html5/spec-ja/common.pl

use strict;

my $data_suffix = q[.dat];
my $data_dir_name = q[data/];
my $data2_dir_name = q[data2/];
my $data2_suffix = q[.dat];
my $lock_suffix = q[.lock];
my $fallback_file_name = $data2_dir_name . 'fallback' . $data2_suffix;
my $patch_file_name = $data2_dir_name . 'modified.txt';

our $UseCVS //= 1;

sub normalize ($) {
  my $s = shift;
  $s =~ s/\s+/ /g;
  $s =~ s/^ //;
  $s =~ s/ $//g;
  return $s;
} # normalize

sub get_hash ($) {
  require Digest::MD5;
  return Digest::MD5::md5_hex (normalize ($_[0]));
} # get_hash

sub create_pattern1 ($) {
  my $s = quotemeta shift;
  $s =~ s/\\\*/(.+)/g;
  return $s;
} # create_pattern1

sub replace_pattern2 ($@) {
  my $s = shift;
  my @arg = @_;
  $s =~ s/\$(\d+)/$arg[$1 - 1]/g;
  return $s;
} # replace_pattern2

sub load_data_file ($$$) {
  my ($data_file_name, $exact_data, $pattern_data) = @_;

  open my $data_file, '<:utf8', $data_file_name
      or die "$0: $data_file_name: $!";
  local $/ = undef;
  my $data = <$data_file>;
  $data =~ s/\x0D?\x0A/\n/g;
  for (split /\n\n+(?=#)/, $data) {
    my ($en, $ja) = split /\n#ja\n/, $_;
    if ($en =~ s/^#en\n//) {
      $exact_data->{normalize ($en)} = $ja;
    } elsif ($en =~ s/^#pattern\n//) {
      $pattern_data->{create_pattern1 (normalize ($en))} = $ja;
    }
  }
} # load_data_file

sub for_each_data_file ($) {
  my ($code) = @_;

  opendir my $data_dir, $data_dir_name or die "$0: $data_dir_name: $!";
  for (sort {$a cmp $b} readdir $data_dir) {
    next if /^\./; # hidden files
    next if /^_/; # editable but not-used files
    my $data_file_name = qq[$data_dir_name$_];
    next unless $data_file_name =~ /\Q$data_suffix\E$/;
    $code->($data_file_name, $_);
  }
} # for_each_data_file

sub read_data_file ($) {
  my $file_name = shift;
  if (-f $file_name) {
    warn "Loading $file_name...\n";
    return do $file_name;
  } else {
    warn "File $file_name not found\n";
    return {};
  }
} # read_data_file

sub write_data_file ($$) {
  my ($file_name, $data) = @_;

  require Data::Dumper;
  local $Data::Dumper::Sortkeys = 1;

  my $had_file = -f $file_name;
  open my $file, '>:encoding(utf8)', $file_name or die "$0: $file_name: $!";
  print $file Data::Dumper::Dumper ($data);
  close $file;
  unless ($had_file) {
    system_ ('cvs', 'add', $file_name) if $UseCVS;
  }
} # write_data_file

sub hash_to_file_name ($) {
  return $data2_dir_name . substr ($_[0], 0, 2) . $data2_suffix;
} # hash_to_file_name

my $Entry = {};
my $ModifiedHash = {};

sub get_entry ($) {
  my $hash = shift;
  
  my $file_name = hash_to_file_name ($hash);
  unless ($Entry->{$file_name}) {
    $Entry->{$file_name} = read_data_file ($file_name);
  }
  
  if ($Entry->{$file_name}->{exact}->{$hash}) {
    return (0, $Entry->{$file_name}->{exact}->{$hash});
  } elsif ($Entry->{$file_name}->{pattern}->{$hash}) {
    return (1, $Entry->{$file_name}->{pattern}->{$hash});
  } else {
    return (undef, undef);
  }
} # get_entry

sub set_entry ($$$) {
  my ($hash, $is_pattern, $value) = @_;
  
  my $file_name = hash_to_file_name ($hash);
  unless ($Entry->{$file_name}) {
    $Entry->{$file_name} = read_data_file ($file_name);    
  }

  unless ($value) {
    delete $Entry->{$file_name}->{exact}->{$hash};
    delete $Entry->{$file_name}->{pattern}->{$hash};
  } elsif ($is_pattern) {
    delete $Entry->{$file_name}->{exact}->{$hash};
    $Entry->{$file_name}->{pattern}->{$hash} = $value;
  } else {
    $Entry->{$file_name}->{exact}->{$hash} = $value;
    delete $Entry->{$file_name}->{pattern}->{$hash};
  }
  $Entry->{$file_name}->{modified} = 1;
  $ModifiedHash->{$hash} = 1;
} # set_entry

use Fcntl ':flock';
my $Lock;

sub lock_entry ($) {
  if ($Lock) {
    die "$0: lock_entry: Another entry is locked";
  }

  my $hash = shift;
  my $file_name = hash_to_file_name ($hash) . $lock_suffix;
  open $Lock, '>', $file_name or die "$0: $file_name: $!";
  flock $Lock, LOCK_EX;
} # lock_entry

sub commit_entries ($) {
  for my $file_name (keys %{$Entry}) {
    if ($Entry->{$file_name}->{modified}) {
      delete $Entry->{$file_name}->{modified};
      write_data_file ($file_name => $Entry->{$file_name});
    }
  }

  open my $file, '>>', $patch_file_name or die "$0: $patch_file_name: $!";
  for (keys %$ModifiedHash) {
    print $file "$_\n";
  }
  close $file;

  my $msg = shift // $0;
  system_ ('cvs', 'commit', -m => $msg, $data2_dir_name) if $UseCVS;
} # commit_entries

sub get_all_entries () {
  opendir my $dir, $data2_dir_name or die "$0: $data2_dir_name: $!";
  for (readdir $dir) {
    next unless /\Q$data2_suffix\E$/;
    my $file_name = $data2_dir_name . $_;
    next if $Entry->{$file_name};

    $Entry->{$file_name} = read_data_file ($file_name);
  }

  return $Entry;
} # get_all_entries

my $FallbackEntry;
sub get_fallback_entry ($) {
  my $hash = shift;
  unless (defined $FallbackEntry) {
    $FallbackEntry = read_data_file ($fallback_file_name);
  }
  return $FallbackEntry->{$hash} // {};
} # get_fallback_entry

sub get_entry_or_fallback_entry ($) {
  my $hash = shift;

  my ($is_pattern, $entry) = get_entry ($hash);
  unless (defined $entry->{en}) {
    $entry = get_fallback_entry ($hash);
  }
  $entry->{tags} ||= [] if defined $entry->{en};
  $entry->{isPattern} = 1 if $is_pattern;

  return $entry;
} # get_entry_or_fallback_entry

sub set_fallback_entry ($$) {
  my ($hash, $value) = @_;
  unless (defined $FallbackEntry) {
    $FallbackEntry = read_data_file ($fallback_file_name);
  }
  $FallbackEntry->{$hash} = $value;
} # set_fallback_entry

sub clear_fallback_entries () {
  $FallbackEntry = {};
} # clear_fallback_entries

sub save_fallback_entries () {
  write_data_file ($fallback_file_name => $FallbackEntry)
      if defined $FallbackEntry;
} # save_fallback_entries

sub get_modified_hashes () {
  open my $file, '<', $patch_file_name or die "$0: $patch_file_name: $!";
  return map {tr/\x0D\x0A//d; $_} <$file>;
} # get_modified_hashes

sub clear_modified_hashes () {
  open my $file, '>', $patch_file_name;
  close $file;
} # clear_modified_hashes

sub htescape ($) {
  my $s = shift;
  $s =~ s/&/&amp;/g;
  $s =~ s/</&lt;/g;
  $s =~ s/"/&quot;/g;
  return $s;
} # htescape

sub system_ (@) {
  (system join (' ', map {quotemeta $_} @_) . " > /dev/null") == 0
      or die "$0: $?";
} # system_

1;

1	wakaba	1.1	use strict;
2
3			my $data_suffix = q[.dat];
4	wakaba	1.7	my $data_dir_name = q[data/];
5	wakaba	1.8	my $data2_dir_name = q[data2/];
6			my $data2_suffix = q[.dat];
7			my $lock_suffix = q[.lock];
8			my $fallback_file_name = $data2_dir_name . 'fallback' . $data2_suffix;
9	wakaba	1.11	my $patch_file_name = $data2_dir_name . 'modified.txt';
10	wakaba	1.8
11			our $UseCVS //= 1;
12	wakaba	1.1
13			sub normalize ($) {
14			my $s = shift;
15			$s =~ s/\s+/ /g;
16			$s =~ s/^ //;
17			$s =~ s/ $//g;
18			return $s;
19			} # normalize
20
21	wakaba	1.8	sub get_hash ($) {
22			require Digest::MD5;
23			return Digest::MD5::md5_hex (normalize ($_[0]));
24			} # get_hash
25
26	wakaba	1.1	sub create_pattern1 ($) {
27			my $s = quotemeta shift;
28			$s =~ s/\\\*/(.+)/g;
29			return $s;
30			} # create_pattern1
31
32			sub replace_pattern2 ($@) {
33			my $s = shift;
34			my @arg = @_;
35			$s =~ s/\$(\d+)/$arg[$1 - 1]/g;
36			return $s;
37			} # replace_pattern2
38
39	wakaba	1.5	sub load_data_file ($$$) {
40			my ($data_file_name, $exact_data, $pattern_data) = @_;
41	wakaba	1.1
42			open my $data_file, '<:utf8', $data_file_name
43			or die "$0: $data_file_name: $!";
44			local $/ = undef;
45			my $data = <$data_file>;
46			$data =~ s/\x0D?\x0A/\n/g;
47			for (split /\n\n+(?=#)/, $data) {
48			my ($en, $ja) = split /\n#ja\n/, $_;
49			if ($en =~ s/^#en\n//) {
50			$exact_data->{normalize ($en)} = $ja;
51			} elsif ($en =~ s/^#pattern\n//) {
52			$pattern_data->{create_pattern1 (normalize ($en))} = $ja;
53			}
54			}
55			} # load_data_file
56
57			sub for_each_data_file ($) {
58			my ($code) = @_;
59
60			opendir my $data_dir, $data_dir_name or die "$0: $data_dir_name: $!";
61			for (sort {$a cmp $b} readdir $data_dir) {
62			next if /^\./; # hidden files
63	wakaba	1.2	next if /^_/; # editable but not-used files
64	wakaba	1.1	my $data_file_name = qq[$data_dir_name$_];
65			next unless $data_file_name =~ /\Q$data_suffix\E$/;
66	wakaba	1.3	$code->($data_file_name, $_);
67	wakaba	1.1	}
68			} # for_each_data_file
69	wakaba	1.8
70			sub read_data_file ($) {
71			my $file_name = shift;
72			if (-f $file_name) {
73			warn "Loading $file_name...\n";
74			return do $file_name;
75			} else {
76			warn "File $file_name not found\n";
77			return {};
78			}
79			} # read_data_file
80
81			sub write_data_file ($$) {
82			my ($file_name, $data) = @_;
83
84			require Data::Dumper;
85			local $Data::Dumper::Sortkeys = 1;
86
87			my $had_file = -f $file_name;
88			open my $file, '>:encoding(utf8)', $file_name or die "$0: $file_name: $!";
89			print $file Data::Dumper::Dumper ($data);
90			close $file;
91			unless ($had_file) {
92	apache	1.9	system_ ('cvs', 'add', $file_name) if $UseCVS;
93	wakaba	1.8	}
94			} # write_data_file
95
96			sub hash_to_file_name ($) {
97			return $data2_dir_name . substr ($_[0], 0, 2) . $data2_suffix;
98			} # hash_to_file_name
99
100			my $Entry = {};
101	wakaba	1.11	my $ModifiedHash = {};
102	wakaba	1.8
103			sub get_entry ($) {
104			my $hash = shift;
105
106			my $file_name = hash_to_file_name ($hash);
107			unless ($Entry->{$file_name}) {
108			$Entry->{$file_name} = read_data_file ($file_name);
109			}
110
111			if ($Entry->{$file_name}->{exact}->{$hash}) {
112			return (0, $Entry->{$file_name}->{exact}->{$hash});
113			} elsif ($Entry->{$file_name}->{pattern}->{$hash}) {
114			return (1, $Entry->{$file_name}->{pattern}->{$hash});
115			} else {
116			return (undef, undef);
117			}
118			} # get_entry
119
120			sub set_entry ($$$) {
121			my ($hash, $is_pattern, $value) = @_;
122
123			my $file_name = hash_to_file_name ($hash);
124			unless ($Entry->{$file_name}) {
125			$Entry->{$file_name} = read_data_file ($file_name);
126			}
127
128			unless ($value) {
129			delete $Entry->{$file_name}->{exact}->{$hash};
130			delete $Entry->{$file_name}->{pattern}->{$hash};
131			} elsif ($is_pattern) {
132			delete $Entry->{$file_name}->{exact}->{$hash};
133			$Entry->{$file_name}->{pattern}->{$hash} = $value;
134			} else {
135			$Entry->{$file_name}->{exact}->{$hash} = $value;
136			delete $Entry->{$file_name}->{pattern}->{$hash};
137			}
138			$Entry->{$file_name}->{modified} = 1;
139	wakaba	1.11	$ModifiedHash->{$hash} = 1;
140	wakaba	1.8	} # set_entry
141
142			use Fcntl ':flock';
143			my $Lock;
144
145			sub lock_entry ($) {
146			if ($Lock) {
147			die "$0: lock_entry: Another entry is locked";
148			}
149
150			my $hash = shift;
151			my $file_name = hash_to_file_name ($hash) . $lock_suffix;
152			open $Lock, '>', $file_name or die "$0: $file_name: $!";
153			flock $Lock, LOCK_EX;
154			} # lock_entry
155
156			sub commit_entries ($) {
157			for my $file_name (keys %{$Entry}) {
158			if ($Entry->{$file_name}->{modified}) {
159	wakaba	1.11	delete $Entry->{$file_name}->{modified};
160	wakaba	1.8	write_data_file ($file_name => $Entry->{$file_name});
161			}
162			}
163
164	wakaba	1.11	open my $file, '>>', $patch_file_name or die "$0: $patch_file_name: $!";
165			for (keys %$ModifiedHash) {
166			print $file "$_\n";
167			}
168			close $file;
169
170	wakaba	1.8	my $msg = shift // $0;
171	wakaba	1.10	system_ ('cvs', 'commit', -m => $msg, $data2_dir_name) if $UseCVS;
172	wakaba	1.8	} # commit_entries
173
174			sub get_all_entries () {
175			opendir my $dir, $data2_dir_name or die "$0: $data2_dir_name: $!";
176			for (readdir $dir) {
177			next unless /\Q$data2_suffix\E$/;
178			my $file_name = $data2_dir_name . $_;
179			next if $Entry->{$file_name};
180
181			$Entry->{$file_name} = read_data_file ($file_name);
182			}
183
184			return $Entry;
185			} # get_all_entries
186
187			my $FallbackEntry;
188			sub get_fallback_entry ($) {
189			my $hash = shift;
190			unless (defined $FallbackEntry) {
191			$FallbackEntry = read_data_file ($fallback_file_name);
192			}
193			return $FallbackEntry->{$hash} // {};
194			} # get_fallback_entry
195
196	wakaba	1.11	sub get_entry_or_fallback_entry ($) {
197			my $hash = shift;
198
199			my ($is_pattern, $entry) = get_entry ($hash);
200			unless (defined $entry->{en}) {
201			$entry = get_fallback_entry ($hash);
202			}
203			$entry->{tags} \|\|= [] if defined $entry->{en};
204			$entry->{isPattern} = 1 if $is_pattern;
205
206			return $entry;
207			} # get_entry_or_fallback_entry
208
209	wakaba	1.8	sub set_fallback_entry ($$) {
210			my ($hash, $value) = @_;
211			unless (defined $FallbackEntry) {
212			$FallbackEntry = read_data_file ($fallback_file_name);
213			}
214			$FallbackEntry->{$hash} = $value;
215			} # set_fallback_entry
216
217			sub clear_fallback_entries () {
218			$FallbackEntry = {};
219			} # clear_fallback_entries
220
221			sub save_fallback_entries () {
222			write_data_file ($fallback_file_name => $FallbackEntry)
223			if defined $FallbackEntry;
224			} # save_fallback_entries
225	wakaba	1.11
226			sub get_modified_hashes () {
227			open my $file, '<', $patch_file_name or die "$0: $patch_file_name: $!";
228			return map {tr/\x0D\x0A//d; $_} <$file>;
229			} # get_modified_hashes
230
231			sub clear_modified_hashes () {
232			open my $file, '>', $patch_file_name;
233			close $file;
234			} # clear_modified_hashes
235	wakaba	1.8
236			sub htescape ($) {
237			my $s = shift;
238			$s =~ s/&/&/g;
239			$s =~ s/</</g;
240			$s =~ s/"/"/g;
241			return $s;
242			} # htescape
243	apache	1.9
244			sub system_ (@) {
245	wakaba	1.10	(system join (' ', map {quotemeta $_} @_) . " > /dev/null") == 0
246			or die "$0: $?";
247	apache	1.9	} # system_
248	wakaba	1.1
249			1;
250