This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
mktables: Suppress output files that shouldn't be written
[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
9be22ca5
KW
7# There was an attempt when this was first rewritten to make it 5.8
8# compatible, but that has now been abandoned, and newer constructs are used
9# as convenient.
6c68572b 10
cdcef19a 11my $start_time;
98dc9551 12BEGIN { # Get the time the script started running; do it at compilation to
cdcef19a
KW
13 # get it as close as possible
14 $start_time= time;
15}
16
23e33b60 17require 5.010_001;
d73e5302 18use strict;
99870f4d 19use warnings;
cf25bb62 20use Carp;
bd9ebcfd 21use Config;
99870f4d
KW
22use File::Find;
23use File::Path;
d07a55ed 24use File::Spec;
99870f4d 25use Text::Tabs;
6b64c11c 26use re "/aa";
99870f4d
KW
27
28sub DEBUG () { 0 } # Set to 0 for production; 1 for development
bd9ebcfd 29my $debugging_build = $Config{"ccflags"} =~ /-DDEBUGGING/;
99870f4d 30
74cd47d0
KW
31sub NON_ASCII_PLATFORM { ord("A") != 65 }
32
99870f4d
KW
33##########################################################################
34#
35# mktables -- create the runtime Perl Unicode files (lib/unicore/.../*.pl),
36# from the Unicode database files (lib/unicore/.../*.txt), It also generates
232ed87f 37# a pod file and .t files, depending on option parameters.
99870f4d
KW
38#
39# The structure of this file is:
40# First these introductory comments; then
41# code needed for everywhere, such as debugging stuff; then
42# code to handle input parameters; then
43# data structures likely to be of external interest (some of which depend on
44# the input parameters, so follows them; then
45# more data structures and subroutine and package (class) definitions; then
46# the small actual loop to process the input files and finish up; then
47# a __DATA__ section, for the .t tests
48#
232ed87f
KW
49# This program works on all releases of Unicode so far. The outputs have been
50# scrutinized most intently for release 5.1. The others have been checked for
51# somewhat more than just sanity. It can handle all non-provisional Unicode
52# character properties in those releases.
99870f4d 53#
99870f4d
KW
54# This program is mostly about Unicode character (or code point) properties.
55# A property describes some attribute or quality of a code point, like if it
56# is lowercase or not, its name, what version of Unicode it was first defined
57# in, or what its uppercase equivalent is. Unicode deals with these disparate
58# possibilities by making all properties into mappings from each code point
59# into some corresponding value. In the case of it being lowercase or not,
60# the mapping is either to 'Y' or 'N' (or various synonyms thereof). Each
61# property maps each Unicode code point to a single value, called a "property
232ed87f
KW
62# value". (Some more recently defined properties, map a code point to a set
63# of values.)
99870f4d
KW
64#
65# When using a property in a regular expression, what is desired isn't the
66# mapping of the code point to its property's value, but the reverse (or the
67# mathematical "inverse relation"): starting with the property value, "Does a
68# code point map to it?" These are written in a "compound" form:
69# \p{property=value}, e.g., \p{category=punctuation}. This program generates
70# files containing the lists of code points that map to each such regular
71# expression property value, one file per list
72#
73# There is also a single form shortcut that Perl adds for many of the commonly
74# used properties. This happens for all binary properties, plus script,
75# general_category, and block properties.
76#
77# Thus the outputs of this program are files. There are map files, mostly in
78# the 'To' directory; and there are list files for use in regular expression
79# matching, all in subdirectories of the 'lib' directory, with each
80# subdirectory being named for the property that the lists in it are for.
81# Bookkeeping, test, and documentation files are also generated.
82
83my $matches_directory = 'lib'; # Where match (\p{}) files go.
84my $map_directory = 'To'; # Where map files go.
85
86# DATA STRUCTURES
87#
88# The major data structures of this program are Property, of course, but also
89# Table. There are two kinds of tables, very similar to each other.
90# "Match_Table" is the data structure giving the list of code points that have
91# a particular property value, mentioned above. There is also a "Map_Table"
92# data structure which gives the property's mapping from code point to value.
93# There are two structures because the match tables need to be combined in
94# various ways, such as constructing unions, intersections, complements, etc.,
95# and the map ones don't. And there would be problems, perhaps subtle, if
96# a map table were inadvertently operated on in some of those ways.
97# The use of separate classes with operations defined on one but not the other
98# prevents accidentally confusing the two.
99#
100# At the heart of each table's data structure is a "Range_List", which is just
101# an ordered list of "Ranges", plus ancillary information, and methods to
102# operate on them. A Range is a compact way to store property information.
103# Each range has a starting code point, an ending code point, and a value that
104# is meant to apply to all the code points between the two end points,
105# inclusive. For a map table, this value is the property value for those
106# code points. Two such ranges could be written like this:
107# 0x41 .. 0x5A, 'Upper',
108# 0x61 .. 0x7A, 'Lower'
109#
110# Each range also has a type used as a convenience to classify the values.
111# Most ranges in this program will be Type 0, or normal, but there are some
112# ranges that have a non-zero type. These are used only in map tables, and
113# are for mappings that don't fit into the normal scheme of things. Mappings
114# that require a hash entry to communicate with utf8.c are one example;
115# another example is mappings for charnames.pm to use which indicate a name
232ed87f 116# that is algorithmically determinable from its code point (and the reverse).
99870f4d
KW
117# These are used to significantly compact these tables, instead of listing
118# each one of the tens of thousands individually.
119#
120# In a match table, the value of a range is irrelevant (and hence the type as
121# well, which will always be 0), and arbitrarily set to the null string.
122# Using the example above, there would be two match tables for those two
123# entries, one named Upper would contain the 0x41..0x5A range, and the other
124# named Lower would contain 0x61..0x7A.
125#
126# Actually, there are two types of range lists, "Range_Map" is the one
127# associated with map tables, and "Range_List" with match tables.
232ed87f
KW
128# Again, this is so that methods can be defined on one and not the others so
129# as to prevent operating on them in incorrect ways.
99870f4d
KW
130#
131# Eventually, most tables are written out to files to be read by utf8_heavy.pl
132# in the perl core. All tables could in theory be written, but some are
133# suppressed because there is no current practical use for them. It is easy
134# to change which get written by changing various lists that are near the top
135# of the actual code in this file. The table data structures contain enough
136# ancillary information to allow them to be treated as separate entities for
137# writing, such as the path to each one's file. There is a heading in each
138# map table that gives the format of its entries, and what the map is for all
139# the code points missing from it. (This allows tables to be more compact.)
678f13d5 140#
99870f4d
KW
141# The Property data structure contains one or more tables. All properties
142# contain a map table (except the $perl property which is a
143# pseudo-property containing only match tables), and any properties that
144# are usable in regular expression matches also contain various matching
145# tables, one for each value the property can have. A binary property can
146# have two values, True and False (or Y and N, which are preferred by Unicode
147# terminology). Thus each of these properties will have a map table that
148# takes every code point and maps it to Y or N (but having ranges cuts the
149# number of entries in that table way down), and two match tables, one
150# which has a list of all the code points that map to Y, and one for all the
232ed87f 151# code points that map to N. (For each binary property, a third table is also
99870f4d 152# generated for the pseudo Perl property. It contains the identical code
232ed87f
KW
153# points as the Y table, but can be written in regular expressions, not in the
154# compound form, but in a "single" form like \p{IsUppercase}.) Many
155# properties are binary, but some properties have several possible values,
156# some have many, and properties like Name have a different value for every
157# named code point. Those will not, unless the controlling lists are changed,
158# have their match tables written out. But all the ones which can be used in
159# regular expression \p{} and \P{} constructs will. Prior to 5.14, generally
160# a property would have either its map table or its match tables written but
161# not both. Again, what gets written is controlled by lists which can easily
162# be changed. Starting in 5.14, advantage was taken of this, and all the map
163# tables needed to reconstruct the Unicode db are now written out, while
164# suppressing the Unicode .txt files that contain the data. Our tables are
165# much more compact than the .txt files, so a significant space savings was
166# achieved. Also, tables are not written out that are trivially derivable
167# from tables that do get written. So, there typically is no file containing
168# the code points not matched by a binary property (the table for \P{} versus
169# lowercase \p{}), since you just need to invert the True table to get the
170# False table.
171
172# Properties have a 'Type', like 'binary', or 'string', or 'enum' depending on
173# how many match tables there are and the content of the maps. This 'Type' is
c12f2655
KW
174# different than a range 'Type', so don't get confused by the two concepts
175# having the same name.
678f13d5 176#
99870f4d
KW
177# For information about the Unicode properties, see Unicode's UAX44 document:
178
179my $unicode_reference_url = 'http://www.unicode.org/reports/tr44/';
180
181# As stated earlier, this program will work on any release of Unicode so far.
182# Most obvious problems in earlier data have NOT been corrected except when
be864b6c 183# necessary to make Perl or this program work reasonably, and to keep out
232ed87f
KW
184# potential security issues. For example, no folding information was given in
185# early releases, so this program substitutes lower case instead, just so that
186# a regular expression with the /i option will do something that actually
187# gives the right results in many cases. There are also a couple other
188# corrections for version 1.1.5, commented at the point they are made. As an
189# example of corrections that weren't made (but could be) is this statement
190# from DerivedAge.txt: "The supplementary private use code points and the
191# non-character code points were assigned in version 2.0, but not specifically
192# listed in the UCD until versions 3.0 and 3.1 respectively." (To be precise
193# it was 3.0.1 not 3.0.0) More information on Unicode version glitches is
194# further down in these introductory comments.
99870f4d 195#
232ed87f
KW
196# This program works on all non-provisional properties as of the current
197# Unicode release, though the files for some are suppressed for various
198# reasons. You can change which are output by changing lists in this program.
678f13d5 199#
dc85bd38 200# The old version of mktables emphasized the term "Fuzzy" to mean Unicode's
99870f4d
KW
201# loose matchings rules (from Unicode TR18):
202#
203# The recommended names for UCD properties and property values are in
204# PropertyAliases.txt [Prop] and PropertyValueAliases.txt
205# [PropValue]. There are both abbreviated names and longer, more
206# descriptive names. It is strongly recommended that both names be
207# recognized, and that loose matching of property names be used,
208# whereby the case distinctions, whitespace, hyphens, and underbar
209# are ignored.
232ed87f 210#
99870f4d
KW
211# The program still allows Fuzzy to override its determination of if loose
212# matching should be used, but it isn't currently used, as it is no longer
213# needed; the calculations it makes are good enough.
678f13d5 214#
99870f4d
KW
215# SUMMARY OF HOW IT WORKS:
216#
217# Process arguments
218#
219# A list is constructed containing each input file that is to be processed
220#
221# Each file on the list is processed in a loop, using the associated handler
222# code for each:
223# The PropertyAliases.txt and PropValueAliases.txt files are processed
224# first. These files name the properties and property values.
225# Objects are created of all the property and property value names
226# that the rest of the input should expect, including all synonyms.
227# The other input files give mappings from properties to property
228# values. That is, they list code points and say what the mapping
229# is under the given property. Some files give the mappings for
230# just one property; and some for many. This program goes through
232ed87f
KW
231# each file and populates the properties and their map tables from
232# them. Some properties are listed in more than one file, and
233# Unicode has set up a precedence as to which has priority if there
234# is a conflict. Thus the order of processing matters, and this
235# program handles the conflict possibility by processing the
236# overriding input files last, so that if necessary they replace
237# earlier values.
99870f4d
KW
238# After this is all done, the program creates the property mappings not
239# furnished by Unicode, but derivable from what it does give.
240# The tables of code points that match each property value in each
241# property that is accessible by regular expressions are created.
242# The Perl-defined properties are created and populated. Many of these
243# require data determined from the earlier steps
244# Any Perl-defined synonyms are created, and name clashes between Perl
678f13d5 245# and Unicode are reconciled and warned about.
99870f4d
KW
246# All the properties are written to files
247# Any other files are written, and final warnings issued.
678f13d5 248#
99870f4d
KW
249# For clarity, a number of operators have been overloaded to work on tables:
250# ~ means invert (take all characters not in the set). The more
251# conventional '!' is not used because of the possibility of confusing
252# it with the actual boolean operation.
253# + means union
254# - means subtraction
255# & means intersection
256# The precedence of these is the order listed. Parentheses should be
257# copiously used. These are not a general scheme. The operations aren't
258# defined for a number of things, deliberately, to avoid getting into trouble.
259# Operations are done on references and affect the underlying structures, so
260# that the copy constructors for them have been overloaded to not return a new
261# clone, but the input object itself.
678f13d5 262#
99870f4d
KW
263# The bool operator is deliberately not overloaded to avoid confusion with
264# "should it mean if the object merely exists, or also is non-empty?".
99870f4d
KW
265#
266# WHY CERTAIN DESIGN DECISIONS WERE MADE
678f13d5
KW
267#
268# This program needs to be able to run under miniperl. Therefore, it uses a
269# minimum of other modules, and hence implements some things itself that could
270# be gotten from CPAN
271#
272# This program uses inputs published by the Unicode Consortium. These can
273# change incompatibly between releases without the Perl maintainers realizing
274# it. Therefore this program is now designed to try to flag these. It looks
275# at the directories where the inputs are, and flags any unrecognized files.
276# It keeps track of all the properties in the files it handles, and flags any
277# that it doesn't know how to handle. It also flags any input lines that
278# don't match the expected syntax, among other checks.
279#
280# It is also designed so if a new input file matches one of the known
281# templates, one hopefully just needs to add it to a list to have it
282# processed.
283#
284# As mentioned earlier, some properties are given in more than one file. In
285# particular, the files in the extracted directory are supposedly just
286# reformattings of the others. But they contain information not easily
9be22ca5
KW
287# derivable from the other files, including results for Unihan (which isn't
288# usually available to this program) and for unassigned code points. They
678f13d5
KW
289# also have historically had errors or been incomplete. In an attempt to
290# create the best possible data, this program thus processes them first to
291# glean information missing from the other files; then processes those other
292# files to override any errors in the extracted ones. Much of the design was
293# driven by this need to store things and then possibly override them.
294#
295# It tries to keep fatal errors to a minimum, to generate something usable for
296# testing purposes. It always looks for files that could be inputs, and will
297# warn about any that it doesn't know how to handle (the -q option suppresses
298# the warning).
99870f4d 299#
678f13d5
KW
300# Why is there more than one type of range?
301# This simplified things. There are some very specialized code points that
302# have to be handled specially for output, such as Hangul syllable names.
303# By creating a range type (done late in the development process), it
304# allowed this to be stored with the range, and overridden by other input.
305# Originally these were stored in another data structure, and it became a
306# mess trying to decide if a second file that was for the same property was
307# overriding the earlier one or not.
308#
309# Why are there two kinds of tables, match and map?
310# (And there is a base class shared by the two as well.) As stated above,
311# they actually are for different things. Development proceeded much more
312# smoothly when I (khw) realized the distinction. Map tables are used to
313# give the property value for every code point (actually every code point
314# that doesn't map to a default value). Match tables are used for regular
315# expression matches, and are essentially the inverse mapping. Separating
316# the two allows more specialized methods, and error checks so that one
317# can't just take the intersection of two map tables, for example, as that
318# is nonsensical.
99870f4d 319#
232ed87f
KW
320# What about 'fate' and 'status'. The concept of a table's fate was created
321# late when it became clear that something more was needed. The difference
322# between this and 'status' is unclean, and could be improved if someone
323# wanted to spend the effort.
324#
23e33b60
KW
325# DEBUGGING
326#
678f13d5
KW
327# This program is written so it will run under miniperl. Occasionally changes
328# will cause an error where the backtrace doesn't work well under miniperl.
329# To diagnose the problem, you can instead run it under regular perl, if you
330# have one compiled.
331#
332# There is a good trace facility. To enable it, first sub DEBUG must be set
333# to return true. Then a line like
334#
335# local $to_trace = 1 if main::DEBUG;
336#
232ed87f
KW
337# can be added to enable tracing in its lexical scope (plus dynamic) or until
338# you insert another line:
678f13d5
KW
339#
340# local $to_trace = 0 if main::DEBUG;
341#
232ed87f 342# To actually trace, use a line like "trace $a, @b, %c, ...;
678f13d5
KW
343#
344# Some of the more complex subroutines already have trace statements in them.
345# Permanent trace statements should be like:
346#
347# trace ... if main::DEBUG && $to_trace;
348#
349# If there is just one or a few files that you're debugging, you can easily
350# cause most everything else to be skipped. Change the line
351#
352# my $debug_skip = 0;
353#
354# to 1, and every file whose object is in @input_file_objects and doesn't have
232ed87f
KW
355# a, 'non_skip => 1,' in its constructor will be skipped. However, skipping
356# Jamo.txt or UnicodeData.txt will likely cause fatal errors.
678f13d5 357#
b4a0206c 358# To compare the output tables, it may be useful to specify the -annotate
97a8a595
KW
359# flag. (As of this writing, this can't be done on a clean workspace, due to
360# requirements in Text::Tabs used in this option; so first run mktables
361# without this option.) This option adds comment lines to each table, one for
362# each non-algorithmically named character giving, currently its code point,
363# name, and graphic representation if printable (and you have a font that
364# knows about it). This makes it easier to see what the particular code
365# points are in each output table. Non-named code points are annotated with a
366# description of their status, and contiguous ones with the same description
367# will be output as a range rather than individually. Algorithmically named
368# characters are also output as ranges, except when there are just a few
369# contiguous ones.
c4019d52 370#
99870f4d
KW
371# FUTURE ISSUES
372#
373# The program would break if Unicode were to change its names so that
374# interior white space, underscores, or dashes differences were significant
375# within property and property value names.
376#
377# It might be easier to use the xml versions of the UCD if this program ever
378# would need heavy revision, and the ability to handle old versions was not
379# required.
380#
381# There is the potential for name collisions, in that Perl has chosen names
382# that Unicode could decide it also likes. There have been such collisions in
383# the past, with mostly Perl deciding to adopt the Unicode definition of the
384# name. However in the 5.2 Unicode beta testing, there were a number of such
385# collisions, which were withdrawn before the final release, because of Perl's
386# and other's protests. These all involved new properties which began with
387# 'Is'. Based on the protests, Unicode is unlikely to try that again. Also,
388# many of the Perl-defined synonyms, like Any, Word, etc, are listed in a
389# Unicode document, so they are unlikely to be used by Unicode for another
390# purpose. However, they might try something beginning with 'In', or use any
391# of the other Perl-defined properties. This program will warn you of name
392# collisions, and refuse to generate tables with them, but manual intervention
393# will be required in this event. One scheme that could be implemented, if
394# necessary, would be to have this program generate another file, or add a
395# field to mktables.lst that gives the date of first definition of a property.
396# Each new release of Unicode would use that file as a basis for the next
397# iteration. And the Perl synonym addition code could sort based on the age
398# of the property, so older properties get priority, and newer ones that clash
399# would be refused; hence existing code would not be impacted, and some other
400# synonym would have to be used for the new property. This is ugly, and
401# manual intervention would certainly be easier to do in the short run; lets
402# hope it never comes to this.
678f13d5 403#
99870f4d
KW
404# A NOTE ON UNIHAN
405#
406# This program can generate tables from the Unihan database. But it doesn't
407# by default, letting the CPAN module Unicode::Unihan handle them. Prior to
408# version 5.2, this database was in a single file, Unihan.txt. In 5.2 the
409# database was split into 8 different files, all beginning with the letters
410# 'Unihan'. This program will read those file(s) if present, but it needs to
411# know which of the many properties in the file(s) should have tables created
412# for them. It will create tables for any properties listed in
413# PropertyAliases.txt and PropValueAliases.txt, plus any listed in the
414# @cjk_properties array and the @cjk_property_values array. Thus, if a
415# property you want is not in those files of the release you are building
416# against, you must add it to those two arrays. Starting in 4.0, the
417# Unicode_Radical_Stroke was listed in those files, so if the Unihan database
418# is present in the directory, a table will be generated for that property.
419# In 5.2, several more properties were added. For your convenience, the two
5f7264c7 420# arrays are initialized with all the 6.0 listed properties that are also in
99870f4d
KW
421# earlier releases. But these are commented out. You can just uncomment the
422# ones you want, or use them as a template for adding entries for other
423# properties.
424#
425# You may need to adjust the entries to suit your purposes. setup_unihan(),
426# and filter_unihan_line() are the functions where this is done. This program
427# already does some adjusting to make the lines look more like the rest of the
428# Unicode DB; You can see what that is in filter_unihan_line()
429#
430# There is a bug in the 3.2 data file in which some values for the
431# kPrimaryNumeric property have commas and an unexpected comment. A filter
9be22ca5
KW
432# could be added to correct these; or for a particular installation, the
433# Unihan.txt file could be edited to fix them.
99870f4d 434#
678f13d5
KW
435# HOW TO ADD A FILE TO BE PROCESSED
436#
437# A new file from Unicode needs to have an object constructed for it in
438# @input_file_objects, probably at the end or at the end of the extracted
439# ones. The program should warn you if its name will clash with others on
440# restrictive file systems, like DOS. If so, figure out a better name, and
441# add lines to the README.perl file giving that. If the file is a character
232ed87f 442# property, it should be in the format that Unicode has implicitly
678f13d5
KW
443# standardized for such files for the more recently introduced ones.
444# If so, the Input_file constructor for @input_file_objects can just be the
445# file name and release it first appeared in. If not, then it should be
446# possible to construct an each_line_handler() to massage the line into the
447# standardized form.
448#
449# For non-character properties, more code will be needed. You can look at
450# the existing entries for clues.
451#
452# UNICODE VERSIONS NOTES
453#
454# The Unicode UCD has had a number of errors in it over the versions. And
455# these remain, by policy, in the standard for that version. Therefore it is
456# risky to correct them, because code may be expecting the error. So this
457# program doesn't generally make changes, unless the error breaks the Perl
458# core. As an example, some versions of 2.1.x Jamo.txt have the wrong value
459# for U+1105, which causes real problems for the algorithms for Jamo
460# calculations, so it is changed here.
461#
462# But it isn't so clear cut as to what to do about concepts that are
463# introduced in a later release; should they extend back to earlier releases
464# where the concept just didn't exist? It was easier to do this than to not,
465# so that's what was done. For example, the default value for code points not
466# in the files for various properties was probably undefined until changed by
467# some version. No_Block for blocks is such an example. This program will
468# assign No_Block even in Unicode versions that didn't have it. This has the
469# benefit that code being written doesn't have to special case earlier
470# versions; and the detriment that it doesn't match the Standard precisely for
471# the affected versions.
472#
473# Here are some observations about some of the issues in early versions:
474#
232ed87f
KW
475# Prior to version 3.0, there were 3 character decompositions. These are not
476# handled by Unicode::Normalize, nor will it compile when presented a version
477# that has them. However, you can trivially get it to compile by simply
478# ignoring those decompositions, by changing the croak to a carp. At the time
9be22ca5
KW
479# of this writing, the line (in cpan/Unicode-Normalize/Normalize.pm or
480# cpan/Unicode-Normalize/mkheader) reads
232ed87f
KW
481#
482# croak("Weird Canonical Decomposition of U+$h");
483#
28807e1d 484# Simply comment it out. It will compile, but will not know about any three
9be22ca5 485# character decompositions.
232ed87f
KW
486
487# The number of code points in \p{alpha=True} halved in 2.1.9. It turns out
488# that the reason is that the CJK block starting at 4E00 was removed from
489# PropList, and was not put back in until 3.1.0. The Perl extension (the
490# single property name \p{alpha}) has the correct values. But the compound
491# form is simply not generated until 3.1, as it can be argued that prior to
492# this release, this was not an official property. The comments for
493# filter_old_style_proplist() give more details.
678f13d5
KW
494#
495# Unicode introduced the synonym Space for White_Space in 4.1. Perl has
496# always had a \p{Space}. In release 3.2 only, they are not synonymous. The
497# reason is that 3.2 introduced U+205F=medium math space, which was not
498# classed as white space, but Perl figured out that it should have been. 4.0
499# reclassified it correctly.
500#
501# Another change between 3.2 and 4.0 is the CCC property value ATBL. In 3.2
232ed87f
KW
502# this was erroneously a synonym for 202 (it should be 200). In 4.0, ATB
503# became 202, and ATBL was left with no code points, as all the ones that
504# mapped to 202 stayed mapped to 202. Thus if your program used the numeric
505# name for the class, it would not have been affected, but if it used the
506# mnemonic, it would have been.
678f13d5 507#
9be22ca5 508# \p{Script=Hrkt} (Katakana_Or_Hiragana) came in 4.0.1. Before that, code
678f13d5
KW
509# points which eventually came to have this script property value, instead
510# mapped to "Unknown". But in the next release all these code points were
511# moved to \p{sc=common} instead.
9be22ca5
KW
512
513# The tests furnished by Unicode for testing WordBreak and SentenceBreak
514# generate errors in 5.0 and earlier.
99870f4d
KW
515#
516# The default for missing code points for BidiClass is complicated. Starting
517# in 3.1.1, the derived file DBidiClass.txt handles this, but this program
518# tries to do the best it can for earlier releases. It is done in
519# process_PropertyAliases()
520#
232ed87f
KW
521# In version 2.1.2, the entry in UnicodeData.txt:
522# 0275;LATIN SMALL LETTER BARRED O;Ll;0;L;;;;;N;;;;019F;
523# should instead be
524# 0275;LATIN SMALL LETTER BARRED O;Ll;0;L;;;;;N;;;019F;;019F
525# Without this change, there are casing problems for this character.
526#
7803ad2d
KW
527# Search for $string_compare_versions to see how to compare changes to
528# properties between Unicode versions
529#
99870f4d
KW
530##############################################################################
531
532my $UNDEF = ':UNDEF:'; # String to print out for undefined values in tracing
533 # and errors
534my $MAX_LINE_WIDTH = 78;
535
536# Debugging aid to skip most files so as to not be distracted by them when
537# concentrating on the ones being debugged. Add
538# non_skip => 1,
539# to the constructor for those files you want processed when you set this.
540# Files with a first version number of 0 are special: they are always
c12f2655
KW
541# processed regardless of the state of this flag. Generally, Jamo.txt and
542# UnicodeData.txt must not be skipped if you want this program to not die
543# before normal completion.
99870f4d
KW
544my $debug_skip = 0;
545
e9c4b4f8
KW
546
547# Normally these are suppressed.
548my $write_Unicode_deprecated_tables = 0;
549
99870f4d
KW
550# Set to 1 to enable tracing.
551our $to_trace = 0;
552
553{ # Closure for trace: debugging aid
554 my $print_caller = 1; # ? Include calling subroutine name
555 my $main_with_colon = 'main::';
556 my $main_colon_length = length($main_with_colon);
557
558 sub trace {
559 return unless $to_trace; # Do nothing if global flag not set
560
561 my @input = @_;
562
563 local $DB::trace = 0;
564 $DB::trace = 0; # Quiet 'used only once' message
565
566 my $line_number;
567
568 # Loop looking up the stack to get the first non-trace caller
569 my $caller_line;
570 my $caller_name;
571 my $i = 0;
572 do {
573 $line_number = $caller_line;
574 (my $pkg, my $file, $caller_line, my $caller) = caller $i++;
575 $caller = $main_with_colon unless defined $caller;
576
577 $caller_name = $caller;
578
579 # get rid of pkg
580 $caller_name =~ s/.*:://;
581 if (substr($caller_name, 0, $main_colon_length)
582 eq $main_with_colon)
583 {
584 $caller_name = substr($caller_name, $main_colon_length);
585 }
586
587 } until ($caller_name ne 'trace');
588
589 # If the stack was empty, we were called from the top level
590 $caller_name = 'main' if ($caller_name eq ""
591 || $caller_name eq 'trace');
592
593 my $output = "";
9346f59a 594 #print STDERR __LINE__, ": ", join ", ", @input, "\n";
99870f4d 595 foreach my $string (@input) {
99870f4d
KW
596 if (ref $string eq 'ARRAY' || ref $string eq 'HASH') {
597 $output .= simple_dumper($string);
598 }
599 else {
600 $string = "$string" if ref $string;
601 $string = $UNDEF unless defined $string;
602 chomp $string;
603 $string = '""' if $string eq "";
604 $output .= " " if $output ne ""
605 && $string ne ""
606 && substr($output, -1, 1) ne " "
607 && substr($string, 0, 1) ne " ";
608 $output .= $string;
609 }
610 }
611
99f78760
KW
612 print STDERR sprintf "%4d: ", $line_number if defined $line_number;
613 print STDERR "$caller_name: " if $print_caller;
99870f4d
KW
614 print STDERR $output, "\n";
615 return;
616 }
617}
618
619# This is for a rarely used development feature that allows you to compare two
620# versions of the Unicode standard without having to deal with changes caused
9be22ca5
KW
621# by the code points introduced in the later version. You probably also want
622# to use the -annotate option when using this. Change the 0 to a string
623# containing a SINGLE dotted Unicode release number (e.g. "2.1"). Only code
624# points introduced in that release and earlier will be used; later ones are
625# thrown away. You use the version number of the earliest one you want to
c12f2655
KW
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 861# unless explicitly added.
68ab5aec 862if ($v_version ge v5.2.0 && ! $write_Unicode_deprecated_tables) {
99870f4d 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
9be22ca5
KW
886# Enum values for to_output_map() method in the Map_Table package. (0 is don't
887# output)
272501f6
KW
888my $EXTERNAL_MAP = 1;
889my $INTERNAL_MAP = 2;
ce712c88 890my $OUTPUT_ADJUSTED = 3;
272501f6 891
fcf1973c
KW
892# To override computed values for writing the map tables for these properties.
893# The default for enum map tables is to write them out, so that the Unicode
894# .txt files can be removed, but all the data to compute any property value
895# for any code point is available in a more compact form.
896my %global_to_output_map = (
897 # Needed by UCD.pm, but don't want to publicize that it exists, so won't
c12f2655
KW
898 # get stuck supporting it if things change. Since it is a STRING
899 # property, it normally would be listed in the pod, but INTERNAL_MAP
900 # suppresses that.
fcf1973c
KW
901 Unicode_1_Name => $INTERNAL_MAP,
902
903 Present_In => 0, # Suppress, as easily computed from Age
74cd47d0
KW
904 Block => (NON_ASCII_PLATFORM) ? 1 : 0, # Suppress, as Blocks.txt is
905 # retained, but needed for
906 # non-ASCII
53d34b6c
KW
907
908 # Suppress, as mapping can be found instead from the
909 # Perl_Decomposition_Mapping file
910 Decomposition_Type => 0,
fcf1973c
KW
911);
912
99870f4d 913# Properties that this program ignores.
230e0c16
KW
914my @unimplemented_properties;
915
916# With this release, it is automatically handled if the Unihan db is
917# downloaded
431c1f00 918push @unimplemented_properties, 'Unicode_Radical_Stroke' if $v_version lt v5.2.0;
d73e5302 919
99870f4d
KW
920# There are several types of obsolete properties defined by Unicode. These
921# must be hand-edited for every new Unicode release.
922my %why_deprecated; # Generates a deprecated warning message if used.
923my %why_stabilized; # Documentation only
924my %why_obsolete; # Documentation only
925
926{ # Closure
8364f9eb 927 my $simple = 'Perl uses the more complete version';
99870f4d
KW
928 my $unihan = 'Unihan properties are by default not enabled in the Perl core. Instead use CPAN: Unicode::Unihan';
929
930 my $other_properties = 'other properties';
931 my $contributory = "Used by Unicode internally for generating $other_properties and not intended to be used stand-alone";
5d294d41 932 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
933
934 %why_deprecated = (
5f7264c7 935 'Grapheme_Link' => 'Deprecated by Unicode: Duplicates ccc=vr (Canonical_Combining_Class=Virama)',
99870f4d
KW
936 'Jamo_Short_Name' => $contributory,
937 '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',
938 'Other_Alphabetic' => $contributory,
939 'Other_Default_Ignorable_Code_Point' => $contributory,
940 'Other_Grapheme_Extend' => $contributory,
941 'Other_ID_Continue' => $contributory,
942 'Other_ID_Start' => $contributory,
943 'Other_Lowercase' => $contributory,
944 'Other_Math' => $contributory,
945 'Other_Uppercase' => $contributory,
e22aaf5c
KW
946 'Expands_On_NFC' => $why_no_expand,
947 'Expands_On_NFD' => $why_no_expand,
948 'Expands_On_NFKC' => $why_no_expand,
949 'Expands_On_NFKD' => $why_no_expand,
99870f4d
KW
950 );
951
952 %why_suppressed = (
5f7264c7 953 # There is a lib/unicore/Decomposition.pl (used by Normalize.pm) which
99870f4d
KW
954 # contains the same information, but without the algorithmically
955 # determinable Hangul syllables'. This file is not published, so it's
956 # existence is not noted in the comment.
12fee290 957 'Decomposition_Mapping' => 'Accessible via Unicode::Normalize or prop_invmap() or charprop() in Unicode::UCD::',
99870f4d 958
ac71d2a0 959 'Indic_Matra_Category' => "Withdrawn by Unicode while still provisional",
3111abc0 960
5f8d1a89
KW
961 # Don't suppress ISO_Comment, as otherwise special handling is needed
962 # to differentiate between it and gc=c, which can be written as 'isc',
963 # which is the same characters as ISO_Comment's short name.
99870f4d 964
12fee290 965 'Name' => "Accessible via \\N{...} or 'use charnames;' or charprop() or prop_invmap() in Unicode::UCD::",
e0b29447 966
12fee290
KW
967 'Simple_Case_Folding' => "$simple. Can access this through casefold(), charprop(), or prop_invmap() in Unicode::UCD",
968 'Simple_Lowercase_Mapping' => "$simple. Can access this through charinfo(), charprop(), or prop_invmap() in Unicode::UCD",
969 'Simple_Titlecase_Mapping' => "$simple. Can access this through charinfo(), charprop(), or prop_invmap() in Unicode::UCD",
970 'Simple_Uppercase_Mapping' => "$simple. Can access this through charinfo(), charprop(), or prop_invmap() in Unicode::UCD",
99870f4d 971
dac6f618 972 FC_NFKC_Closure => 'Deprecated by Unicode, and 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 => $CODE_POINT,
ec2cd995 1109 General_Category => ($v_version le 6.3.0) ? 'Cn' : 'Unassigned',
99870f4d
KW
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
346a20cf
KW
1199my $MAX_UNICODE_CODEPOINT_STRING = ($v_version ge v2.0.0)
1200 ? "10FFFF"
1201 : "FFFF";
6189eadc
KW
1202my $MAX_UNICODE_CODEPOINT = hex $MAX_UNICODE_CODEPOINT_STRING;
1203my $MAX_UNICODE_CODEPOINTS = $MAX_UNICODE_CODEPOINT + 1;
99870f4d 1204
2d88a86a
KW
1205# We work with above-Unicode code points, up to UV_MAX. But when you get
1206# that high, above IV_MAX, some operations don't work, and you can easily get
1207# overflow. Therefore for internal use, we use a much smaller number,
1208# translating it to UV_MAX only for output. The exact number is immaterial
1209# (all Unicode code points are treated exactly the same), but the algorithm
1210# requires it to be at least 2 * $MAX_UNICODE_CODEPOINTS + 1;
1211my $MAX_WORKING_CODEPOINTS= $MAX_UNICODE_CODEPOINT * 8;
1212my $MAX_WORKING_CODEPOINT = $MAX_WORKING_CODEPOINTS - 1;
1213my $MAX_WORKING_CODEPOINT_STRING = sprintf("%X", $MAX_WORKING_CODEPOINT);
1214
1215my $MAX_PLATFORM_CODEPOINT = ~0;
1216
99870f4d
KW
1217# Matches legal code point. 4-6 hex numbers, If there are 6, the first
1218# two must be 10; if there are 5, the first must not be a 0. Written this way
92199589
KW
1219# to decrease backtracking. The first regex allows the code point to be at
1220# the end of a word, but to work properly, the word shouldn't end with a valid
1221# hex character. The second one won't match a code point at the end of a
1222# word, and doesn't have the run-on issue
8c32d378
KW
1223my $run_on_code_point_re =
1224 qr/ (?: 10[0-9A-F]{4} | [1-9A-F][0-9A-F]{4} | [0-9A-F]{4} ) \b/x;
1225my $code_point_re = qr/\b$run_on_code_point_re/;
99870f4d
KW
1226
1227# This matches the beginning of the line in the Unicode db files that give the
1228# defaults for code points not listed (i.e., missing) in the file. The code
1229# depends on this ending with a semi-colon, so it can assume it is a valid
1230# field when the line is split() by semi-colons
346a20cf 1231my $missing_defaults_prefix = qr/^#\s+\@missing:\s+0000\.\.10FFFF\s*;/;
99870f4d
KW
1232
1233# Property types. Unicode has more types, but these are sufficient for our
1234# purposes.
1235my $UNKNOWN = -1; # initialized to illegal value
1236my $NON_STRING = 1; # Either binary or enum
1237my $BINARY = 2;
06f26c45
KW
1238my $FORCED_BINARY = 3; # Not a binary property, but, besides its normal
1239 # tables, additional true and false tables are
1240 # generated so that false is anything matching the
1241 # default value, and true is everything else.
1242my $ENUM = 4; # Include catalog
1243my $STRING = 5; # Anything else: string or misc
99870f4d
KW
1244
1245# Some input files have lines that give default values for code points not
1246# contained in the file. Sometimes these should be ignored.
1247my $NO_DEFAULTS = 0; # Must evaluate to false
1248my $NOT_IGNORED = 1;
1249my $IGNORED = 2;
1250
1251# Range types. Each range has a type. Most ranges are type 0, for normal,
1252# and will appear in the main body of the tables in the output files, but
1253# there are other types of ranges as well, listed below, that are specially
1254# handled. There are pseudo-types as well that will never be stored as a
1255# type, but will affect the calculation of the type.
1256
1257# 0 is for normal, non-specials
1258my $MULTI_CP = 1; # Sequence of more than code point
1259my $HANGUL_SYLLABLE = 2;
1260my $CP_IN_NAME = 3; # The NAME contains the code point appended to it.
1261my $NULL = 4; # The map is to the null string; utf8.c can't
1262 # handle these, nor is there an accepted syntax
1263 # for them in \p{} constructs
f86864ac 1264my $COMPUTE_NO_MULTI_CP = 5; # Pseudo-type; means that ranges that would
99870f4d
KW
1265 # otherwise be $MULTI_CP type are instead type 0
1266
1267# process_generic_property_file() can accept certain overrides in its input.
1268# Each of these must begin AND end with $CMD_DELIM.
1269my $CMD_DELIM = "\a";
1270my $REPLACE_CMD = 'replace'; # Override the Replace
1271my $MAP_TYPE_CMD = 'map_type'; # Override the Type
1272
1273my $NO = 0;
1274my $YES = 1;
1275
1276# Values for the Replace argument to add_range.
1277# $NO # Don't replace; add only the code points not
1278 # already present.
1279my $IF_NOT_EQUIVALENT = 1; # Replace only under certain conditions; details in
1280 # the comments at the subroutine definition.
1281my $UNCONDITIONALLY = 2; # Replace without conditions.
9470941f 1282my $MULTIPLE_BEFORE = 4; # Don't replace, but add a duplicate record if
99870f4d 1283 # already there
7f4b1e25
KW
1284my $MULTIPLE_AFTER = 5; # Don't replace, but add a duplicate record if
1285 # already there
1286my $CROAK = 6; # Die with an error if is already there
99870f4d
KW
1287
1288# Flags to give property statuses. The phrases are to remind maintainers that
1289# if the flag is changed, the indefinite article referring to it in the
1290# documentation may need to be as well.
1291my $NORMAL = "";
99870f4d
KW
1292my $DEPRECATED = 'D';
1293my $a_bold_deprecated = "a 'B<$DEPRECATED>'";
1294my $A_bold_deprecated = "A 'B<$DEPRECATED>'";
1295my $DISCOURAGED = 'X';
1296my $a_bold_discouraged = "an 'B<$DISCOURAGED>'";
1297my $A_bold_discouraged = "An 'B<$DISCOURAGED>'";
1298my $STRICTER = 'T';
1299my $a_bold_stricter = "a 'B<$STRICTER>'";
1300my $A_bold_stricter = "A 'B<$STRICTER>'";
1301my $STABILIZED = 'S';
1302my $a_bold_stabilized = "an 'B<$STABILIZED>'";
1303my $A_bold_stabilized = "An 'B<$STABILIZED>'";
1304my $OBSOLETE = 'O';
1305my $a_bold_obsolete = "an 'B<$OBSOLETE>'";
1306my $A_bold_obsolete = "An 'B<$OBSOLETE>'";
1307
5c6d9947
KW
1308# Aliases can also have an extra status:
1309my $INTERNAL_ALIAS = 'P';
1310
99870f4d
KW
1311my %status_past_participles = (
1312 $DISCOURAGED => 'discouraged',
99870f4d
KW
1313 $STABILIZED => 'stabilized',
1314 $OBSOLETE => 'obsolete',
37e2e78e 1315 $DEPRECATED => 'deprecated',
5c6d9947 1316 $INTERNAL_ALIAS => 'reserved for Perl core internal use only',
99870f4d
KW
1317);
1318
395dfc19
KW
1319# Table fates. These are somewhat ordered, so that fates < $MAP_PROXIED should be
1320# externally documented.
301ba948 1321my $ORDINARY = 0; # The normal fate.
395dfc19
KW
1322my $MAP_PROXIED = 1; # The map table for the property isn't written out,
1323 # but there is a file written that can be used to
1324 # reconstruct this table
3cdaf629 1325my $INTERNAL_ONLY = 2; # The file for this table is written out, but it is
301ba948 1326 # for Perl's internal use only
277b7b16
KW
1327my $LEGACY_ONLY = 3; # Like $INTERNAL_ONLY, but not actually used by Perl.
1328 # Is for backwards compatibility for applications that
1329 # read the file directly, so it's format is
1330 # unchangeable.
1331my $SUPPRESSED = 4; # The file for this table is not written out, and as a
3cdaf629
KW
1332 # result, we don't bother to do many computations on
1333 # it.
277b7b16 1334my $PLACEHOLDER = 5; # Like $SUPPRESSED, but we go through all the
3cdaf629
KW
1335 # computations anyway, as the values are needed for
1336 # things to work. This happens when we have Perl
1337 # extensions that depend on Unicode tables that
1338 # wouldn't normally be in a given Unicode version.
301ba948 1339
f5817e0a
KW
1340# The format of the values of the tables:
1341my $EMPTY_FORMAT = "";
99870f4d
KW
1342my $BINARY_FORMAT = 'b';
1343my $DECIMAL_FORMAT = 'd';
1344my $FLOAT_FORMAT = 'f';
1345my $INTEGER_FORMAT = 'i';
1346my $HEX_FORMAT = 'x';
1347my $RATIONAL_FORMAT = 'r';
1348my $STRING_FORMAT = 's';
d11155ec 1349my $ADJUST_FORMAT = 'a';
24303724 1350my $HEX_ADJUST_FORMAT = 'ax';
a14f3cb1 1351my $DECOMP_STRING_FORMAT = 'c';
c3ff2976 1352my $STRING_WHITE_SPACE_LIST = 'sw';
99870f4d
KW
1353
1354my %map_table_formats = (
1355 $BINARY_FORMAT => 'binary',
1356 $DECIMAL_FORMAT => 'single decimal digit',
1357 $FLOAT_FORMAT => 'floating point number',
1358 $INTEGER_FORMAT => 'integer',
add63c13 1359 $HEX_FORMAT => 'non-negative hex whole number; a code point',
99870f4d 1360 $RATIONAL_FORMAT => 'rational: an integer or a fraction',
1a9d544b 1361 $STRING_FORMAT => 'string',
d11155ec 1362 $ADJUST_FORMAT => 'some entries need adjustment',
24303724 1363 $HEX_ADJUST_FORMAT => 'mapped value in hex; some entries need adjustment',
92f9d56c 1364 $DECOMP_STRING_FORMAT => 'Perl\'s internal (Normalize.pm) decomposition mapping',
c3ff2976 1365 $STRING_WHITE_SPACE_LIST => 'string, but some elements are interpreted as a list; white space occurs only as list item separators'
99870f4d
KW
1366);
1367
1368# Unicode didn't put such derived files in a separate directory at first.
1369my $EXTRACTED_DIR = (-d 'extracted') ? 'extracted' : "";
1370my $EXTRACTED = ($EXTRACTED_DIR) ? "$EXTRACTED_DIR/" : "";
1371my $AUXILIARY = 'auxiliary';
1372
3854b4b8
KW
1373# Hashes and arrays that will eventually go into Heavy.pl for the use of
1374# utf8_heavy.pl and into UCD.pl for the use of UCD.pm
99870f4d
KW
1375my %loose_to_file_of; # loosely maps table names to their respective
1376 # files
1377my %stricter_to_file_of; # same; but for stricter mapping.
315bfd4e 1378my %loose_property_to_file_of; # Maps a loose property name to its map file
e1dc048d 1379my %strict_property_to_file_of; # Same, but strict
3854b4b8
KW
1380my @inline_definitions = "V0"; # Each element gives a definition of a unique
1381 # inversion list. When a definition is inlined,
1382 # its value in the hash it's in (one of the two
1383 # defined just above) will include an index into
1384 # this array. The 0th element is initialized to
9be22ca5 1385 # the definition for a zero length inversion list
89cf10cc
KW
1386my %file_to_swash_name; # Maps the file name to its corresponding key name
1387 # in the hash %utf8::SwashInfo
99870f4d
KW
1388my %nv_floating_to_rational; # maps numeric values floating point numbers to
1389 # their rational equivalent
c12f2655
KW
1390my %loose_property_name_of; # Loosely maps (non_string) property names to
1391 # standard form
e1dc048d
KW
1392my %strict_property_name_of; # Strictly maps (non_string) property names to
1393 # standard form
86a52d1e 1394my %string_property_loose_to_name; # Same, for string properties.
c15fda25
KW
1395my %loose_defaults; # keys are of form "prop=value", where 'prop' is
1396 # the property name in standard loose form, and
1397 # 'value' is the default value for that property,
1398 # also in standard loose form.
9e4a1e86
KW
1399my %loose_to_standard_value; # loosely maps table names to the canonical
1400 # alias for them
2df7880f
KW
1401my %ambiguous_names; # keys are alias names (in standard form) that
1402 # have more than one possible meaning.
d963b40d
KW
1403my %combination_property; # keys are alias names (in standard form) that
1404 # have both a map table, and a binary one that
1405 # yields true for all non-null maps.
5d1df013
KW
1406my %prop_aliases; # Keys are standard property name; values are each
1407 # one's aliases
1e863613
KW
1408my %prop_value_aliases; # Keys of top level are standard property name;
1409 # values are keys to another hash, Each one is
1410 # one of the property's values, in standard form.
1411 # The values are that prop-val's aliases.
2df7880f 1412my %ucd_pod; # Holds entries that will go into the UCD section of the pod
99870f4d 1413
d867ccfb
KW
1414# Most properties are immune to caseless matching, otherwise you would get
1415# nonsensical results, as properties are a function of a code point, not
1416# everything that is caselessly equivalent to that code point. For example,
1417# Changes_When_Case_Folded('s') should be false, whereas caselessly it would
1418# be true because 's' and 'S' are equivalent caselessly. However,
1419# traditionally, [:upper:] and [:lower:] are equivalent caselessly, so we
1420# extend that concept to those very few properties that are like this. Each
1421# such property will match the full range caselessly. They are hard-coded in
1422# the program; it's not worth trying to make it general as it's extremely
1423# unlikely that they will ever change.
1424my %caseless_equivalent_to;
1425
99870f4d
KW
1426# These constants names and values were taken from the Unicode standard,
1427# version 5.1, section 3.12. They are used in conjunction with Hangul
6e5a209b
KW
1428# syllables. The '_string' versions are so generated tables can retain the
1429# hex format, which is the more familiar value
1430my $SBase_string = "0xAC00";
1431my $SBase = CORE::hex $SBase_string;
1432my $LBase_string = "0x1100";
1433my $LBase = CORE::hex $LBase_string;
1434my $VBase_string = "0x1161";
1435my $VBase = CORE::hex $VBase_string;
1436my $TBase_string = "0x11A7";
1437my $TBase = CORE::hex $TBase_string;
99870f4d
KW
1438my $SCount = 11172;
1439my $LCount = 19;
1440my $VCount = 21;
1441my $TCount = 28;
1442my $NCount = $VCount * $TCount;
1443
1444# For Hangul syllables; These store the numbers from Jamo.txt in conjunction
1445# with the above published constants.
1446my %Jamo;
1447my %Jamo_L; # Leading consonants
1448my %Jamo_V; # Vowels
1449my %Jamo_T; # Trailing consonants
1450
bb1dd3da
KW
1451# For code points whose name contains its ordinal as a '-ABCD' suffix.
1452# The key is the base name of the code point, and the value is an
1453# array giving all the ranges that use this base name. Each range
1454# is actually a hash giving the 'low' and 'high' values of it.
1455my %names_ending_in_code_point;
1456my %loose_names_ending_in_code_point; # Same as above, but has blanks, dashes
1457 # removed from the names
1458# Inverse mapping. The list of ranges that have these kinds of
1459# names. Each element contains the low, high, and base names in an
1460# anonymous hash.
1461my @code_points_ending_in_code_point;
1462
6b5ab373
KW
1463# To hold Unicode's normalization test suite
1464my @normalization_tests;
1465
bb1dd3da
KW
1466# Boolean: does this Unicode version have the hangul syllables, and are we
1467# writing out a table for them?
1468my $has_hangul_syllables = 0;
1469
1470# Does this Unicode version have code points whose names end in their
1471# respective code points, and are we writing out a table for them? 0 for no;
1472# otherwise points to first property that a table is needed for them, so that
1473# if multiple tables are needed, we don't create duplicates
1474my $needing_code_points_ending_in_code_point = 0;
1475
37e2e78e 1476my @backslash_X_tests; # List of tests read in for testing \X
06ae2722 1477my @SB_tests; # List of tests read in for testing \b{sb}
ae3bb8ea 1478my @WB_tests; # List of tests read in for testing \b{wb}
99870f4d
KW
1479my @unhandled_properties; # Will contain a list of properties found in
1480 # the input that we didn't process.
f86864ac 1481my @match_properties; # Properties that have match tables, to be
99870f4d
KW
1482 # listed in the pod
1483my @map_properties; # Properties that get map files written
1484my @named_sequences; # NamedSequences.txt contents.
1485my %potential_files; # Generated list of all .txt files in the directory
1486 # structure so we can warn if something is being
1487 # ignored.
1488my @files_actually_output; # List of files we generated.
1489my @more_Names; # Some code point names are compound; this is used
1490 # to store the extra components of them.
1491my $MIN_FRACTION_LENGTH = 3; # How many digits of a floating point number at
1492 # the minimum before we consider it equivalent to a
1493 # candidate rational
1494my $MAX_FLOATING_SLOP = 10 ** - $MIN_FRACTION_LENGTH; # And in floating terms
1495
1496# These store references to certain commonly used property objects
3c88a801 1497my $ccc;
99870f4d
KW
1498my $gc;
1499my $perl;
1500my $block;
3e20195b
KW
1501my $perl_charname;
1502my $print;
ac7dbdb6 1503my $All;
bc0c431d 1504my $Assigned; # All assigned characters in this Unicode release
359523e2 1505my $script;
99870f4d
KW
1506
1507# Are there conflicting names because of beginning with 'In_', or 'Is_'
1508my $has_In_conflicts = 0;
1509my $has_Is_conflicts = 0;
1510
1511sub internal_file_to_platform ($) {
1512 # Convert our file paths which have '/' separators to those of the
1513 # platform.
1514
1515 my $file = shift;
1516 return undef unless defined $file;
1517
1518 return File::Spec->join(split '/', $file);
d07a55ed 1519}
5beb625e 1520
99870f4d
KW
1521sub file_exists ($) { # platform independent '-e'. This program internally
1522 # uses slash as a path separator.
1523 my $file = shift;
1524 return 0 if ! defined $file;
1525 return -e internal_file_to_platform($file);
1526}
5beb625e 1527
99870f4d 1528sub objaddr($) {
23e33b60
KW
1529 # Returns the address of the blessed input object.
1530 # It doesn't check for blessedness because that would do a string eval
1531 # every call, and the program is structured so that this is never called
1532 # for a non-blessed object.
99870f4d 1533
23e33b60 1534 no overloading; # If overloaded, numifying below won't work.
99870f4d
KW
1535
1536 # Numifying a ref gives its address.
051df77b 1537 return pack 'J', $_[0];
99870f4d
KW
1538}
1539
558712cf 1540# These are used only if $annotate is true.
c4019d52
KW
1541# The entire range of Unicode characters is examined to populate these
1542# after all the input has been processed. But most can be skipped, as they
1543# have the same descriptive phrases, such as being unassigned
1544my @viacode; # Contains the 1 million character names
1545my @printable; # boolean: And are those characters printable?
1546my @annotate_char_type; # Contains a type of those characters, specifically
1547 # for the purposes of annotation.
1548my $annotate_ranges; # A map of ranges of code points that have the same
98dc9551 1549 # name for the purposes of annotation. They map to the
c4019d52
KW
1550 # upper edge of the range, so that the end point can
1551 # be immediately found. This is used to skip ahead to
1552 # the end of a range, and avoid processing each
1553 # individual code point in it.
1554my $unassigned_sans_noncharacters; # A Range_List of the unassigned
1555 # characters, but excluding those which are
1556 # also noncharacter code points
1557
1558# The annotation types are an extension of the regular range types, though
1559# some of the latter are folded into one. Make the new types negative to
1560# avoid conflicting with the regular types
1561my $SURROGATE_TYPE = -1;
1562my $UNASSIGNED_TYPE = -2;
1563my $PRIVATE_USE_TYPE = -3;
1564my $NONCHARACTER_TYPE = -4;
1565my $CONTROL_TYPE = -5;
2d88a86a
KW
1566my $ABOVE_UNICODE_TYPE = -6;
1567my $UNKNOWN_TYPE = -7; # Used only if there is a bug in this program
c4019d52
KW
1568
1569sub populate_char_info ($) {
558712cf 1570 # Used only with the $annotate option. Populates the arrays with the
c4019d52
KW
1571 # input code point's info that are needed for outputting more detailed
1572 # comments. If calling context wants a return, it is the end point of
1573 # any contiguous range of characters that share essentially the same info
1574
1575 my $i = shift;
1576 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
1577
1578 $viacode[$i] = $perl_charname->value_of($i) || "";
1579
1580 # A character is generally printable if Unicode says it is,
1581 # but below we make sure that most Unicode general category 'C' types
1582 # aren't.
1583 $printable[$i] = $print->contains($i);
1584
1585 $annotate_char_type[$i] = $perl_charname->type_of($i) || 0;
1586
1587 # Only these two regular types are treated specially for annotations
1588 # purposes
1589 $annotate_char_type[$i] = 0 if $annotate_char_type[$i] != $CP_IN_NAME
1590 && $annotate_char_type[$i] != $HANGUL_SYLLABLE;
1591
1592 # Give a generic name to all code points that don't have a real name.
1593 # We output ranges, if applicable, for these. Also calculate the end
1594 # point of the range.
1595 my $end;
1596 if (! $viacode[$i]) {
1d025d66 1597 my $nonchar;
2d88a86a
KW
1598 if ($i > $MAX_UNICODE_CODEPOINT) {
1599 $viacode[$i] = 'Above-Unicode';
1600 $annotate_char_type[$i] = $ABOVE_UNICODE_TYPE;
1601 $printable[$i] = 0;
1602 $end = $MAX_WORKING_CODEPOINT;
1603 }
1604 elsif ($gc-> table('Private_use')->contains($i)) {
c4019d52
KW
1605 $viacode[$i] = 'Private Use';
1606 $annotate_char_type[$i] = $PRIVATE_USE_TYPE;
1607 $printable[$i] = 0;
1608 $end = $gc->table('Private_Use')->containing_range($i)->end;
1609 }
1d025d66
KW
1610 elsif ((defined ($nonchar =
1611 Property::property_ref('Noncharacter_Code_Point'))
1612 && $nonchar->table('Y')->contains($i)))
c4019d52
KW
1613 {
1614 $viacode[$i] = 'Noncharacter';
1615 $annotate_char_type[$i] = $NONCHARACTER_TYPE;
1616 $printable[$i] = 0;
1617 $end = property_ref('Noncharacter_Code_Point')->table('Y')->
1618 containing_range($i)->end;
1619 }
1620 elsif ($gc-> table('Control')->contains($i)) {
c71dea7f 1621 $viacode[$i] = property_ref('Name_Alias')->value_of($i) || 'Control';
c4019d52
KW
1622 $annotate_char_type[$i] = $CONTROL_TYPE;
1623 $printable[$i] = 0;
c4019d52
KW
1624 }
1625 elsif ($gc-> table('Unassigned')->contains($i)) {
c4019d52
KW
1626 $annotate_char_type[$i] = $UNASSIGNED_TYPE;
1627 $printable[$i] = 0;
1d025d66
KW
1628 if ($v_version lt v2.0.0) { # No blocks in earliest releases
1629 $viacode[$i] = 'Unassigned';
1630 $end = $gc-> table('Unassigned')->containing_range($i)->end;
1631 }
1632 else {
1633 $viacode[$i] = 'Unassigned, block=' . $block-> value_of($i);
c4019d52 1634
bf06c733
KW
1635 # Because we name the unassigned by the blocks they are in, it
1636 # can't go past the end of that block, and it also can't go
1637 # past the unassigned range it is in. The special table makes
1638 # sure that the non-characters, which are unassigned, are
1639 # separated out.
1640 $end = min($block->containing_range($i)->end,
1641 $unassigned_sans_noncharacters->
1642 containing_range($i)->end);
1d025d66
KW
1643 }
1644 }
35e2b416 1645 elsif ($perl->table('_Perl_Surrogate')->contains($i)) {
1d025d66
KW
1646 $viacode[$i] = 'Surrogate';
1647 $annotate_char_type[$i] = $SURROGATE_TYPE;
1648 $printable[$i] = 0;
1649 $end = $gc->table('Surrogate')->containing_range($i)->end;
13ca76ff
KW
1650 }
1651 else {
1652 Carp::my_carp_bug("Can't figure out how to annotate "
1653 . sprintf("U+%04X", $i)
1654 . ". Proceeding anyway.");
c4019d52
KW
1655 $viacode[$i] = 'UNKNOWN';
1656 $annotate_char_type[$i] = $UNKNOWN_TYPE;
1657 $printable[$i] = 0;
1658 }
1659 }
1660
1661 # Here, has a name, but if it's one in which the code point number is
1662 # appended to the name, do that.
1663 elsif ($annotate_char_type[$i] == $CP_IN_NAME) {
1664 $viacode[$i] .= sprintf("-%04X", $i);
1665 $end = $perl_charname->containing_range($i)->end;
1666 }
1667
1668 # And here, has a name, but if it's a hangul syllable one, replace it with
1669 # the correct name from the Unicode algorithm
1670 elsif ($annotate_char_type[$i] == $HANGUL_SYLLABLE) {
1671 use integer;
1672 my $SIndex = $i - $SBase;
1673 my $L = $LBase + $SIndex / $NCount;
1674 my $V = $VBase + ($SIndex % $NCount) / $TCount;
1675 my $T = $TBase + $SIndex % $TCount;
1676 $viacode[$i] = "HANGUL SYLLABLE $Jamo{$L}$Jamo{$V}";
1677 $viacode[$i] .= $Jamo{$T} if $T != $TBase;
1678 $end = $perl_charname->containing_range($i)->end;
1679 }
1680
1681 return if ! defined wantarray;
1682 return $i if ! defined $end; # If not a range, return the input
1683
1684 # Save this whole range so can find the end point quickly
1685 $annotate_ranges->add_map($i, $end, $end);
1686
1687 return $end;
1688}
1689
23e33b60
KW
1690# Commented code below should work on Perl 5.8.
1691## This 'require' doesn't necessarily work in miniperl, and even if it does,
1692## the native perl version of it (which is what would operate under miniperl)
1693## is extremely slow, as it does a string eval every call.
7e017d6d 1694#my $has_fast_scalar_util = $^X !~ /miniperl/
23e33b60
KW
1695# && defined eval "require Scalar::Util";
1696#
1697#sub objaddr($) {
1698# # Returns the address of the blessed input object. Uses the XS version if
1699# # available. It doesn't check for blessedness because that would do a
1700# # string eval every call, and the program is structured so that this is
1701# # never called for a non-blessed object.
1702#
1703# return Scalar::Util::refaddr($_[0]) if $has_fast_scalar_util;
1704#
1705# # Check at least that is a ref.
1706# my $pkg = ref($_[0]) or return undef;
1707#
1708# # Change to a fake package to defeat any overloaded stringify
1709# bless $_[0], 'main::Fake';
1710#
1711# # Numifying a ref gives its address.
051df77b 1712# my $addr = pack 'J', $_[0];
23e33b60
KW
1713#
1714# # Return to original class
1715# bless $_[0], $pkg;
1716# return $addr;
1717#}
1718
99870f4d
KW
1719sub max ($$) {
1720 my $a = shift;
1721 my $b = shift;
1722 return $a if $a >= $b;
1723 return $b;
1724}
1725
1726sub min ($$) {
1727 my $a = shift;
1728 my $b = shift;
1729 return $a if $a <= $b;
1730 return $b;
1731}
1732
1733sub clarify_number ($) {
1734 # This returns the input number with underscores inserted every 3 digits
1735 # in large (5 digits or more) numbers. Input must be entirely digits, not
1736 # checked.
1737
1738 my $number = shift;
1739 my $pos = length($number) - 3;
1740 return $number if $pos <= 1;
1741 while ($pos > 0) {
1742 substr($number, $pos, 0) = '_';
1743 $pos -= 3;
5beb625e 1744 }
99870f4d 1745 return $number;
99598c8c
JH
1746}
1747
731cb813
KW
1748sub clarify_code_point_count ($) {
1749 # This is like clarify_number(), but the input is assumed to be a count of
1750 # code points, rather than a generic number.
1751
2d88a86a
KW
1752 my $append = "";
1753
1754 my $number = shift;
1755 if ($number > $MAX_UNICODE_CODEPOINTS) {
1756 $number -= ($MAX_WORKING_CODEPOINTS - $MAX_UNICODE_CODEPOINTS);
1757 return "All above-Unicode code points" if $number == 0;
1758 $append = " + all above-Unicode code points";
1759 }
1760 return clarify_number($number) . $append;
731cb813 1761}
12ac2576 1762
99870f4d 1763package Carp;
7ebf06b3 1764
99870f4d
KW
1765# These routines give a uniform treatment of messages in this program. They
1766# are placed in the Carp package to cause the stack trace to not include them,
1767# although an alternative would be to use another package and set @CARP_NOT
1768# for it.
12ac2576 1769
99870f4d 1770our $Verbose = 1 if main::DEBUG; # Useful info when debugging
12ac2576 1771
99f78760
KW
1772# This is a work-around suggested by Nicholas Clark to fix a problem with Carp
1773# and overload trying to load Scalar:Util under miniperl. See
1774# http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2009-11/msg01057.html
1775undef $overload::VERSION;
1776
99870f4d
KW
1777sub my_carp {
1778 my $message = shift || "";
1779 my $nofold = shift || 0;
7ebf06b3 1780
99870f4d
KW
1781 if ($message) {
1782 $message = main::join_lines($message);
1783 $message =~ s/^$0: *//; # Remove initial program name
1784 $message =~ s/[.;,]+$//; # Remove certain ending punctuation
1785 $message = "\n$0: $message;";
12ac2576 1786
99870f4d
KW
1787 # Fold the message with program name, semi-colon end punctuation
1788 # (which looks good with the message that carp appends to it), and a
1789 # hanging indent for continuation lines.
1790 $message = main::simple_fold($message, "", 4) unless $nofold;
1791 $message =~ s/\n$//; # Remove the trailing nl so what carp
1792 # appends is to the same line
1793 }
12ac2576 1794
99870f4d 1795 return $message if defined wantarray; # If a caller just wants the msg
12ac2576 1796
99870f4d
KW
1797 carp $message;
1798 return;
1799}
7ebf06b3 1800
99870f4d
KW
1801sub my_carp_bug {
1802 # This is called when it is clear that the problem is caused by a bug in
1803 # this program.
7ebf06b3 1804
99870f4d
KW
1805 my $message = shift;
1806 $message =~ s/^$0: *//;
1807 $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");
1808 carp $message;
1809 return;
1810}
7ebf06b3 1811
99870f4d
KW
1812sub carp_too_few_args {
1813 if (@_ != 2) {
1814 my_carp_bug("Wrong number of arguments: to 'carp_too_few_arguments'. No action taken.");
1815 return;
12ac2576 1816 }
7ebf06b3 1817
99870f4d
KW
1818 my $args_ref = shift;
1819 my $count = shift;
7ebf06b3 1820
99870f4d
KW
1821 my_carp_bug("Need at least $count arguments to "
1822 . (caller 1)[3]
1823 . ". Instead got: '"
1824 . join ', ', @$args_ref
1825 . "'. No action taken.");
1826 return;
12ac2576
JP
1827}
1828
99870f4d
KW
1829sub carp_extra_args {
1830 my $args_ref = shift;
1831 my_carp_bug("Too many arguments to 'carp_extra_args': (" . join(', ', @_) . "); Extras ignored.") if @_;
12ac2576 1832
99870f4d
KW
1833 unless (ref $args_ref) {
1834 my_carp_bug("Argument to 'carp_extra_args' ($args_ref) must be a ref. Not checking arguments.");
1835 return;
1836 }
1837 my ($package, $file, $line) = caller;
1838 my $subroutine = (caller 1)[3];
cf25bb62 1839
99870f4d
KW
1840 my $list;
1841 if (ref $args_ref eq 'HASH') {
1842 foreach my $key (keys %$args_ref) {
1843 $args_ref->{$key} = $UNDEF unless defined $args_ref->{$key};
cf25bb62 1844 }
99870f4d 1845 $list = join ', ', each %{$args_ref};
cf25bb62 1846 }
99870f4d
KW
1847 elsif (ref $args_ref eq 'ARRAY') {
1848 foreach my $arg (@$args_ref) {
1849 $arg = $UNDEF unless defined $arg;
1850 }
1851 $list = join ', ', @$args_ref;
1852 }
1853 else {
1854 my_carp_bug("Can't cope with ref "
1855 . ref($args_ref)
1856 . " . argument to 'carp_extra_args'. Not checking arguments.");
1857 return;
1858 }
1859
1860 my_carp_bug("Unrecognized parameters in options: '$list' to $subroutine. Skipped.");
1861 return;
d73e5302
JH
1862}
1863
99870f4d
KW
1864package main;
1865
1866{ # Closure
1867
1868 # This program uses the inside-out method for objects, as recommended in
1b6b3fa9
KW
1869 # "Perl Best Practices". (This is the best solution still, since this has
1870 # to run under miniperl.) This closure aids in generating those. There
99870f4d
KW
1871 # are two routines. setup_package() is called once per package to set
1872 # things up, and then set_access() is called for each hash representing a
1873 # field in the object. These routines arrange for the object to be
1874 # properly destroyed when no longer used, and for standard accessor
1875 # functions to be generated. If you need more complex accessors, just
1876 # write your own and leave those accesses out of the call to set_access().
1877 # More details below.
1878
1879 my %constructor_fields; # fields that are to be used in constructors; see
1880 # below
1881
1882 # The values of this hash will be the package names as keys to other
1883 # hashes containing the name of each field in the package as keys, and
1884 # references to their respective hashes as values.
1885 my %package_fields;
1886
1887 sub setup_package {
1888 # Sets up the package, creating standard DESTROY and dump methods
1889 # (unless already defined). The dump method is used in debugging by
1890 # simple_dumper().
1891 # The optional parameters are:
1892 # a) a reference to a hash, that gets populated by later
1893 # set_access() calls with one of the accesses being
1894 # 'constructor'. The caller can then refer to this, but it is
1895 # not otherwise used by these two routines.
1896 # b) a reference to a callback routine to call during destruction
1897 # of the object, before any fields are actually destroyed
1898
1899 my %args = @_;
1900 my $constructor_ref = delete $args{'Constructor_Fields'};
1901 my $destroy_callback = delete $args{'Destroy_Callback'};
1902 Carp::carp_extra_args(\@_) if main::DEBUG && %args;
1903
1904 my %fields;
1905 my $package = (caller)[0];
1906
1907 $package_fields{$package} = \%fields;
1908 $constructor_fields{$package} = $constructor_ref;
1909
1910 unless ($package->can('DESTROY')) {
1911 my $destroy_name = "${package}::DESTROY";
1912 no strict "refs";
1913
1914 # Use typeglob to give the anonymous subroutine the name we want
1915 *$destroy_name = sub {
1916 my $self = shift;
ffe43484 1917 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
1918
1919 $self->$destroy_callback if $destroy_callback;
1920 foreach my $field (keys %{$package_fields{$package}}) {
1921 #print STDERR __LINE__, ": Destroying ", ref $self, " ", sprintf("%04X", $addr), ": ", $field, "\n";
1922 delete $package_fields{$package}{$field}{$addr};
1923 }
1924 return;
1925 }
1926 }
1927
1928 unless ($package->can('dump')) {
1929 my $dump_name = "${package}::dump";
1930 no strict "refs";
1931 *$dump_name = sub {
1932 my $self = shift;
1933 return dump_inside_out($self, $package_fields{$package}, @_);
1934 }
1935 }
1936 return;
1937 }
1938
1939 sub set_access {
1940 # Arrange for the input field to be garbage collected when no longer
1941 # needed. Also, creates standard accessor functions for the field
1942 # based on the optional parameters-- none if none of these parameters:
1943 # 'addable' creates an 'add_NAME()' accessor function.
1944 # 'readable' or 'readable_array' creates a 'NAME()' accessor
1945 # function.
1946 # 'settable' creates a 'set_NAME()' accessor function.
1947 # 'constructor' doesn't create an accessor function, but adds the
1948 # field to the hash that was previously passed to
1949 # setup_package();
1950 # Any of the accesses can be abbreviated down, so that 'a', 'ad',
1951 # 'add' etc. all mean 'addable'.
1952 # The read accessor function will work on both array and scalar
1953 # values. If another accessor in the parameter list is 'a', the read
1954 # access assumes an array. You can also force it to be array access
1955 # by specifying 'readable_array' instead of 'readable'
1956 #
1957 # A sort-of 'protected' access can be set-up by preceding the addable,
1958 # readable or settable with some initial portion of 'protected_' (but,
1959 # the underscore is required), like 'p_a', 'pro_set', etc. The
1960 # "protection" is only by convention. All that happens is that the
1961 # accessor functions' names begin with an underscore. So instead of
1962 # calling set_foo, the call is _set_foo. (Real protection could be
c1739a4a 1963 # accomplished by having a new subroutine, end_package, called at the
99870f4d
KW
1964 # end of each package, and then storing the __LINE__ ranges and
1965 # checking them on every accessor. But that is way overkill.)
1966
1967 # We create anonymous subroutines as the accessors and then use
1968 # typeglobs to assign them to the proper package and name
1969
1970 my $name = shift; # Name of the field
1971 my $field = shift; # Reference to the inside-out hash containing the
1972 # field
1973
1974 my $package = (caller)[0];
1975
1976 if (! exists $package_fields{$package}) {
1977 croak "$0: Must call 'setup_package' before 'set_access'";
1978 }
d73e5302 1979
99870f4d
KW
1980 # Stash the field so DESTROY can get it.
1981 $package_fields{$package}{$name} = $field;
cf25bb62 1982
99870f4d
KW
1983 # Remaining arguments are the accessors. For each...
1984 foreach my $access (@_) {
1985 my $access = lc $access;
cf25bb62 1986
99870f4d 1987 my $protected = "";
cf25bb62 1988
99870f4d
KW
1989 # Match the input as far as it goes.
1990 if ($access =~ /^(p[^_]*)_/) {
1991 $protected = $1;
1992 if (substr('protected_', 0, length $protected)
1993 eq $protected)
1994 {
1995
1996 # Add 1 for the underscore not included in $protected
1997 $access = substr($access, length($protected) + 1);
1998 $protected = '_';
1999 }
2000 else {
2001 $protected = "";
2002 }
2003 }
2004
2005 if (substr('addable', 0, length $access) eq $access) {
2006 my $subname = "${package}::${protected}add_$name";
2007 no strict "refs";
2008
2009 # add_ accessor. Don't add if already there, which we
2010 # determine using 'eq' for scalars and '==' otherwise.
2011 *$subname = sub {
2012 use strict "refs";
2013 return Carp::carp_too_few_args(\@_, 2) if main::DEBUG && @_ < 2;
2014 my $self = shift;
2015 my $value = shift;
ffe43484 2016 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
2017 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
2018 if (ref $value) {
f998e60c 2019 return if grep { $value == $_ } @{$field->{$addr}};
99870f4d
KW
2020 }
2021 else {
f998e60c 2022 return if grep { $value eq $_ } @{$field->{$addr}};
99870f4d 2023 }
f998e60c 2024 push @{$field->{$addr}}, $value;
99870f4d
KW
2025 return;
2026 }
2027 }
2028 elsif (substr('constructor', 0, length $access) eq $access) {
2029 if ($protected) {
2030 Carp::my_carp_bug("Can't set-up 'protected' constructors")
2031 }
2032 else {
2033 $constructor_fields{$package}{$name} = $field;
2034 }
2035 }
2036 elsif (substr('readable_array', 0, length $access) eq $access) {
2037
2038 # Here has read access. If one of the other parameters for
2039 # access is array, or this one specifies array (by being more
2040 # than just 'readable_'), then create a subroutine that
2041 # assumes the data is an array. Otherwise just a scalar
2042 my $subname = "${package}::${protected}$name";
2043 if (grep { /^a/i } @_
2044 or length($access) > length('readable_'))
2045 {
2046 no strict "refs";
2047 *$subname = sub {
2048 use strict "refs";
23e33b60 2049 Carp::carp_extra_args(\@_) if main::DEBUG && @_ > 1;
ffe43484 2050 my $addr = do { no overloading; pack 'J', $_[0]; };
99870f4d
KW
2051 if (ref $field->{$addr} ne 'ARRAY') {
2052 my $type = ref $field->{$addr};
2053 $type = 'scalar' unless $type;
2054 Carp::my_carp_bug("Trying to read $name as an array when it is a $type. Big problems.");
2055 return;
2056 }
2057 return scalar @{$field->{$addr}} unless wantarray;
2058
2059 # Make a copy; had problems with caller modifying the
2060 # original otherwise
2061 my @return = @{$field->{$addr}};
2062 return @return;
2063 }
2064 }
2065 else {
2066
2067 # Here not an array value, a simpler function.
2068 no strict "refs";
2069 *$subname = sub {
2070 use strict "refs";
23e33b60 2071 Carp::carp_extra_args(\@_) if main::DEBUG && @_ > 1;
f998e60c 2072 no overloading;
051df77b 2073 return $field->{pack 'J', $_[0]};
99870f4d
KW
2074 }
2075 }
2076 }
2077 elsif (substr('settable', 0, length $access) eq $access) {
2078 my $subname = "${package}::${protected}set_$name";
2079 no strict "refs";
2080 *$subname = sub {
2081 use strict "refs";
23e33b60
KW
2082 if (main::DEBUG) {
2083 return Carp::carp_too_few_args(\@_, 2) if @_ < 2;
2084 Carp::carp_extra_args(\@_) if @_ > 2;
2085 }
2086 # $self is $_[0]; $value is $_[1]
f998e60c 2087 no overloading;
051df77b 2088 $field->{pack 'J', $_[0]} = $_[1];
99870f4d
KW
2089 return;
2090 }
2091 }
2092 else {
2093 Carp::my_carp_bug("Unknown accessor type $access. No accessor set.");
2094 }
cf25bb62 2095 }
99870f4d 2096 return;
cf25bb62 2097 }
99870f4d
KW
2098}
2099
2100package Input_file;
2101
2102# All input files use this object, which stores various attributes about them,
2103# and provides for convenient, uniform handling. The run method wraps the
2104# processing. It handles all the bookkeeping of opening, reading, and closing
2105# the file, returning only significant input lines.
2106#
2107# Each object gets a handler which processes the body of the file, and is
74cd47d0
KW
2108# called by run(). All character property files must use the generic,
2109# default handler, which has code scrubbed to handle things you might not
2110# expect, including automatic EBCDIC handling. For files that don't deal with
2111# mapping code points to a property value, such as test files,
2112# PropertyAliases, PropValueAliases, and named sequences, you can override the
2113# handler to be a custom one. Such a handler should basically be a
2114# while(next_line()) {...} loop.
99870f4d
KW
2115#
2116# You can also set up handlers to
537124e4 2117# 1) call before the first line is read, for pre processing
83b68635
KW
2118# 2) call to adjust each line of the input before the main handler gets
2119# them. This can be automatically generated, if appropriately simple
2120# enough, by specifiying a Properties parameter in the constructor.
99870f4d 2121# 3) call upon EOF before the main handler exits its loop
537124e4 2122# 4) call at the end, for post processing
99870f4d
KW
2123#
2124# $_ is used to store the input line, and is to be filtered by the
2125# each_line_handler()s. So, if the format of the line is not in the desired
2126# format for the main handler, these are used to do that adjusting. They can
2127# be stacked (by enclosing them in an [ anonymous array ] in the constructor,
2128# so the $_ output of one is used as the input to the next. None of the other
2129# handlers are stackable, but could easily be changed to be so.
2130#
2131# Most of the handlers can call insert_lines() or insert_adjusted_lines()
2132# which insert the parameters as lines to be processed before the next input
2133# file line is read. This allows the EOF handler to flush buffers, for
2134# example. The difference between the two routines is that the lines inserted
2135# by insert_lines() are subjected to the each_line_handler()s. (So if you
9be22ca5
KW
2136# called it from such a handler, you would get infinite recursion without some
2137# mechanism to prevent that.) Lines inserted by insert_adjusted_lines() go
2138# directly to the main handler without any adjustments. If the
2139# post-processing handler calls any of these, there will be no effect. Some
2140# error checking for these conditions could be added, but it hasn't been done.
99870f4d
KW
2141#
2142# carp_bad_line() should be called to warn of bad input lines, which clears $_
2143# to prevent further processing of the line. This routine will output the
2144# message as a warning once, and then keep a count of the lines that have the
2145# same message, and output that count at the end of the file's processing.
2146# This keeps the number of messages down to a manageable amount.
2147#
2148# get_missings() should be called to retrieve any @missing input lines.
2149# Messages will be raised if this isn't done if the options aren't to ignore
2150# missings.
2151
2152sub trace { return main::trace(@_); }
2153
99870f4d
KW
2154{ # Closure
2155 # Keep track of fields that are to be put into the constructor.
2156 my %constructor_fields;
2157
2158 main::setup_package(Constructor_Fields => \%constructor_fields);
2159
2160 my %file; # Input file name, required
2161 main::set_access('file', \%file, qw{ c r });
2162
2163 my %first_released; # Unicode version file was first released in, required
2164 main::set_access('first_released', \%first_released, qw{ c r });
2165
2166 my %handler; # Subroutine to process the input file, defaults to
2167 # 'process_generic_property_file'
2168 main::set_access('handler', \%handler, qw{ c });
2169
2170 my %property;
2171 # name of property this file is for. defaults to none, meaning not
2172 # applicable, or is otherwise determinable, for example, from each line.
696609bf 2173 main::set_access('property', \%property, qw{ c r });
99870f4d
KW
2174
2175 my %optional;
2176 # If this is true, the file is optional. If not present, no warning is
2177 # output. If it is present, the string given by this parameter is
2178 # evaluated, and if false the file is not processed.
2179 main::set_access('optional', \%optional, 'c', 'r');
2180
2181 my %non_skip;
2182 # This is used for debugging, to skip processing of all but a few input
2183 # files. Add 'non_skip => 1' to the constructor for those files you want
2184 # processed when you set the $debug_skip global.
2185 main::set_access('non_skip', \%non_skip, 'c');
2186
37e2e78e 2187 my %skip;
09ca89ce
KW
2188 # This is used to skip processing of this input file semi-permanently,
2189 # when it evaluates to true. The value should be the reason the file is
2190 # being skipped. It is used for files that we aren't planning to process
2191 # anytime soon, but want to allow to be in the directory and not raise a
2192 # message that we are not handling. Mostly for test files. This is in
2193 # contrast to the non_skip element, which is supposed to be used very
2194 # temporarily for debugging. Sets 'optional' to 1. Also, files that we
2195 # pretty much will never look at can be placed in the global
1fec9f60 2196 # %ignored_files instead. Ones used here will be added to %skipped files
37e2e78e
KW
2197 main::set_access('skip', \%skip, 'c');
2198
99870f4d
KW
2199 my %each_line_handler;
2200 # list of subroutines to look at and filter each non-comment line in the
2201 # file. defaults to none. The subroutines are called in order, each is
2202 # to adjust $_ for the next one, and the final one adjusts it for
2203 # 'handler'
2204 main::set_access('each_line_handler', \%each_line_handler, 'c');
2205
83b68635
KW
2206 my %properties; # Optional ordered list of the properties that occur in each
2207 # meaningful line of the input file. If present, an appropriate
2208 # each_line_handler() is automatically generated and pushed onto the stack
2209 # of such handlers. This is useful when a file contains multiple
2210 # proerties per line, but no other special considerations are necessary.
2211 # The special value "<ignored>" means to discard the corresponding input
2212 # field.
2213 # Any @missing lines in the file should also match this syntax; no such
2214 # files exist as of 6.3. But if it happens in a future release, the code
2215 # could be expanded to properly parse them.
2216 main::set_access('properties', \%properties, qw{ c r });
2217
99870f4d
KW
2218 my %has_missings_defaults;
2219 # ? Are there lines in the file giving default values for code points
2220 # missing from it?. Defaults to NO_DEFAULTS. Otherwise NOT_IGNORED is
2221 # the norm, but IGNORED means it has such lines, but the handler doesn't
2222 # use them. Having these three states allows us to catch changes to the
83b68635
KW
2223 # UCD that this program should track. XXX This could be expanded to
2224 # specify the syntax for such lines, like %properties above.
99870f4d
KW
2225 main::set_access('has_missings_defaults',
2226 \%has_missings_defaults, qw{ c r });
2227
2228 my %pre_handler;
2229 # Subroutine to call before doing anything else in the file. If undef, no
2230 # such handler is called.
2231 main::set_access('pre_handler', \%pre_handler, qw{ c });
2232
2233 my %eof_handler;
2234 # Subroutine to call upon getting an EOF on the input file, but before
2235 # that is returned to the main handler. This is to allow buffers to be
2236 # flushed. The handler is expected to call insert_lines() or
2237 # insert_adjusted() with the buffered material
2238 main::set_access('eof_handler', \%eof_handler, qw{ c r });
2239
2240 my %post_handler;
2241 # Subroutine to call after all the lines of the file are read in and
9be22ca5
KW
2242 # processed. If undef, no such handler is called. Note that this cannot
2243 # add lines to be processed; instead use eof_handler
99870f4d
KW
2244 main::set_access('post_handler', \%post_handler, qw{ c });
2245
2246 my %progress_message;
2247 # Message to print to display progress in lieu of the standard one
2248 main::set_access('progress_message', \%progress_message, qw{ c });
2249
2250 my %handle;
2251 # cache open file handle, internal. Is undef if file hasn't been
2252 # processed at all, empty if has;
2253 main::set_access('handle', \%handle);
2254
2255 my %added_lines;
2256 # cache of lines added virtually to the file, internal
2257 main::set_access('added_lines', \%added_lines);
2258
74cd47d0
KW
2259 my %remapped_lines;
2260 # cache of lines added virtually to the file, internal
2261 main::set_access('remapped_lines', \%remapped_lines);
2262
99870f4d
KW
2263 my %errors;
2264 # cache of errors found, internal
2265 main::set_access('errors', \%errors);
2266
2267 my %missings;
2268 # storage of '@missing' defaults lines
2269 main::set_access('missings', \%missings);
2270
74cd47d0
KW
2271 sub _next_line;
2272 sub _next_line_with_remapped_range;
2273
99870f4d
KW
2274 sub new {
2275 my $class = shift;
2276
2277 my $self = bless \do{ my $anonymous_scalar }, $class;
ffe43484 2278 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
2279
2280 # Set defaults
2281 $handler{$addr} = \&main::process_generic_property_file;
2282 $non_skip{$addr} = 0;
37e2e78e 2283 $skip{$addr} = 0;
99870f4d
KW
2284 $has_missings_defaults{$addr} = $NO_DEFAULTS;
2285 $handle{$addr} = undef;
2286 $added_lines{$addr} = [ ];
74cd47d0 2287 $remapped_lines{$addr} = [ ];
99870f4d
KW
2288 $each_line_handler{$addr} = [ ];
2289 $errors{$addr} = { };
2290 $missings{$addr} = [ ];
2291
2292 # Two positional parameters.
99f78760 2293 return Carp::carp_too_few_args(\@_, 2) if main::DEBUG && @_ < 2;
99870f4d
KW
2294 $file{$addr} = main::internal_file_to_platform(shift);
2295 $first_released{$addr} = shift;
2296
71bd4c0b
KW
2297 undef $file{$addr} if $first_released{$addr} gt $v_version;
2298
99870f4d
KW
2299 # The rest of the arguments are key => value pairs
2300 # %constructor_fields has been set up earlier to list all possible
2301 # ones. Either set or push, depending on how the default has been set
2302 # up just above.
2303 my %args = @_;
2304 foreach my $key (keys %args) {
2305 my $argument = $args{$key};
2306
2307 # Note that the fields are the lower case of the constructor keys
2308 my $hash = $constructor_fields{lc $key};
2309 if (! defined $hash) {
2310 Carp::my_carp_bug("Unrecognized parameters '$key => $argument' to new() for $self. Skipped");
2311 next;
2312 }
2313 if (ref $hash->{$addr} eq 'ARRAY') {
2314 if (ref $argument eq 'ARRAY') {
2315 foreach my $argument (@{$argument}) {
2316 next if ! defined $argument;
2317 push @{$hash->{$addr}}, $argument;
2318 }
2319 }
2320 else {
2321 push @{$hash->{$addr}}, $argument if defined $argument;
2322 }
2323 }
2324 else {
2325 $hash->{$addr} = $argument;
2326 }
2327 delete $args{$key};
2328 };
2329
2330 # If the file has a property for it, it means that the property is not
2331 # listed in the file's entries. So add a handler to the list of line
2332 # handlers to insert the property name into the lines, to provide a
2333 # uniform interface to the final processing subroutine.
2334 # the final code doesn't have to worry about that.
2335 if ($property{$addr}) {
2336 push @{$each_line_handler{$addr}}, \&_insert_property_into_line;
2337 }
2338
2339 if ($non_skip{$addr} && ! $debug_skip && $verbosity) {
2340 print "Warning: " . __PACKAGE__ . " constructor for $file{$addr} has useless 'non_skip' in it\n";
a3a8c5f0 2341 }
99870f4d 2342
09ca89ce
KW
2343 # If skipping, set to optional, and add to list of ignored files,
2344 # including its reason
2345 if ($skip{$addr}) {
2346 $optional{$addr} = 1;
71bd4c0b 2347 $skipped_files{$file{$addr}} = $skip{$addr} if $file{$addr};
09ca89ce 2348 }
83b68635
KW
2349 elsif ($properties{$addr}) {
2350
9be22ca5
KW
2351 # Similarly, there may be more than one property represented on
2352 # each line, with no clue but the constructor input what those
2353 # might be. Add a handler for each line in the input so that it
2354 # creates a separate input line for each property in those input
2355 # lines, thus making them suitable to handle generically.
83b68635
KW
2356
2357 push @{$each_line_handler{$addr}},
2358 sub {
2359 my $file = shift;
2360 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
2361
2362 my @fields = split /\s*;\s*/, $_, -1;
2363
2364 if (@fields - 1 > @{$properties{$addr}}) {
2365 $file->carp_bad_line('Extra fields');
2366 $_ = "";
2367 return;
2368 }
2369 my $range = shift @fields; # 0th element is always the
2370 # range
2371
2372 # The next fields in the input line correspond
2373 # respectively to the stored properties.
2374 for my $i (0 .. @{$properties{$addr}} - 1) {
2375 my $property_name = $properties{$addr}[$i];
2376 next if $property_name eq '<ignored>';
2377 $file->insert_adjusted_lines(
2378 "$range; $property_name; $fields[$i]");
2379 }
2380 $_ = "";
2381
2382 return;
2383 };
2384 }
37e2e78e 2385
9be22ca5 2386 { # On non-ascii platforms, we use a special pre-handler
74cd47d0
KW
2387 no strict;
2388 no warnings 'once';
2389 *next_line = (main::NON_ASCII_PLATFORM)
2390 ? *_next_line_with_remapped_range
2391 : *_next_line;
2392 }
2393
99870f4d 2394 return $self;
d73e5302
JH
2395 }
2396
cf25bb62 2397
99870f4d
KW
2398 use overload
2399 fallback => 0,
2400 qw("") => "_operator_stringify",
2401 "." => \&main::_operator_dot,
1285127e 2402 ".=" => \&main::_operator_dot_equal,
99870f4d 2403 ;
cf25bb62 2404
99870f4d
KW
2405 sub _operator_stringify {
2406 my $self = shift;
cf25bb62 2407
99870f4d 2408 return __PACKAGE__ . " object for " . $self->file;
d73e5302 2409 }
d73e5302 2410
99870f4d
KW
2411 # flag to make sure extracted files are processed early
2412 my $seen_non_extracted_non_age = 0;
d73e5302 2413
99870f4d
KW
2414 sub run {
2415 # Process the input object $self. This opens and closes the file and
2416 # calls all the handlers for it. Currently, this can only be called
2417 # once per file, as it destroy's the EOF handler
d73e5302 2418
99870f4d
KW
2419 my $self = shift;
2420 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
b6922eda 2421
ffe43484 2422 my $addr = do { no overloading; pack 'J', $self; };
b6922eda 2423
99870f4d 2424 my $file = $file{$addr};
d73e5302 2425
99870f4d
KW
2426 # Don't process if not expecting this file (because released later
2427 # than this Unicode version), and isn't there. This means if someone
2428 # copies it into an earlier version's directory, we will go ahead and
2429 # process it.
71bd4c0b
KW
2430 return if $first_released{$addr} gt $v_version
2431 && (! defined $file || ! -e $file);
99870f4d
KW
2432
2433 # If in debugging mode and this file doesn't have the non-skip
2434 # flag set, and isn't one of the critical files, skip it.
2435 if ($debug_skip
2436 && $first_released{$addr} ne v0
2437 && ! $non_skip{$addr})
2438 {
2439 print "Skipping $file in debugging\n" if $verbosity;
2440 return;
2441 }
2442
2443 # File could be optional
37e2e78e 2444 if ($optional{$addr}) {
99870f4d
KW
2445 return unless -e $file;
2446 my $result = eval $optional{$addr};
2447 if (! defined $result) {
2448 Carp::my_carp_bug("Got '$@' when tried to eval $optional{$addr}. $file Skipped.");
2449 return;
2450 }
2451 if (! $result) {
2452 if ($verbosity) {
2453 print STDERR "Skipping processing input file '$file' because '$optional{$addr}' is not true\n";
2454 }
2455 return;
2456 }
2457 }
2458
2459 if (! defined $file || ! -e $file) {
2460
2461 # If the file doesn't exist, see if have internal data for it
2462 # (based on first_released being 0).
2463 if ($first_released{$addr} eq v0) {
2464 $handle{$addr} = 'pretend_is_open';
2465 }
2466 else {
2467 if (! $optional{$addr} # File could be optional
2468 && $v_version ge $first_released{$addr})
2469 {
f71c7390 2470 print STDERR "Skipping processing input file '$file' because not found\n";
99870f4d
KW
2471 }
2472 return;
2473 }
2474 }
2475 else {
99870f4d 2476 if ($seen_non_extracted_non_age) {
9be22ca5
KW
2477 if ($file =~ /$EXTRACTED/i) # Some platforms may change the
2478 # case of the file's name
2479 {
1675ea0d 2480 Carp::my_carp_bug(main::join_lines(<<END
99f78760 2481$file should be processed just after the 'Prop...Alias' files, and before
99870f4d
KW
2482anything not in the $EXTRACTED_DIR directory. Proceeding, but the results may
2483have subtle problems
2484END
2485 ));
2486 }
2487 }
2488 elsif ($EXTRACTED_DIR
2489 && $first_released{$addr} ne v0
517956bf
CB
2490 && $file !~ /$EXTRACTED/i
2491 && lc($file) ne 'dage.txt')
99870f4d
KW
2492 {
2493 # We don't set this (by the 'if' above) if we have no
2494 # extracted directory, so if running on an early version,
2495 # this test won't work. Not worth worrying about.
2496 $seen_non_extracted_non_age = 1;
2497 }
2498
9be22ca5 2499 # Mark the file as having being processed, and warn if it
99870f4d
KW
2500 # isn't a file we are expecting. As we process the files,
2501 # they are deleted from the hash, so any that remain at the
2502 # end of the program are files that we didn't process.
517956bf 2503 my $fkey = File::Spec->rel2abs($file);
faf3cf6b
KW
2504 my $expecting = delete $potential_files{lc($fkey)};
2505
678f13d5
KW
2506 Carp::my_carp("Was not expecting '$file'.") if
2507 ! $expecting
99870f4d
KW
2508 && ! defined $handle{$addr};
2509
37e2e78e
KW
2510 # Having deleted from expected files, we can quit if not to do
2511 # anything. Don't print progress unless really want verbosity
2512 if ($skip{$addr}) {
2513 print "Skipping $file.\n" if $verbosity >= $VERBOSE;
2514 return;
2515 }
2516
9be22ca5
KW
2517 # Here, we are going to process the file. Open it, converting the
2518 # slashes used in this program into the proper form for the OS
99870f4d
KW
2519 my $file_handle;
2520 if (not open $file_handle, "<", $file) {
2521 Carp::my_carp("Can't open $file. Skipping: $!");
d3664eb9 2522 return;
99870f4d
KW
2523 }
2524 $handle{$addr} = $file_handle; # Cache the open file handle
9e65c3f4 2525
96f226dc 2526 if ($v_version ge v3.2.0 && lc($file) ne 'unicodedata.txt') {
9be22ca5
KW
2527
2528 # UnicodeData.txt has no version marker; the others started
2529 # getting it in 3.2. Unihan files have the version somewhere
2530 # in the first comment block; the other files have it as the
2531 # very first line
96f226dc 2532 if ($file !~ /^Unihan/i) {
cafe9cf0
KW
2533 $_ = <$file_handle>;
2534 if ($_ !~ / - $string_version \. /x) {
2535 chomp;
2536 $_ =~ s/^#\s*//;
2537 die Carp::my_carp("File '$file' is version '$_'. It should be version $string_version");
2538 }
96f226dc
KW
2539 }
2540 else {
2541 while (<$file_handle>) {
2542 if ($_ !~ /^#/) {
2543 Carp::my_carp_bug("Could not find the expected version info in file '$file'");
2544 last;
2545 }
2546 chomp;
2547 $_ =~ s/^#\s*//;
2548 next if $_ !~ / version: /x;
2549 last if $_ =~ /$string_version/;
2550 die Carp::my_carp("File '$file' is '$_'. It should be version $string_version");
2551 }
2552 }
9e65c3f4 2553 }
99870f4d
KW
2554 }
2555
2556 if ($verbosity >= $PROGRESS) {
2557 if ($progress_message{$addr}) {
2558 print "$progress_message{$addr}\n";
2559 }
2560 else {
2561 # If using a virtual file, say so.
2562 print "Processing ", (-e $file)
2563 ? $file
2564 : "substitute $file",
2565 "\n";
2566 }
2567 }
2568
2569
2570 # Call any special handler for before the file.
2571 &{$pre_handler{$addr}}($self) if $pre_handler{$addr};
2572
2573 # Then the main handler
2574 &{$handler{$addr}}($self);
2575
2576 # Then any special post-file handler.
2577 &{$post_handler{$addr}}($self) if $post_handler{$addr};
2578
2579 # If any errors have been accumulated, output the counts (as the first
2580 # error message in each class was output when it was encountered).
2581 if ($errors{$addr}) {
2582 my $total = 0;
2583 my $types = 0;
2584 foreach my $error (keys %{$errors{$addr}}) {
2585 $total += $errors{$addr}->{$error};
2586 delete $errors{$addr}->{$error};
2587 $types++;
2588 }
2589 if ($total > 1) {
2590 my $message
2591 = "A total of $total lines had errors in $file. ";
2592
2593 $message .= ($types == 1)
2594 ? '(Only the first one was displayed.)'
2595 : '(Only the first of each type was displayed.)';
2596 Carp::my_carp($message);
2597 }
2598 }
2599
2600 if (@{$missings{$addr}}) {
2601 Carp::my_carp_bug("Handler for $file didn't look at all the \@missing lines. Generated tables likely are wrong");
2602 }
2603
2604 # If a real file handle, close it.
2605 close $handle{$addr} or Carp::my_carp("Can't close $file: $!") if
2606 ref $handle{$addr};
2607 $handle{$addr} = ""; # Uses empty to indicate that has already seen
2608 # the file, as opposed to undef
2609 return;
2610 }
2611
74cd47d0 2612 sub _next_line {
99870f4d
KW
2613 # Sets $_ to be the next logical input line, if any. Returns non-zero
2614 # if such a line exists. 'logical' means that any lines that have
2615 # been added via insert_lines() will be returned in $_ before the file
2616 # is read again.
2617
2618 my $self = shift;
2619 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
2620
ffe43484 2621 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
2622
2623 # Here the file is open (or if the handle is not a ref, is an open
2624 # 'virtual' file). Get the next line; any inserted lines get priority
2625 # over the file itself.
2626 my $adjusted;
2627
2628 LINE:
2629 while (1) { # Loop until find non-comment, non-empty line
2630 #local $to_trace = 1 if main::DEBUG;
2631 my $inserted_ref = shift @{$added_lines{$addr}};
2632 if (defined $inserted_ref) {
2633 ($adjusted, $_) = @{$inserted_ref};
2634 trace $adjusted, $_ if main::DEBUG && $to_trace;
2635 return 1 if $adjusted;
2636 }
2637 else {
2638 last if ! ref $handle{$addr}; # Don't read unless is real file
2639 last if ! defined ($_ = readline $handle{$addr});
2640 }
2641 chomp;
2642 trace $_ if main::DEBUG && $to_trace;
2643
2644 # See if this line is the comment line that defines what property
2645 # value that code points that are not listed in the file should
2646 # have. The format or existence of these lines is not guaranteed
2647 # by Unicode since they are comments, but the documentation says
2648 # that this was added for machine-readability, so probably won't
2649 # change. This works starting in Unicode Version 5.0. They look
2650 # like:
2651 #
2652 # @missing: 0000..10FFFF; Not_Reordered
2653 # @missing: 0000..10FFFF; Decomposition_Mapping; <code point>
2654 # @missing: 0000..10FFFF; ; NaN
2655 #
2656 # Save the line for a later get_missings() call.
2657 if (/$missing_defaults_prefix/) {
2658 if ($has_missings_defaults{$addr} == $NO_DEFAULTS) {
2659 $self->carp_bad_line("Unexpected \@missing line. Assuming no missing entries");
2660 }
2661 elsif ($has_missings_defaults{$addr} == $NOT_IGNORED) {
2662 my @defaults = split /\s* ; \s*/x, $_;
2663
2664 # The first field is the @missing, which ends in a
2665 # semi-colon, so can safely shift.
2666 shift @defaults;
2667
2668 # Some of these lines may have empty field placeholders
2669 # which get in the way. An example is:
2670 # @missing: 0000..10FFFF; ; NaN
2671 # Remove them. Process starting from the top so the
2672 # splice doesn't affect things still to be looked at.
2673 for (my $i = @defaults - 1; $i >= 0; $i--) {
2674 next if $defaults[$i] ne "";
2675 splice @defaults, $i, 1;
2676 }
2677
2678 # What's left should be just the property (maybe) and the
2679 # default. Having only one element means it doesn't have
2680 # the property.
2681 my $default;
2682 my $property;
2683 if (@defaults >= 1) {
2684 if (@defaults == 1) {
2685 $default = $defaults[0];
2686 }
2687 else {
2688 $property = $defaults[0];
2689 $default = $defaults[1];
2690 }
2691 }
2692
2693 if (@defaults < 1
2694 || @defaults > 2
2695 || ($default =~ /^</
2696 && $default !~ /^<code *point>$/i
09f8d0ac
KW
2697 && $default !~ /^<none>$/i
2698 && $default !~ /^<script>$/i))
99870f4d
KW
2699 {
2700 $self->carp_bad_line("Unrecognized \@missing line: $_. Assuming no missing entries");
2701 }
2702 else {
2703
2704 # If the property is missing from the line, it should
2705 # be the one for the whole file
2706 $property = $property{$addr} if ! defined $property;
2707
2708 # Change <none> to the null string, which is what it
2709 # really means. If the default is the code point
2710 # itself, set it to <code point>, which is what
2711 # Unicode uses (but sometimes they've forgotten the
2712 # space)
2713 if ($default =~ /^<none>$/i) {
2714 $default = "";
2715 }
2716 elsif ($default =~ /^<code *point>$/i) {
2717 $default = $CODE_POINT;
2718 }
09f8d0ac
KW
2719 elsif ($default =~ /^<script>$/i) {
2720
2721 # Special case this one. Currently is from
2722 # ScriptExtensions.txt, and means for all unlisted
2723 # code points, use their Script property values.
2724 # For the code points not listed in that file, the
2725 # default value is 'Unknown'.
2726 $default = "Unknown";
2727 }
99870f4d
KW
2728
2729 # Store them as a sub-arrays with both components.
2730 push @{$missings{$addr}}, [ $default, $property ];
2731 }
2732 }
2733
2734 # There is nothing for the caller to process on this comment
2735 # line.
2736 next;
2737 }
2738
2739 # Remove comments and trailing space, and skip this line if the
2740 # result is empty
2741 s/#.*//;
2742 s/\s+$//;
2743 next if /^$/;
2744
2745 # Call any handlers for this line, and skip further processing of
2746 # the line if the handler sets the line to null.
2747 foreach my $sub_ref (@{$each_line_handler{$addr}}) {
2748 &{$sub_ref}($self);
2749 next LINE if /^$/;
2750 }
2751
2752 # Here the line is ok. return success.
2753 return 1;
2754 } # End of looping through lines.
2755
2756 # If there is an EOF handler, call it (only once) and if it generates
2757 # more lines to process go back in the loop to handle them.
2758 if ($eof_handler{$addr}) {
2759 &{$eof_handler{$addr}}($self);
2760 $eof_handler{$addr} = ""; # Currently only get one shot at it.
2761 goto LINE if $added_lines{$addr};
2762 }
2763
2764 # Return failure -- no more lines.
2765 return 0;
2766
2767 }
2768
74cd47d0
KW
2769 sub _next_line_with_remapped_range {
2770 my $self = shift;
2771 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
2772
2773 # like _next_line(), but for use on non-ASCII platforms. It sets $_
2774 # to be the next logical input line, if any. Returns non-zero if such
2775 # a line exists. 'logical' means that any lines that have been added
2776 # via insert_lines() will be returned in $_ before the file is read
2777 # again.
2778 #
2779 # The difference from _next_line() is that this remaps the Unicode
2780 # code points in the input to those of the native platform. Each
2781 # input line contains a single code point, or a single contiguous
2782 # range of them This routine splits each range into its individual
2783 # code points and caches them. It returns the cached values,
2784 # translated into their native equivalents, one at a time, for each
2785 # call, before reading the next line. Since native values can only be
2786 # a single byte wide, no translation is needed for code points above
2787 # 0xFF, and ranges that are entirely above that number are not split.
2788 # If an input line contains the range 254-1000, it would be split into
2789 # three elements: 254, 255, and 256-1000. (The downstream table
2790 # insertion code will sort and coalesce the individual code points
2791 # into appropriate ranges.)
2792
2793 my $addr = do { no overloading; pack 'J', $self; };
2794
2795 while (1) {
2796
2797 # Look in cache before reading the next line. Return any cached
2798 # value, translated
2799 my $inserted = shift @{$remapped_lines{$addr}};
2800 if (defined $inserted) {
2801 trace $inserted if main::DEBUG && $to_trace;
2802 $_ = $inserted =~ s/^ ( \d+ ) /sprintf("%04X", utf8::unicode_to_native($1))/xer;
2803 trace $_ if main::DEBUG && $to_trace;
2804 return 1;
2805 }
2806
2807 # Get the next line.
2808 return 0 unless _next_line($self);
2809
2810 # If there is a special handler for it, return the line,
2811 # untranslated. This should happen only for files that are
2812 # special, not being code-point related, such as property names.
2813 return 1 if $handler{$addr}
2814 != \&main::process_generic_property_file;
2815
2816 my ($range, $property_name, $map, @remainder)
2817 = split /\s*;\s*/, $_, -1; # -1 => retain trailing null fields
2818
2819 if (@remainder
2820 || ! defined $property_name
2821 || $range !~ /^ ($code_point_re) (?:\.\. ($code_point_re) )? $/x)
2822 {
2823 Carp::my_carp_bug("Unrecognized input line '$_'. Ignored");
2824 }
2825
2826 my $low = hex $1;
2827 my $high = (defined $2) ? hex $2 : $low;
2828
2829 # If the input maps the range to another code point, remap the
2830 # target if it is between 0 and 255.
2831 my $tail;
2832 if (defined $map) {
2833 $map =~ s/\b 00 ( [0-9A-F]{2} ) \b/sprintf("%04X", utf8::unicode_to_native(hex $1))/gxe;
2834 $tail = "$property_name; $map";
2835 $_ = "$range; $tail";
2836 }
2837 else {
2838 $tail = $property_name;
2839 }
2840
2841 # If entire range is above 255, just return it, unchanged (except
2842 # any mapped-to code point, already changed above)
2843 return 1 if $low > 255;
2844
2845 # Cache an entry for every code point < 255. For those in the
2846 # range above 255, return a dummy entry for just that portion of
2847 # the range. Note that this will be out-of-order, but that is not
2848 # a problem.
2849 foreach my $code_point ($low .. $high) {
2850 if ($code_point > 255) {
2851 $_ = sprintf "%04X..%04X; $tail", $code_point, $high;
2852 return 1;
2853 }
2854 push @{$remapped_lines{$addr}}, "$code_point; $tail";
2855 }
2856 } # End of looping through lines.
2857
2858 # NOTREACHED
2859 }
2860
99870f4d
KW
2861# Not currently used, not fully tested.
2862# sub peek {
2863# # Non-destructive look-ahead one non-adjusted, non-comment, non-blank
2864# # record. Not callable from an each_line_handler(), nor does it call
2865# # an each_line_handler() on the line.
2866#
2867# my $self = shift;
ffe43484 2868# my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
2869#
2870# foreach my $inserted_ref (@{$added_lines{$addr}}) {
2871# my ($adjusted, $line) = @{$inserted_ref};
2872# next if $adjusted;
2873#
2874# # Remove comments and trailing space, and return a non-empty
2875# # resulting line
2876# $line =~ s/#.*//;
2877# $line =~ s/\s+$//;
2878# return $line if $line ne "";
2879# }
2880#
2881# return if ! ref $handle{$addr}; # Don't read unless is real file
2882# while (1) { # Loop until find non-comment, non-empty line
2883# local $to_trace = 1 if main::DEBUG;
2884# trace $_ if main::DEBUG && $to_trace;
2885# return if ! defined (my $line = readline $handle{$addr});
2886# chomp $line;
2887# push @{$added_lines{$addr}}, [ 0, $line ];
2888#
2889# $line =~ s/#.*//;
2890# $line =~ s/\s+$//;
2891# return $line if $line ne "";
2892# }
2893#
2894# return;
2895# }
2896
2897
2898 sub insert_lines {
2899 # Lines can be inserted so that it looks like they were in the input
2900 # file at the place it was when this routine is called. See also
2901 # insert_adjusted_lines(). Lines inserted via this routine go through
2902 # any each_line_handler()
2903
2904 my $self = shift;
2905
2906 # Each inserted line is an array, with the first element being 0 to
2907 # indicate that this line hasn't been adjusted, and needs to be
2908 # processed.
f998e60c 2909 no overloading;
051df77b 2910 push @{$added_lines{pack 'J', $self}}, map { [ 0, $_ ] } @_;
99870f4d
KW
2911 return;
2912 }
2913
2914 sub insert_adjusted_lines {
2915 # Lines can be inserted so that it looks like they were in the input
2916 # file at the place it was when this routine is called. See also
2917 # insert_lines(). Lines inserted via this routine are already fully
2918 # adjusted, ready to be processed; each_line_handler()s handlers will
2919 # not be called. This means this is not a completely general
2920 # facility, as only the last each_line_handler on the stack should
2921 # call this. It could be made more general, by passing to each of the
2922 # line_handlers their position on the stack, which they would pass on
2923 # to this routine, and that would replace the boolean first element in
2924 # the anonymous array pushed here, so that the next_line routine could
2925 # use that to call only those handlers whose index is after it on the
2926 # stack. But this is overkill for what is needed now.
2927
2928 my $self = shift;
2929 trace $_[0] if main::DEBUG && $to_trace;
2930
2931 # Each inserted line is an array, with the first element being 1 to
2932 # indicate that this line has been adjusted
f998e60c 2933 no overloading;
051df77b 2934 push @{$added_lines{pack 'J', $self}}, map { [ 1, $_ ] } @_;
99870f4d
KW
2935 return;
2936 }
2937
2938 sub get_missings {
2939 # Returns the stored up @missings lines' values, and clears the list.
2940 # The values are in an array, consisting of the default in the first
2941 # element, and the property in the 2nd. However, since these lines
2942 # can be stacked up, the return is an array of all these arrays.
2943
2944 my $self = shift;
2945 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
2946
ffe43484 2947 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
2948
2949 # If not accepting a list return, just return the first one.
2950 return shift @{$missings{$addr}} unless wantarray;
2951
2952 my @return = @{$missings{$addr}};
2953 undef @{$missings{$addr}};
2954 return @return;
2955 }
2956
2957 sub _insert_property_into_line {
2958 # Add a property field to $_, if this file requires it.
2959
f998e60c 2960 my $self = shift;
ffe43484 2961 my $addr = do { no overloading; pack 'J', $self; };
f998e60c 2962 my $property = $property{$addr};
99870f4d
KW
2963 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
2964
2965 $_ =~ s/(;|$)/; $property$1/;
2966 return;
2967 }
2968
2969 sub carp_bad_line {
2970 # Output consistent error messages, using either a generic one, or the
2971 # one given by the optional parameter. To avoid gazillions of the
2972 # same message in case the syntax of a file is way off, this routine
2973 # only outputs the first instance of each message, incrementing a
2974 # count so the totals can be output at the end of the file.
2975
2976 my $self = shift;
2977 my $message = shift;
2978 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
2979
ffe43484 2980 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
2981
2982 $message = 'Unexpected line' unless $message;
2983
2984 # No trailing punctuation so as to fit with our addenda.
2985 $message =~ s/[.:;,]$//;
2986
2987 # If haven't seen this exact message before, output it now. Otherwise
2988 # increment the count of how many times it has occurred
2989 unless ($errors{$addr}->{$message}) {
2990 Carp::my_carp("$message in '$_' in "
f998e60c 2991 . $file{$addr}
99870f4d
KW
2992 . " at line $.. Skipping this line;");
2993 $errors{$addr}->{$message} = 1;
2994 }
2995 else {
2996 $errors{$addr}->{$message}++;
2997 }
2998
2999 # Clear the line to prevent any further (meaningful) processing of it.
3000 $_ = "";
3001
3002 return;
3003 }
3004} # End closure
3005
3006package Multi_Default;
3007
3008# Certain properties in early versions of Unicode had more than one possible
3009# default for code points missing from the files. In these cases, one
3010# default applies to everything left over after all the others are applied,
3011# and for each of the others, there is a description of which class of code
3012# points applies to it. This object helps implement this by storing the
3013# defaults, and for all but that final default, an eval string that generates
3014# the class that it applies to.
3015
3016
3017{ # Closure
3018
3019 main::setup_package();
3020
3021 my %class_defaults;
3022 # The defaults structure for the classes
3023 main::set_access('class_defaults', \%class_defaults);
3024
3025 my %other_default;
3026 # The default that applies to everything left over.
3027 main::set_access('other_default', \%other_default, 'r');
3028
3029
3030 sub new {
3031 # The constructor is called with default => eval pairs, terminated by
3032 # the left-over default. e.g.
3033 # Multi_Default->new(
3034 # 'T' => '$gc->table("Mn") + $gc->table("Cf") - 0x200C
3035 # - 0x200D',
3036 # 'R' => 'some other expression that evaluates to code points',
3037 # .
3038 # .
3039 # .
3040 # 'U'));
3041
3042 my $class = shift;
3043
3044 my $self = bless \do{my $anonymous_scalar}, $class;
ffe43484 3045 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3046
3047 while (@_ > 1) {
3048 my $default = shift;
3049 my $eval = shift;
3050 $class_defaults{$addr}->{$default} = $eval;
3051 }
3052
3053 $other_default{$addr} = shift;
3054
3055 return $self;
3056 }
3057
3058 sub get_next_defaults {
3059 # Iterates and returns the next class of defaults.
3060 my $self = shift;
3061 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3062
ffe43484 3063 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3064
3065 return each %{$class_defaults{$addr}};
3066 }
3067}
3068
3069package Alias;
3070
3071# An alias is one of the names that a table goes by. This class defines them
3072# including some attributes. Everything is currently setup in the
3073# constructor.
3074
3075
3076{ # Closure
3077
3078 main::setup_package();
3079
3080 my %name;
3081 main::set_access('name', \%name, 'r');
3082
3083 my %loose_match;
c12f2655 3084 # Should this name match loosely or not.
99870f4d
KW
3085 main::set_access('loose_match', \%loose_match, 'r');
3086
33e96e72
KW
3087 my %make_re_pod_entry;
3088 # Some aliases should not get their own entries in the re section of the
3089 # pod, because they are covered by a wild-card, and some we want to
3090 # discourage use of. Binary
f82fe4ba 3091 main::set_access('make_re_pod_entry', \%make_re_pod_entry, 'r', 's');
99870f4d 3092
fd1e3e84
KW
3093 my %ucd;
3094 # Is this documented to be accessible via Unicode::UCD
3095 main::set_access('ucd', \%ucd, 'r', 's');
3096
99870f4d
KW
3097 my %status;
3098 # Aliases have a status, like deprecated, or even suppressed (which means
3099 # they don't appear in documentation). Enum
3100 main::set_access('status', \%status, 'r');
3101
0eac1e20 3102 my %ok_as_filename;
99870f4d
KW
3103 # Similarly, some aliases should not be considered as usable ones for
3104 # external use, such as file names, or we don't want documentation to
3105 # recommend them. Boolean
0eac1e20 3106 main::set_access('ok_as_filename', \%ok_as_filename, 'r');
99870f4d
KW
3107
3108 sub new {
3109 my $class = shift;
3110
3111 my $self = bless \do { my $anonymous_scalar }, $class;
ffe43484 3112 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3113
3114 $name{$addr} = shift;
3115 $loose_match{$addr} = shift;
33e96e72 3116 $make_re_pod_entry{$addr} = shift;
0eac1e20 3117 $ok_as_filename{$addr} = shift;
99870f4d 3118 $status{$addr} = shift;
fd1e3e84 3119 $ucd{$addr} = shift;
99870f4d
KW
3120
3121 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3122
3123 # Null names are never ok externally
0eac1e20 3124 $ok_as_filename{$addr} = 0 if $name{$addr} eq "";
99870f4d
KW
3125
3126 return $self;
3127 }
3128}
3129
3130package Range;
3131
3132# A range is the basic unit for storing code points, and is described in the
3133# comments at the beginning of the program. Each range has a starting code
3134# point; an ending code point (not less than the starting one); a value
3135# that applies to every code point in between the two end-points, inclusive;
3136# and an enum type that applies to the value. The type is for the user's
3137# convenience, and has no meaning here, except that a non-zero type is
3138# considered to not obey the normal Unicode rules for having standard forms.
3139#
3140# The same structure is used for both map and match tables, even though in the
3141# latter, the value (and hence type) is irrelevant and could be used as a
3142# comment. In map tables, the value is what all the code points in the range
3143# map to. Type 0 values have the standardized version of the value stored as
3144# well, so as to not have to recalculate it a lot.
3145
3146sub trace { return main::trace(@_); }
3147
3148{ # Closure
3149
3150 main::setup_package();
3151
3152 my %start;
3153 main::set_access('start', \%start, 'r', 's');
3154
3155 my %end;
3156 main::set_access('end', \%end, 'r', 's');
3157
3158 my %value;
3159 main::set_access('value', \%value, 'r');
3160
3161 my %type;
3162 main::set_access('type', \%type, 'r');
3163
3164 my %standard_form;
3165 # The value in internal standard form. Defined only if the type is 0.
3166 main::set_access('standard_form', \%standard_form);
3167
3168 # Note that if these fields change, the dump() method should as well
3169
3170 sub new {
3171 return Carp::carp_too_few_args(\@_, 3) if main::DEBUG && @_ < 3;
3172 my $class = shift;
3173
3174 my $self = bless \do { my $anonymous_scalar }, $class;
ffe43484 3175 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3176
3177 $start{$addr} = shift;
3178 $end{$addr} = shift;
3179
3180 my %args = @_;
3181
3182 my $value = delete $args{'Value'}; # Can be 0
3183 $value = "" unless defined $value;
3184 $value{$addr} = $value;
3185
3186 $type{$addr} = delete $args{'Type'} || 0;
3187
3188 Carp::carp_extra_args(\%args) if main::DEBUG && %args;
3189
99870f4d
KW
3190 return $self;
3191 }
3192
3193 use overload
3194 fallback => 0,
3195 qw("") => "_operator_stringify",
3196 "." => \&main::_operator_dot,
1285127e 3197 ".=" => \&main::_operator_dot_equal,
99870f4d
KW
3198 ;
3199
3200 sub _operator_stringify {
3201 my $self = shift;
ffe43484 3202 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3203
3204 # Output it like '0041..0065 (value)'
3205 my $return = sprintf("%04X", $start{$addr})
3206 . '..'
3207 . sprintf("%04X", $end{$addr});
3208 my $value = $value{$addr};
3209 my $type = $type{$addr};
3210 $return .= ' (';
3211 $return .= "$value";
3212 $return .= ", Type=$type" if $type != 0;
3213 $return .= ')';
3214
3215 return $return;
3216 }
3217
3218 sub standard_form {
c292d35a
NC
3219 # Calculate the standard form only if needed, and cache the result.
3220 # The standard form is the value itself if the type is special.
3221 # This represents a considerable CPU and memory saving - at the time
3222 # of writing there are 368676 non-special objects, but the standard
3223 # form is only requested for 22047 of them - ie about 6%.
99870f4d
KW
3224
3225 my $self = shift;
3226 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3227
ffe43484 3228 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3229
3230 return $standard_form{$addr} if defined $standard_form{$addr};
c292d35a
NC
3231
3232 my $value = $value{$addr};
3233 return $value if $type{$addr};
3234 return $standard_form{$addr} = main::standardize($value);
99870f4d
KW
3235 }
3236
3237 sub dump {
3238 # Human, not machine readable. For machine readable, comment out this
3239 # entire routine and let the standard one take effect.
3240 my $self = shift;
3241 my $indent = shift;
3242 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3243
ffe43484 3244 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3245
3246 my $return = $indent
3247 . sprintf("%04X", $start{$addr})
3248 . '..'
3249 . sprintf("%04X", $end{$addr})
3250 . " '$value{$addr}';";
3251 if (! defined $standard_form{$addr}) {
3252 $return .= "(type=$type{$addr})";
3253 }
3254 elsif ($standard_form{$addr} ne $value{$addr}) {
3255 $return .= "(standard '$standard_form{$addr}')";
3256 }
3257 return $return;
3258 }
3259} # End closure
3260
3261package _Range_List_Base;
3262
3263# Base class for range lists. A range list is simply an ordered list of
3264# ranges, so that the ranges with the lowest starting numbers are first in it.
3265#
3266# When a new range is added that is adjacent to an existing range that has the
3267# same value and type, it merges with it to form a larger range.
3268#
3269# Ranges generally do not overlap, except that there can be multiple entries
3270# of single code point ranges. This is because of NameAliases.txt.
3271#
3272# In this program, there is a standard value such that if two different
3273# values, have the same standard value, they are considered equivalent. This
3274# value was chosen so that it gives correct results on Unicode data
3275
3276# There are a number of methods to manipulate range lists, and some operators
3277# are overloaded to handle them.
3278
99870f4d
KW
3279sub trace { return main::trace(@_); }
3280
3281{ # Closure
3282
3283 our $addr;
3284
5b348b71
KW
3285 # Max is initialized to a negative value that isn't adjacent to 0, for
3286 # simpler tests
3287 my $max_init = -2;
3288
99870f4d
KW
3289 main::setup_package();
3290
3291 my %ranges;
3292 # The list of ranges
3293 main::set_access('ranges', \%ranges, 'readable_array');
3294
3295 my %max;
3296 # The highest code point in the list. This was originally a method, but
3297 # actual measurements said it was used a lot.
3298 main::set_access('max', \%max, 'r');
3299
3300 my %each_range_iterator;
3301 # Iterator position for each_range()
3302 main::set_access('each_range_iterator', \%each_range_iterator);
3303
3304 my %owner_name_of;
3305 # Name of parent this is attached to, if any. Solely for better error
3306 # messages.
3307 main::set_access('owner_name_of', \%owner_name_of, 'p_r');
3308
3309 my %_search_ranges_cache;
3310 # A cache of the previous result from _search_ranges(), for better
3311 # performance
3312 main::set_access('_search_ranges_cache', \%_search_ranges_cache);
3313
3314 sub new {
3315 my $class = shift;
3316 my %args = @_;
3317
3318 # Optional initialization data for the range list.
3319 my $initialize = delete $args{'Initialize'};
3320
3321 my $self;
3322
3323 # Use _union() to initialize. _union() returns an object of this
3324 # class, which means that it will call this constructor recursively.
3325 # But it won't have this $initialize parameter so that it won't
3326 # infinitely loop on this.
3327 return _union($class, $initialize, %args) if defined $initialize;
3328
3329 $self = bless \do { my $anonymous_scalar }, $class;
ffe43484 3330 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3331
3332 # Optional parent object, only for debug info.
3333 $owner_name_of{$addr} = delete $args{'Owner'};
3334 $owner_name_of{$addr} = "" if ! defined $owner_name_of{$addr};
3335
3336 # Stringify, in case it is an object.
3337 $owner_name_of{$addr} = "$owner_name_of{$addr}";
3338
3339 # This is used only for error messages, and so a colon is added
3340 $owner_name_of{$addr} .= ": " if $owner_name_of{$addr} ne "";
3341
3342 Carp::carp_extra_args(\%args) if main::DEBUG && %args;
3343
5b348b71 3344 $max{$addr} = $max_init;
99870f4d
KW
3345
3346 $_search_ranges_cache{$addr} = 0;
3347 $ranges{$addr} = [];
3348
3349 return $self;
3350 }
3351
3352 use overload
3353 fallback => 0,
3354 qw("") => "_operator_stringify",
3355 "." => \&main::_operator_dot,
1285127e 3356 ".=" => \&main::_operator_dot_equal,
99870f4d
KW
3357 ;
3358
3359 sub _operator_stringify {
3360 my $self = shift;
ffe43484 3361 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3362
3363 return "Range_List attached to '$owner_name_of{$addr}'"
3364 if $owner_name_of{$addr};
3365 return "anonymous Range_List " . \$self;
3366 }
3367
3368 sub _union {
3369 # Returns the union of the input code points. It can be called as
3370 # either a constructor or a method. If called as a method, the result
3371 # will be a new() instance of the calling object, containing the union
3372 # of that object with the other parameter's code points; if called as
d59563d0 3373 # a constructor, the first parameter gives the class that the new object
99870f4d
KW
3374 # should be, and the second parameter gives the code points to go into
3375 # it.
3376 # In either case, there are two parameters looked at by this routine;
3377 # any additional parameters are passed to the new() constructor.
3378 #
3379 # The code points can come in the form of some object that contains
3380 # ranges, and has a conventionally named method to access them; or
3381 # they can be an array of individual code points (as integers); or
3382 # just a single code point.
3383 #
3384 # If they are ranges, this routine doesn't make any effort to preserve
3198cc57
KW
3385 # the range values and types of one input over the other. Therefore
3386 # this base class should not allow _union to be called from other than
99870f4d
KW
3387 # initialization code, so as to prevent two tables from being added
3388 # together where the range values matter. The general form of this
3389 # routine therefore belongs in a derived class, but it was moved here
3390 # to avoid duplication of code. The failure to overload this in this
3391 # class keeps it safe.
3198cc57
KW
3392 #
3393 # It does make the effort during initialization to accept tables with
3394 # multiple values for the same code point, and to preserve the order
3395 # of these. If there is only one input range or range set, it doesn't
3396 # sort (as it should already be sorted to the desired order), and will
3397 # accept multiple values per code point. Otherwise it will merge
3398 # multiple values into a single one.
99870f4d
KW
3399
3400 my $self;
3401 my @args; # Arguments to pass to the constructor
3402
3403 my $class = shift;
3404
3405 # If a method call, will start the union with the object itself, and
3406 # the class of the new object will be the same as self.
3407 if (ref $class) {
3408 $self = $class;
3409 $class = ref $self;
3410 push @args, $self;
3411 }
3412
3413 # Add the other required parameter.
3414 push @args, shift;
3415 # Rest of parameters are passed on to the constructor
3416
3417 # Accumulate all records from both lists.
3418 my @records;
3198cc57 3419 my $input_count = 0;
99870f4d
KW
3420 for my $arg (@args) {
3421 #local $to_trace = 0 if main::DEBUG;
3422 trace "argument = $arg" if main::DEBUG && $to_trace;
3423 if (! defined $arg) {
3424 my $message = "";
3425 if (defined $self) {
f998e60c 3426 no overloading;
051df77b 3427 $message .= $owner_name_of{pack 'J', $self};
99870f4d 3428 }
ada6088e 3429 Carp::my_carp_bug($message . "Undefined argument to _union. No union done.");
99870f4d
KW
3430 return;
3431 }
3198cc57 3432
99870f4d
KW
3433 $arg = [ $arg ] if ! ref $arg;
3434 my $type = ref $arg;
3435 if ($type eq 'ARRAY') {
3436 foreach my $element (@$arg) {
3437 push @records, Range->new($element, $element);
3198cc57 3438 $input_count++;
99870f4d
KW
3439 }
3440 }
3441 elsif ($arg->isa('Range')) {
3442 push @records, $arg;
3198cc57 3443 $input_count++;
99870f4d
KW
3444 }
3445 elsif ($arg->can('ranges')) {
3446 push @records, $arg->ranges;
3198cc57 3447 $input_count++;
99870f4d
KW
3448 }
3449 else {
3450 my $message = "";
3451 if (defined $self) {
f998e60c 3452 no overloading;
051df77b 3453 $message .= $owner_name_of{pack 'J', $self};
99870f4d
KW
3454 }
3455 Carp::my_carp_bug($message . "Cannot take the union of a $type. No union done.");
3456 return;
3457 }
3458 }
3459
3460 # Sort with the range containing the lowest ordinal first, but if
3461 # two ranges start at the same code point, sort with the bigger range
3462 # of the two first, because it takes fewer cycles.
3198cc57
KW
3463 if ($input_count > 1) {
3464 @records = sort { ($a->start <=> $b->start)
99870f4d
KW
3465 or
3466 # if b is shorter than a, b->end will be
3467 # less than a->end, and we want to select
3468 # a, so want to return -1
3469 ($b->end <=> $a->end)
3470 } @records;
3198cc57 3471 }
99870f4d
KW
3472
3473 my $new = $class->new(@_);
3474
3475 # Fold in records so long as they add new information.
3476 for my $set (@records) {
3477 my $start = $set->start;
3478 my $end = $set->end;
d59563d0 3479 my $value = $set->value;
3198cc57 3480 my $type = $set->type;
99870f4d 3481 if ($start > $new->max) {
3198cc57 3482 $new->_add_delete('+', $start, $end, $value, Type => $type);
99870f4d
KW
3483 }
3484 elsif ($end > $new->max) {
3198cc57
KW
3485 $new->_add_delete('+', $new->max +1, $end, $value,
3486 Type => $type);
3487 }
3488 elsif ($input_count == 1) {
3489 # Here, overlaps existing range, but is from a single input,
3490 # so preserve the multiple values from that input.
3491 $new->_add_delete('+', $start, $end, $value, Type => $type,
3492 Replace => $MULTIPLE_AFTER);
99870f4d
KW
3493 }
3494 }
3495
3496 return $new;
3497 }
3498
3499 sub range_count { # Return the number of ranges in the range list
3500 my $self = shift;
3501 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3502
f998e60c 3503 no overloading;
051df77b 3504 return scalar @{$ranges{pack 'J', $self}};
99870f4d
KW
3505 }
3506
3507 sub min {
3508 # Returns the minimum code point currently in the range list, or if
3509 # the range list is empty, 2 beyond the max possible. This is a
3510 # method because used so rarely, that not worth saving between calls,
3511 # and having to worry about changing it as ranges are added and
3512 # deleted.
3513
3514 my $self = shift;
3515 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3516
ffe43484 3517 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3518
3519 # If the range list is empty, return a large value that isn't adjacent
3520 # to any that could be in the range list, for simpler tests
2d88a86a 3521 return $MAX_WORKING_CODEPOINT + 2 unless scalar @{$ranges{$addr}};
99870f4d
KW
3522 return $ranges{$addr}->[0]->start;
3523 }
3524
3525 sub contains {
3526 # Boolean: Is argument in the range list? If so returns $i such that:
3527 # range[$i]->end < $codepoint <= range[$i+1]->end
3528 # which is one beyond what you want; this is so that the 0th range
3529 # doesn't return false
3530 my $self = shift;
3531 my $codepoint = shift;
3532 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3533
99870f4d
KW
3534 my $i = $self->_search_ranges($codepoint);
3535 return 0 unless defined $i;
3536
3537 # The search returns $i, such that
3538 # range[$i-1]->end < $codepoint <= range[$i]->end
3539 # So is in the table if and only iff it is at least the start position
3540 # of range $i.
f998e60c 3541 no overloading;
051df77b 3542 return 0 if $ranges{pack 'J', $self}->[$i]->start > $codepoint;
99870f4d
KW
3543 return $i + 1;
3544 }
3545
2f7a8815
KW
3546 sub containing_range {
3547 # Returns the range object that contains the code point, undef if none
3548
3549 my $self = shift;
3550 my $codepoint = shift;
3551 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3552
3553 my $i = $self->contains($codepoint);
3554 return unless $i;
3555
3556 # contains() returns 1 beyond where we should look
3557 no overloading;
3558 return $ranges{pack 'J', $self}->[$i-1];
3559 }
3560
99870f4d
KW
3561 sub value_of {
3562 # Returns the value associated with the code point, undef if none
3563
3564 my $self = shift;
3565 my $codepoint = shift;
3566 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3567
d69c231b
KW
3568 my $range = $self->containing_range($codepoint);
3569 return unless defined $range;
99870f4d 3570
d69c231b 3571 return $range->value;
99870f4d
KW
3572 }
3573
0a9dbafc
KW
3574 sub type_of {
3575 # Returns the type of the range containing the code point, undef if
3576 # the code point is not in the table
3577
3578 my $self = shift;
3579 my $codepoint = shift;
3580 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3581
3582 my $range = $self->containing_range($codepoint);
3583 return unless defined $range;
3584
3585 return $range->type;
3586 }
3587
99870f4d
KW
3588 sub _search_ranges {
3589 # Find the range in the list which contains a code point, or where it
3590 # should go if were to add it. That is, it returns $i, such that:
3591 # range[$i-1]->end < $codepoint <= range[$i]->end
3592 # Returns undef if no such $i is possible (e.g. at end of table), or
3593 # if there is an error.
3594
3595 my $self = shift;
3596 my $code_point = shift;
3597 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3598
ffe43484 3599 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3600
3601 return if $code_point > $max{$addr};
3602 my $r = $ranges{$addr}; # The current list of ranges
3603 my $range_list_size = scalar @$r;
3604 my $i;
3605
3606 use integer; # want integer division
3607
3608 # Use the cached result as the starting guess for this one, because,
3609 # an experiment on 5.1 showed that 90% of the time the cache was the
3610 # same as the result on the next call (and 7% it was one less).
3611 $i = $_search_ranges_cache{$addr};
3612 $i = 0 if $i >= $range_list_size; # Reset if no longer valid (prob.
3613 # from an intervening deletion
3614 #local $to_trace = 1 if main::DEBUG;
3615 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);
3616 return $i if $code_point <= $r->[$i]->end
3617 && ($i == 0 || $r->[$i-1]->end < $code_point);
3618
3619 # Here the cache doesn't yield the correct $i. Try adding 1.
3620 if ($i < $range_list_size - 1
3621 && $r->[$i]->end < $code_point &&
3622 $code_point <= $r->[$i+1]->end)
3623 {
3624 $i++;
3625 trace "next \$i is correct: $i" if main::DEBUG && $to_trace;
3626 $_search_ranges_cache{$addr} = $i;
3627 return $i;
3628 }
3629
3630 # Here, adding 1 also didn't work. We do a binary search to
3631 # find the correct position, starting with current $i
3632 my $lower = 0;
3633 my $upper = $range_list_size - 1;
3634 while (1) {
3635 trace "top of loop i=$i:", sprintf("%04X", $r->[$lower]->start), "[$lower] .. ", sprintf("%04X", $r->[$i]->start), "[$i] .. ", sprintf("%04X", $r->[$upper]->start), "[$upper]" if main::DEBUG && $to_trace;
3636
3637 if ($code_point <= $r->[$i]->end) {
3638
3639 # Here we have met the upper constraint. We can quit if we
3640 # also meet the lower one.
3641 last if $i == 0 || $r->[$i-1]->end < $code_point;
3642
3643 $upper = $i; # Still too high.
3644
3645 }
3646 else {
3647
3648 # Here, $r[$i]->end < $code_point, so look higher up.
3649 $lower = $i;
3650 }
3651
3652 # Split search domain in half to try again.
3653 my $temp = ($upper + $lower) / 2;
3654
3655 # No point in continuing unless $i changes for next time
3656 # in the loop.
3657 if ($temp == $i) {
3658
3659 # We can't reach the highest element because of the averaging.
3660 # So if one below the upper edge, force it there and try one
3661 # more time.
3662 if ($i == $range_list_size - 2) {
3663
3664 trace "Forcing to upper edge" if main::DEBUG && $to_trace;
3665 $i = $range_list_size - 1;
3666
3667 # Change $lower as well so if fails next time through,
3668 # taking the average will yield the same $i, and we will
3669 # quit with the error message just below.
3670 $lower = $i;
3671 next;
3672 }
3673 Carp::my_carp_bug("$owner_name_of{$addr}Can't find where the range ought to go. No action taken.");
3674 return;
3675 }
3676 $i = $temp;
3677 } # End of while loop
3678
3679 if (main::DEBUG && $to_trace) {
3680 trace 'i-1=[', $i-1, ']', $r->[$i-1] if $i;
3681 trace "i= [ $i ]", $r->[$i];
3682 trace 'i+1=[', $i+1, ']', $r->[$i+1] if $i < $range_list_size - 1;
3683 }
3684
3685 # Here we have found the offset. Cache it as a starting point for the
3686 # next call.
3687 $_search_ranges_cache{$addr} = $i;
3688 return $i;
3689 }
3690
3691 sub _add_delete {
3692 # Add, replace or delete ranges to or from a list. The $type
3693 # parameter gives which:
3694 # '+' => insert or replace a range, returning a list of any changed
3695 # ranges.
3696 # '-' => delete a range, returning a list of any deleted ranges.
3697 #
3698 # The next three parameters give respectively the start, end, and
3699 # value associated with the range. 'value' should be null unless the
3700 # operation is '+';
3701 #
3702 # The range list is kept sorted so that the range with the lowest
3703 # starting position is first in the list, and generally, adjacent
c1739a4a 3704 # ranges with the same values are merged into a single larger one (see
99870f4d
KW
3705 # exceptions below).
3706 #
c1739a4a 3707 # There are more parameters; all are key => value pairs:
99870f4d
KW
3708 # Type gives the type of the value. It is only valid for '+'.
3709 # All ranges have types; if this parameter is omitted, 0 is
3710 # assumed. Ranges with type 0 are assumed to obey the
3711 # Unicode rules for casing, etc; ranges with other types are
3712 # not. Otherwise, the type is arbitrary, for the caller's
3713 # convenience, and looked at only by this routine to keep
3714 # adjacent ranges of different types from being merged into
3715 # a single larger range, and when Replace =>
3716 # $IF_NOT_EQUIVALENT is specified (see just below).
3717 # Replace determines what to do if the range list already contains
3718 # ranges which coincide with all or portions of the input
3719 # range. It is only valid for '+':
3720 # => $NO means that the new value is not to replace
3721 # any existing ones, but any empty gaps of the
3722 # range list coinciding with the input range
3723 # will be filled in with the new value.
3724 # => $UNCONDITIONALLY means to replace the existing values with
3725 # this one unconditionally. However, if the
3726 # new and old values are identical, the
3727 #