more selectors for policies statistics

eprints1.pl

@@ -112,8 +112,8 @@ my $op_mode= 'unknown';
 
 # ======================================================================
 # BEGIN OT2UT: Othes to Utheses 
-# my $ot2ut_context= 'ot2ut-entw'; # TODO: parametrize
-my $ot2ut_context= 'ot2ut-test'; # TODO: parametrize
+my $ot2ut_context= 'ot2ut-entw'; # TODO: parametrize
+# my $ot2ut_context= 'ot2ut-test'; # TODO: parametrize
 my $oma_sleep_time= 10;
 
 my %map_ot2ut_roles=
@@ -169,6 +169,9 @@ my $ignore_errors= 0;
 if ($0 eq './ot2ut.pl') { $op_mode= 'ot2ut'; $do_upload= 1; $MAX_SYNC= 1; }
 if ($0 eq './oma.pl')   { $op_mode= 'oma';   $do_upload= 1; }
 
+my $flag_add_utheses_policies= 1;
+my $flag_add_identifiers= 1;
+
 my %doc_embargo_dates;
 # END OT2UT: Othes to Utheses 
 # ======================================================================
@@ -394,6 +397,10 @@ elsif ($op_mode eq 'debug_filenames')
 {
   debug_filenames();
 }
+elsif ($op_mode eq 'update-policies')
+{
+  update_policies();
+}
 elsif ($op_mode eq 'policies-stats')
 {
   policies_stats();
@@ -3427,7 +3434,7 @@ mysql> select count(*), einverstaendnis, sperre, full_text_status
   {
     $policies->{abstract_locked}= $policies->{keywords_locked}= 1;
 
-=begin commment
+=begin comment
 
 mysql> select count(*), abstract_nicht_anzeigen from eprint group by abstract_nicht_anzeigen;
 +----------+-------------------------+
@@ -3907,7 +3914,7 @@ sub update_policies
       utheses_upload_result_json_path => $utheses_upload_result_json_path,
     };
 
-    map { $full_data->{$_}= $row->{$_} } qw(eprint_status einverstaendnis sperre full_text_status);
+    map { $full_data->{$_}= $row->{$_} } qw(eprint_status einverstaendnis sperre full_text_status urn doi);
 
     # $show++ if ($docs->{cnt_embargo} > 0);
     # $show++ if ($docs->{cnt_public} > 0 && $docs->{cnt_restricted} > 0);
@@ -3982,7 +3989,7 @@ sub policies_stats
     elsif ($context eq 'ot2ut-entw') { $context= 'entw'; }
     elsif ($context eq 'ot2ut-prod') { $context= 'prod'; }
 
-    $synced{$eprint_id}->{$context}= $lastmod;
+    $synced{$eprint_id}->{$context}= [ $lastmod, $utheses_id ];
   }
 
   # MAIN PART: analyze othes policies collection
@@ -3990,6 +3997,15 @@ sub policies_stats
   my $cur_utp= $col_utp->find({});
   
   my @columns= qw(eprint_status einverstaendnis sperre hds fts docs pub restr errors);
+  if ($flag_add_utheses_policies)
+  {
+    push (@columns, qw(lrq kwl absl ftl));
+  }
+  if ($flag_add_identifiers)
+  {
+    push (@columns, qw(urn doi));
+  }
+
   my $cctab= new cctab(columns => \@columns);
 
   my $max; #= 1000;
@@ -4000,8 +4016,8 @@ sub policies_stats
     last unless (defined ($row_utp));
     # print __LINE__, " row_utp: ", Dumper($row_utp); last;
 
-    my ($eprint_id, $eprint_status, $date_sperre, $einverstaendnis, $lastmod, $full_text_status, $sperre, $cnt_errors, $cnt_warnings, $docs)=
-       map { $row_utp->{$_} } qw(eprint_id eprint_status date_sperre einverstaendnis lastmod full_text_status sperre cnt_errors cnt_warnings docs);
+    my ($eprint_id, $eprint_status, $date_sperre, $einverstaendnis, $lastmod, $full_text_status, $sperre, $cnt_errors, $cnt_warnings, $docs, $utp, $urn, $doi)=
+       map { $row_utp->{$_} } qw(eprint_id eprint_status date_sperre einverstaendnis lastmod full_text_status sperre cnt_errors cnt_warnings docs ut_public urn doi);
 
     my ($cnt_docs, $cnt_public, $cnt_restricted)=
        map { $docs->{$_} } qw(cnt_docs cnt_public cnt_restricted);
@@ -4014,22 +4030,50 @@ sub policies_stats
       $has_date_sperre=($date_sperre lt $ts_now) ? 'past' : 'future';
     }
 
-    my $bucket= $cctab->bucket($eprint_status, $einverstaendnis, $sperre, $has_date_sperre, $full_text_status, $cnt_docs, $cnt_public, $cnt_restricted, $has_errors);
+    my @bucket_selectors= ($eprint_status, $einverstaendnis, $sperre, $has_date_sperre, $full_text_status, $cnt_docs, $cnt_public, $cnt_restricted, $has_errors);
+
+    my ($utptp, $lrq, $kwl, $absl, $ftl);
+    my $utpt= $utp->{thesis};
+    if (defined ($utpt))
+    {
+      ($utptp, $urn, $doi)= map { $utpt->{$_} } qw(policies urn doi);
+    }
+    if (defined ($utptp))
+    {
+      ($lrq, $kwl, $absl, $ftl)= map { $utptp->{$_} } qw(lock_request keywords_locked abstract_locked fulltext_locked);
+
+    }
+    if ($flag_add_utheses_policies)
+    {
+      push (@bucket_selectors, $lrq, $kwl, $absl, $ftl);
+    }
+
+    if ($flag_add_identifiers)
+    {
+      $urn= 'present' if (defined ($urn));
+      $doi= 'present' if (defined ($doi));
+
+      push (@bucket_selectors, $urn, $doi);
+    }
+
+    my $bucket= $cctab->bucket(@bucket_selectors);
 
     $bucket->{othes}++;
     push (@{$bucket->{othes_ids}} => $eprint_id);
 
-    my $found_utheses= 0;
+    if (exists ($synced{$eprint_id}))
+    {
       foreach my $context (@contexts)
       {
         if (exists ($synced{$eprint_id}->{$context}))
         {
+          my $x= $synced{$eprint_id}->{$context};
           $bucket->{$context}++;
-        push (@{$bucket->{$context . '_ids'}} => $eprint_id);
-        $found_utheses++;
+          push (@{$bucket->{$context . '_ids'}} => [ $eprint_id, $x->[1] ]);
+        }
       }
+      delete($synced{$eprint_id});
     }
-    delete($synced{$eprint_id}) if (exists ($synced{$eprint_id}));
 
     last if (defined($max) && --$max <= 0);
   }