Handle more complex data

Author: nigelhorne

bin/extract

@@ -118,6 +118,13 @@ sub extract_family_info
 		my $children_text = $1;
 		$children_text =~ s/, grandmother.+//;
 		$family{children} = extract_people_section($children_text);
+	} elsif($text =~ /sons,?\s+([a-z]+)\s+and\s+([a-z]+)/i) {
+		my @children;
+		push @children, { name => $1, sex => 'M' }, { name => $2, sex => 'M' };
+		if($text =~ /daughter,?\s([a-z]+)/i) {
+			push @children, { 'name' => $1, 'sex' => 'F' }
+		}
+		$family{children} = \@children if @children;
 	} else {
 		my @children;
 		while($text =~ /\b(son|daughter)s?,\s*([A-Z][a-z]+(?:\s+\([A-Z][a-z]+\))?)\s*(?:and their children ([^.;]+))?/g) {
@@ -127,8 +134,11 @@ sub extract_family_info
 			if(my @grandchildren = $grandkids ? split /\s*,\s*|\s+and\s+/, $grandkids : ()) {
 				push @children, {
 					name => $child,
+					sex => $sex,
 					grandchildren => \@grandchildren,
 				};
+			} elsif(($sex eq 'F') && ($child =~ /(.+)\s+\((.+)\)/)) {
+				push @children, { name => $1, sex => 'F', spouse => { name => $2, sex => 'M' } }
 			} else {
 				push @children, { name => $child, sex => $sex }
 			}

gedcom

@@ -13685,6 +13685,13 @@ sub extract_family_info
 		my $children_text = $1;
 		$children_text =~ s/, grandmother.+//;
 		$family{children} = extract_people_section($children_text);
+	} elsif($text =~ /sons,?\s+([a-z]+)\s+and\s+([a-z]+)/i) {
+		my @children;
+		push @children, { name => $1, sex => 'M' }, { name => $2, sex => 'M' };
+		if($text =~ /daughter,?\s([a-z]+)/i) {
+			push @children, { 'name' => $1, 'sex' => 'F' }
+		}
+		$family{children} = \@children if @children;
 	} else {
 		my @children;
 		while($text =~ /\b(son|daughter)s?,\s*([A-Z][a-z]+(?:\s+\([A-Z][a-z]+\))?)\s*(?:and their children ([^.;]+))?/g) {
@@ -13694,8 +13701,11 @@ sub extract_family_info
 			if(my @grandchildren = $grandkids ? split /\s*,\s*|\s+and\s+/, $grandkids : ()) {
 				push @children, {
 					name => $child,
+					sex => $sex,
 					grandchildren => \@grandchildren,
 				};
+			} elsif(($sex eq 'F') && ($child =~ /(.+)\s+\((.+)\)/)) {
+				push @children, { name => $1, sex => 'F', spouse => { name => $2, sex => 'M' } }
 			} else {
 				push @children, { name => $child, sex => $sex }
 			}

tests/extract.t

@@ -141,20 +141,20 @@ cmp_deeply($foo,
 		}, 'sisters' => [
 			  { 'name' => 'Phyllis Huestis' }
 		], 'children' => [
-			   { 'name' => 'Boyd R', 'sex' => 'M' },
-			   { 'name' => 'Anne (Malcolm) Maxwell', 'sex' => 'F' }
-		 ], 'brothers' => [
+			   { 'name' => 'Anne', 'sex' => 'F', spouse => { 'name' => 'Malcolm', 'sex' => 'M' } },	# spouse should be 'Malcolm Maxwell'
+			   { 'name' => 'Boyd', 'sex' => 'M' },	# should be Boyd R
+		], 'brothers' => [
 			   {
-			     'status' => 'living',
-			     'name' => 'Ralph N.'
+				'status' => 'living',
+				'name' => 'Ralph N.'
 			   }
-		 ], 'death' => {
+		], 'death' => {
 			'place' => 'home',
 			'date' => 'April 16, 2007'
-	      }, 'birth' => {
+		}, 'birth' => {
 			'place' => 'Fredericton',
 			'date' => 'Jul 19, 1925'
-	      }, 'spouse' => [
+		}, 'spouse' => [
 			 {
 			   'name' => 'Ruth Howland',
 			   'married' => {
@@ -189,9 +189,9 @@ diag(Data::Dumper->new([$foo])->Dump()) if($ENV{'TEST_VERBOSE'});
 cmp_deeply($foo,
 	{
 	   'children' => [
-			   { 'name' => 'Christopher Cloud', 'sex' => 'M' },
-			   { 'name' => 'Thomas Cloud', 'sex' => 'M' },
-			   { 'name' => 'Marsha Cloud', 'sex' => 'F' }
+			   { 'name' => 'Christopher', 'sex' => 'M' },
+			   { 'name' => 'Thomas', 'sex' => 'M' },
+			   { 'name' => 'Marsha', 'sex' => 'F' }
 			 ],
 	   'birth' => {
 			'date' => '1937/05/21',