Food groups facet and export (#6358)

* food groups / groupes alimentaires * add index on food_groups_tags * add food_groups_tags * added strings for food groups * fix export_database.pl and add food_groups * export food_groups * stats for food groups
openfoodfacts · Jan 28, 2022 · 204ac1c · 204ac1c
1 parent 2e8bbd8
commit 204ac1c
Show file tree

Hide file tree

Showing 3 changed files with 33 additions and 5 deletions.
diff --git a/lib/ProductOpener/Config_off.pm b/lib/ProductOpener/Config_off.pm
@@ -650,6 +650,7 @@ periods_after_opening nutrients food_groups
 	nova_group
 	pnns_groups_1
 	pnns_groups_2
+	food_groups
 	states
 	brand_owner
 	ecoscore_score_fr

diff --git a/lib/ProductOpener/DataQualityFood.pm b/lib/ProductOpener/DataQualityFood.pm
@@ -56,6 +56,8 @@ use ProductOpener::Tags qw(:all);
 use ProductOpener::Food qw(:all);
 use ProductOpener::Ecoscore qw(:all);
 
+use Data::DeepAccess qw(deep_exists);
+
 use Log::Any qw($log);
 
 =head1 HARDCODED BRAND LISTS
@@ -1158,6 +1160,30 @@ sub check_ecoscore_data($) {
 }
 
 
+=head2 check_food_groups( PRODUCT_REF )
+
+Add info tags about food groups.
+
+=cut
+
+sub check_food_groups($) {
+
+	my $product_ref = shift;
+
+	for (my $level = 1; $level <= 3; $level++) {
+
+		if (deep_exists($product_ref, "food_groups_tags", $level - 1)) {
+			push @{$product_ref->{data_quality_info_tags}}, 'en:food-groups-' . $level . '-known';
+		}
+		else {
+			push @{$product_ref->{data_quality_info_tags}}, 'en:food-groups-' . $level . '-unknown';
+		}
+	}
+
+	return;
+}
+
+
 =head2 check_quality_food( PRODUCT_REF )
 
 Run all quality checks defined in the module.
@@ -1180,6 +1206,7 @@ sub check_quality_food($) {
 	check_categories($product_ref);
 	compare_nutriscore_with_value_from_producer($product_ref);
 	check_ecoscore_data($product_ref);
+	check_food_groups($product_ref);
 
 	return;
 }

diff --git a/scripts/export_database.pl b/scripts/export_database.pl
@@ -46,13 +46,13 @@
 use ProductOpener::Food qw/:all/;
 use ProductOpener::Ingredients qw/:all/;
 use ProductOpener::Data qw/:all/;
+use ProductOpener::Text qw/:all/;
 
 # for RDF export: replace xml_escape() with xml_escape_NFC()
 use Unicode::Normalize;
 use URI::Escape::XS;
 
 use CGI qw/:cgi :form escapeHTML/;
-use URI::Escape::XS;
 use Storable qw/dclone/;
 use Encode;
 use JSON::PP;
@@ -94,7 +94,7 @@ sub sanitize_field_content {
 }
 
 
-my %tags_fields = (packaging => 1, brands => 1, categories => 1, labels => 1, origins => 1, manufacturing_places => 1, emb_codes=>1, cities=>1, allergen=>1, traces => 1, additives => 1, ingredients_from_palm_oil => 1, ingredients_that_may_be_from_palm_oil => 1, countries => 1, states=>1);
+my %tags_fields = (packaging => 1, brands => 1, categories => 1, labels => 1, origins => 1, manufacturing_places => 1, emb_codes=>1, cities=>1, allergen=>1, traces => 1, additives => 1, ingredients_from_palm_oil => 1, ingredients_that_may_be_from_palm_oil => 1, countries => 1, states=>1, food_groups=>1);
 
 
 my %langs = ();
@@ -146,7 +146,7 @@ sub sanitize_field_content {
 	print STDERR "Write file: $csv_filename.temp\n";
 	print STDERR "Write file: $rdf_filename.temp\n";
 
-	open (my $OUT, ">:encoding(UTF-8)", "$csv_filename.temp");
+	open (my $OUT, ">:encoding(UTF-8)", "$csv_filename.temp") or die("Cannot write $csv_filename.temp: $!\n");
 	open (my $RDF, ">:encoding(UTF-8)", "$rdf_filename.temp");
 	open (my $BAD, ">:encoding(UTF-8)", "$log_filename");
 
@@ -290,9 +290,9 @@ sub sanitize_field_content {
 				$field_value = $product_ref->{$field . "_" . $l};
 			}
 
-			# Eco-Score
+			# Eco-Score data is stored in ecoscore_data.(grades|scores).(language code)
 			if (($field =~ /^ecoscore_(score|grade)_(\w\w)/) and (defined $product_ref->{ecoscore_data})) {
-				$field_value = $product_ref->{ecoscore_data}{$1 . "_" . $2};
+				$field_value = ($product_ref->{ecoscore_data}{$1 . "s"}{$2} // "");
 			}
 
 			if ($field_value ne '') {