This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
regcomp.c: comment typo and rewording
[perl5.git] / lib / unicore / mktables
CommitLineData
d73e5302 1#!/usr/bin/perl -w
99870f4d
KW
2
3# !!!!!!!!!!!!!! IF YOU MODIFY THIS FILE !!!!!!!!!!!!!!!!!!!!!!!!!
4# Any files created or read by this program should be listed in 'mktables.lst'
5# Use -makelist to regenerate it.
6
23e33b60
KW
7# Needs 'no overloading' to run faster on miniperl. Code commented out at the
8# subroutine objaddr can be used instead to work as far back (untested) as
f998e60c
KW
9# 5.8: needs pack "U". But almost all occurrences of objaddr have been
10# removed in favor of using 'no overloading'. You also would have to go
11# through and replace occurrences like:
ffe43484 12# my $addr = do { no overloading; pack 'J', $self; }
f998e60c
KW
13# with
14# my $addr = main::objaddr $self;
6c68572b 15# (or reverse commit 9b01bafde4b022706c3d6f947a0963f821b2e50b
051df77b
NC
16# that instituted the change to main::objaddr, and subsequent commits that
17# changed 0+$self to pack 'J', $self.)
6c68572b 18
cdcef19a 19my $start_time;
98dc9551 20BEGIN { # Get the time the script started running; do it at compilation to
cdcef19a
KW
21 # get it as close as possible
22 $start_time= time;
23}
24
23e33b60 25require 5.010_001;
d73e5302 26use strict;
99870f4d 27use warnings;
cf25bb62 28use Carp;
bd9ebcfd 29use Config;
99870f4d
KW
30use File::Find;
31use File::Path;
d07a55ed 32use File::Spec;
99870f4d 33use Text::Tabs;
6b64c11c 34use re "/aa";
99870f4d
KW
35
36sub DEBUG () { 0 } # Set to 0 for production; 1 for development
bd9ebcfd 37my $debugging_build = $Config{"ccflags"} =~ /-DDEBUGGING/;
99870f4d 38
74cd47d0
KW
39sub NON_ASCII_PLATFORM { ord("A") != 65 }
40
99870f4d
KW
41##########################################################################
42#
43# mktables -- create the runtime Perl Unicode files (lib/unicore/.../*.pl),
44# from the Unicode database files (lib/unicore/.../*.txt), It also generates
232ed87f 45# a pod file and .t files, depending on option parameters.
99870f4d
KW
46#
47# The structure of this file is:
48# First these introductory comments; then
49# code needed for everywhere, such as debugging stuff; then
50# code to handle input parameters; then
51# data structures likely to be of external interest (some of which depend on
52# the input parameters, so follows them; then
53# more data structures and subroutine and package (class) definitions; then
54# the small actual loop to process the input files and finish up; then
55# a __DATA__ section, for the .t tests
56#
232ed87f
KW
57# This program works on all releases of Unicode so far. The outputs have been
58# scrutinized most intently for release 5.1. The others have been checked for
59# somewhat more than just sanity. It can handle all non-provisional Unicode
60# character properties in those releases.
99870f4d 61#
99870f4d
KW
62# This program is mostly about Unicode character (or code point) properties.
63# A property describes some attribute or quality of a code point, like if it
64# is lowercase or not, its name, what version of Unicode it was first defined
65# in, or what its uppercase equivalent is. Unicode deals with these disparate
66# possibilities by making all properties into mappings from each code point
67# into some corresponding value. In the case of it being lowercase or not,
68# the mapping is either to 'Y' or 'N' (or various synonyms thereof). Each
69# property maps each Unicode code point to a single value, called a "property
232ed87f
KW
70# value". (Some more recently defined properties, map a code point to a set
71# of values.)
99870f4d
KW
72#
73# When using a property in a regular expression, what is desired isn't the
74# mapping of the code point to its property's value, but the reverse (or the
75# mathematical "inverse relation"): starting with the property value, "Does a
76# code point map to it?" These are written in a "compound" form:
77# \p{property=value}, e.g., \p{category=punctuation}. This program generates
78# files containing the lists of code points that map to each such regular
79# expression property value, one file per list
80#
81# There is also a single form shortcut that Perl adds for many of the commonly
82# used properties. This happens for all binary properties, plus script,
83# general_category, and block properties.
84#
85# Thus the outputs of this program are files. There are map files, mostly in
86# the 'To' directory; and there are list files for use in regular expression
87# matching, all in subdirectories of the 'lib' directory, with each
88# subdirectory being named for the property that the lists in it are for.
89# Bookkeeping, test, and documentation files are also generated.
90
91my $matches_directory = 'lib'; # Where match (\p{}) files go.
92my $map_directory = 'To'; # Where map files go.
93
94# DATA STRUCTURES
95#
96# The major data structures of this program are Property, of course, but also
97# Table. There are two kinds of tables, very similar to each other.
98# "Match_Table" is the data structure giving the list of code points that have
99# a particular property value, mentioned above. There is also a "Map_Table"
100# data structure which gives the property's mapping from code point to value.
101# There are two structures because the match tables need to be combined in
102# various ways, such as constructing unions, intersections, complements, etc.,
103# and the map ones don't. And there would be problems, perhaps subtle, if
104# a map table were inadvertently operated on in some of those ways.
105# The use of separate classes with operations defined on one but not the other
106# prevents accidentally confusing the two.
107#
108# At the heart of each table's data structure is a "Range_List", which is just
109# an ordered list of "Ranges", plus ancillary information, and methods to
110# operate on them. A Range is a compact way to store property information.
111# Each range has a starting code point, an ending code point, and a value that
112# is meant to apply to all the code points between the two end points,
113# inclusive. For a map table, this value is the property value for those
114# code points. Two such ranges could be written like this:
115# 0x41 .. 0x5A, 'Upper',
116# 0x61 .. 0x7A, 'Lower'
117#
118# Each range also has a type used as a convenience to classify the values.
119# Most ranges in this program will be Type 0, or normal, but there are some
120# ranges that have a non-zero type. These are used only in map tables, and
121# are for mappings that don't fit into the normal scheme of things. Mappings
122# that require a hash entry to communicate with utf8.c are one example;
123# another example is mappings for charnames.pm to use which indicate a name
232ed87f 124# that is algorithmically determinable from its code point (and the reverse).
99870f4d
KW
125# These are used to significantly compact these tables, instead of listing
126# each one of the tens of thousands individually.
127#
128# In a match table, the value of a range is irrelevant (and hence the type as
129# well, which will always be 0), and arbitrarily set to the null string.
130# Using the example above, there would be two match tables for those two
131# entries, one named Upper would contain the 0x41..0x5A range, and the other
132# named Lower would contain 0x61..0x7A.
133#
134# Actually, there are two types of range lists, "Range_Map" is the one
135# associated with map tables, and "Range_List" with match tables.
232ed87f
KW
136# Again, this is so that methods can be defined on one and not the others so
137# as to prevent operating on them in incorrect ways.
99870f4d
KW
138#
139# Eventually, most tables are written out to files to be read by utf8_heavy.pl
140# in the perl core. All tables could in theory be written, but some are
141# suppressed because there is no current practical use for them. It is easy
142# to change which get written by changing various lists that are near the top
143# of the actual code in this file. The table data structures contain enough
144# ancillary information to allow them to be treated as separate entities for
145# writing, such as the path to each one's file. There is a heading in each
146# map table that gives the format of its entries, and what the map is for all
147# the code points missing from it. (This allows tables to be more compact.)
678f13d5 148#
99870f4d
KW
149# The Property data structure contains one or more tables. All properties
150# contain a map table (except the $perl property which is a
151# pseudo-property containing only match tables), and any properties that
152# are usable in regular expression matches also contain various matching
153# tables, one for each value the property can have. A binary property can
154# have two values, True and False (or Y and N, which are preferred by Unicode
155# terminology). Thus each of these properties will have a map table that
156# takes every code point and maps it to Y or N (but having ranges cuts the
157# number of entries in that table way down), and two match tables, one
158# which has a list of all the code points that map to Y, and one for all the
232ed87f 159# code points that map to N. (For each binary property, a third table is also
99870f4d 160# generated for the pseudo Perl property. It contains the identical code
232ed87f
KW
161# points as the Y table, but can be written in regular expressions, not in the
162# compound form, but in a "single" form like \p{IsUppercase}.) Many
163# properties are binary, but some properties have several possible values,
164# some have many, and properties like Name have a different value for every
165# named code point. Those will not, unless the controlling lists are changed,
166# have their match tables written out. But all the ones which can be used in
167# regular expression \p{} and \P{} constructs will. Prior to 5.14, generally
168# a property would have either its map table or its match tables written but
169# not both. Again, what gets written is controlled by lists which can easily
170# be changed. Starting in 5.14, advantage was taken of this, and all the map
171# tables needed to reconstruct the Unicode db are now written out, while
172# suppressing the Unicode .txt files that contain the data. Our tables are
173# much more compact than the .txt files, so a significant space savings was
174# achieved. Also, tables are not written out that are trivially derivable
175# from tables that do get written. So, there typically is no file containing
176# the code points not matched by a binary property (the table for \P{} versus
177# lowercase \p{}), since you just need to invert the True table to get the
178# False table.
179
180# Properties have a 'Type', like 'binary', or 'string', or 'enum' depending on
181# how many match tables there are and the content of the maps. This 'Type' is
c12f2655
KW
182# different than a range 'Type', so don't get confused by the two concepts
183# having the same name.
678f13d5 184#
99870f4d
KW
185# For information about the Unicode properties, see Unicode's UAX44 document:
186
187my $unicode_reference_url = 'http://www.unicode.org/reports/tr44/';
188
189# As stated earlier, this program will work on any release of Unicode so far.
190# Most obvious problems in earlier data have NOT been corrected except when
be864b6c 191# necessary to make Perl or this program work reasonably, and to keep out
232ed87f
KW
192# potential security issues. For example, no folding information was given in
193# early releases, so this program substitutes lower case instead, just so that
194# a regular expression with the /i option will do something that actually
195# gives the right results in many cases. There are also a couple other
196# corrections for version 1.1.5, commented at the point they are made. As an
197# example of corrections that weren't made (but could be) is this statement
198# from DerivedAge.txt: "The supplementary private use code points and the
199# non-character code points were assigned in version 2.0, but not specifically
200# listed in the UCD until versions 3.0 and 3.1 respectively." (To be precise
201# it was 3.0.1 not 3.0.0) More information on Unicode version glitches is
202# further down in these introductory comments.
99870f4d 203#
232ed87f
KW
204# This program works on all non-provisional properties as of the current
205# Unicode release, though the files for some are suppressed for various
206# reasons. You can change which are output by changing lists in this program.
678f13d5 207#
dc85bd38 208# The old version of mktables emphasized the term "Fuzzy" to mean Unicode's
99870f4d
KW
209# loose matchings rules (from Unicode TR18):
210#
211# The recommended names for UCD properties and property values are in
212# PropertyAliases.txt [Prop] and PropertyValueAliases.txt
213# [PropValue]. There are both abbreviated names and longer, more
214# descriptive names. It is strongly recommended that both names be
215# recognized, and that loose matching of property names be used,
216# whereby the case distinctions, whitespace, hyphens, and underbar
217# are ignored.
232ed87f 218#
99870f4d
KW
219# The program still allows Fuzzy to override its determination of if loose
220# matching should be used, but it isn't currently used, as it is no longer
221# needed; the calculations it makes are good enough.
678f13d5 222#
99870f4d
KW
223# SUMMARY OF HOW IT WORKS:
224#
225# Process arguments
226#
227# A list is constructed containing each input file that is to be processed
228#
229# Each file on the list is processed in a loop, using the associated handler
230# code for each:
231# The PropertyAliases.txt and PropValueAliases.txt files are processed
232# first. These files name the properties and property values.
233# Objects are created of all the property and property value names
234# that the rest of the input should expect, including all synonyms.
235# The other input files give mappings from properties to property
236# values. That is, they list code points and say what the mapping
237# is under the given property. Some files give the mappings for
238# just one property; and some for many. This program goes through
232ed87f
KW
239# each file and populates the properties and their map tables from
240# them. Some properties are listed in more than one file, and
241# Unicode has set up a precedence as to which has priority if there
242# is a conflict. Thus the order of processing matters, and this
243# program handles the conflict possibility by processing the
244# overriding input files last, so that if necessary they replace
245# earlier values.
99870f4d
KW
246# After this is all done, the program creates the property mappings not
247# furnished by Unicode, but derivable from what it does give.
248# The tables of code points that match each property value in each
249# property that is accessible by regular expressions are created.
250# The Perl-defined properties are created and populated. Many of these
251# require data determined from the earlier steps
252# Any Perl-defined synonyms are created, and name clashes between Perl
678f13d5 253# and Unicode are reconciled and warned about.
99870f4d
KW
254# All the properties are written to files
255# Any other files are written, and final warnings issued.
678f13d5 256#
99870f4d
KW
257# For clarity, a number of operators have been overloaded to work on tables:
258# ~ means invert (take all characters not in the set). The more
259# conventional '!' is not used because of the possibility of confusing
260# it with the actual boolean operation.
261# + means union
262# - means subtraction
263# & means intersection
264# The precedence of these is the order listed. Parentheses should be
265# copiously used. These are not a general scheme. The operations aren't
266# defined for a number of things, deliberately, to avoid getting into trouble.
267# Operations are done on references and affect the underlying structures, so
268# that the copy constructors for them have been overloaded to not return a new
269# clone, but the input object itself.
678f13d5 270#
99870f4d
KW
271# The bool operator is deliberately not overloaded to avoid confusion with
272# "should it mean if the object merely exists, or also is non-empty?".
99870f4d
KW
273#
274# WHY CERTAIN DESIGN DECISIONS WERE MADE
678f13d5
KW
275#
276# This program needs to be able to run under miniperl. Therefore, it uses a
277# minimum of other modules, and hence implements some things itself that could
278# be gotten from CPAN
279#
280# This program uses inputs published by the Unicode Consortium. These can
281# change incompatibly between releases without the Perl maintainers realizing
282# it. Therefore this program is now designed to try to flag these. It looks
283# at the directories where the inputs are, and flags any unrecognized files.
284# It keeps track of all the properties in the files it handles, and flags any
285# that it doesn't know how to handle. It also flags any input lines that
286# don't match the expected syntax, among other checks.
287#
288# It is also designed so if a new input file matches one of the known
289# templates, one hopefully just needs to add it to a list to have it
290# processed.
291#
292# As mentioned earlier, some properties are given in more than one file. In
293# particular, the files in the extracted directory are supposedly just
294# reformattings of the others. But they contain information not easily
295# derivable from the other files, including results for Unihan, which this
296# program doesn't ordinarily look at, and for unassigned code points. They
297# also have historically had errors or been incomplete. In an attempt to
298# create the best possible data, this program thus processes them first to
299# glean information missing from the other files; then processes those other
300# files to override any errors in the extracted ones. Much of the design was
301# driven by this need to store things and then possibly override them.
302#
303# It tries to keep fatal errors to a minimum, to generate something usable for
304# testing purposes. It always looks for files that could be inputs, and will
305# warn about any that it doesn't know how to handle (the -q option suppresses
306# the warning).
99870f4d 307#
678f13d5
KW
308# Why is there more than one type of range?
309# This simplified things. There are some very specialized code points that
310# have to be handled specially for output, such as Hangul syllable names.
311# By creating a range type (done late in the development process), it
312# allowed this to be stored with the range, and overridden by other input.
313# Originally these were stored in another data structure, and it became a
314# mess trying to decide if a second file that was for the same property was
315# overriding the earlier one or not.
316#
317# Why are there two kinds of tables, match and map?
318# (And there is a base class shared by the two as well.) As stated above,
319# they actually are for different things. Development proceeded much more
320# smoothly when I (khw) realized the distinction. Map tables are used to
321# give the property value for every code point (actually every code point
322# that doesn't map to a default value). Match tables are used for regular
323# expression matches, and are essentially the inverse mapping. Separating
324# the two allows more specialized methods, and error checks so that one
325# can't just take the intersection of two map tables, for example, as that
326# is nonsensical.
99870f4d 327#
232ed87f
KW
328# What about 'fate' and 'status'. The concept of a table's fate was created
329# late when it became clear that something more was needed. The difference
330# between this and 'status' is unclean, and could be improved if someone
331# wanted to spend the effort.
332#
23e33b60
KW
333# DEBUGGING
334#
678f13d5
KW
335# This program is written so it will run under miniperl. Occasionally changes
336# will cause an error where the backtrace doesn't work well under miniperl.
337# To diagnose the problem, you can instead run it under regular perl, if you
338# have one compiled.
339#
340# There is a good trace facility. To enable it, first sub DEBUG must be set
341# to return true. Then a line like
342#
343# local $to_trace = 1 if main::DEBUG;
344#
232ed87f
KW
345# can be added to enable tracing in its lexical scope (plus dynamic) or until
346# you insert another line:
678f13d5
KW
347#
348# local $to_trace = 0 if main::DEBUG;
349#
232ed87f 350# To actually trace, use a line like "trace $a, @b, %c, ...;
678f13d5
KW
351#
352# Some of the more complex subroutines already have trace statements in them.
353# Permanent trace statements should be like:
354#
355# trace ... if main::DEBUG && $to_trace;
356#
357# If there is just one or a few files that you're debugging, you can easily
358# cause most everything else to be skipped. Change the line
359#
360# my $debug_skip = 0;
361#
362# to 1, and every file whose object is in @input_file_objects and doesn't have
232ed87f
KW
363# a, 'non_skip => 1,' in its constructor will be skipped. However, skipping
364# Jamo.txt or UnicodeData.txt will likely cause fatal errors.
678f13d5 365#
b4a0206c 366# To compare the output tables, it may be useful to specify the -annotate
97a8a595
KW
367# flag. (As of this writing, this can't be done on a clean workspace, due to
368# requirements in Text::Tabs used in this option; so first run mktables
369# without this option.) This option adds comment lines to each table, one for
370# each non-algorithmically named character giving, currently its code point,
371# name, and graphic representation if printable (and you have a font that
372# knows about it). This makes it easier to see what the particular code
373# points are in each output table. Non-named code points are annotated with a
374# description of their status, and contiguous ones with the same description
375# will be output as a range rather than individually. Algorithmically named
376# characters are also output as ranges, except when there are just a few
377# contiguous ones.
c4019d52 378#
99870f4d
KW
379# FUTURE ISSUES
380#
381# The program would break if Unicode were to change its names so that
382# interior white space, underscores, or dashes differences were significant
383# within property and property value names.
384#
385# It might be easier to use the xml versions of the UCD if this program ever
386# would need heavy revision, and the ability to handle old versions was not
387# required.
388#
389# There is the potential for name collisions, in that Perl has chosen names
390# that Unicode could decide it also likes. There have been such collisions in
391# the past, with mostly Perl deciding to adopt the Unicode definition of the
392# name. However in the 5.2 Unicode beta testing, there were a number of such
393# collisions, which were withdrawn before the final release, because of Perl's
394# and other's protests. These all involved new properties which began with
395# 'Is'. Based on the protests, Unicode is unlikely to try that again. Also,
396# many of the Perl-defined synonyms, like Any, Word, etc, are listed in a
397# Unicode document, so they are unlikely to be used by Unicode for another
398# purpose. However, they might try something beginning with 'In', or use any
399# of the other Perl-defined properties. This program will warn you of name
400# collisions, and refuse to generate tables with them, but manual intervention
401# will be required in this event. One scheme that could be implemented, if
402# necessary, would be to have this program generate another file, or add a
403# field to mktables.lst that gives the date of first definition of a property.
404# Each new release of Unicode would use that file as a basis for the next
405# iteration. And the Perl synonym addition code could sort based on the age
406# of the property, so older properties get priority, and newer ones that clash
407# would be refused; hence existing code would not be impacted, and some other
408# synonym would have to be used for the new property. This is ugly, and
409# manual intervention would certainly be easier to do in the short run; lets
410# hope it never comes to this.
678f13d5 411#
99870f4d
KW
412# A NOTE ON UNIHAN
413#
414# This program can generate tables from the Unihan database. But it doesn't
415# by default, letting the CPAN module Unicode::Unihan handle them. Prior to
416# version 5.2, this database was in a single file, Unihan.txt. In 5.2 the
417# database was split into 8 different files, all beginning with the letters
418# 'Unihan'. This program will read those file(s) if present, but it needs to
419# know which of the many properties in the file(s) should have tables created
420# for them. It will create tables for any properties listed in
421# PropertyAliases.txt and PropValueAliases.txt, plus any listed in the
422# @cjk_properties array and the @cjk_property_values array. Thus, if a
423# property you want is not in those files of the release you are building
424# against, you must add it to those two arrays. Starting in 4.0, the
425# Unicode_Radical_Stroke was listed in those files, so if the Unihan database
426# is present in the directory, a table will be generated for that property.
427# In 5.2, several more properties were added. For your convenience, the two
5f7264c7 428# arrays are initialized with all the 6.0 listed properties that are also in
99870f4d
KW
429# earlier releases. But these are commented out. You can just uncomment the
430# ones you want, or use them as a template for adding entries for other
431# properties.
432#
433# You may need to adjust the entries to suit your purposes. setup_unihan(),
434# and filter_unihan_line() are the functions where this is done. This program
435# already does some adjusting to make the lines look more like the rest of the
436# Unicode DB; You can see what that is in filter_unihan_line()
437#
438# There is a bug in the 3.2 data file in which some values for the
439# kPrimaryNumeric property have commas and an unexpected comment. A filter
440# could be added for these; or for a particular installation, the Unihan.txt
441# file could be edited to fix them.
99870f4d 442#
678f13d5
KW
443# HOW TO ADD A FILE TO BE PROCESSED
444#
445# A new file from Unicode needs to have an object constructed for it in
446# @input_file_objects, probably at the end or at the end of the extracted
447# ones. The program should warn you if its name will clash with others on
448# restrictive file systems, like DOS. If so, figure out a better name, and
449# add lines to the README.perl file giving that. If the file is a character
232ed87f 450# property, it should be in the format that Unicode has implicitly
678f13d5
KW
451# standardized for such files for the more recently introduced ones.
452# If so, the Input_file constructor for @input_file_objects can just be the
453# file name and release it first appeared in. If not, then it should be
454# possible to construct an each_line_handler() to massage the line into the
455# standardized form.
456#
457# For non-character properties, more code will be needed. You can look at
458# the existing entries for clues.
459#
460# UNICODE VERSIONS NOTES
461#
462# The Unicode UCD has had a number of errors in it over the versions. And
463# these remain, by policy, in the standard for that version. Therefore it is
464# risky to correct them, because code may be expecting the error. So this
465# program doesn't generally make changes, unless the error breaks the Perl
466# core. As an example, some versions of 2.1.x Jamo.txt have the wrong value
467# for U+1105, which causes real problems for the algorithms for Jamo
468# calculations, so it is changed here.
469#
470# But it isn't so clear cut as to what to do about concepts that are
471# introduced in a later release; should they extend back to earlier releases
472# where the concept just didn't exist? It was easier to do this than to not,
473# so that's what was done. For example, the default value for code points not
474# in the files for various properties was probably undefined until changed by
475# some version. No_Block for blocks is such an example. This program will
476# assign No_Block even in Unicode versions that didn't have it. This has the
477# benefit that code being written doesn't have to special case earlier
478# versions; and the detriment that it doesn't match the Standard precisely for
479# the affected versions.
480#
481# Here are some observations about some of the issues in early versions:
482#
232ed87f
KW
483# Prior to version 3.0, there were 3 character decompositions. These are not
484# handled by Unicode::Normalize, nor will it compile when presented a version
485# that has them. However, you can trivially get it to compile by simply
486# ignoring those decompositions, by changing the croak to a carp. At the time
487# of this writing, the line (in cpan/Unicode-Normalize/mkheader) reads
488#
489# croak("Weird Canonical Decomposition of U+$h");
490#
491# Simply change to a carp. It will compile, but will not know about any three
492# character decomposition.
493
494# The number of code points in \p{alpha=True} halved in 2.1.9. It turns out
495# that the reason is that the CJK block starting at 4E00 was removed from
496# PropList, and was not put back in until 3.1.0. The Perl extension (the
497# single property name \p{alpha}) has the correct values. But the compound
498# form is simply not generated until 3.1, as it can be argued that prior to
499# this release, this was not an official property. The comments for
500# filter_old_style_proplist() give more details.
678f13d5
KW
501#
502# Unicode introduced the synonym Space for White_Space in 4.1. Perl has
503# always had a \p{Space}. In release 3.2 only, they are not synonymous. The
504# reason is that 3.2 introduced U+205F=medium math space, which was not
505# classed as white space, but Perl figured out that it should have been. 4.0
506# reclassified it correctly.
507#
508# Another change between 3.2 and 4.0 is the CCC property value ATBL. In 3.2
232ed87f
KW
509# this was erroneously a synonym for 202 (it should be 200). In 4.0, ATB
510# became 202, and ATBL was left with no code points, as all the ones that
511# mapped to 202 stayed mapped to 202. Thus if your program used the numeric
512# name for the class, it would not have been affected, but if it used the
513# mnemonic, it would have been.
678f13d5
KW
514#
515# \p{Script=Hrkt} (Katakana_Or_Hiragana) came in 4.0.1. Before that code
516# points which eventually came to have this script property value, instead
517# mapped to "Unknown". But in the next release all these code points were
518# moved to \p{sc=common} instead.
99870f4d
KW
519#
520# The default for missing code points for BidiClass is complicated. Starting
521# in 3.1.1, the derived file DBidiClass.txt handles this, but this program
522# tries to do the best it can for earlier releases. It is done in
523# process_PropertyAliases()
524#
232ed87f
KW
525# In version 2.1.2, the entry in UnicodeData.txt:
526# 0275;LATIN SMALL LETTER BARRED O;Ll;0;L;;;;;N;;;;019F;
527# should instead be
528# 0275;LATIN SMALL LETTER BARRED O;Ll;0;L;;;;;N;;;019F;;019F
529# Without this change, there are casing problems for this character.
530#
99870f4d
KW
531##############################################################################
532
533my $UNDEF = ':UNDEF:'; # String to print out for undefined values in tracing
534 # and errors
535my $MAX_LINE_WIDTH = 78;
536
537# Debugging aid to skip most files so as to not be distracted by them when
538# concentrating on the ones being debugged. Add
539# non_skip => 1,
540# to the constructor for those files you want processed when you set this.
541# Files with a first version number of 0 are special: they are always
c12f2655
KW
542# processed regardless of the state of this flag. Generally, Jamo.txt and
543# UnicodeData.txt must not be skipped if you want this program to not die
544# before normal completion.
99870f4d
KW
545my $debug_skip = 0;
546
e9c4b4f8
KW
547
548# Normally these are suppressed.
549my $write_Unicode_deprecated_tables = 0;
550
99870f4d
KW
551# Set to 1 to enable tracing.
552our $to_trace = 0;
553
554{ # Closure for trace: debugging aid
555 my $print_caller = 1; # ? Include calling subroutine name
556 my $main_with_colon = 'main::';
557 my $main_colon_length = length($main_with_colon);
558
559 sub trace {
560 return unless $to_trace; # Do nothing if global flag not set
561
562 my @input = @_;
563
564 local $DB::trace = 0;
565 $DB::trace = 0; # Quiet 'used only once' message
566
567 my $line_number;
568
569 # Loop looking up the stack to get the first non-trace caller
570 my $caller_line;
571 my $caller_name;
572 my $i = 0;
573 do {
574 $line_number = $caller_line;
575 (my $pkg, my $file, $caller_line, my $caller) = caller $i++;
576 $caller = $main_with_colon unless defined $caller;
577
578 $caller_name = $caller;
579
580 # get rid of pkg
581 $caller_name =~ s/.*:://;
582 if (substr($caller_name, 0, $main_colon_length)
583 eq $main_with_colon)
584 {
585 $caller_name = substr($caller_name, $main_colon_length);
586 }
587
588 } until ($caller_name ne 'trace');
589
590 # If the stack was empty, we were called from the top level
591 $caller_name = 'main' if ($caller_name eq ""
592 || $caller_name eq 'trace');
593
594 my $output = "";
595 foreach my $string (@input) {
596 #print STDERR __LINE__, ": ", join ", ", @input, "\n";
597 if (ref $string eq 'ARRAY' || ref $string eq 'HASH') {
598 $output .= simple_dumper($string);
599 }
600 else {
601 $string = "$string" if ref $string;
602 $string = $UNDEF unless defined $string;
603 chomp $string;
604 $string = '""' if $string eq "";
605 $output .= " " if $output ne ""
606 && $string ne ""
607 && substr($output, -1, 1) ne " "
608 && substr($string, 0, 1) ne " ";
609 $output .= $string;
610 }
611 }
612
99f78760
KW
613 print STDERR sprintf "%4d: ", $line_number if defined $line_number;
614 print STDERR "$caller_name: " if $print_caller;
99870f4d
KW
615 print STDERR $output, "\n";
616 return;
617 }
618}
619
620# This is for a rarely used development feature that allows you to compare two
621# versions of the Unicode standard without having to deal with changes caused
c12f2655
KW
622# by the code points introduced in the later version. Change the 0 to a
623# string containing a SINGLE dotted Unicode release number (e.g. "2.1"). Only
624# code points introduced in that release and earlier will be used; later ones
625# are thrown away. You use the version number of the earliest one you want to
626# compare; then run this program on directory structures containing each
627# release, and compare the outputs. These outputs will therefore include only
628# the code points common to both releases, and you can see the changes caused
629# just by the underlying release semantic changes. For versions earlier than
630# 3.2, you must copy a version of DAge.txt into the directory.
631my $string_compare_versions = DEBUG && 0; # e.g., "2.1";
99870f4d
KW
632my $compare_versions = DEBUG
633 && $string_compare_versions
634 && pack "C*", split /\./, $string_compare_versions;
635
636sub uniques {
637 # Returns non-duplicated input values. From "Perl Best Practices:
638 # Encapsulated Cleverness". p. 455 in first edition.
639
640 my %seen;
0e407844
NC
641 # Arguably this breaks encapsulation, if the goal is to permit multiple
642 # distinct objects to stringify to the same value, and be interchangeable.
643 # However, for this program, no two objects stringify identically, and all
644 # lists passed to this function are either objects or strings. So this
645 # doesn't affect correctness, but it does give a couple of percent speedup.
646 no overloading;
99870f4d
KW
647 return grep { ! $seen{$_}++ } @_;
648}
649
650$0 = File::Spec->canonpath($0);
651
652my $make_test_script = 0; # ? Should we output a test script
6b5ab373 653my $make_norm_test_script = 0; # ? Should we output a normalization test script
99870f4d
KW
654my $write_unchanged_files = 0; # ? Should we update the output files even if
655 # we don't think they have changed
656my $use_directory = ""; # ? Should we chdir somewhere.
657my $pod_directory; # input directory to store the pod file.
658my $pod_file = 'perluniprops';
659my $t_path; # Path to the .t test file
660my $file_list = 'mktables.lst'; # File to store input and output file names.
661 # This is used to speed up the build, by not
662 # executing the main body of the program if
663 # nothing on the list has changed since the
664 # previous build
665my $make_list = 1; # ? Should we write $file_list. Set to always
666 # make a list so that when the pumpking is
667 # preparing a release, s/he won't have to do
668 # special things
669my $glob_list = 0; # ? Should we try to include unknown .txt files
670 # in the input.
bd9ebcfd
KW
671my $output_range_counts = $debugging_build; # ? Should we include the number
672 # of code points in ranges in
673 # the output
558712cf 674my $annotate = 0; # ? Should character names be in the output
9ef2b94f 675
99870f4d
KW
676# Verbosity levels; 0 is quiet
677my $NORMAL_VERBOSITY = 1;
678my $PROGRESS = 2;
679my $VERBOSE = 3;
680
681my $verbosity = $NORMAL_VERBOSITY;
682
0458fbc1
KW
683# Stored in mktables.lst so that if this program is called with different
684# options, will regenerate even if the files otherwise look like they're
685# up-to-date.
686my $command_line_arguments = join " ", @ARGV;
687
99870f4d
KW
688# Process arguments
689while (@ARGV) {
cf25bb62
JH
690 my $arg = shift @ARGV;
691 if ($arg eq '-v') {
99870f4d
KW
692 $verbosity = $VERBOSE;
693 }
694 elsif ($arg eq '-p') {
695 $verbosity = $PROGRESS;
696 $| = 1; # Flush buffers as we go.
697 }
698 elsif ($arg eq '-q') {
699 $verbosity = 0;
700 }
701 elsif ($arg eq '-w') {
702 $write_unchanged_files = 1; # update the files even if havent changed
703 }
704 elsif ($arg eq '-check') {
6ae7e459
YO
705 my $this = shift @ARGV;
706 my $ok = shift @ARGV;
707 if ($this ne $ok) {
708 print "Skipping as check params are not the same.\n";
709 exit(0);
710 }
00a8df5c 711 }
99870f4d
KW
712 elsif ($arg eq '-P' && defined ($pod_directory = shift)) {
713 -d $pod_directory or croak "Directory '$pod_directory' doesn't exist";
714 }
3df51b85
KW
715 elsif ($arg eq '-maketest' || ($arg eq '-T' && defined ($t_path = shift)))
716 {
99870f4d 717 $make_test_script = 1;
99870f4d 718 }
6b5ab373
KW
719 elsif ($arg eq '-makenormtest')
720 {
721 $make_norm_test_script = 1;
722 }
99870f4d
KW
723 elsif ($arg eq '-makelist') {
724 $make_list = 1;
725 }
726 elsif ($arg eq '-C' && defined ($use_directory = shift)) {
727 -d $use_directory or croak "Unknown directory '$use_directory'";
728 }
729 elsif ($arg eq '-L') {
730
731 # Existence not tested until have chdir'd
732 $file_list = shift;
733 }
734 elsif ($arg eq '-globlist') {
735 $glob_list = 1;
736 }
737 elsif ($arg eq '-c') {
738 $output_range_counts = ! $output_range_counts
739 }
b4a0206c 740 elsif ($arg eq '-annotate') {
558712cf 741 $annotate = 1;
bd9ebcfd
KW
742 $debugging_build = 1;
743 $output_range_counts = 1;
9ef2b94f 744 }
99870f4d
KW
745 else {
746 my $with_c = 'with';
747 $with_c .= 'out' if $output_range_counts; # Complements the state
748 croak <<END;
749usage: $0 [-c|-p|-q|-v|-w] [-C dir] [-L filelist] [ -P pod_dir ]
750 [ -T test_file_path ] [-globlist] [-makelist] [-maketest]
751 [-check A B ]
752 -c : Output comments $with_c number of code points in ranges
753 -q : Quiet Mode: Only output serious warnings.
754 -p : Set verbosity level to normal plus show progress.
755 -v : Set Verbosity level high: Show progress and non-serious
756 warnings
757 -w : Write files regardless
758 -C dir : Change to this directory before proceeding. All relative paths
759 except those specified by the -P and -T options will be done
760 with respect to this directory.
761 -P dir : Output $pod_file file to directory 'dir'.
3df51b85 762 -T path : Create a test script as 'path'; overrides -maketest
99870f4d
KW
763 -L filelist : Use alternate 'filelist' instead of standard one
764 -globlist : Take as input all non-Test *.txt files in current and sub
765 directories
3df51b85
KW
766 -maketest : Make test script 'TestProp.pl' in current (or -C directory),
767 overrides -T
99870f4d 768 -makelist : Rewrite the file list $file_list based on current setup
b4a0206c 769 -annotate : Output an annotation for each character in the table files;
97a8a595
KW
770 useful for debugging mktables, looking at diffs; but is slow
771 and memory intensive
99870f4d
KW
772 -check A B : Executes $0 only if A and B are the same
773END
774 }
775}
776
777# Stores the most-recently changed file. If none have changed, can skip the
778# build
aeab6150 779my $most_recent = (stat $0)[9]; # Do this before the chdir!
99870f4d
KW
780
781# Change directories now, because need to read 'version' early.
782if ($use_directory) {
3df51b85 783 if ($pod_directory && ! File::Spec->file_name_is_absolute($pod_directory)) {
99870f4d
KW
784 $pod_directory = File::Spec->rel2abs($pod_directory);
785 }
3df51b85 786 if ($t_path && ! File::Spec->file_name_is_absolute($t_path)) {
99870f4d 787 $t_path = File::Spec->rel2abs($t_path);
00a8df5c 788 }
99870f4d 789 chdir $use_directory or croak "Failed to chdir to '$use_directory':$!";
3df51b85 790 if ($pod_directory && File::Spec->file_name_is_absolute($pod_directory)) {
99870f4d 791 $pod_directory = File::Spec->abs2rel($pod_directory);
02b1aeec 792 }
3df51b85 793 if ($t_path && File::Spec->file_name_is_absolute($t_path)) {
99870f4d 794 $t_path = File::Spec->abs2rel($t_path);
02b1aeec 795 }
00a8df5c
YO
796}
797
99870f4d
KW
798# Get Unicode version into regular and v-string. This is done now because
799# various tables below get populated based on it. These tables are populated
800# here to be near the top of the file, and so easily seeable by those needing
801# to modify things.
802open my $VERSION, "<", "version"
803 or croak "$0: can't open required file 'version': $!\n";
804my $string_version = <$VERSION>;
805close $VERSION;
806chomp $string_version;
807my $v_version = pack "C*", split /\./, $string_version; # v string
808
809# The following are the complete names of properties with property values that
810# are known to not match any code points in some versions of Unicode, but that
811# may change in the future so they should be matchable, hence an empty file is
812# generated for them.
a9c9e371
KW
813my @tables_that_may_be_empty;
814push @tables_that_may_be_empty, 'Joining_Type=Left_Joining'
815 if $v_version lt v6.3.0;
99870f4d
KW
816push @tables_that_may_be_empty, 'Script=Common' if $v_version le v4.0.1;
817push @tables_that_may_be_empty, 'Title' if $v_version lt v2.0.0;
818push @tables_that_may_be_empty, 'Script=Katakana_Or_Hiragana'
819 if $v_version ge v4.1.0;
82aed44a
KW
820push @tables_that_may_be_empty, 'Script_Extensions=Katakana_Or_Hiragana'
821 if $v_version ge v6.0.0;
f583b44c
KW
822push @tables_that_may_be_empty, 'Grapheme_Cluster_Break=Prepend'
823 if $v_version ge v6.1.0;
1e958ea9
KW
824push @tables_that_may_be_empty, 'Canonical_Combining_Class=CCC133'
825 if $v_version ge v6.2.0;
99870f4d
KW
826
827# The lists below are hashes, so the key is the item in the list, and the
828# value is the reason why it is in the list. This makes generation of
829# documentation easier.
830
831my %why_suppressed; # No file generated for these.
832
833# Files aren't generated for empty extraneous properties. This is arguable.
834# Extraneous properties generally come about because a property is no longer
835# used in a newer version of Unicode. If we generated a file without code
836# points, programs that used to work on that property will still execute
837# without errors. It just won't ever match (or will always match, with \P{}).
838# This means that the logic is now likely wrong. I (khw) think its better to
839# find this out by getting an error message. Just move them to the table
840# above to change this behavior
841my %why_suppress_if_empty_warn_if_not = (
842
843 # It is the only property that has ever officially been removed from the
844 # Standard. The database never contained any code points for it.
845 'Special_Case_Condition' => 'Obsolete',
846
847 # Apparently never official, but there were code points in some versions of
848 # old-style PropList.txt
849 'Non_Break' => 'Obsolete',
850);
851
852# These would normally go in the warn table just above, but they were changed
853# a long time before this program was written, so warnings about them are
854# moot.
855if ($v_version gt v3.2.0) {
856 push @tables_that_may_be_empty,
857 'Canonical_Combining_Class=Attached_Below_Left'
858}
859
5f7264c7 860# These are listed in the Property aliases file in 6.0, but Unihan is ignored
99870f4d
KW
861# unless explicitly added.
862if ($v_version ge v5.2.0) {
863 my $unihan = 'Unihan; remove from list if using Unihan';
ea25a9b2 864 foreach my $table (qw (
99870f4d
KW
865 kAccountingNumeric
866 kOtherNumeric
867 kPrimaryNumeric
868 kCompatibilityVariant
869 kIICore
870 kIRG_GSource
871 kIRG_HSource
872 kIRG_JSource
873 kIRG_KPSource
874 kIRG_MSource
875 kIRG_KSource
876 kIRG_TSource
877 kIRG_USource
878 kIRG_VSource
879 kRSUnicode
ea25a9b2 880 ))
99870f4d
KW
881 {
882 $why_suppress_if_empty_warn_if_not{$table} = $unihan;
883 }
ca12659b
NC
884}
885
272501f6
KW
886# Enum values for to_output_map() method in the Map_Table package.
887my $EXTERNAL_MAP = 1;
888my $INTERNAL_MAP = 2;
ce712c88 889my $OUTPUT_ADJUSTED = 3;
272501f6 890
fcf1973c
KW
891# To override computed values for writing the map tables for these properties.
892# The default for enum map tables is to write them out, so that the Unicode
893# .txt files can be removed, but all the data to compute any property value
894# for any code point is available in a more compact form.
895my %global_to_output_map = (
896 # Needed by UCD.pm, but don't want to publicize that it exists, so won't
c12f2655
KW
897 # get stuck supporting it if things change. Since it is a STRING
898 # property, it normally would be listed in the pod, but INTERNAL_MAP
899 # suppresses that.
fcf1973c
KW
900 Unicode_1_Name => $INTERNAL_MAP,
901
902 Present_In => 0, # Suppress, as easily computed from Age
74cd47d0
KW
903 Block => (NON_ASCII_PLATFORM) ? 1 : 0, # Suppress, as Blocks.txt is
904 # retained, but needed for
905 # non-ASCII
53d34b6c
KW
906
907 # Suppress, as mapping can be found instead from the
908 # Perl_Decomposition_Mapping file
909 Decomposition_Type => 0,
fcf1973c
KW
910);
911
99870f4d 912# Properties that this program ignores.
230e0c16
KW
913my @unimplemented_properties;
914
915# With this release, it is automatically handled if the Unihan db is
916# downloaded
917push @unimplemented_properties, 'Unicode_Radical_Stroke' if $v_version le v5.2.0;
d73e5302 918
99870f4d
KW
919# There are several types of obsolete properties defined by Unicode. These
920# must be hand-edited for every new Unicode release.
921my %why_deprecated; # Generates a deprecated warning message if used.
922my %why_stabilized; # Documentation only
923my %why_obsolete; # Documentation only
924
925{ # Closure
8364f9eb 926 my $simple = 'Perl uses the more complete version';
99870f4d
KW
927 my $unihan = 'Unihan properties are by default not enabled in the Perl core. Instead use CPAN: Unicode::Unihan';
928
929 my $other_properties = 'other properties';
930 my $contributory = "Used by Unicode internally for generating $other_properties and not intended to be used stand-alone";
5d294d41 931 my $why_no_expand = "Deprecated by Unicode. These are characters that expand to more than one character in the specified normalization form, but whether they actually take up more bytes or not depends on the encoding being used. For example, a UTF-8 encoded character may expand to a different number of bytes than a UTF-32 encoded character.";
99870f4d
KW
932
933 %why_deprecated = (
5f7264c7 934 'Grapheme_Link' => 'Deprecated by Unicode: Duplicates ccc=vr (Canonical_Combining_Class=Virama)',
99870f4d
KW
935 'Jamo_Short_Name' => $contributory,
936 'Line_Break=Surrogate' => 'Deprecated by Unicode because surrogates should never appear in well-formed text, and therefore shouldn\'t be the basis for line breaking',
937 'Other_Alphabetic' => $contributory,
938 'Other_Default_Ignorable_Code_Point' => $contributory,
939 'Other_Grapheme_Extend' => $contributory,
940 'Other_ID_Continue' => $contributory,
941 'Other_ID_Start' => $contributory,
942 'Other_Lowercase' => $contributory,
943 'Other_Math' => $contributory,
944 'Other_Uppercase' => $contributory,
e22aaf5c
KW
945 'Expands_On_NFC' => $why_no_expand,
946 'Expands_On_NFD' => $why_no_expand,
947 'Expands_On_NFKC' => $why_no_expand,
948 'Expands_On_NFKD' => $why_no_expand,
99870f4d
KW
949 );
950
951 %why_suppressed = (
5f7264c7 952 # There is a lib/unicore/Decomposition.pl (used by Normalize.pm) which
99870f4d
KW
953 # contains the same information, but without the algorithmically
954 # determinable Hangul syllables'. This file is not published, so it's
955 # existence is not noted in the comment.
e0b29447 956 'Decomposition_Mapping' => 'Accessible via Unicode::Normalize or Unicode::UCD::prop_invmap()',
99870f4d 957
3111abc0
KW
958 'Indic_Matra_Category' => "Provisional",
959 'Indic_Syllabic_Category' => "Provisional",
960
5f8d1a89
KW
961 # Don't suppress ISO_Comment, as otherwise special handling is needed
962 # to differentiate between it and gc=c, which can be written as 'isc',
963 # which is the same characters as ISO_Comment's short name.
99870f4d 964
fbb93542 965 'Name' => "Accessible via \\N{...} or 'use charnames;' or Unicode::UCD::prop_invmap()",
e0b29447
KW
966
967 'Simple_Case_Folding' => "$simple. Can access this through Unicode::UCD::casefold or Unicode::UCD::prop_invmap()",
968 'Simple_Lowercase_Mapping' => "$simple. Can access this through Unicode::UCD::charinfo or Unicode::UCD::prop_invmap()",
969 'Simple_Titlecase_Mapping' => "$simple. Can access this through Unicode::UCD::charinfo or Unicode::UCD::prop_invmap()",
970 'Simple_Uppercase_Mapping' => "$simple. Can access this through Unicode::UCD::charinfo or Unicode::UCD::prop_invmap()",
99870f4d 971
5f7264c7 972 FC_NFKC_Closure => 'Supplanted in usage by NFKC_Casefold; otherwise not useful',
99870f4d
KW
973 );
974
1704a0ea
KW
975 foreach my $property (
976
977 # The following are suppressed because they were made contributory
978 # or deprecated by Unicode before Perl ever thought about
979 # supporting them.
980 'Jamo_Short_Name',
981 'Grapheme_Link',
982 'Expands_On_NFC',
983 'Expands_On_NFD',
984 'Expands_On_NFKC',
985 'Expands_On_NFKD',
986
987 # The following are suppressed because they have been marked
988 # as deprecated for a sufficient amount of time
989 'Other_Alphabetic',
990 'Other_Default_Ignorable_Code_Point',
991 'Other_Grapheme_Extend',
992 'Other_ID_Continue',
993 'Other_ID_Start',
994 'Other_Lowercase',
995 'Other_Math',
996 'Other_Uppercase',
e22aaf5c 997 ) {
99870f4d
KW
998 $why_suppressed{$property} = $why_deprecated{$property};
999 }
cf25bb62 1000
99870f4d
KW
1001 # Customize the message for all the 'Other_' properties
1002 foreach my $property (keys %why_deprecated) {
1003 next if (my $main_property = $property) !~ s/^Other_//;
1004 $why_deprecated{$property} =~ s/$other_properties/the $main_property property (which should be used instead)/;
1005 }
1006}
1007
e9c4b4f8
KW
1008if ($write_Unicode_deprecated_tables) {
1009 foreach my $property (keys %why_suppressed) {
1010 delete $why_suppressed{$property} if $property =~
1011 / ^ Other | Grapheme /x;
1012 }
1013}
1014
99870f4d
KW
1015if ($v_version ge 4.0.0) {
1016 $why_stabilized{'Hyphen'} = 'Use the Line_Break property instead; see www.unicode.org/reports/tr14';
5f7264c7
KW
1017 if ($v_version ge 6.0.0) {
1018 $why_deprecated{'Hyphen'} = 'Supplanted by Line_Break property values; see www.unicode.org/reports/tr14';
1019 }
99870f4d 1020}
5f7264c7 1021if ($v_version ge 5.2.0 && $v_version lt 6.0.0) {
99870f4d 1022 $why_obsolete{'ISO_Comment'} = 'Code points for it have been removed';
5f7264c7 1023 if ($v_version ge 6.0.0) {
63f74647 1024 $why_deprecated{'ISO_Comment'} = 'No longer needed for Unicode\'s internal chart generation; otherwise not useful, and code points for it have been removed';
5f7264c7 1025 }
99870f4d
KW
1026}
1027
1028# Probably obsolete forever
1029if ($v_version ge v4.1.0) {
82aed44a
KW
1030 $why_suppressed{'Script=Katakana_Or_Hiragana'} = 'Obsolete. All code points previously matched by this have been moved to "Script=Common".';
1031}
1032if ($v_version ge v6.0.0) {
caa75395 1033 $why_suppressed{'Script=Katakana_Or_Hiragana'} .= ' Consider instead using "Script_Extensions=Katakana" or "Script_Extensions=Hiragana" (or both)';
2b352efd 1034 $why_suppressed{'Script_Extensions=Katakana_Or_Hiragana'} = 'All code points that would be matched by this are matched by either "Script_Extensions=Katakana" or "Script_Extensions=Hiragana"';
99870f4d
KW
1035}
1036
1037# This program can create files for enumerated-like properties, such as
1038# 'Numeric_Type'. This file would be the same format as for a string
1039# property, with a mapping from code point to its value, so you could look up,
1040# for example, the script a code point is in. But no one so far wants this
1041# mapping, or they have found another way to get it since this is a new
1042# feature. So no file is generated except if it is in this list.
1043my @output_mapped_properties = split "\n", <<END;
1044END
1045
c12f2655
KW
1046# If you are using the Unihan database in a Unicode version before 5.2, you
1047# need to add the properties that you want to extract from it to this table.
1048# For your convenience, the properties in the 6.0 PropertyAliases.txt file are
1049# listed, commented out
99870f4d
KW
1050my @cjk_properties = split "\n", <<'END';
1051#cjkAccountingNumeric; kAccountingNumeric
1052#cjkOtherNumeric; kOtherNumeric
1053#cjkPrimaryNumeric; kPrimaryNumeric
1054#cjkCompatibilityVariant; kCompatibilityVariant
1055#cjkIICore ; kIICore
1056#cjkIRG_GSource; kIRG_GSource
1057#cjkIRG_HSource; kIRG_HSource
1058#cjkIRG_JSource; kIRG_JSource
1059#cjkIRG_KPSource; kIRG_KPSource
1060#cjkIRG_KSource; kIRG_KSource
1061#cjkIRG_TSource; kIRG_TSource
1062#cjkIRG_USource; kIRG_USource
1063#cjkIRG_VSource; kIRG_VSource
1064#cjkRSUnicode; kRSUnicode ; Unicode_Radical_Stroke; URS
1065END
1066
1067# Similarly for the property values. For your convenience, the lines in the
5f7264c7 1068# 6.0 PropertyAliases.txt file are listed. Just remove the first BUT NOT both
c12f2655 1069# '#' marks (for Unicode versions before 5.2)
99870f4d
KW
1070my @cjk_property_values = split "\n", <<'END';
1071## @missing: 0000..10FFFF; cjkAccountingNumeric; NaN
1072## @missing: 0000..10FFFF; cjkCompatibilityVariant; <code point>
1073## @missing: 0000..10FFFF; cjkIICore; <none>
1074## @missing: 0000..10FFFF; cjkIRG_GSource; <none>
1075## @missing: 0000..10FFFF; cjkIRG_HSource; <none>
1076## @missing: 0000..10FFFF; cjkIRG_JSource; <none>
1077## @missing: 0000..10FFFF; cjkIRG_KPSource; <none>
1078## @missing: 0000..10FFFF; cjkIRG_KSource; <none>
1079## @missing: 0000..10FFFF; cjkIRG_TSource; <none>
1080## @missing: 0000..10FFFF; cjkIRG_USource; <none>
1081## @missing: 0000..10FFFF; cjkIRG_VSource; <none>
1082## @missing: 0000..10FFFF; cjkOtherNumeric; NaN
1083## @missing: 0000..10FFFF; cjkPrimaryNumeric; NaN
1084## @missing: 0000..10FFFF; cjkRSUnicode; <none>
1085END
1086
1087# The input files don't list every code point. Those not listed are to be
1088# defaulted to some value. Below are hard-coded what those values are for
1089# non-binary properties as of 5.1. Starting in 5.0, there are
caa75395 1090# machine-parsable comment lines in the files that give the defaults; so this
99870f4d
KW
1091# list shouldn't have to be extended. The claim is that all missing entries
1092# for binary properties will default to 'N'. Unicode tried to change that in
1093# 5.2, but the beta period produced enough protest that they backed off.
1094#
1095# The defaults for the fields that appear in UnicodeData.txt in this hash must
1096# be in the form that it expects. The others may be synonyms.
1097my $CODE_POINT = '<code point>';
1098my %default_mapping = (
1099 Age => "Unassigned",
1100 # Bidi_Class => Complicated; set in code
1101 Bidi_Mirroring_Glyph => "",
1102 Block => 'No_Block',
1103 Canonical_Combining_Class => 0,
1104 Case_Folding => $CODE_POINT,
1105 Decomposition_Mapping => $CODE_POINT,
1106 Decomposition_Type => 'None',
1107 East_Asian_Width => "Neutral",
1108 FC_NFKC_Closure => $CODE_POINT,
1109 General_Category => 'Cn',
1110 Grapheme_Cluster_Break => 'Other',
1111 Hangul_Syllable_Type => 'NA',
1112 ISO_Comment => "",
1113 Jamo_Short_Name => "",
1114 Joining_Group => "No_Joining_Group",
1115 # Joining_Type => Complicated; set in code
1116 kIICore => 'N', # Is converted to binary
1117 #Line_Break => Complicated; set in code
1118 Lowercase_Mapping => $CODE_POINT,
1119 Name => "",
1120 Name_Alias => "",
1121 NFC_QC => 'Yes',
1122 NFD_QC => 'Yes',
1123 NFKC_QC => 'Yes',
1124 NFKD_QC => 'Yes',
1125 Numeric_Type => 'None',
1126 Numeric_Value => 'NaN',
1127 Script => ($v_version le 4.1.0) ? 'Common' : 'Unknown',
1128 Sentence_Break => 'Other',
1129 Simple_Case_Folding => $CODE_POINT,
1130 Simple_Lowercase_Mapping => $CODE_POINT,
1131 Simple_Titlecase_Mapping => $CODE_POINT,
1132 Simple_Uppercase_Mapping => $CODE_POINT,
1133 Titlecase_Mapping => $CODE_POINT,
1134 Unicode_1_Name => "",
1135 Unicode_Radical_Stroke => "",
1136 Uppercase_Mapping => $CODE_POINT,
1137 Word_Break => 'Other',
1138);
1139
232ed87f
KW
1140# Below are files that Unicode furnishes, but this program ignores, and why.
1141# NormalizationCorrections.txt requires some more explanation. It documents
1142# the cumulative fixes to erroneous normalizations in earlier Unicode
1143# versions. Its main purpose is so that someone running on an earlier version
1144# can use this file to override what got published in that earlier release.
1145# It would be easy for mktables to read and handle this file. But all the
1146# corrections in it should already be in the other files for the release it
1147# is. To get it to actually mean something useful, someone would have to be
1148# using an earlier Unicode release, and copy it to the files for that release
1149# and recomplile. So far there has been no demand to do that, so this hasn't
1150# been implemented.
99870f4d 1151my %ignored_files = (
73ba1144
KW
1152 'CJKRadicals.txt' => 'Maps the kRSUnicode property values to corresponding code points',
1153 'Index.txt' => 'Alphabetical index of Unicode characters',
1154 'NamedSqProv.txt' => 'Named sequences proposed for inclusion in a later version of the Unicode Standard; if you need them now, you can append this file to F<NamedSequences.txt> and recompile perl',
1155 'NamesList.txt' => 'Annotated list of characters',
524a8e5e 1156 'NamesList.html' => 'Describes the format and contents of F<NamesList.txt>',
73ba1144
KW
1157 'NormalizationCorrections.txt' => 'Documentation of corrections already incorporated into the Unicode data base',
1158 'Props.txt' => 'Only in very early releases; is a subset of F<PropList.txt> (which is used instead)',
1159 'ReadMe.txt' => 'Documentation',
1160 'StandardizedVariants.txt' => 'Certain glyph variations for character display are standardized. This lists the non-Unihan ones; the Unihan ones are also not used by Perl, and are in a separate Unicode data base L<http://www.unicode.org/ivd>',
524a8e5e 1161 'StandardizedVariants.html' => 'Provides a visual display of the standard variant sequences derived from F<StandardizedVariants.txt>.',
73ba1144 1162 'EmojiSources.txt' => 'Maps certain Unicode code points to their legacy Japanese cell-phone values',
8ee2793f 1163 'USourceData.txt' => 'Documentation of status and cross reference of proposals for encoding by Unicode of Unihan characters',
524a8e5e 1164 'USourceGlyphs.pdf' => 'Pictures of the characters in F<USourceData.txt>',
73ba1144
KW
1165 'auxiliary/WordBreakTest.html' => 'Documentation of validation tests',
1166 'auxiliary/SentenceBreakTest.html' => 'Documentation of validation tests',
1167 'auxiliary/GraphemeBreakTest.html' => 'Documentation of validation tests',
1168 'auxiliary/LineBreakTest.html' => 'Documentation of validation tests',
99870f4d
KW
1169);
1170
1fec9f60
KW
1171my %skipped_files; # List of files that we skip
1172
678f13d5 1173### End of externally interesting definitions, except for @input_file_objects
99870f4d
KW
1174
1175my $HEADER=<<"EOF";
1176# !!!!!!! DO NOT EDIT THIS FILE !!!!!!!
3df51b85
KW
1177# This file is machine-generated by $0 from the Unicode
1178# database, Version $string_version. Any changes made here will be lost!
cf25bb62
JH
1179EOF
1180
126c3d4e 1181my $INTERNAL_ONLY_HEADER = <<"EOF";
99870f4d
KW
1182
1183# !!!!!!! INTERNAL PERL USE ONLY !!!!!!!
fac53429
KW
1184# This file is for internal use by core Perl only. The format and even the
1185# name or existence of this file are subject to change without notice. Don't
d9ae3878
KW
1186# use it directly. Use Unicode::UCD to access the Unicode character data
1187# base.
99870f4d
KW
1188EOF
1189
1190my $DEVELOPMENT_ONLY=<<"EOF";
1191# !!!!!!! DEVELOPMENT USE ONLY !!!!!!!
1192# This file contains information artificially constrained to code points
1193# present in Unicode release $string_compare_versions.
1194# IT CANNOT BE RELIED ON. It is for use during development only and should
23e33b60 1195# not be used for production.
b6922eda
KW
1196
1197EOF
1198
6189eadc
KW
1199my $MAX_UNICODE_CODEPOINT_STRING = "10FFFF";
1200my $MAX_UNICODE_CODEPOINT = hex $MAX_UNICODE_CODEPOINT_STRING;
1201my $MAX_UNICODE_CODEPOINTS = $MAX_UNICODE_CODEPOINT + 1;
99870f4d
KW
1202
1203# Matches legal code point. 4-6 hex numbers, If there are 6, the first
1204# two must be 10; if there are 5, the first must not be a 0. Written this way
92199589
KW
1205# to decrease backtracking. The first regex allows the code point to be at
1206# the end of a word, but to work properly, the word shouldn't end with a valid
1207# hex character. The second one won't match a code point at the end of a
1208# word, and doesn't have the run-on issue
8c32d378
KW
1209my $run_on_code_point_re =
1210 qr/ (?: 10[0-9A-F]{4} | [1-9A-F][0-9A-F]{4} | [0-9A-F]{4} ) \b/x;
1211my $code_point_re = qr/\b$run_on_code_point_re/;
99870f4d
KW
1212
1213# This matches the beginning of the line in the Unicode db files that give the
1214# defaults for code points not listed (i.e., missing) in the file. The code
1215# depends on this ending with a semi-colon, so it can assume it is a valid
1216# field when the line is split() by semi-colons
1217my $missing_defaults_prefix =
6189eadc 1218 qr/^#\s+\@missing:\s+0000\.\.$MAX_UNICODE_CODEPOINT_STRING\s*;/;
99870f4d
KW
1219
1220# Property types. Unicode has more types, but these are sufficient for our
1221# purposes.
1222my $UNKNOWN = -1; # initialized to illegal value
1223my $NON_STRING = 1; # Either binary or enum
1224my $BINARY = 2;
06f26c45
KW
1225my $FORCED_BINARY = 3; # Not a binary property, but, besides its normal
1226 # tables, additional true and false tables are
1227 # generated so that false is anything matching the
1228 # default value, and true is everything else.
1229my $ENUM = 4; # Include catalog
1230my $STRING = 5; # Anything else: string or misc
99870f4d
KW
1231
1232# Some input files have lines that give default values for code points not
1233# contained in the file. Sometimes these should be ignored.
1234my $NO_DEFAULTS = 0; # Must evaluate to false
1235my $NOT_IGNORED = 1;
1236my $IGNORED = 2;
1237
1238# Range types. Each range has a type. Most ranges are type 0, for normal,
1239# and will appear in the main body of the tables in the output files, but
1240# there are other types of ranges as well, listed below, that are specially
1241# handled. There are pseudo-types as well that will never be stored as a
1242# type, but will affect the calculation of the type.
1243
1244# 0 is for normal, non-specials
1245my $MULTI_CP = 1; # Sequence of more than code point
1246my $HANGUL_SYLLABLE = 2;
1247my $CP_IN_NAME = 3; # The NAME contains the code point appended to it.
1248my $NULL = 4; # The map is to the null string; utf8.c can't
1249 # handle these, nor is there an accepted syntax
1250 # for them in \p{} constructs
f86864ac 1251my $COMPUTE_NO_MULTI_CP = 5; # Pseudo-type; means that ranges that would
99870f4d
KW
1252 # otherwise be $MULTI_CP type are instead type 0
1253
1254# process_generic_property_file() can accept certain overrides in its input.
1255# Each of these must begin AND end with $CMD_DELIM.
1256my $CMD_DELIM = "\a";
1257my $REPLACE_CMD = 'replace'; # Override the Replace
1258my $MAP_TYPE_CMD = 'map_type'; # Override the Type
1259
1260my $NO = 0;
1261my $YES = 1;
1262
1263# Values for the Replace argument to add_range.
1264# $NO # Don't replace; add only the code points not
1265 # already present.
1266my $IF_NOT_EQUIVALENT = 1; # Replace only under certain conditions; details in
1267 # the comments at the subroutine definition.
1268my $UNCONDITIONALLY = 2; # Replace without conditions.
9470941f 1269my $MULTIPLE_BEFORE = 4; # Don't replace, but add a duplicate record if
99870f4d 1270 # already there
7f4b1e25
KW
1271my $MULTIPLE_AFTER = 5; # Don't replace, but add a duplicate record if
1272 # already there
1273my $CROAK = 6; # Die with an error if is already there
99870f4d
KW
1274
1275# Flags to give property statuses. The phrases are to remind maintainers that
1276# if the flag is changed, the indefinite article referring to it in the
1277# documentation may need to be as well.
1278my $NORMAL = "";
99870f4d
KW
1279my $DEPRECATED = 'D';
1280my $a_bold_deprecated = "a 'B<$DEPRECATED>'";
1281my $A_bold_deprecated = "A 'B<$DEPRECATED>'";
1282my $DISCOURAGED = 'X';
1283my $a_bold_discouraged = "an 'B<$DISCOURAGED>'";
1284my $A_bold_discouraged = "An 'B<$DISCOURAGED>'";
1285my $STRICTER = 'T';
1286my $a_bold_stricter = "a 'B<$STRICTER>'";
1287my $A_bold_stricter = "A 'B<$STRICTER>'";
1288my $STABILIZED = 'S';
1289my $a_bold_stabilized = "an 'B<$STABILIZED>'";
1290my $A_bold_stabilized = "An 'B<$STABILIZED>'";
1291my $OBSOLETE = 'O';
1292my $a_bold_obsolete = "an 'B<$OBSOLETE>'";
1293my $A_bold_obsolete = "An 'B<$OBSOLETE>'";
1294
1295my %status_past_participles = (
1296 $DISCOURAGED => 'discouraged',
99870f4d
KW
1297 $STABILIZED => 'stabilized',
1298 $OBSOLETE => 'obsolete',
37e2e78e 1299 $DEPRECATED => 'deprecated',
99870f4d
KW
1300);
1301
395dfc19
KW
1302# Table fates. These are somewhat ordered, so that fates < $MAP_PROXIED should be
1303# externally documented.
301ba948 1304my $ORDINARY = 0; # The normal fate.
395dfc19
KW
1305my $MAP_PROXIED = 1; # The map table for the property isn't written out,
1306 # but there is a file written that can be used to
1307 # reconstruct this table
3cdaf629 1308my $INTERNAL_ONLY = 2; # The file for this table is written out, but it is
301ba948 1309 # for Perl's internal use only
277b7b16
KW
1310my $LEGACY_ONLY = 3; # Like $INTERNAL_ONLY, but not actually used by Perl.
1311 # Is for backwards compatibility for applications that
1312 # read the file directly, so it's format is
1313 # unchangeable.
1314my $SUPPRESSED = 4; # The file for this table is not written out, and as a
3cdaf629
KW
1315 # result, we don't bother to do many computations on
1316 # it.
277b7b16 1317my $PLACEHOLDER = 5; # Like $SUPPRESSED, but we go through all the
3cdaf629
KW
1318 # computations anyway, as the values are needed for
1319 # things to work. This happens when we have Perl
1320 # extensions that depend on Unicode tables that
1321 # wouldn't normally be in a given Unicode version.
301ba948 1322
f5817e0a
KW
1323# The format of the values of the tables:
1324my $EMPTY_FORMAT = "";
99870f4d
KW
1325my $BINARY_FORMAT = 'b';
1326my $DECIMAL_FORMAT = 'd';
1327my $FLOAT_FORMAT = 'f';
1328my $INTEGER_FORMAT = 'i';
1329my $HEX_FORMAT = 'x';
1330my $RATIONAL_FORMAT = 'r';
1331my $STRING_FORMAT = 's';
d11155ec 1332my $ADJUST_FORMAT = 'a';
24303724 1333my $HEX_ADJUST_FORMAT = 'ax';
a14f3cb1 1334my $DECOMP_STRING_FORMAT = 'c';
c3ff2976 1335my $STRING_WHITE_SPACE_LIST = 'sw';
99870f4d
KW
1336
1337my %map_table_formats = (
1338 $BINARY_FORMAT => 'binary',
1339 $DECIMAL_FORMAT => 'single decimal digit',
1340 $FLOAT_FORMAT => 'floating point number',
1341 $INTEGER_FORMAT => 'integer',
add63c13 1342 $HEX_FORMAT => 'non-negative hex whole number; a code point',
99870f4d 1343 $RATIONAL_FORMAT => 'rational: an integer or a fraction',
1a9d544b 1344 $STRING_FORMAT => 'string',
d11155ec 1345 $ADJUST_FORMAT => 'some entries need adjustment',
24303724 1346 $HEX_ADJUST_FORMAT => 'mapped value in hex; some entries need adjustment',
92f9d56c 1347 $DECOMP_STRING_FORMAT => 'Perl\'s internal (Normalize.pm) decomposition mapping',
c3ff2976 1348 $STRING_WHITE_SPACE_LIST => 'string, but some elements are interpreted as a list; white space occurs only as list item separators'
99870f4d
KW
1349);
1350
1351# Unicode didn't put such derived files in a separate directory at first.
1352my $EXTRACTED_DIR = (-d 'extracted') ? 'extracted' : "";
1353my $EXTRACTED = ($EXTRACTED_DIR) ? "$EXTRACTED_DIR/" : "";
1354my $AUXILIARY = 'auxiliary';
1355
1356# Hashes that will eventually go into Heavy.pl for the use of utf8_heavy.pl
9e4a1e86 1357# and into UCD.pl for the use of UCD.pm
99870f4d
KW
1358my %loose_to_file_of; # loosely maps table names to their respective
1359 # files
1360my %stricter_to_file_of; # same; but for stricter mapping.
315bfd4e 1361my %loose_property_to_file_of; # Maps a loose property name to its map file
89cf10cc
KW
1362my %file_to_swash_name; # Maps the file name to its corresponding key name
1363 # in the hash %utf8::SwashInfo
99870f4d
KW
1364my %nv_floating_to_rational; # maps numeric values floating point numbers to
1365 # their rational equivalent
c12f2655
KW
1366my %loose_property_name_of; # Loosely maps (non_string) property names to
1367 # standard form
86a52d1e 1368my %string_property_loose_to_name; # Same, for string properties.
c15fda25
KW
1369my %loose_defaults; # keys are of form "prop=value", where 'prop' is
1370 # the property name in standard loose form, and
1371 # 'value' is the default value for that property,
1372 # also in standard loose form.
9e4a1e86
KW
1373my %loose_to_standard_value; # loosely maps table names to the canonical
1374 # alias for them
2df7880f
KW
1375my %ambiguous_names; # keys are alias names (in standard form) that
1376 # have more than one possible meaning.
5d1df013
KW
1377my %prop_aliases; # Keys are standard property name; values are each
1378 # one's aliases
1e863613
KW
1379my %prop_value_aliases; # Keys of top level are standard property name;
1380 # values are keys to another hash, Each one is
1381 # one of the property's values, in standard form.
1382 # The values are that prop-val's aliases.
2df7880f 1383my %ucd_pod; # Holds entries that will go into the UCD section of the pod
99870f4d 1384
d867ccfb
KW
1385# Most properties are immune to caseless matching, otherwise you would get
1386# nonsensical results, as properties are a function of a code point, not
1387# everything that is caselessly equivalent to that code point. For example,
1388# Changes_When_Case_Folded('s') should be false, whereas caselessly it would
1389# be true because 's' and 'S' are equivalent caselessly. However,
1390# traditionally, [:upper:] and [:lower:] are equivalent caselessly, so we
1391# extend that concept to those very few properties that are like this. Each
1392# such property will match the full range caselessly. They are hard-coded in
1393# the program; it's not worth trying to make it general as it's extremely
1394# unlikely that they will ever change.
1395my %caseless_equivalent_to;
1396
99870f4d
KW
1397# These constants names and values were taken from the Unicode standard,
1398# version 5.1, section 3.12. They are used in conjunction with Hangul
6e5a209b
KW
1399# syllables. The '_string' versions are so generated tables can retain the
1400# hex format, which is the more familiar value
1401my $SBase_string = "0xAC00";
1402my $SBase = CORE::hex $SBase_string;
1403my $LBase_string = "0x1100";
1404my $LBase = CORE::hex $LBase_string;
1405my $VBase_string = "0x1161";
1406my $VBase = CORE::hex $VBase_string;
1407my $TBase_string = "0x11A7";
1408my $TBase = CORE::hex $TBase_string;
99870f4d
KW
1409my $SCount = 11172;
1410my $LCount = 19;
1411my $VCount = 21;
1412my $TCount = 28;
1413my $NCount = $VCount * $TCount;
1414
1415# For Hangul syllables; These store the numbers from Jamo.txt in conjunction
1416# with the above published constants.
1417my %Jamo;
1418my %Jamo_L; # Leading consonants
1419my %Jamo_V; # Vowels
1420my %Jamo_T; # Trailing consonants
1421
bb1dd3da
KW
1422# For code points whose name contains its ordinal as a '-ABCD' suffix.
1423# The key is the base name of the code point, and the value is an
1424# array giving all the ranges that use this base name. Each range
1425# is actually a hash giving the 'low' and 'high' values of it.
1426my %names_ending_in_code_point;
1427my %loose_names_ending_in_code_point; # Same as above, but has blanks, dashes
1428 # removed from the names
1429# Inverse mapping. The list of ranges that have these kinds of
1430# names. Each element contains the low, high, and base names in an
1431# anonymous hash.
1432my @code_points_ending_in_code_point;
1433
6b5ab373
KW
1434# To hold Unicode's normalization test suite
1435my @normalization_tests;
1436
bb1dd3da
KW
1437# Boolean: does this Unicode version have the hangul syllables, and are we
1438# writing out a table for them?
1439my $has_hangul_syllables = 0;
1440
1441# Does this Unicode version have code points whose names end in their
1442# respective code points, and are we writing out a table for them? 0 for no;
1443# otherwise points to first property that a table is needed for them, so that
1444# if multiple tables are needed, we don't create duplicates
1445my $needing_code_points_ending_in_code_point = 0;
1446
37e2e78e 1447my @backslash_X_tests; # List of tests read in for testing \X
99870f4d
KW
1448my @unhandled_properties; # Will contain a list of properties found in
1449 # the input that we didn't process.
f86864ac 1450my @match_properties; # Properties that have match tables, to be
99870f4d
KW
1451 # listed in the pod
1452my @map_properties; # Properties that get map files written
1453my @named_sequences; # NamedSequences.txt contents.
1454my %potential_files; # Generated list of all .txt files in the directory
1455 # structure so we can warn if something is being
1456 # ignored.
1457my @files_actually_output; # List of files we generated.
1458my @more_Names; # Some code point names are compound; this is used
1459 # to store the extra components of them.
1460my $MIN_FRACTION_LENGTH = 3; # How many digits of a floating point number at
1461 # the minimum before we consider it equivalent to a
1462 # candidate rational
1463my $MAX_FLOATING_SLOP = 10 ** - $MIN_FRACTION_LENGTH; # And in floating terms
1464
1465# These store references to certain commonly used property objects
3c88a801 1466my $ccc;
99870f4d
KW
1467my $gc;
1468my $perl;
1469my $block;
3e20195b
KW
1470my $perl_charname;
1471my $print;
ac7dbdb6 1472my $All;
359523e2 1473my $script;
99870f4d
KW
1474
1475# Are there conflicting names because of beginning with 'In_', or 'Is_'
1476my $has_In_conflicts = 0;
1477my $has_Is_conflicts = 0;
1478
1479sub internal_file_to_platform ($) {
1480 # Convert our file paths which have '/' separators to those of the
1481 # platform.
1482
1483 my $file = shift;
1484 return undef unless defined $file;
1485
1486 return File::Spec->join(split '/', $file);
d07a55ed 1487}
5beb625e 1488
99870f4d
KW
1489sub file_exists ($) { # platform independent '-e'. This program internally
1490 # uses slash as a path separator.
1491 my $file = shift;
1492 return 0 if ! defined $file;
1493 return -e internal_file_to_platform($file);
1494}
5beb625e 1495
99870f4d 1496sub objaddr($) {
23e33b60
KW
1497 # Returns the address of the blessed input object.
1498 # It doesn't check for blessedness because that would do a string eval
1499 # every call, and the program is structured so that this is never called
1500 # for a non-blessed object.
99870f4d 1501
23e33b60 1502 no overloading; # If overloaded, numifying below won't work.
99870f4d
KW
1503
1504 # Numifying a ref gives its address.
051df77b 1505 return pack 'J', $_[0];
99870f4d
KW
1506}
1507
558712cf 1508# These are used only if $annotate is true.
c4019d52
KW
1509# The entire range of Unicode characters is examined to populate these
1510# after all the input has been processed. But most can be skipped, as they
1511# have the same descriptive phrases, such as being unassigned
1512my @viacode; # Contains the 1 million character names
1513my @printable; # boolean: And are those characters printable?
1514my @annotate_char_type; # Contains a type of those characters, specifically
1515 # for the purposes of annotation.
1516my $annotate_ranges; # A map of ranges of code points that have the same
98dc9551 1517 # name for the purposes of annotation. They map to the
c4019d52
KW
1518 # upper edge of the range, so that the end point can
1519 # be immediately found. This is used to skip ahead to
1520 # the end of a range, and avoid processing each
1521 # individual code point in it.
1522my $unassigned_sans_noncharacters; # A Range_List of the unassigned
1523 # characters, but excluding those which are
1524 # also noncharacter code points
1525
1526# The annotation types are an extension of the regular range types, though
1527# some of the latter are folded into one. Make the new types negative to
1528# avoid conflicting with the regular types
1529my $SURROGATE_TYPE = -1;
1530my $UNASSIGNED_TYPE = -2;
1531my $PRIVATE_USE_TYPE = -3;
1532my $NONCHARACTER_TYPE = -4;
1533my $CONTROL_TYPE = -5;
1534my $UNKNOWN_TYPE = -6; # Used only if there is a bug in this program
1535
1536sub populate_char_info ($) {
558712cf 1537 # Used only with the $annotate option. Populates the arrays with the
c4019d52
KW
1538 # input code point's info that are needed for outputting more detailed
1539 # comments. If calling context wants a return, it is the end point of
1540 # any contiguous range of characters that share essentially the same info
1541
1542 my $i = shift;
1543 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
1544
1545 $viacode[$i] = $perl_charname->value_of($i) || "";
1546
1547 # A character is generally printable if Unicode says it is,
1548 # but below we make sure that most Unicode general category 'C' types
1549 # aren't.
1550 $printable[$i] = $print->contains($i);
1551
1552 $annotate_char_type[$i] = $perl_charname->type_of($i) || 0;
1553
1554 # Only these two regular types are treated specially for annotations
1555 # purposes
1556 $annotate_char_type[$i] = 0 if $annotate_char_type[$i] != $CP_IN_NAME
1557 && $annotate_char_type[$i] != $HANGUL_SYLLABLE;
1558
1559 # Give a generic name to all code points that don't have a real name.
1560 # We output ranges, if applicable, for these. Also calculate the end
1561 # point of the range.
1562 my $end;
1563 if (! $viacode[$i]) {
1d025d66
KW
1564 my $nonchar;
1565 if ($gc-> table('Private_use')->contains($i)) {
c4019d52
KW
1566 $viacode[$i] = 'Private Use';
1567 $annotate_char_type[$i] = $PRIVATE_USE_TYPE;
1568 $printable[$i] = 0;
1569 $end = $gc->table('Private_Use')->containing_range($i)->end;
1570 }
1d025d66
KW
1571 elsif ((defined ($nonchar =
1572 Property::property_ref('Noncharacter_Code_Point'))
1573 && $nonchar->table('Y')->contains($i)))
c4019d52
KW
1574 {
1575 $viacode[$i] = 'Noncharacter';
1576 $annotate_char_type[$i] = $NONCHARACTER_TYPE;
1577 $printable[$i] = 0;
1578 $end = property_ref('Noncharacter_Code_Point')->table('Y')->
1579 containing_range($i)->end;
1580 }
1581 elsif ($gc-> table('Control')->contains($i)) {
c71dea7f 1582 $viacode[$i] = property_ref('Name_Alias')->value_of($i) || 'Control';
c4019d52
KW
1583 $annotate_char_type[$i] = $CONTROL_TYPE;
1584 $printable[$i] = 0;
c4019d52
KW
1585 }
1586 elsif ($gc-> table('Unassigned')->contains($i)) {
c4019d52
KW
1587 $annotate_char_type[$i] = $UNASSIGNED_TYPE;
1588 $printable[$i] = 0;
1d025d66
KW
1589 if ($v_version lt v2.0.0) { # No blocks in earliest releases
1590 $viacode[$i] = 'Unassigned';
1591 $end = $gc-> table('Unassigned')->containing_range($i)->end;
1592 }
1593 else {
1594 $viacode[$i] = 'Unassigned, block=' . $block-> value_of($i);
c4019d52 1595
bf06c733
KW
1596 # Because we name the unassigned by the blocks they are in, it
1597 # can't go past the end of that block, and it also can't go
1598 # past the unassigned range it is in. The special table makes
1599 # sure that the non-characters, which are unassigned, are
1600 # separated out.
1601 $end = min($block->containing_range($i)->end,
1602 $unassigned_sans_noncharacters->
1603 containing_range($i)->end);
1d025d66
KW
1604 }
1605 }
1606 elsif ($v_version lt v2.0.0) { # No surrogates in earliest releases
1607 $viacode[$i] = $gc->value_of($i);
1608 $annotate_char_type[$i] = $UNKNOWN_TYPE;
1609 $printable[$i] = 0;
1610 }
1611 elsif ($gc-> table('Surrogate')->contains($i)) {
1612 $viacode[$i] = 'Surrogate';
1613 $annotate_char_type[$i] = $SURROGATE_TYPE;
1614 $printable[$i] = 0;
1615 $end = $gc->table('Surrogate')->containing_range($i)->end;
13ca76ff
KW
1616 }
1617 else {
1618 Carp::my_carp_bug("Can't figure out how to annotate "
1619 . sprintf("U+%04X", $i)
1620 . ". Proceeding anyway.");
c4019d52
KW
1621 $viacode[$i] = 'UNKNOWN';
1622 $annotate_char_type[$i] = $UNKNOWN_TYPE;
1623 $printable[$i] = 0;
1624 }
1625 }
1626
1627 # Here, has a name, but if it's one in which the code point number is
1628 # appended to the name, do that.
1629 elsif ($annotate_char_type[$i] == $CP_IN_NAME) {
1630 $viacode[$i] .= sprintf("-%04X", $i);
1631 $end = $perl_charname->containing_range($i)->end;
1632 }
1633
1634 # And here, has a name, but if it's a hangul syllable one, replace it with
1635 # the correct name from the Unicode algorithm
1636 elsif ($annotate_char_type[$i] == $HANGUL_SYLLABLE) {
1637 use integer;
1638 my $SIndex = $i - $SBase;
1639 my $L = $LBase + $SIndex / $NCount;
1640 my $V = $VBase + ($SIndex % $NCount) / $TCount;
1641 my $T = $TBase + $SIndex % $TCount;
1642 $viacode[$i] = "HANGUL SYLLABLE $Jamo{$L}$Jamo{$V}";
1643 $viacode[$i] .= $Jamo{$T} if $T != $TBase;
1644 $end = $perl_charname->containing_range($i)->end;
1645 }
1646
1647 return if ! defined wantarray;
1648 return $i if ! defined $end; # If not a range, return the input
1649
1650 # Save this whole range so can find the end point quickly
1651 $annotate_ranges->add_map($i, $end, $end);
1652
1653 return $end;
1654}
1655
23e33b60
KW
1656# Commented code below should work on Perl 5.8.
1657## This 'require' doesn't necessarily work in miniperl, and even if it does,
1658## the native perl version of it (which is what would operate under miniperl)
1659## is extremely slow, as it does a string eval every call.
7e017d6d 1660#my $has_fast_scalar_util = $^X !~ /miniperl/
23e33b60
KW
1661# && defined eval "require Scalar::Util";
1662#
1663#sub objaddr($) {
1664# # Returns the address of the blessed input object. Uses the XS version if
1665# # available. It doesn't check for blessedness because that would do a
1666# # string eval every call, and the program is structured so that this is
1667# # never called for a non-blessed object.
1668#
1669# return Scalar::Util::refaddr($_[0]) if $has_fast_scalar_util;
1670#
1671# # Check at least that is a ref.
1672# my $pkg = ref($_[0]) or return undef;
1673#
1674# # Change to a fake package to defeat any overloaded stringify
1675# bless $_[0], 'main::Fake';
1676#
1677# # Numifying a ref gives its address.
051df77b 1678# my $addr = pack 'J', $_[0];
23e33b60
KW
1679#
1680# # Return to original class
1681# bless $_[0], $pkg;
1682# return $addr;
1683#}
1684
99870f4d
KW
1685sub max ($$) {
1686 my $a = shift;
1687 my $b = shift;
1688 return $a if $a >= $b;
1689 return $b;
1690}
1691
1692sub min ($$) {
1693 my $a = shift;
1694 my $b = shift;
1695 return $a if $a <= $b;
1696 return $b;
1697}
1698
1699sub clarify_number ($) {
1700 # This returns the input number with underscores inserted every 3 digits
1701 # in large (5 digits or more) numbers. Input must be entirely digits, not
1702 # checked.
1703
1704 my $number = shift;
1705 my $pos = length($number) - 3;
1706 return $number if $pos <= 1;
1707 while ($pos > 0) {
1708 substr($number, $pos, 0) = '_';
1709 $pos -= 3;
5beb625e 1710 }
99870f4d 1711 return $number;
99598c8c
JH
1712}
1713
731cb813
KW
1714sub clarify_code_point_count ($) {
1715 # This is like clarify_number(), but the input is assumed to be a count of
1716 # code points, rather than a generic number.
1717
1718 return clarify_number(shift);
1719}
12ac2576 1720
99870f4d 1721package Carp;
7ebf06b3 1722
99870f4d
KW
1723# These routines give a uniform treatment of messages in this program. They
1724# are placed in the Carp package to cause the stack trace to not include them,
1725# although an alternative would be to use another package and set @CARP_NOT
1726# for it.
12ac2576 1727
99870f4d 1728our $Verbose = 1 if main::DEBUG; # Useful info when debugging
12ac2576 1729
99f78760
KW
1730# This is a work-around suggested by Nicholas Clark to fix a problem with Carp
1731# and overload trying to load Scalar:Util under miniperl. See
1732# http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2009-11/msg01057.html
1733undef $overload::VERSION;
1734
99870f4d
KW
1735sub my_carp {
1736 my $message = shift || "";
1737 my $nofold = shift || 0;
7ebf06b3 1738
99870f4d
KW
1739 if ($message) {
1740 $message = main::join_lines($message);
1741 $message =~ s/^$0: *//; # Remove initial program name
1742 $message =~ s/[.;,]+$//; # Remove certain ending punctuation
1743 $message = "\n$0: $message;";
12ac2576 1744
99870f4d
KW
1745 # Fold the message with program name, semi-colon end punctuation
1746 # (which looks good with the message that carp appends to it), and a
1747 # hanging indent for continuation lines.
1748 $message = main::simple_fold($message, "", 4) unless $nofold;
1749 $message =~ s/\n$//; # Remove the trailing nl so what carp
1750 # appends is to the same line
1751 }
12ac2576 1752
99870f4d 1753 return $message if defined wantarray; # If a caller just wants the msg
12ac2576 1754
99870f4d
KW
1755 carp $message;
1756 return;
1757}
7ebf06b3 1758
99870f4d
KW
1759sub my_carp_bug {
1760 # This is called when it is clear that the problem is caused by a bug in
1761 # this program.
7ebf06b3 1762
99870f4d
KW
1763 my $message = shift;
1764 $message =~ s/^$0: *//;
1765 $message = my_carp("Bug in $0. Please report it by running perlbug or if that is unavailable, by sending email to perbug\@perl.org:\n$message");
1766 carp $message;
1767 return;
1768}
7ebf06b3 1769
99870f4d
KW
1770sub carp_too_few_args {
1771 if (@_ != 2) {
1772 my_carp_bug("Wrong number of arguments: to 'carp_too_few_arguments'. No action taken.");
1773 return;
12ac2576 1774 }
7ebf06b3 1775
99870f4d
KW
1776 my $args_ref = shift;
1777 my $count = shift;
7ebf06b3 1778
99870f4d
KW
1779 my_carp_bug("Need at least $count arguments to "
1780 . (caller 1)[3]
1781 . ". Instead got: '"
1782 . join ', ', @$args_ref
1783 . "'. No action taken.");
1784 return;
12ac2576
JP
1785}
1786
99870f4d
KW
1787sub carp_extra_args {
1788 my $args_ref = shift;
1789 my_carp_bug("Too many arguments to 'carp_extra_args': (" . join(', ', @_) . "); Extras ignored.") if @_;
12ac2576 1790
99870f4d
KW
1791 unless (ref $args_ref) {
1792 my_carp_bug("Argument to 'carp_extra_args' ($args_ref) must be a ref. Not checking arguments.");
1793 return;
1794 }
1795 my ($package, $file, $line) = caller;
1796 my $subroutine = (caller 1)[3];
cf25bb62 1797
99870f4d
KW
1798 my $list;
1799 if (ref $args_ref eq 'HASH') {
1800 foreach my $key (keys %$args_ref) {
1801 $args_ref->{$key} = $UNDEF unless defined $args_ref->{$key};
cf25bb62 1802 }
99870f4d 1803 $list = join ', ', each %{$args_ref};
cf25bb62 1804 }
99870f4d
KW
1805 elsif (ref $args_ref eq 'ARRAY') {
1806 foreach my $arg (@$args_ref) {
1807 $arg = $UNDEF unless defined $arg;
1808 }
1809 $list = join ', ', @$args_ref;
1810 }
1811 else {
1812 my_carp_bug("Can't cope with ref "
1813 . ref($args_ref)
1814 . " . argument to 'carp_extra_args'. Not checking arguments.");
1815 return;
1816 }
1817
1818 my_carp_bug("Unrecognized parameters in options: '$list' to $subroutine. Skipped.");
1819 return;
d73e5302
JH
1820}
1821
99870f4d
KW
1822package main;
1823
1824{ # Closure
1825
1826 # This program uses the inside-out method for objects, as recommended in
1827 # "Perl Best Practices". This closure aids in generating those. There
1828 # are two routines. setup_package() is called once per package to set
1829 # things up, and then set_access() is called for each hash representing a
1830 # field in the object. These routines arrange for the object to be
1831 # properly destroyed when no longer used, and for standard accessor
1832 # functions to be generated. If you need more complex accessors, just
1833 # write your own and leave those accesses out of the call to set_access().
1834 # More details below.
1835
1836 my %constructor_fields; # fields that are to be used in constructors; see
1837 # below
1838
1839 # The values of this hash will be the package names as keys to other
1840 # hashes containing the name of each field in the package as keys, and
1841 # references to their respective hashes as values.
1842 my %package_fields;
1843
1844 sub setup_package {
1845 # Sets up the package, creating standard DESTROY and dump methods
1846 # (unless already defined). The dump method is used in debugging by
1847 # simple_dumper().
1848 # The optional parameters are:
1849 # a) a reference to a hash, that gets populated by later
1850 # set_access() calls with one of the accesses being
1851 # 'constructor'. The caller can then refer to this, but it is
1852 # not otherwise used by these two routines.
1853 # b) a reference to a callback routine to call during destruction
1854 # of the object, before any fields are actually destroyed
1855
1856 my %args = @_;
1857 my $constructor_ref = delete $args{'Constructor_Fields'};
1858 my $destroy_callback = delete $args{'Destroy_Callback'};
1859 Carp::carp_extra_args(\@_) if main::DEBUG && %args;
1860
1861 my %fields;
1862 my $package = (caller)[0];
1863
1864 $package_fields{$package} = \%fields;
1865 $constructor_fields{$package} = $constructor_ref;
1866
1867 unless ($package->can('DESTROY')) {
1868 my $destroy_name = "${package}::DESTROY";
1869 no strict "refs";
1870
1871 # Use typeglob to give the anonymous subroutine the name we want
1872 *$destroy_name = sub {
1873 my $self = shift;
ffe43484 1874 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
1875
1876 $self->$destroy_callback if $destroy_callback;
1877 foreach my $field (keys %{$package_fields{$package}}) {
1878 #print STDERR __LINE__, ": Destroying ", ref $self, " ", sprintf("%04X", $addr), ": ", $field, "\n";
1879 delete $package_fields{$package}{$field}{$addr};
1880 }
1881 return;
1882 }
1883 }
1884
1885 unless ($package->can('dump')) {
1886 my $dump_name = "${package}::dump";
1887 no strict "refs";
1888 *$dump_name = sub {
1889 my $self = shift;
1890 return dump_inside_out($self, $package_fields{$package}, @_);
1891 }
1892 }
1893 return;
1894 }
1895
1896 sub set_access {
1897 # Arrange for the input field to be garbage collected when no longer
1898 # needed. Also, creates standard accessor functions for the field
1899 # based on the optional parameters-- none if none of these parameters:
1900 # 'addable' creates an 'add_NAME()' accessor function.
1901 # 'readable' or 'readable_array' creates a 'NAME()' accessor
1902 # function.
1903 # 'settable' creates a 'set_NAME()' accessor function.
1904 # 'constructor' doesn't create an accessor function, but adds the
1905 # field to the hash that was previously passed to
1906 # setup_package();
1907 # Any of the accesses can be abbreviated down, so that 'a', 'ad',
1908 # 'add' etc. all mean 'addable'.
1909 # The read accessor function will work on both array and scalar
1910 # values. If another accessor in the parameter list is 'a', the read
1911 # access assumes an array. You can also force it to be array access
1912 # by specifying 'readable_array' instead of 'readable'
1913 #
1914 # A sort-of 'protected' access can be set-up by preceding the addable,
1915 # readable or settable with some initial portion of 'protected_' (but,
1916 # the underscore is required), like 'p_a', 'pro_set', etc. The
1917 # "protection" is only by convention. All that happens is that the
1918 # accessor functions' names begin with an underscore. So instead of
1919 # calling set_foo, the call is _set_foo. (Real protection could be
c1739a4a 1920 # accomplished by having a new subroutine, end_package, called at the
99870f4d
KW
1921 # end of each package, and then storing the __LINE__ ranges and
1922 # checking them on every accessor. But that is way overkill.)
1923
1924 # We create anonymous subroutines as the accessors and then use
1925 # typeglobs to assign them to the proper package and name
1926
1927 my $name = shift; # Name of the field
1928 my $field = shift; # Reference to the inside-out hash containing the
1929 # field
1930
1931 my $package = (caller)[0];
1932
1933 if (! exists $package_fields{$package}) {
1934 croak "$0: Must call 'setup_package' before 'set_access'";
1935 }
d73e5302 1936
99870f4d
KW
1937 # Stash the field so DESTROY can get it.
1938 $package_fields{$package}{$name} = $field;
cf25bb62 1939
99870f4d
KW
1940 # Remaining arguments are the accessors. For each...
1941 foreach my $access (@_) {
1942 my $access = lc $access;
cf25bb62 1943
99870f4d 1944 my $protected = "";
cf25bb62 1945
99870f4d
KW
1946 # Match the input as far as it goes.
1947 if ($access =~ /^(p[^_]*)_/) {
1948 $protected = $1;
1949 if (substr('protected_', 0, length $protected)
1950 eq $protected)
1951 {
1952
1953 # Add 1 for the underscore not included in $protected
1954 $access = substr($access, length($protected) + 1);
1955 $protected = '_';
1956 }
1957 else {
1958 $protected = "";
1959 }
1960 }
1961
1962 if (substr('addable', 0, length $access) eq $access) {
1963 my $subname = "${package}::${protected}add_$name";
1964 no strict "refs";
1965
1966 # add_ accessor. Don't add if already there, which we
1967 # determine using 'eq' for scalars and '==' otherwise.
1968 *$subname = sub {
1969 use strict "refs";
1970 return Carp::carp_too_few_args(\@_, 2) if main::DEBUG && @_ < 2;
1971 my $self = shift;
1972 my $value = shift;
ffe43484 1973 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
1974 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
1975 if (ref $value) {
f998e60c 1976 return if grep { $value == $_ } @{$field->{$addr}};
99870f4d
KW
1977 }
1978 else {
f998e60c 1979 return if grep { $value eq $_ } @{$field->{$addr}};
99870f4d 1980 }
f998e60c 1981 push @{$field->{$addr}}, $value;
99870f4d
KW
1982 return;
1983 }
1984 }
1985 elsif (substr('constructor', 0, length $access) eq $access) {
1986 if ($protected) {
1987 Carp::my_carp_bug("Can't set-up 'protected' constructors")
1988 }
1989 else {
1990 $constructor_fields{$package}{$name} = $field;
1991 }
1992 }
1993 elsif (substr('readable_array', 0, length $access) eq $access) {
1994
1995 # Here has read access. If one of the other parameters for
1996 # access is array, or this one specifies array (by being more
1997 # than just 'readable_'), then create a subroutine that
1998 # assumes the data is an array. Otherwise just a scalar
1999 my $subname = "${package}::${protected}$name";
2000 if (grep { /^a/i } @_
2001 or length($access) > length('readable_'))
2002 {
2003 no strict "refs";
2004 *$subname = sub {
2005 use strict "refs";
23e33b60 2006 Carp::carp_extra_args(\@_) if main::DEBUG && @_ > 1;
ffe43484 2007 my $addr = do { no overloading; pack 'J', $_[0]; };
99870f4d
KW
2008 if (ref $field->{$addr} ne 'ARRAY') {
2009 my $type = ref $field->{$addr};
2010 $type = 'scalar' unless $type;
2011 Carp::my_carp_bug("Trying to read $name as an array when it is a $type. Big problems.");
2012 return;
2013 }
2014 return scalar @{$field->{$addr}} unless wantarray;
2015
2016 # Make a copy; had problems with caller modifying the
2017 # original otherwise
2018 my @return = @{$field->{$addr}};
2019 return @return;
2020 }
2021 }
2022 else {
2023
2024 # Here not an array value, a simpler function.
2025 no strict "refs";
2026 *$subname = sub {
2027 use strict "refs";
23e33b60 2028 Carp::carp_extra_args(\@_) if main::DEBUG && @_ > 1;
f998e60c 2029 no overloading;
051df77b 2030 return $field->{pack 'J', $_[0]};
99870f4d
KW
2031 }
2032 }
2033 }
2034 elsif (substr('settable', 0, length $access) eq $access) {
2035 my $subname = "${package}::${protected}set_$name";
2036 no strict "refs";
2037 *$subname = sub {
2038 use strict "refs";
23e33b60
KW
2039 if (main::DEBUG) {
2040 return Carp::carp_too_few_args(\@_, 2) if @_ < 2;
2041 Carp::carp_extra_args(\@_) if @_ > 2;
2042 }
2043 # $self is $_[0]; $value is $_[1]
f998e60c 2044 no overloading;
051df77b 2045 $field->{pack 'J', $_[0]} = $_[1];
99870f4d
KW
2046 return;
2047 }
2048 }
2049 else {
2050 Carp::my_carp_bug("Unknown accessor type $access. No accessor set.");
2051 }
cf25bb62 2052 }
99870f4d 2053 return;
cf25bb62 2054 }
99870f4d
KW
2055}
2056
2057package Input_file;
2058
2059# All input files use this object, which stores various attributes about them,
2060# and provides for convenient, uniform handling. The run method wraps the
2061# processing. It handles all the bookkeeping of opening, reading, and closing
2062# the file, returning only significant input lines.
2063#
2064# Each object gets a handler which processes the body of the file, and is
74cd47d0
KW
2065# called by run(). All character property files must use the generic,
2066# default handler, which has code scrubbed to handle things you might not
2067# expect, including automatic EBCDIC handling. For files that don't deal with
2068# mapping code points to a property value, such as test files,
2069# PropertyAliases, PropValueAliases, and named sequences, you can override the
2070# handler to be a custom one. Such a handler should basically be a
2071# while(next_line()) {...} loop.
99870f4d
KW
2072#
2073# You can also set up handlers to
537124e4 2074# 1) call before the first line is read, for pre processing
83b68635
KW
2075# 2) call to adjust each line of the input before the main handler gets
2076# them. This can be automatically generated, if appropriately simple
2077# enough, by specifiying a Properties parameter in the constructor.
99870f4d 2078# 3) call upon EOF before the main handler exits its loop
537124e4 2079# 4) call at the end, for post processing
99870f4d
KW
2080#
2081# $_ is used to store the input line, and is to be filtered by the
2082# each_line_handler()s. So, if the format of the line is not in the desired
2083# format for the main handler, these are used to do that adjusting. They can
2084# be stacked (by enclosing them in an [ anonymous array ] in the constructor,
2085# so the $_ output of one is used as the input to the next. None of the other
2086# handlers are stackable, but could easily be changed to be so.
2087#
2088# Most of the handlers can call insert_lines() or insert_adjusted_lines()
2089# which insert the parameters as lines to be processed before the next input
2090# file line is read. This allows the EOF handler to flush buffers, for
2091# example. The difference between the two routines is that the lines inserted
2092# by insert_lines() are subjected to the each_line_handler()s. (So if you
2093# called it from such a handler, you would get infinite recursion.) Lines
2094# inserted by insert_adjusted_lines() go directly to the main handler without
2095# any adjustments. If the post-processing handler calls any of these, there
2096# will be no effect. Some error checking for these conditions could be added,
2097# but it hasn't been done.
2098#
2099# carp_bad_line() should be called to warn of bad input lines, which clears $_
2100# to prevent further processing of the line. This routine will output the
2101# message as a warning once, and then keep a count of the lines that have the
2102# same message, and output that count at the end of the file's processing.
2103# This keeps the number of messages down to a manageable amount.
2104#
2105# get_missings() should be called to retrieve any @missing input lines.
2106# Messages will be raised if this isn't done if the options aren't to ignore
2107# missings.
2108
2109sub trace { return main::trace(@_); }
2110
99870f4d
KW
2111{ # Closure
2112 # Keep track of fields that are to be put into the constructor.
2113 my %constructor_fields;
2114
2115 main::setup_package(Constructor_Fields => \%constructor_fields);
2116
2117 my %file; # Input file name, required
2118 main::set_access('file', \%file, qw{ c r });
2119
2120 my %first_released; # Unicode version file was first released in, required
2121 main::set_access('first_released', \%first_released, qw{ c r });
2122
2123 my %handler; # Subroutine to process the input file, defaults to
2124 # 'process_generic_property_file'
2125 main::set_access('handler', \%handler, qw{ c });
2126
2127 my %property;
2128 # name of property this file is for. defaults to none, meaning not
2129 # applicable, or is otherwise determinable, for example, from each line.
696609bf 2130 main::set_access('property', \%property, qw{ c r });
99870f4d
KW
2131
2132 my %optional;
2133 # If this is true, the file is optional. If not present, no warning is
2134 # output. If it is present, the string given by this parameter is
2135 # evaluated, and if false the file is not processed.
2136 main::set_access('optional', \%optional, 'c', 'r');
2137
2138 my %non_skip;
2139 # This is used for debugging, to skip processing of all but a few input
2140 # files. Add 'non_skip => 1' to the constructor for those files you want
2141 # processed when you set the $debug_skip global.
2142 main::set_access('non_skip', \%non_skip, 'c');
2143
37e2e78e 2144 my %skip;
09ca89ce
KW
2145 # This is used to skip processing of this input file semi-permanently,
2146 # when it evaluates to true. The value should be the reason the file is
2147 # being skipped. It is used for files that we aren't planning to process
2148 # anytime soon, but want to allow to be in the directory and not raise a
2149 # message that we are not handling. Mostly for test files. This is in
2150 # contrast to the non_skip element, which is supposed to be used very
2151 # temporarily for debugging. Sets 'optional' to 1. Also, files that we
2152 # pretty much will never look at can be placed in the global
1fec9f60 2153 # %ignored_files instead. Ones used here will be added to %skipped files
37e2e78e
KW
2154 main::set_access('skip', \%skip, 'c');
2155
99870f4d
KW
2156 my %each_line_handler;
2157 # list of subroutines to look at and filter each non-comment line in the
2158 # file. defaults to none. The subroutines are called in order, each is
2159 # to adjust $_ for the next one, and the final one adjusts it for
2160 # 'handler'
2161 main::set_access('each_line_handler', \%each_line_handler, 'c');
2162
83b68635
KW
2163 my %properties; # Optional ordered list of the properties that occur in each
2164 # meaningful line of the input file. If present, an appropriate
2165 # each_line_handler() is automatically generated and pushed onto the stack
2166 # of such handlers. This is useful when a file contains multiple
2167 # proerties per line, but no other special considerations are necessary.
2168 # The special value "<ignored>" means to discard the corresponding input
2169 # field.
2170 # Any @missing lines in the file should also match this syntax; no such
2171 # files exist as of 6.3. But if it happens in a future release, the code
2172 # could be expanded to properly parse them.
2173 main::set_access('properties', \%properties, qw{ c r });
2174
99870f4d
KW
2175 my %has_missings_defaults;
2176 # ? Are there lines in the file giving default values for code points
2177 # missing from it?. Defaults to NO_DEFAULTS. Otherwise NOT_IGNORED is
2178 # the norm, but IGNORED means it has such lines, but the handler doesn't
2179 # use them. Having these three states allows us to catch changes to the
83b68635
KW
2180 # UCD that this program should track. XXX This could be expanded to
2181 # specify the syntax for such lines, like %properties above.
99870f4d
KW
2182 main::set_access('has_missings_defaults',
2183 \%has_missings_defaults, qw{ c r });
2184
2185 my %pre_handler;
2186 # Subroutine to call before doing anything else in the file. If undef, no
2187 # such handler is called.
2188 main::set_access('pre_handler', \%pre_handler, qw{ c });
2189
2190 my %eof_handler;
2191 # Subroutine to call upon getting an EOF on the input file, but before
2192 # that is returned to the main handler. This is to allow buffers to be
2193 # flushed. The handler is expected to call insert_lines() or
2194 # insert_adjusted() with the buffered material
2195 main::set_access('eof_handler', \%eof_handler, qw{ c r });
2196
2197 my %post_handler;
2198 # Subroutine to call after all the lines of the file are read in and
2199 # processed. If undef, no such handler is called.
2200 main::set_access('post_handler', \%post_handler, qw{ c });
2201
2202 my %progress_message;
2203 # Message to print to display progress in lieu of the standard one
2204 main::set_access('progress_message', \%progress_message, qw{ c });
2205
2206 my %handle;
2207 # cache open file handle, internal. Is undef if file hasn't been
2208 # processed at all, empty if has;
2209 main::set_access('handle', \%handle);
2210
2211 my %added_lines;
2212 # cache of lines added virtually to the file, internal
2213 main::set_access('added_lines', \%added_lines);
2214
74cd47d0
KW
2215 my %remapped_lines;
2216 # cache of lines added virtually to the file, internal
2217 main::set_access('remapped_lines', \%remapped_lines);
2218
99870f4d
KW
2219 my %errors;
2220 # cache of errors found, internal
2221 main::set_access('errors', \%errors);
2222
2223 my %missings;
2224 # storage of '@missing' defaults lines
2225 main::set_access('missings', \%missings);
2226
74cd47d0
KW
2227 sub _next_line;
2228 sub _next_line_with_remapped_range;
2229
99870f4d
KW
2230 sub new {
2231 my $class = shift;
2232
2233 my $self = bless \do{ my $anonymous_scalar }, $class;
ffe43484 2234 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
2235
2236 # Set defaults
2237 $handler{$addr} = \&main::process_generic_property_file;
2238 $non_skip{$addr} = 0;
37e2e78e 2239 $skip{$addr} = 0;
99870f4d
KW
2240 $has_missings_defaults{$addr} = $NO_DEFAULTS;
2241 $handle{$addr} = undef;
2242 $added_lines{$addr} = [ ];
74cd47d0 2243 $remapped_lines{$addr} = [ ];
99870f4d
KW
2244 $each_line_handler{$addr} = [ ];
2245 $errors{$addr} = { };
2246 $missings{$addr} = [ ];
2247
2248 # Two positional parameters.
99f78760 2249 return Carp::carp_too_few_args(\@_, 2) if main::DEBUG && @_ < 2;
99870f4d
KW
2250 $file{$addr} = main::internal_file_to_platform(shift);
2251 $first_released{$addr} = shift;
2252
2253 # The rest of the arguments are key => value pairs
2254 # %constructor_fields has been set up earlier to list all possible
2255 # ones. Either set or push, depending on how the default has been set
2256 # up just above.
2257 my %args = @_;
2258 foreach my $key (keys %args) {
2259 my $argument = $args{$key};
2260
2261 # Note that the fields are the lower case of the constructor keys
2262 my $hash = $constructor_fields{lc $key};
2263 if (! defined $hash) {
2264 Carp::my_carp_bug("Unrecognized parameters '$key => $argument' to new() for $self. Skipped");
2265 next;
2266 }
2267 if (ref $hash->{$addr} eq 'ARRAY') {
2268 if (ref $argument eq 'ARRAY') {
2269 foreach my $argument (@{$argument}) {
2270 next if ! defined $argument;
2271 push @{$hash->{$addr}}, $argument;
2272 }
2273 }
2274 else {
2275 push @{$hash->{$addr}}, $argument if defined $argument;
2276 }
2277 }
2278 else {
2279 $hash->{$addr} = $argument;
2280 }
2281 delete $args{$key};
2282 };
2283
2284 # If the file has a property for it, it means that the property is not
2285 # listed in the file's entries. So add a handler to the list of line
2286 # handlers to insert the property name into the lines, to provide a
2287 # uniform interface to the final processing subroutine.
2288 # the final code doesn't have to worry about that.
2289 if ($property{$addr}) {
2290 push @{$each_line_handler{$addr}}, \&_insert_property_into_line;
2291 }
2292
2293 if ($non_skip{$addr} && ! $debug_skip && $verbosity) {
2294 print "Warning: " . __PACKAGE__ . " constructor for $file{$addr} has useless 'non_skip' in it\n";
a3a8c5f0 2295 }
99870f4d 2296
09ca89ce
KW
2297 # If skipping, set to optional, and add to list of ignored files,
2298 # including its reason
2299 if ($skip{$addr}) {
2300 $optional{$addr} = 1;
1fec9f60 2301 $skipped_files{$file{$addr}} = $skip{$addr}
09ca89ce 2302 }
83b68635
KW
2303 elsif ($properties{$addr}) {
2304
2305 # Add a handler for each line in the input so that it creates a
2306 # separate input line for each property in those input lines, thus
2307 # making them suitable for process_generic_property_file().
2308
2309 push @{$each_line_handler{$addr}},
2310 sub {
2311 my $file = shift;
2312 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
2313
2314 my @fields = split /\s*;\s*/, $_, -1;
2315
2316 if (@fields - 1 > @{$properties{$addr}}) {
2317 $file->carp_bad_line('Extra fields');
2318 $_ = "";
2319 return;
2320 }
2321 my $range = shift @fields; # 0th element is always the
2322 # range
2323
2324 # The next fields in the input line correspond
2325 # respectively to the stored properties.
2326 for my $i (0 .. @{$properties{$addr}} - 1) {
2327 my $property_name = $properties{$addr}[$i];
2328 next if $property_name eq '<ignored>';
2329 $file->insert_adjusted_lines(
2330 "$range; $property_name; $fields[$i]");
2331 }
2332 $_ = "";
2333
2334 return;
2335 };
2336 }
37e2e78e 2337
74cd47d0
KW
2338 { # On non-ascii platforms, we use a special handler
2339 no strict;
2340 no warnings 'once';
2341 *next_line = (main::NON_ASCII_PLATFORM)
2342 ? *_next_line_with_remapped_range
2343 : *_next_line;
2344 }
2345
99870f4d 2346 return $self;
d73e5302
JH
2347 }
2348
cf25bb62 2349
99870f4d
KW
2350 use overload
2351 fallback => 0,
2352 qw("") => "_operator_stringify",
2353 "." => \&main::_operator_dot,
1285127e 2354 ".=" => \&main::_operator_dot_equal,
99870f4d 2355 ;
cf25bb62 2356
99870f4d
KW
2357 sub _operator_stringify {
2358 my $self = shift;
cf25bb62 2359
99870f4d 2360 return __PACKAGE__ . " object for " . $self->file;
d73e5302 2361 }
d73e5302 2362
99870f4d
KW
2363 # flag to make sure extracted files are processed early
2364 my $seen_non_extracted_non_age = 0;
d73e5302 2365
99870f4d
KW
2366 sub run {
2367 # Process the input object $self. This opens and closes the file and
2368 # calls all the handlers for it. Currently, this can only be called
2369 # once per file, as it destroy's the EOF handler
d73e5302 2370
99870f4d
KW
2371 my $self = shift;
2372 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
b6922eda 2373
ffe43484 2374 my $addr = do { no overloading; pack 'J', $self; };
b6922eda 2375
99870f4d 2376 my $file = $file{$addr};
d73e5302 2377
99870f4d
KW
2378 # Don't process if not expecting this file (because released later
2379 # than this Unicode version), and isn't there. This means if someone
2380 # copies it into an earlier version's directory, we will go ahead and
2381 # process it.
2382 return if $first_released{$addr} gt $v_version && ! -e $file;
2383
2384 # If in debugging mode and this file doesn't have the non-skip
2385 # flag set, and isn't one of the critical files, skip it.
2386 if ($debug_skip
2387 && $first_released{$addr} ne v0
2388 && ! $non_skip{$addr})
2389 {
2390 print "Skipping $file in debugging\n" if $verbosity;
2391 return;
2392 }
2393
2394 # File could be optional
37e2e78e 2395 if ($optional{$addr}) {
99870f4d
KW
2396 return unless -e $file;
2397 my $result = eval $optional{$addr};
2398 if (! defined $result) {
2399 Carp::my_carp_bug("Got '$@' when tried to eval $optional{$addr}. $file Skipped.");
2400 return;
2401 }
2402 if (! $result) {
2403 if ($verbosity) {
2404 print STDERR "Skipping processing input file '$file' because '$optional{$addr}' is not true\n";
2405 }
2406 return;
2407 }
2408 }
2409
2410 if (! defined $file || ! -e $file) {
2411
2412 # If the file doesn't exist, see if have internal data for it
2413 # (based on first_released being 0).
2414 if ($first_released{$addr} eq v0) {
2415 $handle{$addr} = 'pretend_is_open';
2416 }
2417 else {
2418 if (! $optional{$addr} # File could be optional
2419 && $v_version ge $first_released{$addr})
2420 {
2421 print STDERR "Skipping processing input file '$file' because not found\n" if $v_version ge $first_released{$addr};
2422 }
2423 return;
2424 }
2425 }
2426 else {
2427
37e2e78e
KW
2428 # Here, the file exists. Some platforms may change the case of
2429 # its name
99870f4d 2430 if ($seen_non_extracted_non_age) {
517956bf 2431 if ($file =~ /$EXTRACTED/i) {
1675ea0d 2432 Carp::my_carp_bug(main::join_lines(<<END
99f78760 2433$file should be processed just after the 'Prop...Alias' files, and before
99870f4d
KW
2434anything not in the $EXTRACTED_DIR directory. Proceeding, but the results may
2435have subtle problems
2436END
2437 ));
2438 }
2439 }
2440 elsif ($EXTRACTED_DIR
2441 && $first_released{$addr} ne v0
517956bf
CB
2442 && $file !~ /$EXTRACTED/i
2443 && lc($file) ne 'dage.txt')
99870f4d
KW
2444 {
2445 # We don't set this (by the 'if' above) if we have no
2446 # extracted directory, so if running on an early version,
2447 # this test won't work. Not worth worrying about.
2448 $seen_non_extracted_non_age = 1;
2449 }
2450
2451 # And mark the file as having being processed, and warn if it
2452 # isn't a file we are expecting. As we process the files,
2453 # they are deleted from the hash, so any that remain at the
2454 # end of the program are files that we didn't process.
517956bf 2455 my $fkey = File::Spec->rel2abs($file);
faf3cf6b
KW
2456 my $expecting = delete $potential_files{lc($fkey)};
2457
678f13d5
KW
2458 Carp::my_carp("Was not expecting '$file'.") if
2459 ! $expecting
99870f4d
KW
2460 && ! defined $handle{$addr};
2461
37e2e78e
KW
2462 # Having deleted from expected files, we can quit if not to do
2463 # anything. Don't print progress unless really want verbosity
2464 if ($skip{$addr}) {
2465 print "Skipping $file.\n" if $verbosity >= $VERBOSE;
2466 return;
2467 }
2468
99870f4d
KW
2469 # Open the file, converting the slashes used in this program
2470 # into the proper form for the OS
2471 my $file_handle;
2472 if (not open $file_handle, "<", $file) {
2473 Carp::my_carp("Can't open $file. Skipping: $!");
2474 return 0;
2475 }
2476 $handle{$addr} = $file_handle; # Cache the open file handle
9e65c3f4
KW
2477
2478 if ($v_version ge v3.2.0 && lc($file) ne 'unicodedata.txt') {
2479 $_ = <$file_handle>;
2480 if ($_ !~ / - $string_version \. /x) {
2481 chomp;
2482 $_ =~ s/^#\s*//;
2483 die Carp::my_carp("File '$file' is version '$_'. It should be version $string_version");
2484 }
2485 }
99870f4d
KW
2486 }
2487
2488 if ($verbosity >= $PROGRESS) {
2489 if ($progress_message{$addr}) {
2490 print "$progress_message{$addr}\n";
2491 }
2492 else {
2493 # If using a virtual file, say so.
2494 print "Processing ", (-e $file)
2495 ? $file
2496 : "substitute $file",
2497 "\n";
2498 }
2499 }
2500
2501
2502 # Call any special handler for before the file.
2503 &{$pre_handler{$addr}}($self) if $pre_handler{$addr};
2504
2505 # Then the main handler
2506 &{$handler{$addr}}($self);
2507
2508 # Then any special post-file handler.
2509 &{$post_handler{$addr}}($self) if $post_handler{$addr};
2510
2511 # If any errors have been accumulated, output the counts (as the first
2512 # error message in each class was output when it was encountered).
2513 if ($errors{$addr}) {
2514 my $total = 0;
2515 my $types = 0;
2516 foreach my $error (keys %{$errors{$addr}}) {
2517 $total += $errors{$addr}->{$error};
2518 delete $errors{$addr}->{$error};
2519 $types++;
2520 }
2521 if ($total > 1) {
2522 my $message
2523 = "A total of $total lines had errors in $file. ";
2524
2525 $message .= ($types == 1)
2526 ? '(Only the first one was displayed.)'
2527 : '(Only the first of each type was displayed.)';
2528 Carp::my_carp($message);
2529 }
2530 }
2531
2532 if (@{$missings{$addr}}) {
2533 Carp::my_carp_bug("Handler for $file didn't look at all the \@missing lines. Generated tables likely are wrong");
2534 }
2535
2536 # If a real file handle, close it.
2537 close $handle{$addr} or Carp::my_carp("Can't close $file: $!") if
2538 ref $handle{$addr};
2539 $handle{$addr} = ""; # Uses empty to indicate that has already seen
2540 # the file, as opposed to undef
2541 return;
2542 }
2543
74cd47d0 2544 sub _next_line {
99870f4d
KW
2545 # Sets $_ to be the next logical input line, if any. Returns non-zero
2546 # if such a line exists. 'logical' means that any lines that have
2547 # been added via insert_lines() will be returned in $_ before the file
2548 # is read again.
2549
2550 my $self = shift;
2551 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
2552
ffe43484 2553 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
2554
2555 # Here the file is open (or if the handle is not a ref, is an open
2556 # 'virtual' file). Get the next line; any inserted lines get priority
2557 # over the file itself.
2558 my $adjusted;
2559
2560 LINE:
2561 while (1) { # Loop until find non-comment, non-empty line
2562 #local $to_trace = 1 if main::DEBUG;
2563 my $inserted_ref = shift @{$added_lines{$addr}};
2564 if (defined $inserted_ref) {
2565 ($adjusted, $_) = @{$inserted_ref};
2566 trace $adjusted, $_ if main::DEBUG && $to_trace;
2567 return 1 if $adjusted;
2568 }
2569 else {
2570 last if ! ref $handle{$addr}; # Don't read unless is real file
2571 last if ! defined ($_ = readline $handle{$addr});
2572 }
2573 chomp;
2574 trace $_ if main::DEBUG && $to_trace;
2575
2576 # See if this line is the comment line that defines what property
2577 # value that code points that are not listed in the file should
2578 # have. The format or existence of these lines is not guaranteed
2579 # by Unicode since they are comments, but the documentation says
2580 # that this was added for machine-readability, so probably won't
2581 # change. This works starting in Unicode Version 5.0. They look
2582 # like:
2583 #
2584 # @missing: 0000..10FFFF; Not_Reordered
2585 # @missing: 0000..10FFFF; Decomposition_Mapping; <code point>
2586 # @missing: 0000..10FFFF; ; NaN
2587 #
2588 # Save the line for a later get_missings() call.
2589 if (/$missing_defaults_prefix/) {
2590 if ($has_missings_defaults{$addr} == $NO_DEFAULTS) {
2591 $self->carp_bad_line("Unexpected \@missing line. Assuming no missing entries");
2592 }
2593 elsif ($has_missings_defaults{$addr} == $NOT_IGNORED) {
2594 my @defaults = split /\s* ; \s*/x, $_;
2595
2596 # The first field is the @missing, which ends in a
2597 # semi-colon, so can safely shift.
2598 shift @defaults;
2599
2600 # Some of these lines may have empty field placeholders
2601 # which get in the way. An example is:
2602 # @missing: 0000..10FFFF; ; NaN
2603 # Remove them. Process starting from the top so the
2604 # splice doesn't affect things still to be looked at.
2605 for (my $i = @defaults - 1; $i >= 0; $i--) {
2606 next if $defaults[$i] ne "";
2607 splice @defaults, $i, 1;
2608 }
2609
2610 # What's left should be just the property (maybe) and the
2611 # default. Having only one element means it doesn't have
2612 # the property.
2613 my $default;
2614 my $property;
2615 if (@defaults >= 1) {
2616 if (@defaults == 1) {
2617 $default = $defaults[0];
2618 }
2619 else {
2620 $property = $defaults[0];
2621 $default = $defaults[1];
2622 }
2623 }
2624
2625 if (@defaults < 1
2626 || @defaults > 2
2627 || ($default =~ /^</
2628 && $default !~ /^<code *point>$/i
09f8d0ac
KW
2629 && $default !~ /^<none>$/i
2630 && $default !~ /^<script>$/i))
99870f4d
KW
2631 {
2632 $self->carp_bad_line("Unrecognized \@missing line: $_. Assuming no missing entries");
2633 }
2634 else {
2635
2636 # If the property is missing from the line, it should
2637 # be the one for the whole file
2638 $property = $property{$addr} if ! defined $property;
2639
2640 # Change <none> to the null string, which is what it
2641 # really means. If the default is the code point
2642 # itself, set it to <code point>, which is what
2643 # Unicode uses (but sometimes they've forgotten the
2644 # space)
2645 if ($default =~ /^<none>$/i) {
2646 $default = "";
2647 }
2648 elsif ($default =~ /^<code *point>$/i) {
2649 $default = $CODE_POINT;
2650 }
09f8d0ac
KW
2651 elsif ($default =~ /^<script>$/i) {
2652
2653 # Special case this one. Currently is from
2654 # ScriptExtensions.txt, and means for all unlisted
2655 # code points, use their Script property values.
2656 # For the code points not listed in that file, the
2657 # default value is 'Unknown'.
2658 $default = "Unknown";
2659 }
99870f4d
KW
2660
2661 # Store them as a sub-arrays with both components.
2662 push @{$missings{$addr}}, [ $default, $property ];
2663 }
2664 }
2665
2666 # There is nothing for the caller to process on this comment
2667 # line.
2668 next;
2669 }
2670
2671 # Remove comments and trailing space, and skip this line if the
2672 # result is empty
2673 s/#.*//;
2674 s/\s+$//;
2675 next if /^$/;
2676
2677 # Call any handlers for this line, and skip further processing of
2678 # the line if the handler sets the line to null.
2679 foreach my $sub_ref (@{$each_line_handler{$addr}}) {
2680 &{$sub_ref}($self);
2681 next LINE if /^$/;
2682 }
2683
2684 # Here the line is ok. return success.
2685 return 1;
2686 } # End of looping through lines.
2687
2688 # If there is an EOF handler, call it (only once) and if it generates
2689 # more lines to process go back in the loop to handle them.
2690 if ($eof_handler{$addr}) {
2691 &{$eof_handler{$addr}}($self);
2692 $eof_handler{$addr} = ""; # Currently only get one shot at it.
2693 goto LINE if $added_lines{$addr};
2694 }
2695
2696 # Return failure -- no more lines.
2697 return 0;
2698
2699 }
2700
74cd47d0
KW
2701 sub _next_line_with_remapped_range {
2702 my $self = shift;
2703 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
2704
2705 # like _next_line(), but for use on non-ASCII platforms. It sets $_
2706 # to be the next logical input line, if any. Returns non-zero if such
2707 # a line exists. 'logical' means that any lines that have been added
2708 # via insert_lines() will be returned in $_ before the file is read
2709 # again.
2710 #
2711 # The difference from _next_line() is that this remaps the Unicode
2712 # code points in the input to those of the native platform. Each
2713 # input line contains a single code point, or a single contiguous
2714 # range of them This routine splits each range into its individual
2715 # code points and caches them. It returns the cached values,
2716 # translated into their native equivalents, one at a time, for each
2717 # call, before reading the next line. Since native values can only be
2718 # a single byte wide, no translation is needed for code points above
2719 # 0xFF, and ranges that are entirely above that number are not split.
2720 # If an input line contains the range 254-1000, it would be split into
2721 # three elements: 254, 255, and 256-1000. (The downstream table
2722 # insertion code will sort and coalesce the individual code points
2723 # into appropriate ranges.)
2724
2725 my $addr = do { no overloading; pack 'J', $self; };
2726
2727 while (1) {
2728
2729 # Look in cache before reading the next line. Return any cached
2730 # value, translated
2731 my $inserted = shift @{$remapped_lines{$addr}};
2732 if (defined $inserted) {
2733 trace $inserted if main::DEBUG && $to_trace;
2734 $_ = $inserted =~ s/^ ( \d+ ) /sprintf("%04X", utf8::unicode_to_native($1))/xer;
2735 trace $_ if main::DEBUG && $to_trace;
2736 return 1;
2737 }
2738
2739 # Get the next line.
2740 return 0 unless _next_line($self);
2741
2742 # If there is a special handler for it, return the line,
2743 # untranslated. This should happen only for files that are
2744 # special, not being code-point related, such as property names.
2745 return 1 if $handler{$addr}
2746 != \&main::process_generic_property_file;
2747
2748 my ($range, $property_name, $map, @remainder)
2749 = split /\s*;\s*/, $_, -1; # -1 => retain trailing null fields
2750
2751 if (@remainder
2752 || ! defined $property_name
2753 || $range !~ /^ ($code_point_re) (?:\.\. ($code_point_re) )? $/x)
2754 {
2755 Carp::my_carp_bug("Unrecognized input line '$_'. Ignored");
2756 }
2757
2758 my $low = hex $1;
2759 my $high = (defined $2) ? hex $2 : $low;
2760
2761 # If the input maps the range to another code point, remap the
2762 # target if it is between 0 and 255.
2763 my $tail;
2764 if (defined $map) {
2765 $map =~ s/\b 00 ( [0-9A-F]{2} ) \b/sprintf("%04X", utf8::unicode_to_native(hex $1))/gxe;
2766 $tail = "$property_name; $map";
2767 $_ = "$range; $tail";
2768 }
2769 else {
2770 $tail = $property_name;
2771 }
2772
2773 # If entire range is above 255, just return it, unchanged (except
2774 # any mapped-to code point, already changed above)
2775 return 1 if $low > 255;
2776
2777 # Cache an entry for every code point < 255. For those in the
2778 # range above 255, return a dummy entry for just that portion of
2779 # the range. Note that this will be out-of-order, but that is not
2780 # a problem.
2781 foreach my $code_point ($low .. $high) {
2782 if ($code_point > 255) {
2783 $_ = sprintf "%04X..%04X; $tail", $code_point, $high;
2784 return 1;
2785 }
2786 push @{$remapped_lines{$addr}}, "$code_point; $tail";
2787 }
2788 } # End of looping through lines.
2789
2790 # NOTREACHED
2791 }
2792
99870f4d
KW
2793# Not currently used, not fully tested.
2794# sub peek {
2795# # Non-destructive look-ahead one non-adjusted, non-comment, non-blank
2796# # record. Not callable from an each_line_handler(), nor does it call
2797# # an each_line_handler() on the line.
2798#
2799# my $self = shift;
ffe43484 2800# my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
2801#
2802# foreach my $inserted_ref (@{$added_lines{$addr}}) {
2803# my ($adjusted, $line) = @{$inserted_ref};
2804# next if $adjusted;
2805#
2806# # Remove comments and trailing space, and return a non-empty
2807# # resulting line
2808# $line =~ s/#.*//;
2809# $line =~ s/\s+$//;
2810# return $line if $line ne "";
2811# }
2812#
2813# return if ! ref $handle{$addr}; # Don't read unless is real file
2814# while (1) { # Loop until find non-comment, non-empty line
2815# local $to_trace = 1 if main::DEBUG;
2816# trace $_ if main::DEBUG && $to_trace;
2817# return if ! defined (my $line = readline $handle{$addr});
2818# chomp $line;
2819# push @{$added_lines{$addr}}, [ 0, $line ];
2820#
2821# $line =~ s/#.*//;
2822# $line =~ s/\s+$//;
2823# return $line if $line ne "";
2824# }
2825#
2826# return;
2827# }
2828
2829
2830 sub insert_lines {
2831 # Lines can be inserted so that it looks like they were in the input
2832 # file at the place it was when this routine is called. See also
2833 # insert_adjusted_lines(). Lines inserted via this routine go through
2834 # any each_line_handler()
2835
2836 my $self = shift;
2837
2838 # Each inserted line is an array, with the first element being 0 to
2839 # indicate that this line hasn't been adjusted, and needs to be
2840 # processed.
f998e60c 2841 no overloading;
051df77b 2842 push @{$added_lines{pack 'J', $self}}, map { [ 0, $_ ] } @_;
99870f4d
KW
2843 return;
2844 }
2845
2846 sub insert_adjusted_lines {
2847 # Lines can be inserted so that it looks like they were in the input
2848 # file at the place it was when this routine is called. See also
2849 # insert_lines(). Lines inserted via this routine are already fully
2850 # adjusted, ready to be processed; each_line_handler()s handlers will
2851 # not be called. This means this is not a completely general
2852 # facility, as only the last each_line_handler on the stack should
2853 # call this. It could be made more general, by passing to each of the
2854 # line_handlers their position on the stack, which they would pass on
2855 # to this routine, and that would replace the boolean first element in
2856 # the anonymous array pushed here, so that the next_line routine could
2857 # use that to call only those handlers whose index is after it on the
2858 # stack. But this is overkill for what is needed now.
2859
2860 my $self = shift;
2861 trace $_[0] if main::DEBUG && $to_trace;
2862
2863 # Each inserted line is an array, with the first element being 1 to
2864 # indicate that this line has been adjusted
f998e60c 2865 no overloading;
051df77b 2866 push @{$added_lines{pack 'J', $self}}, map { [ 1, $_ ] } @_;
99870f4d
KW
2867 return;
2868 }
2869
2870 sub get_missings {
2871 # Returns the stored up @missings lines' values, and clears the list.
2872 # The values are in an array, consisting of the default in the first
2873 # element, and the property in the 2nd. However, since these lines
2874 # can be stacked up, the return is an array of all these arrays.
2875
2876 my $self = shift;
2877 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
2878
ffe43484 2879 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
2880
2881 # If not accepting a list return, just return the first one.
2882 return shift @{$missings{$addr}} unless wantarray;
2883
2884 my @return = @{$missings{$addr}};
2885 undef @{$missings{$addr}};
2886 return @return;
2887 }
2888
2889 sub _insert_property_into_line {
2890 # Add a property field to $_, if this file requires it.
2891
f998e60c 2892 my $self = shift;
ffe43484 2893 my $addr = do { no overloading; pack 'J', $self; };
f998e60c 2894 my $property = $property{$addr};
99870f4d
KW
2895 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
2896
2897 $_ =~ s/(;|$)/; $property$1/;
2898 return;
2899 }
2900
2901 sub carp_bad_line {
2902 # Output consistent error messages, using either a generic one, or the
2903 # one given by the optional parameter. To avoid gazillions of the
2904 # same message in case the syntax of a file is way off, this routine
2905 # only outputs the first instance of each message, incrementing a
2906 # count so the totals can be output at the end of the file.
2907
2908 my $self = shift;
2909 my $message = shift;
2910 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
2911
ffe43484 2912 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
2913
2914 $message = 'Unexpected line' unless $message;
2915
2916 # No trailing punctuation so as to fit with our addenda.
2917 $message =~ s/[.:;,]$//;
2918
2919 # If haven't seen this exact message before, output it now. Otherwise
2920 # increment the count of how many times it has occurred
2921 unless ($errors{$addr}->{$message}) {
2922 Carp::my_carp("$message in '$_' in "
f998e60c 2923 . $file{$addr}
99870f4d
KW
2924 . " at line $.. Skipping this line;");
2925 $errors{$addr}->{$message} = 1;
2926 }
2927 else {
2928 $errors{$addr}->{$message}++;
2929 }
2930
2931 # Clear the line to prevent any further (meaningful) processing of it.
2932 $_ = "";
2933
2934 return;
2935 }
2936} # End closure
2937
2938package Multi_Default;
2939
2940# Certain properties in early versions of Unicode had more than one possible
2941# default for code points missing from the files. In these cases, one
2942# default applies to everything left over after all the others are applied,
2943# and for each of the others, there is a description of which class of code
2944# points applies to it. This object helps implement this by storing the
2945# defaults, and for all but that final default, an eval string that generates
2946# the class that it applies to.
2947
2948
2949{ # Closure
2950
2951 main::setup_package();
2952
2953 my %class_defaults;
2954 # The defaults structure for the classes
2955 main::set_access('class_defaults', \%class_defaults);
2956
2957 my %other_default;
2958 # The default that applies to everything left over.
2959 main::set_access('other_default', \%other_default, 'r');
2960
2961
2962 sub new {
2963 # The constructor is called with default => eval pairs, terminated by
2964 # the left-over default. e.g.
2965 # Multi_Default->new(
2966 # 'T' => '$gc->table("Mn") + $gc->table("Cf") - 0x200C
2967 # - 0x200D',
2968 # 'R' => 'some other expression that evaluates to code points',
2969 # .
2970 # .
2971 # .
2972 # 'U'));
2973
2974 my $class = shift;
2975
2976 my $self = bless \do{my $anonymous_scalar}, $class;
ffe43484 2977 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
2978
2979 while (@_ > 1) {
2980 my $default = shift;
2981 my $eval = shift;
2982 $class_defaults{$addr}->{$default} = $eval;
2983 }
2984
2985 $other_default{$addr} = shift;
2986
2987 return $self;
2988 }
2989
2990 sub get_next_defaults {
2991 # Iterates and returns the next class of defaults.
2992 my $self = shift;
2993 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
2994
ffe43484 2995 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
2996
2997 return each %{$class_defaults{$addr}};
2998 }
2999}
3000
3001package Alias;
3002
3003# An alias is one of the names that a table goes by. This class defines them
3004# including some attributes. Everything is currently setup in the
3005# constructor.
3006
3007
3008{ # Closure
3009
3010 main::setup_package();
3011
3012 my %name;
3013 main::set_access('name', \%name, 'r');
3014
3015 my %loose_match;
c12f2655 3016 # Should this name match loosely or not.
99870f4d
KW
3017 main::set_access('loose_match', \%loose_match, 'r');
3018
33e96e72
KW
3019 my %make_re_pod_entry;
3020 # Some aliases should not get their own entries in the re section of the
3021 # pod, because they are covered by a wild-card, and some we want to
3022 # discourage use of. Binary
f82fe4ba 3023 main::set_access('make_re_pod_entry', \%make_re_pod_entry, 'r', 's');
99870f4d 3024
fd1e3e84
KW
3025 my %ucd;
3026 # Is this documented to be accessible via Unicode::UCD
3027 main::set_access('ucd', \%ucd, 'r', 's');
3028
99870f4d
KW
3029 my %status;
3030 # Aliases have a status, like deprecated, or even suppressed (which means
3031 # they don't appear in documentation). Enum
3032 main::set_access('status', \%status, 'r');
3033
0eac1e20 3034 my %ok_as_filename;
99870f4d
KW
3035 # Similarly, some aliases should not be considered as usable ones for
3036 # external use, such as file names, or we don't want documentation to
3037 # recommend them. Boolean
0eac1e20 3038 main::set_access('ok_as_filename', \%ok_as_filename, 'r');
99870f4d
KW
3039
3040 sub new {
3041 my $class = shift;
3042
3043 my $self = bless \do { my $anonymous_scalar }, $class;
ffe43484 3044 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3045
3046 $name{$addr} = shift;
3047 $loose_match{$addr} = shift;
33e96e72 3048 $make_re_pod_entry{$addr} = shift;
0eac1e20 3049 $ok_as_filename{$addr} = shift;
99870f4d 3050 $status{$addr} = shift;
fd1e3e84 3051 $ucd{$addr} = shift;
99870f4d
KW
3052
3053 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3054
3055 # Null names are never ok externally
0eac1e20 3056 $ok_as_filename{$addr} = 0 if $name{$addr} eq "";
99870f4d
KW
3057
3058 return $self;
3059 }
3060}
3061
3062package Range;
3063
3064# A range is the basic unit for storing code points, and is described in the
3065# comments at the beginning of the program. Each range has a starting code
3066# point; an ending code point (not less than the starting one); a value
3067# that applies to every code point in between the two end-points, inclusive;
3068# and an enum type that applies to the value. The type is for the user's
3069# convenience, and has no meaning here, except that a non-zero type is
3070# considered to not obey the normal Unicode rules for having standard forms.
3071#
3072# The same structure is used for both map and match tables, even though in the
3073# latter, the value (and hence type) is irrelevant and could be used as a
3074# comment. In map tables, the value is what all the code points in the range
3075# map to. Type 0 values have the standardized version of the value stored as
3076# well, so as to not have to recalculate it a lot.
3077
3078sub trace { return main::trace(@_); }
3079
3080{ # Closure
3081
3082 main::setup_package();
3083
3084 my %start;
3085 main::set_access('start', \%start, 'r', 's');
3086
3087 my %end;
3088 main::set_access('end', \%end, 'r', 's');
3089
3090 my %value;
3091 main::set_access('value', \%value, 'r');
3092
3093 my %type;
3094 main::set_access('type', \%type, 'r');
3095
3096 my %standard_form;
3097 # The value in internal standard form. Defined only if the type is 0.
3098 main::set_access('standard_form', \%standard_form);
3099
3100 # Note that if these fields change, the dump() method should as well
3101
3102 sub new {
3103 return Carp::carp_too_few_args(\@_, 3) if main::DEBUG && @_ < 3;
3104 my $class = shift;
3105
3106 my $self = bless \do { my $anonymous_scalar }, $class;
ffe43484 3107 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3108
3109 $start{$addr} = shift;
3110 $end{$addr} = shift;
3111
3112 my %args = @_;
3113
3114 my $value = delete $args{'Value'}; # Can be 0
3115 $value = "" unless defined $value;
3116 $value{$addr} = $value;
3117
3118 $type{$addr} = delete $args{'Type'} || 0;
3119
3120 Carp::carp_extra_args(\%args) if main::DEBUG && %args;
3121
99870f4d
KW
3122 return $self;
3123 }
3124
3125 use overload
3126 fallback => 0,
3127 qw("") => "_operator_stringify",
3128 "." => \&main::_operator_dot,
1285127e 3129 ".=" => \&main::_operator_dot_equal,
99870f4d
KW
3130 ;
3131
3132 sub _operator_stringify {
3133 my $self = shift;
ffe43484 3134 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3135
3136 # Output it like '0041..0065 (value)'
3137 my $return = sprintf("%04X", $start{$addr})
3138 . '..'
3139 . sprintf("%04X", $end{$addr});
3140 my $value = $value{$addr};
3141 my $type = $type{$addr};
3142 $return .= ' (';
3143 $return .= "$value";
3144 $return .= ", Type=$type" if $type != 0;
3145 $return .= ')';
3146
3147 return $return;
3148 }
3149
3150 sub standard_form {
c292d35a
NC
3151 # Calculate the standard form only if needed, and cache the result.
3152 # The standard form is the value itself if the type is special.
3153 # This represents a considerable CPU and memory saving - at the time
3154 # of writing there are 368676 non-special objects, but the standard
3155 # form is only requested for 22047 of them - ie about 6%.
99870f4d
KW
3156
3157 my $self = shift;
3158 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3159
ffe43484 3160 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3161
3162 return $standard_form{$addr} if defined $standard_form{$addr};
c292d35a
NC
3163
3164 my $value = $value{$addr};
3165 return $value if $type{$addr};
3166 return $standard_form{$addr} = main::standardize($value);
99870f4d
KW
3167 }
3168
3169 sub dump {
3170 # Human, not machine readable. For machine readable, comment out this
3171 # entire routine and let the standard one take effect.
3172 my $self = shift;
3173 my $indent = shift;
3174 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3175
ffe43484 3176 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3177
3178 my $return = $indent
3179 . sprintf("%04X", $start{$addr})
3180 . '..'
3181 . sprintf("%04X", $end{$addr})
3182 . " '$value{$addr}';";
3183 if (! defined $standard_form{$addr}) {
3184 $return .= "(type=$type{$addr})";
3185 }
3186 elsif ($standard_form{$addr} ne $value{$addr}) {
3187 $return .= "(standard '$standard_form{$addr}')";
3188 }
3189 return $return;
3190 }
3191} # End closure
3192
3193package _Range_List_Base;
3194
3195# Base class for range lists. A range list is simply an ordered list of
3196# ranges, so that the ranges with the lowest starting numbers are first in it.
3197#
3198# When a new range is added that is adjacent to an existing range that has the
3199# same value and type, it merges with it to form a larger range.
3200#
3201# Ranges generally do not overlap, except that there can be multiple entries
3202# of single code point ranges. This is because of NameAliases.txt.
3203#
3204# In this program, there is a standard value such that if two different
3205# values, have the same standard value, they are considered equivalent. This
3206# value was chosen so that it gives correct results on Unicode data
3207
3208# There are a number of methods to manipulate range lists, and some operators
3209# are overloaded to handle them.
3210
99870f4d
KW
3211sub trace { return main::trace(@_); }
3212
3213{ # Closure
3214
3215 our $addr;
3216
5b348b71
KW
3217 # Max is initialized to a negative value that isn't adjacent to 0, for
3218 # simpler tests
3219 my $max_init = -2;
3220
99870f4d
KW
3221 main::setup_package();
3222
3223 my %ranges;
3224 # The list of ranges
3225 main::set_access('ranges', \%ranges, 'readable_array');
3226
3227 my %max;
3228 # The highest code point in the list. This was originally a method, but
3229 # actual measurements said it was used a lot.
3230 main::set_access('max', \%max, 'r');
3231
3232 my %each_range_iterator;
3233 # Iterator position for each_range()
3234 main::set_access('each_range_iterator', \%each_range_iterator);
3235
3236 my %owner_name_of;
3237 # Name of parent this is attached to, if any. Solely for better error
3238 # messages.
3239 main::set_access('owner_name_of', \%owner_name_of, 'p_r');
3240
3241 my %_search_ranges_cache;
3242 # A cache of the previous result from _search_ranges(), for better
3243 # performance
3244 main::set_access('_search_ranges_cache', \%_search_ranges_cache);
3245
3246 sub new {
3247 my $class = shift;
3248 my %args = @_;
3249
3250 # Optional initialization data for the range list.
3251 my $initialize = delete $args{'Initialize'};
3252
3253 my $self;
3254
3255 # Use _union() to initialize. _union() returns an object of this
3256 # class, which means that it will call this constructor recursively.
3257 # But it won't have this $initialize parameter so that it won't
3258 # infinitely loop on this.
3259 return _union($class, $initialize, %args) if defined $initialize;
3260
3261 $self = bless \do { my $anonymous_scalar }, $class;
ffe43484 3262 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3263
3264 # Optional parent object, only for debug info.
3265 $owner_name_of{$addr} = delete $args{'Owner'};
3266 $owner_name_of{$addr} = "" if ! defined $owner_name_of{$addr};
3267
3268 # Stringify, in case it is an object.
3269 $owner_name_of{$addr} = "$owner_name_of{$addr}";
3270
3271 # This is used only for error messages, and so a colon is added
3272 $owner_name_of{$addr} .= ": " if $owner_name_of{$addr} ne "";
3273
3274 Carp::carp_extra_args(\%args) if main::DEBUG && %args;
3275
5b348b71 3276 $max{$addr} = $max_init;
99870f4d
KW
3277
3278 $_search_ranges_cache{$addr} = 0;
3279 $ranges{$addr} = [];
3280
3281 return $self;
3282 }
3283
3284 use overload
3285 fallback => 0,
3286 qw("") => "_operator_stringify",
3287 "." => \&main::_operator_dot,
1285127e 3288 ".=" => \&main::_operator_dot_equal,
99870f4d
KW
3289 ;
3290
3291 sub _operator_stringify {
3292 my $self = shift;
ffe43484 3293 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3294
3295 return "Range_List attached to '$owner_name_of{$addr}'"
3296 if $owner_name_of{$addr};
3297 return "anonymous Range_List " . \$self;
3298 }
3299
3300 sub _union {
3301 # Returns the union of the input code points. It can be called as
3302 # either a constructor or a method. If called as a method, the result
3303 # will be a new() instance of the calling object, containing the union
3304 # of that object with the other parameter's code points; if called as
d59563d0 3305 # a constructor, the first parameter gives the class that the new object
99870f4d
KW
3306 # should be, and the second parameter gives the code points to go into
3307 # it.
3308 # In either case, there are two parameters looked at by this routine;
3309 # any additional parameters are passed to the new() constructor.
3310 #
3311 # The code points can come in the form of some object that contains
3312 # ranges, and has a conventionally named method to access them; or
3313 # they can be an array of individual code points (as integers); or
3314 # just a single code point.
3315 #
3316 # If they are ranges, this routine doesn't make any effort to preserve
3198cc57
KW
3317 # the range values and types of one input over the other. Therefore
3318 # this base class should not allow _union to be called from other than
99870f4d
KW
3319 # initialization code, so as to prevent two tables from being added
3320 # together where the range values matter. The general form of this
3321 # routine therefore belongs in a derived class, but it was moved here
3322 # to avoid duplication of code. The failure to overload this in this
3323 # class keeps it safe.
3198cc57
KW
3324 #
3325 # It does make the effort during initialization to accept tables with
3326 # multiple values for the same code point, and to preserve the order
3327 # of these. If there is only one input range or range set, it doesn't
3328 # sort (as it should already be sorted to the desired order), and will
3329 # accept multiple values per code point. Otherwise it will merge
3330 # multiple values into a single one.
99870f4d
KW
3331
3332 my $self;
3333 my @args; # Arguments to pass to the constructor
3334
3335 my $class = shift;
3336
3337 # If a method call, will start the union with the object itself, and
3338 # the class of the new object will be the same as self.
3339 if (ref $class) {
3340 $self = $class;
3341 $class = ref $self;
3342 push @args, $self;
3343 }
3344
3345 # Add the other required parameter.
3346 push @args, shift;
3347 # Rest of parameters are passed on to the constructor
3348
3349 # Accumulate all records from both lists.
3350 my @records;
3198cc57 3351 my $input_count = 0;
99870f4d
KW
3352 for my $arg (@args) {
3353 #local $to_trace = 0 if main::DEBUG;
3354 trace "argument = $arg" if main::DEBUG && $to_trace;
3355 if (! defined $arg) {
3356 my $message = "";
3357 if (defined $self) {
f998e60c 3358 no overloading;
051df77b 3359 $message .= $owner_name_of{pack 'J', $self};
99870f4d 3360 }
ada6088e 3361 Carp::my_carp_bug($message . "Undefined argument to _union. No union done.");
99870f4d
KW
3362 return;
3363 }
3198cc57 3364
99870f4d
KW
3365 $arg = [ $arg ] if ! ref $arg;
3366 my $type = ref $arg;
3367 if ($type eq 'ARRAY') {
3368 foreach my $element (@$arg) {
3369 push @records, Range->new($element, $element);
3198cc57 3370 $input_count++;
99870f4d
KW
3371 }
3372 }
3373 elsif ($arg->isa('Range')) {
3374 push @records, $arg;
3198cc57 3375 $input_count++;
99870f4d
KW
3376 }
3377 elsif ($arg->can('ranges')) {
3378 push @records, $arg->ranges;
3198cc57 3379 $input_count++;
99870f4d
KW
3380 }
3381 else {
3382 my $message = "";
3383 if (defined $self) {
f998e60c 3384 no overloading;
051df77b 3385 $message .= $owner_name_of{pack 'J', $self};
99870f4d
KW
3386 }
3387 Carp::my_carp_bug($message . "Cannot take the union of a $type. No union done.");
3388 return;
3389 }
3390 }
3391
3392 # Sort with the range containing the lowest ordinal first, but if
3393 # two ranges start at the same code point, sort with the bigger range
3394 # of the two first, because it takes fewer cycles.
3198cc57
KW
3395 if ($input_count > 1) {
3396 @records = sort { ($a->start <=> $b->start)
99870f4d
KW
3397 or
3398 # if b is shorter than a, b->end will be
3399 # less than a->end, and we want to select
3400 # a, so want to return -1
3401 ($b->end <=> $a->end)
3402 } @records;
3198cc57 3403 }
99870f4d
KW
3404
3405 my $new = $class->new(@_);
3406
3407 # Fold in records so long as they add new information.
3408 for my $set (@records) {
3409 my $start = $set->start;
3410 my $end = $set->end;
d59563d0 3411 my $value = $set->value;
3198cc57 3412 my $type = $set->type;
99870f4d 3413 if ($start > $new->max) {
3198cc57 3414 $new->_add_delete('+', $start, $end, $value, Type => $type);
99870f4d
KW
3415 }
3416 elsif ($end > $new->max) {
3198cc57
KW
3417 $new->_add_delete('+', $new->max +1, $end, $value,
3418 Type => $type);
3419 }
3420 elsif ($input_count == 1) {
3421 # Here, overlaps existing range, but is from a single input,
3422 # so preserve the multiple values from that input.
3423 $new->_add_delete('+', $start, $end, $value, Type => $type,
3424 Replace => $MULTIPLE_AFTER);
99870f4d
KW
3425 }
3426 }
3427
3428 return $new;
3429 }
3430
3431 sub range_count { # Return the number of ranges in the range list
3432 my $self = shift;
3433 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3434
f998e60c 3435 no overloading;
051df77b 3436 return scalar @{$ranges{pack 'J', $self}};
99870f4d
KW
3437 }
3438
3439 sub min {
3440 # Returns the minimum code point currently in the range list, or if
3441 # the range list is empty, 2 beyond the max possible. This is a
3442 # method because used so rarely, that not worth saving between calls,
3443 # and having to worry about changing it as ranges are added and
3444 # deleted.
3445
3446 my $self = shift;
3447 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3448
ffe43484 3449 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3450
3451 # If the range list is empty, return a large value that isn't adjacent
3452 # to any that could be in the range list, for simpler tests
6189eadc 3453 return $MAX_UNICODE_CODEPOINT + 2 unless scalar @{$ranges{$addr}};
99870f4d
KW
3454 return $ranges{$addr}->[0]->start;
3455 }
3456
3457 sub contains {
3458 # Boolean: Is argument in the range list? If so returns $i such that:
3459 # range[$i]->end < $codepoint <= range[$i+1]->end
3460 # which is one beyond what you want; this is so that the 0th range
3461 # doesn't return false
3462 my $self = shift;
3463 my $codepoint = shift;
3464 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3465
99870f4d
KW
3466 my $i = $self->_search_ranges($codepoint);
3467 return 0 unless defined $i;
3468
3469 # The search returns $i, such that
3470 # range[$i-1]->end < $codepoint <= range[$i]->end
3471 # So is in the table if and only iff it is at least the start position
3472 # of range $i.
f998e60c 3473 no overloading;
051df77b 3474 return 0 if $ranges{pack 'J', $self}->[$i]->start > $codepoint;
99870f4d
KW
3475 return $i + 1;
3476 }
3477
2f7a8815
KW
3478 sub containing_range {
3479 # Returns the range object that contains the code point, undef if none
3480
3481 my $self = shift;
3482 my $codepoint = shift;
3483 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3484
3485 my $i = $self->contains($codepoint);
3486 return unless $i;
3487
3488 # contains() returns 1 beyond where we should look
3489 no overloading;
3490 return $ranges{pack 'J', $self}->[$i-1];
3491 }
3492
99870f4d
KW
3493 sub value_of {
3494 # Returns the value associated with the code point, undef if none
3495
3496 my $self = shift;
3497 my $codepoint = shift;
3498 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3499
d69c231b
KW
3500 my $range = $self->containing_range($codepoint);
3501 return unless defined $range;
99870f4d 3502
d69c231b 3503 return $range->value;
99870f4d
KW
3504 }
3505
0a9dbafc
KW
3506 sub type_of {
3507 # Returns the type of the range containing the code point, undef if
3508 # the code point is not in the table
3509
3510 my $self = shift;
3511 my $codepoint = shift;
3512 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3513
3514 my $range = $self->containing_range($codepoint);
3515 return unless defined $range;
3516
3517 return $range->type;
3518 }
3519
99870f4d
KW
3520 sub _search_ranges {
3521 # Find the range in the list which contains a code point, or where it
3522 # should go if were to add it. That is, it returns $i, such that:
3523 # range[$i-1]->end < $codepoint <= range[$i]->end
3524 # Returns undef if no such $i is possible (e.g. at end of table), or
3525 # if there is an error.
3526
3527 my $self = shift;
3528 my $code_point = shift;
3529 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3530
ffe43484 3531 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3532
3533 return if $code_point > $max{$addr};
3534 my $r = $ranges{$addr}; # The current list of ranges
3535 my $range_list_size = scalar @$r;
3536 my $i;
3537
3538 use integer; # want integer division
3539
3540 # Use the cached result as the starting guess for this one, because,
3541 # an experiment on 5.1 showed that 90% of the time the cache was the
3542 # same as the result on the next call (and 7% it was one less).
3543 $i = $_search_ranges_cache{$addr};
3544 $i = 0 if $i >= $range_list_size; # Reset if no longer valid (prob.
3545 # from an intervening deletion
3546 #local $to_trace = 1 if main::DEBUG;
3547 trace "previous \$i is still valid: $i" if main::DEBUG && $to_trace && $code_point <= $r->[$i]->end && ($i == 0 || $r->[$i-1]->end < $code_point);
3548 return $i if $code_point <= $r->[$i]->end
3549 && ($i == 0 || $r->[$i-1]->end < $code_point);
3550
3551 # Here the cache doesn't yield the correct $i. Try adding 1.
3552 if ($i < $range_list_size - 1
3553 && $r->[$i]->end < $code_point &&
3554 $code_point <= $r->[$i+1]->end)
3555 {
3556 $i++;
3557 trace "next \$i is correct: $i" if main::DEBUG && $to_trace;
3558 $_search_ranges_cache{$addr} = $i;
3559 return $i;
3560 }
3561
3562 # Here, adding 1 also didn't work. We do a binary search to
3563 # find the correct position, starting with current $i
3564 my $lower = 0;
3565 my $upper = $range_list_size - 1;
3566 while (1) {
3567 trace "top of loop i=$i:", sprintf("%04X", $r->[$lower]->start), "[$lower] .. ", sprintf("%04X", $r->[$i]->start), "[$i] .. ", sprintf("%04X", $r->[$upper]->start), "[$upper]" if main::DEBUG && $to_trace;
3568
3569 if ($code_point <= $r->[$i]->end) {
3570
3571 # Here we have met the upper constraint. We can quit if we
3572 # also meet the lower one.
3573 last if $i == 0 || $r->[$i-1]->end < $code_point;
3574
3575 $upper = $i; # Still too high.
3576
3577 }
3578 else {
3579
3580 # Here, $r[$i]->end < $code_point, so look higher up.
3581 $lower = $i;
3582 }
3583
3584 # Split search domain in half to try again.
3585 my $temp = ($upper + $lower) / 2;
3586
3587 # No point in continuing unless $i changes for next time
3588 # in the loop.
3589 if ($temp == $i) {
3590
3591 # We can't reach the highest element because of the averaging.
3592 # So if one below the upper edge, force it there and try one
3593 # more time.
3594 if ($i == $range_list_size - 2) {
3595
3596 trace "Forcing to upper edge" if main::DEBUG && $to_trace;
3597 $i = $range_list_size - 1;
3598
3599 # Change $lower as well so if fails next time through,
3600 # taking the average will yield the same $i, and we will
3601 # quit with the error message just below.
3602 $lower = $i;
3603 next;
3604 }
3605 Carp::my_carp_bug("$owner_name_of{$addr}Can't find where the range ought to go. No action taken.");
3606 return;
3607 }
3608 $i = $temp;
3609 } # End of while loop
3610
3611 if (main::DEBUG && $to_trace) {
3612 trace 'i-1=[', $i-1, ']', $r->[$i-1] if $i;
3613 trace "i= [ $i ]", $r->[$i];
3614 trace 'i+1=[', $i+1, ']', $r->[$i+1] if $i < $range_list_size - 1;
3615 }
3616
3617 # Here we have found the offset. Cache it as a starting point for the
3618 # next call.
3619 $_search_ranges_cache{$addr} = $i;
3620 return $i;
3621 }
3622
3623 sub _add_delete {
3624 # Add, replace or delete ranges to or from a list. The $type
3625 # parameter gives which:
3626 # '+' => insert or replace a range, returning a list of any changed
3627 # ranges.
3628 # '-' => delete a range, returning a list of any deleted ranges.
3629 #
3630 # The next three parameters give respectively the start, end, and
3631 # value associated with the range. 'value' should be null unless the
3632 # operation is '+';
3633 #
3634 # The range list is kept sorted so that the range with the lowest
3635 # starting position is first in the list, and generally, adjacent
c1739a4a 3636 # ranges with the same values are merged into a single larger one (see
99870f4d
KW
3637 # exceptions below).
3638 #
c1739a4a 3639 # There are more parameters; all are key => value pairs:
99870f4d
KW
3640 # Type gives the type of the value. It is only valid for '+'.
3641 # All ranges have types; if this parameter is omitted, 0 is
3642 # assumed. Ranges with type 0 are assumed to obey the
3643 # Unicode rules for casing, etc; ranges with other types are
3644 # not. Otherwise, the type is arbitrary, for the caller's
3645 # convenience, and looked at only by this routine to keep
3646 # adjacent ranges of different types from being merged into
3647 # a single larger range, and when Replace =>
3648 # $IF_NOT_EQUIVALENT is specified (see just below).
3649 # Replace determines what to do if the range list already contains
3650 # ranges which coincide with all or portions of the input
3651 # range. It is only valid for '+':
3652 # => $NO means that the new value is not to replace
3653 # any existing ones, but any empty gaps of the
3654 # range list coinciding with the input range
3655 # will be filled in with the new value.
3656 # => $UNCONDITIONALLY means to replace the existing values with
3657 # this one unconditionally. However, if the
3658 # new and old values are identical, the
3659 # replacement is skipped to save cycles
3660 # => $IF_NOT_EQUIVALENT means to replace the existing values
d59563d0 3661 # (the default) with this one if they are not equivalent.
99870f4d 3662 # Ranges are equivalent if their types are the
c1739a4a 3663 # same, and they are the same string; or if
99870f4d
KW
3664 # both are type 0 ranges, if their Unicode
3665 # standard forms are identical. In this last
3666 # case, the routine chooses the more "modern"
3667 # one to use. This is because some of the
3668 # older files are formatted with values that
3669 # are, for example, ALL CAPs, whereas the
3670 # derived files have a more modern style,
3671 # which looks better. By looking for this
3672 # style when the pre-existing and replacement
3673 # standard forms are the same, we can move to
3674 # the modern style
9470941f 3675 # => $MULTIPLE_BEFORE means that if this range duplicates an
99870f4d
KW
3676 # existing one, but has a different value,
3677 # don't replace the existing one, but insert
3678 # this, one so that the same range can occur
53d84487
KW
3679 # multiple times. They are stored LIFO, so
3680 # that the final one inserted is the first one
3681 # returned in an ordered search of the table.
6901521e
KW
3682 # If this is an exact duplicate, including the
3683 # value, the original will be moved to be
3684 # first, before any other duplicate ranges
3685 # with different values.
7f4b1e25
KW
3686 # => $MULTIPLE_AFTER is like $MULTIPLE_BEFORE, but is stored
3687 # FIFO, so that this one is inserted after all
6901521e
KW
3688 # others that currently exist. If this is an
3689 # exact duplicate, including value, of an
3690 # existing range, this one is discarded
3691 # (leaving the existing one in its original,
3692 # higher priority position
99870f4d
KW
3693 # => anything else is the same as => $IF_NOT_EQUIVALENT
3694 #
c1739a4a
KW
3695 # "same value" means identical for non-type-0 ranges, and it means
3696 # having the same standard forms for type-0 ranges.
99870f4d
KW
3697
3698 return Carp::carp_too_few_args(\@_, 5) if main::DEBUG && @_ < 5;
3699
3700 my $self = shift;
3701 my $operation = shift; # '+' for add/replace; '-' for delete;
3702 my $start = shift;
3703 my $end = shift;
3704 my $value = shift;
3705
3706 my %args = @_;
3707
3708 $value = "" if not defined $value; # warning: $value can be "0"
3709
3710 my $replace = delete $args{'Replace'};
3711 $replace = $IF_NOT_EQUIVALENT unless defined $replace;
3712
3713 my $type = delete $args{'Type'};
3714 $type = 0 unless defined $type;
3715
3716 Carp::carp_extra_args(\%args) if main::DEBUG && %args;
3717
ffe43484 3718 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3719
3720 if ($operation ne '+' && $operation ne '-') {
3721 Carp::my_carp_bug("$owner_name_of{$addr}First parameter to _add_delete must be '+' or '-'. No action taken.");
3722 return;
3723 }
3724 unless (defined $start && defined $end) {
3725 Carp::my_carp_bug("$owner_name_of{$addr}Undefined start and/or end to _add_delete. No action taken.");
3726 return;
3727 }
3728 unless ($end >= $start) {
3729 Carp::my_carp_bug("$owner_name_of{$addr}End of range (" . sprintf("%04X", $end) . ") must not be before start (" . sprintf("%04X", $start) . "). No action taken.");
3730 return;
3731 }
556ca434
KW
3732 if ($end > $MAX_UNICODE_CODEPOINT && $operation eq '+') {
3733 Carp::my_carp("$owner_name_of{$addr}Warning: Range '" . sprintf("%04X..%04X", $start, $end) . ") is above the Unicode maximum of " . sprintf("%04X", $MAX_UNICODE_CODEPOINT) . ". Adding it anyway");
3734 }
99870f4d
KW
3735 #local $to_trace = 1 if main::DEBUG;
3736
3737 if ($operation eq '-') {
3738 if ($replace != $IF_NOT_EQUIVALENT) {
3739 Carp::my_carp_bug("$owner_name_of{$addr}Replace => \$IF_NOT_EQUIVALENT is required when deleting a range from a range list. Assuming Replace => \$IF_NOT_EQUIVALENT.");
3740 $replace = $IF_NOT_EQUIVALENT;
3741 }
3742 if ($type) {
3743 Carp::my_carp_bug("$owner_name_of{$addr}Type => 0 is required when deleting a range from a range list. Assuming Type => 0.");
3744 $type = 0;
3745 }
3746 if ($value ne "") {
3747 Carp::my_carp_bug("$owner_name_of{$addr}Value => \"\" is required when deleting a range from a range list. Assuming Value => \"\".");
3748 $value = "";
3749 }
3750 }
3751
3752 my $r = $ranges{$addr}; # The current list of ranges
3753 my $range_list_size = scalar @$r; # And its size
3754 my $max = $max{$addr}; # The current high code point in
3755 # the list of ranges
3756
3757 # Do a special case requiring fewer machine cycles when the new range
3758 # starts after the current highest point. The Unicode input data is
3759 # structured so this is common.
3760 if ($start > $max) {
3761
52d4d76a 3762 trace "$owner_name_of{$addr} $operation", sprintf("%04X..%04X (%s) type=%d; prev max=%04X", $start, $end, $value, $type, $max) if main::DEBUG && $to_trace;