Code Index:

package Lingua::AM::Abbreviate
- Expand
- Contract
__END__
EOF

NAME

AM::Abbreviate - Expand or Contract Amharic Abbreviations

SYNOPSIS

  use Lingua::AM::Abbreviate;

	while ( $string = <> ) {  # some UTF8 string
		if ( $contracted = Contract ( $string ) ) {
			print "$string => $contracted\n";
		} elsif ( $expanded = Expand ( $string ) ) {
			print "$string => $expanded\n";
		}
	}

DESCRIPTION

AM::Abbreviate provides two routines, "Expand" and "Contract", to assist in Amharic translation or spell checking. Each routine expects an Amharic string in UTF8 encoding as an argument and returns an expansion or contraction if found.

AUTHOR

Daniel Yacob, LibEth@EthiopiaOnline.Net

package Lingua::AM::Abbreviate; require 5.000; require Exporter; @ISA = qw(Exporter); @EXPORT = qw( Contract Expand ); $VERSION = '0.01'; %AbbreviatedAmharic = ( 'á‰µ/á‰¤á‰µ' => 'á‰µáˆáˆ…áˆá‰µ á‰¤á‰µ', 'áŒ½/á‰¤á‰µ' => 'áŒ½áˆ…áˆá‰µ á‰¤á‰µ', 'áˆ/á‰¤á‰µ' => 'áˆáŠáˆ á‰¤á‰µ', 'áˆ˜/á‰¤á‰µ' => 'áˆ˜áˆ¥áˆªá‹« á‰¤á‰µ', 'á/á‰¤á‰µ' => 'ááˆá‹µ á‰¤á‰µ', 'áˆ /áŒá‹´áˆ¬áˆ½áŠ•' => 'áˆ áˆ«á‰°áŠ› áŒá‹´áˆ¬áˆ½áŠ•', 'á‰¤/áŠ' => 'á‰¤á‰°áŠáˆáˆµá‰µá‹«áŠ•', 'á‰¤/áˆ˜' => 'á‰¤á‰°áˆ˜áŠ•áŒáˆµá‰µ', 'áˆ˜/á‰…' => 'áˆ˜áŒ¸áˆá á‰…á‹±áˆµ', 'áˆ/' => 'áˆáŠá‰µáˆ', 'áŒˆ/' => 'áŒˆá‰¥áˆ¨', 'á‹ˆ/' => 'á‹ˆáˆá‹°', 'á‰°/' => 'á‰°áŠáˆˆ', # á‹ˆá‹áˆ á‰°áˆµá‹ 'áŠƒ/' => 'áŠƒá‹áˆˆ', 'áŠ®/áˆ' => 'áŠ®áˆˆáŠ”áˆ', 'áŒ„/áˆ' => 'áŒ„áŠ”áˆ«áˆ', 'áˆ»/' => 'áˆ»áˆá‰ áˆ', 'á‹ˆ/áˆ®' => 'á‹ˆá‹á‹˜áˆ®', 'á‹ˆ/áˆªá‰µ' => 'á‹ˆá‹á‹˜áˆªá‰µ', 'áˆš/áˆ' => 'áˆšáˆµá‰°áˆ', 'áˆš/áˆµ' => 'áˆšáˆµáˆµ', 'á•/áˆ' => 'á•áˆ®áŒáˆ°áˆ', 'á•/á‰µ' => 'á•áˆ¬á‹šá‹³áŠ•á‰µ', 'áŒ /áˆš' => 'áŒ á‰…áˆ‹á‹ áˆšáŠ’áˆµá‰µáˆ', 'áŒ /áˆšáŠ’áˆµá‰µáˆ' => 'áŒ á‰…áˆ‹á‹ áˆšáŠ’áˆµá‰µáˆ', 'áŒ /áˆš/á‰¢áˆ®' => 'áŒ á‰…áˆ‹á‹ áˆšáŠ’áˆµá‰µáˆ á‰¢áˆ®', 'áˆš/áˆ©' => 'áˆšáŠ’áˆµá‰µáˆ©', 'á‹“/áˆ' => 'á‹“áˆ˜á‰° áˆáˆ…áˆ¨á‰µ', 'á‹“/á‹“' => 'á‹“áˆ˜á‰° á‹“áˆˆáˆ', 'áŠ /áŠ ' => 'áŠ á‹²áˆµ áŠ á‰£á‰£', 'á‹¶/áˆ' => 'á‹¶áŠá‰°áˆ', 'áˆ†/áˆ' => 'áˆ†áˆµá’á‰³áˆ' ); sub Expand { local ($term) = $_[0]; if ( $AbbreviatedAmharic{$term} ) { return ( $AbbreviatedAmharic{$term} ); } else { $prefix = $term =~ s/^([áˆˆá‰ áŠ¨á‹¨])//; if ( $AbbreviatedAmharic{$term} ) { return ( "$prefix$AbbreviatedAmharic{$term}" ); } else { $term = $_[0]; ($prefix, $term) = split ( /\//, $term, 2 ); $prefix .= "/"; if ( $AbbreviatedAmharic{$prefix} ) { return ( "$AbbreviatedAmharic{$prefix}$term" ); } } } } sub Contract { local ($term) = shift; foreach $key (keys %AbbreviatedAmharic) { return ( $key ) if ( $AbbreviatedAmharic{$key} eq $term ); # print "Contract: $AbbreviatedAmharic{$key} = $term\n"; if ( $term =~ /^$AbbreviatedAmharic{$key}/ ) { $term =~ s/$AbbreviatedAmharic{$key}//; return ( "$key$term" ); } } } # # ma/bEt # d/bEt # me/`se/ma # qe/ge/ma # m/wana as in "m/wana SeHefi" ######################################################### # Do not change this, Do not put anything below this. # File must return "true" value at termination 1; ########################################################## __END__