tito Memo

* camelKWIC.pl

一行一単語を入力として'_'やcamel caseで区切られた単語のKWIC風見出しを出力するperlスクリプト

 use strict;
 
 my %word;
 my %key;
 my (@a,@b);
 
 $"="";
 while(<>){
     chop;
     $word{$_}=1;
     @a=split(/(?:_|(?<=[a-z0-9])(?=[A-Z]))/,$_);
 
     foreach my $k(@a){
         $key{$k}=1 if(length $k>1);
     }
 }
 
 foreach my $i( sort { lc $a cmp lc $b || $a cmp $b } keys %key){
     print "==== $i ====\n";
     foreach ( sort { lc $a cmp lc $b || $a cmp $b } keys %word){
         if( /$i/ ){
             my @a=split(/(?:(?=_)|(?<=_)|(?<=[a-z0-9])(?=[A-Z]))/,$_);
             my $j;
             for( $j=0; $j <@a;++$j){
                 if( $a[$j] eq $i ){
                     printf "%20s%s\n","@a[0..$j-1]","@a[$j..$#a]";
                 }
             }
 
 #              while( /(\A|(?<=[_a-z0-9])(?=[A-Z])|(?<=_)(?=[A-Za-z0-9]))/g){
 #                  my($l,$r)=($`,$');
 #                  printf "%20s%s\n", $l, $r if( $r=~/^$i/);
 #              }
         }
     }
 }

 $ cat foo 
 aSampleTest
 Sample_test_word
 word_Sample_test
 test_test_test
 
 $ perl camelKWIC.pl <foo
 ==== Sample ====
                    aSampleTest
                     Sample_test_word
                word_Sample_test
 ==== Test ====
              aSampleTest
 ==== test ====
              Sample_test_word
                     test_test_test
                test_test_test
           test_test_test
         word_Sample_test
 ==== word ====
         Sample_test_word
                     word_Sample_test