Improve RTL detection (#3682)

- Use plaintext
- Strip out URLs
- Strip out mentions
- Strip out hashtags
- Strip out whitespace from "overall" count
- Consistent between JS and Ruby

2 files changed, 6 insertions, 2 deletions

diff --git a/app/javascript/mastodon/components/status_content.js b/app/javascript/mastodon/components/status_content.js
index 06e25b36a..d22854288 100644
--- a/app/javascript/mastodon/components/status_content.js
+++ b/app/javascript/mastodon/components/status_content.js
@@ -100,7 +100,7 @@ class StatusContent extends React.PureComponent {
     const spoilerContent = { __html: emojify(escapeTextContentForBrowser(status.get('spoiler_text', ''))) };
     const directionStyle = { direction: 'ltr' };
 
-    if (isRtl(status.get('content'))) {
+    if (isRtl(status.get('search_index'))) {
       directionStyle.direction = 'rtl';
     }
 
diff --git a/app/javascript/mastodon/rtl.js b/app/javascript/mastodon/rtl.js
index 8f14bb338..00870a15d 100644
--- a/app/javascript/mastodon/rtl.js
+++ b/app/javascript/mastodon/rtl.js
@@ -17,11 +17,15 @@ export function isRtl(text) {
     return false;
   }
 
+  text = text.replace(/(?:^|[^\/\w])@([a-z0-9_]+(@[a-z0-9\.\-]+)?)/ig, '');
+  text = text.replace(/(?:^|[^\/\w])#([\S]+)/ig, '');
+  text = text.replace(/\s+/g, '');
+
   const matches = text.match(rtlChars);
 
   if (!matches) {
     return false;
   }
 
-  return matches.length / text.trim().length > 0.3;
+  return matches.length / text.length > 0.3;
 };