Changeset 433

Timestamp:

1999-08-04T17:23:41+12:00 (25 years ago)

Author:

sjboddie

Message:

added gzip option to import.pl

Location:

trunk/gsdl

Files:

• bin/script/import.pl (modified) (3 diffs)
• perllib/docsave.pm (modified) (7 diffs)
• perllib/plugins/GMLPlug.pm (modified) (2 diffs)

trunk/gsdl/bin/script/import.pl

@@ -27 +27 @@
     print STDERR "   -removeold             Will remove the old contents of the archives\n";
     print STDERR "                          directory -- use with care\n";
+    print STDERR "   -gzip                  Use gzip to compress resulting gml documents\n";
     print STDERR "   -maxdocs number        Maximum number of documents to import\n\n";
 }
@@ -40 +41 @@
              'keepold', \$keepold,
              'removeold', \$removeold,
+             'gzip', \$gzip,
              'maxdocs/\d+/', \$maxdocs)) {
     &print_usage();
@@ -110 +112 @@
 
     # create a docsave object to process the documents
-    $processor = new docsave ($collection, $archive_info, $verbosity);
+    $processor = new docsave ($collection, $archive_info, $verbosity, $gzip);
     $processor->setarchivedir ($archivedir);
 

trunk/gsdl/perllib/docsave.pm

@@ -15 +15 @@
 
 sub new {
-    my ($class, $collection, $archive_info, $verbosity) = @_;
+    my ($class, $collection, $archive_info, $verbosity, $gzip) = @_;
     my $self = new docproc ();
 
@@ -21 +21 @@
     $self->{'archive_info'} = $archive_info;
     $self->{'verbosity'} = $verbosity;
+    $self->{'gzip'} = $gzip;
 
     # set a default for the archive directory
@@ -50 +51 @@
     # same one.
     $doc_dir = $doc_info->[0];
-    $doc_dir =~ s/\/?doc\.gml$//;
+    $doc_dir =~ s/\/?doc\.gml(\.gz)?$//;
 
     } else {
@@ -63 +64 @@
         }
     } while ($doc_dir_rest ne "" && 
-         (-d "$archive_dir/$doc_dir.dir" || 
+         ((-d &util::filename_cat ($archive_dir, "$doc_dir.dir")) || 
           ($self->{'archive_info'}->size() >= 1024 && $doc_dir_num < 2)));
     $doc_dir .= ".dir";
@@ -75 +76 @@
     foreach $assoc_file (@{$doc_obj->get_assoc_files()}) {
     if (-e $assoc_file->[0]) {
-        &util::cp ($assoc_file->[0], "$archive_dir/$doc_dir/$assoc_file->[1]");
+        my $afile = &util::filename_cat($archive_dir, $doc_dir, $assoc_file->[1]);
+        &util::cp ($assoc_file->[0], $afile);
         $doc_obj->add_metadata ($doc_obj->get_top_section(), 
                     "gsdlassocfile", 
@@ -81 +83 @@
     } else {
         print STDERR "docsave::process couldn't copy the associated file " .
-        "$assoc_file->[0] to $archive_dir/$doc_dir/$assoc_file->[1]\n"
+        "$assoc_file->[0] to $afile\n";
     }
     }
 
     # save this document
-    if (!open (OUTDOC, ">$archive_dir/$doc_dir/doc.gml")) {
-    print STDERR "docsave::process could not write to file " . 
-        "$archive_dir/$doc_dir/doc.gml\n";
+    my $doc_file = &util::filename_cat ($archive_dir, $doc_dir, "doc.gml");
+    my $short_doc_file = &util::filename_cat ($doc_dir, "doc.gml");
+
+    if (!open (OUTDOC, ">$doc_file")) {
+    print STDERR "docsave::process could not write to file $docfile\n";
     return;
     }
@@ -94 +98 @@
     close OUTDOC;
 
+    if ($self->{'gzip'}) {
+    `gzip $doc_file`;
+    $doc_file .= ".gz";
+    $short_doc_file .= ".gz";
+    if (!-e $doc_file) {
+        print STDERR "error while gzipping: $doc_file doesn't exist\n";
+        return;
+    }
+    }
+
     # store reference in the archive_info
-    $self->{'archive_info'}->add_info($OID, 
-                      "$doc_dir/doc.gml");
+    $self->{'archive_info'}->add_info($OID, $short_doc_file);
     
 }

trunk/gsdl/perllib/plugins/GMLPlug.pm

@@ -47 +47 @@
 
     # see if this is a gml book
-    return undef unless (-f $fullname && $fullname =~ /\.gml$/i);
+    return undef unless (-f $fullname && $fullname =~ /\.gml(\.gz)?$/i);
 
-    my ($parent_dir) = $fullname =~ /^(.*)\/[^\/]+.gml$/;
+    my ($parent_dir, $gz) = $fullname =~ /^(.*)\/[^\/]+.gml(\.gz)?$/i;
+
+    if (defined $gz && $gz =~ /\.gz/i) {
+    $gz = 1;
+    } else {
+    $gz = 0;
+    }
 
     # create a new document
@@ -60 +66 @@
     my $gml = "";
     my $line = "";
-    if (!open (INFILE, $fullname)) {
-    print STDERR "GMLPlug::read - couldn't read $fullname\n";
-    return undef;
+
+    if ($gz) {
+    if (!open (INFILE, "zcat $fullname |")) {
+        print STDERR "GMLPlug::read - zcat couldn't read $fullname\n";
+        return undef;
+    }
+    } else {
+    if (!open (INFILE, $fullname)) {
+        print STDERR "GMLPlug::read - couldn't read $fullname\n";
+        return undef;
+    }
     }
+
     while (defined ($line = <INFILE>)) {
     $gml .= $line;