From 1aacb01385f9e6a5ec5fce9a51abce9d7cbfb69e Mon Sep 17 00:00:00 2001
From: benbenben2 <110821832+benbenben2@users.noreply.github.com>
Date: Wed, 15 Nov 2023 09:39:04 +0100
Subject: [PATCH] fix: fix_avoid_eiweiss_false_positive_for_allergens (#9317)

fix_avoid_eiweiss_false_positive_for_allergens
---
 lib/ProductOpener/Ingredients.pm |  4 ++++
 taxonomies/ingredients.txt       |  6 +++---
 tests/unit/allergens.t           | 15 +++++++++++++++
 3 files changed, 22 insertions(+), 3 deletions(-)

diff --git a/lib/ProductOpener/Ingredients.pm b/lib/ProductOpener/Ingredients.pm
index 3f93f6213e6d0..38b67ad721fa7 100644
--- a/lib/ProductOpener/Ingredients.pm
+++ b/lib/ProductOpener/Ingredients.pm
@@ -6689,6 +6689,10 @@ sub detect_allergens_from_text ($product_ref) {
 			$text =~ s/\b___([^,;_\(\)\[\]]+?)___\b/replace_allergen($language,$product_ref,$1,$`)/iesg;
 			$text =~ s/\b__([^,;_\(\)\[\]]+?)__\b/replace_allergen($language,$product_ref,$1,$`)/iesg;
 			$text =~ s/\b_([^,;_\(\)\[\]]+?)_\b/replace_allergen($language,$product_ref,$1,$`)/iesg;
+			# _Weizen_eiweiß is not caught in last regex because of \b (word boundary).
+			if ($language eq 'de') {
+				$text =~ s/\b_([^,;_\(\)\[\]]+?)_/replace_allergen($language,$product_ref,$1,$`)/iesg;
+			}
 
 			# allergens in all caps, with other ingredients not in all caps
 
diff --git a/taxonomies/ingredients.txt b/taxonomies/ingredients.txt
index e6f4aec69c7aa..2216f7fe6b460 100644
--- a/taxonomies/ingredients.txt
+++ b/taxonomies/ingredients.txt
@@ -22150,7 +22150,7 @@ cs:pšenice
 cv:тулă
 cy:gwenith
 da:hvede
-de:weizen
+de:weizen, weizen art
 dv:ގޮދަން
 el:σιτάρι
 es:trigo
@@ -30941,7 +30941,7 @@ en:wheat protein
 bg:пшеничен протеин
 ca:proteïna de blat
 cs:pšeničná bílkovina, pšeničná bilkovina
-de:Weizenprotein, Weizeneiweiß, Weizeneiweiss
+de:Weizenprotein, Weizeneiweiß, Weizeneiweiss, Weizen eiweiß
 es:Proteina de trigo
 fi:vehnäproteiini
 fr:protéine de blé, protéines de blé, protéines de froment
@@ -87042,7 +87042,7 @@ bn:সয়া সস
 ca:salsa de soia
 cs:sójová omáčka
 da:sojasovs, sojasauce
-de:Sojasauce, Sojasoße, Sojasosse, Soja Sosse, Sojasoßen, Sojasossen, Sojasaucen
+de:Sojasauce, Sojasoße, Sojasosse, Soja Sosse, Sojasoßen, Sojasossen, Sojasaucen, Soja soße
 el:Σάλτσα σόγιας
 eo:sojsaŭco
 es:salsa de soya, salsa de soja,shoyu
diff --git a/tests/unit/allergens.t b/tests/unit/allergens.t
index 50ed520670f77..676570fc2fba3 100644
--- a/tests/unit/allergens.t
+++ b/tests/unit/allergens.t
@@ -518,4 +518,19 @@ detect_allergens_from_text($product_ref);
 
 is($product_ref->{ingredients_text_with_allergens_en}, "Whole Grain Oat Flakes (65.0%)");
 
+# German and underscores, see issue #8386
+$product_ref = {
+	lc => "de",
+	lang => "de",
+	ingredients_text_de =>
+		"Seitan 65% (_Weizen_eiweiß, Wasser), Rapsöl, Kidneybohnen, _Dinkel_vollkornmehl (_Weizen_art), Apfelessig, Gewürze, Tomatenmark, _Soja_soße (Wasser, _Soja_bohnen, Salz, _Weizen_mehl), Kartoffelstärke, Salz."
+};
+
+compute_languages($product_ref);
+detect_allergens_from_text($product_ref);
+
+diag explain $product_ref->{allergens_tags};
+
+is_deeply($product_ref->{allergens_tags}, ['en:gluten', 'en:soybeans',]) || diag explain $product_ref->{allergens_tags};
+
 done_testing();