| 1 | #!/usr/bin/perl -w | ||
| 2 | # | ||
| 3 | # Perl package exporting a function "unconj" that un-conjugates | ||
| 4 | # Portuguese verbs. | ||
| 5 | # | ||
| 6 | # Author : Etienne Grossmann (etienne@isr.ist.utl.pt) | ||
| 7 | # | ||
| 8 | # Date : September 1999 onwards. | ||
| 9 | # | ||
| 10 | # | ||
| 11 | package Lingua::PT::UnConjugate ; | ||
| 12 | |||
| 13 | =head1 NAME | ||
| 14 | |||
| 15 | Lingua::PT::UnConjugate - Recognition of the conjugated forms of | ||
| 16 | portuguese verbs. | ||
| 17 | |||
| 18 | =head1 DESCRIPTION | ||
| 19 | |||
| 20 | This module provides functions for the recognition of the conjugated | ||
| 21 | forms of portuguese verbs. | ||
| 22 | |||
| 23 | =head1 BUGS | ||
| 24 | |||
| 25 | Composed tenses are not recognized. The verb list contains many | ||
| 26 | non-verbs that I have not removed yet. | ||
| 27 | |||
| 28 | =cut | ||
| 29 | |||
| 30 | use Lingua::PT::Conjugate qw( conjug %long_tense ) ; | ||
| 31 | |||
| 32 | use Lingua::PT::Infinitives ; | ||
| 33 | use Lingua::PT::VerbSuffixes ; | ||
| 34 | import Lingua::PT::Accent_iso_8859_1 qw(asc2iso); | ||
| 35 | use Exporter ; | ||
| 36 | @ISA = qw(Exporter); | ||
| 37 | # Yes, this package is a namespace polluter. | ||
| 38 | @EXPORT = qw( unconj ); | ||
| 39 | @EXPORT_OK = qw( unconj list_entries string_entries ); | ||
| 40 | |||
| 41 | |||
| 42 | BEGIN { | ||
| 43 | # Suffixes and Infinitives. | ||
| 44 | |||
| 45 | # ####################### VOCALS, CONSONANTS ##################### | ||
| 46 | # Vocals and Consonants | ||
| 47 | $vocs = "aeiouáàäâãéèëêíìïîóòöôõúùüû"; | ||
| 48 | $cons = 'qwrtypsdfghjklzxcvbnm'; | ||
| 49 | $letter = "ç$vocs$cons"; | ||
| 50 | $lpat = "[$letter]" ; | ||
| 51 | # Equivalent accent-matching regexp | ||
| 52 | %equiv = ( "a"=>"[aáàäâã]", | ||
| 53 | "á"=>"[aáàäâã]", | ||
| 54 | "à"=>"[aáàäâã]", | ||
| 55 | "ä"=>"[aáàäâã]", | ||
| 56 | "â"=>"[aáàäâã]", | ||
| 57 | "ã"=>"[aáàäâã]", | ||
| 58 | "e"=>"[eéèëê]", | ||
| 59 | "é"=>"[eéèëê]", | ||
| 60 | "è"=>"[eéèëê]", | ||
| 61 | "ë"=>"[eéèëê]", | ||
| 62 | "ê"=>"[eéèëê]", | ||
| 63 | "i"=>"[iíìïî]", | ||
| 64 | "í"=>"[iíìïî]", | ||
| 65 | "ì"=>"[iíìïî]", | ||
| 66 | "ï"=>"[iíìïî]", | ||
| 67 | "î"=>"[iíìïî]", | ||
| 68 | "o"=>"[oóòöôõ]", | ||
| 69 | "ó"=>"[oóòöôõ]", | ||
| 70 | "ò"=>"[oóòöôõ]", | ||
| 71 | "ö"=>"[oóòöôõ]", | ||
| 72 | "ô"=>"[oóòöôõ]", | ||
| 73 | "õ"=>"[oóòöôõ]", | ||
| 74 | "u"=>"[uúùüû]", | ||
| 75 | "ú"=>"[uúùüû]", | ||
| 76 | "ù"=>"[uúùüû]", | ||
| 77 | "ü"=>"[uúùüû]", | ||
| 78 | "û"=>"[uúùüû]", | ||
| 79 | "c"=>"[cç]", | ||
| 80 | "ç"=>"[cç]", | ||
| 81 | ) ; | ||
| 82 | $equivk = join "", "[", keys(%equiv), "]" ; | ||
| 83 | |||
| 84 | # Lower_case | ||
| 85 | %mylc = split "", | ||
| 86 | "ÇçÁáÀàÄäÂâÃãÉéÈèËëÊêÍíÌìÏïÎîÓóÒòÖöÔôÕõÚúÙùÜüÛû"; | ||
| 87 | $mylck = join "", "[", keys(%mylc), "]" ; | ||
| 88 | # print "$equivk\n$mylck\n" ; | ||
| 89 | |||
| 90 | } | ||
| 91 | |||
| 92 | sub my_lc # lc() for accentuated characters too | ||
| 93 | { | ||
| 94 | my $a = shift ; | ||
| 95 | $a = lc($a) ; | ||
| 96 | $a =~ s/($mylck)/$mylc{$1}/g ; | ||
| 97 | return $a ; | ||
| 98 | } | ||
| 99 | |||
| 100 | # $r = regexify( $w ) | ||
| 101 | # $r is a regex that will match any ending substring of $w | ||
| 102 | sub regexify | ||
| 103 | { | ||
| 104 | my $r = shift ; | ||
| 105 | my $r0 = $r ; | ||
| 106 | while( $r =~ s/($lpat+)($lpat)/(\?:$1)\?$2/ ){} | ||
| 107 | # print "regexify : $r0 -> $r\n" ; | ||
| 108 | return $r ; | ||
| 109 | } | ||
| 110 | |||
| 111 | =head1 SYNOPSIS | ||
| 112 | |||
| 113 | =head1 C<$verb_forms = unconj( [-a] , $string )> | ||
| 114 | |||
| 115 | Attempts to recognize a conjugated form of a Portuguese verb, and | ||
| 116 | returns the result as a reference to hash : if the element | ||
| 117 | |||
| 118 | C<$verb_forms-E<gt>{$infinitive}-E<gt>{$tense}-E<gt>[$person]> | ||
| 119 | |||
| 120 | is true, then the conjugation of the verb "$infinitive" at the tense | ||
| 121 | "$tense" and the person "$person" should yield "$string". | ||
| 122 | |||
| 123 | =head2 OPTIONS | ||
| 124 | |||
| 125 | The first argument may an option : | ||
| 126 | |||
| 127 | =over 4 | ||
| 128 | |||
| 129 | =item -a : Try to match accentuation errors. | ||
| 130 | |||
| 131 | =item -A : If no match is found, try matching with option -a. | ||
| 132 | |||
| 133 | =back | ||
| 134 | |||
| 135 | =cut | ||
| 136 | |||
| 137 | sub unconj | ||
| 138 | { | ||
| 139 | my $acc = 0 ; # Check errors in accentuation ? | ||
| 140 | my $ret = 0 ; # Retry in case of failure ? | ||
| 141 | while( $_[0] =~ /^-[aA]$/ ) # Get options | ||
| 142 | { | ||
| 143 | my $opt = shift ; | ||
| 144 | $opt =~ s/-//; | ||
| 145 | $acc = 1 if $opt =~ /a/ ; | ||
| 146 | $ret = 1 if $opt =~ /A/ ; | ||
| 147 | # print "unconj : option $opt\n" ; | ||
| 148 | } | ||
| 149 | my $v0 = shift ; | ||
| 150 | # print "unconj : $v0\n" ; | ||
| 151 | my $v = asc2iso( $v0 ); # No ascii-style accents | ||
| 152 | $v = my_lc($v) ; | ||
| 153 | # my @res = (); | ||
| 154 | my %res = (); | ||
| 155 | |||
| 156 | my $p = regexify( $v ) ; | ||
| 157 | if( $acc ) | ||
| 158 | { | ||
| 159 | ## HERE : assume letters are isolated in $p | ||
| 160 | $p =~ s/\b($equivk)\b/$equiv{$1}/g ; | ||
| 161 | ## print "$p\n" ; | ||
| 162 | } | ||
| 163 | my $p2 = "($p .*)" ; | ||
| 164 | my @matches = $verb_suffixes =~ /^$p2/mg ; | ||
| 165 | |||
| 166 | push @matches, " cfut,1,", " cfut,3," if $infinitives =~ /^$v$/m ; | ||
| 167 | |||
| 168 | # print join "\n", @matches,"\n" ; | ||
| 169 | foreach $m (@matches) | ||
| 170 | { | ||
| 171 | my ($s,$t,$p,$r) = $m =~ /^(\S*) (\w+),(\d+),(.*)/ ; | ||
| 172 | my @endings = split ",",$r ; | ||
| 173 | @endings = ("") unless @endings ; | ||
| 174 | # print "Found $m ",0+@endings," endings\n"; | ||
| 175 | # print "-- $s, $t, $p, $r\n" ; | ||
| 176 | my $root = $s ? substr( $v, 0, -length($s) ) : $v ; | ||
| 177 | foreach (@endings) | ||
| 178 | { | ||
| 179 | my $i = $root . $_ ; | ||
| 180 | # $i =~ s/r+r$/r/ ; # Why? | ||
| 181 | ##print "Trying : $i, $t, $p, $root + $_\n" ; | ||
| 182 | next unless $infinitives =~ /^$i$/m ; | ||
| 183 | my $check = conjug("xs",$i,$t,$p) ; | ||
| 184 | $check =~ s/($equivk)/$equiv{$1}/g if $acc ; | ||
| 185 | # print "Check $check\n" if $acc ; | ||
| 186 | # print "Checking $i, $t, $p against $check\n" ; | ||
| 187 | next unless $v =~ /^$check$/ ; | ||
| 188 | # print "Found $i, $t, $p in $m\n" ; | ||
| 189 | # push @res , [$i, $t, $p] ; | ||
| 190 | $res{$i}->{$t}->[$p] = 1 ; | ||
| 191 | } | ||
| 192 | } | ||
| 193 | %res = %{unconj("-a",$v0)} if $ret && !$acc && !keys(%res) ; | ||
| 194 | return \%res ; | ||
| 195 | } | ||
| 196 | |||
| 197 | =head1 C<@res = string_entries( ['l'], \%verb_forms )> | ||
| 198 | |||
| 199 | Convert a hash of recognized forms into a list of strings | ||
| 200 | C<"$verb, $tense, $person">. | ||
| 201 | |||
| 202 | If the first argument is a 'l', then long forms of verb names will | ||
| 203 | be used. | ||
| 204 | |||
| 205 | =cut | ||
| 206 | |||
| 207 | sub string_entries | ||
| 208 | { | ||
| 209 | my $long = 0 ; | ||
| 210 | if( $_[0] eq 'l' ) # Accept a "long verb name" option | ||
| 211 | { | ||
| 212 | $long = 1 ; | ||
| 213 | shift ; | ||
| 214 | } | ||
| 215 | my $vdb = shift ; | ||
| 216 | my ($w,$x,$y,$z); | ||
| 217 | my @res = (); | ||
| 218 | |||
| 219 | my $vcnt = 0 ; | ||
| 220 | |||
| 221 | while( ($w,$x) = each %$vdb ) { | ||
| 222 | next if $w eq " " ; | ||
| 223 | next unless $w ; | ||
| 224 | while( ($t,$y) = each %$x ) { | ||
| 225 | foreach $p (1..6) | ||
| 226 | { | ||
| 227 | # print "." if $vcnt % 50 == 49 ; | ||
| 228 | # print "\n".sprintf("%-6d ",$vcnt) | ||
| 229 | # if $vcnt %1000 == 999 ; | ||
| 230 | next unless defined( $y->[$p] ); | ||
| 231 | # HERE : A bug ? If I don't check for "defined" the | ||
| 232 | # first time $long_tense{$t} is used, it is undef'd. | ||
| 233 | $t = $long_tense{$t} if $long && | ||
| 234 | defined( $long_tense{$t} ); | ||
| 235 | push @res, "$w, $t, $p" ; | ||
| 236 | } | ||
| 237 | } | ||
| 238 | } | ||
| 239 | return sort @res ; | ||
| 240 | } | ||
| 241 | |||
| 242 | =head1 C<@res = list_entries( ['l'], \%verb_forms )> | ||
| 243 | |||
| 244 | Convert a hash of recognized forms into a list of triplets | ||
| 245 | C<[ $verb, $tense, $person ]>. | ||
| 246 | |||
| 247 | If the first argument is a 'l', then long forms of verb names will | ||
| 248 | be used. | ||
| 249 | |||
| 250 | =cut | ||
| 251 | |||
| 252 | sub list_entries | ||
| 253 | { | ||
| 254 | my $long = 0 ; | ||
| 255 | if( $_[0] eq 'l' ) # Accept a "long verb name" option | ||
| 256 | { | ||
| 257 | $long = 1 ; | ||
| 258 | shift ; | ||
| 259 | } | ||
| 260 | my $vdb = shift ; | ||
| 261 | my ($w,$x,$y,$z); | ||
| 262 | my @res = (); | ||
| 263 | |||
| 264 | my $vcnt = 0 ; | ||
| 265 | |||
| 266 | while( ($w,$x) = each %$vdb ) { | ||
| 267 | next if $w eq " " ; | ||
| 268 | next unless $w ; | ||
| 269 | while( ($t,$y) = each %$x ) { | ||
| 270 | foreach $p (1..6) | ||
| 271 | { | ||
| 272 | # print "." if $vcnt % 50 == 49 ; | ||
| 273 | # print "\n".sprintf("%-6d ",$vcnt) | ||
| 274 | # if $vcnt %1000 == 999 ; | ||
| 275 | next unless defined( $y->[$p] ); | ||
| 276 | $t = $long_tense{$t} if $long && | ||
| 277 | defined( $long_tense{$t} ); | ||
| 278 | push @res, [$w, $t, $p ] ; | ||
| 279 | } | ||
| 280 | } | ||
| 281 | } | ||
| 282 | return sort @res ; | ||
| 283 | } | ||
| 284 | |||
| 285 | =head1 SEE ALSO : unconj, conjug, treinar. | ||
| 286 | |||
| 287 | =head1 AUTHOR Etienne Grossmann, 1999 [etienne@isr.ist.utl.pt] | ||
| 288 | |||
| 289 | =head1 CREDITS | ||
| 290 | |||
| 291 | Thanks to Soraia Almeida (salmeida@logos.it) from the Logos project | ||
| 292 | (http://www.logos.it) and Ulisses Pinto and José João Almeida from | ||
| 293 | Projecto Natura (http://shiva.di.uminho.pt/~jj/pln) who made Ispell | ||
| 294 | available. | ||
| 295 | |||
| 296 | A big part of the list of verb infinitives comes from files used in | ||
| 297 | Ispell (http://shiva.di.uminho.pt/~jj/pln) and in Logos | ||
| 298 | (http://www.verba.org, http://www.logos.it). these projects. Some | ||
| 299 | verbs were removed and others added by hand. | ||
| 300 | |||
| 301 | =cut | ||
| 302 | |||
| 303 | |||
| 304 | 1; | ||
| 305 | |||
| 306 | |||
| 307 | |||
| 308 | |||
| 309 |