Changeset 2480 for trunk/gsdl/perllib/mgppbuildproc.pm
- Timestamp:
- 2001-05-29T10:51:39+12:00 (23 years ago)
- File:
-
- 1 edited
Legend:
- Unmodified
- Added
- Removed
-
trunk/gsdl/perllib/mgppbuildproc.pm
r1917 r2480 67 67 $self->{'num_processed_bytes'} = 0; 68 68 $self->{'outhandle'} = $outhandle; 69 $self->{'store_text'} = 1; 69 70 $self->{'dontindex'} = {}; 70 71 $self->{'indexfieldmap'} = {}; … … 175 176 176 177 return $self->{'indexing_text'}; 178 } 179 180 sub set_store_text { 181 my $self = shift (@_); 182 my ($store_text) = @_; 183 184 $self->{'store_text'} = $store_text; 177 185 } 178 186 … … 547 555 if ($self->{'indexing_text'}) { #tag the text with <Text>...</Text>, add the <Paragraph> tags and strip out html if needed 548 556 $new_text .= "<TX>\n"; 549 $tmp_text .= $doc_obj->get_text ($section) ;557 $tmp_text .= $doc_obj->get_text ($section) if $self->{'store_text'}; 550 558 $tmp_text = $self->preprocess_text($tmp_text, $self->{'strip_html'}, $self->{'levels'}->{'Paragraph'}); 551 559 … … 556 564 } 557 565 else { # leave html stuff in, and dont add Paragraph tags - never retrieve paras at the moment 558 $new_text .= $doc_obj->get_text ($section) ;566 $new_text .= $doc_obj->get_text ($section) if $self->{'store_text'}; 559 567 #if ($self->{'levels'}->{'Paragraph'}) { 560 568 #$self->find_paragraphs($new_text); … … 569 577 my ($mfield, $mvalue) = (@$pair); 570 578 # check fields here, maybe others dont want - change to use dontindex!! 571 if ($mfield ne "Identifier" && $mfield ne "classifytype" && 579 if ($mfield ne "Identifier" && $mfield ne "classifytype" && $mfield ne "assocfilepath" && 572 580 $mfield !~ /^gsdl/ && defined $mvalue && $mvalue ne "") { 573 581
Note:
See TracChangeset
for help on using the changeset viewer.