Changeset 617

Timestamp:

1999-09-22T10:00:53+12:00 (25 years ago)

Author:

sjboddie

Message:

a few fixes

Location:

trunk/gsdl/perllib/plugins

Files:

• GBTPlug.pm (modified) (1 diff)
• HTMLPlug.pm (modified) (3 diffs)

trunk/gsdl/perllib/plugins/GBTPlug.pm

@@ -96 +96 @@
     $text = &cnseg::segment($text);
 
-    $doc_obj->add_utf8_text($cursection, $text);
+    $doc_obj->add_utf8_text($cursection, "<pre>$text</pre>");
 
     # add Title metadata (filename);

trunk/gsdl/perllib/plugins/HTMLPlug.pm

@@ -93 +93 @@
     my $text = "";
     my $line = "";
-    my $donehead = 0;
     my $title = "";
     while (defined ($line = <FILE>)) {
@@ -99 +98 @@
     }
 
+    # we'll use the worthless alarm thingy to temporarily replace 
+    # '\n' so we'd better check it doesn't occur naturally
+    if ($text =~ /\a/) {
+    print STDERR "HTMLPlug::read - 'WARNING '\a' character occurs in text!!\n";
+    }
+
     # remove line breaks
-    $text =~ s/\s+/ /g;
+    $text =~ s/\n/\a/g;
 
     # see if there's a <title> tag
@@ -124 +129 @@
     $text =~ s/^.*?<body[^>]*>//i;
 
+    # and any other unwanted tags
+    $text =~ s/<(\/p|\/html|\/body)>//g;
+
     # fix up the image links
     $text =~ s/(<img[^>]*?src\s*=\s*\"?)([^\">]+)(\"?[^>]*>)/
     &replace_image_links($absdir, $doc_obj, $1, $2, $3)/ige;
 
-    # add a newline at the beginning of each paragraph
-    $text =~ s/(.)\s*<p\b/$1\n\n<p/gi;
-    
-    # add a newline every 80 characters at a word boundary
-    # Note: this regular expression puts a line feed before
-    # the last word in each section, even when it is not
-    # needed.
-    $text =~ s/(.{1,80})\s/$1\n/g;
+    # put line breaks back in
+    $text =~ s/\a/\n/g;
 
     $doc_obj->add_text ($cursection, $text);