re2/make_perl_groups.pl

#!/usr/bin/perl
# Copyright 2008 The RE2 Authors.  All Rights Reserved.
# Use of this source code is governed by a BSD-style
# license that can be found in the LICENSE file.

# Generate table entries giving character ranges
# for POSIX/Perl character classes.  Rather than
# figure out what the definition is, it is easier to ask
# Perl about each letter from 0-128 and write down
# its answer.

@posixclasses = (
	"[:alnum:]",
	"[:alpha:]",
	"[:ascii:]",
	"[:blank:]",
	"[:cntrl:]",
	"[:digit:]",
	"[:graph:]",
	"[:lower:]",
	"[:print:]",
	"[:punct:]",
	"[:space:]",
	"[:upper:]",
	"[:word:]",
	"[:xdigit:]",
);

@perlclasses = (
	"\\d",
	"\\s",
	"\\w",
);

%overrides = (
	# Prior to Perl 5.18, \s did not match vertical tab.
	# RE2 preserves that original behaviour.
	"\\s:11" => 0,
);

sub ComputeClass($) {
  my ($cname) = @_;
  my @ranges;
  my $regexp = qr/[$cname]/;
  my $start = -1;
  for (my $i=0; $i<=129; $i++) {
    if ($i == 129) { $i = 256; }
    if ($i <= 128 && ($overrides{"$cname:$i"} // chr($i) =~ $regexp)) {
      if ($start < 0) {
        $start = $i;
      }
    } else {
      if ($start >= 0) {
        push @ranges, [$start, $i-1];
      }
      $start = -1;
    }
  }
  return @ranges;
}

sub PrintClass($$@) {
  my ($cnum, $cname, @ranges) = @_;
  print "static const URange16 code${cnum}[] = {  /* $cname */\n";
  for (my $i=0; $i<@ranges; $i++) {
    my @a = @{$ranges[$i]};
    printf "\t{ 0x%x, 0x%x },\n", $a[0], $a[1];
  }
  print "};\n";
  my $n = @ranges;
  my $escname = $cname;
  $escname =~ s/\\/\\\\/g;
  $negname = $escname;
  if ($negname =~ /:/) {
    $negname =~ s/:/:^/;
  } else {
    $negname =~ y/a-z/A-Z/;
  }
  return "{ \"$escname\", +1, code$cnum, $n }", "{ \"$negname\", -1, code$cnum, $n }";
}

my $cnum = 0;

sub PrintClasses($@) {
  my ($pname, @classes) = @_;
  my @entries;
  foreach my $cname (@classes) {
    my @ranges = ComputeClass($cname);
    push @entries, PrintClass(++$cnum, $cname, @ranges);
  }
  print "const UGroup ${pname}_groups[] = {\n";
  foreach my $e (@entries) {
    print "\t$e,\n";
  }
  print "};\n";
  my $count = @entries;
  print "const int num_${pname}_groups = $count;\n";
}

print <<EOF;
// GENERATED BY make_perl_groups.pl; DO NOT EDIT.
// make_perl_groups.pl >perl_groups.cc

#include "re2/unicode_groups.h"

namespace re2 {

EOF

PrintClasses("perl", @perlclasses);
PrintClasses("posix", @posixclasses);

print <<EOF;

}  // namespace re2
EOF