gtsocial-umbx

commit 664713ddd4f7236fde0759cf7a0e04a434417876
parent 9e7d022a06779a03e3eaaadad6cc33423f46892b
Author: tobi <31960611+tsmethurst@users.noreply.github.com>
Date:   Sun,  3 Jul 2022 11:03:03 +0200

[bugfix] Make hashtag regex work with non-ascii characters (#682)


Diffstat:
M
internal/regexes/regexes.go
 | 
4
++--
M
internal/util/statustools_test.go
 | 
9
+++++++--

2 files changed, 9 insertions(+), 4 deletions(-)
diff --git a/internal/regexes/regexes.go b/internal/regexes/regexes.go
@@ -71,8 +71,8 @@ var (
 	// MentionFinder extracts mentions from a piece of text.
 	MentionFinder = regexp.MustCompile(mentionFinder)
 
-	// hashtag regex can be played with here: https://regex101.com/r/bPxeca/1
-	hashtagFinder = fmt.Sprintf(`(?:^|\s)(?:#*)(#[a-zA-Z0-9]{1,%d})(?:#|\b)`, maximumHashtagLength)
+	// hashtag regex can be played with here: https://regex101.com/r/bpyGlj/1
+	hashtagFinder = fmt.Sprintf(`(?:^|\s)(?:#*)(#[\p{L}\p{N}]{1,%d})(?:#|\b)`, maximumHashtagLength)
 	// HashtagFinder finds possible hashtags in a string.
 	// It returns just the string part of the hashtag, not the # symbol.
 	HashtagFinder = regexp.MustCompile(hashtagFinder)
diff --git a/internal/util/statustools_test.go b/internal/util/statustools_test.go
@@ -83,15 +83,20 @@ func (suite *StatusTestSuite) TestDeriveHashtagsOK() {
 
 #ThisShouldAlsoWork #not_this_though
 
-#111111 thisalsoshouldn'twork#### ##`
+#111111 thisalsoshouldn'twork#### ##
+
+#alimentación, #saúde
+`
 
 	tags := util.DeriveHashtagsFromText(statusText)
-	assert.Len(suite.T(), tags, 5)
+	assert.Len(suite.T(), tags, 7)
 	assert.Equal(suite.T(), "testing123", tags[0])
 	assert.Equal(suite.T(), "also", tags[1])
 	assert.Equal(suite.T(), "thisshouldwork", tags[2])
 	assert.Equal(suite.T(), "ThisShouldAlsoWork", tags[3])
 	assert.Equal(suite.T(), "111111", tags[4])
+	assert.Equal(suite.T(), "alimentación", tags[5])
+	assert.Equal(suite.T(), "saúde", tags[6])
 }
 
 func (suite *StatusTestSuite) TestDeriveEmojiOK() {