fix: escape issue in additives knowledge panel - remove latex formulas

lib/ProductOpener/KnowledgePanels.pm

@@ -216,11 +216,13 @@ The function converts the multiline string into a single line string.
 
 sub convert_multiline_string_to_singleline ($line) {
 
+	# Escape " and \ unless they have been escaped already
+	# negative look behind to not convert \n to \\n or \" to \\" or \\ to \\\\
+	$line =~ s/(?<!\\)("|\\)/\\$1/g;
+
 	# \R will match all Unicode newline sequence
 	$line =~ s/\R/\\n/sg;
-	# Escape quotes unless they have been escaped already
-	# negative look behind to not convert \" to \\"
-	$line =~ s/(?<!\\)"/\\"/g;
+
 	return '"' . $line . '"';
 }
 
@@ -1290,6 +1292,26 @@ sub create_ingredients_analysis_panel ($product_ref, $target_lc, $target_cc, $op
 	return;
 }
 
+sub remove_latex_sequences ($string) {
+
+	# Some wikipedia abstracts have chemical formulas like {\\displaystyle \\mathrm {NaNO_{3}} +\\mathrm {Pb} \\to \\mathrm {NaNO_{2}} +\\mathrm {PbO} }
+	# In practice, we remove everything between { }
+
+	$string =~ s/
+        (                   
+        {                   # match an opening {
+            (?:
+                [^{}]++     # one or more non angle brackets, non backtracking
+                  |
+                (?1)        # found { or }, so recurse to capture group 1
+            )*
+        }                   # match a closing }
+        )                   
+        //xg;
+
+	return $string;
+}
+
 =head2 add_taxonomy_properties_in_target_languages_to_object ( $object_ref, $tagtype, $tagid, $properties_ref, $target_lcs_ref )
 
 This function adds to the hash ref $object_ref (for instance a data structure passed to a template) the values
@@ -1332,7 +1354,7 @@ sub add_taxonomy_properties_in_target_languages_to_object ($object_ref, $tagtype
 			}
 		}
 		if (defined $property_value) {
-			$object_ref->{$property} = $property_value;
+			$object_ref->{$property} = remove_latex_sequences($property_value);
 			$object_ref->{$property . "_lc"} = $property_lc;
 			$object_ref->{$property . "_language"}
 				= display_taxonomy_tag($target_lcs_ref->[0], "languages", $language_codes{$property_lc});

templates/api/knowledge-panels/environment/carbon_footprint.tt.json

@@ -29,7 +29,7 @@
                 "type": "summary",
                 "html": `
                     <p>[% lang('source_ademe_agribalyse_for_category') %] 
-                    <a href=\"https://agribalyse.ademe.fr/app/aliments/[% product.ecoscore_data.agribalyse.code %]\">[% panel.agribalyse_category_name.dquote %]</a>
+                    <a href="https://agribalyse.ademe.fr/app/aliments/[% product.ecoscore_data.agribalyse.code %]">[% panel.agribalyse_category_name.dquote %]</a>
                     ([% lang('source_ademe_agribalyse') %])
                     </p>
                     `

templates/api/knowledge-panels/environment/ecoscore/agribalyse.tt.json

@@ -16,7 +16,7 @@
                 "type": "summary",
                 "html": `
                     <p>[% lang('categories_s') FILTER ucfirst %][% sep %]: 
-                    <a href=\"https://agribalyse.ademe.fr/app/aliments/[% product.ecoscore_data.agribalyse.code %]\">[% panel.agribalyse_category_name %]</a>
+                    <a href="https://agribalyse.ademe.fr/app/aliments/[% product.ecoscore_data.agribalyse.code %]">[% panel.agribalyse_category_name %]</a>
                     </p>
                     <ul>
                         <li>

tests/unit/knowledge_panels.t

@@ -0,0 +1,27 @@
+#!/usr/bin/perl -w
+
+use Modern::Perl '2017';
+use utf8;
+
+use Test::More;
+use Log::Any::Adapter 'TAP';
+
+use ProductOpener::KnowledgePanels qw/:all/;
+
+is(
+	ProductOpener::KnowledgePanels::convert_multiline_string_to_singleline('
+test
+muti-line string
+a slash A / B and anti-slash \
+HTML <a href="https://url.com">test</a>
+'),
+	'"\ntest\nmuti-line string\na slash A / B and anti-slash \\\\\nHTML <a href=\"https://url.com\">test</a>\n"'
+);
+
+is(
+	ProductOpener::KnowledgePanels::convert_multiline_string_to_singleline(
+		'<a href="https://agribalyse.ademe.fr/app/aliments/[% product.ecoscore_data.agribalyse.code %]">'),
+	'"<a href=\"https://agribalyse.ademe.fr/app/aliments/[% product.ecoscore_data.agribalyse.code %]\">"'
+);
+
+done_testing();