Changeset 840 for trunk

Timestamp:

1999-12-13T16:57:18+13:00 (24 years ago)

Author:

davidb

Message:

Optimisations to make plugin go faster

File:

• trunk/gsdl/perllib/plugins/GMLPlug.pm (modified) (6 diffs)

trunk/gsdl/perllib/plugins/GMLPlug.pm

@@ -25 +25 @@
 
 # plugin which processes a GML format document
+# assumes that gml tags are all in lower-case.
 
 package GMLPlug;
@@ -33 +34 @@
 use doc;
 
-
 sub BEGIN {
     @ISA = ('BasPlug');
@@ -44 +44 @@
     return bless $self, $class;
 }
+
 
 sub is_recursive {
@@ -68 +69 @@
 sub read {
     my $self = shift (@_);
-    my ($pluginfo, $base_dir, $file, $metadata, $processor) = @_;
+    my ($pluginfo, $base_dir, $file, $metadata, $processor, $maxdocs) = @_;
     my $fullname = &util::filename_cat ($base_dir, $file);
 
     # see if this is a gml book
-    return undef unless (-f $fullname && $fullname =~ /\.gml(\.gz)?$/i);
-
-    my ($parent_dir, $gz) = $fullname =~ /^(.*?)[\/\\][^\/\\]+.gml(\.gz)?$/i;
-
-    if (defined $gz && $gz =~ /\.gz/i) {
+    return undef unless (-f $fullname && $fullname =~ /\.gml(\.gz)?$/io);
+
+    my ($parent_dir, $gz) = $fullname =~ /^(.*?)[\/\\][^\/\\]+.gml(\.gz)?$/io;
+
+    if (defined $gz && $gz =~ /\.gz/io) {
     $gz = 1;
     } else {
     $gz = 0;
     }
-
-    # create a new document
-    my $doc_obj = new doc ();
-    my $section = $doc_obj->get_top_section();
 
     print STDERR "GMLPlug: processing $file\n";
@@ -104 +101 @@
     }
 
+    undef $/;
     while (defined ($line = <INFILE>)) {
     $gml .= $line;
     }
+    $/ = "\n";
+
     close (INFILE);
 
-    # process the document
-    my $firstsection = 1;
-    while ($gml =~ /\S/) {
-    if ($gml =~ s/^\s*<gsdlsection([^>]*)>(.*?)(<\/?gsdlsection)/$3/is) {
-        my $tags = $1;
-        $tags = "" unless defined $tags;
-        my $text = &_unescape_text($2);
-
-        # create the section (unless this is the first section)
-        if (!$firstsection) {
-        $tags =~ s/gsdlnum\s*=\s*\"?(\d+)\"?//i;
-        if (defined $1) {
-            $section .= ".$1";
-            $doc_obj->create_named_section($section);
-        } else {
-            $section = $doc_obj->insert_section($doc_obj->get_end_child($section));
-        }
+    my @gml_sections = split("</gsdlsection>",$gml);
+    $gml = shift(@gml_sections);
+
+    my $no_docs = 0;
+    my $src_filename = "";
+
+    while (1)
+    {
+    # create a new document
+    my $doc_obj = new doc ();
+    my $section = $doc_obj->get_top_section();
+
+    # process the document
+    my $firstsection = 1;
+    while (1) 
+    {
+        my $tags = "";
+        my $text = "";
+
+        my @indenting_sections = split("<gsdlsection",$gml);
+        shift(@indenting_sections); # first entry is trivially empty
+####        print STDERR "**** no indenting sections = ", scalar(@indenting_sections), "\n";
+
+        foreach $gml (@indenting_sections)
+        {
+####        print STDERR "\n\n\n\n!!!!!!!!!!! gml = $gml\n\n\n";
+
+        if ($gml =~ m/^\s*([^>]*)>(.*)$/so)
+        {
+            $tags = $1 if (defined $1);
+            $text = &GMLPlug::_unescape_text($2);
+        }
+        else
+        {
+            print STDERR "GMLPlug::read - error in file $fullname\n";
+            print STDERR "text: \"$gml\"\n";
+            last;
+        }
+
+        # create the section (unless this is the first section)
+        if ($firstsection) 
+        {
+            $firstsection = 0;
+            
+            print STDERR "  0 of $src_filename\n" if ($no_docs==1);
+            $tags =~ m/gsdlsourcefilename\s*=\s*(\"([^\"]*)\")|(\w+)/o;
+            $src_filename = $2 || $3;
+            print STDERR "  $no_docs of $src_filename\n" if ($no_docs>=1);
+        }
+        else
+        {
+####            print STDERR "*** tags = $tags\n";
+
+            $tags =~ s/gsdlnum\s*=\s*\"?(\d+)\"?//o;
+            if (defined $1) 
+            {
+            $section .= ".$1";
+            $doc_obj->create_named_section($section);
+            } 
+            else 
+            {
+            $section = $doc_obj->insert_section($doc_obj->get_end_child($section));
+            }
+        }
+        
+
+        # add the tags
+        while ((defined $tags) 
+               && ($tags =~ s/^\s*(\w+)\s*=\s*\"([^\"]*)\"//o)) 
+        {
+            $doc_obj->add_utf8_metadata($section, $1 , 
+                        &GMLPlug::_unescape_text($2)) 
+            if (defined $1 and defined $2);
+        }
+        
+        # add the text
+        $doc_obj->add_utf8_text($section, $text) 
+            if ((defined $text) && ($text ne ""));      
         }
-        $firstsection = 0;
-
-        # add the tags
-        while ((defined $tags) && ($tags =~ s/^\s*(\w+)\s*=\s*\"([^\"]*)\"//)) {
-        $doc_obj->add_utf8_metadata($section, $1 , &_unescape_text($2)) 
-            if (defined $1 and defined $2);
-        }
-
-        # add the text
-        $doc_obj->add_utf8_text($section, $text) 
-        if ((defined $text) && ($text ne ""));
-
-    } elsif ($gml =~ s/^\s*<\/gsdlsection>//) {
+
+        $gml = shift(@gml_sections); # get next bit of data
+        last if (!defined $gml);
+###     print "####### before section = $section\n";
         $section = $doc_obj->get_parent_section ($section);
-
-    } else {
-        print STDERR "GMLPlug::read - error in file $fullname\n";
-        print STDERR "text: \"$gml\"\n";
-        last;
-    }
-    }
-
+###     print "####### after section = $section\n";
+        #### last if ($section eq "");
+    }
+
+<<<<<<< GMLPlug.pm
+    # add the associated files
+    $assoc_files 
+        = $doc_obj->get_metadata($doc_obj->get_top_section(), 
+                     "gsdlassocfile");
+    my ($assoc_file_info);
+    foreach $assoc_file_info (@$assoc_files) {
+        my ($assoc_file, $mime_type) = split (":", $assoc_file_info);
+        $doc_obj->associate_file(&util::filename_cat($parent_dir, $assoc_file),
+                     $assoc_file, $mime_type);
+=======
     # add the associated files
     $assoc_files = $doc_obj->get_metadata($doc_obj->get_top_section(), "gsdlassocfile");
@@ -157 +216 @@
     $doc_obj->associate_file(&util::filename_cat($parent_dir, $assoc_file),
                  &util::filename_cat($dir, $assoc_file), $mime_type);
+>>>>>>> 1.10
                  
-    }
-    $doc_obj->delete_metadata($doc_obj->get_top_section(), "gsdlassocfile");
-
-    # add metadata
-    foreach $field (keys(%$metadata)) {
-    # $metadata->{$field} may be an array reference
-    if (ref ($metadata->{$field}) eq "ARRAY") {
-        map { 
-        $doc_obj->add_metadata ($doc_obj->get_top_section(), $field, $_); 
-        } @{$metadata->{$field}};
-    } else {
-        $doc_obj->add_metadata ($doc_obj->get_top_section(), $field, $metadata->{$field}); 
-    }
-    }
-
-
-    # assume the document has an OID
-
-    # process the document
-    $processor->process($doc_obj, $file);
-        
-    return 1; # processed the file
+    }
+    $doc_obj->delete_metadata($doc_obj->get_top_section(), "gsdlassocfile");
+
+    # add metadata
+    foreach $field (keys(%$metadata)) {
+        # $metadata->{$field} may be an array reference
+        if (ref ($metadata->{$field}) eq "ARRAY") {
+        map { 
+            $doc_obj->add_metadata ($doc_obj->get_top_section(), $field, $_); 
+        } @{$metadata->{$field}};
+        } else {
+        $doc_obj->add_metadata ($doc_obj->get_top_section(), $field, $metadata->{$field}); 
+        }
+    }
+
+
+    # assume the document has an OID
+
+    # process the document
+    $processor->process($doc_obj, $file);
+
+    $no_docs++;
+
+    last if (defined $maxdocs && $maxdocs =~ /\d/ && $no_docs >= $maxdocs);
+
+    last if (!defined $gml);
+    }
+
+    return $no_docs; # no of docs processed
 }