Changeset 10218 for trunk/gsdl/perllib/classify/Phind.pm

Timestamp:

2005-07-06T15:27:45+12:00 (19 years ago)

Author:

kjdon

Message:

Jeffrey's new parsing modifications, committed approx 6 July, 15.16

File:

• trunk/gsdl/perllib/classify/Phind.pm (modified) (6 diffs)

trunk/gsdl/perllib/classify/Phind.pm

@@ -150 +150 @@
 
 sub new {
-    my $class = shift (@_);
-    my $self = new BasClas($class, @_);
-
-    # 14-05-02 To allow for proper inheritance of arguments - John Thompson
-    my $option_list = $self->{'option_list'};
-    push( @{$option_list}, $options );
-
-    if ($self->{'info_only'}) {
-    # created from classinfo.pl - don't need to parse the arguments
-    return bless $self, $class;
-    }
-
-    my $out = $self->{'outhandle'};
+    my ($class) = shift (@_);
+    my ($classifierslist,$inputargs,$hashArgOptLists) = @_;
+    push(@$classifierslist, $class);
+
+    if(defined $arguments){ push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});}
+    if(defined $options) { push(@{$hashArgOptLists->{"OptList"}},$options)};
+
+    my $self = (defined $hashArgOptLists)? new BasClas($classifierslist,$inputargs,$hashArgOptLists): new BasClas($classifierslist,$inputargs);
 
     # Ensure the Phind generate scripts are in place
@@ -168 +163 @@
     $file1 .= ".exe" if $ENV{'GSDLOS'} =~ /^windows$/;
     my $src = &util::filename_cat($ENV{'GSDLHOME'}, "src", "phind", "generate");
-
     if (!(-e $file1)) {
     print STDERR "Phind.pm: ERROR: The Phind \"suffix\" program is not installed.\n\n";
     exit(1);
     }
-
-    # Parse classifier arguments
-    my $builddir = "";
-    if (!parsargv::parse(\@_, 
-             q^text/.*/section:Title,section:text^, \$self->{'indexes'},
-             q^title/.*/Title^, \$self->{'titlefield'},
-             q^buttonname/.*/Phrase^, \$self->{'buttonname'},
-             q^language/.*/en^, \$language,
-             q^savephrases/.*/^, \$self->{'savephrases'},
-             q^suffixmode/\d/1^, \$self->{'suffixmode'},
-             q^min_occurs/\d/2^, \$self->{'min_occurs'},
-             q^thesaurus/.*/^, \$self->{'thesaurus'},
-             q^untidy^, \$self->{'untidy'},
-             "allow_extra_options")) {
-    
-    print STDERR "\nIncorrect options passed to $class, check your collect.cfg file\n";
-    $self->print_txt_usage("");  # Use default resource bundle
-    die "\n";
-    }
-    
-    # classifier information
-    $self->{'collection'} = $ENV{'GSDLCOLLECTION'};
-
-    # limit languages
-    $self->{'language_exp'} = $language;
-    
-    # collection directories
-    $self->{'collectiondir'} = $ENV{'GSDLCOLLECTDIR'};
-
+    
+    # Transfer value from Auto Parsing to the variable name that used in previous GreenStone.
+    $self->{"indexes"} = $self->{"text"};
+
+    # Further setup
+    $self->{'collection'} = $ENV{'GSDLCOLLECTION'}; # classifier information
+    $self->{'collectiondir'} = $ENV{'GSDLCOLLECTDIR'}; # collection directories
     if (! defined $self->{'builddir'}) {
     $self->{'builddir'} = &util::filename_cat($ENV{'GSDLCOLLECTDIR'}, "building");
     } 
-
     $self->{'total'} = 0;
-   
+    
+    # Clean out the unused keys
+    delete $self->{"text"};
+
     return bless $self, $class;
 }
@@ -265 +239 @@
     my $top_section = $doc_obj->get_top_section();
 
-    my $titlefield = $self->{'titlefield'};
+    my $titlefield = $self->{'title'};
     
     my $title = $doc_obj->get_metadata_element ($top_section, $titlefield);
@@ -276 +250 @@
     # Only consider the file if it is in the correct language
     my $doclanguage = $doc_obj->get_metadata_element ($top_section, "Language");
-    my $phrlanguage = $self->{'language_exp'};
+    my $phrlanguage = $self->{'language'};
     return if ($doclanguage && ($doclanguage !~ /$phrlanguage/i));
     
@@ -282 +256 @@
     $self->{'total'} ++;
     print "file $self->{'total'}: $file\n" if ($self->{'$verbosity'});
-    
 
     # Store document details
@@ -698 +671 @@
     my $phinddir = $self->{'phinddir'};
 
-    my $language_exp = $self->{'language_exp'};
+    my $language_exp = $self->{'language'};
 
     my ($w, $l, $line, $word);