[LON-CAPA-cvs] cvs: loncom /localize/localize checksimilar_1file.pl

[bisitz]

bisitz		Mon Jan 21 15:29:31 2013 EDT

  Modified files:              
    /loncom/localize/localize	checksimilar_1file.pl 
  Log:
  - Detect even more similarities by extending list of similar characters
    (include improvements from checksimilar_2files.pl 1.4)
  - Offer switch to include check for similar phrases, e.g. course/community, as done in checksimilar_2files.pl
  
  
Index: loncom/localize/localize/checksimilar_1file.pl
diff -u loncom/localize/localize/checksimilar_1file.pl:1.1 loncom/localize/localize/checksimilar_1file.pl:1.2
--- loncom/localize/localize/checksimilar_1file.pl:1.1	Tue Mar  9 15:16:26 2010
+++ loncom/localize/localize/checksimilar_1file.pl	Mon Jan 21 15:29:30 2013
@@ -1,6 +1,6 @@
 #!/usr/bin/perl
 # The LearningOnline Network with CAPA
-# $Id: checksimilar_1file.pl,v 1.1 2010/03/09 15:16:26 wenzelju Exp $
+# $Id: checksimilar_1file.pl,v 1.2 2013/01/21 15:29:30 bisitz Exp $
 
 use strict;
 use warnings;
@@ -9,6 +9,9 @@
 #### Checks, if there are similar keys in the inputfile (for example de.pm)
 ####
 
+####--------Configuration--------####
+# Include check for similar phrases -> set to 1
+my $inclphrases = 0;
 
 
 ####--------Subroutines--------####
@@ -29,10 +32,18 @@
     return %filecontent;
 }
 
-sub similarities{
+sub similarities {
    my $text = shift;
-   $text =~ s/[.,\_\-?!:]//g;
-
+   $text =~ s/\[_\d\]//g; # translation parameters
+   $text =~ s/[.,\_\-?!: \/]//g; # punctuation
+   if ($inclphrases) {
+       $text =~ s/course/X002X/gi;
+       $text =~ s/community/X002X/gi;
+       $text =~ s/communities/X001X/gi;    
+       $text =~ s/member/X003X/gi;
+       $text =~ s/student/X003X/gi;
+       $text =~ s/students/X003X/gi;
+   }
    return $text;
 }