Changeset 34109 for main

Timestamp:

2020-05-04T08:39:22+12:00 (4 years ago)

Author:

kjdon

Message:

tidied this up a bit. Now we leave in _textmonth00_ if the month is undefined. in gs2 this is resolved to empty string, in gs3 I have added 'undefined' string. when classifying the doc, sanity check the date rather than doing it later on, in multiple places in the code. if we are classifying bymonth and the month is absent or invalid, then set it to 00. no longer allows dates that don't at least start with yyyy.

File:

• main/trunk/greenstone2/perllib/classify/DateList.pm (modified) (9 diffs)

main/trunk/greenstone2/perllib/classify/DateList.pm

@@ -28 +28 @@
 # date is assumed to be in the form yyyymmdd
 
-# at present dates are split by year - this should change
-# jrm21 - added option "bymonth", which splits by year and month.
-
-# 23/09/03 Added some more options -kjdon.
-# these include:
+# options
+# -bymonth: splits by month as well as year
 # -nogroup, which makes each year (or year+month) an individual entry in 
 # the horizontal list and prevents compaction
@@ -111 +108 @@
     $self->{'sort'} = $self->strip_ex_from_metadata($self->{'sort'});
  
-    # now can have comma separated list of Dates - we just use the first one (for now)
+    # now can have comma separated list of Dates - we just use the first value found (for now)
     my @meta_list = split(/,/, $self->{"metadata"});
     $self->{'meta_list'} = \@meta_list;
@@ -136 +133 @@
 
     my $doc_OID = $doc_obj->get_OID();
-
+    my $outhandle = $self->{'outhandle'};
+    my $verbosity = $self->{'verbosity'};
     # find the first available metadata
     my $date;
@@ -147 +145 @@
     # if this document doesn't contain Date element we won't 
     # include it in this classification
+    print $outhandle "DateList: $doc_OID has no date, not including it\n" if $verbosity >=2;
     return;
     }
 
+    # sanity check date
+    if ($date !~ /^\d\d\d\d.*/) {
+    print $outhandle "DateList: $doc_OID date: '$date' malformed: expected it to start with yyyy; not classifying\n" if $verbosity >=2;
+    return;
+    }
+    if ($self->{'bymonth'}) {
+    # check that we have valid month - if not, set it to 00 == undefined
+    if ($date !~ /^\d\d\d\d-?\d\d/) {
+    print $outhandle "DateList $doc_OID date: '$date' has no month (expecting yyyymm... or yyyy-mm...), setting date to yyyy-00\n" if $verbosity >=2;
+    $date =~ s/^(\d\d\d\d).*$/$1-00/;
+    } else {
+        my ($year, $month) = $date =~ /^(\d\d\d\d)-?(\d\d)/;
+        if ($month > 12) {
+        print $outhandle "DateList $doc_OID date: '$date' has invalid month, setting date to $year-00\n" if $verbosity >=2;
+        $date = "$year-00";
+        }
+    }
+    }
+    
     my $sort_other = "";
     if (defined $self->{'sort'} && $self->{'sort'} ne "") {
@@ -157 +175 @@
     
     if (defined $self->{'list'}->{$doc_OID}) {
-    my $outhandle = $self->{'outhandle'};
-    print $outhandle "WARNING: DateList::classify called multiple times for $doc_OID\n";
+    
+    print $outhandle "WARNING: DateList::classify called multiple times for $doc_OID, overwriting previous stored date value ($self->{'list'}->{$doc_OID}) with $date$sort_other \n";
     } 
-        
+    
     $self->{'list'}->{$doc_OID} = "$date$sort_other";
 
@@ -218 +236 @@
     }
 
+    
 
     if ($self->{'bymonth'}) {
@@ -227 +246 @@
         my $date = $self->{'list'}->{$classification};
         $date =~ s/^(\d\d\d\d)-?(\d\d).*$/$1 _textmonth$2_/;
-        # sanity check if month is zero
-        if ($date =~ /00_$/) {
-            $date =~ s/^(\d\d\d\d).*$/$1/g;
-        }
         $classhash->{$date} = [] unless defined $classhash->{$date};
         push (@{$classhash->{$date}}, $classification);
@@ -240 +255 @@
         $date =~ s/^(\d\d\d\d)-?(\d\d).*$/$1 _textmonth$2_/;
         my ($year, $month)=($1,$2);
-        # sanity check if month is zero
-        if ($date =~ /00_$/) {
-            $date =~ s/^(\d\d\d\d).*$/$1/g;
-        }
         # create subclass if it doesn't already exist
         $classhash->{$year} = () unless defined $classhash->{$year};
@@ -268 +279 @@
           foreach my $subsubclass (@subsubclasslist) {
               my $monthname=$subsubclass;
-              if ($monthname >= 1 && $monthname <= 12) {
-              $monthname="_textmonth" . $monthname . "_";
-              }
+              $monthname="_textmonth" . $monthname . "_";
               my $monthclassify=$self->get_entry($monthname, $self->{'childtype'});
               push (@{$yearclassify->{'contains'}}, $monthclassify);