xref: /openbsd-src/gnu/usr.bin/perl/lib/_charnames.pm (revision 50b7afb2c2c0993b0894d4e34bf857cb13ed9c80)
1# !!!!!!!   INTERNAL PERL USE ONLY   !!!!!!!
2# This helper module is for internal use by core Perl only.  This module is
3# subject to change or removal at any time without notice.  Don't use it
4# directly.  Use the public <charnames> module instead.
5
6package _charnames;
7use strict;
8use warnings;
9use File::Spec;
10our $VERSION = '1.36';
11use unicore::Name;    # mktables-generated algorithmically-defined names
12
13use bytes ();          # for $bytes::hint_bits
14use re "/aa";          # Everything in here should be ASCII
15
16$Carp::Internal{ (__PACKAGE__) } = 1;
17
18# Translate between Unicode character names and their code points.  This is a
19# submodule of package <charnames>, used to allow \N{...} to be autoloaded,
20# but it was decided not to autoload the various functions in charnames; the
21# splitting allows this behavior.
22#
23# The official names with their code points are stored in a table in
24# lib/unicore/Name.pl which is read in as a large string (almost 3/4 Mb in
25# Unicode 6.0).  Each code point/name combination is separated by a \n in the
26# string.  (Some of the CJK and the Hangul syllable names are determined
27# instead algorithmically via subroutines stored instead in
28# lib/unicore/Name.pm).  Because of the large size of this table, it isn't
29# converted into hashes for faster lookup.
30#
31# But, user defined aliases are stored in their own hashes, as are Perl
32# extensions to the official names.  These are checked first before looking at
33# the official table.
34#
35# Basically, the table is grepped for the input code point (viacode()) or
36# name (the other functions), and the corresponding value on the same line is
37# returned.  The grepping is done by turning the input into a regular
38# expression.  Thus, the same table does double duty, used by both name and
39# code point lookup.  (If we were to have hashes, we would need two, one for
40# each lookup direction.)
41#
42# For loose name matching, the logical thing would be to have a table
43# with all the ignorable characters squeezed out, and then grep it with the
44# similiarly-squeezed input name.  (And this is in fact how the lookups are
45# done with the small Perl extension hashes.)  But since we need to be able to
46# go from code point to official name, the original table would still need to
47# exist.  Due to the large size of the table, it was decided to not read
48# another very large string into memory for a second table.  Instead, the
49# regular expression of the input name is modified to have optional spaces and
50# dashes between characters.  For example, in strict matching, the regular
51# expression would be:
52#   qr/\tDIGIT ONE$/m
53# Under loose matching, the blank would be squeezed out, and the re would be:
54#   qr/\tD[- ]?I[- ]?G[- ]?I[- ]?T[- ]?O[- ]?N[- ]?E$/m
55# which matches a blank or dash between any characters in the official table.
56#
57# This is also how script lookup is done.  Basically the re looks like
58#   qr/ (?:LATIN|GREEK|CYRILLIC) (?:SMALL )?LETTER $name/
59# where $name is the loose or strict regex for the remainder of the name.
60
61# The hashes are stored as utf8 strings.  This makes it easier to deal with
62# sequences.  I (khw) also tried making Name.pl utf8, but it slowed things
63# down by a factor of 7.  I then tried making Name.pl store the ut8
64# equivalents but not calling them utf8.  That led to similar speed as leaving
65# it alone, but since that is harder for a human to parse, I left it as-is.
66
67my %system_aliases = (
68
69    'SINGLE-SHIFT 2'                => pack("U", 0x8E),
70    'SINGLE-SHIFT 3'                => pack("U", 0x8F),
71    'PRIVATE USE 1'                 => pack("U", 0x91),
72    'PRIVATE USE 2'                 => pack("U", 0x92),
73);
74
75# These are the aliases above that differ under :loose and :full matching
76# because the :full versions have blanks or hyphens in them.
77#my %loose_system_aliases = (
78#);
79
80#my %deprecated_aliases;
81#$deprecated_aliases{'BELL'} = pack("U", 0x07) if $^V lt v5.17.0;
82
83#my %loose_deprecated_aliases = (
84#);
85
86# These are special cased in :loose matching, differing only in a medial
87# hyphen
88my $HANGUL_JUNGSEONG_O_E_utf8 = pack("U", 0x1180);
89my $HANGUL_JUNGSEONG_OE_utf8 = pack("U", 0x116C);
90
91
92my $txt;  # The table of official character names
93
94my %full_names_cache; # Holds already-looked-up names, so don't have to
95# re-look them up again.  The previous versions of charnames had scoping
96# bugs.  For example if we use script A in one scope and find and cache
97# what Z resolves to, we can't use that cache in a different scope that
98# uses script B instead of A, as Z might be an entirely different letter
99# there; or there might be different aliases in effect in different
100# scopes, or :short may be in effect or not effect in different scopes,
101# or various combinations thereof.  This was solved in this version
102# mostly by moving things to %^H.  But some things couldn't be moved
103# there.  One of them was the cache of runtime looked-up names, in part
104# because %^H is read-only at runtime.  I (khw) don't know why the cache
105# was run-time only in the previous versions: perhaps oversight; perhaps
106# that compile time looking doesn't happen in a loop so didn't think it
107# was worthwhile; perhaps not wanting to make the cache too large.  But
108# I decided to make it compile time as well; this could easily be
109# changed.
110# Anyway, this hash is not scoped, and is added to at runtime.  It
111# doesn't have scoping problems because the data in it is restricted to
112# official names, which are always invariant, and we only set it and
113# look at it at during :full lookups, so is unaffected by any other
114# scoped options.  I put this in to maintain parity with the older
115# version.  If desired, a %short_names cache could also be made, as well
116# as one for each script, say in %script_names_cache, with each key
117# being a hash for a script named in a 'use charnames' statement.  I
118# decided not to do that for now, just because it's added complication,
119# and because I'm just trying to maintain parity, not extend it.
120
121# Like %full_names_cache, but for use when :loose is in effect.  There needs
122# to be two caches because :loose may not be in effect for a scope, and a
123# loose name could inappropriately be returned when only exact matching is
124# called for.
125my %loose_names_cache;
126
127# Designed so that test decimal first, and then hex.  Leading zeros
128# imply non-decimal, as do non-[0-9]
129my $decimal_qr = qr/^[1-9]\d*$/;
130
131# Returns the hex number in $1.
132my $hex_qr = qr/^(?:[Uu]\+|0[xX])?([[:xdigit:]]+)$/;
133
134sub croak
135{
136  require Carp; goto &Carp::croak;
137} # croak
138
139sub carp
140{
141  require Carp; goto &Carp::carp;
142} # carp
143
144sub alias (@) # Set up a single alias
145{
146  my @errors;
147
148  my $alias = ref $_[0] ? $_[0] : { @_ };
149  foreach my $name (sort keys %$alias) {  # Sort only because it helps having
150                                          # deterministic output for
151                                          # t/lib/charnames/alias
152    my $value = $alias->{$name};
153    next unless defined $value;          # Omit if screwed up.
154
155    # Is slightly slower to just after this statement see if it is
156    # decimal, since we already know it is after having converted from
157    # hex, but makes the code easier to maintain, and is called
158    # infrequently, only at compile-time
159    if ($value !~ $decimal_qr && $value =~ $hex_qr) {
160      $value = CORE::hex $1;
161    }
162    if ($value =~ $decimal_qr) {
163        no warnings qw(non_unicode surrogate nonchar); # Allow any of these
164        $^H{charnames_ord_aliases}{$name} = pack("U", $value);
165
166        # Use a canonical form.
167        $^H{charnames_inverse_ords}{sprintf("%05X", $value)} = $name;
168    }
169    else {
170        # This regex needs to be sync'd with the code in toke.c that checks
171        # for the same thing
172        if ($name !~ / ^
173                       \p{_Perl_Charname_Begin}
174                       \p{_Perl_Charname_Continue}*
175                       $ /x) {
176
177          push @errors, $name;
178        }
179        else {
180          $^H{charnames_name_aliases}{$name} = $value;
181
182          if (warnings::enabled('deprecated')) {
183            if ($name =~ / ( .* \s ) ( \s* ) $ /x) {
184              carp "Trailing white-space in a charnames alias definition is deprecated; marked by <-- HERE in '$1 <-- HERE " . $2 . "'";
185            }
186
187            # Use '+' instead of '*' in this regex, because any trailing
188            # blanks have already been warned about.
189            if ($name =~ / ( .*? \s{2} ) ( .+ ) /x) {
190              carp "A sequence of multiple spaces in a charnames alias definition is deprecated; marked by <-- HERE in '$1 <-- HERE " . $2 . "'";
191            }
192          }
193        }
194    }
195  }
196
197  # We find and output all errors from this :alias definition, rather than
198  # failing on the first one, so fewer runs are needed to get it to compile
199  if (@errors) {
200    foreach my $name (@errors) {
201      my $ok = "";
202      $ok = $1 if $name =~ / ^ ( \p{Alpha} [-\p{XPosixWord} ():\xa0]* ) /x;
203      my $first_bad = substr($name, length($ok), 1);
204      $name = "Invalid character in charnames alias definition; marked by <-- HERE in '$ok$first_bad<-- HERE " . substr($name, length($ok) + 1) . "'";
205    }
206    croak join "\n", @errors;
207  }
208
209  return;
210} # alias
211
212sub not_legal_use_bytes_msg {
213  my ($name, $utf8) = @_;
214  my $return;
215
216  if (length($utf8) == 1) {
217    $return = sprintf("Character 0x%04x with name '%s' is", ord $utf8, $name);
218  } else {
219    $return = sprintf("String with name '%s' (and ordinals %s) contains character(s)", $name, join(" ", map { sprintf "0x%04X", ord $_ } split(//, $utf8)));
220  }
221  return $return . " above 0xFF with 'use bytes' in effect";
222}
223
224sub alias_file ($)  # Reads a file containing alias definitions
225{
226  my ($arg, $file) = @_;
227  if (-f $arg && File::Spec->file_name_is_absolute ($arg)) {
228    $file = $arg;
229  }
230  elsif ($arg =~ m/ ^ \p{_Perl_IDStart} \p{_Perl_IDCont}* $/x) {
231    $file = "unicore/${arg}_alias.pl";
232  }
233  else {
234    croak "Charnames alias file names can only have identifier characters";
235  }
236  if (my @alias = do $file) {
237    @alias == 1 && !defined $alias[0] and
238      croak "$file cannot be used as alias file for charnames";
239    @alias % 2 and
240      croak "$file did not return a (valid) list of alias pairs";
241    alias (@alias);
242    return (1);
243  }
244  0;
245} # alias_file
246
247# For use when don't import anything.  This structure must be kept in
248# sync with the one that import() fills up.
249my %dummy_H = (
250                charnames_stringified_names => "",
251                charnames_stringified_ords => "",
252                charnames_scripts => "",
253                charnames_full => 1,
254                charnames_loose => 0,
255                charnames_short => 0,
256              );
257
258
259sub lookup_name ($$$) {
260  my ($name, $wants_ord, $runtime) = @_;
261
262  # Lookup the name or sequence $name in the tables.  If $wants_ord is false,
263  # returns the string equivalent of $name; if true, returns the ordinal value
264  # instead, but in this case $name must not be a sequence; otherwise undef is
265  # returned and a warning raised.  $runtime is 0 if compiletime, otherwise
266  # gives the number of stack frames to go back to get the application caller
267  # info.
268  # If $name is not found, returns undef in runtime with no warning; and in
269  # compiletime, the Unicode replacement character, with a warning.
270
271  # It looks first in the aliases, then in the large table of official Unicode
272  # names.
273
274  my $utf8;       # The string result
275  my $save_input;
276
277  if ($runtime) {
278
279    my $hints_ref = (caller($runtime))[10];
280
281    # If we didn't import anything (which happens with 'use charnames ()',
282    # substitute a dummy structure.
283    $hints_ref = \%dummy_H if ! defined $hints_ref
284                              || (! defined $hints_ref->{charnames_full}
285                                  && ! defined $hints_ref->{charnames_loose});
286
287    # At runtime, but currently not at compile time, $^H gets
288    # stringified, so un-stringify back to the original data structures.
289    # These get thrown away by perl before the next invocation
290    # Also fill in the hash with the non-stringified data.
291    # N.B.  New fields must be also added to %dummy_H
292
293    %{$^H{charnames_name_aliases}} = split ',',
294                                      $hints_ref->{charnames_stringified_names};
295    %{$^H{charnames_ord_aliases}} = split ',',
296                                      $hints_ref->{charnames_stringified_ords};
297    $^H{charnames_scripts} = $hints_ref->{charnames_scripts};
298    $^H{charnames_full} = $hints_ref->{charnames_full};
299    $^H{charnames_loose} = $hints_ref->{charnames_loose};
300    $^H{charnames_short} = $hints_ref->{charnames_short};
301  }
302
303  my $loose = $^H{charnames_loose};
304  my $lookup_name;  # Input name suitably modified for grepping for in the
305                    # table
306
307  # User alias should be checked first or else can't override ours, and if we
308  # were to add any, could conflict with theirs.
309  if (exists $^H{charnames_ord_aliases}{$name}) {
310    $utf8 = $^H{charnames_ord_aliases}{$name};
311  }
312  elsif (exists $^H{charnames_name_aliases}{$name}) {
313    $name = $^H{charnames_name_aliases}{$name};
314    $save_input = $lookup_name = $name;  # Cache the result for any error
315                                         # message
316    # The aliases are documented to not match loosely, so change loose match
317    # into full.
318    if ($loose) {
319      $loose = 0;
320      $^H{charnames_full} = 1;
321    }
322  }
323  else {
324
325    # Here, not a user alias.  That means that loose matching may be in
326    # effect; will have to modify the input name.
327    $lookup_name = $name;
328    if ($loose) {
329      $lookup_name = uc $lookup_name;
330
331      # Squeeze out all underscores
332      $lookup_name =~ s/_//g;
333
334      # Remove all medial hyphens
335      $lookup_name =~ s/ (?<= \S  ) - (?= \S  )//gx;
336
337      # Squeeze out all spaces
338      $lookup_name =~ s/\s//g;
339    }
340
341    # Here, $lookup_name has been modified as necessary for looking in the
342    # hashes.  Check the system alias files next.  Most of these aliases are
343    # the same for both strict and loose matching.  To save space, the ones
344    # which differ are in their own separate hash, which is checked if loose
345    # matching is selected and the regular match fails.  To save time, the
346    # loose hashes could be expanded to include all aliases, and there would
347    # only have to be one check.  But if someone specifies :loose, they are
348    # interested in convenience over speed, and the time for this second check
349    # is miniscule compared to the rest of the routine.
350    if (exists $system_aliases{$lookup_name}) {
351      $utf8 = $system_aliases{$lookup_name};
352    }
353    # There are currently no entries in this hash, so don't waste time looking
354    # for them.  But the code is retained for the unlikely possibility that
355    # some will be added in the future.
356#    elsif ($loose && exists $loose_system_aliases{$lookup_name}) {
357#      $utf8 = $loose_system_aliases{$lookup_name};
358#    }
359#    if (exists $deprecated_aliases{$lookup_name}) {
360#      require warnings;
361#      warnings::warnif('deprecated',
362#                       "Unicode character name \"$name\" is deprecated, use \""
363#                       . viacode(ord $deprecated_aliases{$lookup_name})
364#                       . "\" instead");
365#      $utf8 = $deprecated_aliases{$lookup_name};
366#    }
367    # There are currently no entries in this hash, so don't waste time looking
368    # for them.  But the code is retained for the unlikely possibility that
369    # some will be added in the future.
370#    elsif ($loose && exists $loose_deprecated_aliases{$lookup_name}) {
371#      require warnings;
372#      warnings::warnif('deprecated',
373#                       "Unicode character name \"$name\" is deprecated, use \""
374#                       . viacode(ord $loose_deprecated_aliases{$lookup_name})
375#                       . "\" instead");
376#      $utf8 = $loose_deprecated_aliases{$lookup_name};
377#    }
378  }
379
380  my @off;  # Offsets into table of pattern match begin and end
381
382  # If haven't found it yet...
383  if (! defined $utf8) {
384
385    # See if has looked this input up earlier.
386    if (! $loose && $^H{charnames_full} && exists $full_names_cache{$name}) {
387      $utf8 = $full_names_cache{$name};
388    }
389    elsif ($loose && exists $loose_names_cache{$name}) {
390      $utf8 = $loose_names_cache{$name};
391    }
392    else { # Here, must do a look-up
393
394      # If full or loose matching succeeded, points to where to cache the
395      # result
396      my $cache_ref;
397
398      ## Suck in the code/name list as a big string.
399      ## Lines look like:
400      ##     "00052\tLATIN CAPITAL LETTER R\n"
401      # or
402      #      "0052 0303\tLATIN CAPITAL LETTER R WITH TILDE\n"
403      $txt = do "unicore/Name.pl" unless $txt;
404
405      ## @off will hold the index into the code/name string of the start and
406      ## end of the name as we find it.
407
408      ## If :loose, look for a loose match; if :full, look for the name
409      ## exactly
410      # First, see if the name is one which is algorithmically determinable.
411      # The subroutine is included in Name.pl.  The table contained in
412      # $txt doesn't contain these.  Experiments show that checking
413      # for these before checking for the regular names has no
414      # noticeable impact on performance for the regular names, but
415      # the other way around slows down finding these immensely.
416      # Algorithmically determinables are not placed in the cache because
417      # that uses up memory, and finding these again is fast.
418      if (($loose || $^H{charnames_full})
419          && (defined (my $ord = charnames::name_to_code_point_special($lookup_name, $loose))))
420      {
421        $utf8 = pack("U", $ord);
422      }
423      else {
424
425        # Not algorithmically determinable; look up in the table.  The name
426        # will be turned into a regex, so quote any meta characters.
427        $lookup_name = quotemeta $lookup_name;
428
429        if ($loose) {
430
431          # For loose matches, $lookup_name has already squeezed out the
432          # non-essential characters.  We have to add in code to make the
433          # squeezed version match the non-squeezed equivalent in the table.
434          # The only remaining hyphens are ones that start or end a word in
435          # the original.  They have been quoted in $lookup_name so they look
436          # like "\-".  Change all other characters except the backslash
437          # quotes for any metacharacters, and the final character, so that
438          # e.g., COLON gets transformed into: /C[- ]?O[- ]?L[- ]?O[- ]?N/
439          $lookup_name =~ s/ (?! \\ -)    # Don't do this to the \- sequence
440                             ( [^-\\] )   # Nor the "-" within that sequence,
441                                          # nor the "\" that quotes metachars,
442                                          # but otherwise put the char into $1
443                             (?=.)        # And don't do it for the final char
444                           /$1\[- \]?/gx; # And add an optional blank or
445                                          # '-' after each $1 char
446
447          # Those remaining hyphens were originally at the beginning or end of
448          # a word, so they can match either a blank before or after, but not
449          # both.  (Keep in mind that they have been quoted, so are a '\-'
450          # sequence)
451          $lookup_name =~ s/\\ -/(?:- | -)/xg;
452        }
453
454        # Do the lookup in the full table if asked for, and if succeeds
455        # save the offsets and set where to cache the result.
456        if (($loose || $^H{charnames_full}) && $txt =~ /\t$lookup_name$/m) {
457          @off = ($-[0] + 1, $+[0]);    # The 1 is for the tab
458          $cache_ref = ($loose) ? \%loose_names_cache : \%full_names_cache;
459        }
460        else {
461
462          # Here, didn't look for, or didn't find the name.
463          # If :short is allowed, see if input is like "greek:Sigma".
464          # Keep in mind that $lookup_name has had the metas quoted.
465          my $scripts_trie = "";
466          my $name_has_uppercase;
467          if (($^H{charnames_short})
468              && $lookup_name =~ /^ (?: \\ \s)*   # Quoted space
469                                    (.+?)         # $1 = the script
470                                    (?: \\ \s)*
471                                    \\ :          # Quoted colon
472                                    (?: \\ \s)*
473                                    (.+?)         # $2 = the name
474                                    (?: \\ \s)* $
475                                  /xs)
476          {
477              # Even in non-loose matching, the script traditionally has been
478              # case insensitve
479              $scripts_trie = "\U$1";
480              $lookup_name = $2;
481
482              # Use original name to find its input casing, but ignore the
483              # script part of that to make the determination.
484              $save_input = $name if ! defined $save_input;
485              $name =~ s/.*?://;
486              $name_has_uppercase = $name =~ /[[:upper:]]/;
487          }
488          else { # Otherwise look in allowed scripts
489              $scripts_trie = $^H{charnames_scripts};
490
491              # Use original name to find its input casing
492              $name_has_uppercase = $name =~ /[[:upper:]]/;
493          }
494
495          my $case = $name_has_uppercase ? "CAPITAL" : "SMALL";
496          return if (! $scripts_trie || $txt !~
497             /\t (?: $scripts_trie ) \ (?:$case\ )? LETTER \ \U$lookup_name $/xm);
498
499          # Here have found the input name in the table.
500          @off = ($-[0] + 1, $+[0]);  # The 1 is for the tab
501        }
502
503        # Here, the input name has been found; we haven't set up the output,
504        # but we know where in the string
505        # the name starts.  The string is set up so that for single characters
506        # (and not named sequences), the name is preceded immediately by a
507        # tab and 5 hex digits for its code, with a \n before those.  Named
508        # sequences won't have the 7th preceding character be a \n.
509        # (Actually, for the very first entry in the table this isn't strictly
510        # true: subtracting 7 will yield -1, and the substr below will
511        # therefore yield the very last character in the table, which should
512        # also be a \n, so the statement works anyway.)
513        if (substr($txt, $off[0] - 7, 1) eq "\n") {
514          $utf8 = pack("U", CORE::hex substr($txt, $off[0] - 6, 5));
515
516          # Handle the single loose matching special case, in which two names
517          # differ only by a single medial hyphen.  If the original had a
518          # hyphen (or more) in the right place, then it is that one.
519          $utf8 = $HANGUL_JUNGSEONG_O_E_utf8
520                  if $loose
521                     && $utf8 eq $HANGUL_JUNGSEONG_OE_utf8
522                     && $name =~ m/O \s* - [-\s]* E/ix;
523                     # Note that this wouldn't work if there were a 2nd
524                     # OE in the name
525        }
526        else {
527
528          # Here, is a named sequence.  Need to go looking for the beginning,
529          # which is just after the \n from the previous entry in the table.
530          # The +1 skips past that newline, or, if the rindex() fails, to put
531          # us to an offset of zero.
532          my $charstart = rindex($txt, "\n", $off[0] - 7) + 1;
533          $utf8 = pack("U*", map { CORE::hex }
534              split " ", substr($txt, $charstart, $off[0] - $charstart - 1));
535        }
536      }
537
538      # Cache the input so as to not have to search the large table
539      # again, but only if it came from the one search that we cache.
540      # (Haven't bothered with the pain of sorting out scoping issues for the
541      # scripts searches.)
542      $cache_ref->{$name} = $utf8 if defined $cache_ref;
543    }
544  }
545
546
547  # Here, have the utf8.  If the return is to be an ord, must be any single
548  # character.
549  if ($wants_ord) {
550    return ord($utf8) if length $utf8 == 1;
551  }
552  else {
553
554    # Here, wants string output.  If utf8 is acceptable, just return what
555    # we've got; otherwise attempt to convert it to non-utf8 and return that.
556    my $in_bytes = ($runtime)
557                   ? (caller $runtime)[8] & $bytes::hint_bits
558                   : $^H & $bytes::hint_bits;
559    return $utf8 if (! $in_bytes || utf8::downgrade($utf8, 1)) # The 1 arg
560                                                  # means don't die on failure
561  }
562
563  # Here, there is an error:  either there are too many characters, or the
564  # result string needs to be non-utf8, and at least one character requires
565  # utf8.  Prefer any official name over the input one for the error message.
566  if (@off) {
567    $name = substr($txt, $off[0], $off[1] - $off[0]) if @off;
568  }
569  else {
570    $name = (defined $save_input) ? $save_input : $_[0];
571  }
572
573  if ($wants_ord) {
574    # Only way to get here in this case is if result too long.  Message
575    # assumes that our only caller that requires single char result is
576    # vianame.
577    carp "charnames::vianame() doesn't handle named sequences ($name).  Use charnames::string_vianame() instead";
578    return;
579  }
580
581  # Only other possible failure here is from use bytes.
582  if ($runtime) {
583    carp not_legal_use_bytes_msg($name, $utf8);
584    return;
585  } else {
586    croak not_legal_use_bytes_msg($name, $utf8);
587  }
588
589} # lookup_name
590
591sub charnames {
592
593  # For \N{...}.  Looks up the character name and returns the string
594  # representation of it.
595
596  # The first 0 arg means wants a string returned; the second that we are in
597  # compile time
598  return lookup_name($_[0], 0, 0);
599}
600
601sub import
602{
603  shift; ## ignore class name
604
605  if (not @_) {
606    carp("'use charnames' needs explicit imports list");
607  }
608  $^H{charnames} = \&charnames ;
609  $^H{charnames_ord_aliases} = {};
610  $^H{charnames_name_aliases} = {};
611  $^H{charnames_inverse_ords} = {};
612  # New fields must be added to %dummy_H, and the code in lookup_name()
613  # that copies fields from the runtime structure
614
615  ##
616  ## fill %h keys with our @_ args.
617  ##
618  my ($promote, %h, @args) = (0);
619  while (my $arg = shift) {
620    if ($arg eq ":alias") {
621      @_ or
622        croak ":alias needs an argument in charnames";
623      my $alias = shift;
624      if (ref $alias) {
625        ref $alias eq "HASH" or
626          croak "Only HASH reference supported as argument to :alias";
627        alias ($alias);
628        $promote = 1;
629        next;
630      }
631      if ($alias =~ m{:(\w+)$}) {
632        $1 eq "full" || $1 eq "loose" || $1 eq "short" and
633          croak ":alias cannot use existing pragma :$1 (reversed order?)";
634        alias_file ($1) and $promote = 1;
635        next;
636      }
637      alias_file ($alias) and $promote = 1;
638      next;
639    }
640    if (substr($arg, 0, 1) eq ':'
641      and ! ($arg eq ":full" || $arg eq ":short" || $arg eq ":loose"))
642    {
643      warn "unsupported special '$arg' in charnames";
644      next;
645    }
646    push @args, $arg;
647  }
648
649  @args == 0 && $promote and @args = (":full");
650  @h{@args} = (1) x @args;
651
652  # Don't leave these undefined as are tested for in lookup_names
653  $^H{charnames_full} = delete $h{':full'} || 0;
654  $^H{charnames_loose} = delete $h{':loose'} || 0;
655  $^H{charnames_short} = delete $h{':short'} || 0;
656  my @scripts = map { uc quotemeta } keys %h;
657
658  ##
659  ## If utf8? warnings are enabled, and some scripts were given,
660  ## see if at least we can find one letter from each script.
661  ##
662  if (warnings::enabled('utf8') && @scripts) {
663    $txt = do "unicore/Name.pl" unless $txt;
664
665    for my $script (@scripts) {
666      if (not $txt =~ m/\t$script (?:CAPITAL |SMALL )?LETTER /) {
667        warnings::warn('utf8',  "No such script: '$script'");
668        $script = quotemeta $script;  # Escape it, for use in the re.
669      }
670    }
671  }
672
673  # %^H gets stringified, so serialize it ourselves so can extract the
674  # real data back later.
675  $^H{charnames_stringified_ords} = join ",", %{$^H{charnames_ord_aliases}};
676  $^H{charnames_stringified_names} = join ",", %{$^H{charnames_name_aliases}};
677  $^H{charnames_stringified_inverse_ords} = join ",", %{$^H{charnames_inverse_ords}};
678
679  # Modify the input script names for loose name matching if that is also
680  # specified, similar to the way the base character name is prepared.  They
681  # don't (currently, and hopefully never will) have dashes.  These go into a
682  # regex, and have already been uppercased and quotemeta'd.  Squeeze out all
683  # input underscores, blanks, and dashes.  Then convert so will match a blank
684  # between any characters.
685  if ($^H{charnames_loose}) {
686    for (my $i = 0; $i < @scripts; $i++) {
687      $scripts[$i] =~ s/[_ -]//g;
688      $scripts[$i] =~ s/ ( [^\\] ) (?= . ) /$1\\ ?/gx;
689    }
690  }
691
692  $^H{charnames_scripts} = join "|", @scripts;  # Stringifiy them as a trie
693} # import
694
695# Cache of already looked-up values.  This is set to only contain
696# official values, and user aliases can't override them, so scoping is
697# not an issue.
698my %viacode;
699
700sub viacode {
701
702  # Returns the name of the code point argument
703
704  if (@_ != 1) {
705    carp "charnames::viacode() expects one argument";
706    return;
707  }
708
709  my $arg = shift;
710
711  # This is derived from Unicode::UCD, where it is nearly the same as the
712  # function _getcode(), but here it makes sure that even a hex argument
713  # has the proper number of leading zeros, which is critical in
714  # matching against $txt below
715  # Must check if decimal first; see comments at that definition
716  my $hex;
717  if ($arg =~ $decimal_qr) {
718    $hex = sprintf "%05X", $arg;
719  } elsif ($arg =~ $hex_qr) {
720    # Below is the line that differs from the _getcode() source
721    $hex = sprintf "%05X", hex $1;
722  } else {
723    carp("unexpected arg \"$arg\" to charnames::viacode()");
724    return;
725  }
726
727  return $viacode{$hex} if exists $viacode{$hex};
728
729  my $return;
730
731  # If the code point is above the max in the table, there's no point
732  # looking through it.  Checking the length first is slightly faster
733  if (length($hex) <= 5 || CORE::hex($hex) <= 0x10FFFF) {
734    $txt = do "unicore/Name.pl" unless $txt;
735
736    # See if the name is algorithmically determinable.
737    my $algorithmic = charnames::code_point_to_name_special(CORE::hex $hex);
738    if (defined $algorithmic) {
739      $viacode{$hex} = $algorithmic;
740      return $algorithmic;
741    }
742
743    # Return the official name, if exists.  It's unclear to me (khw) at
744    # this juncture if it is better to return a user-defined override, so
745    # leaving it as is for now.
746    if ($txt =~ m/^$hex\t/m) {
747
748        # The name starts with the next character and goes up to the
749        # next new-line.  Using capturing parentheses above instead of
750        # @+ more than doubles the execution time in Perl 5.13
751        $return = substr($txt, $+[0], index($txt, "\n", $+[0]) - $+[0]);
752
753        # If not one of these 4 code points, return what we've found.
754        if ($hex !~ / ^ 000 (?: 8[014] | 99 ) $ /x) {
755          $viacode{$hex} = $return;
756          return $return;
757        }
758
759        # For backwards compatibility, we don't return the official name of
760        # the 4 code points if there are user-defined aliases for them -- so
761        # continue looking.
762    }
763  }
764
765  # See if there is a user name for it, before giving up completely.
766  # First get the scoped aliases, give up if have none.
767  my $H_ref = (caller(1))[10];
768  return if ! defined $return
769              && (! defined $H_ref
770                  || ! exists $H_ref->{charnames_stringified_inverse_ords});
771
772  my %code_point_aliases;
773  if (defined $H_ref->{charnames_stringified_inverse_ords}) {
774    %code_point_aliases = split ',',
775                          $H_ref->{charnames_stringified_inverse_ords};
776    return $code_point_aliases{$hex} if exists $code_point_aliases{$hex};
777  }
778
779  # Here there is no user-defined alias, return any official one.
780  return $return if defined $return;
781
782  if (CORE::hex($hex) > 0x10FFFF
783      && warnings::enabled('non_unicode'))
784  {
785      carp "Unicode characters only allocated up to U+10FFFF (you asked for U+$hex)";
786  }
787  return;
788
789} # _viacode
790
7911;
792
793# ex: set ts=8 sts=2 sw=2 et:
794