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;

## SEE ALSO: |Makefile|.
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	1.12
10			## SEE ALSO: \|Makefile\|.
11	wakaba	1.11	my $patch_file_name = $data2_dir_name . 'modified.txt';
12	wakaba	1.8
13			our $UseCVS //= 1;
14	wakaba	1.1
15			sub normalize ($) {
16			my $s = shift;
17			$s =~ s/\s+/ /g;
18			$s =~ s/^ //;
19			$s =~ s/ $//g;
20			return $s;
21			} # normalize
22
23	wakaba	1.8	sub get_hash ($) {
24			require Digest::MD5;
25			return Digest::MD5::md5_hex (normalize ($_[0]));
26			} # get_hash
27
28	wakaba	1.1	sub create_pattern1 ($) {
29			my $s = quotemeta shift;
30			$s =~ s/\\\*/(.+)/g;
31			return $s;
32			} # create_pattern1
33
34			sub replace_pattern2 ($@) {
35			my $s = shift;
36			my @arg = @_;
37			$s =~ s/\$(\d+)/$arg[$1 - 1]/g;
38			return $s;
39			} # replace_pattern2
40
41	wakaba	1.5	sub load_data_file ($$$) {
42			my ($data_file_name, $exact_data, $pattern_data) = @_;
43	wakaba	1.1
44			open my $data_file, '<:utf8', $data_file_name
45			or die "$0: $data_file_name: $!";
46			local $/ = undef;
47			my $data = <$data_file>;
48			$data =~ s/\x0D?\x0A/\n/g;
49			for (split /\n\n+(?=#)/, $data) {
50			my ($en, $ja) = split /\n#ja\n/, $_;
51			if ($en =~ s/^#en\n//) {
52			$exact_data->{normalize ($en)} = $ja;
53			} elsif ($en =~ s/^#pattern\n//) {
54			$pattern_data->{create_pattern1 (normalize ($en))} = $ja;
55			}
56			}
57			} # load_data_file
58
59			sub for_each_data_file ($) {
60			my ($code) = @_;
61
62			opendir my $data_dir, $data_dir_name or die "$0: $data_dir_name: $!";
63			for (sort {$a cmp $b} readdir $data_dir) {
64			next if /^\./; # hidden files
65	wakaba	1.2	next if /^_/; # editable but not-used files
66	wakaba	1.1	my $data_file_name = qq[$data_dir_name$_];
67			next unless $data_file_name =~ /\Q$data_suffix\E$/;
68	wakaba	1.3	$code->($data_file_name, $_);
69	wakaba	1.1	}
70			} # for_each_data_file
71	wakaba	1.8
72			sub read_data_file ($) {
73			my $file_name = shift;
74			if (-f $file_name) {
75			warn "Loading $file_name...\n";
76			return do $file_name;
77			} else {
78			warn "File $file_name not found\n";
79			return {};
80			}
81			} # read_data_file
82
83			sub write_data_file ($$) {
84			my ($file_name, $data) = @_;
85
86			require Data::Dumper;
87			local $Data::Dumper::Sortkeys = 1;
88
89			my $had_file = -f $file_name;
90			open my $file, '>:encoding(utf8)', $file_name or die "$0: $file_name: $!";
91			print $file Data::Dumper::Dumper ($data);
92			close $file;
93			unless ($had_file) {
94	apache	1.9	system_ ('cvs', 'add', $file_name) if $UseCVS;
95	wakaba	1.8	}
96			} # write_data_file
97
98			sub hash_to_file_name ($) {
99			return $data2_dir_name . substr ($_[0], 0, 2) . $data2_suffix;
100			} # hash_to_file_name
101
102			my $Entry = {};
103	wakaba	1.11	my $ModifiedHash = {};
104	wakaba	1.8
105			sub get_entry ($) {
106			my $hash = shift;
107
108			my $file_name = hash_to_file_name ($hash);
109			unless ($Entry->{$file_name}) {
110			$Entry->{$file_name} = read_data_file ($file_name);
111			}
112
113			if ($Entry->{$file_name}->{exact}->{$hash}) {
114			return (0, $Entry->{$file_name}->{exact}->{$hash});
115			} elsif ($Entry->{$file_name}->{pattern}->{$hash}) {
116			return (1, $Entry->{$file_name}->{pattern}->{$hash});
117			} else {
118			return (undef, undef);
119			}
120			} # get_entry
121
122			sub set_entry ($$$) {
123			my ($hash, $is_pattern, $value) = @_;
124
125			my $file_name = hash_to_file_name ($hash);
126			unless ($Entry->{$file_name}) {
127			$Entry->{$file_name} = read_data_file ($file_name);
128			}
129
130			unless ($value) {
131			delete $Entry->{$file_name}->{exact}->{$hash};
132			delete $Entry->{$file_name}->{pattern}->{$hash};
133			} elsif ($is_pattern) {
134			delete $Entry->{$file_name}->{exact}->{$hash};
135			$Entry->{$file_name}->{pattern}->{$hash} = $value;
136			} else {
137			$Entry->{$file_name}->{exact}->{$hash} = $value;
138			delete $Entry->{$file_name}->{pattern}->{$hash};
139			}
140			$Entry->{$file_name}->{modified} = 1;
141	wakaba	1.11	$ModifiedHash->{$hash} = 1;
142	wakaba	1.8	} # set_entry
143
144			use Fcntl ':flock';
145			my $Lock;
146
147			sub lock_entry ($) {
148			if ($Lock) {
149			die "$0: lock_entry: Another entry is locked";
150			}
151
152			my $hash = shift;
153			my $file_name = hash_to_file_name ($hash) . $lock_suffix;
154			open $Lock, '>', $file_name or die "$0: $file_name: $!";
155			flock $Lock, LOCK_EX;
156			} # lock_entry
157
158			sub commit_entries ($) {
159			for my $file_name (keys %{$Entry}) {
160			if ($Entry->{$file_name}->{modified}) {
161	wakaba	1.11	delete $Entry->{$file_name}->{modified};
162	wakaba	1.8	write_data_file ($file_name => $Entry->{$file_name});
163			}
164			}
165
166	wakaba	1.11	open my $file, '>>', $patch_file_name or die "$0: $patch_file_name: $!";
167			for (keys %$ModifiedHash) {
168			print $file "$_\n";
169			}
170			close $file;
171
172	wakaba	1.8	my $msg = shift // $0;
173	wakaba	1.10	system_ ('cvs', 'commit', -m => $msg, $data2_dir_name) if $UseCVS;
174	wakaba	1.8	} # commit_entries
175
176			sub get_all_entries () {
177			opendir my $dir, $data2_dir_name or die "$0: $data2_dir_name: $!";
178			for (readdir $dir) {
179			next unless /\Q$data2_suffix\E$/;
180			my $file_name = $data2_dir_name . $_;
181			next if $Entry->{$file_name};
182
183			$Entry->{$file_name} = read_data_file ($file_name);
184			}
185
186			return $Entry;
187			} # get_all_entries
188
189			my $FallbackEntry;
190			sub get_fallback_entry ($) {
191			my $hash = shift;
192			unless (defined $FallbackEntry) {
193			$FallbackEntry = read_data_file ($fallback_file_name);
194			}
195			return $FallbackEntry->{$hash} // {};
196			} # get_fallback_entry
197
198	wakaba	1.11	sub get_entry_or_fallback_entry ($) {
199			my $hash = shift;
200
201			my ($is_pattern, $entry) = get_entry ($hash);
202			unless (defined $entry->{en}) {
203			$entry = get_fallback_entry ($hash);
204			}
205			$entry->{tags} \|\|= [] if defined $entry->{en};
206			$entry->{isPattern} = 1 if $is_pattern;
207
208			return $entry;
209			} # get_entry_or_fallback_entry
210
211	wakaba	1.8	sub set_fallback_entry ($$) {
212			my ($hash, $value) = @_;
213			unless (defined $FallbackEntry) {
214			$FallbackEntry = read_data_file ($fallback_file_name);
215			}
216			$FallbackEntry->{$hash} = $value;
217			} # set_fallback_entry
218
219			sub clear_fallback_entries () {
220			$FallbackEntry = {};
221			} # clear_fallback_entries
222
223			sub save_fallback_entries () {
224			write_data_file ($fallback_file_name => $FallbackEntry)
225			if defined $FallbackEntry;
226			} # save_fallback_entries
227	wakaba	1.11
228			sub get_modified_hashes () {
229			open my $file, '<', $patch_file_name or die "$0: $patch_file_name: $!";
230			return map {tr/\x0D\x0A//d; $_} <$file>;
231			} # get_modified_hashes
232
233			sub clear_modified_hashes () {
234			open my $file, '>', $patch_file_name;
235			close $file;
236			} # clear_modified_hashes
237	wakaba	1.8
238			sub htescape ($) {
239			my $s = shift;
240			$s =~ s/&/&/g;
241			$s =~ s/</</g;
242			$s =~ s/"/"/g;
243			return $s;
244			} # htescape
245	apache	1.9
246			sub system_ (@) {
247	wakaba	1.10	(system join (' ', map {quotemeta $_} @_) . " > /dev/null") == 0
248			or die "$0: $?";
249	apache	1.9	} # system_
250	wakaba	1.1
251			1;
252