forked from cybrespace/mastodon
Exclude URLs from text analysis (#11759)
By the added regex, URLs, including the one without http or even www like mysite.com will be removed from the toot's body so only the real text of the toot will be analyzed for RTL detection
This commit is contained in:
parent
bdca8da8eb
commit
e974d4923f
|
@ -20,6 +20,7 @@ export function isRtl(text) {
|
||||||
text = text.replace(/(?:^|[^\/\w])@([a-z0-9_]+(@[a-z0-9\.\-]+)?)/ig, '');
|
text = text.replace(/(?:^|[^\/\w])@([a-z0-9_]+(@[a-z0-9\.\-]+)?)/ig, '');
|
||||||
text = text.replace(/(?:^|[^\/\w])#([\S]+)/ig, '');
|
text = text.replace(/(?:^|[^\/\w])#([\S]+)/ig, '');
|
||||||
text = text.replace(/\s+/g, '');
|
text = text.replace(/\s+/g, '');
|
||||||
|
text = text.replace(/(\w\S+\.\w{2,}\S*)/g, '');
|
||||||
|
|
||||||
const matches = text.match(rtlChars);
|
const matches = text.match(rtlChars);
|
||||||
|
|
||||||
|
|
Loading…
Reference in New Issue