taxonomy: taking into account the ingredient origin when retrieving t…

…he corresponding Ecobalyse code (#11108)

### What

Goal : being able to take into account the origin when retrieving the
corresponding ecobalyse code for a given ingredient

---------

Co-authored-by: Stéphane Gigandet <stephane@openfoodfacts.org>
Co-authored-by: Open Food Facts Bot <contact@openfoodfacts.org>
Co-authored-by: Alex Garel <alex@openfoodfacts.org>

lib/ProductOpener/Ingredients.pm

@@ -3458,7 +3458,7 @@ sub get_missing_ciqual_codes ($ingredients_ref) {
 
 =head2 get_missing_ecobalyse_ids ($ingredients_ref)
 
-Assign a ecobalyse_id or a ciqual_proxy_food_code to ingredients and sub ingredients. (NOTE : this is a first version that'll soon be improved)
+Assign a ecobalyse_code or a ecobalyse_proxy_code to ingredients and sub ingredients. (NOTE : this is a first version that'll soon be improved)
 
 =head3 Arguments
 
@@ -3495,10 +3495,36 @@ sub get_missing_ecobalyse_ids ($ingredients_ref) {
 
 		# List of suffixes
 		my @suffixes = ();
+		# If the ingredient is both organic and French...
+		if (    (defined $ingredient_ref->{labels})
+			and ($ingredient_ref->{labels} =~ /\ben:organic\b/)
+			and (defined $ingredient_ref->{origins})
+			and (get_geographical_area($ingredient_ref->{origins}) eq "fr"))
+		{
+			push @suffixes, "_labels_en_organic_origins_en_france";
+			push @suffixes, "_labels_en_organic_origins_en_european_union";
+		}
+		# If the ingredient is both organic and European...
+		if (    (defined $ingredient_ref->{labels})
+			and ($ingredient_ref->{labels} =~ /\ben:organic\b/)
+			and (defined $ingredient_ref->{origins})
+			and (get_geographical_area($ingredient_ref->{origins}) eq "eu"))
+		{
+			push @suffixes, "_labels_en_organic_origins_en_european_union";
+		}
 		# If the ingredient is organic...
 		if ((defined $ingredient_ref->{labels}) and ($ingredient_ref->{labels} =~ /\ben:organic\b/)) {
 			push @suffixes, "_labels_en_organic";
 		}
+		# If the ingredient is French...
+		if ((defined $ingredient_ref->{origins}) and (get_geographical_area($ingredient_ref->{origins}) eq "fr")) {
+			push @suffixes, "_origins_en_france";
+			push @suffixes, "_origins_en_european_union";
+		}
+		# If the ingredient is European...
+		if ((defined $ingredient_ref->{origins}) and (get_geographical_area($ingredient_ref->{origins}) eq "eu")) {
+			push @suffixes, "_origins_en_european_union";
+		}
 		push @suffixes, '';
 
 		# First try an exact match, and then a proxy match
@@ -3535,6 +3561,35 @@ sub get_missing_ecobalyse_ids ($ingredients_ref) {
 	return @ingredients_without_ecobalyse_ids;
 }
 
+=head2 get_geographical_area ($originid)
+
+Retrieve the geographical area for ecobalyse. (NOTE : this is a first version that'll soon be improved)
+
+=head3 Arguments
+
+=head4 $originid
+
+reference to the name of the country 
+
+=head3 Return values
+
+=head4 $ecobalyse_area
+
+=cut
+
+sub get_geographical_area ($originid) {
+	# Getting information about the country
+	my $ecobalyse_area = "";
+	if (get_inherited_property("countries", $originid, "ecobalyse_is_part_of_eu") eq "yes") {
+		$ecobalyse_area = "eu";
+	}
+	if ($originid eq "en:france") {
+		$ecobalyse_area = "fr";
+	}
+
+	return $ecobalyse_area;
+}
+
 =head2 estimate_ingredients_percent_service ( $product_ref, $updated_product_fields_ref, $errors_ref )
 
 Compute minimum and maximum percent ranges and percent estimates for each ingredient and sub ingredient.

stop_words.txt

@@ -71,6 +71,7 @@ dont
 dropdown
 du
 ecoscore
+ecobalyse
 Ecobalyse
 EAN
 EANs

tests/unit/expected_test_results/ecoscore/carrots-plastic.json

@@ -403,7 +403,7 @@
    "ingredients" : [
       {
          "ciqual_food_code" : "20009",
-         "ecobalyse_code" : "carrot-non-eu",
+         "ecobalyse_code" : "carrot-fr",
          "id" : "en:carrot",
          "is_in_taxonomy" : 1,
          "origins" : "en:france",

tests/unit/expected_test_results/ecoscore/carrots.json

@@ -407,7 +407,7 @@
    "ingredients" : [
       {
          "ciqual_food_code" : "20009",
-         "ecobalyse_code" : "carrot-non-eu",
+         "ecobalyse_code" : "carrot-fr",
          "id" : "en:carrot",
          "is_in_taxonomy" : 1,
          "origins" : "en:france",

tests/unit/expected_test_results/ecoscore/fr-viande-porcine-francaise.json

@@ -422,7 +422,7 @@
       },
       {
          "ciqual_proxy_food_code" : "28205",
-         "ecobalyse_code" : "cooked-ham",
+         "ecobalyse_code" : "cooked-ham-fr",
          "id" : "en:cooked-ham",
          "is_in_taxonomy" : 1,
          "origins" : "en:france",

tests/unit/expected_test_results/ecoscore/origins-of-ingredients-partly-specified.json

@@ -402,7 +402,7 @@
    "ingredients" : [
       {
          "ciqual_food_code" : "13000",
-         "ecobalyse_code" : "apricot-eu",
+         "ecobalyse_code" : "apricot-fr",
          "id" : "en:apricot",
          "is_in_taxonomy" : 1,
          "origins" : "en:france",

tests/unit/expected_test_results/ecoscore/origins-of-ingredients-specified.json

@@ -406,7 +406,7 @@
    "ingredients" : [
       {
          "ciqual_food_code" : "13000",
-         "ecobalyse_code" : "apricot-eu",
+         "ecobalyse_code" : "apricot-fr",
          "id" : "en:apricot",
          "is_in_taxonomy" : 1,
          "origins" : "en:france",

tests/unit/expected_test_results/ingredients/fr-origins-labels.json

@@ -2,7 +2,7 @@
    "ingredients" : [
       {
          "ciqual_food_code" : "13014",
-         "ecobalyse_code" : "strawberry-non-eu",
+         "ecobalyse_code" : "strawberry-fr-offseason",
          "id" : "en:strawberry",
          "is_in_taxonomy" : 1,
          "origins" : "en:france",

tests/unit/expected_test_results/ingredients/fr-viande-porcine-francaise.json

@@ -15,7 +15,7 @@
       },
       {
          "ciqual_proxy_food_code" : "28205",
-         "ecobalyse_code" : "cooked-ham",
+         "ecobalyse_code" : "cooked-ham-fr",
          "id" : "en:cooked-ham",
          "is_in_taxonomy" : 1,
          "origins" : "en:france",