fixed parsing for science direct html with more than one <a href=> per one <tr>

[webpac] / WebPac.pm

diff --git a/WebPac.pm b/WebPac.pm
index 8139173..c187dee 100644 (file)
--- a/WebPac.pm
+++ b/WebPac.pm
@@ -28,20 +28,24 @@ my $MAX_HITS = $cfg_global->val('webpac', 'max_hits') || 0;
 my $ON_PAGE =$cfg_global->val('webpac', 'on_page') || 10;
 my $MIN_WILDCARD =$cfg_global->val('webpac', 'min_wildcard') || 1;
 my $TEMPLATE =$cfg_global->val('webpac', 'template');
-my $UNAC_FILTER =$cfg_global->val('global', 'unac_filter');
+my $UNAC_FILTER =$cfg_global->val('global', 'my_unac_filter');
 my $BASE_PATH =$cfg_global->val('webpac', 'base_path');
 # for pager
 my $pages_per_set = $cfg_global->val('webpac', 'pages_per_set') || 10;
 
+Text::Iconv->raise_error(0);     # Conversion errors raise exceptions
+
+my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET);
 
 if ($UNAC_FILTER) {
        require $UNAC_FILTER;
+} else {
+       sub WebPac::my_unac_string {
+               my ($charset, $string) = (@_);
+               return $string;
+       }
 }
 
-Text::Iconv->raise_error(0);     # Conversion errors raise exceptions
-
-my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET);
-
 # use path from cgi script to support templates in subdirs
 sub url_ex {
        my $q = shift || die "suff2file needs CGI object!";
@@ -255,7 +259,7 @@ sub show_results_list {
                while (my $search = shift @param_vals) {
                        my $s;
                        # remove accents
-                       $search = unac_string($CHARSET,$search);
+                       $search = my_unac_string($CHARSET,$search);
                        while ($search =~ s/\s*("[^"]+")\s*/ /) {
                                $s .= "$1 ";
                        }