Fixing upload json multi categorical data from field book

lib/CXGN/Phenotype.pm

@@ -478,6 +478,7 @@ sub store_external_references {
 
     if (! $self->phenotype_id()) {
         #print STDERR "Can't store external references on this phenotype because there is no phenotype_id\n";
+        return;
     }
 
     my $external_references_type_id = SGN::Model::Cvterm->get_cvterm_row($self->schema(), 'phenotype_external_references', 'phenotype_property')->cvterm_id();
@@ -508,6 +509,7 @@ sub store_additional_info {
 
     if (! $self->phenotype_id()) {
         print STDERR "Can't store additional info on this phenotype because there is no phenotype_id\n";
+        return;
     }
 
     my $phenotype_additional_info_type_id = SGN::Model::Cvterm->get_cvterm_row($self->schema(), 'phenotype_additional_info', 'phenotype_property')->cvterm_id();

lib/CXGN/Phenotypes/ParseUpload/Plugin/FieldBook.pm

@@ -20,6 +20,7 @@ use Moose;
 use File::Slurp;
 use Text::CSV;
 use Data::Dumper;
+use JSON qw(decode_json);
 
 sub name {
     return "field book";
@@ -261,7 +262,7 @@ sub parse {
 
         my $unit_value = $row[$header_column_info{$unit_col}];
         my $trait = $row[$header_column_info{'trait'}];
-        my $value = $row[$header_column_info{'value'}];
+        my $value = _normalize_fieldbook_value($row[$header_column_info{'value'}]);
         my $timestamp = defined $header_column_info{'timestamp'} ? $row[$header_column_info{'timestamp'}] : '';
         my $collector = defined $header_column_info{'person'} ? $row[$header_column_info{'person'}] : '';
 
@@ -314,4 +315,34 @@ sub parse {
     return \%parse_result;
 }
 
+sub _normalize_fieldbook_value {
+    my $value = shift;
+    return $value if !defined($value) || $value !~ /^\s*[\[\{]/;
+
+    my $decoded = eval { decode_json($value) };
+    return $value if $@;
+
+    if (ref($decoded) eq "ARRAY") {
+        my @values = map { _fieldbook_json_value($_) } @$decoded;
+        @values = grep { defined($_) && $_ ne "" } @values;
+        return @values ? join(":", @values) : "";
+    }
+
+    my $decoded_value = _fieldbook_json_value($decoded);
+    return defined($decoded_value) ? $decoded_value : $value;
+}
+
+sub _fieldbook_json_value {
+    my $entry = shift;
+    return if !defined($entry);
+
+    if (ref($entry) eq "HASH") {
+        return $entry->{value} if defined($entry->{value});
+        return $entry->{label} if defined($entry->{label});
+        return;
+    }
+
+    return ref($entry) ? undef : $entry;
+}
+
 1;

lib/CXGN/Phenotypes/StorePhenotypes.pm

@@ -620,7 +620,8 @@ sub check_measurement {
     if (ref($value_array) eq 'ARRAY') {
         # the entry represents trait + timestamp
         #
-        $trait_value = $value_array->[0];
+        $trait_value = _normalize_json_trait_value($value_array->[0]);
+        $value_array->[0] = $trait_value;
         $timestamp = $value_array->[1];
     }
     elsif (ref($value_array) eq "HASH") {
@@ -631,7 +632,7 @@ sub check_measurement {
     else {
         # it's a scalar. It really shouldn't be I guess?
         #
-        $trait_value = $value_array;
+        $trait_value = _normalize_json_trait_value($value_array);
     }
     #print STDERR "$plot_name, $trait_name, $trait_value\n";
     if ( $trait_name ne "notes" ) {
@@ -996,7 +997,8 @@ sub store {
 
                         my $phenotype_object = CXGN::Phenotype->new( { schema => $schema });
 
-                        my $trait_value = $value_array->[0];
+                        my $trait_value = _normalize_json_trait_value($value_array->[0]);
+                        $value_array->[0] = $trait_value;
 
                         $phenotype_object->value($trait_value);
                         my $timestamp = $value_array->[1];
@@ -1620,6 +1622,36 @@ sub get_trait_props {
     return \%property_by_cvterm_id;
 }
 
+sub _normalize_json_trait_value {
+    my $value = shift;
+    return $value if !defined($value) || $value !~ /^\s*[\[\{]/;
+
+    my $decoded = eval { decode_json($value) };
+    return $value if $@;
+
+    if (ref($decoded) eq "ARRAY") {
+        my @values = map { _json_trait_entry_value($_) } @$decoded;
+        @values = grep { defined($_) && $_ ne "" } @values;
+        return @values ? join(":", @values) : "";
+    }
+
+    my $decoded_value = _json_trait_entry_value($decoded);
+    return defined($decoded_value) ? $decoded_value : $value;
+}
+
+sub _json_trait_entry_value {
+    my $entry = shift;
+    return if !defined($entry);
+
+    if (ref($entry) eq "HASH") {
+        return $entry->{value} if defined($entry->{value});
+        return $entry->{label} if defined($entry->{label});
+        return;
+    }
+
+    return ref($entry) ? undef : $entry;
+}
+
 ###
 1;
 ###