Changeset 1700 for trunk/gsdl/perllib

Timestamp:

2000-11-27T08:39:17+13:00 (24 years ago)

Author:

say1

Message:

changed PSPlug to extract CreationDate, Title and Pages info.

File:

• trunk/gsdl/perllib/plugins/PSPlug.pm (modified) (3 diffs)

trunk/gsdl/perllib/plugins/PSPlug.pm

@@ -27 +27 @@
 
 use ConvertToPlug;
+use sorttools;
 
 sub BEGIN {
@@ -40 +41 @@
     my $self = new ConvertToPlug ($class, ("-convert_to","text",@_), "--", "-title_sub", 'Page\s+\d+');
     
+    if (!parsargv::parse(\@_, 
+             q^extract_date^, \$self->{'extract_date'},
+             q^extract_pages^, \$self->{'extract_pages'},
+             q^extract_title^, \$self->{'extract_title'},
+             "allow_extra_options")) {
+    print STDERR "\nIncorrect options passed to HTMLPlug, check your collect.cfg configuration file\n";
+    &print_usage();
+    die "\n";
+    }
+
     return bless $self, $class;
+}
+
+sub print_usage {
+    print STDERR "\n  usage: plugin PSPlug [options]\n\n";
+    print STDERR "  options:\n";
+    print STDERR "   -extract_date               Extract date from PS header\n";
+    print STDERR "   -extract_pages              Extract pages from PS header\n";
+    print STDERR "   -extract_title              Extract title from PS header\n";
 }
 
@@ -55 +74 @@
     return q^(?i)\.ps$^;
 }
-    
+
+sub extract_metadata_from_postscript {
+    my $self = shift (@_);
+    my $filename = shift (@_);
+    my $doc = shift (@_);
+    my $section = $doc->get_top_section();
+
+    my $title_found = 0;
+    my $pages_found = 0;
+    my $date_found = 0;
+
+    print STDERR "PSPlug: extracting PostScript metadata from \"$filename\"\n" 
+    if $self->{'verbosity'} > 1;
+
+    open(INPUT, "<$filename");
+    my $date;
+
+    while(my $line =<INPUT>) {
+    if ($self->{'extract_title'} && !$title_found) {
+        foreach my $word ($line =~ m|Title: ([-A-Za-z0-9@/\/\(\):,. ]*)|g) {
+        if ($word ne "") {
+            $doc->add_utf8_metadata($section, "Title", $word );
+            print "\"$word\" ===>>> \"Title\" = \"$word\"\n";
+            $title_found = 1;
+        }
+        }
+    }
+    if ($self->{'extract_date'} && !$date_found) {
+            foreach my $word ($line =~ m/(Creation[-A-Za-z0-9@\/\(\):,. ]*)/g) {
+                if ($word =~ m/ ([A-Za-z][A-Za-z][A-Za-z]) ([0-9 ][0-9])  ?[0-9: ]+ ([0-9]{4})/) {
+                    $date = &sorttools::format_date($2,$1,$3);
+            $doc->add_utf8_metadata($section, "Date", $date );
+                    print "\"$word\" ===>>> \"Date\" = \"$date\"\n";
+                }
+                if ($word =~ m/D:([0-9]{4})([0-9]{2})([0-9]{2})[0-9]{6}\)/) {
+                    $date = &sorttools::format_date($3,$2,$1);
+            $doc->add_utf8_metadata($section, "Date", $date );
+                    print "\"$word\" ===>>> \"Date\" = \"$date\"\n";
+                }
+                if ($word =~ m/CreationDate: ([0-9]{4}) ([A-Za-z][A-Za-z][A-Za-z]) ([0-9 ][0-9]) [0-9:]*/) {
+                    $date = &sorttools::format_date($3,$2,$1);
+            $doc->add_utf8_metadata($section, "Date", $date );
+                    print "\"$word\" ===>>> \"Date\" = \"$date\"\n";
+                }
+        $date_found = 1;
+            }
+    }
+    if ($self->{'extract_pages'} && !$pages_found) {
+        foreach my $word ($line =~ m/(Pages: [0-9]*)/g) {
+                my $digits = $word;
+                $digits =~ s/[^0-9]//g;
+        if ($digits ne "" && $digits ne "0") {
+            $doc->add_utf8_metadata($section, "Pages", $digits );
+            print "\"$word\" ===>>> \"Pages\" = $digits\n";
+            $pages_found = 1;
+        }
+            }
+    }
+    }
+}
+
 
 # do plugin specific processing of doc_obj for HTML type
 sub process {
     my $self = shift (@_);
+    my ($trash, $trash2, $path, $file, $trash3, $doc) = @_;
 
     print STDERR "PSPlug: passing $_[3] onto $self->{'convert_to'} Plug\n" 
     if $self->{'verbosity'} > 1;
     
+    &extract_metadata_from_postscript($self,"$path/$file", $doc);
+
     return ConvertToPlug::process_type($self,"ps",@_);
 }