Changeset 25155 for gs2-extensions

Timestamp:

2012-02-28T16:40:01+13:00 (12 years ago)

Author:

papitha

Message:

Tidy up of code to give controlled output (verbosity>=2)

Location:

gs2-extensions/ngramj/perllib

Files:

• ngramj.pm (modified) (6 diffs)
• plugins/ReadTextFile.pm (modified) (3 diffs)

gs2-extensions/ngramj/perllib/ngramj.pm

@@ -33 +33 @@
 
 use strict;
-
+no strict 'refs'; # allow filehandles to be variables and viceversa
 
 sub new {
     my $class = shift (@_);
-
-    my $self = {};
+    my ($verbosity,$outhandle) = @_;
+    
+    my $self = { 'verbosity' => $verbosity, 'outhandle' => $outhandle };
 
     my $ngram_jar = &util::filename_cat($ENV{'GSDLHOME'},"ext","ngramj","jars","cngram.jar");
@@ -45 +46 @@
     
     $self->{'java_cmd'} = $java_cmd;
-    
   
     return bless $self, $class;
@@ -87 +87 @@
                 my @lang_array = split(/\s+/,$lang_group);
                 
+                my @lang_summary = ( "++ Ngram language probabilities:\n++  ");
+                
                 foreach my $l (@lang_array) {
-                    print STDERR "l = $l\n";
+                    push(@lang_summary,$l);
                     my ($lang,$score) = ($l =~ m/^(.+):(.+)$/);
                     
@@ -95 +97 @@
                     
                     push(@$lang_encode_pairs,$lang_pair);
+                }
+                push(@lang_summary,"\n");
+                
+                if ($self->{'verbosity'}>=2) {
+                    my $outhandle = $self->{'outhandle'};
+                    my $lang_summary_str = join(" ",@lang_summary);
+                    print $outhandle $lang_summary_str;
                 }
             }
@@ -105 +114 @@
         return undef;
     }
-    
-    
    
     &util::rm($tmp_txt_filename);
@@ -119 +126 @@
 
     return $self->classify_contents($contents_ref,$filename,$filter_by_encoding);
-    
 }
    

gs2-extensions/ngramj/perllib/plugins/ReadTextFile.pm

@@ -120 +120 @@
     my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
 
-    print STDERR "**** In Pei Jones Local version of ReadTextFile\n";
+    my $verbosity = $self->{'verbosity'};
+    if ($verbosity>=2) {
+        my $outhandle = $self->{'outhandle'};
+        print $outhandle "++ Using Ngram-Java version of ReadTextFile.pm ++\n";
+    }
     
     my $outhandle = $self->{'outhandle'};
@@ -329 +333 @@
     my ($filename) = @_;
 
-    $self->{'textcat'} = new ngramj() if (!defined($self->{'textcat'}));
+    $self->{'textcat'} = new ngramj($self->{'verbosity'},$self->{'outhandle'}) if (!defined($self->{'textcat'}));
     
     my ($language, $encoding, $extracted_encoding);
@@ -629 +633 @@
 
     else { # need to use textcat to get either the language, or get both language and encoding
-    $self->{'ngramj'} = new ngramj() if (!defined($self->{'ngramj'}));
+    $self->{'ngramj'} = new ngramj($self->{'verbosity'},$self->{'outhandle'}) if (!defined($self->{'ngramj'}));
     
     if($found_html_encoding) { # know encoding, find language by limiting search to known encoding