The Perl Toolchain Summit needs more sponsors. If your company depends on Perl, please support this very important event.

NAME

Activator::Dictionary

SYNOPSIS

Configure your dictionary using Activator::Registry. See "CONFIGURATION OVERVIEW" below.

Using explicit realms and languages:

  use Activator::Dictionary;
  my $dict = Activator::Dictionary->get_dict( $lang );
  my $val  = $dict->lookup( $key, $realm );

Or, configure defaults in Activator::Registry config file:

  'Activator::Registry':
    'Activator::Dictionary':
      default_lang:  'en'
      default_realm: 'my_realm'

Then:

  use Activator::Dictionary;
  my $dict = Activator::Dictionary->get_dict();
  my $val  = $dict->lookup( $key );

DESCRIPTION

This module provides simple lookup of key/value pairs for intended for internationalization/localization(I18N/L10N). It is also useful for separating the progamming of a project from the creation of the text used by it. The object created by this module is a per-process singleton that uses dictionary definintions from a simple space-delimeted file or database table(s). The dictionary is completely maintained in memory and loads realms and languages dynamically on an as-needed basis, so this module may not be appropriate for extremely large lexicons or for projects that create large numbers of program instances. That being said, it can be relatively memory efficient when used for a single language deployment in an application that provides multiple language support.

An Activator::Dictionary object can have multiple realms: that is, you could have a 'web' dictionary for the website text, an 'error' dictionary for backend job messages, and any number of other realms needed for your application. This allows you to separate the translatable texts from each other so that, for example, the web frontend of your application could give a user friendly message using the 'web' realm, and the backend could use the 'error' realm to log something much more useful to a technician.

Note that there can be great amounts of complexity localizing language within an application. This module is for the simple cases, where you just have key/value lookups. If you need complex conjugations, object sensitive pluralization, you should look into the existing Locale::Maketext, or the upcoming Activator::Lexicon module. It is highly recommended that you read http://search.cpan.org/dist/Locale-Maketext/lib/Locale/Maketext/TPJ13.pod before making a decision as to which localization method your application needs.

CONFIGURATION OVERVIEW

  'Activator::Registry':          # uses Activator::Registry
    'Activator::Dictionary':
      default_lang:  'en'         # default language for get_dict()*
      default_realm: 'my_realm'   # default realm for lookup()*
      fail_mode:     [ die ]      # die instead of returning undef
                                    for lookup failures*
      dict_files:    '<path>'     # path to definition files**
      dict_tables:   [ t1, t2 ]   # database definition table(s)**
      db_alias:      'db'         # Activator::DB alias to use***

   * optional
  ** either dict_files OR dict_tables MUST be defined
 *** db_alias required when dict_tables defined

DICTIONARY FILE CONFIGURATION

Configure your dictionary in your project registry:

  'Activator::Registry':
    'Activator::Dictionary':
      dict_files:         '/path/to/definitions/files'

Then create dictionary definition files for realms in the dictionary path as such:

 <dict_files path>/<lang>/<realm>.dict

Dictionary File Format

To create a dictionary file, create a file named <realm>.dict containing key/value pairs separated by whitespace. Keys can have any non-whitespace character in them. The amount of whitespace between key and value can be any length and can be tab or space characters (more specifically, any character that matches /\s/). Keys and values must be on the same line.

For example:

  error.bummer        A bummer of an error occured
  foo-html            <p>this is the foo paragraph</p>
  welcome_msg         Welcome to Activatory::Dictionary!!
  answer              42

Empty lines and any line that the first non-whitespace character is # will be ignored. Leading whitespace for keys will be ignored as well, so that you can indent however you see fit.

Leading and trailing whitespace are stripped from values. If the value for some key must begin or end with white space, wrap the value portion of the line with double quotes. Any value that begins with a double quote will have a trailing double quote stripped.

Examples: key1 value1 # value eq 'value1' # key2 value2 # ignored # key3 value3 # ignored # ignored key4 multiple words # value eq 'multiple words' key5 " value5 is quoted" # value eq ' value5 is quoted' key6 ""OMG!" "quotes!"" # value eq '"OMG!" "quotes!"' key7 " whitespaced " # value eq ' whitespaced '

DATABASE CONFIGURATION

If you would rather that your dictionary definitions are in a database, or need more complex values than can be reasonably contained within a single line, create a table of any name with this schema:

 CREATE TABLE db_table_name (
   # primary column must end with '_id'
   *_id          serial,
   lang          enum('en','de','es') default 'en',
   realm         text NOT NULL,
   key_prefix    text NOT NULL,
   last_modified datetime NOT NULL,

   # Then, define any attributes of the key that you any way you want,
   # excepting that they cannot end with the string '_id', or be the
   # same as any of the cols in the above section (aka: the previous
   # columns use reserved words):
   col_1 varchar(256) NOT NULL,
   col_2 text NOT NULL,
   col_3 text NOT NULL,
   col_4 int,
   col_5 text,

   # insure realm/key/lang integrity in your DB's way. This is MySQL:
   UNIQUE KEY IDX_db_dictionary_1 (realm,key_prefix,lang)
 );

The schema is designed to allow all realms to be in one table, but you can spread it accross as many tables as you like, provided they are in the same database.

NOTE: When using database for definitions key_prefix cannot have a period in it.

Add the table(s) to use to the registry:

  'Activator::Registry':
    'Activator::Dictionary':
      dict_tables: [ table1, table2 ]
      db_alias: 'Activator::DB alias to use'

Note that you can use dict_files and dict_tables in any combination.

RESERVED WORDS FOR REALMS

When naming realms, follow these guidelines:

  • Use more than 2 characters, to not confuse realms with languages.

  • Do not use the word config for a realm

TODO: enforce this guidance programatically

LOOKUP FEATURES

Using a Default Realm and/or Language

In some applications, it is inconvenient to have to pass the realm as an argument for every lookup call when there is one common realm that is nearly always used. You can define a default language and/or realm as such:

  'Activator::Registry':
    'Activator::Dictionary':
      default_lang:  'en'         # optional
      default_realm: 'my_realm'   # optional

Not passing the $lang or $realm arguments will then use the registry key(s):

  my $dict = Activator::Dictionary->get_dict();  # sets lang to en
  $dict->lookup( $key );                         # returns 'my_realm' value

Database Dictionary Lookups

When using database dictionary definitions, you must define the target field you are interested in with dot notation:

  $dict->lookup( $key_prefix );        # fails
  $dict->lookup( "$key_prefix.$col" ); # succeeds

For this reason, it is required that you not use period in the key_prefix column.

Failure Mode

Instead of returning undef for non-existent keys, you can configure this module to fail via one or more of these methods:

  die     : throws Activator::Exception::Dictionary('key', 'missing')
  key     : returns the requested key itself
  ''      : returns empty string
  <lang>  : return the value for <lang> in the requested realm
  <realm> : return the value for <realm>

Examples:

  $db->lookup( $key, $realm1 );  # value does not exist

  fail_mode: [ realm2, de, key ]

   return value for $key in realm2 if it exists
   return value for $key in realm1 in german if it exists
   return $key

  fail_mode: [ realm2, die ]

   return value for $key in realm2 if it exists
   throw Activator::Exception::Dictionary

  fail_mode: [ realm2, realm3 ]

   return value for $key in realm2 if it exists
   return value for $key in realm3 if it exists
   return undef (fallback to default failure mode)

  fail_mode: [ '' ]

   return empty string

DISABLING LOAD WARNING

When loading dictionary files, you may sometimes see:

  [WARN] Couldn't load dictionary from file for <lang>

If you are using files for one language, and the DB for another, this could get really annoying since you KNOW THIS TO BE TRUE. The workaround is to set the log level for this message an alternate level of FATAL, ERROR, WARN, INFO, DEBUG, or TRACE. For example:

  $dict->{LOG_LEVEL_FOR_FILE_LOAD} = 'INFO';

METHODS

lookup($key, $realm)

OO Usage:

  my $dict = Activator::Dictionary->get_dict( $lang );
  $dict->lookup( $key, $realm );
  $dict->lookup( $key2, $realm );

Static Usage:

  Activator::Dictionary->use_lang( $lang );
  Activator::Dictionary->lookup( $key, $realm );
  Activator::Dictionary->lookup( $key2, $realm );

Returns the value for $key in $realm. Returns undef when the key does not exist, but you can configure this module to do something different (see "Failure Mode" below). If realm does not exist, throws Activator::Exception::Dictionary no matter the failure mode.

get_dict( $lang )

Returns a reference to the Activator::Dictionary object. Sets all future lookups to use the $lang passed in. If $lang is not passed in, uses 'Activator::Dictionary' registry value for 'default_lang'. If $lang cannot be determined, throws Activator::Exception::Dictionary.

new( $lang )

Creates a dictionary object. Not very useful, as all it does is create an uninitialized instance of an Activator::Dictionary object.

SEE ALSO

Activator::Log, Activator::Exception, Activator::DB, Exception::Class::TryCatch, Class::StrongSingleton

AUTHOR

Karim A. Nassar

COPYRIGHT

Copyright (c) 2007 Karim A. Nassar <karim.nassar@acm.org>

You may distribute under the terms of either the GNU General Public License or the Artistic License, as specified in the Perl README file.