Changeset 1279 for branches/New_Config_Format-branch/gsdl/perllib/plugins/HBPlug.pm

Timestamp:

2000-07-13T10:21:53+12:00 (24 years ago)

Author:

sjboddie

Message:

merged changes to trunk into New_Config_Format branch

File:

• branches/New_Config_Format-branch/gsdl/perllib/plugins/HBPlug.pm (modified) (8 diffs)

branches/New_Config_Format-branch/gsdl/perllib/plugins/HBPlug.pm

@@ -24 +24 @@
 ###########################################################################
 
-# plugin which process an HTML book directory
+# plugin which processes an HTML book directory
+
+# This plugin is used by the Humanity Library collections and does not handle
+# input encodings other than ascii or extended ascii
+
+# this code is kind of ugly and could no doubt be made to run faster, by leaving
+# it in this state I hope to encourage people to make their collections use
+# HBSPlug instead ;-)
+
+# Use HBSPlug if creating a new collection and marking up files like the
+# Humanity Library collections. HBSPlug accepts all input encodings but
+# expects the marked up files to be cleaner than those used by the
+# Humanity Library collections
 
 package HBPlug;
 
-use plugin;
 use ghtml;
 use BasPlug;
 use util;
-use lang;
 use doc;
-use cfgread;
 
 
@@ -43 +52 @@
 sub new {
     my ($class) = @_;
-    $self = new BasPlug ();
+    my $self = new BasPlug ("HBPlug", @_);
 
     return bless $self, $class;
 }
 
-sub is_recursive {
-    my $self = shift (@_);
-
-    return 0; # this is not a recursive plugin
-}
+sub init {
+    my $self = shift (@_);
+    my ($verbosity) = @_;
+
+    $self->BasPlug::init();
+
+    # this plugin only handles ascii encodings
+    if ($self->{'input_encoding'} !~ /^(iso_8859_1|ascii)$/) {
+    die "ERROR: HBPlug can handle only iso_8859_1 or ascii encodings.\n" .
+        $self->{'input_encoding'} . " is not an acceptable input_encoding value\n";
+    }
+}
+
+# this is included only to prevent warnings being printed out 
+# from BasPlug::init. The process_exp is not used by this plugin
+sub get_default_process_exp {
+    my $self = shift (@_);
+
+    return "This plugin does not use a process_exp\n";
+}
+
 
 sub HB_read_html_file {
@@ -65 +90 @@
 
     my $foundbody = 0;
-    $self->HB_gettext (\$foundbody, $text, FILE);
+    $self->HB_gettext (\$foundbody, $text, "FILE");
     close FILE;
 
@@ -72 +97 @@
     $foundbody = 1;
     open (FILE, $htmlfile) || return;
-    $self->HB_gettext (\$foundbody, $text, FILE);   
+    $self->HB_gettext (\$foundbody, $text, "FILE"); 
     close FILE;
     }
@@ -159 +184 @@
 }
 
+# if input_encoding is ascii we can call add_utf8_metadata
+# directly but if it's iso_8859_1 (the default) we need to call
+# add_metadata so that the ascii2utf8 conversion is done first
+# this should speed things up a little if processing an ascii only
+# document with input_encoding set to ascii
+sub HB_add_metadata {
+    my $self = shift (@_);
+    my ($doc_obj, $cursection, $field, $value) = @_;
+
+    if ($self->{'input_encoding'} eq "ascii") {
+    $doc_obj->add_utf8_metadata ($cursection, $field, $value);
+    } else {
+    $doc_obj->add_metadata ($cursection, $field, $value);
+    }
+}
 
 # return number of files processed, undef if can't process
@@ -192 +232 @@
 
     # add metadata for top level of document
-    foreach $field (keys(%$metadata)) {
+    foreach my $field (keys(%$metadata)) {
     # $metadata->{$field} may be an array reference
     if (ref ($metadata->{$field}) eq "ARRAY") {
         map { 
-        $doc_obj->add_metadata ($cursection, $field, $_); 
+        $self->HB_add_metadata ($doc_obj, $cursection, $field, $_); 
         } @{$metadata->{$field}};
     } else {
-        $doc_obj->add_metadata ($cursection, $field, $metadata->{$field}); 
+        $self->HB_add_metadata ($doc_obj, $cursection, $field, $metadata->{$field}); 
     }
     }
@@ -240 +280 @@
 
         # add the metadata to this section
-        $doc_obj->add_metadata ($cursection, "Title", $title);
+        $self->HB_add_metadata ($doc_obj, $cursection, "Title", $title);
 
         # clean up the section html
@@ -251 +291 @@
 
         # add the text for this section
-        $doc_obj->add_text ($cursection, $sectiontext);
-        
+        if ($self->{'input_encoding'} eq "ascii") {
+        $doc_obj->add_utf8_text ($cursection, $sectiontext);
+        } else {
+        $doc_obj->add_text ($cursection, $sectiontext);
+        }
     } else {
         print STDERR "WARNING - leftover text\n" , $self->shorten($html),