C4/Languages.pm

   1 package C4::Languages;
   2
   3 # Copyright 2006 (C) LibLime
   4 # Joshua Ferraro <jmf@liblime.com>
   5 # Portions Copyright 2009 Chris Cormack and the Koha Dev Team
   6 # This file is part of Koha.
   7 #
   8 # Koha is free software; you can redistribute it and/or modify it under the
   9 # terms of the GNU General Public License as published by the Free Software
  10 # Foundation; either version 2 of the License, or (at your option) any later
  11 # version.
  12 #
  13 # Koha is distributed in the hope that it will be useful, but WITHOUT ANY
  14 # WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
  15 # A PARTICULAR PURPOSE.  See the GNU General Public License for more details.
  16 #
  17 # You should have received a copy of the GNU General Public License along with
  18 # Koha; if not, write to the Free Software Foundation, Inc., 59 Temple Place,
  19 # Suite 330, Boston, MA  02111-1307 USA
  20
  21
  22 use strict;
  23 #use warnings;   #FIXME: turn off warnings before release
  24 use Carp;
  25 use C4::Context;
  26 use vars qw($VERSION @ISA @EXPORT @EXPORT_OK %EXPORT_TAGS $DEBUG);
  27
  28 eval {
  29     my $servers = C4::Context->config('memcached_servers');
  30     if ($servers) {
  31         require Memoize::Memcached;
  32         import Memoize::Memcached qw(memoize_memcached);
  33
  34         my $memcached = {
  35             servers    => [ $servers ],
  36             key_prefix => C4::Context->config('memcached_namespace') || 'koha',
  37         };
  38
  39         memoize_memcached('getTranslatedLanguages', memcached => $memcached, expire_time => 600); #cache for 10 minutes
  40         memoize_memcached('getFrameworkLanguages' , memcached => $memcached, expire_time => 600);
  41         memoize_memcached('getAllLanguages',        memcached => $memcached, expire_time => 600);
  42     }
  43 };
  44
  45 BEGIN {
  46     $VERSION = 3.00;
  47     require Exporter;
  48     @ISA    = qw(Exporter);
  49     @EXPORT = qw(
  50         &getFrameworkLanguages
  51         &getTranslatedLanguages
  52         &getAllLanguages
  53     );
  54     @EXPORT_OK = qw(getFrameworkLanguages getTranslatedLanguages getAllLanguages get_bidi regex_lang_subtags language_get_description accept_language);
  55     $DEBUG = 0;
  56 }
  57
  58 =head1 NAME
  59
  60 C4::Languages - Perl Module containing language list functions for Koha
  61
  62 =head1 SYNOPSIS
  63
  64 use C4::Languages;
  65
  66 =head1 DESCRIPTION
  67
  68 =head1 FUNCTIONS
  69
  70 =head2 getFrameworkLanguages
  71
  72 Returns a reference to an array of hashes:
  73
  74  my $languages = getFrameworkLanguages();
  75  for my $language(@$languages) {
  76     print "$language->{language_code}\n"; # language code in iso 639-2
  77     print "$language->{language_name}\n"; # language name in native script
  78     print "$language->{language_locale_name}\n"; # language name in current locale
  79  }
  80
  81 =cut
  82
  83 sub getFrameworkLanguages {
  84     # get a hash with all language codes, names, and locale names
  85     my $all_languages = getAllLanguages();
  86     my @languages;
  87
  88     # find the available directory names
  89     my $dir=C4::Context->config('intranetdir')."/installer/data/";
  90     opendir (MYDIR,$dir);
  91     my @listdir= grep { !/^\.|CVS/ && -d "$dir/$_"} readdir(MYDIR);
  92     closedir MYDIR;
  93
  94     # pull out all data for the dir names that exist
  95     for my $dirname (@listdir) {
  96         for my $language_set (@$all_languages) {
  97
  98             if ($dirname eq $language_set->{language_code}) {
  99                 push @languages, {
 100                     'language_code'=>$dirname,
 101                     'language_description'=>$language_set->{language_description},
 102                     'native_descrition'=>$language_set->{language_native_description} }
 103             }
 104         }
 105     }
 106     return \@languages;
 107 }
 108
 109 =head2 getTranslatedLanguages
 110
 111 Returns a reference to an array of hashes:
 112
 113  my $languages = getTranslatedLanguages();
 114  print "Available translated languages:\n";
 115  for my $language(@$trlanguages) {
 116     print "$language->{language_code}\n"; # language code in iso 639-2
 117     print "$language->{language_name}\n"; # language name in native script
 118     print "$language->{language_locale_name}\n"; # language name in current locale
 119  }
 120
 121 =cut
 122
 123 sub getTranslatedLanguages {
 124     my ($interface, $theme, $current_language, $which) = @_;
 125     my $htdocs;
 126     my $all_languages = getAllLanguages();
 127     my @languages;
 128     my @enabled_languages;
 129
 130     if ($interface && $interface eq 'opac' ) {
 131         @enabled_languages = split ",", C4::Context->preference('opaclanguages');
 132         $htdocs = C4::Context->config('opachtdocs');
 133         if ( $theme and -d "$htdocs/$theme" ) {
 134             (@languages) = _get_language_dirs($htdocs,$theme);
 135         }
 136         else {
 137             for my $theme ( _get_themes('opac') ) {
 138                 push @languages, _get_language_dirs($htdocs,$theme);
 139             }
 140         }
 141     }
 142     elsif ($interface && $interface eq 'intranet' ) {
 143         @enabled_languages = split ",", C4::Context->preference('language');
 144         $htdocs = C4::Context->config('intrahtdocs');
 145         if ( $theme and -d "$htdocs/$theme" ) {
 146             @languages = _get_language_dirs($htdocs,$theme);
 147         }
 148         else {
 149             foreach my $theme ( _get_themes('intranet') ) {
 150                 push @languages, _get_language_dirs($htdocs,$theme);
 151             }
 152         }
 153     }
 154     else {
 155         @enabled_languages = split ",", C4::Context->preference('opaclanguages');
 156         my $htdocs = C4::Context->config('intrahtdocs');
 157         foreach my $theme ( _get_themes('intranet') ) {
 158             push @languages, _get_language_dirs($htdocs,$theme);
 159         }
 160         $htdocs = C4::Context->config('opachtdocs');
 161         foreach my $theme ( _get_themes('opac') ) {
 162             push @languages, _get_language_dirs($htdocs,$theme);
 163         }
 164         my %seen;
 165         $seen{$_}++ for @languages;
 166         @languages = keys %seen;
 167     }
 168     return _build_languages_arrayref($all_languages,\@languages,$current_language,\@enabled_languages);
 169 }
 170
 171 =head2 getAllLanguages
 172
 173 Returns a reference to an array of hashes:
 174
 175  my $alllanguages = getAllLanguages();
 176  print "Available translated languages:\n";
 177  for my $language(@$alllanguages) {
 178     print "$language->{language_code}\n";
 179     print "$language->{language_name}\n";
 180     print "$language->{language_locale_name}\n";
 181  }
 182
 183 =cut
 184
 185 sub getAllLanguages {
 186     my @languages_loop;
 187     my $dbh=C4::Context->dbh;
 188     my $current_language = shift || 'en';
 189     my $sth = $dbh->prepare('SELECT * FROM language_subtag_registry WHERE type=\'language\'');
 190     $sth->execute();
 191     while (my $language_subtag_registry = $sth->fetchrow_hashref) {
 192
 193         # pull out all the script descriptions for each language
 194         my $sth2= $dbh->prepare("SELECT * FROM language_descriptions LEFT JOIN language_rfc4646_to_iso639 on language_rfc4646_to_iso639.rfc4646_subtag = language_descriptions.subtag WHERE type='language' AND subtag =? AND language_descriptions.lang = ?");
 195         $sth2->execute($language_subtag_registry->{subtag},$current_language);
 196
 197         my $sth3 = $dbh->prepare("SELECT description FROM language_descriptions WHERE type='language' AND subtag=? AND lang=?");
 198
 199         # add the correct description info
 200         while (my $language_descriptions = $sth2->fetchrow_hashref) {
 201             $sth3->execute($language_subtag_registry->{subtag},$language_subtag_registry->{subtag});
 202             my $native_description;
 203             while (my $description = $sth3->fetchrow_hashref) {
 204                 $native_description = $description->{description};
 205             }
 206
 207             # fill in the ISO6329 code
 208             $language_subtag_registry->{iso639_2_code} = $language_descriptions->{iso639_2_code};
 209             # fill in the native description of the language, as well as the current language's translation of that if it exists
 210             if ($native_description) {
 211                 $language_subtag_registry->{language_description} = $native_description;
 212                 $language_subtag_registry->{language_description}.=" ($language_descriptions->{description})" if $language_descriptions->{description};
 213             }
 214             else {
 215                 $language_subtag_registry->{language_description} = $language_descriptions->{description};
 216             }
 217         }
 218         push @languages_loop, $language_subtag_registry;
 219     }
 220     return \@languages_loop;
 221 }
 222
 223 =head2 _get_themes
 224
 225 Internal function, returns an array of all available themes.
 226
 227   (@themes) = &_get_themes('opac');
 228   (@themes) = &_get_themes('intranet');
 229
 230 =cut
 231
 232 sub _get_themes {
 233     my $interface = shift;
 234     my $htdocs;
 235     my @themes;
 236     if ( $interface eq 'intranet' ) {
 237         $htdocs = C4::Context->config('intrahtdocs');
 238     }
 239     else {
 240         $htdocs = C4::Context->config('opachtdocs');
 241     }
 242     opendir D, "$htdocs";
 243     my @dirlist = readdir D;
 244     foreach my $directory (@dirlist) {
 245         # if there's an en dir, it's a valid theme
 246         -d "$htdocs/$directory/en" and push @themes, $directory;
 247     }
 248     return @themes;
 249 }
 250
 251 =head2 _get_language_dirs
 252
 253 Internal function, returns an array of directory names, excluding non-language directories
 254
 255 =cut
 256
 257 sub _get_language_dirs {
 258     my ($htdocs,$theme) = @_;
 259     my @lang_strings;
 260     opendir D, "$htdocs/$theme";
 261     for my $lang_string ( readdir D ) {
 262         next if $lang_string =~/^\./;
 263         next if $lang_string eq 'all';
 264         next if $lang_string =~/png$/;
 265         next if $lang_string =~/css$/;
 266         next if $lang_string =~/CVS$/;
 267         next if $lang_string =~/\.txt$/i;     #Don't read the readme.txt !
 268         next if $lang_string =~/img|images|famfam|sound/;
 269         push @lang_strings, $lang_string;
 270     }
 271         return (@lang_strings);
 272 }
 273
 274 =head2 _build_languages_arrayref
 275
 276 Internal function for building the ref to array of hashes
 277
 278 FIXME: this could be rewritten and simplified using map
 279
 280 =cut
 281
 282 sub _build_languages_arrayref {
 283         my ($all_languages,$translated_languages,$current_language,$enabled_languages) = @_;
 284         my @translated_languages = @$translated_languages;
 285         my @languages_loop; # the final reference to an array of hashrefs
 286         my @enabled_languages = @$enabled_languages;
 287         # how many languages are enabled, if one, take note, some contexts won't need to display it
 288         my %seen_languages; # the language tags we've seen
 289         my %found_languages;
 290         my $language_groups;
 291         my $track_language_groups;
 292         my $current_language_regex = regex_lang_subtags($current_language);
 293         # Loop through the translated languages
 294         for my $translated_language (@translated_languages) {
 295             # separate the language string into its subtag types
 296             my $language_subtags_hashref = regex_lang_subtags($translated_language);
 297
 298             # is this language string 'enabled'?
 299             for my $enabled_language (@enabled_languages) {
 300                 #warn "Checking out if $translated_language eq $enabled_language";
 301                 $language_subtags_hashref->{'enabled'} = 1 if $translated_language eq $enabled_language;
 302             }
 303
 304             # group this language, key by langtag
 305             $language_subtags_hashref->{'sublanguage_current'} = 1 if $translated_language eq $current_language;
 306             $language_subtags_hashref->{'rfc4646_subtag'} = $translated_language;
 307             $language_subtags_hashref->{'native_description'} = language_get_description($language_subtags_hashref->{language},$language_subtags_hashref->{language},'language');
 308             $language_subtags_hashref->{'script_description'} = language_get_description($language_subtags_hashref->{script},$language_subtags_hashref->{'language'},'script');
 309             $language_subtags_hashref->{'region_description'} = language_get_description($language_subtags_hashref->{region},$language_subtags_hashref->{'language'},'region');
 310             $language_subtags_hashref->{'variant_description'} = language_get_description($language_subtags_hashref->{variant},$language_subtags_hashref->{'language'},'variant');
 311             $track_language_groups->{$language_subtags_hashref->{'language'}}++;
 312             push ( @{ $language_groups->{$language_subtags_hashref->{language}} }, $language_subtags_hashref );
 313         }
 314         # $key is a language subtag like 'en'
 315         while( my ($key, $value) = each %$language_groups) {
 316
 317             # is this language group enabled? are any of the languages within it enabled?
 318             my $enabled;
 319             for my $enabled_language (@enabled_languages) {
 320                 my $regex_enabled_language = regex_lang_subtags($enabled_language);
 321                 $enabled = 1 if $key eq $regex_enabled_language->{language};
 322             }
 323             push @languages_loop,  {
 324                             # this is only use if there is one
 325                             rfc4646_subtag => @$value[0]->{rfc4646_subtag},
 326                             native_description => language_get_description($key,$key,'language'),
 327                             language => $key,
 328                             sublanguages_loop => $value,
 329                             plural => $track_language_groups->{$key} >1 ? 1 : 0,
 330                             current => $current_language_regex->{language} eq $key ? 1 : 0,
 331                             group_enabled => $enabled,
 332                            };
 333         }
 334         return \@languages_loop;
 335 }
 336
 337 sub language_get_description {
 338     my ($script,$lang,$type) = @_;
 339     my $dbh = C4::Context->dbh;
 340     my $desc;
 341     my $sth = $dbh->prepare("SELECT description FROM language_descriptions WHERE subtag=? AND lang=? AND type=?");
 342     #warn "QUERY: SELECT description FROM language_descriptions WHERE subtag=$script AND lang=$lang AND type=$type";
 343     $sth->execute($script,$lang,$type);
 344     while (my $descriptions = $sth->fetchrow_hashref) {
 345         $desc = $descriptions->{'description'};
 346     }
 347     unless ($desc) {
 348         $sth = $dbh->prepare("SELECT description FROM language_descriptions WHERE subtag=? AND lang=? AND type=?");
 349         $sth->execute($script,'en',$type);
 350         while (my $descriptions = $sth->fetchrow_hashref) {
 351             $desc = $descriptions->{'description'};
 352         }
 353     }
 354     return $desc;
 355 }
 356 =head2 regex_lang_subtags
 357
 358 This internal sub takes a string composed according to RFC 4646 as
 359 an input and returns a reference to a hash containing keys and values
 360 for ( language, script, region, variant, extension, privateuse )
 361
 362 =cut
 363
 364 sub regex_lang_subtags {
 365     my $string = shift;
 366
 367     # Regex for recognizing RFC 4646 well-formed tags
 368     # http://www.rfc-editor.org/rfc/rfc4646.txt
 369
 370     # regexes based on : http://unicode.org/cldr/data/tools/java/org/unicode/cldr/util/data/langtagRegex.txt
 371     # The structure requires no forward references, so it reverses the order.
 372     # The uppercase comments are fragments copied from RFC 4646
 373     #
 374     # Note: the tool requires that any real "=" or "#" or ";" in the regex be escaped.
 375
 376     my $alpha   = qr/[a-zA-Z]/ ;    # ALPHA
 377     my $digit   = qr/[0-9]/ ;   # DIGIT
 378     my $alphanum    = qr/[a-zA-Z0-9]/ ; # ALPHA / DIGIT
 379     my $x   = qr/[xX]/ ;    # private use singleton
 380     my $singleton = qr/[a-w y-z A-W Y-Z]/ ; # other singleton
 381     my $s   = qr/[-]/ ; # separator -- lenient parsers will use [-_]
 382
 383     # Now do the components. The structure is slightly different to allow for capturing the right components.
 384     # The notation (?:....) is a non-capturing version of (...): so the "?:" can be deleted if someone doesn't care about capturing.
 385
 386     my $extlang = qr{(?: $s $alpha{3} )}x ; # *3("-" 3ALPHA)
 387     my $language    = qr{(?: $alpha{2,3} | $alpha{4,8} )}x ;
 388     #my $language   = qr{(?: $alpha{2,3}$extlang{0,3} | $alpha{4,8} )}x ;   # (2*3ALPHA [ extlang ]) / 4ALPHA / 5*8ALPHA
 389
 390     my $script  = qr{(?: $alpha{4} )}x ;    # 4ALPHA
 391
 392     my $region  = qr{(?: $alpha{2} | $digit{3} )}x ;     # 2ALPHA / 3DIGIT
 393
 394     my $variantSub  = qr{(?: $digit$alphanum{3} | $alphanum{5,8} )}x ;  # *("-" variant), 5*8alphanum / (DIGIT 3alphanum)
 395     my $variant = qr{(?: $variantSub (?: $s$variantSub )* )}x ; # *("-" variant), 5*8alphanum / (DIGIT 3alphanum)
 396
 397     my $extensionSub    = qr{(?: $singleton (?: $s$alphanum{2,8} )+ )}x ;   # singleton 1*("-" (2*8alphanum))
 398     my $extension   = qr{(?: $extensionSub (?: $s$extensionSub )* )}x ; # singleton 1*("-" (2*8alphanum))
 399
 400     my $privateuse  = qr{(?: $x (?: $s$alphanum{1,8} )+ )}x ;   # ("x"/"X") 1*("-" (1*8alphanum))
 401
 402     # Define certain grandfathered codes, since otherwise the regex is pretty useless.
 403     # Since these are limited, this is safe even later changes to the registry --
 404     # the only oddity is that it might change the type of the tag, and thus
 405     # the results from the capturing groups.
 406     # http://www.iana.org/assignments/language-subtag-registry
 407     # Note that these have to be compared case insensitively, requiring (?i) below.
 408
 409     my $grandfathered   = qr{(?: (?i)
 410         en $s GB $s oed
 411     |   i $s (?: ami | bnn | default | enochian | hak | klingon | lux | mingo | navajo | pwn | tao | tay | tsu )
 412     |   sgn $s (?: BE $s fr | BE $s nl | CH $s de)
 413 )}x;
 414
 415     # For well-formedness, we don't need the ones that would otherwise pass, so they are commented out here
 416
 417     #   |   art $s lojban
 418     #   |   cel $s gaulish
 419     #   |   en $s (?: boont | GB $s oed | scouse )
 420     #   |   no $s (?: bok | nyn)
 421     #   |   zh $s (?: cmn | cmn $s Hans | cmn $s Hant | gan | guoyu | hakka | min | min $s nan | wuu | xiang | yue)
 422
 423     # Here is the final breakdown, with capturing groups for each of these components
 424     # The language, variants, extensions, grandfathered, and private-use may have interior '-'
 425
 426     #my $root = qr{(?: ($language) (?: $s ($script) )? 40% (?: $s ($region) )? 40% (?: $s ($variant) )? 10% (?: $s ($extension) )? 5% (?: $s ($privateuse) )? 5% ) 90% | ($grandfathered) 5% | ($privateuse) 5% };
 427
 428     $string =~  qr{^ (?:($language)) (?:$s($script))? (?:$s($region))?  (?:$s($variant))?  (?:$s($extension))?  (?:$s($privateuse))? $}xi;  # |($grandfathered) | ($privateuse) $}xi;
 429     my %subtag = (
 430         'rfc4646_subtag' => $string,
 431         'language' => $1,
 432         'script' => $2,
 433         'region' => $3,
 434         'variant' => $4,
 435         'extension' => $5,
 436         'privateuse' => $6,
 437     );
 438     return \%subtag;
 439 }
 440
 441 # Script Direction Resources:
 442 # http://www.w3.org/International/questions/qa-scripts
 443 sub get_bidi {
 444     my ($language_script)= @_;
 445     my $dbh = C4::Context->dbh;
 446     my $bidi;
 447     my $sth = $dbh->prepare('SELECT bidi FROM language_script_bidi WHERE rfc4646_subtag=?');
 448     $sth->execute($language_script);
 449     while (my $result = $sth->fetchrow_hashref) {
 450         $bidi = $result->{'bidi'};
 451     }
 452     return $bidi;
 453 };
 454
 455 sub accept_language {
 456     # referenced http://search.cpan.org/src/CGILMORE/I18N-AcceptLanguage-1.04/lib/I18N/AcceptLanguage.pm
 457     # FIXME: since this is only used in Output.pm as of Jan 8 2008, maybe it should be IN Output.pm
 458     my ($clientPreferences,$supportedLanguages) = @_;
 459     my @languages = ();
 460     if ($clientPreferences) {
 461         # There should be no whitespace anways, but a cleanliness/sanity check
 462         $clientPreferences =~ s/\s//g;
 463         # Prepare the list of client-acceptable languages
 464         foreach my $tag (split(/,/, $clientPreferences)) {
 465             my ($language, $quality) = split(/\;/, $tag);
 466             $quality =~ s/^q=//i if $quality;
 467             $quality = 1 unless $quality;
 468             next if $quality <= 0;
 469             # We want to force the wildcard to be last
 470             $quality = 0 if ($language eq '*');
 471             # Pushing lowercase language here saves processing later
 472             push(@languages, { quality => $quality,
 473                language => $language,
 474                lclanguage => lc($language) });
 475         }
 476     } else {
 477         carp "accept_language(x,y) called with no clientPreferences (x).";
 478     }
 479     # Prepare the list of server-supported languages
 480     my %supportedLanguages = ();
 481     my %secondaryLanguages = ();
 482     foreach my $language (@$supportedLanguages) {
 483         # warn "Language supported: " . $language->{language};
 484         my $subtag = $language->{rfc4646_subtag};
 485         $supportedLanguages{lc($subtag)} = $subtag;
 486         if ( $subtag =~ /^([^-]+)-?/ ) {
 487             $secondaryLanguages{lc($1)} = $subtag;
 488         }
 489     }
 490
 491     # Reverse sort the list, making best quality at the front of the array
 492     @languages = sort { $b->{quality} <=> $a->{quality} } @languages;
 493     my $secondaryMatch = '';
 494     foreach my $tag (@languages) {
 495         if (exists($supportedLanguages{$tag->{lclanguage}})) {
 496             # Client en-us eq server en-us
 497             return $supportedLanguages{$tag->{language}} if exists($supportedLanguages{$tag->{language}});
 498             return $supportedLanguages{$tag->{lclanguage}};
 499         } elsif (exists($secondaryLanguages{$tag->{lclanguage}})) {
 500             # Client en eq server en-us
 501             return $secondaryLanguages{$tag->{language}} if exists($secondaryLanguages{$tag->{language}});
 502             return $supportedLanguages{$tag->{lclanguage}};
 503         } elsif ($tag->{lclanguage} =~ /^([^-]+)-/ && exists($secondaryLanguages{$1}) && $secondaryMatch eq '') {
 504             # Client en-gb eq server en-us
 505             $secondaryMatch = $secondaryLanguages{$1};
 506         } elsif ($tag->{lclanguage} =~ /^([^-]+)-/ && exists($secondaryLanguages{$1}) && $secondaryMatch eq '') {
 507             # FIXME: We just checked the exact same conditional!
 508             # Client en-us eq server en
 509             $secondaryMatch = $supportedLanguages{$1};
 510         } elsif ($tag->{lclanguage} eq '*') {
 511         # * matches every language not already specified.
 512         # It doesn't care which we pick, so let's pick the default,
 513         # if available, then the first in the array.
 514         #return $acceptor->defaultLanguage() if $acceptor->defaultLanguage();
 515         return $supportedLanguages->[0];
 516         }
 517     }
 518     # No primary matches. Secondary? (ie, en-us requested and en supported)
 519     return $secondaryMatch if $secondaryMatch;
 520     return undef;   # else, we got nothing.
 521 }
 522 1;
 523
 524 __END__
 525
 526 =head1 AUTHOR
 527
 528 Joshua Ferraro
 529
 530 =cut