Fix parsing non-ascii tags

This commit is contained in:
tusooa 2023-07-21 13:54:10 -04:00
parent 510392e4ca
commit 0b0b1dabdf
No known key found for this signature in database
GPG Key ID: 42AEC43D48433C51
3 changed files with 12 additions and 2 deletions

View File

@ -0,0 +1 @@
Fix parsing non-ascii tags

View File

@ -14,8 +14,11 @@ export const mentionMatchesUrl = (attention, url) => {
* @param {string} url
*/
export const extractTagFromUrl = (url) => {
const regex = /tag[s]*\/(\w+)$/g
const result = regex.exec(url)
const decoded = decodeURI(url)
// https://git.pleroma.social/pleroma/elixir-libraries/linkify/-/blob/master/lib/linkify/parser.ex
// https://www.pcre.org/original/doc/html/pcrepattern.html
const regex = /tag[s]*\/([\p{L}\p{N}_]*[\p{Alphabetic}_·\u{200c}][\p{L}\p{N}_·\p{M}\u{200c}]*)$/ug
const result = regex.exec(decoded)
if (!result) {
return false
}

View File

@ -78,5 +78,11 @@ describe('MatcherService', () => {
expect(MatcherService.extractTagFromUrl(url)).to.eql(false)
})
it('should return tag name from non-ascii tags', () => {
const url = encodeURI('https://website.com/tag/喵喵喵')
expect(MatcherService.extractTagFromUrl(url)).to.eql('喵喵喵')
})
})
})