feat: update product_type (#11173)

### What Automatically change the product type of non-food categorized products Copy pasted all non-*something* on top of each categories taxonomies. Also copy pasted categories from food to other flavors (for example, en: Dry Dog Food is moved from taxonomies/food/categories.txt to taxonomies/petfood/categories.txt ### Remark The following, I could not add, but would be a nice to have: - rm tag from previous flavor/project after migration. For example, after changing product_type from food to beauty we would like to remove 1) 'en:non-food-products' and 'en:open-beauty-facts' but also categories from food 2) "en:desserts", "en:pies", "en:sweet-pies", "en:apple-pies" for example I tried to do something with **remove_tag** function from **Tags.pm**. It removes element of the **categories_tags** array, but **categories** (string) and **categories_hierarchy** (array) remain. And tags are still visible on the website. As I did not want to do some update of the **categories** string, I decided not to include that in the present PR. ### Related issue(s) and discussion - Fixes #11094 --------- Co-authored-by: Stéphane Gigandet <[email protected]> Co-authored-by: Pierre Slamich <[email protected]>
openfoodfacts · Feb 14, 2025 · d1648b3 · d1648b3
1 parent 03b105a
commit d1648b3
Show file tree

Hide file tree

Showing 8 changed files with 362 additions and 262 deletions.
diff --git a/lib/ProductOpener/Config_off.pm b/lib/ProductOpener/Config_off.pm
@@ -613,7 +613,6 @@ $options{categories_exempted_from_nutriscore} = [
 		en:spices
 		en:sugar-substitutes
 		en:vinegars
-		en:pet-food
 		en:non-food-products
 	)
 ];

diff --git a/lib/ProductOpener/Products.pm b/lib/ProductOpener/Products.pm
@@ -95,6 +95,7 @@ BEGIN {
 		&compute_completeness_and_missing_tags
 		&compute_product_history_and_completeness
 		&compute_languages
+		&review_product_type
 		&compute_changes_diff_text
 		&compute_data_sources
 		&compute_sort_keys
@@ -3000,6 +3001,59 @@ sub compute_languages ($product_ref) {
 	return;
 }
 
+=head2 review_product_type ( $product_ref )
+
+Reviews the product type based on the presence of specific tags in the categories field.
+Updates the product type if necessary.
+
+=head3 Arguments
+
+=head4 Product reference $product_ref
+
+A reference to a hash containing the product details.
+
+=cut
+
+sub review_product_type ($product_ref) {
+
+	my $error;
+
+	my $expected_type;
+	if (has_tag($product_ref, "categories", "en:open-beauty-facts")) {
+		$expected_type = "beauty";
+	}
+	elsif (has_tag($product_ref, "categories", "en:open-food-facts")) {
+		$expected_type = "food";
+	}
+	elsif (has_tag($product_ref, "categories", "en:open-pet-food-facts")) {
+		$expected_type = "petfood";
+	}
+	elsif (has_tag($product_ref, "categories", "en:open-products-facts")) {
+		$expected_type = "product";
+	}
+
+	if ($expected_type and ($product_ref->{product_type} ne $expected_type)) {
+		$error = change_product_type($product_ref, $expected_type);
+	}
+
+	if ($error) {
+		$log->error("review_product_type - error", {error => $error, product_ref => $product_ref});
+	}
+	else {
+		# We remove the tag en:incorrect-product-type and its children before the product is stored on the server of the new type
+		remove_tag($product_ref, "categories", "en:incorrect-product-type");
+		remove_tag($product_ref, "categories", "en:open-beauty-facts");
+		remove_tag($product_ref, "categories", "en:open-food-facts");
+		remove_tag($product_ref, "categories", "en:open-pet-food-facts");
+		remove_tag($product_ref, "categories", "en:open-products-facts");
+		remove_tag($product_ref, "categories", "en:non-food-products");
+		remove_tag($product_ref, "categories", "en:non-pet-food-products");
+		remove_tag($product_ref, "categories", "en:non-beauty-products");
+	}
+
+	return;
+}
+
 =head2 process_product_edit_rules ($product_ref)
 
 Process the edit_rules (see C<@edit_rules> in in Config file).
@@ -3723,6 +3777,11 @@ sub analyze_and_enrich_product_data ($product_ref, $response_ref) {
 
 	compute_languages($product_ref);    # need languages for allergens detection and cleaning ingredients
 
+	# change the product type of non-food categorized products (issue #11094)
+	if (has_tag($product_ref, "categories", "en:incorrect-product-type")) {
+		review_product_type($product_ref);
+	}
+
 	# Run special analysis, score calculations that it specific to the product type
 
 	if (($options{product_type} eq "food")) {

diff --git a/scripts/update_all_products.pl b/scripts/update_all_products.pl
@@ -1381,6 +1381,12 @@
 			assign_ciqual_codes($product_ref);
 		}
 
+		# if we have an old_product_type (if change_product_type() was called),
+		# we need to use store_product() so that the product is removed from the old MongoDB collection and added to the new one
+		if (defined $product_ref->{old_product_type}) {
+			$product_values_changed = 1;
+		}
+
 		my $any_change = $product_values_changed;
 		if (not $pretend) {
 			if (!$any_change) {

diff --git a/taxonomies/beauty/categories.txt b/taxonomies/beauty/categories.txt
@@ -2,8 +2,48 @@ stopwords:fr: aux, au, de, le, du, la, a, et, avec, pour
 
 synonyms:en: flavoured, flavored
 
+en: Incorrect product type
+bg: Неправилен product type
+ca: Tipus de producte incorrecte
+de: Falscher product type
+es: Tipo de producto incorrecto
+fi: Väärä product type
+fr: Mauvais product type
+hr: Pogrešan product type
+it: Tipo di prodotto errato
+nl: Verkeerd product type
+pt: Tipo de produto incorreto
+ru: Неправильный product type
+
+< en: Incorrect product type
+en: Non beauty products
+bg: Некозметични продукти
+ca: Productes no de bellesa
+de: Non-Beauty-Produkte, Non Beauty Produkte
+es: Productos no de belleza
+fi: ei kauneustuote
+fr: Non beauté
+hr: Neprehrambeni proizvodi za ljepotu
+it: Prodotti non di bellezza
+nl: Non-beauty
+pt: Produtos não de beleza
+ru: Непродукты для красоты, Не косметические продукты
+
+< en:Non beauty products
+en: Open Food Facts, OFF
+xx: Open Food Facts, OFF
+
+< en:Non beauty products
+en: Open Pet Food Facts, OPFF
+xx: Open Pet Food Facts, OPFF
+
+< en:Non beauty products
+en: Open Products Facts, OPF
+xx: Open Products Facts, OPF
+
 # Can you add translations in all European, Asian languages (and more), following alphabetical order for language ordering, except for english which comes first ?
 
+
 ###### CATEGORIE HYGIENE
 
 en: Hygiene
@@ -1295,7 +1335,6 @@ uz: Makiyaj
 vi: Trang điểm
 zh: 化妆品
 
-
 #### Maquillage pour les yeux
 
 < en:Makeup
@@ -1484,8 +1523,8 @@ en: Men perfumes, Men perfume, Men fragrances, Men fragrance, Male perfume, Male
 fr: Parfums homme, Parfums hommes, Parfum homme, Parfum hommes
 
 < en:Perfumes
-en: Women perfumes, Women fragrances, Female perfumes
-fr: Parfums femme
+en: Women perfumes, Women perfume, Women fragrances, Women fragrance, Female perfume, Female perfumes
+fr: Parfums femme, Parfum femmes, Parfum femme, Parfum femmes
 
 < en:Perfumes
 en: Kids perfumes, Kids fragrances