @short_names, $short_enum; |
| 573 | $short_enum_name{$enum} = $short_enum; |
| 574 | $need_explanation{$enum} = $short_enum unless $is_official_name; |
| 575 | } |
| 576 | } # End of calculating short enum names for certain properties |
| 577 | |
| 578 | # Assign a value to each element of the enum type we are creating. |
| 579 | # The default value always gets 0; the others are arbitrarily |
| 580 | # assigned, but for the properties which have the extra table, it is |
| 581 | # in the order we have computed above so the rows and columns appear |
| 582 | # alphabetically by heading abbreviation. |
| 583 | my $enum_val = 0; |
| 584 | my $canonical_default = prop_value_aliases($prop_name, $default); |
| 585 | $default = $canonical_default if defined $canonical_default; |
| 586 | $enums{$default} = $enum_val++; |
| 587 | |
| 588 | for my $enum (sort { ($name =~ $property_needs_table_re) |
| 589 | ? lc $short_enum_name{$a} |
| 590 | cmp lc $short_enum_name{$b} |
| 591 | : lc $a cmp lc $b |
| 592 | } @enums) |
| 593 | { |
| 594 | $enums{$enum} = $enum_val++ unless exists $enums{$enum}; |
| 595 | } |
| 596 | |
| 597 | # Now calculate the data for the special tables output for these |
| 598 | # properties. |
| 599 | if ($name =~ $property_needs_table_re) { |
| 600 | |
| 601 | # The data includes the hashes %gcb_enums, %lb_enums, etc. |
| 602 | # Similarly we calculate column headings for the tables. |
| 603 | # |
| 604 | # We use string evals to allow the same code to work on |
| 605 | # all the tables |
| 606 | |
| 607 | # Skip if we've already done this code, which populated |
| 608 | # this hash |
| 609 | if (eval "! \%${type}_enums") { |
| 610 | |
| 611 | # For each enum in the type ... |
| 612 | foreach my $enum (keys %enums) { |
| 613 | my $value = $enums{$enum}; |
| 614 | my $short_enum = $short_enum_name{$enum}; |
| 615 | |
| 616 | # Remember the mapping from the property value |
| 617 | # (enum) name to its value. |
| 618 | eval "\$${type}_enums{$enum} = $value"; |
| 619 | die $@ if $@; |
| 620 | |
| 621 | # Remember the inverse mapping to the short name |
| 622 | # so that we can properly label the generated |
| 623 | # table's rows and columns |
| 624 | eval "\$${type}_short_enums[$value] = '$short_enum'"; |
| 625 | die $@ if $@; |
| 626 | |
| 627 | # And note the abbreviations that need explanation |
| 628 | if ($need_explanation{$enum}) { |
| 629 | eval "\$${type}_abbreviations{$short_enum} = '$enum'"; |
| 630 | die $@ if $@; |
| 631 | } |
| 632 | } |
| 633 | |
| 634 | # Each unused enum has the same value. They all are collapsed |
| 635 | # into one row and one column, named $unused_table_hdr. |
| 636 | if (@unused_enums) { |
| 637 | eval "\$${type}_short_enums['$unused_enum_value'] = '$unused_table_hdr'"; |
| 638 | die $@ if $@; |
| 639 | |
| 640 | foreach my $enum (@unused_enums) { |
| 641 | eval "\$${type}_enums{$enum} = $unused_enum_value"; |
| 642 | die $@ if $@; |
| 643 | } |
| 644 | } |
| 645 | } |
| 646 | } |
| 647 | |
| 648 | # The short property names tend to be two lower case letters, but it |
| 649 | # looks better for those if they are upper. XXX |
| 650 | $short_name = uc($short_name) if length($short_name) < 3 |
| 651 | || substr($short_name, 0, 1) =~ /[[:lower:]]/; |
| 652 | $name_prefix = "${short_name}_"; |
| 653 | |
| 654 | # Start the enum definition for this map |
| 655 | my @enum_definition; |
| 656 | my @enum_list; |
| 657 | foreach my $enum (keys %enums) { |
| 658 | $enum_list[$enums{$enum}] = $enum; |
| 659 | } |
| 660 | foreach my $i (0 .. @enum_list - 1) { |
| 661 | push @enum_definition, ",\n" if $i > 0; |
| 662 | |
| 663 | my $name = $enum_list[$i]; |
| 664 | push @enum_definition, "\t${name_prefix}$name = $i"; |
| 665 | } |
| 666 | if (@unused_enums) { |
| 667 | foreach my $unused (@unused_enums) { |
| 668 | push @enum_definition, |
| 669 | ",\n\t${name_prefix}$unused = $unused_enum_value"; |
| 670 | } |
| 671 | } |
| 672 | |
| 673 | # For an 'l' property, we need extra enums, because some of the |
| 674 | # elements are lists. Each such distinct list is placed in its own |
| 675 | # auxiliary map table. Here, we go through the inversion map, and for |
| 676 | # each distinct list found, create an enum value for it, numbered -1, |
| 677 | # -2, .... |
| 678 | my %multiples; |
| 679 | my $aux_table_prefix = "AUX_TABLE_"; |
| 680 | if ($input_format =~ /l/) { |
| 681 | foreach my $element (@$invmap) { |
| 682 | |
| 683 | # A regular scalar is not one of the lists we're looking for |
| 684 | # at this stage. |
| 685 | next unless ref $element; |
| 686 | |
| 687 | my $joined; |
| 688 | if ($input_format =~ /a/) { # These are already ordered |
| 689 | $joined = join ",", @$element; |
| 690 | } |
| 691 | else { |
| 692 | $joined = join ",", sort caselessly @$element; |
| 693 | } |
| 694 | my $already_found = exists $multiples{$joined}; |
| 695 | |
| 696 | my $i; |
| 697 | if ($already_found) { # Use any existing one |
| 698 | $i = $multiples{$joined}; |
| 699 | } |
| 700 | else { # Otherwise increment to get a new table number |
| 701 | $i = keys(%multiples) + 1; |
| 702 | $multiples{$joined} = $i; |
| 703 | } |
| 704 | |
| 705 | # This changes the inversion map for this entry to not be the |
| 706 | # list |
| 707 | $element = "use_$aux_table_prefix$i"; |
| 708 | |
| 709 | # And add to the enum values |
| 710 | if (! $already_found) { |
| 711 | push @enum_definition, ",\n\t${name_prefix}$element = -$i"; |
| 712 | } |
| 713 | } |
| 714 | } |
| 715 | |
| 716 | $enum_declaration_type = "${name_prefix}enum"; |
| 717 | |
| 718 | # Finished with the enum definition. Inversion map stuff is used only |
| 719 | # by regexec or utf-8 (if it is for code points) , unless it is in the |
| 720 | # enum exception list |
| 721 | my $where = (exists $where_to_define_enums{$name}) |
| 722 | ? $where_to_define_enums{$name} |
| 723 | : ($input_format =~ /a/) |
| 724 | ? 'PERL_IN_UTF8_C' |
| 725 | : 'PERL_IN_REGEXEC_C'; |
| 726 | |
| 727 | if (! exists $public_enums{$name}) { |
| 728 | switch_pound_if($name, $where, $charset); |
| 729 | } |
| 730 | else { |
| 731 | end_charset_pound_if; |
| 732 | end_file_pound_if; |
| 733 | start_charset_pound_if($charset, 1); |
| 734 | } |
| 735 | |
| 736 | # If the enum only contains one element, that is a dummy, default one |
| 737 | if (scalar @enum_definition > 1) { |
| 738 | |
| 739 | # Currently unneeded |
| 740 | #print $out_fh "\n#define ${name_prefix}ENUM_COUNT ", |
| 741 | # ..scalar keys %enums, "\n"; |
| 742 | |
| 743 | if ($input_format =~ /l/) { |
| 744 | print $out_fh |
| 745 | "\n", |
| 746 | "/* Negative enum values indicate the need to use an", |
| 747 | " auxiliary table\n", |
| 748 | " * consisting of the list of enums this one expands to.", |
| 749 | " The absolute\n", |
| 750 | " * values of the negative enums are indices into a table", |
| 751 | " of the auxiliary\n", |
| 752 | " * tables' addresses */"; |
| 753 | } |
| 754 | print $out_fh "\ntypedef enum {\n"; |
| 755 | print $out_fh join "", @enum_definition; |
| 756 | print $out_fh "\n"; |
| 757 | print $out_fh "} $enum_declaration_type;\n"; |
| 758 | } |
| 759 | |
| 760 | switch_pound_if($name, $where, $charset); |
| 761 | |
| 762 | # The inversion lists here have to be UV because inversion lists are |
| 763 | # capable of storing any code point, and even though the the ones here |
| 764 | # are only Unicode ones, which need just 21 bits, they are linked to |
| 765 | # directly, rather than copied. The inversion map and aux tables also |
| 766 | # only need be 21 bits, and so we can get away with declaring them |
| 767 | # 32-bits to save a little space and memory (on some 64-bit |
| 768 | # platforms), as they are copied. |
| 769 | $invmap_declaration_type = ($input_format =~ /s/) |
| 770 | ? $enum_declaration_type |
| 771 | : "I32"; |
| 772 | $aux_declaration_type = ($input_format =~ /s/) |
| 773 | ? $enum_declaration_type |
| 774 | : "U32"; |
| 775 | |
| 776 | $output_format = "${name_prefix}%s"; |
| 777 | |
| 778 | # If there are auxiliary tables, output them. |
| 779 | if (%multiples) { |
| 780 | |
| 781 | print $out_fh "\n#define HAS_${name_prefix}AUX_TABLES\n"; |
| 782 | |
| 783 | # Invert keys and values |
| 784 | my %inverted_mults; |
| 785 | while (my ($key, $value) = each %multiples) { |
| 786 | $inverted_mults{$value} = $key; |
| 787 | } |
| 788 | |
| 789 | # Output them in sorted order |
| 790 | my @sorted_table_list = sort { $a <=> $b } keys %inverted_mults; |
| 791 | |
| 792 | # Keep track of how big each aux table is |
| 793 | my @aux_counts; |
| 794 | |
| 795 | # Output each aux table. |
| 796 | foreach my $table_number (@sorted_table_list) { |
| 797 | my $table = $inverted_mults{$table_number}; |
| 798 | output_table_header($out_fh, |
| 799 | $aux_declaration_type, |
| 800 | "$name_prefix$aux_table_prefix$table_number"); |
| 801 | |
| 802 | # Earlier, we joined the elements of this table together with |
| 803 | # a comma |
| 804 | my @elements = split ",", $table; |
| 805 | |
| 806 | $aux_counts[$table_number] = scalar @elements; |
| 807 | for my $i (0 .. @elements - 1) { |
| 808 | print $out_fh ",\n" if $i > 0; |
| 809 | if ($input_format =~ /a/) { |
| 810 | printf $out_fh "\t0x%X", $elements[$i]; |
| 811 | } |
| 812 | else { |
| 813 | print $out_fh "\t${name_prefix}$elements[$i]"; |
| 814 | } |
| 815 | } |
| 816 | |
| 817 | print $out_fh "\n"; |
| 818 | output_table_trailer(); |
| 819 | } |
| 820 | |
| 821 | # Output the table that is indexed by the absolute value of the |
| 822 | # aux table enum and contains pointers to the tables output just |
| 823 | # above |
| 824 | output_table_header($out_fh, "$aux_declaration_type *", |
| 825 | "${name_prefix}${aux_table_prefix}ptrs"); |
| 826 | print $out_fh "\tNULL,\t/* Placeholder */\n"; |
| 827 | for my $i (1 .. @sorted_table_list) { |
| 828 | print $out_fh ",\n" if $i > 1; |
| 829 | print $out_fh "\t$name_prefix$aux_table_prefix$i"; |
| 830 | } |
| 831 | print $out_fh "\n"; |
| 832 | output_table_trailer(); |
| 833 | |
| 834 | print $out_fh |
| 835 | "\n/* Parallel table to the above, giving the number of elements" |
| 836 | . " in each table\n * pointed to */\n"; |
| 837 | output_table_header($out_fh, "U8", |
| 838 | "${name_prefix}${aux_table_prefix}lengths"); |
| 839 | print $out_fh "\t0,\t/* Placeholder */\n"; |
| 840 | for my $i (1 .. @sorted_table_list) { |
| 841 | print $out_fh ",\n" if $i > 1; |
| 842 | print $out_fh |
| 843 | "\t$aux_counts[$i]\t/* $name_prefix$aux_table_prefix$i */"; |
| 844 | } |
| 845 | print $out_fh "\n"; |
| 846 | output_table_trailer(); |
| 847 | } # End of outputting the auxiliary and associated tables |
| 848 | |
| 849 | # The scx property used in regexec.c needs a specialized table which |
| 850 | # is most convenient to output here, while the data structures set up |
| 851 | # above are still extant. This table contains the code point that is |
| 852 | # the zero digit of each script, indexed by script enum value. |
| 853 | if (lc $short_name eq 'scx') { |
| 854 | my @decimals_invlist = prop_invlist("Numeric_Type=Decimal"); |
| 855 | my %script_zeros; |
| 856 | |
| 857 | # Find all the decimal digits. The 0 of each range is always the |
| 858 | # 0th element, except in some early Unicode releases, so check for |
| 859 | # that. |
| 860 | for (my $i = 0; $i < @decimals_invlist; $i += 2) { |
| 861 | my $code_point = $decimals_invlist[$i]; |
| 862 | next if num(chr($code_point)) ne '0'; |
| 863 | |
| 864 | # Turn the scripts this zero is in into a list. |
| 865 | my @scripts = split ",", |
| 866 | charprop($code_point, "_Perl_SCX", '_perl_core_internal_ok'); |
| 867 | $code_point = sprintf("0x%x", $code_point); |
| 868 | |
| 869 | foreach my $script (@scripts) { |
| 870 | if (! exists $script_zeros{$script}) { |
| 871 | $script_zeros{$script} = $code_point; |
| 872 | } |
| 873 | elsif (ref $script_zeros{$script}) { |
| 874 | push $script_zeros{$script}->@*, $code_point; |
| 875 | } |
| 876 | else { # Turn into a list if this is the 2nd zero of the |
| 877 | # script |
| 878 | my $existing = $script_zeros{$script}; |
| 879 | undef $script_zeros{$script}; |
| 880 | push $script_zeros{$script}->@*, $existing, $code_point; |
| 881 | } |
| 882 | } |
| 883 | } |
| 884 | |
| 885 | # @script_zeros contains the zero, sorted by the script's enum |
| 886 | # value |
| 887 | my @script_zeros; |
| 888 | foreach my $script (keys %script_zeros) { |
| 889 | my $enum_value = $enums{$script}; |
| 890 | $script_zeros[$enum_value] = $script_zeros{$script}; |
| 891 | } |
| 892 | |
| 893 | print $out_fh |
| 894 | "\n/* This table, indexed by the script enum, gives the zero" |
| 895 | . " code point for that\n * script; 0 if the script has multiple" |
| 896 | . " digit sequences. Scripts without a\n * digit sequence use" |
| 897 | . " ASCII [0-9], hence are marked '0' */\n"; |
| 898 | output_table_header($out_fh, "UV", "script_zeros"); |
| 899 | for my $i (0 .. @script_zeros - 1) { |
| 900 | my $code_point = $script_zeros[$i]; |
| 901 | if (defined $code_point) { |
| 902 | $code_point = " 0" if ref $code_point; |
| 903 | print $out_fh "\t$code_point"; |
| 904 | } |
| 905 | elsif (lc $enum_list[$i] eq 'inherited') { |
| 906 | print $out_fh "\t 0"; |
| 907 | } |
| 908 | else { # The only digits a script without its own set accepts |
| 909 | # is [0-9] |
| 910 | print $out_fh "\t'0'"; |
| 911 | } |
| 912 | print $out_fh "," if $i < @script_zeros - 1; |
| 913 | print $out_fh "\t/* $enum_list[$i] */"; |
| 914 | print $out_fh "\n"; |
| 915 | } |
| 916 | output_table_trailer(); |
| 917 | } # End of special handling of scx |
| 918 | } |
| 919 | else { |
| 920 | die "'$input_format' invmap() format for '$prop_name' unimplemented"; |
| 921 | } |
| 922 | |
| 923 | die "No inversion map for $prop_name" unless defined $invmap |
| 924 | && ref $invmap eq 'ARRAY' |
| 925 | && $count; |
| 926 | |
| 927 | # Now output the inversion map proper |
| 928 | $charset = "for $charset" if $charset; |
| 929 | output_table_header($out_fh, $invmap_declaration_type, |
| 930 | "${name}_invmap", |
| 931 | $charset); |
| 932 | |
| 933 | # The main body are the scalars passed in to this routine. |
| 934 | for my $i (0 .. $count - 1) { |
| 935 | my $element = $invmap->[$i]; |
| 936 | my $full_element_name = prop_value_aliases($prop_name, $element); |
| 937 | if ($input_format =~ /a/ && $element !~ /\D/) { |
| 938 | $element = ($element == 0) |
| 939 | ? 0 |
| 940 | : sprintf("0x%X", $element); |
| 941 | } |
| 942 | else { |
| 943 | $element = $full_element_name if defined $full_element_name; |
| 944 | $element = $name_prefix . $element; |
| 945 | } |
| 946 | print $out_fh "\t$element"; |
| 947 | print $out_fh "," if $i < $count - 1; |
| 948 | print $out_fh "\n"; |
| 949 | } |
| 950 | output_table_trailer(); |
| 951 | } |
| 952 | |
| 953 | sub mk_invlist_from_sorted_cp_list { |
| 954 | |
| 955 | # Returns an inversion list constructed from the sorted input array of |
| 956 | # code points |
| 957 | |
| 958 | my $list_ref = shift; |
| 959 | |
| 960 | return unless @$list_ref; |
| 961 | |
| 962 | # Initialize to just the first element |
| 963 | my @invlist = ( $list_ref->[0], $list_ref->[0] + 1); |
| 964 | |
| 965 | # For each succeeding element, if it extends the previous range, adjust |
| 966 | # up, otherwise add it. |
| 967 | for my $i (1 .. @$list_ref - 1) { |
| 968 | if ($invlist[-1] == $list_ref->[$i]) { |
| 969 | $invlist[-1]++; |
| 970 | } |
| 971 | else { |
| 972 | push @invlist, $list_ref->[$i], $list_ref->[$i] + 1; |
| 973 | } |
| 974 | } |
| 975 | return @invlist; |
| 976 | } |
| 977 | |
| 978 | print "Reading Case Folding rules.\n" if DEBUG; |
| 979 | # Read in the Case Folding rules, and construct arrays of code points for the |
| 980 | # properties we need. |
| 981 | my ($cp_ref, $folds_ref, $format, $default) = prop_invmap("Case_Folding"); |
| 982 | die "Could not find inversion map for Case_Folding" unless defined $format; |
| 983 | die "Incorrect format '$format' for Case_Folding inversion map" |
| 984 | unless $format eq 'al' |
| 985 | || $format eq 'a'; |
| 986 | print "Finished reading Case Folding rules.\n" if DEBUG; |
| 987 | |
| 988 | |
| 989 | sub _Perl_IVCF { |
| 990 | |
| 991 | # This creates a map of the inversion of case folding. i.e., given a |
| 992 | # character, it gives all the other characters that fold to it. |
| 993 | # |
| 994 | # Inversion maps function kind of like a hash, with the inversion list |
| 995 | # specifying the buckets (keys) and the inversion maps specifying the |
| 996 | # contents of the corresponding bucket. Effectively this function just |
| 997 | # swaps the keys and values of the case fold hash. But there are |
| 998 | # complications. Most importantly, More than one character can each have |
| 999 | # the same fold. This is solved by having a list of characters that fold |
| 1000 | # to a given one. |
| 1001 | |
| 1002 | my %new; |
| 1003 | |
| 1004 | # Go through the inversion list. |
| 1005 | for (my $i = 0; $i < @$cp_ref; $i++) { |
| 1006 | |
| 1007 | # Skip if nothing folds to this |
| 1008 | next if $folds_ref->[$i] == 0; |
| 1009 | |
| 1010 | # This entry which is valid from here to up (but not including) the |
| 1011 | # next entry is for the next $count characters, so that, for example, |
| 1012 | # A-Z is represented by one entry. |
| 1013 | my $cur_list = $cp_ref->[$i]; |
| 1014 | my $count = $cp_ref->[$i+1] - $cur_list; |
| 1015 | |
| 1016 | # The fold of [$i] can be not just a single character, but a sequence |
| 1017 | # of multiple ones. We deal with those here by just creating a string |
| 1018 | # consisting of them. Otherwise, we use the single code point [$i] |
| 1019 | # folds to. |
| 1020 | my $cur_map = (ref $folds_ref->[$i]) |
| 1021 | ? join "", map { chr } $folds_ref->[$i]->@* |
| 1022 | : $folds_ref->[$i]; |
| 1023 | |
| 1024 | # Expand out this range |
| 1025 | while ($count > 0) { |
| 1026 | push @{$new{$cur_map}}, $cur_list; |
| 1027 | |
| 1028 | # A multiple-character fold is a string, and shouldn't need |
| 1029 | # incrementing anyway |
| 1030 | if (ref $folds_ref->[$i]) { |
| 1031 | die sprintf("Case fold for %x is multiple chars; should have" |
| 1032 | . " a count of 1, but instead it was $count", $count) |
| 1033 | unless $count == 1; |
| 1034 | } |
| 1035 | else { |
| 1036 | $cur_map++; |
| 1037 | $cur_list++; |
| 1038 | } |
| 1039 | $count--; |
| 1040 | } |
| 1041 | } |
| 1042 | |
| 1043 | # Now go through and make some adjustments. We add synthetic entries for |
| 1044 | # three cases. |
| 1045 | # 1) If the fold of a Latin1-range character is above that range, some |
| 1046 | # coding in regexec.c can be saved by creating a reverse map here. The |
| 1047 | # impetus for this is that U+B5 (MICRO SIGN) folds to the Greek small |
| 1048 | # mu (U+3BC). That fold isn't done at regex pattern compilation time |
| 1049 | # if it means that the pattern would have to be translated into UTF-8, |
| 1050 | # whose operation is slower. At run time, having this reverse |
| 1051 | # translation eliminates some special cases in the code. |
| 1052 | # 2) Two or more code points can fold to the same multiple character, |
| 1053 | # sequence, as U+FB05 and U+FB06 both fold to 'st'. This code is only |
| 1054 | # for single character folds, but FB05 and FB06 are single characters |
| 1055 | # that are equivalent folded, so we add entries so that they are |
| 1056 | # considered to fold to each other |
| 1057 | # 3) If two or more above-Latin1 code points fold to the same Latin1 range |
| 1058 | # one, we also add entries so that they are considered to fold to each |
| 1059 | # other. This is so that under /aa or /l matching, where folding to |
| 1060 | # their Latin1 range code point is illegal, they still can fold to each |
| 1061 | # other. This situation happens in Unicode 3.0.1, but probably no |
| 1062 | # other version. |
| 1063 | foreach my $fold (keys %new) { |
| 1064 | my $folds_to_string = $fold =~ /\D/; |
| 1065 | |
| 1066 | # If the bucket contains only one element, convert from an array to a |
| 1067 | # scalar |
| 1068 | if (scalar $new{$fold}->@* == 1) { |
| 1069 | $new{$fold} = $new{$fold}[0]; |
| 1070 | |
| 1071 | # Handle case 1) above: if there were a Latin1 range code point |
| 1072 | # whose fold is above that range, this creates an extra entry that |
| 1073 | # maps the other direction, and would save some special case code. |
| 1074 | # (The one current case of this is handled in the else clause |
| 1075 | # below.) |
| 1076 | $new{$new{$fold}} = $fold if $new{$fold} < 256 && $fold > 255; |
| 1077 | } |
| 1078 | else { |
| 1079 | |
| 1080 | # Handle case 1) when there are multiple things that fold to an |
| 1081 | # above-Latin1 code point, at least one of which is in Latin1. |
| 1082 | if (! $folds_to_string && $fold > 255) { |
| 1083 | foreach my $cp ($new{$fold}->@*) { |
| 1084 | if ($cp < 256) { |
| 1085 | my @new_entry = grep { $_ != $cp } $new{$fold}->@*; |
| 1086 | push @new_entry, $fold; |
| 1087 | $new{$cp}->@* = @new_entry; |
| 1088 | } |
| 1089 | } |
| 1090 | } |
| 1091 | |
| 1092 | # Otherwise, sort numerically. This places the highest code point |
| 1093 | # in the list at the tail end. This is because Unicode keeps the |
| 1094 | # lowercase code points as higher ordinals than the uppercase, at |
| 1095 | # least for the ones that matter so far. These are synthetic |
| 1096 | # entries, and we want to predictably have the lowercase (which is |
| 1097 | # more likely to be what gets folded to) in the same corresponding |
| 1098 | # position, so that other code can rely on that. If some new |
| 1099 | # version of Unicode came along that violated this, we might have |
| 1100 | # to change so that the sort is based on upper vs lower instead. |
| 1101 | # (The lower-comes-after isn't true of native EBCDIC, but here we |
| 1102 | # are dealing strictly with Unicode values). |
| 1103 | @{$new{$fold}} = sort { $a <=> $b } $new{$fold}->@* |
| 1104 | unless $folds_to_string; |
| 1105 | # We will be working with a copy of this sorted entry. |
| 1106 | my @source_list = $new{$fold}->@*; |
| 1107 | if (! $folds_to_string) { |
| 1108 | |
| 1109 | # This handles situation 2) listed above, which only arises if |
| 1110 | # what is being folded-to (the fold) is in the Latin1 range. |
| 1111 | if ($fold > 255 ) { |
| 1112 | undef @source_list; |
| 1113 | } |
| 1114 | else { |
| 1115 | # And it only arises if there are two or more folders that |
| 1116 | # fold to it above Latin1. We look at just those. |
| 1117 | @source_list = grep { $_ > 255 } @source_list; |
| 1118 | undef @source_list if @source_list == 1; |
| 1119 | } |
| 1120 | } |
| 1121 | |
| 1122 | # Here, we've found the items we want to set up synthetic folds |
| 1123 | # for. Add entries so that each folds to each other. |
| 1124 | foreach my $cp (@source_list) { |
| 1125 | my @rest = grep { $cp != $_ } @source_list; |
| 1126 | if (@rest == 1) { |
| 1127 | $new{$cp} = $rest[0]; |
| 1128 | } |
| 1129 | else { |
| 1130 | push @{$new{$cp}}, @rest; |
| 1131 | } |
| 1132 | } |
| 1133 | } |
| 1134 | |
| 1135 | # We don't otherwise deal with multiple-character folds |
| 1136 | delete $new{$fold} if $folds_to_string; |
| 1137 | } |
| 1138 | |
| 1139 | |
| 1140 | # Now we have a hash that is the inversion of the case fold property. |
| 1141 | # First find the maximum number of code points that fold to the same one. |
| 1142 | foreach my $fold_to (keys %new) { |
| 1143 | if (ref $new{$fold_to}) { |
| 1144 | my $folders_count = scalar @{$new{$fold_to}}; |
| 1145 | $max_fold_froms = $folders_count if $folders_count > $max_fold_froms; |
| 1146 | } |
| 1147 | } |
| 1148 | |
| 1149 | # Then convert the hash to an inversion map. |
| 1150 | my @sorted_folds = sort { $a <=> $b } keys %new; |
| 1151 | my (@invlist, @invmap); |
| 1152 | |
| 1153 | # We know that nothing folds to the controls (whose ordinals start at 0). |
| 1154 | # And the first real entries are the lowest in the hash. |
| 1155 | push @invlist, 0, $sorted_folds[0]; |
| 1156 | push @invmap, 0, $new{$sorted_folds[0]}; |
| 1157 | |
| 1158 | # Go through the remainder of the hash keys (which are the folded code |
| 1159 | # points) |
| 1160 | for (my $i = 1; $i < @sorted_folds; $i++) { |
| 1161 | |
| 1162 | # Get the current one, and the one prior to it. |
| 1163 | my $fold = $sorted_folds[$i]; |
| 1164 | my $prev_fold = $sorted_folds[$i-1]; |
| 1165 | |
| 1166 | # If the current one is not just 1 away from the prior one, we close |
| 1167 | # out the range containing the previous fold, and know that the gap |
| 1168 | # doesn't have anything that folds. |
| 1169 | if ($fold - 1 != $prev_fold) { |
| 1170 | push @invlist, $prev_fold + 1; |
| 1171 | push @invmap, 0; |
| 1172 | |
| 1173 | # And start a new range |
| 1174 | push @invlist, $fold; |
| 1175 | push @invmap, $new{$fold}; |
| 1176 | } |
| 1177 | elsif ($new{$fold} - 1 != $new{$prev_fold}) { |
| 1178 | |
| 1179 | # Here the current fold is just 1 greater than the previous, but |
| 1180 | # the new map isn't correspondingly 1 greater than the previous, |
| 1181 | # the old range is ended, but since there is no gap, we don't have |
| 1182 | # to insert anything else. |
| 1183 | push @invlist, $fold; |
| 1184 | push @invmap, $new{$fold}; |
| 1185 | |
| 1186 | } # else { Otherwise, this new entry just extends the previous } |
| 1187 | |
| 1188 | die "In IVCF: $invlist[-1] <= $invlist[-2]" |
| 1189 | if $invlist[-1] <= $invlist[-2]; |
| 1190 | } |
| 1191 | |
| 1192 | # And add an entry that indicates that everything above this, to infinity, |
| 1193 | # does not have a case fold. |
| 1194 | push @invlist, $sorted_folds[-1] + 1; |
| 1195 | push @invmap, 0; |
| 1196 | |
| 1197 | push @invlist, 0x110000; |
| 1198 | push @invmap, 0; |
| 1199 | |
| 1200 | # All Unicode versions have some places where multiple code points map to |
| 1201 | # the same one, so the format always has an 'l' |
| 1202 | return \@invlist, \@invmap, 'al', $default; |
| 1203 | } |
| 1204 | |
| 1205 | sub prop_name_for_cmp ($) { # Sort helper |
| 1206 | my $name = shift; |
| 1207 | |
| 1208 | # Returns the input lowercased, with non-alphas removed, as well as |
| 1209 | # everything starting with a comma |
| 1210 | |
| 1211 | $name =~ s/,.*//; |
| 1212 | $name =~ s/[[:^alpha:]]//g; |
| 1213 | return lc $name; |
| 1214 | } |
| 1215 | |
| 1216 | sub UpperLatin1 { |
| 1217 | my @return = mk_invlist_from_sorted_cp_list([ 128 .. 255 ]); |
| 1218 | return \@return; |
| 1219 | } |
| 1220 | |
| 1221 | sub _Perl_CCC_non0_non230 { |
| 1222 | |
| 1223 | # Create an inversion list of code points with non-zero canonical |
| 1224 | # combining class that also don't have 230 as the class number. This is |
| 1225 | # part of a Unicode Standard rule |
| 1226 | |
| 1227 | my @nonzeros = prop_invlist("ccc=0"); |
| 1228 | shift @nonzeros; # Invert so is "ccc != 0" |
| 1229 | |
| 1230 | my @return; |
| 1231 | |
| 1232 | # Expand into list of code points, while excluding those with ccc == 230 |
| 1233 | for (my $i = 0; $i < @nonzeros; $i += 2) { |
| 1234 | my $upper = ($i + 1) < @nonzeros |
| 1235 | ? $nonzeros[$i+1] - 1 # In range |
| 1236 | : $Unicode::UCD::MAX_CP; # To infinity. |
| 1237 | for my $j ($nonzeros[$i] .. $upper) { |
| 1238 | my @ccc_names = prop_value_aliases("ccc", charprop($j, "ccc")); |
| 1239 | |
| 1240 | # Final element in @ccc_names will be all numeric |
| 1241 | push @return, $j if $ccc_names[-1] != 230; |
| 1242 | } |
| 1243 | } |
| 1244 | |
| 1245 | @return = sort { $a <=> $b } @return; |
| 1246 | @return = mk_invlist_from_sorted_cp_list(\@return); |
| 1247 | return \@return; |
| 1248 | } |
| 1249 | |
| 1250 | sub output_table_common { |
| 1251 | |
| 1252 | # Common subroutine to actually output the generated rules table. |
| 1253 | |
| 1254 | my ($property, |
| 1255 | $table_value_defines_ref, |
| 1256 | $table_ref, |
| 1257 | $names_ref, |
| 1258 | $abbreviations_ref) = @_; |
| 1259 | my $size = @$table_ref; |
| 1260 | |
| 1261 | # Output the #define list, sorted by numeric value |
| 1262 | if ($table_value_defines_ref) { |
| 1263 | my $max_name_length = 0; |
| 1264 | my @defines; |
| 1265 | |
| 1266 | # Put in order, and at the same time find the longest name |
| 1267 | while (my ($enum, $value) = each %$table_value_defines_ref) { |
| 1268 | $defines[$value] = $enum; |
| 1269 | |
| 1270 | my $length = length $enum; |
| 1271 | $max_name_length = $length if $length > $max_name_length; |
| 1272 | } |
| 1273 | |
| 1274 | print $out_fh "\n"; |
| 1275 | |
| 1276 | # Output, so that the values are vertically aligned in a column after |
| 1277 | # the longest name |
| 1278 | foreach my $i (0 .. @defines - 1) { |
| 1279 | next unless defined $defines[$i]; |
| 1280 | printf $out_fh "#define %-*s %2d\n", |
| 1281 | $max_name_length, |
| 1282 | $defines[$i], |
| 1283 | $i; |
| 1284 | } |
| 1285 | } |
| 1286 | |
| 1287 | my $column_width = 2; # We currently allow 2 digits for the number |
| 1288 | |
| 1289 | # Being above a U8 is not currently handled |
| 1290 | my $table_type = 'U8'; |
| 1291 | |
| 1292 | # If a name is longer than the width set aside for a column, its column |
| 1293 | # needs to have increased spacing so that the name doesn't get truncated |
| 1294 | # nor run into an adjacent column |
| 1295 | my @spacers; |
| 1296 | |
| 1297 | # Is there a row and column for unused values in this release? |
| 1298 | my $has_unused = $names_ref->[$size-1] eq $unused_table_hdr; |
| 1299 | |
| 1300 | for my $i (0 .. $size - 1) { |
| 1301 | no warnings 'numeric'; |
| 1302 | $spacers[$i] = " " x (length($names_ref->[$i]) - $column_width); |
| 1303 | } |
| 1304 | |
| 1305 | output_table_header($out_fh, $table_type, "${property}_table", undef, |
| 1306 | $size, $size); |
| 1307 | |
| 1308 | # Calculate the column heading line |
| 1309 | my $header_line = "/* " |
| 1310 | . (" " x $max_hdr_len) # We let the row heading meld to |
| 1311 | # the '*/' for those that are at |
| 1312 | # the max |
| 1313 | . " " x 3; # Space for '*/ ' |
| 1314 | # Now each column |
| 1315 | for my $i (0 .. $size - 1) { |
| 1316 | $header_line .= sprintf "%s%*s", |
| 1317 | $spacers[$i], |
| 1318 | $column_width + 1, # 1 for the ',' |
| 1319 | $names_ref->[$i]; |
| 1320 | } |
| 1321 | $header_line .= " */\n"; |
| 1322 | |
| 1323 | # If we have annotations, output it now. |
| 1324 | if ($has_unused || scalar %$abbreviations_ref) { |
| 1325 | my $text = ""; |
| 1326 | foreach my $abbr (sort caselessly keys %$abbreviations_ref) { |
| 1327 | $text .= "; " if $text; |
| 1328 | $text .= "'$abbr' stands for '$abbreviations_ref->{$abbr}'"; |
| 1329 | } |
| 1330 | if ($has_unused) { |
| 1331 | $text .= "; $unused_table_hdr stands for 'unused in this Unicode" |
| 1332 | . " release (and the data in its row and column are garbage)" |
| 1333 | } |
| 1334 | |
| 1335 | my $indent = " " x 3; |
| 1336 | $text = $indent . "/* $text */"; |
| 1337 | |
| 1338 | # Wrap the text so that it is no wider than the table, which the |
| 1339 | # header line gives. |
| 1340 | my $output_width = length $header_line; |
| 1341 | while (length $text > $output_width) { |
| 1342 | my $cur_line = substr($text, 0, $output_width); |
| 1343 | |
| 1344 | # Find the first blank back from the right end to wrap at. |
| 1345 | for (my $i = $output_width -1; $i > 0; $i--) { |
| 1346 | if (substr($text, $i, 1) eq " ") { |
| 1347 | print $out_fh substr($text, 0, $i), "\n"; |
| 1348 | |
| 1349 | # Set so will look at just the remaining tail (which will |
| 1350 | # be indented and have a '*' after the indent |
| 1351 | $text = $indent . " * " . substr($text, $i + 1); |
| 1352 | last; |
| 1353 | } |
| 1354 | } |
| 1355 | } |
| 1356 | |
| 1357 | # And any remaining |
| 1358 | print $out_fh $text, "\n" if $text; |
| 1359 | } |
| 1360 | |
| 1361 | # We calculated the header line earlier just to get its width so that we |
| 1362 | # could make sure the annotations fit into that. |
| 1363 | print $out_fh $header_line; |
| 1364 | |
| 1365 | # Now output the bulk of the table. |
| 1366 | for my $i (0 .. $size - 1) { |
| 1367 | |
| 1368 | # First the row heading. |
| 1369 | printf $out_fh "/* %-*s*/ ", $max_hdr_len, $names_ref->[$i]; |
| 1370 | print $out_fh "{"; # Then the brace for this row |
| 1371 | |
| 1372 | # Then each column |
| 1373 | for my $j (0 .. $size -1) { |
| 1374 | print $out_fh $spacers[$j]; |
| 1375 | printf $out_fh "%*d", $column_width, $table_ref->[$i][$j]; |
| 1376 | print $out_fh "," if $j < $size - 1; |
| 1377 | } |
| 1378 | print $out_fh " }"; |
| 1379 | print $out_fh "," if $i < $size - 1; |
| 1380 | print $out_fh "\n"; |
| 1381 | } |
| 1382 | |
| 1383 | output_table_trailer(); |
| 1384 | } |
| 1385 | |
| 1386 | sub output_GCB_table() { |
| 1387 | |
| 1388 | # Create and output the pair table for use in determining Grapheme Cluster |
| 1389 | # Breaks, given in http://www.unicode.org/reports/tr29/. |
| 1390 | my %gcb_actions = ( |
| 1391 | GCB_NOBREAK => 0, |
| 1392 | GCB_BREAKABLE => 1, |
| 1393 | GCB_RI_then_RI => 2, # Rules 12 and 13 |
| 1394 | GCB_EX_then_EM => 3, # Rule 10 |
| 1395 | GCB_Maybe_Emoji_NonBreak => 4, |
| 1396 | ); |
| 1397 | |
| 1398 | # The table is constructed in reverse order of the rules, to make the |
| 1399 | # lower-numbered, higher priority ones override the later ones, as the |
| 1400 | # algorithm stops at the earliest matching rule |
| 1401 | |
| 1402 | my @gcb_table; |
| 1403 | my $table_size = @gcb_short_enums; |
| 1404 | |
| 1405 | # Otherwise, break everywhere. |
| 1406 | # GB99 Any ÷ Any |
| 1407 | for my $i (0 .. $table_size - 1) { |
| 1408 | for my $j (0 .. $table_size - 1) { |
| 1409 | $gcb_table[$i][$j] = 1; |
| 1410 | } |
| 1411 | } |
| 1412 | |
| 1413 | # Do not break within emoji flag sequences. That is, do not break between |
| 1414 | # regional indicator (RI) symbols if there is an odd number of RI |
| 1415 | # characters before the break point. Must be resolved in runtime code. |
| 1416 | # |
| 1417 | # GB12 sot (RI RI)* RI × RI |
| 1418 | # GB13 [^RI] (RI RI)* RI × RI |
| 1419 | $gcb_table[$gcb_enums{'Regional_Indicator'}] |
| 1420 | [$gcb_enums{'Regional_Indicator'}] = $gcb_actions{GCB_RI_then_RI}; |
| 1421 | |
| 1422 | # Post 11.0: GB11 \p{Extended_Pictographic} Extend* ZWJ |
| 1423 | # × \p{Extended_Pictographic} |
| 1424 | $gcb_table[$gcb_enums{'ZWJ'}][$gcb_enums{'ExtPict_XX'}] = |
| 1425 | $gcb_actions{GCB_Maybe_Emoji_NonBreak}; |
| 1426 | |
| 1427 | # This and the rule GB10 obsolete starting with Unicode 11.0, can be left |
| 1428 | # in as there are no code points that match, so the code won't ever get |
| 1429 | # executed. |
| 1430 | # Do not break within emoji modifier sequences or emoji zwj sequences. |
| 1431 | # Pre 11.0: GB11 ZWJ × ( Glue_After_Zwj | E_Base_GAZ ) |
| 1432 | $gcb_table[$gcb_enums{'ZWJ'}][$gcb_enums{'Glue_After_Zwj'}] = 0; |
| 1433 | $gcb_table[$gcb_enums{'ZWJ'}][$gcb_enums{'E_Base_GAZ'}] = 0; |
| 1434 | |
| 1435 | # GB10 ( E_Base | E_Base_GAZ ) Extend* × E_Modifier |
| 1436 | $gcb_table[$gcb_enums{'Extend'}][$gcb_enums{'E_Modifier'}] |
| 1437 | = $gcb_actions{GCB_EX_then_EM}; |
| 1438 | $gcb_table[$gcb_enums{'E_Base'}][$gcb_enums{'E_Modifier'}] = 0; |
| 1439 | $gcb_table[$gcb_enums{'E_Base_GAZ'}][$gcb_enums{'E_Modifier'}] = 0; |
| 1440 | |
| 1441 | # Do not break before extending characters or ZWJ. |
| 1442 | # Do not break before SpacingMarks, or after Prepend characters. |
| 1443 | # GB9b Prepend × |
| 1444 | # GB9a × SpacingMark |
| 1445 | # GB9 × ( Extend | ZWJ ) |
| 1446 | for my $i (0 .. @gcb_table - 1) { |
| 1447 | $gcb_table[$gcb_enums{'Prepend'}][$i] = 0; |
| 1448 | $gcb_table[$i][$gcb_enums{'SpacingMark'}] = 0; |
| 1449 | $gcb_table[$i][$gcb_enums{'Extend'}] = 0; |
| 1450 | $gcb_table[$i][$gcb_enums{'ZWJ'}] = 0; |
| 1451 | } |
| 1452 | |
| 1453 | # Do not break Hangul syllable sequences. |
| 1454 | # GB8 ( LVT | T) × T |
| 1455 | $gcb_table[$gcb_enums{'LVT'}][$gcb_enums{'T'}] = 0; |
| 1456 | $gcb_table[$gcb_enums{'T'}][$gcb_enums{'T'}] = 0; |
| 1457 | |
| 1458 | # GB7 ( LV | V ) × ( V | T ) |
| 1459 | $gcb_table[$gcb_enums{'LV'}][$gcb_enums{'V'}] = 0; |
| 1460 | $gcb_table[$gcb_enums{'LV'}][$gcb_enums{'T'}] = 0; |
| 1461 | $gcb_table[$gcb_enums{'V'}][$gcb_enums{'V'}] = 0; |
| 1462 | $gcb_table[$gcb_enums{'V'}][$gcb_enums{'T'}] = 0; |
| 1463 | |
| 1464 | # GB6 L × ( L | V | LV | LVT ) |
| 1465 | $gcb_table[$gcb_enums{'L'}][$gcb_enums{'L'}] = 0; |
| 1466 | $gcb_table[$gcb_enums{'L'}][$gcb_enums{'V'}] = 0; |
| 1467 | $gcb_table[$gcb_enums{'L'}][$gcb_enums{'LV'}] = 0; |
| 1468 | $gcb_table[$gcb_enums{'L'}][$gcb_enums{'LVT'}] = 0; |
| 1469 | |
| 1470 | # Do not break between a CR and LF. Otherwise, break before and after |
| 1471 | # controls. |
| 1472 | # GB5 ÷ ( Control | CR | LF ) |
| 1473 | # GB4 ( Control | CR | LF ) ÷ |
| 1474 | for my $i (0 .. @gcb_table - 1) { |
| 1475 | $gcb_table[$i][$gcb_enums{'Control'}] = 1; |
| 1476 | $gcb_table[$i][$gcb_enums{'CR'}] = 1; |
| 1477 | $gcb_table[$i][$gcb_enums{'LF'}] = 1; |
| 1478 | $gcb_table[$gcb_enums{'Control'}][$i] = 1; |
| 1479 | $gcb_table[$gcb_enums{'CR'}][$i] = 1; |
| 1480 | $gcb_table[$gcb_enums{'LF'}][$i] = 1; |
| 1481 | } |
| 1482 | |
| 1483 | # GB3 CR × LF |
| 1484 | $gcb_table[$gcb_enums{'CR'}][$gcb_enums{'LF'}] = 0; |
| 1485 | |
| 1486 | # Break at the start and end of text, unless the text is empty |
| 1487 | # GB1 sot ÷ |
| 1488 | # GB2 ÷ eot |
| 1489 | for my $i (0 .. @gcb_table - 1) { |
| 1490 | $gcb_table[$i][$gcb_enums{'EDGE'}] = 1; |
| 1491 | $gcb_table[$gcb_enums{'EDGE'}][$i] = 1; |
| 1492 | } |
| 1493 | $gcb_table[$gcb_enums{'EDGE'}][$gcb_enums{'EDGE'}] = 0; |
| 1494 | |
| 1495 | output_table_common('GCB', \%gcb_actions, |
| 1496 | \@gcb_table, \@gcb_short_enums, \%gcb_abbreviations); |
| 1497 | } |
| 1498 | |
| 1499 | sub output_LB_table() { |
| 1500 | |
| 1501 | # Create and output the enums, #defines, and pair table for use in |
| 1502 | # determining Line Breaks. This uses the default line break algorithm, |
| 1503 | # given in http://www.unicode.org/reports/tr14/, but tailored by example 7 |
| 1504 | # in that page, as the Unicode-furnished tests assume that tailoring. |
| 1505 | |
| 1506 | # The result is really just true or false. But we follow along with tr14, |
| 1507 | # creating a rule which is false for something like X SP* X. That gets |
| 1508 | # encoding 2. The rest of the actions are synthetic ones that indicate |
| 1509 | # some context handling is required. These each are added to the |
| 1510 | # underlying 0, 1, or 2, instead of replacing them, so that the underlying |
| 1511 | # value can be retrieved. Actually only rules from 7 through 18 (which |
| 1512 | # are the ones where space matter) are possible to have 2 added to them. |
| 1513 | # The others below add just 0 or 1. It might be possible for one |
| 1514 | # synthetic rule to be added to another, yielding a larger value. This |
| 1515 | # doesn't happen in the Unicode 8.0 rule set, and as you can see from the |
| 1516 | # names of the middle grouping below, it is impossible for that to occur |
| 1517 | # for them because they all start with mutually exclusive classes. That |
| 1518 | # the final rule can't be added to any of the others isn't obvious from |
| 1519 | # its name, so it is assigned a power of 2 higher than the others can get |
| 1520 | # to so any addition would preserve all data. (And the code will reach an |
| 1521 | # assert(0) on debugging builds should this happen.) |
| 1522 | my %lb_actions = ( |
| 1523 | LB_NOBREAK => 0, |
| 1524 | LB_BREAKABLE => 1, |
| 1525 | LB_NOBREAK_EVEN_WITH_SP_BETWEEN => 2, |
| 1526 | |
| 1527 | LB_CM_ZWJ_foo => 3, # Rule 9 |
| 1528 | LB_SP_foo => 6, # Rule 18 |
| 1529 | LB_PR_or_PO_then_OP_or_HY => 9, # Rule 25 |
| 1530 | LB_SY_or_IS_then_various => 11, # Rule 25 |
| 1531 | LB_HY_or_BA_then_foo => 13, # Rule 21 |
| 1532 | LB_RI_then_RI => 15, # Rule 30a |
| 1533 | |
| 1534 | LB_various_then_PO_or_PR => (1<<5), # Rule 25 |
| 1535 | ); |
| 1536 | |
| 1537 | # Construct the LB pair table. This is based on the rules in |
| 1538 | # http://www.unicode.org/reports/tr14/, but modified as those rules are |
| 1539 | # designed for someone taking a string of text and sequentially going |
| 1540 | # through it to find the break opportunities, whereas, Perl requires |
| 1541 | # determining if a given random spot is a break opportunity, without |
| 1542 | # knowing all the entire string before it. |
| 1543 | # |
| 1544 | # The table is constructed in reverse order of the rules, to make the |
| 1545 | # lower-numbered, higher priority ones override the later ones, as the |
| 1546 | # algorithm stops at the earliest matching rule |
| 1547 | |
| 1548 | my @lb_table; |
| 1549 | my $table_size = @lb_short_enums; |
| 1550 | |
| 1551 | # LB31. Break everywhere else |
| 1552 | for my $i (0 .. $table_size - 1) { |
| 1553 | for my $j (0 .. $table_size - 1) { |
| 1554 | $lb_table[$i][$j] = $lb_actions{'LB_BREAKABLE'}; |
| 1555 | } |
| 1556 | } |
| 1557 | |
| 1558 | # LB30b Do not break between an emoji base (or potential emoji) and an |
| 1559 | # emoji modifier. |
| 1560 | |
| 1561 | # EB × EM |
| 1562 | # [\p{Extended_Pictographic}&\p{Cn}] × EM |
| 1563 | $lb_table[$lb_enums{'E_Base'}][$lb_enums{'E_Modifier'}] |
| 1564 | = $lb_actions{'LB_NOBREAK'}; |
| 1565 | $lb_table[$lb_enums{'Unassigned_Extended_Pictographic_Ideographic'}] |
| 1566 | [$lb_enums{'E_Modifier'}] = $lb_actions{'LB_NOBREAK'}; |
| 1567 | |
| 1568 | # LB30a Break between two regional indicator symbols if and only if there |
| 1569 | # are an even number of regional indicators preceding the position of the |
| 1570 | # break. |
| 1571 | # sot (RI RI)* RI × RI |
| 1572 | # [^RI] (RI RI)* RI × RI |
| 1573 | $lb_table[$lb_enums{'Regional_Indicator'}] |
| 1574 | [$lb_enums{'Regional_Indicator'}] = $lb_actions{'LB_RI_then_RI'}; |
| 1575 | |
| 1576 | # LB30 Do not break between letters, numbers, or ordinary symbols and |
| 1577 | # non-East-Asian opening punctuation nor non-East-Asian closing |
| 1578 | # parentheses. |
| 1579 | |
| 1580 | # (AL | HL | NU) × [OP-[\p{ea=F}\p{ea=W}\p{ea=H}]] |
| 1581 | # (what we call CP and OP here have already been modified by mktables to |
| 1582 | # exclude the ea items |
| 1583 | $lb_table[$lb_enums{'Alphabetic'}][$lb_enums{'Open_Punctuation'}] |
| 1584 | = $lb_actions{'LB_NOBREAK'}; |
| 1585 | $lb_table[$lb_enums{'Hebrew_Letter'}][$lb_enums{'Open_Punctuation'}] |
| 1586 | = $lb_actions{'LB_NOBREAK'}; |
| 1587 | $lb_table[$lb_enums{'Numeric'}][$lb_enums{'Open_Punctuation'}] |
| 1588 | = $lb_actions{'LB_NOBREAK'}; |
| 1589 | |
| 1590 | # [CP-[\p{ea=F}\p{ea=W}\p{ea=H}]] × (AL | HL | NU) |
| 1591 | $lb_table[$lb_enums{'Close_Parenthesis'}][$lb_enums{'Alphabetic'}] |
| 1592 | = $lb_actions{'LB_NOBREAK'}; |
| 1593 | $lb_table[$lb_enums{'Close_Parenthesis'}][$lb_enums{'Hebrew_Letter'}] |
| 1594 | = $lb_actions{'LB_NOBREAK'}; |
| 1595 | $lb_table[$lb_enums{'Close_Parenthesis'}][$lb_enums{'Numeric'}] |
| 1596 | = $lb_actions{'LB_NOBREAK'}; |
| 1597 | |
| 1598 | # LB29 Do not break between numeric punctuation and alphabetics (“e.g.”). |
| 1599 | # IS × (AL | HL) |
| 1600 | $lb_table[$lb_enums{'Infix_Numeric'}][$lb_enums{'Alphabetic'}] |
| 1601 | = $lb_actions{'LB_NOBREAK'}; |
| 1602 | $lb_table[$lb_enums{'Infix_Numeric'}][$lb_enums{'Hebrew_Letter'}] |
| 1603 | = $lb_actions{'LB_NOBREAK'}; |
| 1604 | |
| 1605 | # LB28 Do not break between alphabetics (“at”). |
| 1606 | # (AL | HL) × (AL | HL) |
| 1607 | $lb_table[$lb_enums{'Alphabetic'}][$lb_enums{'Alphabetic'}] |
| 1608 | = $lb_actions{'LB_NOBREAK'}; |
| 1609 | $lb_table[$lb_enums{'Hebrew_Letter'}][$lb_enums{'Alphabetic'}] |
| 1610 | = $lb_actions{'LB_NOBREAK'}; |
| 1611 | $lb_table[$lb_enums{'Alphabetic'}][$lb_enums{'Hebrew_Letter'}] |
| 1612 | = $lb_actions{'LB_NOBREAK'}; |
| 1613 | $lb_table[$lb_enums{'Hebrew_Letter'}][$lb_enums{'Hebrew_Letter'}] |
| 1614 | = $lb_actions{'LB_NOBREAK'}; |
| 1615 | |
| 1616 | # LB27 Treat a Korean Syllable Block the same as ID. |
| 1617 | # (JL | JV | JT | H2 | H3) × PO |
| 1618 | $lb_table[$lb_enums{'JL'}][$lb_enums{'Postfix_Numeric'}] |
| 1619 | = $lb_actions{'LB_NOBREAK'}; |
| 1620 | $lb_table[$lb_enums{'JV'}][$lb_enums{'Postfix_Numeric'}] |
| 1621 | = $lb_actions{'LB_NOBREAK'}; |
| 1622 | $lb_table[$lb_enums{'JT'}][$lb_enums{'Postfix_Numeric'}] |
| 1623 | = $lb_actions{'LB_NOBREAK'}; |
| 1624 | $lb_table[$lb_enums{'H2'}][$lb_enums{'Postfix_Numeric'}] |
| 1625 | = $lb_actions{'LB_NOBREAK'}; |
| 1626 | $lb_table[$lb_enums{'H3'}][$lb_enums{'Postfix_Numeric'}] |
| 1627 | = $lb_actions{'LB_NOBREAK'}; |
| 1628 | |
| 1629 | # PR × (JL | JV | JT | H2 | H3) |
| 1630 | $lb_table[$lb_enums{'Prefix_Numeric'}][$lb_enums{'JL'}] |
| 1631 | = $lb_actions{'LB_NOBREAK'}; |
| 1632 | $lb_table[$lb_enums{'Prefix_Numeric'}][$lb_enums{'JV'}] |
| 1633 | = $lb_actions{'LB_NOBREAK'}; |
| 1634 | $lb_table[$lb_enums{'Prefix_Numeric'}][$lb_enums{'JT'}] |
| 1635 | = $lb_actions{'LB_NOBREAK'}; |
| 1636 | $lb_table[$lb_enums{'Prefix_Numeric'}][$lb_enums{'H2'}] |
| 1637 | = $lb_actions{'LB_NOBREAK'}; |
| 1638 | $lb_table[$lb_enums{'Prefix_Numeric'}][$lb_enums{'H3'}] |
| 1639 | = $lb_actions{'LB_NOBREAK'}; |
| 1640 | |
| 1641 | # LB26 Do not break a Korean syllable. |
| 1642 | # JL × (JL | JV | H2 | H3) |
| 1643 | $lb_table[$lb_enums{'JL'}][$lb_enums{'JL'}] = $lb_actions{'LB_NOBREAK'}; |
| 1644 | $lb_table[$lb_enums{'JL'}][$lb_enums{'JV'}] = $lb_actions{'LB_NOBREAK'}; |
| 1645 | $lb_table[$lb_enums{'JL'}][$lb_enums{'H2'}] = $lb_actions{'LB_NOBREAK'}; |
| 1646 | $lb_table[$lb_enums{'JL'}][$lb_enums{'H3'}] = $lb_actions{'LB_NOBREAK'}; |
| 1647 | |
| 1648 | # (JV | H2) × (JV | JT) |
| 1649 | $lb_table[$lb_enums{'JV'}][$lb_enums{'JV'}] = $lb_actions{'LB_NOBREAK'}; |
| 1650 | $lb_table[$lb_enums{'H2'}][$lb_enums{'JV'}] = $lb_actions{'LB_NOBREAK'}; |
| 1651 | $lb_table[$lb_enums{'JV'}][$lb_enums{'JT'}] = $lb_actions{'LB_NOBREAK'}; |
| 1652 | $lb_table[$lb_enums{'H2'}][$lb_enums{'JT'}] = $lb_actions{'LB_NOBREAK'}; |
| 1653 | |
| 1654 | # (JT | H3) × JT |
| 1655 | $lb_table[$lb_enums{'JT'}][$lb_enums{'JT'}] = $lb_actions{'LB_NOBREAK'}; |
| 1656 | $lb_table[$lb_enums{'H3'}][$lb_enums{'JT'}] = $lb_actions{'LB_NOBREAK'}; |
| 1657 | |
| 1658 | # LB25 Do not break between the following pairs of classes relevant to |
| 1659 | # numbers, as tailored by example 7 in |
| 1660 | # http://www.unicode.org/reports/tr14/#Examples |
| 1661 | # We follow that tailoring because Unicode's test cases expect it |
| 1662 | # (PR | PO) × ( OP | HY )? NU |
| 1663 | $lb_table[$lb_enums{'Prefix_Numeric'}][$lb_enums{'Numeric'}] |
| 1664 | = $lb_actions{'LB_NOBREAK'}; |
| 1665 | $lb_table[$lb_enums{'Postfix_Numeric'}][$lb_enums{'Numeric'}] |
| 1666 | = $lb_actions{'LB_NOBREAK'}; |
| 1667 | |
| 1668 | # Given that (OP | HY )? is optional, we have to test for it in code. |
| 1669 | # We add in the action (instead of overriding) for this, so that in |
| 1670 | # the code we can recover the underlying break value. |
| 1671 | $lb_table[$lb_enums{'Prefix_Numeric'}][$lb_enums{'Open_Punctuation'}] |
| 1672 | += $lb_actions{'LB_PR_or_PO_then_OP_or_HY'}; |
| 1673 | $lb_table[$lb_enums{'Prefix_Numeric'}][$lb_enums{'East_Asian_OP'}] |
| 1674 | += $lb_actions{'LB_PR_or_PO_then_OP_or_HY'}; |
| 1675 | $lb_table[$lb_enums{'Postfix_Numeric'}][$lb_enums{'Open_Punctuation'}] |
| 1676 | += $lb_actions{'LB_PR_or_PO_then_OP_or_HY'}; |
| 1677 | $lb_table[$lb_enums{'Prefix_Numeric'}][$lb_enums{'Hyphen'}] |
| 1678 | += $lb_actions{'LB_PR_or_PO_then_OP_or_HY'}; |
| 1679 | $lb_table[$lb_enums{'Postfix_Numeric'}][$lb_enums{'Hyphen'}] |
| 1680 | += $lb_actions{'LB_PR_or_PO_then_OP_or_HY'}; |
| 1681 | |
| 1682 | # ( OP | HY ) × NU |
| 1683 | $lb_table[$lb_enums{'Open_Punctuation'}][$lb_enums{'Numeric'}] |
| 1684 | = $lb_actions{'LB_NOBREAK'}; |
| 1685 | $lb_table[$lb_enums{'East_Asian_OP'}][$lb_enums{'Numeric'}] |
| 1686 | = $lb_actions{'LB_NOBREAK'}; |
| 1687 | $lb_table[$lb_enums{'Hyphen'}][$lb_enums{'Numeric'}] |
| 1688 | = $lb_actions{'LB_NOBREAK'}; |
| 1689 | |
| 1690 | # NU (NU | SY | IS)* × (NU | SY | IS | CL | CP ) |
| 1691 | # which can be rewritten as: |
| 1692 | # NU (SY | IS)* × (NU | SY | IS | CL | CP ) |
| 1693 | $lb_table[$lb_enums{'Numeric'}][$lb_enums{'Numeric'}] |
| 1694 | = $lb_actions{'LB_NOBREAK'}; |
| 1695 | $lb_table[$lb_enums{'Numeric'}][$lb_enums{'Break_Symbols'}] |
| 1696 | = $lb_actions{'LB_NOBREAK'}; |
| 1697 | $lb_table[$lb_enums{'Numeric'}][$lb_enums{'Infix_Numeric'}] |
| 1698 | = $lb_actions{'LB_NOBREAK'}; |
| 1699 | $lb_table[$lb_enums{'Numeric'}][$lb_enums{'Close_Punctuation'}] |
| 1700 | = $lb_actions{'LB_NOBREAK'}; |
| 1701 | $lb_table[$lb_enums{'Numeric'}][$lb_enums{'Close_Parenthesis'}] |
| 1702 | = $lb_actions{'LB_NOBREAK'}; |
| 1703 | $lb_table[$lb_enums{'Numeric'}][$lb_enums{'East_Asian_CP'}] |
| 1704 | = $lb_actions{'LB_NOBREAK'}; |
| 1705 | |
| 1706 | # Like earlier where we have to test in code, we add in the action so |
| 1707 | # that we can recover the underlying values. This is done in rules |
| 1708 | # below, as well. The code assumes that we haven't added 2 actions. |
| 1709 | # Shoul a later Unicode release break that assumption, then tests |
| 1710 | # should start failing. |
| 1711 | $lb_table[$lb_enums{'Break_Symbols'}][$lb_enums{'Numeric'}] |
| 1712 | += $lb_actions{'LB_SY_or_IS_then_various'}; |
| 1713 | $lb_table[$lb_enums{'Break_Symbols'}][$lb_enums{'Break_Symbols'}] |
| 1714 | += $lb_actions{'LB_SY_or_IS_then_various'}; |
| 1715 | $lb_table[$lb_enums{'Break_Symbols'}][$lb_enums{'Infix_Numeric'}] |
| 1716 | += $lb_actions{'LB_SY_or_IS_then_various'}; |
| 1717 | $lb_table[$lb_enums{'Break_Symbols'}][$lb_enums{'Close_Punctuation'}] |
| 1718 | += $lb_actions{'LB_SY_or_IS_then_various'}; |
| 1719 | $lb_table[$lb_enums{'Break_Symbols'}][$lb_enums{'Close_Parenthesis'}] |
| 1720 | += $lb_actions{'LB_SY_or_IS_then_various'}; |
| 1721 | $lb_table[$lb_enums{'Break_Symbols'}][$lb_enums{'East_Asian_CP'}] |
| 1722 | += $lb_actions{'LB_SY_or_IS_then_various'}; |
| 1723 | $lb_table[$lb_enums{'Infix_Numeric'}][$lb_enums{'Numeric'}] |
| 1724 | += $lb_actions{'LB_SY_or_IS_then_various'}; |
| 1725 | $lb_table[$lb_enums{'Infix_Numeric'}][$lb_enums{'Break_Symbols'}] |
| 1726 | += $lb_actions{'LB_SY_or_IS_then_various'}; |
| 1727 | $lb_table[$lb_enums{'Infix_Numeric'}][$lb_enums{'Infix_Numeric'}] |
| 1728 | += $lb_actions{'LB_SY_or_IS_then_various'}; |
| 1729 | $lb_table[$lb_enums{'Infix_Numeric'}][$lb_enums{'Close_Punctuation'}] |
| 1730 | += $lb_actions{'LB_SY_or_IS_then_various'}; |
| 1731 | $lb_table[$lb_enums{'Infix_Numeric'}][$lb_enums{'Close_Parenthesis'}] |
| 1732 | += $lb_actions{'LB_SY_or_IS_then_various'}; |
| 1733 | $lb_table[$lb_enums{'Infix_Numeric'}][$lb_enums{'East_Asian_CP'}] |
| 1734 | += $lb_actions{'LB_SY_or_IS_then_various'}; |
| 1735 | |
| 1736 | # NU (NU | SY | IS)* (CL | CP)? × (PO | PR) |
| 1737 | # which can be rewritten as: |
| 1738 | # NU (SY | IS)* (CL | CP)? × (PO | PR) |
| 1739 | $lb_table[$lb_enums{'Numeric'}][$lb_enums{'Postfix_Numeric'}] |
| 1740 | = $lb_actions{'LB_NOBREAK'}; |
| 1741 | $lb_table[$lb_enums{'Numeric'}][$lb_enums{'Prefix_Numeric'}] |
| 1742 | = $lb_actions{'LB_NOBREAK'}; |
| 1743 | |
| 1744 | $lb_table[$lb_enums{'Close_Parenthesis'}][$lb_enums{'Postfix_Numeric'}] |
| 1745 | += $lb_actions{'LB_various_then_PO_or_PR'}; |
| 1746 | $lb_table[$lb_enums{'East_Asian_CP'}][$lb_enums{'Postfix_Numeric'}] |
| 1747 | += $lb_actions{'LB_various_then_PO_or_PR'}; |
| 1748 | $lb_table[$lb_enums{'Close_Punctuation'}][$lb_enums{'Postfix_Numeric'}] |
| 1749 | += $lb_actions{'LB_various_then_PO_or_PR'}; |
| 1750 | $lb_table[$lb_enums{'Infix_Numeric'}][$lb_enums{'Postfix_Numeric'}] |
| 1751 | += $lb_actions{'LB_various_then_PO_or_PR'}; |
| 1752 | $lb_table[$lb_enums{'Break_Symbols'}][$lb_enums{'Postfix_Numeric'}] |
| 1753 | += $lb_actions{'LB_various_then_PO_or_PR'}; |
| 1754 | |
| 1755 | $lb_table[$lb_enums{'Close_Parenthesis'}][$lb_enums{'Prefix_Numeric'}] |
| 1756 | += $lb_actions{'LB_various_then_PO_or_PR'}; |
| 1757 | $lb_table[$lb_enums{'East_Asian_CP'}][$lb_enums{'Prefix_Numeric'}] |
| 1758 | += $lb_actions{'LB_various_then_PO_or_PR'}; |
| 1759 | $lb_table[$lb_enums{'Close_Punctuation'}][$lb_enums{'Prefix_Numeric'}] |
| 1760 | += $lb_actions{'LB_various_then_PO_or_PR'}; |
| 1761 | $lb_table[$lb_enums{'Infix_Numeric'}][$lb_enums{'Prefix_Numeric'}] |
| 1762 | += $lb_actions{'LB_various_then_PO_or_PR'}; |
| 1763 | $lb_table[$lb_enums{'Break_Symbols'}][$lb_enums{'Prefix_Numeric'}] |
| 1764 | += $lb_actions{'LB_various_then_PO_or_PR'}; |
| 1765 | |
| 1766 | # LB24 Do not break between numeric prefix/postfix and letters, or between |
| 1767 | # letters and prefix/postfix. |
| 1768 | # (PR | PO) × (AL | HL) |
| 1769 | $lb_table[$lb_enums{'Prefix_Numeric'}][$lb_enums{'Alphabetic'}] |
| 1770 | = $lb_actions{'LB_NOBREAK'}; |
| 1771 | $lb_table[$lb_enums{'Prefix_Numeric'}][$lb_enums{'Hebrew_Letter'}] |
| 1772 | = $lb_actions{'LB_NOBREAK'}; |
| 1773 | $lb_table[$lb_enums{'Postfix_Numeric'}][$lb_enums{'Alphabetic'}] |
| 1774 | = $lb_actions{'LB_NOBREAK'}; |
| 1775 | $lb_table[$lb_enums{'Postfix_Numeric'}][$lb_enums{'Hebrew_Letter'}] |
| 1776 | = $lb_actions{'LB_NOBREAK'}; |
| 1777 | |
| 1778 | # (AL | HL) × (PR | PO) |
| 1779 | $lb_table[$lb_enums{'Alphabetic'}][$lb_enums{'Prefix_Numeric'}] |
| 1780 | = $lb_actions{'LB_NOBREAK'}; |
| 1781 | $lb_table[$lb_enums{'Hebrew_Letter'}][$lb_enums{'Prefix_Numeric'}] |
| 1782 | = $lb_actions{'LB_NOBREAK'}; |
| 1783 | $lb_table[$lb_enums{'Alphabetic'}][$lb_enums{'Postfix_Numeric'}] |
| 1784 | = $lb_actions{'LB_NOBREAK'}; |
| 1785 | $lb_table[$lb_enums{'Hebrew_Letter'}][$lb_enums{'Postfix_Numeric'}] |
| 1786 | = $lb_actions{'LB_NOBREAK'}; |
| 1787 | |
| 1788 | # LB23a Do not break between numeric prefixes and ideographs, or between |
| 1789 | # ideographs and numeric postfixes. |
| 1790 | # PR × (ID | EB | EM) |
| 1791 | $lb_table[$lb_enums{'Prefix_Numeric'}][$lb_enums{'Ideographic'}] |
| 1792 | = $lb_actions{'LB_NOBREAK'}; |
| 1793 | $lb_table[$lb_enums{'Prefix_Numeric'}] |
| 1794 | [$lb_enums{'Unassigned_Extended_Pictographic_Ideographic'}] |
| 1795 | = $lb_actions{'LB_NOBREAK'}; |
| 1796 | $lb_table[$lb_enums{'Prefix_Numeric'}][$lb_enums{'E_Base'}] |
| 1797 | = $lb_actions{'LB_NOBREAK'}; |
| 1798 | $lb_table[$lb_enums{'Prefix_Numeric'}][$lb_enums{'E_Modifier'}] |
| 1799 | = $lb_actions{'LB_NOBREAK'}; |
| 1800 | |
| 1801 | # (ID | EB | EM) × PO |
| 1802 | $lb_table[$lb_enums{'Ideographic'}][$lb_enums{'Postfix_Numeric'}] |
| 1803 | = $lb_actions{'LB_NOBREAK'}; |
| 1804 | $lb_table[$lb_enums{'Unassigned_Extended_Pictographic_Ideographic'}] |
| 1805 | [$lb_enums{'Postfix_Numeric'}] = $lb_actions{'LB_NOBREAK'}; |
| 1806 | $lb_table[$lb_enums{'E_Base'}][$lb_enums{'Postfix_Numeric'}] |
| 1807 | = $lb_actions{'LB_NOBREAK'}; |
| 1808 | $lb_table[$lb_enums{'E_Modifier'}][$lb_enums{'Postfix_Numeric'}] |
| 1809 | = $lb_actions{'LB_NOBREAK'}; |
| 1810 | |
| 1811 | # LB23 Do not break between digits and letters |
| 1812 | # (AL | HL) × NU |
| 1813 | $lb_table[$lb_enums{'Alphabetic'}][$lb_enums{'Numeric'}] |
| 1814 | = $lb_actions{'LB_NOBREAK'}; |
| 1815 | $lb_table[$lb_enums{'Hebrew_Letter'}][$lb_enums{'Numeric'}] |
| 1816 | = $lb_actions{'LB_NOBREAK'}; |
| 1817 | |
| 1818 | # NU × (AL | HL) |
| 1819 | $lb_table[$lb_enums{'Numeric'}][$lb_enums{'Alphabetic'}] |
| 1820 | = $lb_actions{'LB_NOBREAK'}; |
| 1821 | $lb_table[$lb_enums{'Numeric'}][$lb_enums{'Hebrew_Letter'}] |
| 1822 | = $lb_actions{'LB_NOBREAK'}; |
| 1823 | |
| 1824 | # LB22 Do not break before ellipses |
| 1825 | for my $i (0 .. @lb_table - 1) { |
| 1826 | $lb_table[$i][$lb_enums{'Inseparable'}] = $lb_actions{'LB_NOBREAK'}; |
| 1827 | } |
| 1828 | |
| 1829 | # LB21b Don’t break between Solidus and Hebrew letters. |
| 1830 | # SY × HL |
| 1831 | $lb_table[$lb_enums{'Break_Symbols'}][$lb_enums{'Hebrew_Letter'}] |
| 1832 | = $lb_actions{'LB_NOBREAK'}; |
| 1833 | |
| 1834 | # LB21a Don't break after Hebrew + Hyphen. |
| 1835 | # HL (HY | BA) × |
| 1836 | for my $i (0 .. @lb_table - 1) { |
| 1837 | $lb_table[$lb_enums{'Hyphen'}][$i] |
| 1838 | += $lb_actions{'LB_HY_or_BA_then_foo'}; |
| 1839 | $lb_table[$lb_enums{'Break_After'}][$i] |
| 1840 | += $lb_actions{'LB_HY_or_BA_then_foo'}; |
| 1841 | } |
| 1842 | |
| 1843 | # LB21 Do not break before hyphen-minus, other hyphens, fixed-width |
| 1844 | # spaces, small kana, and other non-starters, or after acute accents. |
| 1845 | # × BA |
| 1846 | # × HY |
| 1847 | # × NS |
| 1848 | # BB × |
| 1849 | for my $i (0 .. @lb_table - 1) { |
| 1850 | $lb_table[$i][$lb_enums{'Break_After'}] = $lb_actions{'LB_NOBREAK'}; |
| 1851 | $lb_table[$i][$lb_enums{'Hyphen'}] = $lb_actions{'LB_NOBREAK'}; |
| 1852 | $lb_table[$i][$lb_enums{'Nonstarter'}] = $lb_actions{'LB_NOBREAK'}; |
| 1853 | $lb_table[$lb_enums{'Break_Before'}][$i] = $lb_actions{'LB_NOBREAK'}; |
| 1854 | } |
| 1855 | |
| 1856 | # LB20 Break before and after unresolved CB. |
| 1857 | # ÷ CB |
| 1858 | # CB ÷ |
| 1859 | # Conditional breaks should be resolved external to the line breaking |
| 1860 | # rules. However, the default action is to treat unresolved CB as breaking |
| 1861 | # before and after. |
| 1862 | for my $i (0 .. @lb_table - 1) { |
| 1863 | $lb_table[$i][$lb_enums{'Contingent_Break'}] |
| 1864 | = $lb_actions{'LB_BREAKABLE'}; |
| 1865 | $lb_table[$lb_enums{'Contingent_Break'}][$i] |
| 1866 | = $lb_actions{'LB_BREAKABLE'}; |
| 1867 | } |
| 1868 | |
| 1869 | # LB19 Do not break before or after quotation marks, such as ‘ ” ’. |
| 1870 | # × QU |
| 1871 | # QU × |
| 1872 | for my $i (0 .. @lb_table - 1) { |
| 1873 | $lb_table[$i][$lb_enums{'Quotation'}] = $lb_actions{'LB_NOBREAK'}; |
| 1874 | $lb_table[$lb_enums{'Quotation'}][$i] = $lb_actions{'LB_NOBREAK'}; |
| 1875 | } |
| 1876 | |
| 1877 | # LB18 Break after spaces |
| 1878 | # SP ÷ |
| 1879 | for my $i (0 .. @lb_table - 1) { |
| 1880 | $lb_table[$lb_enums{'Space'}][$i] = $lb_actions{'LB_BREAKABLE'}; |
| 1881 | } |
| 1882 | |
| 1883 | # LB17 Do not break within ‘——’, even with intervening spaces. |
| 1884 | # B2 SP* × B2 |
| 1885 | $lb_table[$lb_enums{'Break_Both'}][$lb_enums{'Break_Both'}] |
| 1886 | = $lb_actions{'LB_NOBREAK_EVEN_WITH_SP_BETWEEN'}; |
| 1887 | |
| 1888 | # LB16 Do not break between closing punctuation and a nonstarter even with |
| 1889 | # intervening spaces. |
| 1890 | # (CL | CP) SP* × NS |
| 1891 | $lb_table[$lb_enums{'Close_Punctuation'}][$lb_enums{'Nonstarter'}] |
| 1892 | = $lb_actions{'LB_NOBREAK_EVEN_WITH_SP_BETWEEN'}; |
| 1893 | $lb_table[$lb_enums{'Close_Parenthesis'}][$lb_enums{'Nonstarter'}] |
| 1894 | = $lb_actions{'LB_NOBREAK_EVEN_WITH_SP_BETWEEN'}; |
| 1895 | $lb_table[$lb_enums{'East_Asian_CP'}][$lb_enums{'Nonstarter'}] |
| 1896 | = $lb_actions{'LB_NOBREAK_EVEN_WITH_SP_BETWEEN'}; |
| 1897 | |
| 1898 | |
| 1899 | # LB15 Do not break within ‘”[’, even with intervening spaces. |
| 1900 | # QU SP* × OP |
| 1901 | $lb_table[$lb_enums{'Quotation'}][$lb_enums{'Open_Punctuation'}] |
| 1902 | = $lb_actions{'LB_NOBREAK_EVEN_WITH_SP_BETWEEN'}; |
| 1903 | $lb_table[$lb_enums{'Quotation'}][$lb_enums{'East_Asian_OP'}] |
| 1904 | = $lb_actions{'LB_NOBREAK_EVEN_WITH_SP_BETWEEN'}; |
| 1905 | |
| 1906 | # LB14 Do not break after ‘[’, even after spaces. |
| 1907 | # OP SP* × |
| 1908 | for my $i (0 .. @lb_table - 1) { |
| 1909 | $lb_table[$lb_enums{'Open_Punctuation'}][$i] |
| 1910 | = $lb_actions{'LB_NOBREAK_EVEN_WITH_SP_BETWEEN'}; |
| 1911 | $lb_table[$lb_enums{'East_Asian_OP'}][$i] |
| 1912 | = $lb_actions{'LB_NOBREAK_EVEN_WITH_SP_BETWEEN'}; |
| 1913 | } |
| 1914 | |
| 1915 | # LB13 Do not break before ‘]’ or ‘!’ or ‘;’ or ‘/’, even after spaces, as |
| 1916 | # tailored by example 7 in http://www.unicode.org/reports/tr14/#Examples |
| 1917 | # [^NU] × CL |
| 1918 | # [^NU] × CP |
| 1919 | # × EX |
| 1920 | # [^NU] × IS |
| 1921 | # [^NU] × SY |
| 1922 | for my $i (0 .. @lb_table - 1) { |
| 1923 | $lb_table[$i][$lb_enums{'Exclamation'}] |
| 1924 | = $lb_actions{'LB_NOBREAK_EVEN_WITH_SP_BETWEEN'}; |
| 1925 | |
| 1926 | next if $i == $lb_enums{'Numeric'}; |
| 1927 | |
| 1928 | $lb_table[$i][$lb_enums{'Close_Punctuation'}] |
| 1929 | = $lb_actions{'LB_NOBREAK_EVEN_WITH_SP_BETWEEN'}; |
| 1930 | $lb_table[$i][$lb_enums{'Close_Parenthesis'}] |
| 1931 | = $lb_actions{'LB_NOBREAK_EVEN_WITH_SP_BETWEEN'}; |
| 1932 | $lb_table[$i][$lb_enums{'East_Asian_CP'}] |
| 1933 | = $lb_actions{'LB_NOBREAK_EVEN_WITH_SP_BETWEEN'}; |
| 1934 | $lb_table[$i][$lb_enums{'Infix_Numeric'}] |
| 1935 | = $lb_actions{'LB_NOBREAK_EVEN_WITH_SP_BETWEEN'}; |
| 1936 | $lb_table[$i][$lb_enums{'Break_Symbols'}] |
| 1937 | = $lb_actions{'LB_NOBREAK_EVEN_WITH_SP_BETWEEN'}; |
| 1938 | } |
| 1939 | |
| 1940 | # LB12a Do not break before NBSP and related characters, except after |
| 1941 | # spaces and hyphens. |
| 1942 | # [^SP BA HY] × GL |
| 1943 | for my $i (0 .. @lb_table - 1) { |
| 1944 | next if $i == $lb_enums{'Space'} |
| 1945 | || $i == $lb_enums{'Break_After'} |
| 1946 | || $i == $lb_enums{'Hyphen'}; |
| 1947 | |
| 1948 | # We don't break, but if a property above has said don't break even |
| 1949 | # with space between, don't override that (also in the next few rules) |
| 1950 | next if $lb_table[$i][$lb_enums{'Glue'}] |
| 1951 | == $lb_actions{'LB_NOBREAK_EVEN_WITH_SP_BETWEEN'}; |
| 1952 | $lb_table[$i][$lb_enums{'Glue'}] = $lb_actions{'LB_NOBREAK'}; |
| 1953 | } |
| 1954 | |
| 1955 | # LB12 Do not break after NBSP and related characters. |
| 1956 | # GL × |
| 1957 | for my $i (0 .. @lb_table - 1) { |
| 1958 | next if $lb_table[$lb_enums{'Glue'}][$i] |
| 1959 | == $lb_actions{'LB_NOBREAK_EVEN_WITH_SP_BETWEEN'}; |
| 1960 | $lb_table[$lb_enums{'Glue'}][$i] = $lb_actions{'LB_NOBREAK'}; |
| 1961 | } |
| 1962 | |
| 1963 | # LB11 Do not break before or after Word joiner and related characters. |
| 1964 | # × WJ |
| 1965 | # WJ × |
| 1966 | for my $i (0 .. @lb_table - 1) { |
| 1967 | if ($lb_table[$i][$lb_enums{'Word_Joiner'}] |
| 1968 | != $lb_actions{'LB_NOBREAK_EVEN_WITH_SP_BETWEEN'}) |
| 1969 | { |
| 1970 | $lb_table[$i][$lb_enums{'Word_Joiner'}] = $lb_actions{'LB_NOBREAK'}; |
| 1971 | } |
| 1972 | if ($lb_table[$lb_enums{'Word_Joiner'}][$i] |
| 1973 | != $lb_actions{'LB_NOBREAK_EVEN_WITH_SP_BETWEEN'}) |
| 1974 | { |
| 1975 | $lb_table[$lb_enums{'Word_Joiner'}][$i] = $lb_actions{'LB_NOBREAK'}; |
| 1976 | } |
| 1977 | } |
| 1978 | |
| 1979 | # Special case this here to avoid having to do a special case in the code, |
| 1980 | # by making this the same as other things with a SP in front of them that |
| 1981 | # don't break, we avoid an extra test |
| 1982 | $lb_table[$lb_enums{'Space'}][$lb_enums{'Word_Joiner'}] |
| 1983 | = $lb_actions{'LB_NOBREAK_EVEN_WITH_SP_BETWEEN'}; |
| 1984 | |
| 1985 | # LB9 and LB10 are done in the same loop |
| 1986 | # |
| 1987 | # LB9 Do not break a combining character sequence; treat it as if it has |
| 1988 | # the line breaking class of the base character in all of the |
| 1989 | # higher-numbered rules. Treat ZWJ as if it were CM |
| 1990 | # Treat X (CM|ZWJ)* as if it were X. |
| 1991 | # where X is any line break class except BK, CR, LF, NL, SP, or ZW. |
| 1992 | |
| 1993 | # LB10 Treat any remaining combining mark or ZWJ as AL. This catches the |
| 1994 | # case where a CM or ZWJ is the first character on the line or follows SP, |
| 1995 | # BK, CR, LF, NL, or ZW. |
| 1996 | for my $i (0 .. @lb_table - 1) { |
| 1997 | |
| 1998 | # When the CM or ZWJ is the first in the pair, we don't know without |
| 1999 | # looking behind whether the CM or ZWJ is going to attach to an |
| 2000 | # earlier character, or not. So have to figure this out at runtime in |
| 2001 | # the code |
| 2002 | $lb_table[$lb_enums{'Combining_Mark'}][$i] |
| 2003 | = $lb_actions{'LB_CM_ZWJ_foo'}; |
| 2004 | $lb_table[$lb_enums{'ZWJ'}][$i] = $lb_actions{'LB_CM_ZWJ_foo'}; |
| 2005 | |
| 2006 | if ( $i == $lb_enums{'Mandatory_Break'} |
| 2007 | || $i == $lb_enums{'EDGE'} |
| 2008 | || $i == $lb_enums{'Carriage_Return'} |
| 2009 | || $i == $lb_enums{'Line_Feed'} |
| 2010 | || $i == $lb_enums{'Next_Line'} |
| 2011 | || $i == $lb_enums{'Space'} |
| 2012 | || $i == $lb_enums{'ZWSpace'}) |
| 2013 | { |
| 2014 | # For these classes, a following CM doesn't combine, and should do |
| 2015 | # whatever 'Alphabetic' would do. |
| 2016 | $lb_table[$i][$lb_enums{'Combining_Mark'}] |
| 2017 | = $lb_table[$i][$lb_enums{'Alphabetic'}]; |
| 2018 | $lb_table[$i][$lb_enums{'ZWJ'}] |
| 2019 | = $lb_table[$i][$lb_enums{'Alphabetic'}]; |
| 2020 | } |
| 2021 | else { |
| 2022 | # For these classes, the CM or ZWJ combines, so doesn't break, |
| 2023 | # inheriting the type of nobreak from the master character. |
| 2024 | if ($lb_table[$i][$lb_enums{'Combining_Mark'}] |
| 2025 | != $lb_actions{'LB_NOBREAK_EVEN_WITH_SP_BETWEEN'}) |
| 2026 | { |
| 2027 | $lb_table[$i][$lb_enums{'Combining_Mark'}] |
| 2028 | = $lb_actions{'LB_NOBREAK'}; |
| 2029 | } |
| 2030 | if ($lb_table[$i][$lb_enums{'ZWJ'}] |
| 2031 | != $lb_actions{'LB_NOBREAK_EVEN_WITH_SP_BETWEEN'}) |
| 2032 | { |
| 2033 | $lb_table[$i][$lb_enums{'ZWJ'}] |
| 2034 | = $lb_actions{'LB_NOBREAK'}; |
| 2035 | } |
| 2036 | } |
| 2037 | } |
| 2038 | |
| 2039 | # LB8a Do not break after a zero width joiner |
| 2040 | # ZWJ × |
| 2041 | for my $i (0 .. @lb_table - 1) { |
| 2042 | $lb_table[$lb_enums{'ZWJ'}][$i] = $lb_actions{'LB_NOBREAK'}; |
| 2043 | } |
| 2044 | |
| 2045 | # LB8 Break before any character following a zero-width space, even if one |
| 2046 | # or more spaces intervene. |
| 2047 | # ZW SP* ÷ |
| 2048 | for my $i (0 .. @lb_table - 1) { |
| 2049 | $lb_table[$lb_enums{'ZWSpace'}][$i] = $lb_actions{'LB_BREAKABLE'}; |
| 2050 | } |
| 2051 | |
| 2052 | # Because of LB8-10, we need to look at context for "SP x", and this must |
| 2053 | # be done in the code. So override the existing rules for that, by adding |
| 2054 | # a constant to get new rules that tell the code it needs to look at |
| 2055 | # context. By adding this action instead of replacing the existing one, |
| 2056 | # we can get back to the original rule if necessary. |
| 2057 | for my $i (0 .. @lb_table - 1) { |
| 2058 | $lb_table[$lb_enums{'Space'}][$i] += $lb_actions{'LB_SP_foo'}; |
| 2059 | } |
| 2060 | |
| 2061 | # LB7 Do not break before spaces or zero width space. |
| 2062 | # × SP |
| 2063 | # × ZW |
| 2064 | for my $i (0 .. @lb_table - 1) { |
| 2065 | $lb_table[$i][$lb_enums{'Space'}] = $lb_actions{'LB_NOBREAK'}; |
| 2066 | $lb_table[$i][$lb_enums{'ZWSpace'}] = $lb_actions{'LB_NOBREAK'}; |
| 2067 | } |
| 2068 | |
| 2069 | # LB6 Do not break before hard line breaks. |
| 2070 | # × ( BK | CR | LF | NL ) |
| 2071 | for my $i (0 .. @lb_table - 1) { |
| 2072 | $lb_table[$i][$lb_enums{'Mandatory_Break'}] = $lb_actions{'LB_NOBREAK'}; |
| 2073 | $lb_table[$i][$lb_enums{'Carriage_Return'}] = $lb_actions{'LB_NOBREAK'}; |
| 2074 | $lb_table[$i][$lb_enums{'Line_Feed'}] = $lb_actions{'LB_NOBREAK'}; |
| 2075 | $lb_table[$i][$lb_enums{'Next_Line'}] = $lb_actions{'LB_NOBREAK'}; |
| 2076 | } |
| 2077 | |
| 2078 | # LB5 Treat CR followed by LF, as well as CR, LF, and NL as hard line breaks. |
| 2079 | # CR × LF |
| 2080 | # CR ! |
| 2081 | # LF ! |
| 2082 | # NL ! |
| 2083 | for my $i (0 .. @lb_table - 1) { |
| 2084 | $lb_table[$lb_enums{'Carriage_Return'}][$i] |
| 2085 | = $lb_actions{'LB_BREAKABLE'}; |
| 2086 | $lb_table[$lb_enums{'Line_Feed'}][$i] = $lb_actions{'LB_BREAKABLE'}; |
| 2087 | $lb_table[$lb_enums{'Next_Line'}][$i] = $lb_actions{'LB_BREAKABLE'}; |
| 2088 | } |
| 2089 | $lb_table[$lb_enums{'Carriage_Return'}][$lb_enums{'Line_Feed'}] |
| 2090 | = $lb_actions{'LB_NOBREAK'}; |
| 2091 | |
| 2092 | # LB4 Always break after hard line breaks. |
| 2093 | # BK ! |
| 2094 | for my $i (0 .. @lb_table - 1) { |
| 2095 | $lb_table[$lb_enums{'Mandatory_Break'}][$i] |
| 2096 | = $lb_actions{'LB_BREAKABLE'}; |
| 2097 | } |
| 2098 | |
| 2099 | # LB3 Always break at the end of text. |
| 2100 | # ! eot |
| 2101 | # LB2 Never break at the start of text. |
| 2102 | # sot × |
| 2103 | for my $i (0 .. @lb_table - 1) { |
| 2104 | $lb_table[$i][$lb_enums{'EDGE'}] = $lb_actions{'LB_BREAKABLE'}; |
| 2105 | $lb_table[$lb_enums{'EDGE'}][$i] = $lb_actions{'LB_NOBREAK'}; |
| 2106 | } |
| 2107 | |
| 2108 | # LB1 Assign a line breaking class to each code point of the input. |
| 2109 | # Resolve AI, CB, CJ, SA, SG, and XX into other line breaking classes |
| 2110 | # depending on criteria outside the scope of this algorithm. |
| 2111 | # |
| 2112 | # In the absence of such criteria all characters with a specific |
| 2113 | # combination of original class and General_Category property value are |
| 2114 | # resolved as follows: |
| 2115 | # Original Resolved General_Category |
| 2116 | # AI, SG, XX AL Any |
| 2117 | # SA CM Only Mn or Mc |
| 2118 | # SA AL Any except Mn and Mc |
| 2119 | # CJ NS Any |
| 2120 | # |
| 2121 | # This is done in mktables, so we never see any of the remapped-from |
| 2122 | # classes. |
| 2123 | |
| 2124 | output_table_common('LB', \%lb_actions, |
| 2125 | \@lb_table, \@lb_short_enums, \%lb_abbreviations); |
| 2126 | } |
| 2127 | |
| 2128 | sub output_WB_table() { |
| 2129 | |
| 2130 | # Create and output the enums, #defines, and pair table for use in |
| 2131 | # determining Word Breaks, given in http://www.unicode.org/reports/tr29/. |
| 2132 | |
| 2133 | # This uses the same mechanism in the other bounds tables generated by |
| 2134 | # this file. The actions that could override a 0 or 1 are added to those |
| 2135 | # numbers; the actions that clearly don't depend on the underlying rule |
| 2136 | # simply overwrite |
| 2137 | my %wb_actions = ( |
| 2138 | WB_NOBREAK => 0, |
| 2139 | WB_BREAKABLE => 1, |
| 2140 | WB_hs_then_hs => 2, |
| 2141 | WB_Ex_or_FO_or_ZWJ_then_foo => 3, |
| 2142 | WB_DQ_then_HL => 4, |
| 2143 | WB_HL_then_DQ => 6, |
| 2144 | WB_LE_or_HL_then_MB_or_ML_or_SQ => 8, |
| 2145 | WB_MB_or_ML_or_SQ_then_LE_or_HL => 10, |
| 2146 | WB_MB_or_MN_or_SQ_then_NU => 12, |
| 2147 | WB_NU_then_MB_or_MN_or_SQ => 14, |
| 2148 | WB_RI_then_RI => 16, |
| 2149 | ); |
| 2150 | |
| 2151 | # Construct the WB pair table. |
| 2152 | # The table is constructed in reverse order of the rules, to make the |
| 2153 | # lower-numbered, higher priority ones override the later ones, as the |
| 2154 | # algorithm stops at the earliest matching rule |
| 2155 | |
| 2156 | my @wb_table; |
| 2157 | my $table_size = @wb_short_enums; |
| 2158 | |
| 2159 | # Otherwise, break everywhere (including around ideographs). |
| 2160 | # WB99 Any ÷ Any |
| 2161 | for my $i (0 .. $table_size - 1) { |
| 2162 | for my $j (0 .. $table_size - 1) { |
| 2163 | $wb_table[$i][$j] = $wb_actions{'WB_BREAKABLE'}; |
| 2164 | } |
| 2165 | } |
| 2166 | |
| 2167 | # Do not break within emoji flag sequences. That is, do not break between |
| 2168 | # regional indicator (RI) symbols if there is an odd number of RI |
| 2169 | # characters before the break point. |
| 2170 | # WB16 [^RI] (RI RI)* RI × RI |
| 2171 | # WB15 sot (RI RI)* RI × RI |
| 2172 | $wb_table[$wb_enums{'Regional_Indicator'}] |
| 2173 | [$wb_enums{'Regional_Indicator'}] = $wb_actions{'WB_RI_then_RI'}; |
| 2174 | |
| 2175 | # Do not break within emoji modifier sequences. |
| 2176 | # WB14 ( E_Base | EBG ) × E_Modifier |
| 2177 | $wb_table[$wb_enums{'E_Base'}][$wb_enums{'E_Modifier'}] |
| 2178 | = $wb_actions{'WB_NOBREAK'}; |
| 2179 | $wb_table[$wb_enums{'E_Base_GAZ'}][$wb_enums{'E_Modifier'}] |
| 2180 | = $wb_actions{'WB_NOBREAK'}; |
| 2181 | |
| 2182 | # Do not break from extenders. |
| 2183 | # WB13b ExtendNumLet × (ALetter | Hebrew_Letter | Numeric | Katakana) |
| 2184 | $wb_table[$wb_enums{'ExtendNumLet'}][$wb_enums{'ALetter'}] |
| 2185 | = $wb_actions{'WB_NOBREAK'}; |
| 2186 | $wb_table[$wb_enums{'ExtendNumLet'}][$wb_enums{'ExtPict_LE'}] |
| 2187 | = $wb_actions{'WB_NOBREAK'}; |
| 2188 | $wb_table[$wb_enums{'ExtendNumLet'}][$wb_enums{'Hebrew_Letter'}] |
| 2189 | = $wb_actions{'WB_NOBREAK'}; |
| 2190 | $wb_table[$wb_enums{'ExtendNumLet'}][$wb_enums{'Numeric'}] |
| 2191 | = $wb_actions{'WB_NOBREAK'}; |
| 2192 | $wb_table[$wb_enums{'ExtendNumLet'}][$wb_enums{'Katakana'}] |
| 2193 | = $wb_actions{'WB_NOBREAK'}; |
| 2194 | |
| 2195 | # WB13a (ALetter | Hebrew_Letter | Numeric | Katakana | ExtendNumLet) |
| 2196 | # × ExtendNumLet |
| 2197 | $wb_table[$wb_enums{'ALetter'}][$wb_enums{'ExtendNumLet'}] |
| 2198 | = $wb_actions{'WB_NOBREAK'}; |
| 2199 | $wb_table[$wb_enums{'ExtPict_LE'}][$wb_enums{'ExtendNumLet'}] |
| 2200 | = $wb_actions{'WB_NOBREAK'}; |
| 2201 | $wb_table[$wb_enums{'Hebrew_Letter'}][$wb_enums{'ExtendNumLet'}] |
| 2202 | = $wb_actions{'WB_NOBREAK'}; |
| 2203 | $wb_table[$wb_enums{'Numeric'}][$wb_enums{'ExtendNumLet'}] |
| 2204 | = $wb_actions{'WB_NOBREAK'}; |
| 2205 | $wb_table[$wb_enums{'Katakana'}][$wb_enums{'ExtendNumLet'}] |
| 2206 | = $wb_actions{'WB_NOBREAK'}; |
| 2207 | $wb_table[$wb_enums{'ExtendNumLet'}][$wb_enums{'ExtendNumLet'}] |
| 2208 | = $wb_actions{'WB_NOBREAK'}; |
| 2209 | |
| 2210 | # Do not break between Katakana. |
| 2211 | # WB13 Katakana × Katakana |
| 2212 | $wb_table[$wb_enums{'Katakana'}][$wb_enums{'Katakana'}] |
| 2213 | = $wb_actions{'WB_NOBREAK'}; |
| 2214 | |
| 2215 | # Do not break within sequences, such as “3.2” or “3,456.789”. |
| 2216 | # WB12 Numeric × (MidNum | MidNumLet | Single_Quote) Numeric |
| 2217 | $wb_table[$wb_enums{'Numeric'}][$wb_enums{'MidNumLet'}] |
| 2218 | += $wb_actions{'WB_NU_then_MB_or_MN_or_SQ'}; |
| 2219 | $wb_table[$wb_enums{'Numeric'}][$wb_enums{'MidNum'}] |
| 2220 | += $wb_actions{'WB_NU_then_MB_or_MN_or_SQ'}; |
| 2221 | $wb_table[$wb_enums{'Numeric'}][$wb_enums{'Single_Quote'}] |
| 2222 | += $wb_actions{'WB_NU_then_MB_or_MN_or_SQ'}; |
| 2223 | |
| 2224 | # WB11 Numeric (MidNum | (MidNumLet | Single_Quote)) × Numeric |
| 2225 | $wb_table[$wb_enums{'MidNumLet'}][$wb_enums{'Numeric'}] |
| 2226 | += $wb_actions{'WB_MB_or_MN_or_SQ_then_NU'}; |
| 2227 | $wb_table[$wb_enums{'MidNum'}][$wb_enums{'Numeric'}] |
| 2228 | += $wb_actions{'WB_MB_or_MN_or_SQ_then_NU'}; |
| 2229 | $wb_table[$wb_enums{'Single_Quote'}][$wb_enums{'Numeric'}] |
| 2230 | += $wb_actions{'WB_MB_or_MN_or_SQ_then_NU'}; |
| 2231 | |
| 2232 | # Do not break within sequences of digits, or digits adjacent to letters |
| 2233 | # (“3a”, or “A3”). |
| 2234 | # WB10 Numeric × (ALetter | Hebrew_Letter) |
| 2235 | $wb_table[$wb_enums{'Numeric'}][$wb_enums{'ALetter'}] |
| 2236 | = $wb_actions{'WB_NOBREAK'}; |
| 2237 | $wb_table[$wb_enums{'Numeric'}][$wb_enums{'ExtPict_LE'}] |
| 2238 | = $wb_actions{'WB_NOBREAK'}; |
| 2239 | $wb_table[$wb_enums{'Numeric'}][$wb_enums{'Hebrew_Letter'}] |
| 2240 | = $wb_actions{'WB_NOBREAK'}; |
| 2241 | |
| 2242 | # WB9 (ALetter | Hebrew_Letter) × Numeric |
| 2243 | $wb_table[$wb_enums{'ALetter'}][$wb_enums{'Numeric'}] |
| 2244 | = $wb_actions{'WB_NOBREAK'}; |
| 2245 | $wb_table[$wb_enums{'ExtPict_LE'}][$wb_enums{'Numeric'}] |
| 2246 | = $wb_actions{'WB_NOBREAK'}; |
| 2247 | $wb_table[$wb_enums{'Hebrew_Letter'}][$wb_enums{'Numeric'}] |
| 2248 | = $wb_actions{'WB_NOBREAK'}; |
| 2249 | |
| 2250 | # WB8 Numeric × Numeric |
| 2251 | $wb_table[$wb_enums{'Numeric'}][$wb_enums{'Numeric'}] |
| 2252 | = $wb_actions{'WB_NOBREAK'}; |
| 2253 | |
| 2254 | # Do not break letters across certain punctuation. |
| 2255 | # WB7c Hebrew_Letter Double_Quote × Hebrew_Letter |
| 2256 | $wb_table[$wb_enums{'Double_Quote'}][$wb_enums{'Hebrew_Letter'}] |
| 2257 | += $wb_actions{'WB_DQ_then_HL'}; |
| 2258 | |
| 2259 | # WB7b Hebrew_Letter × Double_Quote Hebrew_Letter |
| 2260 | $wb_table[$wb_enums{'Hebrew_Letter'}][$wb_enums{'Double_Quote'}] |
| 2261 | += $wb_actions{'WB_HL_then_DQ'}; |
| 2262 | |
| 2263 | # WB7a Hebrew_Letter × Single_Quote |
| 2264 | $wb_table[$wb_enums{'Hebrew_Letter'}][$wb_enums{'Single_Quote'}] |
| 2265 | = $wb_actions{'WB_NOBREAK'}; |
| 2266 | |
| 2267 | # WB7 (ALetter | Hebrew_Letter) (MidLetter | MidNumLet | Single_Quote) |
| 2268 | # × (ALetter | Hebrew_Letter) |
| 2269 | $wb_table[$wb_enums{'MidNumLet'}][$wb_enums{'ALetter'}] |
| 2270 | += $wb_actions{'WB_MB_or_ML_or_SQ_then_LE_or_HL'}; |
| 2271 | $wb_table[$wb_enums{'MidNumLet'}][$wb_enums{'ExtPict_LE'}] |
| 2272 | += $wb_actions{'WB_MB_or_ML_or_SQ_then_LE_or_HL'}; |
| 2273 | $wb_table[$wb_enums{'MidNumLet'}][$wb_enums{'Hebrew_Letter'}] |
| 2274 | += $wb_actions{'WB_MB_or_ML_or_SQ_then_LE_or_HL'}; |
| 2275 | $wb_table[$wb_enums{'MidLetter'}][$wb_enums{'ALetter'}] |
| 2276 | += $wb_actions{'WB_MB_or_ML_or_SQ_then_LE_or_HL'}; |
| 2277 | $wb_table[$wb_enums{'MidLetter'}][$wb_enums{'ExtPict_LE'}] |
| 2278 | += $wb_actions{'WB_MB_or_ML_or_SQ_then_LE_or_HL'}; |
| 2279 | $wb_table[$wb_enums{'MidLetter'}][$wb_enums{'Hebrew_Letter'}] |
| 2280 | += $wb_actions{'WB_MB_or_ML_or_SQ_then_LE_or_HL'}; |
| 2281 | $wb_table[$wb_enums{'Single_Quote'}][$wb_enums{'ALetter'}] |
| 2282 | += $wb_actions{'WB_MB_or_ML_or_SQ_then_LE_or_HL'}; |
| 2283 | $wb_table[$wb_enums{'Single_Quote'}][$wb_enums{'ExtPict_LE'}] |
| 2284 | += $wb_actions{'WB_MB_or_ML_or_SQ_then_LE_or_HL'}; |
| 2285 | $wb_table[$wb_enums{'Single_Quote'}][$wb_enums{'Hebrew_Letter'}] |
| 2286 | += $wb_actions{'WB_MB_or_ML_or_SQ_then_LE_or_HL'}; |
| 2287 | |
| 2288 | # WB6 (ALetter | Hebrew_Letter) × (MidLetter | MidNumLet |
| 2289 | # | Single_Quote) (ALetter | Hebrew_Letter) |
| 2290 | $wb_table[$wb_enums{'ALetter'}][$wb_enums{'MidNumLet'}] |
| 2291 | += $wb_actions{'WB_LE_or_HL_then_MB_or_ML_or_SQ'}; |
| 2292 | $wb_table[$wb_enums{'ExtPict_LE'}][$wb_enums{'MidNumLet'}] |
| 2293 | += $wb_actions{'WB_LE_or_HL_then_MB_or_ML_or_SQ'}; |
| 2294 | $wb_table[$wb_enums{'Hebrew_Letter'}][$wb_enums{'MidNumLet'}] |
| 2295 | += $wb_actions{'WB_LE_or_HL_then_MB_or_ML_or_SQ'}; |
| 2296 | $wb_table[$wb_enums{'ALetter'}][$wb_enums{'MidLetter'}] |
| 2297 | += $wb_actions{'WB_LE_or_HL_then_MB_or_ML_or_SQ'}; |
| 2298 | $wb_table[$wb_enums{'ExtPict_LE'}][$wb_enums{'MidLetter'}] |
| 2299 | += $wb_actions{'WB_LE_or_HL_then_MB_or_ML_or_SQ'}; |
| 2300 | $wb_table[$wb_enums{'Hebrew_Letter'}][$wb_enums{'MidLetter'}] |
| 2301 | += $wb_actions{'WB_LE_or_HL_then_MB_or_ML_or_SQ'}; |
| 2302 | $wb_table[$wb_enums{'ALetter'}][$wb_enums{'Single_Quote'}] |
| 2303 | += $wb_actions{'WB_LE_or_HL_then_MB_or_ML_or_SQ'}; |
| 2304 | $wb_table[$wb_enums{'ExtPict_LE'}][$wb_enums{'Single_Quote'}] |
| 2305 | += $wb_actions{'WB_LE_or_HL_then_MB_or_ML_or_SQ'}; |
| 2306 | $wb_table[$wb_enums{'Hebrew_Letter'}][$wb_enums{'Single_Quote'}] |
| 2307 | += $wb_actions{'WB_LE_or_HL_then_MB_or_ML_or_SQ'}; |
| 2308 | |
| 2309 | # Do not break between most letters. |
| 2310 | # WB5 (ALetter | Hebrew_Letter) × (ALetter | Hebrew_Letter) |
| 2311 | $wb_table[$wb_enums{'ALetter'}][$wb_enums{'ALetter'}] |
| 2312 | = $wb_actions{'WB_NOBREAK'}; |
| 2313 | $wb_table[$wb_enums{'ExtPict_LE'}][$wb_enums{'ALetter'}] |
| 2314 | = $wb_actions{'WB_NOBREAK'}; |
| 2315 | $wb_table[$wb_enums{'ALetter'}][$wb_enums{'Hebrew_Letter'}] |
| 2316 | = $wb_actions{'WB_NOBREAK'}; |
| 2317 | $wb_table[$wb_enums{'ExtPict_LE'}][$wb_enums{'Hebrew_Letter'}] |
| 2318 | = $wb_actions{'WB_NOBREAK'}; |
| 2319 | $wb_table[$wb_enums{'Hebrew_Letter'}][$wb_enums{'ALetter'}] |
| 2320 | = $wb_actions{'WB_NOBREAK'}; |
| 2321 | $wb_table[$wb_enums{'Hebrew_Letter'}][$wb_enums{'ExtPict_LE'}] |
| 2322 | = $wb_actions{'WB_NOBREAK'}; |
| 2323 | $wb_table[$wb_enums{'Hebrew_Letter'}][$wb_enums{'Hebrew_Letter'}] |
| 2324 | = $wb_actions{'WB_NOBREAK'}; |
| 2325 | $wb_table[$wb_enums{'ExtPict_LE'}][$wb_enums{'ExtPict_LE'}] |
| 2326 | = $wb_actions{'WB_NOBREAK'}; |
| 2327 | |
| 2328 | # Ignore Format and Extend characters, except after sot, CR, LF, and |
| 2329 | # Newline. This also has the effect of: Any × (Format | Extend | ZWJ) |
| 2330 | # WB4 X (Extend | Format | ZWJ)* → X |
| 2331 | for my $i (0 .. @wb_table - 1) { |
| 2332 | $wb_table[$wb_enums{'Extend'}][$i] |
| 2333 | = $wb_actions{'WB_Ex_or_FO_or_ZWJ_then_foo'}; |
| 2334 | $wb_table[$wb_enums{'Format'}][$i] |
| 2335 | = $wb_actions{'WB_Ex_or_FO_or_ZWJ_then_foo'}; |
| 2336 | $wb_table[$wb_enums{'ZWJ'}][$i] |
| 2337 | = $wb_actions{'WB_Ex_or_FO_or_ZWJ_then_foo'}; |
| 2338 | } |
| 2339 | for my $i (0 .. @wb_table - 1) { |
| 2340 | $wb_table[$i][$wb_enums{'Extend'}] = $wb_actions{'WB_NOBREAK'}; |
| 2341 | $wb_table[$i][$wb_enums{'Format'}] = $wb_actions{'WB_NOBREAK'}; |
| 2342 | $wb_table[$i][$wb_enums{'ZWJ'}] = $wb_actions{'WB_NOBREAK'}; |
| 2343 | } |
| 2344 | |
| 2345 | # Implied is that these attach to the character before them, except for |
| 2346 | # the characters that mark the end of a region of text. The rules below |
| 2347 | # override the ones set up here, for all the characters that need |
| 2348 | # overriding. |
| 2349 | for my $i (0 .. @wb_table - 1) { |
| 2350 | $wb_table[$i][$wb_enums{'Extend'}] = $wb_actions{'WB_NOBREAK'}; |
| 2351 | $wb_table[$i][$wb_enums{'Format'}] = $wb_actions{'WB_NOBREAK'}; |
| 2352 | } |
| 2353 | |
| 2354 | # Keep horizontal whitespace together |
| 2355 | # Use perl's tailoring instead |
| 2356 | # WB3d WSegSpace × WSegSpace |
| 2357 | #$wb_table[$wb_enums{'WSegSpace'}][$wb_enums{'WSegSpace'}] |
| 2358 | # = $wb_actions{'WB_NOBREAK'}; |
| 2359 | |
| 2360 | # Do not break within emoji zwj sequences. |
| 2361 | # WB3c ZWJ × ( Glue_After_Zwj | EBG ) |
| 2362 | $wb_table[$wb_enums{'ZWJ'}][$wb_enums{'Glue_After_Zwj'}] |
| 2363 | = $wb_actions{'WB_NOBREAK'}; |
| 2364 | $wb_table[$wb_enums{'ZWJ'}][$wb_enums{'E_Base_GAZ'}] |
| 2365 | = $wb_actions{'WB_NOBREAK'}; |
| 2366 | $wb_table[$wb_enums{'ZWJ'}][$wb_enums{'ExtPict_XX'}] |
| 2367 | = $wb_actions{'WB_NOBREAK'}; |
| 2368 | $wb_table[$wb_enums{'ZWJ'}][$wb_enums{'ExtPict_LE'}] |
| 2369 | = $wb_actions{'WB_NOBREAK'}; |
| 2370 | |
| 2371 | # Break before and after newlines |
| 2372 | # WB3b ÷ (Newline | CR | LF) |
| 2373 | # WB3a (Newline | CR | LF) ÷ |
| 2374 | # et. al. |
| 2375 | for my $i ('CR', 'LF', 'Newline', 'Perl_Tailored_HSpace') { |
| 2376 | for my $j (0 .. @wb_table - 1) { |
| 2377 | $wb_table[$j][$wb_enums{$i}] = $wb_actions{'WB_BREAKABLE'}; |
| 2378 | $wb_table[$wb_enums{$i}][$j] = $wb_actions{'WB_BREAKABLE'}; |
| 2379 | } |
| 2380 | } |
| 2381 | |
| 2382 | # But do not break within white space. |
| 2383 | # WB3 CR × LF |
| 2384 | # et.al. |
| 2385 | for my $i ('CR', 'LF', 'Newline', 'Perl_Tailored_HSpace') { |
| 2386 | for my $j ('CR', 'LF', 'Newline', 'Perl_Tailored_HSpace') { |
| 2387 | $wb_table[$wb_enums{$i}][$wb_enums{$j}] = $wb_actions{'WB_NOBREAK'}; |
| 2388 | } |
| 2389 | } |
| 2390 | |
| 2391 | # And do not break horizontal space followed by Extend or Format or ZWJ |
| 2392 | $wb_table[$wb_enums{'Perl_Tailored_HSpace'}][$wb_enums{'Extend'}] |
| 2393 | = $wb_actions{'WB_NOBREAK'}; |
| 2394 | $wb_table[$wb_enums{'Perl_Tailored_HSpace'}][$wb_enums{'Format'}] |
| 2395 | = $wb_actions{'WB_NOBREAK'}; |
| 2396 | $wb_table[$wb_enums{'Perl_Tailored_HSpace'}][$wb_enums{'ZWJ'}] |
| 2397 | = $wb_actions{'WB_NOBREAK'}; |
| 2398 | $wb_table[$wb_enums{'Perl_Tailored_HSpace'}] |
| 2399 | [$wb_enums{'Perl_Tailored_HSpace'}] |
| 2400 | = $wb_actions{'WB_hs_then_hs'}; |
| 2401 | |
| 2402 | # Break at the start and end of text, unless the text is empty |
| 2403 | # WB2 Any ÷ eot |
| 2404 | # WB1 sot ÷ Any |
| 2405 | for my $i (0 .. @wb_table - 1) { |
| 2406 | $wb_table[$i][$wb_enums{'EDGE'}] = $wb_actions{'WB_BREAKABLE'}; |
| 2407 | $wb_table[$wb_enums{'EDGE'}][$i] = $wb_actions{'WB_BREAKABLE'}; |
| 2408 | } |
| 2409 | $wb_table[$wb_enums{'EDGE'}][$wb_enums{'EDGE'}] = 0; |
| 2410 | |
| 2411 | output_table_common('WB', \%wb_actions, |
| 2412 | \@wb_table, \@wb_short_enums, \%wb_abbreviations); |
| 2413 | } |
| 2414 | |
| 2415 | sub sanitize_name ($) { |
| 2416 | # Change the non-word characters in the input string to standardized word |
| 2417 | # equivalents |
| 2418 | # |
| 2419 | my $sanitized = shift; |
| 2420 | $sanitized =~ s/=/__/; |
| 2421 | $sanitized =~ s/&/_AMP_/; |
| 2422 | $sanitized =~ s/\./_DOT_/; |
| 2423 | $sanitized =~ s/-/_MINUS_/; |
| 2424 | $sanitized =~ s!/!_SLASH_!; |
| 2425 | |
| 2426 | return $sanitized; |
| 2427 | } |
| 2428 | |
| 2429 | sub token_name |
| 2430 | { |
| 2431 | my $name = sanitize_name(shift); |
| 2432 | warn "$name contains non-word" if $name =~ /\W/; |
| 2433 | |
| 2434 | return "$table_name_prefix\U$name" |
| 2435 | } |
| 2436 | |
| 2437 | switch_pound_if ('ALL', 'PERL_IN_REGCOMP_C'); |
| 2438 | |
| 2439 | output_invlist("Latin1", [ 0, 256 ]); |
| 2440 | output_invlist("AboveLatin1", [ 256 ]); |
| 2441 | |
| 2442 | if ($num_anyof_code_points == 256) { # Same as Latin1 |
| 2443 | print $out_fh |
| 2444 | "\nstatic const UV * const InBitmap_invlist = Latin1_invlist;\n"; |
| 2445 | } |
| 2446 | else { |
| 2447 | output_invlist("InBitmap", [ 0, $num_anyof_code_points ]); |
| 2448 | } |
| 2449 | |
| 2450 | end_file_pound_if; |
| 2451 | |
| 2452 | # We construct lists for all the POSIX and backslash sequence character |
| 2453 | # classes in two forms: |
| 2454 | # 1) ones which match only in the ASCII range |
| 2455 | # 2) ones which match either in the Latin1 range, or the entire Unicode range |
| 2456 | # |
| 2457 | # These get compiled in, and hence affect the memory footprint of every Perl |
| 2458 | # program, even those not using Unicode. To minimize the size, currently |
| 2459 | # the Latin1 version is generated for the beyond ASCII range except for those |
| 2460 | # lists that are quite small for the entire range, such as for \s, which is 22 |
| 2461 | # UVs long plus 4 UVs (currently) for the header. |
| 2462 | # |
| 2463 | # To save even more memory, the ASCII versions could be derived from the |
| 2464 | # larger ones at runtime, saving some memory (minus the expense of the machine |
| 2465 | # instructions to do so), but these are all small anyway, so their total is |
| 2466 | # about 100 UVs. |
| 2467 | # |
| 2468 | # In the list of properties below that get generated, the L1 prefix is a fake |
| 2469 | # property that means just the Latin1 range of the full property (whose name |
| 2470 | # has an X prefix instead of L1). |
| 2471 | # |
| 2472 | # An initial & means to use the subroutine from this file instead of an |
| 2473 | # official inversion list. |
| 2474 | # |
| 2475 | print "Computing unicode properties\n" if DEBUG; |
| 2476 | |
| 2477 | # Below is the list of property names to generate. '&' means to use the |
| 2478 | # subroutine to generate the inversion list instead of the generic code |
| 2479 | # below. Some properties have a comma-separated list after the name, |
| 2480 | # These are extra enums to add to those found in the Unicode tables. |
| 2481 | no warnings 'qw'; |
| 2482 | # Ignore non-alpha in sort |
| 2483 | my @props; |
| 2484 | push @props, sort { prop_name_for_cmp($a) cmp prop_name_for_cmp($b) } qw( |
| 2485 | &UpperLatin1 |
| 2486 | _Perl_GCB,EDGE,E_Base,E_Base_GAZ,E_Modifier,Glue_After_Zwj,LV,Prepend,Regional_Indicator,SpacingMark,ZWJ,ExtPict_XX |
| 2487 | _Perl_LB,EDGE,Close_Parenthesis,Hebrew_Letter,Next_Line,Regional_Indicator,ZWJ,Contingent_Break,E_Base,E_Modifier,H2,H3,JL,JT,JV,Word_Joiner,East_Asian_CP,East_Asian_OP,Unassigned_Extended_Pictographic_Ideographic |
| 2488 | _Perl_SB,EDGE,SContinue,CR,Extend,LF |
| 2489 | _Perl_WB,Perl_Tailored_HSpace,EDGE,UNKNOWN,CR,Double_Quote,E_Base,E_Base_GAZ,E_Modifier,Extend,Glue_After_Zwj,Hebrew_Letter,LF,MidNumLet,Newline,Regional_Indicator,Single_Quote,ZWJ,ExtPict_XX,ExtPict_LE |
| 2490 | _Perl_SCX,Latin,Inherited,Unknown,Kore,Jpan,Hanb,INVALID |
| 2491 | Lowercase_Mapping |
| 2492 | Titlecase_Mapping |
| 2493 | Uppercase_Mapping |
| 2494 | Simple_Case_Folding |
| 2495 | Case_Folding |
| 2496 | &_Perl_IVCF |
| 2497 | &_Perl_CCC_non0_non230 |
| 2498 | ); |
| 2499 | # NOTE that the convention is that extra enum values come |
| 2500 | # after the property name, separated by commas, with the enums |
| 2501 | # that aren't ever defined by Unicode (with some exceptions) |
| 2502 | # containing at least 4 all-uppercase characters. |
| 2503 | |
| 2504 | # Some of the enums are current official property values that |
| 2505 | # are needed for the rules in constructing certain tables in |
| 2506 | # this file, and perhaps in regexec.c as well. They are here |
| 2507 | # so that things don't crash when compiled on earlier Unicode |
| 2508 | # releases where they don't exist. Thus the rules that use |
| 2509 | # them still get compiled, but no code point actually uses |
| 2510 | # them, hence they won't get exercized on such Unicode |
| 2511 | # versions, but the code will still compile and run, though |
| 2512 | # may not give the precise results that those versions would |
| 2513 | # expect, but reasonable results nonetheless. |
| 2514 | # |
| 2515 | # Other enums are due to the fact that Unicode has in more |
| 2516 | # recent versions added criteria to the rules in these extra |
| 2517 | # tables that are based on factors outside the property |
| 2518 | # values. And those have to be accounted for, essentially by |
| 2519 | # here splitting certain enum equivalence classes based on |
| 2520 | # those extra rules. |
| 2521 | # |
| 2522 | # EDGE is supposed to be a boundary between some types of |
| 2523 | # enums, but khw thinks that isn't valid any more. |
| 2524 | |
| 2525 | my @bin_props; |
| 2526 | my @perl_prop_synonyms; |
| 2527 | my %enums; |
| 2528 | my @deprecated_messages = ""; # Element [0] is a placeholder |
| 2529 | my %deprecated_tags; |
| 2530 | |
| 2531 | my $float_e_format = qr/ ^ -? \d \. \d+ e [-+] \d+ $ /x; |
| 2532 | |
| 2533 | # Create another hash that maps floating point x.yyEzz representation to what |
| 2534 | # %stricter_to_file_of does for the equivalent rational. A typical entry in |
| 2535 | # the latter hash is |
| 2536 | # |
| 2537 | # 'nv=1/2' => 'Nv/1_2', |
| 2538 | # |
| 2539 | # From that, this loop creates an entry |
| 2540 | # |
| 2541 | # 'nv=5.00e-01' => 'Nv/1_2', |
| 2542 | # |
| 2543 | # %stricter_to_file_of contains far more than just the rationals. Instead we |
| 2544 | # use %Unicode::UCD::nv_floating_to_rational which should have an entry for each |
| 2545 | # nv in the former hash. |
| 2546 | my %floating_to_file_of; |
| 2547 | foreach my $key (keys %Unicode::UCD::nv_floating_to_rational) { |
| 2548 | my $value = $Unicode::UCD::nv_floating_to_rational{$key}; |
| 2549 | $floating_to_file_of{$key} = $Unicode::UCD::stricter_to_file_of{"nv=$value"}; |
| 2550 | } |
| 2551 | |
| 2552 | # Properties that are specified with a prop=value syntax |
| 2553 | my @equals_properties; |
| 2554 | |
| 2555 | # Collect all the binary properties from data in lib/unicore |
| 2556 | # Sort so that complements come after the main table, and the shortest |
| 2557 | # names first, finally alphabetically. Also, sort together the tables we want |
| 2558 | # to be kept together, and prefer those with 'posix' in their names, which is |
| 2559 | # what the C code is expecting their names to be. |
| 2560 | foreach my $property (sort |
| 2561 | { exists $keep_together{lc $b} <=> exists $keep_together{lc $a} |
| 2562 | or $b =~ /posix/i <=> $a =~ /posix/i |
| 2563 | or $b =~ /perl/i <=> $a =~ /perl/i |
| 2564 | or $a =~ $float_e_format <=> $b =~ $float_e_format |
| 2565 | or $a =~ /!/ <=> $b =~ /!/ |
| 2566 | or length $a <=> length $b |
| 2567 | or $a cmp $b |
| 2568 | } keys %Unicode::UCD::loose_to_file_of, |
| 2569 | keys %Unicode::UCD::stricter_to_file_of, |
| 2570 | keys %floating_to_file_of |
| 2571 | ) { |
| 2572 | |
| 2573 | # These two hashes map properties to values that can be considered to |
| 2574 | # be checksums. If two properties have the same checksum, they have |
| 2575 | # identical entries. Otherwise they differ in some way. |
| 2576 | my $tag = $Unicode::UCD::loose_to_file_of{$property}; |
| 2577 | $tag = $Unicode::UCD::stricter_to_file_of{$property} unless defined $tag; |
| 2578 | $tag = $floating_to_file_of{$property} unless defined $tag; |
| 2579 | |
| 2580 | # The tag may contain an '!' meaning it is identical to the one formed |
| 2581 | # by removing the !, except that it is inverted. |
| 2582 | my $inverted = $tag =~ s/!//; |
| 2583 | |
| 2584 | # This hash is lacking the property name |
| 2585 | $property = "nv=$property" if $property =~ $float_e_format; |
| 2586 | |
| 2587 | # The list of 'prop=value' entries that this single entry expands to |
| 2588 | my @this_entries; |
| 2589 | |
| 2590 | # Split 'property=value' on the equals sign, with $lhs being the whole |
| 2591 | # thing if there is no '=' |
| 2592 | my ($lhs, $rhs) = $property =~ / ( [^=]* ) ( =? .*) /x; |
| 2593 | |
| 2594 | # $lhs then becomes the property name. |
| 2595 | my $prop_value = $rhs =~ s/ ^ = //rx; |
| 2596 | |
| 2597 | push @equals_properties, $lhs if $prop_value ne ""; |
| 2598 | |
| 2599 | # See if there are any synonyms for this property. |
| 2600 | if (exists $prop_name_aliases{$lhs}) { |
| 2601 | |
| 2602 | # If so, do the combinatorics so that a new entry is added for |
| 2603 | # each legal property combined with the property value (which is |
| 2604 | # $rhs) |
| 2605 | foreach my $alias (@{$prop_name_aliases{$lhs}}) { |
| 2606 | |
| 2607 | # But, there are some ambiguities, like 'script' is a synonym |
| 2608 | # for 'sc', and 'sc' can stand alone, meaning something |
| 2609 | # entirely different than 'script'. 'script' cannot stand |
| 2610 | # alone. Don't add if the potential new lhs is in the hash of |
| 2611 | # stand-alone properties. |
| 2612 | no warnings 'once'; |
| 2613 | next if $rhs eq "" && grep { $alias eq $_ } |
| 2614 | keys %Unicode::UCD::loose_property_to_file_of; |
| 2615 | |
| 2616 | my $new_entry = $alias . $rhs; |
| 2617 | push @this_entries, $new_entry; |
| 2618 | } |
| 2619 | } |
| 2620 | |
| 2621 | # Above, we added the synonyms for the base entry we're now |
| 2622 | # processing. But we haven't dealt with it yet. If we already have a |
| 2623 | # property with the identical characteristics, this becomes just a |
| 2624 | # synonym for it. |
| 2625 | |
| 2626 | if (exists $enums{$tag}) { |
| 2627 | push @this_entries, $property; |
| 2628 | } |
| 2629 | else { # Otherwise, create a new entry. |
| 2630 | |
| 2631 | # Add to the list of properties to generate inversion lists for. |
| 2632 | push @bin_props, uc $property; |
| 2633 | |
| 2634 | # Create a rule for the parser |
| 2635 | if (! exists $keywords{$property}) { |
| 2636 | $keywords{$property} = token_name($property); |
| 2637 | } |
| 2638 | |
| 2639 | # And create an enum for it. |
| 2640 | $enums{$tag} = $table_name_prefix . uc sanitize_name($property); |
| 2641 | |
| 2642 | $perl_tags{$tag} = 1 if exists $keep_together{lc $property}; |
| 2643 | |
| 2644 | # Some properties are deprecated. This hash tells us so, and the |
| 2645 | # warning message to raise if they are used. |
| 2646 | if (exists $Unicode::UCD::why_deprecated{$tag}) { |
| 2647 | $deprecated_tags{$enums{$tag}} = scalar @deprecated_messages; |
| 2648 | push @deprecated_messages, $Unicode::UCD::why_deprecated{$tag}; |
| 2649 | } |
| 2650 | |
| 2651 | # Our sort above should have made sure that we see the |
| 2652 | # non-inverted version first, but this makes sure. |
| 2653 | warn "$property is inverted!!!" if $inverted; |
| 2654 | } |
| 2655 | |
| 2656 | # Everything else is #defined to be the base enum, inversion is |
| 2657 | # indicated by negating the value. |
| 2658 | my $defined_to = ""; |
| 2659 | $defined_to .= "-" if $inverted; |
| 2660 | $defined_to .= $enums{$tag}; |
| 2661 | |
| 2662 | # Go through the entries that evaluate to this. |
| 2663 | @this_entries = uniques @this_entries; |
| 2664 | foreach my $define (@this_entries) { |
| 2665 | |
| 2666 | # There is a rule for the parser for each. |
| 2667 | $keywords{$define} = $defined_to; |
| 2668 | |
| 2669 | # And a #define for all simple names equivalent to a perl property, |
| 2670 | # except those that begin with 'is' or 'in'; |
| 2671 | if (exists $perl_tags{$tag} && $property !~ / ^ i[ns] | = /x) { |
| 2672 | push @perl_prop_synonyms, "#define " |
| 2673 | . $table_name_prefix |
| 2674 | . uc(sanitize_name($define)) |
| 2675 | . " $defined_to"; |
| 2676 | } |
| 2677 | } |
| 2678 | } |
| 2679 | |
| 2680 | @bin_props = sort { exists $keep_together{lc $b} <=> exists $keep_together{lc $a} |
| 2681 | or $a cmp $b |
| 2682 | } @bin_props; |
| 2683 | @perl_prop_synonyms = sort(uniques(@perl_prop_synonyms)); |
| 2684 | push @props, @bin_props; |
| 2685 | |
| 2686 | foreach my $prop (@props) { |
| 2687 | |
| 2688 | # For the Latin1 properties, we change to use the eXtended version of the |
| 2689 | # base property, then go through the result and get rid of everything not |
| 2690 | # in Latin1 (above 255). Actually, we retain the element for the range |
| 2691 | # that crosses the 255/256 boundary if it is one that matches the |
| 2692 | # property. For example, in the Word property, there is a range of code |
| 2693 | # points that start at U+00F8 and goes through U+02C1. Instead of |
| 2694 | # artificially cutting that off at 256 because 256 is the first code point |
| 2695 | # above Latin1, we let the range go to its natural ending. That gives us |
| 2696 | # extra information with no added space taken. But if the range that |
| 2697 | # crosses the boundary is one that doesn't match the property, we don't |
| 2698 | # start a new range above 255, as that could be construed as going to |
| 2699 | # infinity. For example, the Upper property doesn't include the character |
| 2700 | # at 255, but does include the one at 256. We don't include the 256 one. |
| 2701 | my $prop_name = $prop; |
| 2702 | my $is_local_sub = $prop_name =~ s/^&//; |
| 2703 | my $extra_enums = ""; |
| 2704 | $extra_enums = $1 if $prop_name =~ s/, ( .* ) //x; |
| 2705 | my $lookup_prop = $prop_name; |
| 2706 | $prop_name = sanitize_name($prop_name); |
| 2707 | $prop_name = $table_name_prefix . $prop_name |
| 2708 | if grep { lc $lookup_prop eq lc $_ } @bin_props; |
| 2709 | my $l1_only = ($lookup_prop =~ s/^L1Posix/XPosix/ |
| 2710 | or $lookup_prop =~ s/^L1//); |
| 2711 | my $nonl1_only = 0; |
| 2712 | $nonl1_only = $lookup_prop =~ s/^NonL1// unless $l1_only; |
| 2713 | ($lookup_prop, my $has_suffixes) = $lookup_prop =~ / (.*) ( , .* )? /x; |
| 2714 | |
| 2715 | for my $charset (get_supported_code_pages()) { |
| 2716 | @a2n = @{get_a2n($charset)}; |
| 2717 | |
| 2718 | my @invlist; |
| 2719 | my @invmap; |
| 2720 | my $map_format = 0;; |
| 2721 | my $map_default; |
| 2722 | my $maps_to_code_point = 0; |
| 2723 | my $to_adjust = 0; |
| 2724 | my $same_in_all_code_pages; |
| 2725 | if ($is_local_sub) { |
| 2726 | my @return = eval $lookup_prop; |
| 2727 | die $@ if $@; |
| 2728 | my $invlist_ref = shift @return; |
| 2729 | @invlist = @$invlist_ref; |
| 2730 | if (@return) { # If has other values returned , must be an |
| 2731 | # inversion map |
| 2732 | my $invmap_ref = shift @return; |
| 2733 | @invmap = @$invmap_ref; |
| 2734 | $map_format = shift @return; |
| 2735 | $map_default = shift @return; |
| 2736 | } |
| 2737 | } |
| 2738 | else { |
| 2739 | @invlist = prop_invlist($lookup_prop, '_perl_core_internal_ok'); |
| 2740 | if (! @invlist) { |
| 2741 | |
| 2742 | # If couldn't find a non-empty inversion list, see if it is |
| 2743 | # instead an inversion map |
| 2744 | my ($list_ref, $map_ref, $format, $default) |
| 2745 | = prop_invmap($lookup_prop, '_perl_core_internal_ok'); |
| 2746 | if (! $list_ref) { |
| 2747 | # An empty return here could mean an unknown property, or |
| 2748 | # merely that the original inversion list is empty. Call |
| 2749 | # in scalar context to differentiate |
| 2750 | my $count = prop_invlist($lookup_prop, |
| 2751 | '_perl_core_internal_ok'); |
| 2752 | if (defined $count) { |
| 2753 | # Short-circuit an empty inversion list. |
| 2754 | output_invlist($prop_name, \@invlist, $charset); |
| 2755 | last; |
| 2756 | } |
| 2757 | die "Could not find inversion list for '$lookup_prop'" |
| 2758 | } |
| 2759 | else { |
| 2760 | @invlist = @$list_ref; |
| 2761 | @invmap = @$map_ref; |
| 2762 | $map_format = $format; |
| 2763 | $map_default = $default; |
| 2764 | } |
| 2765 | } |
| 2766 | } |
| 2767 | |
| 2768 | if ($map_format) { |
| 2769 | $maps_to_code_point = $map_format =~ / a ($ | [^r] ) /x; |
| 2770 | $to_adjust = $map_format =~ /a/; |
| 2771 | } |
| 2772 | |
| 2773 | # Re-order the Unicode code points to native ones for this platform. |
| 2774 | # This is only needed for code points below 256, because native code |
| 2775 | # points are only in that range. For inversion maps of properties |
| 2776 | # where the mappings are adjusted (format =~ /a/), this reordering |
| 2777 | # could mess up the adjustment pattern that was in the input, so that |
| 2778 | # has to be dealt with. |
| 2779 | # |
| 2780 | # And inversion maps that map to code points need to eventually have |
| 2781 | # all those code points remapped to native, and it's better to do that |
| 2782 | # here, going through the whole list not just those below 256. This |
| 2783 | # is because some inversion maps have adjustments (format =~ /a/) |
| 2784 | # which may be affected by the reordering. This code needs to be done |
| 2785 | # both for when we are translating the inversion lists for < 256, and |
| 2786 | # for the inversion maps for everything. By doing both in this loop, |
| 2787 | # we can share that code. |
| 2788 | # |
| 2789 | # So, we go through everything for an inversion map to code points; |
| 2790 | # otherwise, we can skip any remapping at all if we are going to |
| 2791 | # output only the above-Latin1 values, or if the range spans the whole |
| 2792 | # of 0..256, as the remap will also include all of 0..256 (256 not |
| 2793 | # 255 because a re-ordering could cause 256 to need to be in the same |
| 2794 | # range as 255.) |
| 2795 | if ( (@invmap && $maps_to_code_point) |
| 2796 | || ( @invlist |
| 2797 | && $invlist[0] < 256 |
| 2798 | && ( $invlist[0] != 0 |
| 2799 | || (scalar @invlist != 1 && $invlist[1] < 256)))) |
| 2800 | { |
| 2801 | $same_in_all_code_pages = 0; |
| 2802 | if (! @invmap) { # Straight inversion list |
| 2803 | # Look at all the ranges that start before 257. |
| 2804 | my @latin1_list; |
| 2805 | while (@invlist) { |
| 2806 | last if $invlist[0] > 256; |
| 2807 | my $upper = @invlist > 1 |
| 2808 | ? $invlist[1] - 1 # In range |
| 2809 | |
| 2810 | # To infinity. You may want to stop much much |
| 2811 | # earlier; going this high may expose perl |
| 2812 | # deficiencies with very large numbers. |
| 2813 | : 256; |
| 2814 | for my $j ($invlist[0] .. $upper) { |
| 2815 | push @latin1_list, a2n($j); |
| 2816 | } |
| 2817 | |
| 2818 | shift @invlist; # Shift off the range that's in the list |
| 2819 | shift @invlist; # Shift off the range not in the list |
| 2820 | } |
| 2821 | |
| 2822 | # Here @invlist contains all the ranges in the original that |
| 2823 | # start at code points above 256, and @latin1_list contains |
| 2824 | # all the native code points for ranges that start with a |
| 2825 | # Unicode code point below 257. We sort the latter and |
| 2826 | # convert it to inversion list format. Then simply prepend it |
| 2827 | # to the list of the higher code points. |
| 2828 | @latin1_list = sort { $a <=> $b } @latin1_list; |
| 2829 | @latin1_list = mk_invlist_from_sorted_cp_list(\@latin1_list); |
| 2830 | unshift @invlist, @latin1_list; |
| 2831 | } |
| 2832 | else { # Is an inversion map |
| 2833 | |
| 2834 | # This is a similar procedure as plain inversion list, but has |
| 2835 | # multiple buckets. A plain inversion list just has two |
| 2836 | # buckets, 1) 'in' the list; and 2) 'not' in the list, and we |
| 2837 | # pretty much can ignore the 2nd bucket, as it is completely |
| 2838 | # defined by the 1st. But here, what we do is create buckets |
| 2839 | # which contain the code points that map to each, translated |
| 2840 | # to native and turned into an inversion list. Thus each |
| 2841 | # bucket is an inversion list of native code points that map |
| 2842 | # to it or don't map to it. We use these to create an |
| 2843 | # inversion map for the whole property. |
| 2844 | |
| 2845 | # As mentioned earlier, we use this procedure to not just |
| 2846 | # remap the inversion list to native values, but also the maps |
| 2847 | # of code points to native ones. In the latter case we have |
| 2848 | # to look at the whole of the inversion map (or at least to |
| 2849 | # above Unicode; as the maps of code points above that should |
| 2850 | # all be to the default). |
| 2851 | my $upper_limit = (! $maps_to_code_point) |
| 2852 | ? 256 |
| 2853 | : (Unicode::UCD::UnicodeVersion() eq '1.1.5') |
| 2854 | ? 0xFFFF |
| 2855 | : 0x10FFFF; |
| 2856 | |
| 2857 | my %mapped_lists; # A hash whose keys are the buckets. |
| 2858 | while (@invlist) { |
| 2859 | last if $invlist[0] > $upper_limit; |
| 2860 | |
| 2861 | # This shouldn't actually happen, as prop_invmap() returns |
| 2862 | # an extra element at the end that is beyond $upper_limit |
| 2863 | die "inversion map (for $prop_name) that extends to" |
| 2864 | . " infinity is unimplemented" unless @invlist > 1; |
| 2865 | |
| 2866 | my $bucket; |
| 2867 | |
| 2868 | # A hash key can't be a ref (we are only expecting arrays |
| 2869 | # of scalars here), so convert any such to a string that |
| 2870 | # will be converted back later (using a vertical tab as |
| 2871 | # the separator). |
| 2872 | if (ref $invmap[0]) { |
| 2873 | $bucket = join "\cK", map { a2n($_) } @{$invmap[0]}; |
| 2874 | } |
| 2875 | elsif ( $maps_to_code_point |
| 2876 | && $invmap[0] =~ $integer_or_float_re) |
| 2877 | { |
| 2878 | |
| 2879 | # Do convert to native for maps to single code points. |
| 2880 | # There are some properties that have a few outlier |
| 2881 | # maps that aren't code points, so the above test |
| 2882 | # skips those. 0 is never remapped. |
| 2883 | $bucket = $invmap[0] == 0 ? 0 : a2n($invmap[0]); |
| 2884 | } else { |
| 2885 | $bucket = $invmap[0]; |
| 2886 | } |
| 2887 | |
| 2888 | # We now have the bucket that all code points in the range |
| 2889 | # map to, though possibly they need to be adjusted. Go |
| 2890 | # through the range and put each translated code point in |
| 2891 | # it into its bucket. |
| 2892 | my $base_map = $invmap[0]; |
| 2893 | for my $j ($invlist[0] .. $invlist[1] - 1) { |
| 2894 | if ($to_adjust |
| 2895 | # The 1st code point doesn't need adjusting |
| 2896 | && $j > $invlist[0] |
| 2897 | |
| 2898 | # Skip any non-numeric maps: these are outliers |
| 2899 | # that aren't code points. |
| 2900 | && $base_map =~ $integer_or_float_re |
| 2901 | |
| 2902 | # 'ne' because the default can be a string |
| 2903 | && $base_map ne $map_default) |
| 2904 | { |
| 2905 | # We adjust, by incrementing each the bucket and |
| 2906 | # the map. For code point maps, translate to |
| 2907 | # native |
| 2908 | $base_map++; |
| 2909 | $bucket = ($maps_to_code_point) |
| 2910 | ? a2n($base_map) |
| 2911 | : $base_map; |
| 2912 | } |
| 2913 | |
| 2914 | # Add the native code point to the bucket for the |
| 2915 | # current map |
| 2916 | push @{$mapped_lists{$bucket}}, a2n($j); |
| 2917 | } # End of loop through all code points in the range |
| 2918 | |
| 2919 | # Get ready for the next range |
| 2920 | shift @invlist; |
| 2921 | shift @invmap; |
| 2922 | } # End of loop through all ranges in the map. |
| 2923 | |
| 2924 | # Here, @invlist and @invmap retain all the ranges from the |
| 2925 | # originals that start with code points above $upper_limit. |
| 2926 | # Each bucket in %mapped_lists contains all the code points |
| 2927 | # that map to that bucket. If the bucket is for a map to a |
| 2928 | # single code point, the bucket has been converted to native. |
| 2929 | # If something else (including multiple code points), no |
| 2930 | # conversion is done. |
| 2931 | # |
| 2932 | # Now we recreate the inversion map into %xlated, but this |
| 2933 | # time for the native character set. |
| 2934 | my %xlated; |
| 2935 | foreach my $bucket (keys %mapped_lists) { |
| 2936 | |
| 2937 | # Sort and convert this bucket to an inversion list. The |
| 2938 | # result will be that ranges that start with even-numbered |
| 2939 | # indexes will be for code points that map to this bucket; |
| 2940 | # odd ones map to some other bucket, and are discarded |
| 2941 | # below. |
| 2942 | @{$mapped_lists{$bucket}} |
| 2943 | = sort{ $a <=> $b} @{$mapped_lists{$bucket}}; |
| 2944 | @{$mapped_lists{$bucket}} |
| 2945 | = mk_invlist_from_sorted_cp_list( |
| 2946 | \@{$mapped_lists{$bucket}}); |
| 2947 | |
| 2948 | # Add each even-numbered range in the bucket to %xlated; |
| 2949 | # so that the keys of %xlated become the range start code |
| 2950 | # points, and the values are their corresponding maps. |
| 2951 | while (@{$mapped_lists{$bucket}}) { |
| 2952 | my $range_start = $mapped_lists{$bucket}->[0]; |
| 2953 | if ($bucket =~ /\cK/) { |
| 2954 | @{$xlated{$range_start}} = split /\cK/, $bucket; |
| 2955 | } |
| 2956 | else { |
| 2957 | # If adjusting, and there is more than one thing |
| 2958 | # that maps to the same thing, they must be split |
| 2959 | # so that later the adjusting doesn't think the |
| 2960 | # subsequent items can go away because of the |
| 2961 | # adjusting. |
| 2962 | my $range_end = ( $to_adjust |
| 2963 | && $bucket != $map_default) |
| 2964 | ? $mapped_lists{$bucket}->[1] - 1 |
| 2965 | : $range_start; |
| 2966 | for my $i ($range_start .. $range_end) { |
| 2967 | $xlated{$i} = $bucket; |
| 2968 | } |
| 2969 | } |
| 2970 | shift @{$mapped_lists{$bucket}}; # Discard odd ranges |
| 2971 | shift @{$mapped_lists{$bucket}}; # Get ready for next |
| 2972 | # iteration |
| 2973 | } |
| 2974 | } # End of loop through all the buckets. |
| 2975 | |
| 2976 | # Here %xlated's keys are the range starts of all the code |
| 2977 | # points in the inversion map. Construct an inversion list |
| 2978 | # from them. |
| 2979 | my @new_invlist = sort { $a <=> $b } keys %xlated; |
| 2980 | |
| 2981 | # If the list is adjusted, we want to munge this list so that |
| 2982 | # we only have one entry for where consecutive code points map |
| 2983 | # to consecutive values. We just skip the subsequent entries |
| 2984 | # where this is the case. |
| 2985 | if ($to_adjust) { |
| 2986 | my @temp; |
| 2987 | for my $i (0 .. @new_invlist - 1) { |
| 2988 | next if $i > 0 |
| 2989 | && $new_invlist[$i-1] + 1 == $new_invlist[$i] |
| 2990 | && $xlated{$new_invlist[$i-1]} |
| 2991 | =~ $integer_or_float_re |
| 2992 | && $xlated{$new_invlist[$i]} |
| 2993 | =~ $integer_or_float_re |
| 2994 | && $xlated{$new_invlist[$i-1]} + 1 |
| 2995 | == $xlated{$new_invlist[$i]}; |
| 2996 | push @temp, $new_invlist[$i]; |
| 2997 | } |
| 2998 | @new_invlist = @temp; |
| 2999 | } |
| 3000 | |
| 3001 | # The inversion map comes from %xlated's values. We can |
| 3002 | # unshift each onto the front of the untouched portion, in |
| 3003 | # reverse order of the portion we did process. |
| 3004 | foreach my $start (reverse @new_invlist) { |
| 3005 | unshift @invmap, $xlated{$start}; |
| 3006 | } |
| 3007 | |
| 3008 | # Finally prepend the inversion list we have just constructed |
| 3009 | # to the one that contains anything we didn't process. |
| 3010 | unshift @invlist, @new_invlist; |
| 3011 | } |
| 3012 | } |
| 3013 | elsif (@invmap) { # inversion maps can't cope with this variable |
| 3014 | # being true, even if it could be true |
| 3015 | $same_in_all_code_pages = 0; |
| 3016 | } |
| 3017 | else { |
| 3018 | $same_in_all_code_pages = 1; |
| 3019 | } |
| 3020 | |
| 3021 | # prop_invmap() returns an extra final entry, which we can now |
| 3022 | # discard. |
| 3023 | if (@invmap) { |
| 3024 | pop @invlist; |
| 3025 | pop @invmap; |
| 3026 | } |
| 3027 | |
| 3028 | if ($l1_only) { |
| 3029 | die "Unimplemented to do a Latin-1 only inversion map" if @invmap; |
| 3030 | for my $i (0 .. @invlist - 1 - 1) { |
| 3031 | if ($invlist[$i] > 255) { |
| 3032 | |
| 3033 | # In an inversion list, even-numbered elements give the code |
| 3034 | # points that begin ranges that match the property; |
| 3035 | # odd-numbered give ones that begin ranges that don't match. |
| 3036 | # If $i is odd, we are at the first code point above 255 that |
| 3037 | # doesn't match, which means the range it is ending does |
| 3038 | # match, and crosses the 255/256 boundary. We want to |
| 3039 | # include this ending point, so increment $i, so the |
| 3040 | # splice below includes it. Conversely, if $i is even, it |
| 3041 | # is the first code point above 255 that matches, which |
| 3042 | # means there was no matching range that crossed the |
| 3043 | # boundary, and we don't want to include this code point, |
| 3044 | # so splice before it. |
| 3045 | $i++ if $i % 2 != 0; |
| 3046 | |
| 3047 | # Remove everything past this. |
| 3048 | splice @invlist, $i; |
| 3049 | splice @invmap, $i if @invmap; |
| 3050 | last; |
| 3051 | } |
| 3052 | } |
| 3053 | } |
| 3054 | elsif ($nonl1_only) { |
| 3055 | my $found_nonl1 = 0; |
| 3056 | for my $i (0 .. @invlist - 1 - 1) { |
| 3057 | next if $invlist[$i] < 256; |
| 3058 | |
| 3059 | # Here, we have the first element in the array that indicates an |
| 3060 | # element above Latin1. Get rid of all previous ones. |
| 3061 | splice @invlist, 0, $i; |
| 3062 | splice @invmap, 0, $i if @invmap; |
| 3063 | |
| 3064 | # If this one's index is not divisible by 2, it means that this |
| 3065 | # element is inverting away from being in the list, which means |
| 3066 | # all code points from 256 to this one are in this list (or |
| 3067 | # map to the default for inversion maps) |
| 3068 | if ($i % 2 != 0) { |
| 3069 | unshift @invlist, 256; |
| 3070 | unshift @invmap, $map_default if @invmap; |
| 3071 | } |
| 3072 | $found_nonl1 = 1; |
| 3073 | last; |
| 3074 | } |
| 3075 | if (! $found_nonl1) { |
| 3076 | warn "No non-Latin1 code points in $prop_name"; |
| 3077 | output_invlist($prop_name, []); |
| 3078 | last; |
| 3079 | } |
| 3080 | } |
| 3081 | |
| 3082 | switch_pound_if ($prop_name, 'PERL_IN_REGCOMP_C'); |
| 3083 | start_charset_pound_if($charset, 1) unless $same_in_all_code_pages; |
| 3084 | |
| 3085 | output_invlist($prop_name, \@invlist, ($same_in_all_code_pages) |
| 3086 | ? $applies_to_all_charsets_text |
| 3087 | : $charset); |
| 3088 | |
| 3089 | if (@invmap) { |
| 3090 | output_invmap($prop_name, \@invmap, $lookup_prop, $map_format, |
| 3091 | $map_default, $extra_enums, $charset); |
| 3092 | } |
| 3093 | |
| 3094 | last if $same_in_all_code_pages; |
| 3095 | end_charset_pound_if; |
| 3096 | } |
| 3097 | } |
| 3098 | |
| 3099 | print "Finished computing unicode properties\n" if DEBUG; |
| 3100 | |
| 3101 | print $out_fh "\nconst char * const deprecated_property_msgs[] = {\n\t"; |
| 3102 | print $out_fh join ",\n\t", map { "\"$_\"" } @deprecated_messages; |
| 3103 | print $out_fh "\n};\n"; |
| 3104 | |
| 3105 | switch_pound_if ('binary_invlist_enum', 'PERL_IN_REGCOMP_C'); |
| 3106 | |
| 3107 | my @enums = sort values %enums; |
| 3108 | |
| 3109 | # Save a copy of these before modification |
| 3110 | my @invlist_names = map { "${_}_invlist" } @enums; |
| 3111 | |
| 3112 | # Post-process the enums for deprecated properties. |
| 3113 | if (scalar keys %deprecated_tags) { |
| 3114 | my $seen_deprecated = 0; |
| 3115 | foreach my $enum (@enums) { |
| 3116 | if (grep { $_ eq $enum } keys %deprecated_tags) { |
| 3117 | |
| 3118 | # Change the enum name for this deprecated property to a |
| 3119 | # munged one to act as a placeholder in the typedef. Then |
| 3120 | # make the real name be a #define whose value is such that |
| 3121 | # its modulus with the number of enums yields the index into |
| 3122 | # the table occupied by the placeholder. And so that dividing |
| 3123 | # the #define value by the table length gives an index into |
| 3124 | # the table of deprecation messages for the corresponding |
| 3125 | # warning. |
| 3126 | my $revised_enum = "${enum}_perl_aux"; |
| 3127 | if (! $seen_deprecated) { |
| 3128 | $seen_deprecated = 1; |
| 3129 | print $out_fh "\n"; |
| 3130 | } |
| 3131 | print $out_fh "#define $enum ($revised_enum + (MAX_UNI_KEYWORD_INDEX * $deprecated_tags{$enum}))\n"; |
| 3132 | $enum = $revised_enum; |
| 3133 | } |
| 3134 | } |
| 3135 | } |
| 3136 | |
| 3137 | print $out_fh "\ntypedef enum {\n\tPERL_BIN_PLACEHOLDER = 0,", |
| 3138 | " /* So no real value is zero */\n\t"; |
| 3139 | print $out_fh join ",\n\t", @enums; |
| 3140 | print $out_fh "\n"; |
| 3141 | print $out_fh "} binary_invlist_enum;\n"; |
| 3142 | print $out_fh "\n#define MAX_UNI_KEYWORD_INDEX $enums[-1]\n"; |
| 3143 | |
| 3144 | switch_pound_if ('binary_property_tables', 'PERL_IN_REGCOMP_C'); |
| 3145 | |
| 3146 | output_table_header($out_fh, "UV *", "uni_prop_ptrs"); |
| 3147 | print $out_fh "\tNULL,\t/* Placeholder */\n"; |
| 3148 | print $out_fh "\t"; |
| 3149 | print $out_fh join ",\n\t", @invlist_names; |
| 3150 | print $out_fh "\n"; |
| 3151 | |
| 3152 | output_table_trailer(); |
| 3153 | |
| 3154 | switch_pound_if ('synonym defines', 'PERL_IN_REGCOMP_C'); |
| 3155 | |
| 3156 | print $out_fh join "\n", "\n", |
| 3157 | #'# ifdef DOINIT', |
| 3158 | #"\n", |
| 3159 | "/* Synonyms for perl properties */", |
| 3160 | @perl_prop_synonyms, |
| 3161 | #"\n", |
| 3162 | #"# endif /* DOINIT */", |
| 3163 | "\n"; |
| 3164 | |
| 3165 | switch_pound_if ('Valid property_values', 'PERL_IN_REGCOMP_C'); |
| 3166 | |
| 3167 | # Each entry is a pointer to a table of property values for some property. |
| 3168 | # (Other properties may share this table. The next two data structures allow |
| 3169 | # this sharing to be implemented.) |
| 3170 | my @values_tables = "NULL /* Placeholder so zero index is an error */"; |
| 3171 | |
| 3172 | # Keys are all the values of a property, strung together. The value of each |
| 3173 | # key is its index in @values_tables. This is because many properties have |
| 3174 | # the same values, and this allows the data to appear just once. |
| 3175 | my %joined_values; |
| 3176 | |
| 3177 | # #defines for indices into @values_tables, so can have synonyms resolved by |
| 3178 | # the C compiler. |
| 3179 | my @values_indices; |
| 3180 | |
| 3181 | print "Computing short unicode properties\n" if DEBUG; |
| 3182 | # Go through each property which is specifiable by \p{prop=value}, and create |
| 3183 | # a hash with the keys being the canonicalized short property names, and the |
| 3184 | # values for each property being all possible values that it can take on. |
| 3185 | # Both the full value and its short, canonicalized into lc, sans punctuation |
| 3186 | # version are included. |
| 3187 | my %all_values; |
| 3188 | for my $property (sort { prop_name_for_cmp($a) cmp prop_name_for_cmp($b) } |
| 3189 | uniques @equals_properties) |
| 3190 | { |
| 3191 | # Get and canonicalize the short name for this property. |
| 3192 | my ($short_name) = prop_aliases($property); |
| 3193 | $short_name = lc $short_name; |
| 3194 | $short_name =~ s/[ _-]//g; |
| 3195 | |
| 3196 | # Now look at each value this property can take on |
| 3197 | foreach my $value (prop_values($short_name)) { |
| 3198 | |
| 3199 | # And for each value, look at each synonym for it |
| 3200 | foreach my $alias (prop_value_aliases($short_name, $value)) { |
| 3201 | |
| 3202 | # Add each synonym |
| 3203 | push @{$all_values{$short_name}}, $alias; |
| 3204 | |
| 3205 | # As well as its canonicalized name. khw made the decision to not |
| 3206 | # support the grandfathered L_ Gc property value |
| 3207 | $alias = lc $alias; |
| 3208 | $alias =~ s/[ _-]//g unless $alias =~ $numeric_re; |
| 3209 | push @{$all_values{$short_name}}, $alias; |
| 3210 | } |
| 3211 | } |
| 3212 | } |
| 3213 | print "Finished computing short unicode properties\n" if DEBUG; |
| 3214 | |
| 3215 | # Also include the old style block names, using the recipe given in |
| 3216 | # Unicode::UCD |
| 3217 | foreach my $block (prop_values('block')) { |
| 3218 | push @{$all_values{'blk'}}, charblock((prop_invlist("block=$block"))[0]); |
| 3219 | } |
| 3220 | |
| 3221 | print "Creating property tables\n" if DEBUG; |
| 3222 | # Now create output tables for each property in @equals_properties (the keys |
| 3223 | # in %all_values) each containing that property's possible values as computed |
| 3224 | # just above. |
| 3225 | PROPERTY: |
| 3226 | for my $property (sort { prop_name_for_cmp($a) cmp prop_name_for_cmp($b) |
| 3227 | or $a cmp $b } keys %all_values) |
| 3228 | { |
| 3229 | @{$all_values{$property}} = uniques(@{$all_values{$property}}); |
| 3230 | |
| 3231 | # String together the values for this property, sorted. This string forms |
| 3232 | # a list definition, with each value as an entry in it, indented on a new |
| 3233 | # line. The sorting is used to find properties that take on the exact |
| 3234 | # same values to share this string. |
| 3235 | my $joined = "\t\""; |
| 3236 | $joined .= join "\",\n\t\"", |
| 3237 | sort { ($a =~ $numeric_re && $b =~ $numeric_re) |
| 3238 | ? eval $a <=> eval $b |
| 3239 | : prop_name_for_cmp($a) cmp prop_name_for_cmp($b) |
| 3240 | or $a cmp $b |
| 3241 | } @{$all_values{$property}}; |
| 3242 | # And add a trailing marker |
| 3243 | $joined .= "\",\n\tNULL\n"; |
| 3244 | |
| 3245 | my $table_name = $table_name_prefix . $property . "_values"; |
| 3246 | my $index_name = "${table_name}_index"; |
| 3247 | |
| 3248 | # Add a rule for the parser that is just an empty value. It will need to |
| 3249 | # know to look up empty things in the prop_value_ptrs[] table. |
| 3250 | |
| 3251 | $keywords{"$property="} = $index_name; |
| 3252 | if (exists $prop_name_aliases{$property}) { |
| 3253 | foreach my $alias (@{$prop_name_aliases{$property}}) { |
| 3254 | $keywords{"$alias="} = $index_name; |
| 3255 | } |
| 3256 | } |
| 3257 | |
| 3258 | # Also create rules for the synonyms of this property to point to the same |
| 3259 | # thing |
| 3260 | |
| 3261 | # If this property's values are the same as one we've already computed, |
| 3262 | # use that instead of creating a duplicate. But we add a #define to point |
| 3263 | # to the proper one. |
| 3264 | if (exists $joined_values{$joined}) { |
| 3265 | push @values_indices, "#define $index_name $joined_values{$joined}\n"; |
| 3266 | next PROPERTY; |
| 3267 | } |
| 3268 | |
| 3269 | # And this property, now known to have unique values from any other seen |
| 3270 | # so far is about to be pushed onto @values_tables. Its index is the |
| 3271 | # current count. |
| 3272 | push @values_indices, "#define $index_name " |
| 3273 | . scalar @values_tables . "\n"; |
| 3274 | $joined_values{$joined} = $index_name; |
| 3275 | push @values_tables, $table_name; |
| 3276 | |
| 3277 | # Create the table for this set of values. |
| 3278 | output_table_header($out_fh, "char *", $table_name); |
| 3279 | print $out_fh $joined; |
| 3280 | output_table_trailer(); |
| 3281 | } # End of loop through the properties, and their values |
| 3282 | |
| 3283 | # We have completely determined the table of the unique property values |
| 3284 | output_table_header($out_fh, "char * const *", |
| 3285 | "${table_name_prefix}prop_value_ptrs"); |
| 3286 | print $out_fh join ",\n", @values_tables; |
| 3287 | print $out_fh "\n"; |
| 3288 | output_table_trailer(); |
| 3289 | |
| 3290 | # And the #defines for the indices in it |
| 3291 | print $out_fh "\n\n", join "", @values_indices; |
| 3292 | |
| 3293 | switch_pound_if('Boundary_pair_tables', 'PERL_IN_REGEXEC_C'); |
| 3294 | |
| 3295 | output_GCB_table(); |
| 3296 | output_LB_table(); |
| 3297 | output_WB_table(); |
| 3298 | |
| 3299 | end_file_pound_if; |
| 3300 | |
| 3301 | print "Computing fold data\n" if DEBUG; |
| 3302 | |
| 3303 | print $out_fh <<"EOF"; |
| 3304 | |
| 3305 | /* More than one code point may have the same code point as their fold. This |
| 3306 | * gives the maximum number in the current Unicode release. (The folded-to |
| 3307 | * code point is not included in this count.) For example, both 'S' and |
| 3308 | * \\x{17F} fold to 's', so the number for that fold is 2. Another way to |
| 3309 | * look at it is the maximum length of all the IVCF_AUX_TABLE's */ |
| 3310 | #define MAX_FOLD_FROMS $max_fold_froms |
| 3311 | EOF |
| 3312 | |
| 3313 | my $sources_list = "lib/unicore/mktables.lst"; |
| 3314 | my @sources = qw(regen/mk_invlists.pl |
| 3315 | lib/unicore/mktables |
| 3316 | lib/Unicode/UCD.pm |
| 3317 | regen/charset_translations.pl |
| 3318 | regen/mk_PL_charclass.pl |
| 3319 | ); |
| 3320 | { |
| 3321 | # Depend on mktables’ own sources. It’s a shorter list of files than |
| 3322 | # those that Unicode::UCD uses. Some may not actually have an effect on |
| 3323 | # the output of this program, but easier to just include all of them, and |
| 3324 | # no real harm in doing so, as it is rare for one such to change without |
| 3325 | # the others doing so as well. |
| 3326 | if (! open my $mktables_list, '<', $sources_list) { |
| 3327 | |
| 3328 | # This should force a rebuild once $sources_list exists |
| 3329 | push @sources, $sources_list; |
| 3330 | } |
| 3331 | else { |
| 3332 | while(<$mktables_list>) { |
| 3333 | last if /===/; |
| 3334 | chomp; |
| 3335 | push @sources, "lib/unicore/$_" if /^[^#]/; |
| 3336 | } |
| 3337 | } |
| 3338 | } |
| 3339 | |
| 3340 | read_only_bottom_close_and_rename($out_fh, \@sources); |
| 3341 | |
| 3342 | my %name_to_index; |
| 3343 | for my $i (0 .. @enums - 1) { |
| 3344 | my $loose_name = $enums[$i] =~ s/^$table_name_prefix//r; |
| 3345 | $loose_name = lc $loose_name; |
| 3346 | $loose_name =~ s/__/=/; |
| 3347 | $loose_name =~ s/_dot_/./; |
| 3348 | $loose_name =~ s/_slash_/\//g; |
| 3349 | $name_to_index{$loose_name} = $i + 1; |
| 3350 | } |
| 3351 | # unsanitize, exclude &, maybe add these before sanitize |
| 3352 | for my $i (0 .. @perl_prop_synonyms - 1) { |
| 3353 | my $loose_name_pair = $perl_prop_synonyms[$i] =~ s/#\s*define\s*//r; |
| 3354 | $loose_name_pair =~ s/\b$table_name_prefix//g; |
| 3355 | $loose_name_pair = lc $loose_name_pair; |
| 3356 | $loose_name_pair =~ s/__/=/g; |
| 3357 | $loose_name_pair =~ s/_dot_/./g; |
| 3358 | $loose_name_pair =~ s/_slash_/\//g; |
| 3359 | my ($synonym, $primary) = split / +/, $loose_name_pair; |
| 3360 | $name_to_index{$synonym} = $name_to_index{$primary}; |
| 3361 | } |
| 3362 | |
| 3363 | my $uni_pl = open_new('lib/unicore/uni_keywords.pl', '>', |
| 3364 | {style => '*', by => 'regen/mk_invlists.pl', |
| 3365 | from => "Unicode::UCD"}); |
| 3366 | { |
| 3367 | print $uni_pl "\%Unicode::UCD::uni_prop_ptrs_indices = (\n"; |
| 3368 | for my $name (sort keys %name_to_index) { |
| 3369 | print $uni_pl " '$name' => $name_to_index{$name},\n"; |
| 3370 | } |
| 3371 | print $uni_pl ");\n\n1;\n"; |
| 3372 | } |
| 3373 | |
| 3374 | read_only_bottom_close_and_rename($uni_pl, \@sources); |
| 3375 | |
| 3376 | print "Computing minimal perfect hash for unicode properties.\n" if DEBUG; |
| 3377 | |
| 3378 | if (my $file= $ENV{DUMP_KEYWORDS_FILE}) { |
| 3379 | require Data::Dumper; |
| 3380 | |
| 3381 | open my $ofh, ">", $file |
| 3382 | or die "Failed to open DUMP_KEYWORDS_FILE '$file' for write: $!"; |
| 3383 | print $ofh Data::Dumper->new([\%keywords],['*keywords']) |
| 3384 | ->Sortkeys(1)->Useqq(1)->Dump(); |
| 3385 | close $ofh; |
| 3386 | print "Wrote keywords to '$file'.\n"; |
| 3387 | } |
| 3388 | |
| 3389 | my $keywords_fh = open_new('uni_keywords.h', '>', |
| 3390 | {style => '*', by => 'regen/mk_invlists.pl', |
| 3391 | from => "mph.pl"}); |
| 3392 | |
| 3393 | print $keywords_fh "\n#if defined(PERL_CORE) || defined(PERL_EXT_RE_BUILD)\n\n"; |
| 3394 | |
| 3395 | my $mph= MinimalPerfectHash->new( |
| 3396 | source_hash => \%keywords, |
| 3397 | match_name => "match_uniprop", |
| 3398 | simple_split => $ENV{SIMPLE_SPLIT} // 0, |
| 3399 | randomize_squeeze => $ENV{RANDOMIZE_SQUEEZE} // 1, |
| 3400 | max_same_in_squeeze => $ENV{MAX_SAME} // 5, |
| 3401 | srand_seed => (lc($ENV{SRAND_SEED}//"") eq "auto") |
| 3402 | ? undef |
| 3403 | : $ENV{SRAND_SEED} // 1785235451, # I let perl pick a number |
| 3404 | ); |
| 3405 | $mph->make_mph_with_split_keys(); |
| 3406 | print $keywords_fh $mph->make_algo(); |
| 3407 | |
| 3408 | print $keywords_fh "\n#endif /* #if defined(PERL_CORE) || defined(PERL_EXT_RE_BUILD) */\n"; |
| 3409 | |
| 3410 | push @sources, 'regen/mph.pl'; |
| 3411 | read_only_bottom_close_and_rename($keywords_fh, \@sources); |