Changeset 16263
- Timestamp:
- 2008-07-01T17:13:09+12:00 (15 years ago)
- File:
-
- 1 edited
Legend:
- Unmodified
- Added
- Removed
-
gsdl/trunk/bin/script/lucene_passes.pl
r14924 r16263 31 31 die "GSDLOS not set\n" unless defined $ENV{'GSDLOS'}; 32 32 unshift (@INC, "$ENV{'GSDLHOME'}/perllib"); 33 unshift (@INC, "$ENV{'GSDLHOME'}/perllib/cpan"); 34 unshift (@INC, "$ENV{'GSDLHOME'}/perllib/plugins"); 35 unshift (@INC, "$ENV{'GSDLHOME'}/perllib/classify"); 36 } 37 33 } 34 35 36 use strict; 38 37 use util; 39 use ghtml; 38 40 39 41 40 sub open_java_lucene … … 50 49 my $java_cmd = "$java_lucene $cmd_options $doc_tag_level \"$full_builddir\" $indexdir"; 51 50 52 open (PIPEOUT, "| $java_cmd") or die "$PROGNAME - couldn't run $java_cmd\n"; 53 } 51 open (PIPEOUT, "| $java_cmd") or die "lucene_passes.pl - couldn't run $java_cmd\n"; 52 } 53 54 54 55 55 sub close_java_lucene … … 58 58 } 59 59 60 60 61 sub save_xml_doc 61 62 { 62 63 my ($full_textdir,$output_filename,$doc_xml) = @_; 63 $dir_sep = &util::get_os_dirsep(); 64 65 my $dir_sep = &util::get_os_dirsep(); 64 66 65 67 my $full_output_filename = &util::filename_cat($full_textdir,$output_filename); … … 76 78 } 77 79 80 78 81 sub compress_xml_doc 79 82 { … … 85 88 `gzip $full_output_filename`; 86 89 } 90 87 91 88 92 # This appears to be the callback that gets the xml stream during the … … 178 182 die "\n\nCannot continue"; 179 183 } 180 $removeoid = $argv[$i];181 print STDERR "\n\nWe're removing the document with id: '$ removeoid'\n";184 my $oid = $argv[$i]; 185 print STDERR "\n\nWe're removing the document with id: '$oid'\n"; 182 186 183 187 # Now, to make sure this gets through to the Java executable 184 $create = "-create -remove '$ removeoid'";188 $create = "-create -remove '$oid'"; 185 189 } 186 190 … … 205 209 206 210 if ($filtered_argc < 4) { 207 print STDERR "Usage: $PROGNAME[-create|-verbosity num] \"text\"|\"index\" doc-tag-level build-dir index-name\n";211 print STDERR "Usage: lucene_passes.pl [-create|-verbosity num] \"text\"|\"index\" doc-tag-level build-dir index-name\n"; 208 212 exit 1; 209 213 } … … 215 219 ### print STDERR "**** ARGS = ", join(" ", @argv), "\n"; 216 220 221 # We only need the Lucene handle opened if we are indexing the documents, not if we are just storing the text 222 if ($mode eq "index") { 223 open_java_lucene($doc_tag_level,$full_builddir,$indexdir,$create,$verbosity); 224 } 225 226 print STDERR "Monitoring for input!\n"; 217 227 my $full_textdir = &util::filename_cat($full_builddir,"text"); 218 219 if ($mode eq "index") {220 # don't need the lucene stuff if we are just storing the docs221 open_java_lucene($doc_tag_level,$full_builddir,$indexdir,$create,$verbosity);222 }223 print STDERR "Monitoring for input!\n";224 228 monitor_xml_stream($mode, $full_textdir); 229 225 230 if ($mode eq "index") { 226 231 close_java_lucene(); … … 228 233 } 229 234 230 $PROGNAME = $0;231 $PROGNAME =~ s/^.*\/(.*)$/$1/;232 235 233 236 &main(@ARGV); 234
Note:
See TracChangeset
for help on using the changeset viewer.