Changeset 1244 for trunk/gsdl/perllib/plugins/BasPlug.pm

Timestamp:

2000-06-27T17:10:07+12:00 (24 years ago)

Author:

sjboddie

Message:

Caught up most general plugins (that's the ones in gsdlhome/perllib/plugins)
with changes to BasPlug so that they can all now use the new general plugin
options. Those I didn't do were FoxPlug (as it's not actually used anywhere
and I don't know what it does) and WebPlug (as it's kind of a work in
progress and doesn't really work anyway). All plugins will still work
(including all the collection specific ones that are laying around), some
of them just won't have access to the general options.
I also wrote a short perl script (pluginfo.pl) that prints out all the
options available to a given plugin.

File:

• trunk/gsdl/perllib/plugins/BasPlug.pm (modified) (13 diffs)

trunk/gsdl/perllib/plugins/BasPlug.pm

@@ -33 +33 @@
 use doc;
 
-sub print_usage {
+sub print_general_usage {
     my ($plugin_name) = @_;
 
-    print STDERR "\nThe $plugin_name plugin uses an incorrect general option (general options are those\n";
-    print STDERR "available to all plugins). Check your collect.cfg configuration file.\n";
-    
     print STDERR "\n  usage: plugin $plugin_name [options]\n\n";
-    print STDERR "  currently supported general options are:\n";
     print STDERR "   -input_encoding   The encoding of the source documents. Documents will be\n";
     print STDERR "                     converted from these encodings and stored internally as\n";
@@ -71 +67 @@
 }
 
+# print_usage should be overridden for any sub-classes having 
+# their own plugin specific options
+sub print_usage {
+    print STDERR "\nThis plugin has no plugin specific options\n\n";
+
+}
+
 sub new {
     my $class = shift (@_);
@@ -85 +88 @@
              q^extract_acronyms^, \$self->{'extract_acronyms'},
              "allow_extra_options")) {
-    &print_usage($plugin_name);
+
+    print STDERR "\nThe $plugin_name plugin uses an incorrect general option (general options are those\n";
+    print STDERR "available to all plugins). Check your collect.cfg configuration file.\n";
+        &print_general_usage($plugin_name);
     die "\n";
     }
@@ -103 +109 @@
     # set process_exp and block_exp to defaults unless they were
     # explicitly set
-    if ((!$self->is_recursive()) && 
+
+    if ((!$self->is_recursive()) and 
     (!defined $self->{'process_exp'}) || ($self->{'process_exp'} eq "")) {
-    
+
     $self->{'process_exp'} = $self->get_default_process_exp ();
     if ($self->{'process_exp'} eq "") {
-        warn ref($self) . " Warning: Non-recursive plugin has no process_exp so will have no effect\n";
+        warn ref($self) . " Warning: Non-recursive plugin has no process_exp\n";
     }
     }
@@ -115 +122 @@
     $self->{'block_exp'} = $self->get_default_block_exp ();
     }
+    
+    # handle input_encoding aliases
+    $self->{'input_encoding'} = "iso_8859_1" if $self->{'input_encoding'} eq "Latin1";
+    $self->{'input_encoding'} = "windows_1256" if $self->{'input_encoding'} eq "Arabic";
 }
 
@@ -152 +163 @@
 # process() function and let this read() function keep control.
 #
+# recursive plugins (e.g. RecPlug) and specialized plugins like those
+# capable of processing many documents within a single file (e.g.
+# GMLPlug) should normally implement their own version of read()
+#
 # Return number of files processed, undef if can't process
 # Note that $base_dir might be "" and that $file might 
@@ -165 +180 @@
 
     my $filename = &util::filename_cat($base_dir, $file);
-    return 0 if $filename =~ /$self->{'block_exp'}/;
+    return 0 if $self->{'block_exp'} ne "" && $filename =~ /$self->{'block_exp'}/;
     if ($filename !~ /$self->{'process_exp'}/ || !-f $filename) {
     return undef;
@@ -174 +189 @@
     # create a new document
     my $doc_obj = new doc ($file, "indexed_doc");
-    my $cursection = 
     
     # read in file ($text will be in utf8)
@@ -190 +204 @@
 
     # do plugin specific processing of doc_obj
-    $self->process (\$text, $pluginfo, $base_dir, $file, $metadata, $doc_obj);
-
-    # add text
-    $doc_obj->add_utf8_text ($cursection, $text);
+    return undef unless defined ($self->process (\$text, $pluginfo, $base_dir, $file, $metadata, $doc_obj));
 
     # do any automatic metadata extraction
@@ -207 +218 @@
 }
 
+# returns undef if file is rejected by the plugin
 sub process {
     my $self = shift (@_);
@@ -212 +224 @@
 
     die "Basplug::process function must be implemented in sub-class\n";
+
+    return undef; # never gets here
 }
 
@@ -223 +237 @@
 
     $$textref = "";
-    my $encoding = "";
-    if ($self->{'input_encoding'} =~ /^(Latin1|iso_8859_1)$/) {
-    $encoding = "iso_8859_1";
-    } elsif ($self->{'input_encoding'} =~ /^(Arabic|windows_1256)$/) {
-    $encoding = "windows_1256";
-    } else {
-    $encoding = $self->{'input_encoding'};
-    }
 
     open (FILE, $filename) || die "BasPlug::read_file could not open $filename for reading ($!)\n";
 
-    if ($encoding eq "ascii") {
+    if ($self->{'input_encoding'} eq "ascii") {
     undef $/;
     $$textref = <FILE>;
@@ -241 +247 @@
     my $reader = new multiread();
     $reader->set_handle ('BasPlug::FILE');
-    $reader->set_encoding ($encoding);
+    $reader->set_encoding ($self->{'input_encoding'});
     $reader->read_file ($textref);
 
-    if ($encoding eq "gb") {
+    if ($self->{'input_encoding'} eq "gb") {
         # segment the Chinese words
         $$textref = &cnseg::segment($$textref);