feat: use ingredients processing for NOVA classification

openfoodfacts · Feb 7, 2025 · ae64f24 · ae64f24
1 parent d1346e7
commit ae64f24
Show file tree

Hide file tree

Showing 5 changed files with 60 additions and 2 deletions.
diff --git a/lib/ProductOpener/Food.pm b/lib/ProductOpener/Food.pm
@@ -2692,6 +2692,26 @@ sub compute_nova_group ($product_ref) {
 		}
 	}
 
+	# Go through the nested ingredients structure to check if some ingredients
+	# have a processing that has a nova:en: property
+	if (defined $product_ref->{ingredients}) {
+		# Create a copy of the ingredients structure to avoid modifying the original one
+		my $ingredients_ref = dclone($product_ref->{ingredients});
+		while (my $ingredient_ref = shift @{$ingredients_ref}) {
+			if (defined $ingredient_ref->{processing}) {
+				foreach my $processing (split(/,/, $ingredient_ref->{processing})) {
+					my $nova_group = get_property("ingredients_processing", $processing, "nova:en");
+					if (defined $nova_group) {
+						push @{$matching_tags_for_groups{$nova_group + 0}}, ["ingredients", $ingredient_ref->{id}];
+					}
+				}
+			}
+			if (defined $ingredient_ref->{ingredients}) {
+				push @{$ingredients_ref}, @{$ingredient_ref->{ingredients}};
+			}
+		}
+	}
+
 	# Assign the NOVA group based on matching tags (options in Config.pm and then taxonomies)
 	# First identify group 2 foods, then group 3 and 4
 	# Group 2 foods should not be moved to group 3

diff --git a/taxonomies/food/ingredients.txt b/taxonomies/food/ingredients.txt
@@ -28561,6 +28561,11 @@ it: farina di soia sgrassata, farina di soia senza grassi
 pl: odtłuszczona mąka sojowa
 ro: făină de soia degrasată, faina de soia degrasata
 
+< en: flour
+en: modified flour
+fr: farine modifiée
+nova:en: 4
+
 < en:corn
 < en:flour
 en: corn flour, maize flour, cornflour
@@ -28610,11 +28615,12 @@ hu: bio kukoricaliszt
 it: Farina di mais biologica
 pt: farinha de milho orgânico
 
+< en: modified flour
 < en:corn
-en: modified cornflour
+en: modified corn flour
 bg: модифицирано царевично брашно
 de: Modifiziertes Maismehl
-fr: Amidon de maïs modifié
+fr: Farine de maïs modifié
 hr: modificirano kukuruzno brašno
 it: Farina di mais modificata
 nl: gemodificeerd maïsmeel

diff --git a/taxonomies/ingredients_processing.txt b/taxonomies/ingredients_processing.txt
@@ -2660,6 +2660,7 @@ hr: hidrolizat, hidrolizirane, hidrolizirano, hidrolizirani
 nl: gehydrolyseerd, gehydrolyseerde
 pl: hydrolizat, hydrolizowane
 sv: hydrolyserat
+nova:en: 4
 
 #en:description:Transformation of unsaturated glycerides into saturated glycerides (of oils and fats)
 en: hydrogenated, hardened, partially hardened
@@ -2675,6 +2676,7 @@ it: idrogenato, idrogenati, idrogenata, idrogenate
 nl: Gehydrogeneerde
 pl: uwodorniony, uwodorniona, uwodornione, utwardzony, utwardzone, utwardzona
 sv: hårdat
+nova:en: 4
 
 #<en:hydrogenated
 en: fully hydrogenated
@@ -2688,6 +2690,7 @@ hr: potpuno hidrogenirana, potpuno hidrogenirane, potpuno hidrogenirano, potpuno
 nb: fullherdet
 pl: całkowicie uwodorniony, całkowicie uwodorniona, całkowicie uwodornione, całkowicie utwardzony, całkowicie utwardzona, całkowicie utwardzone
 sv: fullhärdad
+nova:en: 4
 
 #<en:hydrogenated
 en: partially hydrogenated
@@ -2706,10 +2709,12 @@ pl: częściowo uwodorniony, częściowo uwodorniona, częściowo uwodornione, c
 pt: parcialmente hidrogenada
 ro: parțial hidrogenate
 sv: delvis härdad
+nova:en: 4
 
 en: textured
 de: texturiert, texturierte, texturierter, texturiertes, texturierten, texturiertem
 fr: texturé, texturée
+nova:en: 4
 
 en: processed with alkali
 bg: обработено с щелоч
@@ -2727,6 +2732,26 @@ pl: przetworzone z alkalicznym
 pt: processado com álcali
 ro: prelucrat cu alcali
 sv: behandlad med alkali
+nova:en: 4
+
+en: modified
+bg: модифициран, модифицирана, модифицирани, модифицирани
+ca: modificat, modificada, modificats, modificades
+da: modificeret, modificerede, modificerede, modificerede
+de: modifiziert, modifizierte, modifizierten, modifizierten
+es: modificado, modificada, modificados, modificadas
+fi: muokattu, muokattu, muokatut, muokatut
+fr: modifié, modifiée, modifiés, modifiées
+hr: modificiran, modificirana, modificirani, modificirane
+hu: módosított, módosított, módosítottak, módosítottak
+it: modificato, modificata, modificati, modificate
+nl: gewijzigd, gewijzigde, gewijzigden, gewijzigde
+pl: zmodyfikowany, zmodyfikowana, zmodyfikowani, zmodyfikowane
+pt: modificado, modificada, modificados, modificadas
+ro: modificat, modificată, modificați, modificate
+sv: modifierad, modifierad, modifierade, modifierade
+nova:en: 4
+
 
 en: brewed
 de: gebrühter

diff --git a/tests/unit/nova.t b/tests/unit/nova.t
@@ -52,6 +52,12 @@ my @tests = (
 	# allergens in parenthesis should not change the nova score but should appear in allergens
 	[{lc => "hr", ingredients_text_hr => "Krupica od durum pšenice (gluten), voda."}, 1],
 
+	# added nova:en: 4 property to some processings
+	[{lc => "fr", ingredients_text_fr => "Farine de blé modifiée"}, 4],
+	[{lc => "en", ingredients_text_en => "Modified cornflour"}, 4],
+	[{lc => "en", ingredients_text_en => "Modified whey"}, 4],
+	[{lc => "en", ingredients_text_en => "Modified strawberry"}, 4],
+
 );
 
 foreach my $test_ref (@tests) {

diff --git a/tests/unit/packaging_food_contact.t b/tests/unit/packaging_food_contact.t
@@ -6,6 +6,7 @@ use utf8;
 use Test2::V0;
 use Data::Dumper;
 $Data::Dumper::Terse = 1;
+$Data::Dumper::Sortkeys = 1;
 use Log::Any::Adapter 'TAP';
 
 use JSON;