This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
mktables: White-space, comment, debugging changes only
[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
KW
861# unless explicitly added.
862if ($v_version ge v5.2.0) {
863 my $unihan = 'Unihan; remove from list if using Unihan';
ea25a9b2 864 foreach my $table (qw (
99870f4d
KW
865 kAccountingNumeric
866 kOtherNumeric
867 kPrimaryNumeric
868 kCompatibilityVariant
869 kIICore
870 kIRG_GSource
871 kIRG_HSource
872 kIRG_JSource
873 kIRG_KPSource
874 kIRG_MSource
875 kIRG_KSource
876 kIRG_TSource
877 kIRG_USource
878 kIRG_VSource
879 kRSUnicode
ea25a9b2 880 ))
99870f4d
KW
881 {
882 $why_suppress_if_empty_warn_if_not{$table} = $unihan;
883 }
ca12659b
NC
884}
885
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,
1109 General_Category => 'Cn',
1110 Grapheme_Cluster_Break => 'Other',
1111 Hangul_Syllable_Type => 'NA',
1112 ISO_Comment => "",
1113 Jamo_Short_Name => "",
1114 Joining_Group => "No_Joining_Group",
1115 # Joining_Type => Complicated; set in code
1116 kIICore => 'N', # Is converted to binary
1117 #Line_Break => Complicated; set in code
1118 Lowercase_Mapping => $CODE_POINT,
1119 Name => "",
1120 Name_Alias => "",
1121 NFC_QC => 'Yes',
1122 NFD_QC => 'Yes',
1123 NFKC_QC => 'Yes',
1124 NFKD_QC => 'Yes',
1125 Numeric_Type => 'None',
1126 Numeric_Value => 'NaN',
1127 Script => ($v_version le 4.1.0) ? 'Common' : 'Unknown',
1128 Sentence_Break => 'Other',
1129 Simple_Case_Folding => $CODE_POINT,
1130 Simple_Lowercase_Mapping => $CODE_POINT,
1131 Simple_Titlecase_Mapping => $CODE_POINT,
1132 Simple_Uppercase_Mapping => $CODE_POINT,
1133 Titlecase_Mapping => $CODE_POINT,
1134 Unicode_1_Name => "",
1135 Unicode_Radical_Stroke => "",
1136 Uppercase_Mapping => $CODE_POINT,
1137 Word_Break => 'Other',
1138);
1139
232ed87f
KW
1140# Below are files that Unicode furnishes, but this program ignores, and why.
1141# NormalizationCorrections.txt requires some more explanation. It documents
1142# the cumulative fixes to erroneous normalizations in earlier Unicode
1143# versions. Its main purpose is so that someone running on an earlier version
1144# can use this file to override what got published in that earlier release.
1145# It would be easy for mktables to read and handle this file. But all the
1146# corrections in it should already be in the other files for the release it
1147# is. To get it to actually mean something useful, someone would have to be
1148# using an earlier Unicode release, and copy it to the files for that release
1149# and recomplile. So far there has been no demand to do that, so this hasn't
1150# been implemented.
99870f4d 1151my %ignored_files = (
73ba1144
KW
1152 'CJKRadicals.txt' => 'Maps the kRSUnicode property values to corresponding code points',
1153 'Index.txt' => 'Alphabetical index of Unicode characters',
1154 'NamedSqProv.txt' => 'Named sequences proposed for inclusion in a later version of the Unicode Standard; if you need them now, you can append this file to F<NamedSequences.txt> and recompile perl',
1155 'NamesList.txt' => 'Annotated list of characters',
524a8e5e 1156 'NamesList.html' => 'Describes the format and contents of F<NamesList.txt>',
73ba1144
KW
1157 'NormalizationCorrections.txt' => 'Documentation of corrections already incorporated into the Unicode data base',
1158 'Props.txt' => 'Only in very early releases; is a subset of F<PropList.txt> (which is used instead)',
1159 'ReadMe.txt' => 'Documentation',
1160 'StandardizedVariants.txt' => 'Certain glyph variations for character display are standardized. This lists the non-Unihan ones; the Unihan ones are also not used by Perl, and are in a separate Unicode data base L<http://www.unicode.org/ivd>',
524a8e5e 1161 'StandardizedVariants.html' => 'Provides a visual display of the standard variant sequences derived from F<StandardizedVariants.txt>.',
73ba1144 1162 'EmojiSources.txt' => 'Maps certain Unicode code points to their legacy Japanese cell-phone values',
8ee2793f 1163 'USourceData.txt' => 'Documentation of status and cross reference of proposals for encoding by Unicode of Unihan characters',
524a8e5e 1164 'USourceGlyphs.pdf' => 'Pictures of the characters in F<USourceData.txt>',
73ba1144
KW
1165 'auxiliary/WordBreakTest.html' => 'Documentation of validation tests',
1166 'auxiliary/SentenceBreakTest.html' => 'Documentation of validation tests',
1167 'auxiliary/GraphemeBreakTest.html' => 'Documentation of validation tests',
1168 'auxiliary/LineBreakTest.html' => 'Documentation of validation tests',
99870f4d
KW
1169);
1170
1fec9f60
KW
1171my %skipped_files; # List of files that we skip
1172
678f13d5 1173### End of externally interesting definitions, except for @input_file_objects
99870f4d
KW
1174
1175my $HEADER=<<"EOF";
1176# !!!!!!! DO NOT EDIT THIS FILE !!!!!!!
3df51b85
KW
1177# This file is machine-generated by $0 from the Unicode
1178# database, Version $string_version. Any changes made here will be lost!
cf25bb62
JH
1179EOF
1180
126c3d4e 1181my $INTERNAL_ONLY_HEADER = <<"EOF";
99870f4d
KW
1182
1183# !!!!!!! INTERNAL PERL USE ONLY !!!!!!!
fac53429
KW
1184# This file is for internal use by core Perl only. The format and even the
1185# name or existence of this file are subject to change without notice. Don't
d9ae3878
KW
1186# use it directly. Use Unicode::UCD to access the Unicode character data
1187# base.
99870f4d
KW
1188EOF
1189
1190my $DEVELOPMENT_ONLY=<<"EOF";
1191# !!!!!!! DEVELOPMENT USE ONLY !!!!!!!
1192# This file contains information artificially constrained to code points
1193# present in Unicode release $string_compare_versions.
1194# IT CANNOT BE RELIED ON. It is for use during development only and should
23e33b60 1195# not be used for production.
b6922eda
KW
1196
1197EOF
1198
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
1308my %status_past_participles = (
1309 $DISCOURAGED => 'discouraged',
99870f4d
KW
1310 $STABILIZED => 'stabilized',
1311 $OBSOLETE => 'obsolete',
37e2e78e 1312 $DEPRECATED => 'deprecated',
99870f4d
KW
1313);
1314
395dfc19
KW
1315# Table fates. These are somewhat ordered, so that fates < $MAP_PROXIED should be
1316# externally documented.
301ba948 1317my $ORDINARY = 0; # The normal fate.
395dfc19
KW
1318my $MAP_PROXIED = 1; # The map table for the property isn't written out,
1319 # but there is a file written that can be used to
1320 # reconstruct this table
3cdaf629 1321my $INTERNAL_ONLY = 2; # The file for this table is written out, but it is
301ba948 1322 # for Perl's internal use only
277b7b16
KW
1323my $LEGACY_ONLY = 3; # Like $INTERNAL_ONLY, but not actually used by Perl.
1324 # Is for backwards compatibility for applications that
1325 # read the file directly, so it's format is
1326 # unchangeable.
1327my $SUPPRESSED = 4; # The file for this table is not written out, and as a
3cdaf629
KW
1328 # result, we don't bother to do many computations on
1329 # it.
277b7b16 1330my $PLACEHOLDER = 5; # Like $SUPPRESSED, but we go through all the
3cdaf629
KW
1331 # computations anyway, as the values are needed for
1332 # things to work. This happens when we have Perl
1333 # extensions that depend on Unicode tables that
1334 # wouldn't normally be in a given Unicode version.
301ba948 1335
f5817e0a
KW
1336# The format of the values of the tables:
1337my $EMPTY_FORMAT = "";
99870f4d
KW
1338my $BINARY_FORMAT = 'b';
1339my $DECIMAL_FORMAT = 'd';
1340my $FLOAT_FORMAT = 'f';
1341my $INTEGER_FORMAT = 'i';
1342my $HEX_FORMAT = 'x';
1343my $RATIONAL_FORMAT = 'r';
1344my $STRING_FORMAT = 's';
d11155ec 1345my $ADJUST_FORMAT = 'a';
24303724 1346my $HEX_ADJUST_FORMAT = 'ax';
a14f3cb1 1347my $DECOMP_STRING_FORMAT = 'c';
c3ff2976 1348my $STRING_WHITE_SPACE_LIST = 'sw';
99870f4d
KW
1349
1350my %map_table_formats = (
1351 $BINARY_FORMAT => 'binary',
1352 $DECIMAL_FORMAT => 'single decimal digit',
1353 $FLOAT_FORMAT => 'floating point number',
1354 $INTEGER_FORMAT => 'integer',
add63c13 1355 $HEX_FORMAT => 'non-negative hex whole number; a code point',
99870f4d 1356 $RATIONAL_FORMAT => 'rational: an integer or a fraction',
1a9d544b 1357 $STRING_FORMAT => 'string',
d11155ec 1358 $ADJUST_FORMAT => 'some entries need adjustment',
24303724 1359 $HEX_ADJUST_FORMAT => 'mapped value in hex; some entries need adjustment',
92f9d56c 1360 $DECOMP_STRING_FORMAT => 'Perl\'s internal (Normalize.pm) decomposition mapping',
c3ff2976 1361 $STRING_WHITE_SPACE_LIST => 'string, but some elements are interpreted as a list; white space occurs only as list item separators'
99870f4d
KW
1362);
1363
1364# Unicode didn't put such derived files in a separate directory at first.
1365my $EXTRACTED_DIR = (-d 'extracted') ? 'extracted' : "";
1366my $EXTRACTED = ($EXTRACTED_DIR) ? "$EXTRACTED_DIR/" : "";
1367my $AUXILIARY = 'auxiliary';
1368
3854b4b8
KW
1369# Hashes and arrays that will eventually go into Heavy.pl for the use of
1370# utf8_heavy.pl and into UCD.pl for the use of UCD.pm
99870f4d
KW
1371my %loose_to_file_of; # loosely maps table names to their respective
1372 # files
1373my %stricter_to_file_of; # same; but for stricter mapping.
315bfd4e 1374my %loose_property_to_file_of; # Maps a loose property name to its map file
e1dc048d 1375my %strict_property_to_file_of; # Same, but strict
3854b4b8
KW
1376my @inline_definitions = "V0"; # Each element gives a definition of a unique
1377 # inversion list. When a definition is inlined,
1378 # its value in the hash it's in (one of the two
1379 # defined just above) will include an index into
1380 # this array. The 0th element is initialized to
9be22ca5 1381 # the definition for a zero length inversion list
89cf10cc
KW
1382my %file_to_swash_name; # Maps the file name to its corresponding key name
1383 # in the hash %utf8::SwashInfo
99870f4d
KW
1384my %nv_floating_to_rational; # maps numeric values floating point numbers to
1385 # their rational equivalent
c12f2655
KW
1386my %loose_property_name_of; # Loosely maps (non_string) property names to
1387 # standard form
e1dc048d
KW
1388my %strict_property_name_of; # Strictly maps (non_string) property names to
1389 # standard form
86a52d1e 1390my %string_property_loose_to_name; # Same, for string properties.
c15fda25
KW
1391my %loose_defaults; # keys are of form "prop=value", where 'prop' is
1392 # the property name in standard loose form, and
1393 # 'value' is the default value for that property,
1394 # also in standard loose form.
9e4a1e86
KW
1395my %loose_to_standard_value; # loosely maps table names to the canonical
1396 # alias for them
2df7880f
KW
1397my %ambiguous_names; # keys are alias names (in standard form) that
1398 # have more than one possible meaning.
5d1df013
KW
1399my %prop_aliases; # Keys are standard property name; values are each
1400 # one's aliases
1e863613
KW
1401my %prop_value_aliases; # Keys of top level are standard property name;
1402 # values are keys to another hash, Each one is
1403 # one of the property's values, in standard form.
1404 # The values are that prop-val's aliases.
2df7880f 1405my %ucd_pod; # Holds entries that will go into the UCD section of the pod
99870f4d 1406
d867ccfb
KW
1407# Most properties are immune to caseless matching, otherwise you would get
1408# nonsensical results, as properties are a function of a code point, not
1409# everything that is caselessly equivalent to that code point. For example,
1410# Changes_When_Case_Folded('s') should be false, whereas caselessly it would
1411# be true because 's' and 'S' are equivalent caselessly. However,
1412# traditionally, [:upper:] and [:lower:] are equivalent caselessly, so we
1413# extend that concept to those very few properties that are like this. Each
1414# such property will match the full range caselessly. They are hard-coded in
1415# the program; it's not worth trying to make it general as it's extremely
1416# unlikely that they will ever change.
1417my %caseless_equivalent_to;
1418
99870f4d
KW
1419# These constants names and values were taken from the Unicode standard,
1420# version 5.1, section 3.12. They are used in conjunction with Hangul
6e5a209b
KW
1421# syllables. The '_string' versions are so generated tables can retain the
1422# hex format, which is the more familiar value
1423my $SBase_string = "0xAC00";
1424my $SBase = CORE::hex $SBase_string;
1425my $LBase_string = "0x1100";
1426my $LBase = CORE::hex $LBase_string;
1427my $VBase_string = "0x1161";
1428my $VBase = CORE::hex $VBase_string;
1429my $TBase_string = "0x11A7";
1430my $TBase = CORE::hex $TBase_string;
99870f4d
KW
1431my $SCount = 11172;
1432my $LCount = 19;
1433my $VCount = 21;
1434my $TCount = 28;
1435my $NCount = $VCount * $TCount;
1436
1437# For Hangul syllables; These store the numbers from Jamo.txt in conjunction
1438# with the above published constants.
1439my %Jamo;
1440my %Jamo_L; # Leading consonants
1441my %Jamo_V; # Vowels
1442my %Jamo_T; # Trailing consonants
1443
bb1dd3da
KW
1444# For code points whose name contains its ordinal as a '-ABCD' suffix.
1445# The key is the base name of the code point, and the value is an
1446# array giving all the ranges that use this base name. Each range
1447# is actually a hash giving the 'low' and 'high' values of it.
1448my %names_ending_in_code_point;
1449my %loose_names_ending_in_code_point; # Same as above, but has blanks, dashes
1450 # removed from the names
1451# Inverse mapping. The list of ranges that have these kinds of
1452# names. Each element contains the low, high, and base names in an
1453# anonymous hash.
1454my @code_points_ending_in_code_point;
1455
6b5ab373
KW
1456# To hold Unicode's normalization test suite
1457my @normalization_tests;
1458
bb1dd3da
KW
1459# Boolean: does this Unicode version have the hangul syllables, and are we
1460# writing out a table for them?
1461my $has_hangul_syllables = 0;
1462
1463# Does this Unicode version have code points whose names end in their
1464# respective code points, and are we writing out a table for them? 0 for no;
1465# otherwise points to first property that a table is needed for them, so that
1466# if multiple tables are needed, we don't create duplicates
1467my $needing_code_points_ending_in_code_point = 0;
1468
37e2e78e 1469my @backslash_X_tests; # List of tests read in for testing \X
06ae2722 1470my @SB_tests; # List of tests read in for testing \b{sb}
ae3bb8ea 1471my @WB_tests; # List of tests read in for testing \b{wb}
99870f4d
KW
1472my @unhandled_properties; # Will contain a list of properties found in
1473 # the input that we didn't process.
f86864ac 1474my @match_properties; # Properties that have match tables, to be
99870f4d
KW
1475 # listed in the pod
1476my @map_properties; # Properties that get map files written
1477my @named_sequences; # NamedSequences.txt contents.
1478my %potential_files; # Generated list of all .txt files in the directory
1479 # structure so we can warn if something is being
1480 # ignored.
1481my @files_actually_output; # List of files we generated.
1482my @more_Names; # Some code point names are compound; this is used
1483 # to store the extra components of them.
1484my $MIN_FRACTION_LENGTH = 3; # How many digits of a floating point number at
1485 # the minimum before we consider it equivalent to a
1486 # candidate rational
1487my $MAX_FLOATING_SLOP = 10 ** - $MIN_FRACTION_LENGTH; # And in floating terms
1488
1489# These store references to certain commonly used property objects
3c88a801 1490my $ccc;
99870f4d
KW
1491my $gc;
1492my $perl;
1493my $block;
3e20195b
KW
1494my $perl_charname;
1495my $print;
ac7dbdb6 1496my $All;
bc0c431d 1497my $Assigned; # All assigned characters in this Unicode release
359523e2 1498my $script;
99870f4d
KW
1499
1500# Are there conflicting names because of beginning with 'In_', or 'Is_'
1501my $has_In_conflicts = 0;
1502my $has_Is_conflicts = 0;
1503
1504sub internal_file_to_platform ($) {
1505 # Convert our file paths which have '/' separators to those of the
1506 # platform.
1507
1508 my $file = shift;
1509 return undef unless defined $file;
1510
1511 return File::Spec->join(split '/', $file);
d07a55ed 1512}
5beb625e 1513
99870f4d
KW
1514sub file_exists ($) { # platform independent '-e'. This program internally
1515 # uses slash as a path separator.
1516 my $file = shift;
1517 return 0 if ! defined $file;
1518 return -e internal_file_to_platform($file);
1519}
5beb625e 1520
99870f4d 1521sub objaddr($) {
23e33b60
KW
1522 # Returns the address of the blessed input object.
1523 # It doesn't check for blessedness because that would do a string eval
1524 # every call, and the program is structured so that this is never called
1525 # for a non-blessed object.
99870f4d 1526
23e33b60 1527 no overloading; # If overloaded, numifying below won't work.
99870f4d
KW
1528
1529 # Numifying a ref gives its address.
051df77b 1530 return pack 'J', $_[0];
99870f4d
KW
1531}
1532
558712cf 1533# These are used only if $annotate is true.
c4019d52
KW
1534# The entire range of Unicode characters is examined to populate these
1535# after all the input has been processed. But most can be skipped, as they
1536# have the same descriptive phrases, such as being unassigned
1537my @viacode; # Contains the 1 million character names
1538my @printable; # boolean: And are those characters printable?
1539my @annotate_char_type; # Contains a type of those characters, specifically
1540 # for the purposes of annotation.
1541my $annotate_ranges; # A map of ranges of code points that have the same
98dc9551 1542 # name for the purposes of annotation. They map to the
c4019d52
KW
1543 # upper edge of the range, so that the end point can
1544 # be immediately found. This is used to skip ahead to
1545 # the end of a range, and avoid processing each
1546 # individual code point in it.
1547my $unassigned_sans_noncharacters; # A Range_List of the unassigned
1548 # characters, but excluding those which are
1549 # also noncharacter code points
1550
1551# The annotation types are an extension of the regular range types, though
1552# some of the latter are folded into one. Make the new types negative to
1553# avoid conflicting with the regular types
1554my $SURROGATE_TYPE = -1;
1555my $UNASSIGNED_TYPE = -2;
1556my $PRIVATE_USE_TYPE = -3;
1557my $NONCHARACTER_TYPE = -4;
1558my $CONTROL_TYPE = -5;
2d88a86a
KW
1559my $ABOVE_UNICODE_TYPE = -6;
1560my $UNKNOWN_TYPE = -7; # Used only if there is a bug in this program
c4019d52
KW
1561
1562sub populate_char_info ($) {
558712cf 1563 # Used only with the $annotate option. Populates the arrays with the
c4019d52
KW
1564 # input code point's info that are needed for outputting more detailed
1565 # comments. If calling context wants a return, it is the end point of
1566 # any contiguous range of characters that share essentially the same info
1567
1568 my $i = shift;
1569 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
1570
1571 $viacode[$i] = $perl_charname->value_of($i) || "";
1572
1573 # A character is generally printable if Unicode says it is,
1574 # but below we make sure that most Unicode general category 'C' types
1575 # aren't.
1576 $printable[$i] = $print->contains($i);
1577
1578 $annotate_char_type[$i] = $perl_charname->type_of($i) || 0;
1579
1580 # Only these two regular types are treated specially for annotations
1581 # purposes
1582 $annotate_char_type[$i] = 0 if $annotate_char_type[$i] != $CP_IN_NAME
1583 && $annotate_char_type[$i] != $HANGUL_SYLLABLE;
1584
1585 # Give a generic name to all code points that don't have a real name.
1586 # We output ranges, if applicable, for these. Also calculate the end
1587 # point of the range.
1588 my $end;
1589 if (! $viacode[$i]) {
1d025d66 1590 my $nonchar;
2d88a86a
KW
1591 if ($i > $MAX_UNICODE_CODEPOINT) {
1592 $viacode[$i] = 'Above-Unicode';
1593 $annotate_char_type[$i] = $ABOVE_UNICODE_TYPE;
1594 $printable[$i] = 0;
1595 $end = $MAX_WORKING_CODEPOINT;
1596 }
1597 elsif ($gc-> table('Private_use')->contains($i)) {
c4019d52
KW
1598 $viacode[$i] = 'Private Use';
1599 $annotate_char_type[$i] = $PRIVATE_USE_TYPE;
1600 $printable[$i] = 0;
1601 $end = $gc->table('Private_Use')->containing_range($i)->end;
1602 }
1d025d66
KW
1603 elsif ((defined ($nonchar =
1604 Property::property_ref('Noncharacter_Code_Point'))
1605 && $nonchar->table('Y')->contains($i)))
c4019d52
KW
1606 {
1607 $viacode[$i] = 'Noncharacter';
1608 $annotate_char_type[$i] = $NONCHARACTER_TYPE;
1609 $printable[$i] = 0;
1610 $end = property_ref('Noncharacter_Code_Point')->table('Y')->
1611 containing_range($i)->end;
1612 }
1613 elsif ($gc-> table('Control')->contains($i)) {
c71dea7f 1614 $viacode[$i] = property_ref('Name_Alias')->value_of($i) || 'Control';
c4019d52
KW
1615 $annotate_char_type[$i] = $CONTROL_TYPE;
1616 $printable[$i] = 0;
c4019d52
KW
1617 }
1618 elsif ($gc-> table('Unassigned')->contains($i)) {
c4019d52
KW
1619 $annotate_char_type[$i] = $UNASSIGNED_TYPE;
1620 $printable[$i] = 0;
1d025d66
KW
1621 if ($v_version lt v2.0.0) { # No blocks in earliest releases
1622 $viacode[$i] = 'Unassigned';
1623 $end = $gc-> table('Unassigned')->containing_range($i)->end;
1624 }
1625 else {
1626 $viacode[$i] = 'Unassigned, block=' . $block-> value_of($i);
c4019d52 1627
bf06c733
KW
1628 # Because we name the unassigned by the blocks they are in, it
1629 # can't go past the end of that block, and it also can't go
1630 # past the unassigned range it is in. The special table makes
1631 # sure that the non-characters, which are unassigned, are
1632 # separated out.
1633 $end = min($block->containing_range($i)->end,
1634 $unassigned_sans_noncharacters->
1635 containing_range($i)->end);
1d025d66
KW
1636 }
1637 }
35e2b416 1638 elsif ($perl->table('_Perl_Surrogate')->contains($i)) {
1d025d66
KW
1639 $viacode[$i] = 'Surrogate';
1640 $annotate_char_type[$i] = $SURROGATE_TYPE;
1641 $printable[$i] = 0;
1642 $end = $gc->table('Surrogate')->containing_range($i)->end;
13ca76ff
KW
1643 }
1644 else {
1645 Carp::my_carp_bug("Can't figure out how to annotate "
1646 . sprintf("U+%04X", $i)
1647 . ". Proceeding anyway.");
c4019d52
KW
1648 $viacode[$i] = 'UNKNOWN';
1649 $annotate_char_type[$i] = $UNKNOWN_TYPE;
1650 $printable[$i] = 0;
1651 }
1652 }
1653
1654 # Here, has a name, but if it's one in which the code point number is
1655 # appended to the name, do that.
1656 elsif ($annotate_char_type[$i] == $CP_IN_NAME) {
1657 $viacode[$i] .= sprintf("-%04X", $i);
1658 $end = $perl_charname->containing_range($i)->end;
1659 }
1660
1661 # And here, has a name, but if it's a hangul syllable one, replace it with
1662 # the correct name from the Unicode algorithm
1663 elsif ($annotate_char_type[$i] == $HANGUL_SYLLABLE) {
1664 use integer;
1665 my $SIndex = $i - $SBase;
1666 my $L = $LBase + $SIndex / $NCount;
1667 my $V = $VBase + ($SIndex % $NCount) / $TCount;
1668 my $T = $TBase + $SIndex % $TCount;
1669 $viacode[$i] = "HANGUL SYLLABLE $Jamo{$L}$Jamo{$V}";
1670 $viacode[$i] .= $Jamo{$T} if $T != $TBase;
1671 $end = $perl_charname->containing_range($i)->end;
1672 }
1673
1674 return if ! defined wantarray;
1675 return $i if ! defined $end; # If not a range, return the input
1676
1677 # Save this whole range so can find the end point quickly
1678 $annotate_ranges->add_map($i, $end, $end);
1679
1680 return $end;
1681}
1682
23e33b60
KW
1683# Commented code below should work on Perl 5.8.
1684## This 'require' doesn't necessarily work in miniperl, and even if it does,
1685## the native perl version of it (which is what would operate under miniperl)
1686## is extremely slow, as it does a string eval every call.
7e017d6d 1687#my $has_fast_scalar_util = $^X !~ /miniperl/
23e33b60
KW
1688# && defined eval "require Scalar::Util";
1689#
1690#sub objaddr($) {
1691# # Returns the address of the blessed input object. Uses the XS version if
1692# # available. It doesn't check for blessedness because that would do a
1693# # string eval every call, and the program is structured so that this is
1694# # never called for a non-blessed object.
1695#
1696# return Scalar::Util::refaddr($_[0]) if $has_fast_scalar_util;
1697#
1698# # Check at least that is a ref.
1699# my $pkg = ref($_[0]) or return undef;
1700#
1701# # Change to a fake package to defeat any overloaded stringify
1702# bless $_[0], 'main::Fake';
1703#
1704# # Numifying a ref gives its address.
051df77b 1705# my $addr = pack 'J', $_[0];
23e33b60
KW
1706#
1707# # Return to original class
1708# bless $_[0], $pkg;
1709# return $addr;
1710#}
1711
99870f4d
KW
1712sub max ($$) {
1713 my $a = shift;
1714 my $b = shift;
1715 return $a if $a >= $b;
1716 return $b;
1717}
1718
1719sub min ($$) {
1720 my $a = shift;
1721 my $b = shift;
1722 return $a if $a <= $b;
1723 return $b;
1724}
1725
1726sub clarify_number ($) {
1727 # This returns the input number with underscores inserted every 3 digits
1728 # in large (5 digits or more) numbers. Input must be entirely digits, not
1729 # checked.
1730
1731 my $number = shift;
1732 my $pos = length($number) - 3;
1733 return $number if $pos <= 1;
1734 while ($pos > 0) {
1735 substr($number, $pos, 0) = '_';
1736 $pos -= 3;
5beb625e 1737 }
99870f4d 1738 return $number;
99598c8c
JH
1739}
1740
731cb813
KW
1741sub clarify_code_point_count ($) {
1742 # This is like clarify_number(), but the input is assumed to be a count of
1743 # code points, rather than a generic number.
1744
2d88a86a
KW
1745 my $append = "";
1746
1747 my $number = shift;
1748 if ($number > $MAX_UNICODE_CODEPOINTS) {
1749 $number -= ($MAX_WORKING_CODEPOINTS - $MAX_UNICODE_CODEPOINTS);
1750 return "All above-Unicode code points" if $number == 0;
1751 $append = " + all above-Unicode code points";
1752 }
1753 return clarify_number($number) . $append;
731cb813 1754}
12ac2576 1755
99870f4d 1756package Carp;
7ebf06b3 1757
99870f4d
KW
1758# These routines give a uniform treatment of messages in this program. They
1759# are placed in the Carp package to cause the stack trace to not include them,
1760# although an alternative would be to use another package and set @CARP_NOT
1761# for it.
12ac2576 1762
99870f4d 1763our $Verbose = 1 if main::DEBUG; # Useful info when debugging
12ac2576 1764
99f78760
KW
1765# This is a work-around suggested by Nicholas Clark to fix a problem with Carp
1766# and overload trying to load Scalar:Util under miniperl. See
1767# http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2009-11/msg01057.html
1768undef $overload::VERSION;
1769
99870f4d
KW
1770sub my_carp {
1771 my $message = shift || "";
1772 my $nofold = shift || 0;
7ebf06b3 1773
99870f4d
KW
1774 if ($message) {
1775 $message = main::join_lines($message);
1776 $message =~ s/^$0: *//; # Remove initial program name
1777 $message =~ s/[.;,]+$//; # Remove certain ending punctuation
1778 $message = "\n$0: $message;";
12ac2576 1779
99870f4d
KW
1780 # Fold the message with program name, semi-colon end punctuation
1781 # (which looks good with the message that carp appends to it), and a
1782 # hanging indent for continuation lines.
1783 $message = main::simple_fold($message, "", 4) unless $nofold;
1784 $message =~ s/\n$//; # Remove the trailing nl so what carp
1785 # appends is to the same line
1786 }
12ac2576 1787
99870f4d 1788 return $message if defined wantarray; # If a caller just wants the msg
12ac2576 1789
99870f4d
KW
1790 carp $message;
1791 return;
1792}
7ebf06b3 1793
99870f4d
KW
1794sub my_carp_bug {
1795 # This is called when it is clear that the problem is caused by a bug in
1796 # this program.
7ebf06b3 1797
99870f4d
KW
1798 my $message = shift;
1799 $message =~ s/^$0: *//;
1800 $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");
1801 carp $message;
1802 return;
1803}
7ebf06b3 1804
99870f4d
KW
1805sub carp_too_few_args {
1806 if (@_ != 2) {
1807 my_carp_bug("Wrong number of arguments: to 'carp_too_few_arguments'. No action taken.");
1808 return;
12ac2576 1809 }
7ebf06b3 1810
99870f4d
KW
1811 my $args_ref = shift;
1812 my $count = shift;
7ebf06b3 1813
99870f4d
KW
1814 my_carp_bug("Need at least $count arguments to "
1815 . (caller 1)[3]
1816 . ". Instead got: '"
1817 . join ', ', @$args_ref
1818 . "'. No action taken.");
1819 return;
12ac2576
JP
1820}
1821
99870f4d
KW
1822sub carp_extra_args {
1823 my $args_ref = shift;
1824 my_carp_bug("Too many arguments to 'carp_extra_args': (" . join(', ', @_) . "); Extras ignored.") if @_;
12ac2576 1825
99870f4d
KW
1826 unless (ref $args_ref) {
1827 my_carp_bug("Argument to 'carp_extra_args' ($args_ref) must be a ref. Not checking arguments.");
1828 return;
1829 }
1830 my ($package, $file, $line) = caller;
1831 my $subroutine = (caller 1)[3];
cf25bb62 1832
99870f4d
KW
1833 my $list;
1834 if (ref $args_ref eq 'HASH') {
1835 foreach my $key (keys %$args_ref) {
1836 $args_ref->{$key} = $UNDEF unless defined $args_ref->{$key};
cf25bb62 1837 }
99870f4d 1838 $list = join ', ', each %{$args_ref};
cf25bb62 1839 }
99870f4d
KW
1840 elsif (ref $args_ref eq 'ARRAY') {
1841 foreach my $arg (@$args_ref) {
1842 $arg = $UNDEF unless defined $arg;
1843 }
1844 $list = join ', ', @$args_ref;
1845 }
1846 else {
1847 my_carp_bug("Can't cope with ref "
1848 . ref($args_ref)
1849 . " . argument to 'carp_extra_args'. Not checking arguments.");
1850 return;
1851 }
1852
1853 my_carp_bug("Unrecognized parameters in options: '$list' to $subroutine. Skipped.");
1854 return;
d73e5302
JH
1855}
1856
99870f4d
KW
1857package main;
1858
1859{ # Closure
1860
1861 # This program uses the inside-out method for objects, as recommended in
1b6b3fa9
KW
1862 # "Perl Best Practices". (This is the best solution still, since this has
1863 # to run under miniperl.) This closure aids in generating those. There
99870f4d
KW
1864 # are two routines. setup_package() is called once per package to set
1865 # things up, and then set_access() is called for each hash representing a
1866 # field in the object. These routines arrange for the object to be
1867 # properly destroyed when no longer used, and for standard accessor
1868 # functions to be generated. If you need more complex accessors, just
1869 # write your own and leave those accesses out of the call to set_access().
1870 # More details below.
1871
1872 my %constructor_fields; # fields that are to be used in constructors; see
1873 # below
1874
1875 # The values of this hash will be the package names as keys to other
1876 # hashes containing the name of each field in the package as keys, and
1877 # references to their respective hashes as values.
1878 my %package_fields;
1879
1880 sub setup_package {
1881 # Sets up the package, creating standard DESTROY and dump methods
1882 # (unless already defined). The dump method is used in debugging by
1883 # simple_dumper().
1884 # The optional parameters are:
1885 # a) a reference to a hash, that gets populated by later
1886 # set_access() calls with one of the accesses being
1887 # 'constructor'. The caller can then refer to this, but it is
1888 # not otherwise used by these two routines.
1889 # b) a reference to a callback routine to call during destruction
1890 # of the object, before any fields are actually destroyed
1891
1892 my %args = @_;
1893 my $constructor_ref = delete $args{'Constructor_Fields'};
1894 my $destroy_callback = delete $args{'Destroy_Callback'};
1895 Carp::carp_extra_args(\@_) if main::DEBUG && %args;
1896
1897 my %fields;
1898 my $package = (caller)[0];
1899
1900 $package_fields{$package} = \%fields;
1901 $constructor_fields{$package} = $constructor_ref;
1902
1903 unless ($package->can('DESTROY')) {
1904 my $destroy_name = "${package}::DESTROY";
1905 no strict "refs";
1906
1907 # Use typeglob to give the anonymous subroutine the name we want
1908 *$destroy_name = sub {
1909 my $self = shift;
ffe43484 1910 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
1911
1912 $self->$destroy_callback if $destroy_callback;
1913 foreach my $field (keys %{$package_fields{$package}}) {
1914 #print STDERR __LINE__, ": Destroying ", ref $self, " ", sprintf("%04X", $addr), ": ", $field, "\n";
1915 delete $package_fields{$package}{$field}{$addr};
1916 }
1917 return;
1918 }
1919 }
1920
1921 unless ($package->can('dump')) {
1922 my $dump_name = "${package}::dump";
1923 no strict "refs";
1924 *$dump_name = sub {
1925 my $self = shift;
1926 return dump_inside_out($self, $package_fields{$package}, @_);
1927 }
1928 }
1929 return;
1930 }
1931
1932 sub set_access {
1933 # Arrange for the input field to be garbage collected when no longer
1934 # needed. Also, creates standard accessor functions for the field
1935 # based on the optional parameters-- none if none of these parameters:
1936 # 'addable' creates an 'add_NAME()' accessor function.
1937 # 'readable' or 'readable_array' creates a 'NAME()' accessor
1938 # function.
1939 # 'settable' creates a 'set_NAME()' accessor function.
1940 # 'constructor' doesn't create an accessor function, but adds the
1941 # field to the hash that was previously passed to
1942 # setup_package();
1943 # Any of the accesses can be abbreviated down, so that 'a', 'ad',
1944 # 'add' etc. all mean 'addable'.
1945 # The read accessor function will work on both array and scalar
1946 # values. If another accessor in the parameter list is 'a', the read
1947 # access assumes an array. You can also force it to be array access
1948 # by specifying 'readable_array' instead of 'readable'
1949 #
1950 # A sort-of 'protected' access can be set-up by preceding the addable,
1951 # readable or settable with some initial portion of 'protected_' (but,
1952 # the underscore is required), like 'p_a', 'pro_set', etc. The
1953 # "protection" is only by convention. All that happens is that the
1954 # accessor functions' names begin with an underscore. So instead of
1955 # calling set_foo, the call is _set_foo. (Real protection could be
c1739a4a 1956 # accomplished by having a new subroutine, end_package, called at the
99870f4d
KW
1957 # end of each package, and then storing the __LINE__ ranges and
1958 # checking them on every accessor. But that is way overkill.)
1959
1960 # We create anonymous subroutines as the accessors and then use
1961 # typeglobs to assign them to the proper package and name
1962
1963 my $name = shift; # Name of the field
1964 my $field = shift; # Reference to the inside-out hash containing the
1965 # field
1966
1967 my $package = (caller)[0];
1968
1969 if (! exists $package_fields{$package}) {
1970 croak "$0: Must call 'setup_package' before 'set_access'";
1971 }
d73e5302 1972
99870f4d
KW
1973 # Stash the field so DESTROY can get it.
1974 $package_fields{$package}{$name} = $field;
cf25bb62 1975
99870f4d
KW
1976 # Remaining arguments are the accessors. For each...
1977 foreach my $access (@_) {
1978 my $access = lc $access;
cf25bb62 1979
99870f4d 1980 my $protected = "";
cf25bb62 1981
99870f4d
KW
1982 # Match the input as far as it goes.
1983 if ($access =~ /^(p[^_]*)_/) {
1984 $protected = $1;
1985 if (substr('protected_', 0, length $protected)
1986 eq $protected)
1987 {
1988
1989 # Add 1 for the underscore not included in $protected
1990 $access = substr($access, length($protected) + 1);
1991 $protected = '_';
1992 }
1993 else {
1994 $protected = "";
1995 }
1996 }
1997
1998 if (substr('addable', 0, length $access) eq $access) {
1999 my $subname = "${package}::${protected}add_$name";
2000 no strict "refs";
2001
2002 # add_ accessor. Don't add if already there, which we
2003 # determine using 'eq' for scalars and '==' otherwise.
2004 *$subname = sub {
2005 use strict "refs";
2006 return Carp::carp_too_few_args(\@_, 2) if main::DEBUG && @_ < 2;
2007 my $self = shift;
2008 my $value = shift;
ffe43484 2009 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
2010 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
2011 if (ref $value) {
f998e60c 2012 return if grep { $value == $_ } @{$field->{$addr}};
99870f4d
KW
2013 }
2014 else {
f998e60c 2015 return if grep { $value eq $_ } @{$field->{$addr}};
99870f4d 2016 }
f998e60c 2017 push @{$field->{$addr}}, $value;
99870f4d
KW
2018 return;
2019 }
2020 }
2021 elsif (substr('constructor', 0, length $access) eq $access) {
2022 if ($protected) {
2023 Carp::my_carp_bug("Can't set-up 'protected' constructors")
2024 }
2025 else {
2026 $constructor_fields{$package}{$name} = $field;
2027 }
2028 }
2029 elsif (substr('readable_array', 0, length $access) eq $access) {
2030
2031 # Here has read access. If one of the other parameters for
2032 # access is array, or this one specifies array (by being more
2033 # than just 'readable_'), then create a subroutine that
2034 # assumes the data is an array. Otherwise just a scalar
2035 my $subname = "${package}::${protected}$name";
2036 if (grep { /^a/i } @_
2037 or length($access) > length('readable_'))
2038 {
2039 no strict "refs";
2040 *$subname = sub {
2041 use strict "refs";
23e33b60 2042 Carp::carp_extra_args(\@_) if main::DEBUG && @_ > 1;
ffe43484 2043 my $addr = do { no overloading; pack 'J', $_[0]; };
99870f4d
KW
2044 if (ref $field->{$addr} ne 'ARRAY') {
2045 my $type = ref $field->{$addr};
2046 $type = 'scalar' unless $type;
2047 Carp::my_carp_bug("Trying to read $name as an array when it is a $type. Big problems.");
2048 return;
2049 }
2050 return scalar @{$field->{$addr}} unless wantarray;
2051
2052 # Make a copy; had problems with caller modifying the
2053 # original otherwise
2054 my @return = @{$field->{$addr}};
2055 return @return;
2056 }
2057 }
2058 else {
2059
2060 # Here not an array value, a simpler function.
2061 no strict "refs";
2062 *$subname = sub {
2063 use strict "refs";
23e33b60 2064 Carp::carp_extra_args(\@_) if main::DEBUG && @_ > 1;
f998e60c 2065 no overloading;
051df77b 2066 return $field->{pack 'J', $_[0]};
99870f4d
KW
2067 }
2068 }
2069 }
2070 elsif (substr('settable', 0, length $access) eq $access) {
2071 my $subname = "${package}::${protected}set_$name";
2072 no strict "refs";
2073 *$subname = sub {
2074 use strict "refs";
23e33b60
KW
2075 if (main::DEBUG) {
2076 return Carp::carp_too_few_args(\@_, 2) if @_ < 2;
2077 Carp::carp_extra_args(\@_) if @_ > 2;
2078 }
2079 # $self is $_[0]; $value is $_[1]
f998e60c 2080 no overloading;
051df77b 2081 $field->{pack 'J', $_[0]} = $_[1];
99870f4d
KW
2082 return;
2083 }
2084 }
2085 else {
2086 Carp::my_carp_bug("Unknown accessor type $access. No accessor set.");
2087 }
cf25bb62 2088 }
99870f4d 2089 return;
cf25bb62 2090 }
99870f4d
KW
2091}
2092
2093package Input_file;
2094
2095# All input files use this object, which stores various attributes about them,
2096# and provides for convenient, uniform handling. The run method wraps the
2097# processing. It handles all the bookkeeping of opening, reading, and closing
2098# the file, returning only significant input lines.
2099#
2100# Each object gets a handler which processes the body of the file, and is
74cd47d0
KW
2101# called by run(). All character property files must use the generic,
2102# default handler, which has code scrubbed to handle things you might not
2103# expect, including automatic EBCDIC handling. For files that don't deal with
2104# mapping code points to a property value, such as test files,
2105# PropertyAliases, PropValueAliases, and named sequences, you can override the
2106# handler to be a custom one. Such a handler should basically be a
2107# while(next_line()) {...} loop.
99870f4d
KW
2108#
2109# You can also set up handlers to
537124e4 2110# 1) call before the first line is read, for pre processing
83b68635
KW
2111# 2) call to adjust each line of the input before the main handler gets
2112# them. This can be automatically generated, if appropriately simple
2113# enough, by specifiying a Properties parameter in the constructor.
99870f4d 2114# 3) call upon EOF before the main handler exits its loop
537124e4 2115# 4) call at the end, for post processing
99870f4d
KW
2116#
2117# $_ is used to store the input line, and is to be filtered by the
2118# each_line_handler()s. So, if the format of the line is not in the desired
2119# format for the main handler, these are used to do that adjusting. They can
2120# be stacked (by enclosing them in an [ anonymous array ] in the constructor,
2121# so the $_ output of one is used as the input to the next. None of the other
2122# handlers are stackable, but could easily be changed to be so.
2123#
2124# Most of the handlers can call insert_lines() or insert_adjusted_lines()
2125# which insert the parameters as lines to be processed before the next input
2126# file line is read. This allows the EOF handler to flush buffers, for
2127# example. The difference between the two routines is that the lines inserted
2128# by insert_lines() are subjected to the each_line_handler()s. (So if you
9be22ca5
KW
2129# called it from such a handler, you would get infinite recursion without some
2130# mechanism to prevent that.) Lines inserted by insert_adjusted_lines() go
2131# directly to the main handler without any adjustments. If the
2132# post-processing handler calls any of these, there will be no effect. Some
2133# error checking for these conditions could be added, but it hasn't been done.
99870f4d
KW
2134#
2135# carp_bad_line() should be called to warn of bad input lines, which clears $_
2136# to prevent further processing of the line. This routine will output the
2137# message as a warning once, and then keep a count of the lines that have the
2138# same message, and output that count at the end of the file's processing.
2139# This keeps the number of messages down to a manageable amount.
2140#
2141# get_missings() should be called to retrieve any @missing input lines.
2142# Messages will be raised if this isn't done if the options aren't to ignore
2143# missings.
2144
2145sub trace { return main::trace(@_); }
2146
99870f4d
KW
2147{ # Closure
2148 # Keep track of fields that are to be put into the constructor.
2149 my %constructor_fields;
2150
2151 main::setup_package(Constructor_Fields => \%constructor_fields);
2152
2153 my %file; # Input file name, required
2154 main::set_access('file', \%file, qw{ c r });
2155
2156 my %first_released; # Unicode version file was first released in, required
2157 main::set_access('first_released', \%first_released, qw{ c r });
2158
2159 my %handler; # Subroutine to process the input file, defaults to
2160 # 'process_generic_property_file'
2161 main::set_access('handler', \%handler, qw{ c });
2162
2163 my %property;
2164 # name of property this file is for. defaults to none, meaning not
2165 # applicable, or is otherwise determinable, for example, from each line.
696609bf 2166 main::set_access('property', \%property, qw{ c r });
99870f4d
KW
2167
2168 my %optional;
2169 # If this is true, the file is optional. If not present, no warning is
2170 # output. If it is present, the string given by this parameter is
2171 # evaluated, and if false the file is not processed.
2172 main::set_access('optional', \%optional, 'c', 'r');
2173
2174 my %non_skip;
2175 # This is used for debugging, to skip processing of all but a few input
2176 # files. Add 'non_skip => 1' to the constructor for those files you want
2177 # processed when you set the $debug_skip global.
2178 main::set_access('non_skip', \%non_skip, 'c');
2179
37e2e78e 2180 my %skip;
09ca89ce
KW
2181 # This is used to skip processing of this input file semi-permanently,
2182 # when it evaluates to true. The value should be the reason the file is
2183 # being skipped. It is used for files that we aren't planning to process
2184 # anytime soon, but want to allow to be in the directory and not raise a
2185 # message that we are not handling. Mostly for test files. This is in
2186 # contrast to the non_skip element, which is supposed to be used very
2187 # temporarily for debugging. Sets 'optional' to 1. Also, files that we
2188 # pretty much will never look at can be placed in the global
1fec9f60 2189 # %ignored_files instead. Ones used here will be added to %skipped files
37e2e78e
KW
2190 main::set_access('skip', \%skip, 'c');
2191
99870f4d
KW
2192 my %each_line_handler;
2193 # list of subroutines to look at and filter each non-comment line in the
2194 # file. defaults to none. The subroutines are called in order, each is
2195 # to adjust $_ for the next one, and the final one adjusts it for
2196 # 'handler'
2197 main::set_access('each_line_handler', \%each_line_handler, 'c');
2198
83b68635
KW
2199 my %properties; # Optional ordered list of the properties that occur in each
2200 # meaningful line of the input file. If present, an appropriate
2201 # each_line_handler() is automatically generated and pushed onto the stack
2202 # of such handlers. This is useful when a file contains multiple
2203 # proerties per line, but no other special considerations are necessary.
2204 # The special value "<ignored>" means to discard the corresponding input
2205 # field.
2206 # Any @missing lines in the file should also match this syntax; no such
2207 # files exist as of 6.3. But if it happens in a future release, the code
2208 # could be expanded to properly parse them.
2209 main::set_access('properties', \%properties, qw{ c r });
2210
99870f4d
KW
2211 my %has_missings_defaults;
2212 # ? Are there lines in the file giving default values for code points
2213 # missing from it?. Defaults to NO_DEFAULTS. Otherwise NOT_IGNORED is
2214 # the norm, but IGNORED means it has such lines, but the handler doesn't
2215 # use them. Having these three states allows us to catch changes to the
83b68635
KW
2216 # UCD that this program should track. XXX This could be expanded to
2217 # specify the syntax for such lines, like %properties above.
99870f4d
KW
2218 main::set_access('has_missings_defaults',
2219 \%has_missings_defaults, qw{ c r });
2220
2221 my %pre_handler;
2222 # Subroutine to call before doing anything else in the file. If undef, no
2223 # such handler is called.
2224 main::set_access('pre_handler', \%pre_handler, qw{ c });
2225
2226 my %eof_handler;
2227 # Subroutine to call upon getting an EOF on the input file, but before
2228 # that is returned to the main handler. This is to allow buffers to be
2229 # flushed. The handler is expected to call insert_lines() or
2230 # insert_adjusted() with the buffered material
2231 main::set_access('eof_handler', \%eof_handler, qw{ c r });
2232
2233 my %post_handler;
2234 # Subroutine to call after all the lines of the file are read in and
9be22ca5
KW
2235 # processed. If undef, no such handler is called. Note that this cannot
2236 # add lines to be processed; instead use eof_handler
99870f4d
KW
2237 main::set_access('post_handler', \%post_handler, qw{ c });
2238
2239 my %progress_message;
2240 # Message to print to display progress in lieu of the standard one
2241 main::set_access('progress_message', \%progress_message, qw{ c });
2242
2243 my %handle;
2244 # cache open file handle, internal. Is undef if file hasn't been
2245 # processed at all, empty if has;
2246 main::set_access('handle', \%handle);
2247
2248 my %added_lines;
2249 # cache of lines added virtually to the file, internal
2250 main::set_access('added_lines', \%added_lines);
2251
74cd47d0
KW
2252 my %remapped_lines;
2253 # cache of lines added virtually to the file, internal
2254 main::set_access('remapped_lines', \%remapped_lines);
2255
99870f4d
KW
2256 my %errors;
2257 # cache of errors found, internal
2258 main::set_access('errors', \%errors);
2259
2260 my %missings;
2261 # storage of '@missing' defaults lines
2262 main::set_access('missings', \%missings);
2263
74cd47d0
KW
2264 sub _next_line;
2265 sub _next_line_with_remapped_range;
2266
99870f4d
KW
2267 sub new {
2268 my $class = shift;
2269
2270 my $self = bless \do{ my $anonymous_scalar }, $class;
ffe43484 2271 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
2272
2273 # Set defaults
2274 $handler{$addr} = \&main::process_generic_property_file;
2275 $non_skip{$addr} = 0;
37e2e78e 2276 $skip{$addr} = 0;
99870f4d
KW
2277 $has_missings_defaults{$addr} = $NO_DEFAULTS;
2278 $handle{$addr} = undef;
2279 $added_lines{$addr} = [ ];
74cd47d0 2280 $remapped_lines{$addr} = [ ];
99870f4d
KW
2281 $each_line_handler{$addr} = [ ];
2282 $errors{$addr} = { };
2283 $missings{$addr} = [ ];
2284
2285 # Two positional parameters.
99f78760 2286 return Carp::carp_too_few_args(\@_, 2) if main::DEBUG && @_ < 2;
99870f4d
KW
2287 $file{$addr} = main::internal_file_to_platform(shift);
2288 $first_released{$addr} = shift;
2289
71bd4c0b
KW
2290 undef $file{$addr} if $first_released{$addr} gt $v_version;
2291
99870f4d
KW
2292 # The rest of the arguments are key => value pairs
2293 # %constructor_fields has been set up earlier to list all possible
2294 # ones. Either set or push, depending on how the default has been set
2295 # up just above.
2296 my %args = @_;
2297 foreach my $key (keys %args) {
2298 my $argument = $args{$key};
2299
2300 # Note that the fields are the lower case of the constructor keys
2301 my $hash = $constructor_fields{lc $key};
2302 if (! defined $hash) {
2303 Carp::my_carp_bug("Unrecognized parameters '$key => $argument' to new() for $self. Skipped");
2304 next;
2305 }
2306 if (ref $hash->{$addr} eq 'ARRAY') {
2307 if (ref $argument eq 'ARRAY') {
2308 foreach my $argument (@{$argument}) {
2309 next if ! defined $argument;
2310 push @{$hash->{$addr}}, $argument;
2311 }
2312 }
2313 else {
2314 push @{$hash->{$addr}}, $argument if defined $argument;
2315 }
2316 }
2317 else {
2318 $hash->{$addr} = $argument;
2319 }
2320 delete $args{$key};
2321 };
2322
2323 # If the file has a property for it, it means that the property is not
2324 # listed in the file's entries. So add a handler to the list of line
2325 # handlers to insert the property name into the lines, to provide a
2326 # uniform interface to the final processing subroutine.
2327 # the final code doesn't have to worry about that.
2328 if ($property{$addr}) {
2329 push @{$each_line_handler{$addr}}, \&_insert_property_into_line;
2330 }
2331
2332 if ($non_skip{$addr} && ! $debug_skip && $verbosity) {
2333 print "Warning: " . __PACKAGE__ . " constructor for $file{$addr} has useless 'non_skip' in it\n";
a3a8c5f0 2334 }
99870f4d 2335
09ca89ce
KW
2336 # If skipping, set to optional, and add to list of ignored files,
2337 # including its reason
2338 if ($skip{$addr}) {
2339 $optional{$addr} = 1;
71bd4c0b 2340 $skipped_files{$file{$addr}} = $skip{$addr} if $file{$addr};
09ca89ce 2341 }
83b68635
KW
2342 elsif ($properties{$addr}) {
2343
9be22ca5
KW
2344 # Similarly, there may be more than one property represented on
2345 # each line, with no clue but the constructor input what those
2346 # might be. Add a handler for each line in the input so that it
2347 # creates a separate input line for each property in those input
2348 # lines, thus making them suitable to handle generically.
83b68635
KW
2349
2350 push @{$each_line_handler{$addr}},
2351 sub {
2352 my $file = shift;
2353 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
2354
2355 my @fields = split /\s*;\s*/, $_, -1;
2356
2357 if (@fields - 1 > @{$properties{$addr}}) {
2358 $file->carp_bad_line('Extra fields');
2359 $_ = "";
2360 return;
2361 }
2362 my $range = shift @fields; # 0th element is always the
2363 # range
2364
2365 # The next fields in the input line correspond
2366 # respectively to the stored properties.
2367 for my $i (0 .. @{$properties{$addr}} - 1) {
2368 my $property_name = $properties{$addr}[$i];
2369 next if $property_name eq '<ignored>';
2370 $file->insert_adjusted_lines(
2371 "$range; $property_name; $fields[$i]");
2372 }
2373 $_ = "";
2374
2375 return;
2376 };
2377 }
37e2e78e 2378
9be22ca5 2379 { # On non-ascii platforms, we use a special pre-handler
74cd47d0
KW
2380 no strict;
2381 no warnings 'once';
2382 *next_line = (main::NON_ASCII_PLATFORM)
2383 ? *_next_line_with_remapped_range
2384 : *_next_line;
2385 }
2386
99870f4d 2387 return $self;
d73e5302
JH
2388 }
2389
cf25bb62 2390
99870f4d
KW
2391 use overload
2392 fallback => 0,
2393 qw("") => "_operator_stringify",
2394 "." => \&main::_operator_dot,
1285127e 2395 ".=" => \&main::_operator_dot_equal,
99870f4d 2396 ;
cf25bb62 2397
99870f4d
KW
2398 sub _operator_stringify {
2399 my $self = shift;
cf25bb62 2400
99870f4d 2401 return __PACKAGE__ . " object for " . $self->file;
d73e5302 2402 }
d73e5302 2403
99870f4d
KW
2404 # flag to make sure extracted files are processed early
2405 my $seen_non_extracted_non_age = 0;
d73e5302 2406
99870f4d
KW
2407 sub run {
2408 # Process the input object $self. This opens and closes the file and
2409 # calls all the handlers for it. Currently, this can only be called
2410 # once per file, as it destroy's the EOF handler
d73e5302 2411
99870f4d
KW
2412 my $self = shift;
2413 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
b6922eda 2414
ffe43484 2415 my $addr = do { no overloading; pack 'J', $self; };
b6922eda 2416
99870f4d 2417 my $file = $file{$addr};
d73e5302 2418
99870f4d
KW
2419 # Don't process if not expecting this file (because released later
2420 # than this Unicode version), and isn't there. This means if someone
2421 # copies it into an earlier version's directory, we will go ahead and
2422 # process it.
71bd4c0b
KW
2423 return if $first_released{$addr} gt $v_version
2424 && (! defined $file || ! -e $file);
99870f4d
KW
2425
2426 # If in debugging mode and this file doesn't have the non-skip
2427 # flag set, and isn't one of the critical files, skip it.
2428 if ($debug_skip
2429 && $first_released{$addr} ne v0
2430 && ! $non_skip{$addr})
2431 {
2432 print "Skipping $file in debugging\n" if $verbosity;
2433 return;
2434 }
2435
2436 # File could be optional
37e2e78e 2437 if ($optional{$addr}) {
99870f4d
KW
2438 return unless -e $file;
2439 my $result = eval $optional{$addr};
2440 if (! defined $result) {
2441 Carp::my_carp_bug("Got '$@' when tried to eval $optional{$addr}. $file Skipped.");
2442 return;
2443 }
2444 if (! $result) {
2445 if ($verbosity) {
2446 print STDERR "Skipping processing input file '$file' because '$optional{$addr}' is not true\n";
2447 }
2448 return;
2449 }
2450 }
2451
2452 if (! defined $file || ! -e $file) {
2453
2454 # If the file doesn't exist, see if have internal data for it
2455 # (based on first_released being 0).
2456 if ($first_released{$addr} eq v0) {
2457 $handle{$addr} = 'pretend_is_open';
2458 }
2459 else {
2460 if (! $optional{$addr} # File could be optional
2461 && $v_version ge $first_released{$addr})
2462 {
f71c7390 2463 print STDERR "Skipping processing input file '$file' because not found\n";
99870f4d
KW
2464 }
2465 return;
2466 }
2467 }
2468 else {
99870f4d 2469 if ($seen_non_extracted_non_age) {
9be22ca5
KW
2470 if ($file =~ /$EXTRACTED/i) # Some platforms may change the
2471 # case of the file's name
2472 {
1675ea0d 2473 Carp::my_carp_bug(main::join_lines(<<END
99f78760 2474$file should be processed just after the 'Prop...Alias' files, and before
99870f4d
KW
2475anything not in the $EXTRACTED_DIR directory. Proceeding, but the results may
2476have subtle problems
2477END
2478 ));
2479 }
2480 }
2481 elsif ($EXTRACTED_DIR
2482 && $first_released{$addr} ne v0
517956bf
CB
2483 && $file !~ /$EXTRACTED/i
2484 && lc($file) ne 'dage.txt')
99870f4d
KW
2485 {
2486 # We don't set this (by the 'if' above) if we have no
2487 # extracted directory, so if running on an early version,
2488 # this test won't work. Not worth worrying about.
2489 $seen_non_extracted_non_age = 1;
2490 }
2491
9be22ca5 2492 # Mark the file as having being processed, and warn if it
99870f4d
KW
2493 # isn't a file we are expecting. As we process the files,
2494 # they are deleted from the hash, so any that remain at the
2495 # end of the program are files that we didn't process.
517956bf 2496 my $fkey = File::Spec->rel2abs($file);
faf3cf6b
KW
2497 my $expecting = delete $potential_files{lc($fkey)};
2498
678f13d5
KW
2499 Carp::my_carp("Was not expecting '$file'.") if
2500 ! $expecting
99870f4d
KW
2501 && ! defined $handle{$addr};
2502
37e2e78e
KW
2503 # Having deleted from expected files, we can quit if not to do
2504 # anything. Don't print progress unless really want verbosity
2505 if ($skip{$addr}) {
2506 print "Skipping $file.\n" if $verbosity >= $VERBOSE;
2507 return;
2508 }
2509
9be22ca5
KW
2510 # Here, we are going to process the file. Open it, converting the
2511 # slashes used in this program into the proper form for the OS
99870f4d
KW
2512 my $file_handle;
2513 if (not open $file_handle, "<", $file) {
2514 Carp::my_carp("Can't open $file. Skipping: $!");
2515 return 0;
2516 }
2517 $handle{$addr} = $file_handle; # Cache the open file handle
9e65c3f4 2518
96f226dc 2519 if ($v_version ge v3.2.0 && lc($file) ne 'unicodedata.txt') {
9be22ca5
KW
2520
2521 # UnicodeData.txt has no version marker; the others started
2522 # getting it in 3.2. Unihan files have the version somewhere
2523 # in the first comment block; the other files have it as the
2524 # very first line
96f226dc 2525 if ($file !~ /^Unihan/i) {
cafe9cf0
KW
2526 $_ = <$file_handle>;
2527 if ($_ !~ / - $string_version \. /x) {
2528 chomp;
2529 $_ =~ s/^#\s*//;
2530 die Carp::my_carp("File '$file' is version '$_'. It should be version $string_version");
2531 }
96f226dc
KW
2532 }
2533 else {
2534 while (<$file_handle>) {
2535 if ($_ !~ /^#/) {
2536 Carp::my_carp_bug("Could not find the expected version info in file '$file'");
2537 last;
2538 }
2539 chomp;
2540 $_ =~ s/^#\s*//;
2541 next if $_ !~ / version: /x;
2542 last if $_ =~ /$string_version/;
2543 die Carp::my_carp("File '$file' is '$_'. It should be version $string_version");
2544 }
2545 }
9e65c3f4 2546 }
99870f4d
KW
2547 }
2548
2549 if ($verbosity >= $PROGRESS) {
2550 if ($progress_message{$addr}) {
2551 print "$progress_message{$addr}\n";
2552 }
2553 else {
2554 # If using a virtual file, say so.
2555 print "Processing ", (-e $file)
2556 ? $file
2557 : "substitute $file",
2558 "\n";
2559 }
2560 }
2561
2562
2563 # Call any special handler for before the file.
2564 &{$pre_handler{$addr}}($self) if $pre_handler{$addr};
2565
2566 # Then the main handler
2567 &{$handler{$addr}}($self);
2568
2569 # Then any special post-file handler.
2570 &{$post_handler{$addr}}($self) if $post_handler{$addr};
2571
2572 # If any errors have been accumulated, output the counts (as the first
2573 # error message in each class was output when it was encountered).
2574 if ($errors{$addr}) {
2575 my $total = 0;
2576 my $types = 0;
2577 foreach my $error (keys %{$errors{$addr}}) {
2578 $total += $errors{$addr}->{$error};
2579 delete $errors{$addr}->{$error};
2580 $types++;
2581 }
2582 if ($total > 1) {
2583 my $message
2584 = "A total of $total lines had errors in $file. ";
2585
2586 $message .= ($types == 1)
2587 ? '(Only the first one was displayed.)'
2588 : '(Only the first of each type was displayed.)';
2589 Carp::my_carp($message);
2590 }
2591 }
2592
2593 if (@{$missings{$addr}}) {
2594 Carp::my_carp_bug("Handler for $file didn't look at all the \@missing lines. Generated tables likely are wrong");
2595 }
2596
2597 # If a real file handle, close it.
2598 close $handle{$addr} or Carp::my_carp("Can't close $file: $!") if
2599 ref $handle{$addr};
2600 $handle{$addr} = ""; # Uses empty to indicate that has already seen
2601 # the file, as opposed to undef
2602 return;
2603 }
2604
74cd47d0 2605 sub _next_line {
99870f4d
KW
2606 # Sets $_ to be the next logical input line, if any. Returns non-zero
2607 # if such a line exists. 'logical' means that any lines that have
2608 # been added via insert_lines() will be returned in $_ before the file
2609 # is read again.
2610
2611 my $self = shift;
2612 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
2613
ffe43484 2614 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
2615
2616 # Here the file is open (or if the handle is not a ref, is an open
2617 # 'virtual' file). Get the next line; any inserted lines get priority
2618 # over the file itself.
2619 my $adjusted;
2620
2621 LINE:
2622 while (1) { # Loop until find non-comment, non-empty line
2623 #local $to_trace = 1 if main::DEBUG;
2624 my $inserted_ref = shift @{$added_lines{$addr}};
2625 if (defined $inserted_ref) {
2626 ($adjusted, $_) = @{$inserted_ref};
2627 trace $adjusted, $_ if main::DEBUG && $to_trace;
2628 return 1 if $adjusted;
2629 }
2630 else {
2631 last if ! ref $handle{$addr}; # Don't read unless is real file
2632 last if ! defined ($_ = readline $handle{$addr});
2633 }
2634 chomp;
2635 trace $_ if main::DEBUG && $to_trace;
2636
2637 # See if this line is the comment line that defines what property
2638 # value that code points that are not listed in the file should
2639 # have. The format or existence of these lines is not guaranteed
2640 # by Unicode since they are comments, but the documentation says
2641 # that this was added for machine-readability, so probably won't
2642 # change. This works starting in Unicode Version 5.0. They look
2643 # like:
2644 #
2645 # @missing: 0000..10FFFF; Not_Reordered
2646 # @missing: 0000..10FFFF; Decomposition_Mapping; <code point>
2647 # @missing: 0000..10FFFF; ; NaN
2648 #
2649 # Save the line for a later get_missings() call.
2650 if (/$missing_defaults_prefix/) {
2651 if ($has_missings_defaults{$addr} == $NO_DEFAULTS) {
2652 $self->carp_bad_line("Unexpected \@missing line. Assuming no missing entries");
2653 }
2654 elsif ($has_missings_defaults{$addr} == $NOT_IGNORED) {
2655 my @defaults = split /\s* ; \s*/x, $_;
2656
2657 # The first field is the @missing, which ends in a
2658 # semi-colon, so can safely shift.
2659 shift @defaults;
2660
2661 # Some of these lines may have empty field placeholders
2662 # which get in the way. An example is:
2663 # @missing: 0000..10FFFF; ; NaN
2664 # Remove them. Process starting from the top so the
2665 # splice doesn't affect things still to be looked at.
2666 for (my $i = @defaults - 1; $i >= 0; $i--) {
2667 next if $defaults[$i] ne "";
2668 splice @defaults, $i, 1;
2669 }
2670
2671 # What's left should be just the property (maybe) and the
2672 # default. Having only one element means it doesn't have
2673 # the property.
2674 my $default;
2675 my $property;
2676 if (@defaults >= 1) {
2677 if (@defaults == 1) {
2678 $default = $defaults[0];
2679 }
2680 else {
2681 $property = $defaults[0];
2682 $default = $defaults[1];
2683 }
2684 }
2685
2686 if (@defaults < 1
2687 || @defaults > 2
2688 || ($default =~ /^</
2689 && $default !~ /^<code *point>$/i
09f8d0ac
KW
2690 && $default !~ /^<none>$/i
2691 && $default !~ /^<script>$/i))
99870f4d
KW
2692 {
2693 $self->carp_bad_line("Unrecognized \@missing line: $_. Assuming no missing entries");
2694 }
2695 else {
2696
2697 # If the property is missing from the line, it should
2698 # be the one for the whole file
2699 $property = $property{$addr} if ! defined $property;
2700
2701 # Change <none> to the null string, which is what it
2702 # really means. If the default is the code point
2703 # itself, set it to <code point>, which is what
2704 # Unicode uses (but sometimes they've forgotten the
2705 # space)
2706 if ($default =~ /^<none>$/i) {
2707 $default = "";
2708 }
2709 elsif ($default =~ /^<code *point>$/i) {
2710 $default = $CODE_POINT;
2711 }
09f8d0ac
KW
2712 elsif ($default =~ /^<script>$/i) {
2713
2714 # Special case this one. Currently is from
2715 # ScriptExtensions.txt, and means for all unlisted
2716 # code points, use their Script property values.
2717 # For the code points not listed in that file, the
2718 # default value is 'Unknown'.
2719 $default = "Unknown";
2720 }
99870f4d
KW
2721
2722 # Store them as a sub-arrays with both components.
2723 push @{$missings{$addr}}, [ $default, $property ];
2724 }
2725 }
2726
2727 # There is nothing for the caller to process on this comment
2728 # line.
2729 next;
2730 }
2731
2732 # Remove comments and trailing space, and skip this line if the
2733 # result is empty
2734 s/#.*//;
2735 s/\s+$//;
2736 next if /^$/;
2737
2738 # Call any handlers for this line, and skip further processing of
2739 # the line if the handler sets the line to null.
2740 foreach my $sub_ref (@{$each_line_handler{$addr}}) {
2741 &{$sub_ref}($self);
2742 next LINE if /^$/;
2743 }
2744
2745 # Here the line is ok. return success.
2746 return 1;
2747 } # End of looping through lines.
2748
2749 # If there is an EOF handler, call it (only once) and if it generates
2750 # more lines to process go back in the loop to handle them.
2751 if ($eof_handler{$addr}) {
2752 &{$eof_handler{$addr}}($self);
2753 $eof_handler{$addr} = ""; # Currently only get one shot at it.
2754 goto LINE if $added_lines{$addr};
2755 }
2756
2757 # Return failure -- no more lines.
2758 return 0;
2759
2760 }
2761
74cd47d0
KW
2762 sub _next_line_with_remapped_range {
2763 my $self = shift;
2764 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
2765
2766 # like _next_line(), but for use on non-ASCII platforms. It sets $_
2767 # to be the next logical input line, if any. Returns non-zero if such
2768 # a line exists. 'logical' means that any lines that have been added
2769 # via insert_lines() will be returned in $_ before the file is read
2770 # again.
2771 #
2772 # The difference from _next_line() is that this remaps the Unicode
2773 # code points in the input to those of the native platform. Each
2774 # input line contains a single code point, or a single contiguous
2775 # range of them This routine splits each range into its individual
2776 # code points and caches them. It returns the cached values,
2777 # translated into their native equivalents, one at a time, for each
2778 # call, before reading the next line. Since native values can only be
2779 # a single byte wide, no translation is needed for code points above
2780 # 0xFF, and ranges that are entirely above that number are not split.
2781 # If an input line contains the range 254-1000, it would be split into
2782 # three elements: 254, 255, and 256-1000. (The downstream table
2783 # insertion code will sort and coalesce the individual code points
2784 # into appropriate ranges.)
2785
2786 my $addr = do { no overloading; pack 'J', $self; };
2787
2788 while (1) {
2789
2790 # Look in cache before reading the next line. Return any cached
2791 # value, translated
2792 my $inserted = shift @{$remapped_lines{$addr}};
2793 if (defined $inserted) {
2794 trace $inserted if main::DEBUG && $to_trace;
2795 $_ = $inserted =~ s/^ ( \d+ ) /sprintf("%04X", utf8::unicode_to_native($1))/xer;
2796 trace $_ if main::DEBUG && $to_trace;
2797 return 1;
2798 }
2799
2800 # Get the next line.
2801 return 0 unless _next_line($self);
2802
2803 # If there is a special handler for it, return the line,
2804 # untranslated. This should happen only for files that are
2805 # special, not being code-point related, such as property names.
2806 return 1 if $handler{$addr}
2807 != \&main::process_generic_property_file;
2808
2809 my ($range, $property_name, $map, @remainder)
2810 = split /\s*;\s*/, $_, -1; # -1 => retain trailing null fields
2811
2812 if (@remainder
2813 || ! defined $property_name
2814 || $range !~ /^ ($code_point_re) (?:\.\. ($code_point_re) )? $/x)
2815 {
2816 Carp::my_carp_bug("Unrecognized input line '$_'. Ignored");
2817 }
2818
2819 my $low = hex $1;
2820 my $high = (defined $2) ? hex $2 : $low;
2821
2822 # If the input maps the range to another code point, remap the
2823 # target if it is between 0 and 255.
2824 my $tail;
2825 if (defined $map) {
2826 $map =~ s/\b 00 ( [0-9A-F]{2} ) \b/sprintf("%04X", utf8::unicode_to_native(hex $1))/gxe;
2827 $tail = "$property_name; $map";
2828 $_ = "$range; $tail";
2829 }
2830 else {
2831 $tail = $property_name;
2832 }
2833
2834 # If entire range is above 255, just return it, unchanged (except
2835 # any mapped-to code point, already changed above)
2836 return 1 if $low > 255;
2837
2838 # Cache an entry for every code point < 255. For those in the
2839 # range above 255, return a dummy entry for just that portion of
2840 # the range. Note that this will be out-of-order, but that is not
2841 # a problem.
2842 foreach my $code_point ($low .. $high) {
2843 if ($code_point > 255) {
2844 $_ = sprintf "%04X..%04X; $tail", $code_point, $high;
2845 return 1;
2846 }
2847 push @{$remapped_lines{$addr}}, "$code_point; $tail";
2848 }
2849 } # End of looping through lines.
2850
2851 # NOTREACHED
2852 }
2853
99870f4d
KW
2854# Not currently used, not fully tested.
2855# sub peek {
2856# # Non-destructive look-ahead one non-adjusted, non-comment, non-blank
2857# # record. Not callable from an each_line_handler(), nor does it call
2858# # an each_line_handler() on the line.
2859#
2860# my $self = shift;
ffe43484 2861# my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
2862#
2863# foreach my $inserted_ref (@{$added_lines{$addr}}) {
2864# my ($adjusted, $line) = @{$inserted_ref};
2865# next if $adjusted;
2866#
2867# # Remove comments and trailing space, and return a non-empty
2868# # resulting line
2869# $line =~ s/#.*//;
2870# $line =~ s/\s+$//;
2871# return $line if $line ne "";
2872# }
2873#
2874# return if ! ref $handle{$addr}; # Don't read unless is real file
2875# while (1) { # Loop until find non-comment, non-empty line
2876# local $to_trace = 1 if main::DEBUG;
2877# trace $_ if main::DEBUG && $to_trace;
2878# return if ! defined (my $line = readline $handle{$addr});
2879# chomp $line;
2880# push @{$added_lines{$addr}}, [ 0, $line ];
2881#
2882# $line =~ s/#.*//;
2883# $line =~ s/\s+$//;
2884# return $line if $line ne "";
2885# }
2886#
2887# return;
2888# }
2889
2890
2891 sub insert_lines {
2892 # Lines can be inserted so that it looks like they were in the input
2893 # file at the place it was when this routine is called. See also
2894 # insert_adjusted_lines(). Lines inserted via this routine go through
2895 # any each_line_handler()
2896
2897 my $self = shift;
2898
2899 # Each inserted line is an array, with the first element being 0 to
2900 # indicate that this line hasn't been adjusted, and needs to be
2901 # processed.
f998e60c 2902 no overloading;
051df77b 2903 push @{$added_lines{pack 'J', $self}}, map { [ 0, $_ ] } @_;
99870f4d
KW
2904 return;
2905 }
2906
2907 sub insert_adjusted_lines {
2908 # Lines can be inserted so that it looks like they were in the input
2909 # file at the place it was when this routine is called. See also
2910 # insert_lines(). Lines inserted via this routine are already fully
2911 # adjusted, ready to be processed; each_line_handler()s handlers will
2912 # not be called. This means this is not a completely general
2913 # facility, as only the last each_line_handler on the stack should
2914 # call this. It could be made more general, by passing to each of the
2915 # line_handlers their position on the stack, which they would pass on
2916 # to this routine, and that would replace the boolean first element in
2917 # the anonymous array pushed here, so that the next_line routine could
2918 # use that to call only those handlers whose index is after it on the
2919 # stack. But this is overkill for what is needed now.
2920
2921 my $self = shift;
2922 trace $_[0] if main::DEBUG && $to_trace;
2923
2924 # Each inserted line is an array, with the first element being 1 to
2925 # indicate that this line has been adjusted
f998e60c 2926 no overloading;
051df77b 2927 push @{$added_lines{pack 'J', $self}}, map { [ 1, $_ ] } @_;
99870f4d
KW
2928 return;
2929 }
2930
2931 sub get_missings {
2932 # Returns the stored up @missings lines' values, and clears the list.
2933 # The values are in an array, consisting of the default in the first
2934 # element, and the property in the 2nd. However, since these lines
2935 # can be stacked up, the return is an array of all these arrays.
2936
2937 my $self = shift;
2938 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
2939
ffe43484 2940 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
2941
2942 # If not accepting a list return, just return the first one.
2943 return shift @{$missings{$addr}} unless wantarray;
2944
2945 my @return = @{$missings{$addr}};
2946 undef @{$missings{$addr}};
2947 return @return;
2948 }
2949
2950 sub _insert_property_into_line {
2951 # Add a property field to $_, if this file requires it.
2952
f998e60c 2953 my $self = shift;
ffe43484 2954 my $addr = do { no overloading; pack 'J', $self; };
f998e60c 2955 my $property = $property{$addr};
99870f4d
KW
2956 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
2957
2958 $_ =~ s/(;|$)/; $property$1/;
2959 return;
2960 }
2961
2962 sub carp_bad_line {
2963 # Output consistent error messages, using either a generic one, or the
2964 # one given by the optional parameter. To avoid gazillions of the
2965 # same message in case the syntax of a file is way off, this routine
2966 # only outputs the first instance of each message, incrementing a
2967 # count so the totals can be output at the end of the file.
2968
2969 my $self = shift;
2970 my $message = shift;
2971 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
2972
ffe43484 2973 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
2974
2975 $message = 'Unexpected line' unless $message;
2976
2977 # No trailing punctuation so as to fit with our addenda.
2978 $message =~ s/[.:;,]$//;
2979
2980 # If haven't seen this exact message before, output it now. Otherwise
2981 # increment the count of how many times it has occurred
2982 unless ($errors{$addr}->{$message}) {
2983 Carp::my_carp("$message in '$_' in "
f998e60c 2984 . $file{$addr}
99870f4d
KW
2985 . " at line $.. Skipping this line;");
2986 $errors{$addr}->{$message} = 1;
2987 }
2988 else {
2989 $errors{$addr}->{$message}++;
2990 }
2991
2992 # Clear the line to prevent any further (meaningful) processing of it.
2993 $_ = "";
2994
2995 return;
2996 }
2997} # End closure
2998
2999package Multi_Default;
3000
3001# Certain properties in early versions of Unicode had more than one possible
3002# default for code points missing from the files. In these cases, one
3003# default applies to everything left over after all the others are applied,
3004# and for each of the others, there is a description of which class of code
3005# points applies to it. This object helps implement this by storing the
3006# defaults, and for all but that final default, an eval string that generates
3007# the class that it applies to.
3008
3009
3010{ # Closure
3011
3012 main::setup_package();
3013
3014 my %class_defaults;
3015 # The defaults structure for the classes
3016 main::set_access('class_defaults', \%class_defaults);
3017
3018 my %other_default;
3019 # The default that applies to everything left over.
3020 main::set_access('other_default', \%other_default, 'r');
3021
3022
3023 sub new {
3024 # The constructor is called with default => eval pairs, terminated by
3025 # the left-over default. e.g.
3026 # Multi_Default->new(
3027 # 'T' => '$gc->table("Mn") + $gc->table("Cf") - 0x200C
3028 # - 0x200D',
3029 # 'R' => 'some other expression that evaluates to code points',
3030 # .
3031 # .
3032 # .
3033 # 'U'));
3034
3035 my $class = shift;
3036
3037 my $self = bless \do{my $anonymous_scalar}, $class;
ffe43484 3038 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3039
3040 while (@_ > 1) {
3041 my $default = shift;
3042 my $eval = shift;
3043 $class_defaults{$addr}->{$default} = $eval;
3044 }
3045
3046 $other_default{$addr} = shift;
3047
3048 return $self;
3049 }
3050
3051 sub get_next_defaults {
3052 # Iterates and returns the next class of defaults.
3053 my $self = shift;
3054 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3055
ffe43484 3056 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3057
3058 return each %{$class_defaults{$addr}};
3059 }
3060}
3061
3062package Alias;
3063
3064# An alias is one of the names that a table goes by. This class defines them
3065# including some attributes. Everything is currently setup in the
3066# constructor.
3067
3068
3069{ # Closure
3070
3071 main::setup_package();
3072
3073 my %name;
3074 main::set_access('name', \%name, 'r');
3075
3076 my %loose_match;
c12f2655 3077 # Should this name match loosely or not.
99870f4d
KW
3078 main::set_access('loose_match', \%loose_match, 'r');
3079
33e96e72
KW
3080 my %make_re_pod_entry;
3081 # Some aliases should not get their own entries in the re section of the
3082 # pod, because they are covered by a wild-card, and some we want to
3083 # discourage use of. Binary
f82fe4ba 3084 main::set_access('make_re_pod_entry', \%make_re_pod_entry, 'r', 's');
99870f4d 3085
fd1e3e84
KW
3086 my %ucd;
3087 # Is this documented to be accessible via Unicode::UCD
3088 main::set_access('ucd', \%ucd, 'r', 's');
3089
99870f4d
KW
3090 my %status;
3091 # Aliases have a status, like deprecated, or even suppressed (which means
3092 # they don't appear in documentation). Enum
3093 main::set_access('status', \%status, 'r');
3094
0eac1e20 3095 my %ok_as_filename;
99870f4d
KW
3096 # Similarly, some aliases should not be considered as usable ones for
3097 # external use, such as file names, or we don't want documentation to
3098 # recommend them. Boolean
0eac1e20 3099 main::set_access('ok_as_filename', \%ok_as_filename, 'r');
99870f4d
KW
3100
3101 sub new {
3102 my $class = shift;
3103
3104 my $self = bless \do { my $anonymous_scalar }, $class;
ffe43484 3105 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3106
3107 $name{$addr} = shift;
3108 $loose_match{$addr} = shift;
33e96e72 3109 $make_re_pod_entry{$addr} = shift;
0eac1e20 3110 $ok_as_filename{$addr} = shift;
99870f4d 3111 $status{$addr} = shift;
fd1e3e84 3112 $ucd{$addr} = shift;
99870f4d
KW
3113
3114 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3115
3116 # Null names are never ok externally
0eac1e20 3117 $ok_as_filename{$addr} = 0 if $name{$addr} eq "";
99870f4d
KW
3118
3119 return $self;
3120 }
3121}
3122
3123package Range;
3124
3125# A range is the basic unit for storing code points, and is described in the
3126# comments at the beginning of the program. Each range has a starting code
3127# point; an ending code point (not less than the starting one); a value
3128# that applies to every code point in between the two end-points, inclusive;
3129# and an enum type that applies to the value. The type is for the user's
3130# convenience, and has no meaning here, except that a non-zero type is
3131# considered to not obey the normal Unicode rules for having standard forms.
3132#
3133# The same structure is used for both map and match tables, even though in the
3134# latter, the value (and hence type) is irrelevant and could be used as a
3135# comment. In map tables, the value is what all the code points in the range
3136# map to. Type 0 values have the standardized version of the value stored as
3137# well, so as to not have to recalculate it a lot.
3138
3139sub trace { return main::trace(@_); }
3140
3141{ # Closure
3142
3143 main::setup_package();
3144
3145 my %start;
3146 main::set_access('start', \%start, 'r', 's');
3147
3148 my %end;
3149 main::set_access('end', \%end, 'r', 's');
3150
3151 my %value;
3152 main::set_access('value', \%value, 'r');
3153
3154 my %type;
3155 main::set_access('type', \%type, 'r');
3156
3157 my %standard_form;
3158 # The value in internal standard form. Defined only if the type is 0.
3159 main::set_access('standard_form', \%standard_form);
3160
3161 # Note that if these fields change, the dump() method should as well
3162
3163 sub new {
3164 return Carp::carp_too_few_args(\@_, 3) if main::DEBUG && @_ < 3;
3165 my $class = shift;
3166
3167 my $self = bless \do { my $anonymous_scalar }, $class;
ffe43484 3168 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3169
3170 $start{$addr} = shift;
3171 $end{$addr} = shift;
3172
3173 my %args = @_;
3174
3175 my $value = delete $args{'Value'}; # Can be 0
3176 $value = "" unless defined $value;
3177 $value{$addr} = $value;
3178
3179 $type{$addr} = delete $args{'Type'} || 0;
3180
3181 Carp::carp_extra_args(\%args) if main::DEBUG && %args;
3182
99870f4d
KW
3183 return $self;
3184 }
3185
3186 use overload
3187 fallback => 0,
3188 qw("") => "_operator_stringify",
3189 "." => \&main::_operator_dot,
1285127e 3190 ".=" => \&main::_operator_dot_equal,
99870f4d
KW
3191 ;
3192
3193 sub _operator_stringify {
3194 my $self = shift;
ffe43484 3195 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3196
3197 # Output it like '0041..0065 (value)'
3198 my $return = sprintf("%04X", $start{$addr})
3199 . '..'
3200 . sprintf("%04X", $end{$addr});
3201 my $value = $value{$addr};
3202 my $type = $type{$addr};
3203 $return .= ' (';
3204 $return .= "$value";
3205 $return .= ", Type=$type" if $type != 0;
3206 $return .= ')';
3207
3208 return $return;
3209 }
3210
3211 sub standard_form {
c292d35a
NC
3212 # Calculate the standard form only if needed, and cache the result.
3213 # The standard form is the value itself if the type is special.
3214 # This represents a considerable CPU and memory saving - at the time
3215 # of writing there are 368676 non-special objects, but the standard
3216 # form is only requested for 22047 of them - ie about 6%.
99870f4d
KW
3217
3218 my $self = shift;
3219 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3220
ffe43484 3221 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3222
3223 return $standard_form{$addr} if defined $standard_form{$addr};
c292d35a
NC
3224
3225 my $value = $value{$addr};
3226 return $value if $type{$addr};
3227 return $standard_form{$addr} = main::standardize($value);
99870f4d
KW
3228 }
3229
3230 sub dump {
3231 # Human, not machine readable. For machine readable, comment out this
3232 # entire routine and let the standard one take effect.
3233 my $self = shift;
3234 my $indent = shift;
3235 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3236
ffe43484 3237 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3238
3239 my $return = $indent
3240 . sprintf("%04X", $start{$addr})
3241 . '..'
3242 . sprintf("%04X", $end{$addr})
3243 . " '$value{$addr}';";
3244 if (! defined $standard_form{$addr}) {
3245 $return .= "(type=$type{$addr})";
3246 }
3247 elsif ($standard_form{$addr} ne $value{$addr}) {
3248 $return .= "(standard '$standard_form{$addr}')";
3249 }
3250 return $return;
3251 }
3252} # End closure
3253
3254package _Range_List_Base;
3255
3256# Base class for range lists. A range list is simply an ordered list of
3257# ranges, so that the ranges with the lowest starting numbers are first in it.
3258#
3259# When a new range is added that is adjacent to an existing range that has the
3260# same value and type, it merges with it to form a larger range.
3261#
3262# Ranges generally do not overlap, except that there can be multiple entries
3263# of single code point ranges. This is because of NameAliases.txt.
3264#
3265# In this program, there is a standard value such that if two different
3266# values, have the same standard value, they are considered equivalent. This
3267# value was chosen so that it gives correct results on Unicode data
3268
3269# There are a number of methods to manipulate range lists, and some operators
3270# are overloaded to handle them.
3271
99870f4d
KW
3272sub trace { return main::trace(@_); }
3273
3274{ # Closure
3275
3276 our $addr;
3277
5b348b71
KW
3278 # Max is initialized to a negative value that isn't adjacent to 0, for
3279 # simpler tests
3280 my $max_init = -2;
3281
99870f4d
KW
3282 main::setup_package();
3283
3284 my %ranges;
3285 # The list of ranges
3286 main::set_access('ranges', \%ranges, 'readable_array');
3287
3288 my %max;
3289 # The highest code point in the list. This was originally a method, but
3290 # actual measurements said it was used a lot.
3291 main::set_access('max', \%max, 'r');
3292
3293 my %each_range_iterator;
3294 # Iterator position for each_range()
3295 main::set_access('each_range_iterator', \%each_range_iterator);
3296
3297 my %owner_name_of;
3298 # Name of parent this is attached to, if any. Solely for better error
3299 # messages.
3300 main::set_access('owner_name_of', \%owner_name_of, 'p_r');
3301
3302 my %_search_ranges_cache;
3303 # A cache of the previous result from _search_ranges(), for better
3304 # performance
3305 main::set_access('_search_ranges_cache', \%_search_ranges_cache);
3306
3307 sub new {
3308 my $class = shift;
3309 my %args = @_;
3310
3311 # Optional initialization data for the range list.
3312 my $initialize = delete $args{'Initialize'};
3313
3314 my $self;
3315
3316 # Use _union() to initialize. _union() returns an object of this
3317 # class, which means that it will call this constructor recursively.
3318 # But it won't have this $initialize parameter so that it won't
3319 # infinitely loop on this.
3320 return _union($class, $initialize, %args) if defined $initialize;
3321
3322 $self = bless \do { my $anonymous_scalar }, $class;
ffe43484 3323 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3324
3325 # Optional parent object, only for debug info.
3326 $owner_name_of{$addr} = delete $args{'Owner'};
3327 $owner_name_of{$addr} = "" if ! defined $owner_name_of{$addr};
3328
3329 # Stringify, in case it is an object.
3330 $owner_name_of{$addr} = "$owner_name_of{$addr}";
3331
3332 # This is used only for error messages, and so a colon is added
3333 $owner_name_of{$addr} .= ": " if $owner_name_of{$addr} ne "";
3334
3335 Carp::carp_extra_args(\%args) if main::DEBUG && %args;
3336
5b348b71 3337 $max{$addr} = $max_init;
99870f4d
KW
3338
3339 $_search_ranges_cache{$addr} = 0;
3340 $ranges{$addr} = [];
3341
3342 return $self;
3343 }
3344
3345 use overload
3346 fallback => 0,
3347 qw("") => "_operator_stringify",
3348 "." => \&main::_operator_dot,
1285127e 3349 ".=" => \&main::_operator_dot_equal,
99870f4d
KW
3350 ;
3351
3352 sub _operator_stringify {
3353 my $self = shift;
ffe43484 3354 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3355
3356 return "Range_List attached to '$owner_name_of{$addr}'"
3357 if $owner_name_of{$addr};
3358 return "anonymous Range_List " . \$self;
3359 }
3360
3361 sub _union {
3362 # Returns the union of the input code points. It can be called as
3363 # either a constructor or a method. If called as a method, the result
3364 # will be a new() instance of the calling object, containing the union
3365 # of that object with the other parameter's code points; if called as
d59563d0 3366 # a constructor, the first parameter gives the class that the new object
99870f4d
KW
3367 # should be, and the second parameter gives the code points to go into
3368 # it.
3369 # In either case, there are two parameters looked at by this routine;
3370 # any additional parameters are passed to the new() constructor.
3371 #
3372 # The code points can come in the form of some object that contains
3373 # ranges, and has a conventionally named method to access them; or
3374 # they can be an array of individual code points (as integers); or
3375 # just a single code point.
3376 #
3377 # If they are ranges, this routine doesn't make any effort to preserve
3198cc57
KW
3378 # the range values and types of one input over the other. Therefore
3379 # this base class should not allow _union to be called from other than
99870f4d
KW
3380 # initialization code, so as to prevent two tables from being added
3381 # together where the range values matter. The general form of this
3382 # routine therefore belongs in a derived class, but it was moved here
3383 # to avoid duplication of code. The failure to overload this in this
3384 # class keeps it safe.
3198cc57
KW
3385 #
3386 # It does make the effort during initialization to accept tables with
3387 # multiple values for the same code point, and to preserve the order
3388 # of these. If there is only one input range or range set, it doesn't
3389 # sort (as it should already be sorted to the desired order), and will
3390 # accept multiple values per code point. Otherwise it will merge
3391 # multiple values into a single one.
99870f4d
KW
3392
3393 my $self;
3394 my @args; # Arguments to pass to the constructor
3395
3396 my $class = shift;
3397
3398 # If a method call, will start the union with the object itself, and
3399 # the class of the new object will be the same as self.
3400 if (ref $class) {
3401 $self = $class;
3402 $class = ref $self;
3403 push @args, $self;
3404 }
3405
3406 # Add the other required parameter.
3407 push @args, shift;
3408 # Rest of parameters are passed on to the constructor
3409
3410 # Accumulate all records from both lists.
3411 my @records;
3198cc57 3412 my $input_count = 0;
99870f4d
KW
3413 for my $arg (@args) {
3414 #local $to_trace = 0 if main::DEBUG;
3415 trace "argument = $arg" if main::DEBUG && $to_trace;
3416 if (! defined $arg) {
3417 my $message = "";
3418 if (defined $self) {
f998e60c 3419 no overloading;
051df77b 3420 $message .= $owner_name_of{pack 'J', $self};
99870f4d 3421 }
ada6088e 3422 Carp::my_carp_bug($message . "Undefined argument to _union. No union done.");
99870f4d
KW
3423 return;
3424 }
3198cc57 3425
99870f4d
KW
3426 $arg = [ $arg ] if ! ref $arg;
3427 my $type = ref $arg;
3428 if ($type eq 'ARRAY') {
3429 foreach my $element (@$arg) {
3430 push @records, Range->new($element, $element);
3198cc57 3431 $input_count++;
99870f4d
KW
3432 }
3433 }
3434 elsif ($arg->isa('Range')) {
3435 push @records, $arg;
3198cc57 3436 $input_count++;
99870f4d
KW
3437 }
3438 elsif ($arg->can('ranges')) {
3439 push @records, $arg->ranges;
3198cc57 3440 $input_count++;
99870f4d
KW
3441 }
3442 else {
3443 my $message = "";
3444 if (defined $self) {
f998e60c 3445 no overloading;
051df77b 3446 $message .= $owner_name_of{pack 'J', $self};
99870f4d
KW
3447 }
3448 Carp::my_carp_bug($message . "Cannot take the union of a $type. No union done.");
3449 return;
3450 }
3451 }
3452
3453 # Sort with the range containing the lowest ordinal first, but if
3454 # two ranges start at the same code point, sort with the bigger range
3455 # of the two first, because it takes fewer cycles.
3198cc57
KW
3456 if ($input_count > 1) {
3457 @records = sort { ($a->start <=> $b->start)
99870f4d
KW
3458 or
3459 # if b is shorter than a, b->end will be
3460 # less than a->end, and we want to select
3461 # a, so want to return -1
3462 ($b->end <=> $a->end)
3463 } @records;
3198cc57 3464 }
99870f4d
KW
3465
3466 my $new = $class->new(@_);
3467
3468 # Fold in records so long as they add new information.
3469 for my $set (@records) {
3470 my $start = $set->start;
3471 my $end = $set->end;
d59563d0 3472 my $value = $set->value;
3198cc57 3473 my $type = $set->type;
99870f4d 3474 if ($start > $new->max) {
3198cc57 3475 $new->_add_delete('+', $start, $end, $value, Type => $type);
99870f4d
KW
3476 }
3477 elsif ($end > $new->max) {
3198cc57
KW
3478 $new->_add_delete('+', $new->max +1, $end, $value,
3479 Type => $type);
3480 }
3481 elsif ($input_count == 1) {
3482 # Here, overlaps existing range, but is from a single input,
3483 # so preserve the multiple values from that input.
3484 $new->_add_delete('+', $start, $end, $value, Type => $type,
3485 Replace => $MULTIPLE_AFTER);
99870f4d
KW
3486 }
3487 }
3488
3489 return $new;
3490 }
3491
3492 sub range_count { # Return the number of ranges in the range list
3493 my $self = shift;
3494 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3495
f998e60c 3496 no overloading;
051df77b 3497 return scalar @{$ranges{pack 'J', $self}};
99870f4d
KW
3498 }
3499
3500 sub min {
3501 # Returns the minimum code point currently in the range list, or if
3502 # the range list is empty, 2 beyond the max possible. This is a
3503 # method because used so rarely, that not worth saving between calls,
3504 # and having to worry about changing it as ranges are added and
3505 # deleted.
3506
3507 my $self = shift;
3508 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3509
ffe43484 3510 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3511
3512 # If the range list is empty, return a large value that isn't adjacent
3513 # to any that could be in the range list, for simpler tests
2d88a86a 3514 return $MAX_WORKING_CODEPOINT + 2 unless scalar @{$ranges{$addr}};
99870f4d
KW
3515 return $ranges{$addr}->[0]->start;
3516 }
3517
3518 sub contains {
3519 # Boolean: Is argument in the range list? If so returns $i such that:
3520 # range[$i]->end < $codepoint <= range[$i+1]->end
3521 # which is one beyond what you want; this is so that the 0th range
3522 # doesn't return false
3523 my $self = shift;
3524 my $codepoint = shift;
3525 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3526
99870f4d
KW
3527 my $i = $self->_search_ranges($codepoint);
3528 return 0 unless defined $i;
3529
3530 # The search returns $i, such that
3531 # range[$i-1]->end < $codepoint <= range[$i]->end
3532 # So is in the table if and only iff it is at least the start position
3533 # of range $i.
f998e60c 3534 no overloading;
051df77b 3535 return 0 if $ranges{pack 'J', $self}->[$i]->start > $codepoint;
99870f4d
KW
3536 return $i + 1;
3537 }
3538
2f7a8815
KW
3539 sub containing_range {
3540 # Returns the range object that contains the code point, undef if none
3541
3542 my $self = shift;
3543 my $codepoint = shift;
3544 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3545
3546 my $i = $self->contains($codepoint);
3547 return unless $i;
3548
3549 # contains() returns 1 beyond where we should look
3550 no overloading;
3551 return $ranges{pack 'J', $self}->[$i-1];
3552 }
3553
99870f4d
KW
3554 sub value_of {
3555 # Returns the value associated with the code point, undef if none
3556
3557 my $self = shift;
3558 my $codepoint = shift;
3559 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3560
d69c231b
KW
3561 my $range = $self->containing_range($codepoint);
3562 return unless defined $range;
99870f4d 3563
d69c231b 3564 return $range->value;
99870f4d
KW
3565 }
3566
0a9dbafc
KW
3567 sub type_of {
3568 # Returns the type of the range containing the code point, undef if
3569 # the code point is not in the table
3570
3571 my $self = shift;
3572 my $codepoint = shift;
3573 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3574
3575 my $range = $self->containing_range($codepoint);
3576 return unless defined $range;
3577
3578 return $range->type;
3579 }
3580
99870f4d
KW
3581 sub _search_ranges {
3582 # Find the range in the list which contains a code point, or where it
3583 # should go if were to add it. That is, it returns $i, such that:
3584 # range[$i-1]->end < $codepoint <= range[$i]->end
3585 # Returns undef if no such $i is possible (e.g. at end of table), or
3586 # if there is an error.
3587
3588 my $self = shift;
3589 my $code_point = shift;
3590 Carp::carp_extra_args(\@_) if main::DEBUG && @_;
3591
ffe43484 3592 my $addr = do { no overloading; pack 'J', $self; };
99870f4d
KW
3593
3594 return if $code_point > $max{$addr};
3595 my $r = $ranges{$addr}; # The current list of ranges
3596 my $range_list_size = scalar @$r;
3597 my $i;
3598
3599 use integer; # want integer division
3600
3601 # Use the cached result as the starting guess for this one, because,
3602 # an experiment on 5.1 showed that 90% of the time the cache was the
3603 # same as the result on the next call (and 7% it was one less).
3604 $i = $_search_ranges_cache{$addr};
3605 $i = 0 if $i >= $range_list_size; # Reset if no longer valid (prob.
3606 # from an intervening deletion
3607 #local $to_trace = 1 if main::DEBUG;
3608 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);
3609 return $i if $code_point <= $r->[$i]->end
3610 && ($i == 0 || $r->[$i-1]->end < $code_point);
3611
3612 # Here the cache doesn't yield the correct $i. Try adding 1.
3613 if ($i < $range_list_size - 1
3614 && $r->[$i]->end < $code_point &&
3615 $code_point <= $r->[$i+1]->end)
3616 {
3617 $i++;
3618 trace "next \$i is correct: $i" if main::DEBUG && $to_trace;
3619 $_search_ranges_cache{$addr} = $i;
3620 return $i;
3621 }
3622
3623 # Here, adding 1 also didn't work. We do a binary search to
3624 # find the correct position, starting with current $i
3625 my $lower = 0;
3626 my $upper = $range_list_size - 1;
3627 while (1) {
3628 trace "top of loop i=$i:", sprintf("%04X", $r->[$lower]->start), "[$lower] .. ", sprintf("%04X", $r->[$i]->start), "[$i] .. ", sprintf("%04X", $r->[$upper]->start), "[$upper]" if main::DEBUG && $to_trace;
3629
3630 if ($code_point <= $r->[$i]->end) {
3631
3632 # Here we have met the upper constraint. We can quit if we
3633 # also meet the lower one.
3634 last if $i == 0 || $r->[$i-1]->end < $code_point;
3635
3636 $upper = $i; # Still too high.
3637
3638 }
3639 else {
3640
3641 # Here, $r[$i]->end < $code_point, so look higher up.
3642 $lower = $i;
3643 }
3644
3645 # Split search domain in half to try again.
3646 my $temp = ($upper + $lower) / 2;
3647
3648 # No point in continuing unless $i changes for next time
3649 # in the loop.
3650 if ($temp == $i) {
3651
3652 # We can't reach the highest element because of the averaging.
3653 # So if one below the upper edge, force it there and try one
3654 # more time.
3655 if ($i == $range_list_size - 2) {
3656
3657 trace "Forcing to upper edge" if main::DEBUG && $to_trace;
3658 $i = $range_list_size - 1;
3659
3660 # Change $lower as well so if fails next time through,
3661 # taking the average will yield the same $i, and we will
3662 # quit with the error message just below.
3663 $lower = $i;
3664 next;
3665 }
3666 Carp::my_carp_bug("$owner_name_of{$addr}Can't find where the range ought to go. No action taken.");
3667 return;
3668 }
3669 $i = $temp;
3670 } # End of while loop
3671
3672 if (main::DEBUG && $to_trace) {
3673 trace 'i-1=[', $i-1, ']', $r->[$i-1] if $i;
3674 trace "i= [ $i ]", $r->[$i];
3675 trace 'i+1=[', $i+1, ']', $r->[$i+1] if $i < $range_list_size - 1;
3676 }
3677
3678 # Here we have found the offset. Cache it as a starting point for the
3679 # next call.
3680 $_search_ranges_cache{$addr} = $i;
3681 return $i;
3682 }
3683
3684 sub _add_delete {
3685 # Add, replace or delete ranges to or from a list. The $type
3686 # parameter gives which:
3687 # '+' => insert or replace a range, returning a list of any changed
3688 # ranges.
3689 # '-' => delete a range, returning a list of any deleted ranges.
3690 #
3691 # The next three parameters give respectively the start, end, and
3692 # value associated with the range. 'value' should be null unless the
3693 # operation is '+';
3694 #
3695 # The range list is kept sorted so that the range with the lowest
3696 # starting position is first in the list, and generally, adjacent
c1739a4a 3697 # ranges with the same values are merged into a single larger one (see
99870f4d
KW
3698 # exceptions below).
3699 #
c1739a4a 3700 # There are more parameters; all are key => value pairs:
99870f4d
KW
3701 # Type gives the type of the value. It is only valid for '+'.
3702 # All ranges have types; if this parameter is omitted, 0 is
3703 # assumed. Ranges with type 0 are assumed to obey the
3704 # Unicode rules for casing, etc; ranges with other types are
3705 # not. Otherwise, the type is arbitrary, for the caller's
3706 # convenience, and looked at only by this routine to keep
3707 # adjacent ranges of different types from being merged into
3708 # a single larger range, and when Replace =>
3709 # $IF_NOT_EQUIVALENT is specified (see just below).
3710 # Replace determines what to do if the range list already contains
3711 # ranges which coincide with all or portions of the input
3712 # range. It is only valid for '+':
3713 # => $NO means that the new value is not to replace
3714 # any existing ones, but any empty gaps of the
3715 # range list coinciding with the input range
3716 # will be filled in with the new value.
3717 # => $UNCONDITIONALLY means to replace the existing values with
3718 # this one unconditionally. However, if the
3719 # new and old values are identical, the
3720 # replacement is skipped to save cycles
3721 # => $IF_NOT_EQUIVALENT means to replace the existing values
d59563d0 3722 # (the default) with this one if they are not equivalent.
99870f4d 3723 # Ranges are equivalent if their types are the
c1739a4a 3724 # same, and they are the same string; or if
99870f4d
KW
3725 # both are type 0 ranges, if their Unicode
3726 # standard forms are identical. In this last
3727 # case, the routine chooses the more "modern"
3728 # one to use. This is because some of the
3729 # older files are formatted with values that
3730 # are, for example, ALL CAPs, whereas the