feat: Use 'GS1 Barcode Syntax Engine' in READ API v3 (#9050)

cgi/search.pl

@@ -138,31 +138,31 @@
 }
 
 # check if the search term looks like a barcode
-
 if (    (not defined single_param('json'))
 	and (not defined single_param('jsonp'))
 	and (not defined single_param('jqm'))
 	and (not defined single_param('jqm_loadmore'))
 	and (not defined single_param('xml'))
 	and (not defined single_param('rss'))
-	and ($search_terms =~ /^(\d{4,24})$/))
+	and ($search_terms =~ /^(\d{4,24}|(?:[\^(\N{U+001D}\N{U+241D}]|https?:\/\/).+)$/))
 {
 
 	my $code = normalize_code($search_terms);
+	if ((defined $code) and (length($code) > 0)) {
+		my $product_id = product_id_for_owner($Owner_id, $code);
 
-	my $product_id = product_id_for_owner($Owner_id, $code);
-
-	my $product_ref = product_exists($product_id);    # returns 0 if not
+		my $product_ref = product_exists($product_id);    # returns 0 if not
 
-	if ($product_ref) {
-		$log->info("product code exists, redirecting to product page", {code => $code});
-		my $location = product_url($product_ref);
+		if ($product_ref) {
+			$log->info("product code exists, redirecting to product page", {code => $code});
+			my $location = product_url($product_ref);
 
-		my $r = shift;
-		$r->headers_out->set(Location => $location);
-		$r->status(301);
-		return 301;
+			my $r = shift;
+			$r->headers_out->set(Location => $location);
+			$r->status(301);
+			return 301;
 
+		}
 	}
 }
 

lib/ProductOpener/API.pm

@@ -449,13 +449,13 @@ Reference to the response object.
 
 =head3 Return value
 
-Normalized code.
+Normalized code and, if available, GS1 AI data string.
 
 =cut
 
 sub normalize_requested_code ($requested_code, $response_ref) {
 
-	my $code = normalize_code($requested_code);
+	my ($code, $ai_data_str) = &normalize_code_with_gs1_ai($requested_code);
 	$response_ref->{code} = $code;
 
 	# Add a warning if the normalized code is different from the requested code
@@ -470,7 +470,7 @@ sub normalize_requested_code ($requested_code, $response_ref) {
 		);
 	}
 
-	return $code;
+	return ($code, $ai_data_str);
 }
 
 =head2 get_images_to_update($product_ref, $target_lc)

lib/ProductOpener/APIProductRead.pm

@@ -83,7 +83,7 @@ sub read_product_api ($request_ref) {
 			|| "";
 	}
 
-	my $code = normalize_requested_code($request_ref->{code}, $response_ref);
+	my ($code, $ai_data_string) = &normalize_requested_code($request_ref->{code}, $response_ref);
 
 	my $product_ref;
 	my $product_id;

lib/ProductOpener/APIProductWrite.pm

@@ -394,7 +394,7 @@ sub write_product_api ($request_ref) {
 
 		if ($code ne 'test') {
 			# Load the product
-			$code = normalize_requested_code($request_ref->{code}, $response_ref);
+			($code, my $ai_data_string) = &normalize_requested_code($request_ref->{code}, $response_ref);
 
 			# Check if the code is valid
 			if ($code !~ /^\d{4,24}$/) {

lib/ProductOpener/Products.pm

@@ -66,6 +66,7 @@ BEGIN {
 	use vars qw(@ISA @EXPORT_OK %EXPORT_TAGS);
 	@EXPORT_OK = qw(
 		&normalize_code
+		&normalize_code_with_gs1_ai
 		&assign_new_code
 		&split_code
 		&product_id_for_owner
@@ -280,9 +281,36 @@ Normalized version of the code
 sub normalize_code ($code) {
 
 	if (defined $code) {
-		my $gs1_code = _try_normalize_code_gs1($code);
-		if ($gs1_code) {
+		($code, my $gs1_ai_data_str) = &normalize_code_with_gs1_ai($code);
+	}
+	return $code;
+}
+
+=head2 normalize_code_with_gs1_ai()
+
+C<normalize_code_with_gs1_ai()> this function normalizes the product code by:
+- running the given code through normalization method provided by GS1 to format a GS1 data string, or data URI to a GTIN,
+- keeping only digits and removing spaces/dashes etc.,
+- normalizing the length by adding leading zeroes or removing the leading zero (in case of 14 digit codes)
+
+=head3 Arguments
+
+Product Code in the Raw form: $code
+
+=head3 Return Values
+
+Normalized version of the code, and GS1 AI data string of the code, if a valid GS1 string was given as the argument
+
+=cut
+
+sub normalize_code_with_gs1_ai ($code) {
+
+	my $ai_data_str;
+	if (defined $code) {
+		my ($gs1_code, $gs1_ai_data_str) = &_try_normalize_code_gs1($code);
+		if ($gs1_code and $gs1_ai_data_str) {
 			$code = $gs1_code;
+			$ai_data_str = $gs1_ai_data_str;
 		}
 
 		# Keep only digits, remove spaces, dashes and everything else
@@ -308,7 +336,7 @@ sub normalize_code ($code) {
 			$code = $';
 		}
 	}
-	return $code;
+	return ($code, $ai_data_str);
 }
 
 sub _try_normalize_code_gs1 ($code) {
@@ -346,11 +374,12 @@ sub _try_normalize_code_gs1 ($code) {
 	};
 	if ($@) {
 		$log->warn("GS1Parser error", {error => $@}) if $log->is_warn();
+		$code = undef;
 		$ai_data_str = undef;
 	}
 
-	if ((defined $ai_data_str) and ($ai_data_str =~ /^\(01\)(\d{1,14})/)) {
-		return $1;
+	if ((defined $code) and (defined $ai_data_str) and ($ai_data_str =~ /^\(01\)(\d{1,14})/)) {
+		return ($1, $ai_data_str);
 	}
 	else {
 		return;

lib/ProductOpener/Routing.pm

@@ -148,33 +148,33 @@ sub analyze_request ($request_ref) {
 		{query_string => $request_ref->{query_string}})
 		if $log->is_debug();
 
-	# Decode the escaped characters in the query string
-	$request_ref->{query_string} = decode("utf8", URI::Escape::XS::decodeURIComponent($request_ref->{query_string}));
-
-	$log->debug("analyzing query_string, step 3 - components UTF8 decoded",
-		{query_string => $request_ref->{query_string}})
-		if $log->is_debug();
-
-	$request_ref->{page} = 1;
-
 	# some sites like FB can add query parameters, remove all of them
 	# make sure that all query parameters of interest have already been consumed above
 
 	$request_ref->{query_string} =~ s/(\&|\?).*//;
 
-	$log->debug("analyzing query_string, step 4 - removed all query parameters",
+	$log->debug("analyzing query_string, step 3 - removed all query parameters",
 		{query_string => $request_ref->{query_string}})
 		if $log->is_debug();
 
+	# Split query string by "/" to know where it points
+	my @components = ();
+	foreach my $component (split(/\//, $request_ref->{query_string})) {
+		# Decode the escaped characters in the query string
+		push(@components, decode("utf8", URI::Escape::XS::decodeURIComponent($component)));
+	}
+
+	$log->debug("analyzing query_string, step 4 - components split and UTF8 decoded", {components => \@components})
+		if $log->is_debug();
+
+	$request_ref->{page} = 1;
+
 	# if the query request json or xml, either through the json=1 parameter or a .json extension
 	# set the $request_ref->{api} field
 	if ((defined single_param('json')) or (defined single_param('jsonp')) or (defined single_param('xml'))) {
 		$request_ref->{api} = 'v0';
 	}
 
-	# Split query string by "/" to know where it points
-	my @components = split(/\//, $request_ref->{query_string});
-
 	# Root, ex: https://world.openfoodfacts.org/
 	if ($#components < 0) {
 		$request_ref->{text} = 'index';

tests/integration/api_v3_product_read.t

@@ -28,7 +28,7 @@ my @products = (
 	{
 		%{dclone(\%default_product_form)},
 		(
-			code => '200000000034',
+			code => '4260392550101',
 			product_name => "Some product",
 			generic_name => "Tester",
 			ingredients_text => "apple, milk, eggs, palm oil",
@@ -56,90 +56,120 @@ my $tests_ref = [
 	{
 		test_case => 'get-existing-product',
 		method => 'GET',
-		path => '/api/v3/product/200000000034',
+		path => '/api/v3/product/4260392550101',
+		expected_status_code => 200,
+	},
+	{
+		test_case => 'get-existing-product-gs1-caret',
+		method => 'GET',
+		path => '/api/v3/product/%5E0104260392550101',
+		expected_status_code => 200,
+	},
+	{
+		test_case => 'get-existing-product-gs1-fnc1',
+		method => 'GET',
+		path => '/api/v3/product/%1D0104260392550101',
+		expected_status_code => 200,
+	},
+	{
+		test_case => 'get-existing-product-gs1-gs',
+		method => 'GET',
+		path => '/api/v3/product/%E2%90%9D0104260392550101',
+		expected_status_code => 200,
+	},
+	{
+		test_case => 'get-existing-product-gs1-ai-data-str',
+		method => 'GET',
+		path => '/api/v3/product/(01)04260392550101',
+		expected_status_code => 200,
+	},
+	{
+		test_case => 'get-existing-product-gs1-data-uri',
+		method => 'GET',
+		path => '/api/v3/product/https%3A%2F%2Fid.gs1.org%2F01%2F04260392550101%2F10%2FABC%2F21%2F123456%3F17%3D211200',
 		expected_status_code => 200,
 	},
 	{
 		test_case => 'get-specific-fields',
 		method => 'GET',
-		path => '/api/v3/product/200000000034',
+		path => '/api/v3/product/4260392550101',
 		query_string => '?fields=product_name,categories_tags,categories_tags_en',
 		expected_status_code => 200,
 	},
 	{
 		test_case => 'get-images-to-update',
 		method => 'GET',
-		path => '/api/v3/product/200000000034',
+		path => '/api/v3/product/4260392550101',
 		query_string => '?fields=images_to_update_en',
 		expected_status_code => 200,
 	},
 	{
 		test_case => 'get-attribute-groups',
 		method => 'GET',
-		path => '/api/v3/product/200000000034',
+		path => '/api/v3/product/4260392550101',
 		query_string => '?fields=attribute_groups',
 		expected_status_code => 200,
 	},
 	{
 		test_case => 'get-attribute-groups-fr',
 		method => 'GET',
-		path => '/api/v3/product/200000000034',
+		path => '/api/v3/product/4260392550101',
 		query_string => '?fields=attribute_groups&lc=fr',
 		expected_status_code => 200,
 	},
 	{
 		test_case => 'get-knowledge-panels',
 		method => 'GET',
-		path => '/api/v3/product/200000000034',
+		path => '/api/v3/product/4260392550101',
 		query_string => '?fields=knowledge_panels',
 		expected_status_code => 200,
 	},
 	{
 		test_case => 'get-knowledge-panels-fr',
 		method => 'GET',
-		path => '/api/v3/product/200000000034',
+		path => '/api/v3/product/4260392550101',
 		query_string => '?fields=knowledge_panels&lc=fr',
 		expected_status_code => 200,
 	},
 	{
 		test_case => 'get-packagings',
 		method => 'GET',
-		path => '/api/v3/product/200000000034',
+		path => '/api/v3/product/4260392550101',
 		query_string => '?fields=packagings',
 		expected_status_code => 200,
 	},
 	{
 		test_case => 'get-packagings-fr',
 		method => 'GET',
-		path => '/api/v3/product/200000000034',
+		path => '/api/v3/product/4260392550101',
 		query_string => '?fields=packagings&tags_lc=fr',
 		expected_status_code => 200,
 	},
 	{
 		test_case => 'get-fields-raw',
 		method => 'GET',
-		path => '/api/v3/product/200000000034',
+		path => '/api/v3/product/4260392550101',
 		query_string => '?fields=raw',
 		expected_status_code => 200,
 	},
 	{
 		test_case => 'get-fields-all',
 		method => 'GET',
-		path => '/api/v3/product/200000000034',
+		path => '/api/v3/product/4260392550101',
 		query_string => '?fields=all',
 		expected_status_code => 200,
 	},
 	{
 		test_case => 'get-fields-all-knowledge-panels',
 		method => 'GET',
-		path => '/api/v3/product/200000000034',
+		path => '/api/v3/product/4260392550101',
 		query_string => '?fields=all,knowledge_panels',
 		expected_status_code => 200,
 	},
 	{
 		test_case => 'get-fields-attribute-groups-all-knowledge-panels',
 		method => 'GET',
-		path => '/api/v3/product/200000000034',
+		path => '/api/v3/product/4260392550101',
 		query_string => '?fields=attribute_groups,all,knowledge_panels',
 		expected_status_code => 200,
 	},
@@ -148,14 +178,14 @@ my $tests_ref = [
 	{
 		test_case => 'get-auth-good-password',
 		method => 'GET',
-		path => '/api/v3/product/200000000034',
+		path => '/api/v3/product/4260392550101',
 		query_string => '?fields=code,product_name&user_id=tests&password=testtest',
 		expected_status_code => 200,
 	},
 	{
 		test_case => 'get-auth-bad-user-password',
 		method => 'GET',
-		path => '/api/v3/product/200000000034',
+		path => '/api/v3/product/4260392550101',
 		query_string => '?fields=code,product_name&user_id=tests&password=bad_password',
 		expected_status_code => 200,
 	},

tests/integration/expected_test_results/api_v3_product_read/get-attribute-groups-fr.json

@@ -1,5 +1,5 @@
 {
-   "code" : "200000000034",
+   "code" : "4260392550101",
    "errors" : [],
    "product" : {
       "attribute_groups" : [

tests/integration/expected_test_results/api_v3_product_read/get-attribute-groups.json

@@ -1,5 +1,5 @@
 {
-   "code" : "200000000034",
+   "code" : "4260392550101",
    "errors" : [],
    "product" : {
       "attribute_groups" : [

tests/integration/expected_test_results/api_v3_product_read/get-auth-good-password.json

@@ -1,8 +1,8 @@
 {
-   "code" : "200000000034",
+   "code" : "4260392550101",
    "errors" : [],
    "product" : {
-      "code" : "200000000034",
+      "code" : "4260392550101",
       "product_name" : "Some product"
    },
    "result" : {