From d3e53a89ddbefed33771cb42c305dbd200f09636 Mon Sep 17 00:00:00 2001
From: Michael Mayer <michael@lastzero.net>
Date: Wed, 5 May 2021 12:32:49 +0200
Subject: [PATCH] Metadata: Allow single quotes in keywords #1196

---
 pkg/txt/words.go      | 6 +++---
 pkg/txt/words_test.go | 9 +++++++--
 2 files changed, 10 insertions(+), 5 deletions(-)

diff --git a/pkg/txt/words.go b/pkg/txt/words.go
index 6810c3788..11988b047 100644
--- a/pkg/txt/words.go
+++ b/pkg/txt/words.go
@@ -6,7 +6,7 @@ import (
 	"strings"
 )
 
-var KeywordsRegexp = regexp.MustCompile("[\\p{L}\\-]{1,}")
+var KeywordsRegexp = regexp.MustCompile("[\\p{L}\\-']{1,}")
 
 // UnknownWord returns true if the string does not seem to be a real word.
 func UnknownWord(s string) bool {
@@ -34,7 +34,7 @@ func Words(s string) (results []string) {
 	}
 
 	for _, w := range KeywordsRegexp.FindAllString(s, -1) {
-		w = strings.Trim(w, "- ")
+		w = strings.Trim(w, "- '")
 
 		if w == "" || len(w) < 2 && IsLatin(w) {
 			continue
@@ -119,7 +119,7 @@ func UniqueWords(words []string) (results []string) {
 	SortCaseInsensitive(words)
 
 	for _, w := range words {
-		w = strings.Trim(strings.ToLower(w), "- ")
+		w = strings.Trim(strings.ToLower(w), "- '")
 
 		if w == "" || len(w) < 2 && IsLatin(w) || w == last {
 			continue
diff --git a/pkg/txt/words_test.go b/pkg/txt/words_test.go
index 8aa1d0b05..a2807a10a 100644
--- a/pkg/txt/words_test.go
+++ b/pkg/txt/words_test.go
@@ -25,7 +25,7 @@ func TestWords(t *testing.T) {
 	})
 	t.Run("I'm a lazy-brown fox!", func(t *testing.T) {
 		result := Words("I'm a lazy-BRoWN fox!")
-		assert.Equal(t, []string{"lazy-BRoWN", "fox"}, result)
+		assert.Equal(t, []string{"I'm", "lazy-BRoWN", "fox"}, result)
 	})
 	t.Run("no result", func(t *testing.T) {
 		result := Words("x")
@@ -51,6 +51,11 @@ func TestWords(t *testing.T) {
 		result := Words(" -foo- -")
 		assert.Equal(t, []string{"foo"}, result)
 	})
+	t.Run("McDonalds", func(t *testing.T) {
+		result := Words(" McDonald's FOO'bar-'")
+		assert.Equal(t, []string{"McDonald's", "FOO'bar"}, result)
+	})
+	// McDonald's
 }
 
 func TestReplaceSpaces(t *testing.T) {
@@ -97,7 +102,7 @@ func TestAddToWords(t *testing.T) {
 func TestMergeWords(t *testing.T) {
 	t.Run("I'm a lazy-BRoWN fox!", func(t *testing.T) {
 		result := MergeWords("I'm a lazy-BRoWN fox!", "Yellow banana, apple; pan-pot")
-		assert.Equal(t, "apple, banana, fox, lazy-brown, pan-pot, yellow", result)
+		assert.Equal(t, "apple, banana, fox, i'm, lazy-brown, pan-pot, yellow", result)
 	})
 }