Changeset 1220

Timestamp:

2000-06-21T10:16:40+12:00 (24 years ago)

Author:

sjboddie

Message:

Caught HTMLPlug up with the changes I made to BasPlug. HTMLPlug now uses
the new BasPlug::read_file function and may be passed a -input_encoding
option (which may be set to Latin1, Arabic, gb, ascii, unicode, etc.).

File:

• trunk/gsdl/perllib/plugins/HTMLPlug.pm (modified) (9 diffs)

trunk/gsdl/perllib/plugins/HTMLPlug.pm

@@ -81 +81 @@
 sub new {
     my $class = shift (@_);
-    my $self = new BasPlug ();
+    my $self = new BasPlug (@_);
 
     if (!parsargv::parse(\@_, 
@@ -92 +92 @@
              q^w3mir^, \$self->{'w3mir'},
              q^assoc_files/.*/(?i)\.(jpe?g|gif|png|css|pdf)$^, \$self->{'assoc_files'},
-             q^rename_assoc_files^, \$self->{'rename_assoc_files'})) {
+             q^rename_assoc_files^, \$self->{'rename_assoc_files'},
+             "allow_extra_options")) {
+
     &print_usage();
     die "\n";
@@ -132 +134 @@
     my $cursection = $doc_obj->get_top_section();
     
-    # read in HTML file
-    open (FILE, $filename) || die "HTMLPlug::read - can't open $filename\n";
-    undef $/;
-    my $text = <FILE>;
-    $/ = "\n";
-    close FILE;
-    if (!defined $text || $text !~ /\w/) {
+    # read in HTML file ($text will be in utf8)
+    my $text = "";
+    $self->read_file ($filename, \$text);
+
+    if ($text !~ /\w/) {
     print STDERR "HTMLPlug: ERROR: $file contains no text\n" if $self->{'verbosity'};
     return 0;
@@ -153 +153 @@
     my $web_url = "http://$file";
     $web_url =~ s/\\/\//g; # for windows
-    $doc_obj->add_metadata($cursection, "URL", $web_url);
+    $doc_obj->add_utf8_metadata($cursection, "URL", $web_url);
 
     # remove header and footer
@@ -176 +176 @@
     $self->replace_images ($1, $2, $3, $base_dir, $file, $doc_obj, $cursection)/isge;
 
-    $doc_obj->add_text ($cursection, $text);
+    $doc_obj->add_utf8_text ($cursection, $text);
 
     # add an OID
@@ -362 +362 @@
             my $value = $1;
             $value =~ s/\s+/ /gs;
-            $doc_obj->add_metadata($section, $field, $value);
+            $doc_obj->add_utf8_metadata($section, $field, $value);
             next;
         }
@@ -379 +379 @@
             if ($title =~ /\w/) {
             $title =~ s/\s+/ /gs;
-            $doc_obj->add_metadata ($section, $field, $title);
+            $doc_obj->add_utf8_metadata ($section, $field, $title);
             next;
             }
@@ -390 +390 @@
         $tmptext =~ s/<[^>]*>//g;
         my $title = substr ($tmptext, 0, 100);
-        $doc_obj->add_metadata ($section, $field, $title);
+        $doc_obj->add_utf8_metadata ($section, $field, $title);
     }
 
@@ -402 +402 @@
         $tmptext = substr ($tmptext, 0, 200);
         $tmptext =~ s/\s\S*$/.../;
-        $doc_obj->add_metadata ($section, $field, $tmptext);
+        $doc_obj->add_utf8_metadata ($section, $field, $tmptext);
     }
     }