- simplified (and hopefully increased performance) parsing filesize

perl/exit_strategy.pl

@@ -394,10 +394,13 @@ sub check_if_db_conform ($string, $filename) {
 #
 ###############################################################################
   my $compiled_xpath_dmdSec = XML::LibXML::XPathExpression->new('/mets:mets/mets:dmdSec');
-  my $compiled_xpath_amdSec = XML::LibXML::XPathExpression->new('/mets:mets/mets:amdSec[starts-with(@ID, \'REP\')]');
+  my $compiled_xpath_amdSec_rep = XML::LibXML::XPathExpression->new('/mets:mets/mets:amdSec[starts-with(@ID, \'REP\')]');
+  my $compiled_xpath_amdSec_fl = XML::LibXML::XPathExpression->new('/mets:mets/mets:amdSec[starts-with(@ID, \'FL\')]');
   my $compiled_xpath_fileSec = XML::LibXML::XPathExpression->new('/mets:mets/mets:fileSec');
   my $compiled_xpath_titles = XML::LibXML::XPathExpression->new('mets:mdWrap[1]/mets:xmlData[1]/dc:record/dc:title[1]');
+  my $compiled_xpath_fileGrp = XML::LibXML::XPathExpression->new('/mets:mets/mets:fileSec/mets:fileGrp');
   my $compiled_xpath_files = XML::LibXML::XPathExpression->new("mets:file/mets:FLocat/\@xlin:href");
+  my $compiled_xpath_filepids = XML::LibXML::XPathExpression->new('mets:file/@ID');
   my $compiled_xpath_dcrecords = XML::LibXML::XPathExpression->new('mets:mdWrap/mets:xmlData/dc:record/*');
   my $compiled_xpath_repid = XML::LibXML::XPathExpression->new('/mets:mets/mets:amdSec[starts-with(@ID, \'REP\')]/@ID');
   my $str_local_record = "mets:techMD/mets:mdWrap/mets:xmlData/*[local-name()=\'dnx\']/*[local-name()=\'section\']/*[local-name()=\'record\']";
@@ -447,31 +450,33 @@ sub check_if_db_conform ($string, $filename) {
     # if $found == 0 do nothing
     return $repid;
   }
+  sub get_filepids_ref ($xp, $fsp, $repid) {
+    my $filegrp = $xp->findnodes("mets:fileGrp[\@ADMID='$repid']", $fsp)->[0];
+    my @files_nodes = $xp->findnodes($compiled_xpath_filepids, $filegrp);
+    my @filepids =  map { $_->nodeValue; } @files_nodes;
+    return \@filepids;
+  }
 
   sub get_files_ref ($xp, $fsp, $repid){
     my $filegrp = $xp->findnodes("mets:fileGrp[\@ADMID='$repid']", $fsp)->[0];
     my @files_nodes = $xp->findnodes($compiled_xpath_files, $filegrp);
-    my @files =  map { my $tmp= $_->nodeValue; $tmp } @files_nodes;
+    my @files =  map { $_->nodeValue; } @files_nodes;
     return \@files;
   }
 
-  sub get_filesize ($xp, $filepath) {
-    my $xpath_id =<<"XPATH";
-/mets:mets/mets:fileSec/mets:fileGrp/mets:file[mets:FLocat/\@*[
-          namespace-uri()='http://www.w3.org/1999/xlink' and local-name()='href']='$filepath']/\@ID
-XPATH
-    my $xpath =<<"XPATH2";
-        /mets:mets/mets:amdSec/mets:techMD[
-          $xpath_id
-          and
-          starts-with(\@ID,$xpath_id)
-        ]/mets:mdWrap/mets:xmlData/*[namespace-uri()='http://www.exlibrisgroup.com/dps/dnx' and local-name()='dnx']
+  sub get_file_path($xp, $fsp, $filepid) {
+    return $xp->findvalue("mets:fileGrp/mets:file[\@ID=\"$filepid\"]/mets:FLocat/\@xlin:href", $fsp);
+  }
+
+  sub get_filesize ($xp, $filepid) {
+    my $xpath =<<"XPATH";
+        /mets:mets/mets:amdSec[starts-with(\@ID, \'$filepid\')]/mets:techMD[\@ID=\"$filepid-amd-tech\"]
+        /mets:mdWrap/mets:xmlData/*[namespace-uri()='http://www.exlibrisgroup.com/dps/dnx' and local-name()='dnx']
       /*[namespace-uri()='http://www.exlibrisgroup.com/dps/dnx' and local-name()='section']
       /*[namespace-uri()='http://www.exlibrisgroup.com/dps/dnx' and local-name()='record']
       /*[namespace-uri()='http://www.exlibrisgroup.com/dps/dnx' and local-name()='key' and \@id='fileSizeBytes']/text()
-XPATH2
-    my $size = $xp->findvalue($xpath);
-    return $size;
+XPATH
+    return $xp->findvalue($xpath);
   }
 
   sub get_purged_states($xp, $amd) {
@@ -523,7 +528,7 @@ XPATH2
     my $xp = get_xpath_context();
     $xp->setContextNode($dom);
     my $dmdsec = $xp->findnodes($compiled_xpath_dmdSec)->[0];
-    my $amdsec = $xp->findnodes($compiled_xpath_amdSec)->[0];
+    my $amdsec = $xp->findnodes($compiled_xpath_amdSec_rep)->[0];
     my $filesec= $xp->findnodes($compiled_xpath_fileSec)->[0];
     my $ret;
     $ret->{"filename"} = $filename;
@@ -559,11 +564,17 @@ XPATH2
     $ret->{"repid"} = $repid;
     ############################################
     # get all files of LOCAL representation
+    $ret->{"filepids"} = get_filepids_ref($xp, $filesec, $repid);
     $ret->{"files"}  = get_files_ref($xp, $filesec, $repid);
-    foreach my $location (@{$ret->{"files"}}) {
-      $ret->{"sizes"}->{$location} = get_filesize($xp, $location);
-    }
+    foreach my $fpid (@{$ret->{"filepids"}}) {
+      #say "pid=$fpid";
+      my $location = get_file_path($xp, $filesec, $fpid);
+      #say "location=$location";
+      my $size = get_filesize($xp, $fpid);
+      #say "size=$size";
+      $ret->{"sizes"}->{$location} = $size ;
 
+    }
     return $ret;
   }