Update lexer autolinker and extractor, add support for mentioned usernames containing dashes, periods and underscore characters

This commit is contained in:
Daniel Supernault 2020-02-02 23:20:19 -07:00
parent f7f16c185e
commit f911c96d16
No known key found for this signature in database
GPG key ID: 0DEF1C662C9033F7
2 changed files with 181 additions and 2 deletions

View file

@ -162,9 +162,9 @@ abstract class Regex
// look-ahead capture here and don't append $after when we return. // look-ahead capture here and don't append $after when we return.
$tmp['valid_mention_preceding_chars'] = '([^a-zA-Z0-9_!#\$%&*@\/]|^|(?:^|[^a-z0-9_+~.-])RT:?)'; $tmp['valid_mention_preceding_chars'] = '([^a-zA-Z0-9_!#\$%&*@\/]|^|(?:^|[^a-z0-9_+~.-])RT:?)';
$re['valid_mentions_or_lists'] = '/'.$tmp['valid_mention_preceding_chars'].'(['.$tmp['at_signs'].'])([a-z0-9_]{1,20})((\/[a-z][a-z0-9_\-]{0,24})?(?=(.*|$))(?:@[a-z0-9\.\-]+[a-z0-9]+)?)/i'; $re['valid_mentions_or_lists'] = '/'.$tmp['valid_mention_preceding_chars'].'(['.$tmp['at_signs'].'])([a-z0-9_\-.]{1,20})((\/[a-z][a-z0-9_\-]{0,24})?(?=(.*|$))(?:@[a-z0-9\.\-]+[a-z0-9]+)?)/i';
$re['valid_reply'] = '/^(?:['.$tmp['spaces'].'])*['.$tmp['at_signs'].']([a-z0-9_]{1,20})(?=(.*|$))/iu'; $re['valid_reply'] = '/^(?:['.$tmp['spaces'].'])*['.$tmp['at_signs'].']([a-z0-9_\-.]{1,20})(?=(.*|$))/iu';
$re['end_mention_match'] = '/\A(?:['.$tmp['at_signs'].']|['.$tmp['latin_accents'].']|:\/\/)/iu'; $re['end_mention_match'] = '/\A(?:['.$tmp['at_signs'].']|['.$tmp['latin_accents'].']|:\/\/)/iu';
// URL related hash regex collection // URL related hash regex collection

View file

@ -0,0 +1,179 @@
<?php
namespace Tests\Unit\Lexer;
use Tests\TestCase;
use Illuminate\Foundation\Testing\WithFaker;
use App\Util\Lexer\Autolink;
use App\Util\Lexer\Extractor;
class UsernameTest extends TestCase
{
/** @test **/
public function genericUsername()
{
$username = '@dansup';
$entities = Extractor::create()->extract($username);
$autolink = Autolink::create()->autolink($username);
$expectedAutolink = '<a class="u-url mention" href="https://pixelfed.dev/dansup" rel="external nofollow noopener" target="_blank">@dansup</a>';
$expectedEntity = [
"hashtags" => [],
"urls" => [],
"mentions" => [
"dansup",
],
"replyto" => "dansup",
"hashtags_with_indices" => [],
"urls_with_indices" => [],
"mentions_with_indices" => [
[
"screen_name" => "dansup",
"indices" => [
0,
7,
],
],
],
];
$this->assertEquals($expectedAutolink, $autolink);
$this->assertEquals($expectedEntity, $entities);
}
/** @test **/
public function usernameWithPeriod()
{
$username = '@dansup.two';
$autolink = Autolink::create()->autolink($username);
$entities = Extractor::create()->extract($username);
$expectedAutolink = '<a class="u-url mention" href="https://pixelfed.dev/dansup.two" rel="external nofollow noopener" target="_blank">@dansup.two</a>';
$expectedEntity = [
"hashtags" => [],
"urls" => [],
"mentions" => [
"dansup.two",
],
"replyto" => "dansup.two",
"hashtags_with_indices" => [],
"urls_with_indices" => [],
"mentions_with_indices" => [
[
"screen_name" => "dansup.two",
"indices" => [
0,
11,
],
],
],
];
$this->assertEquals($expectedAutolink, $autolink);
$this->assertEquals($expectedEntity, $entities);
}
/** @test **/
public function usernameWithDash()
{
$username = '@dansup-too';
$autolink = Autolink::create()->autolink($username);
$entities = Extractor::create()->extract($username);
$expectedAutolink = '<a class="u-url mention" href="https://pixelfed.dev/dansup-too" rel="external nofollow noopener" target="_blank">@dansup-too</a>';
$expectedEntity = [
"hashtags" => [],
"urls" => [],
"mentions" => [
"dansup-too",
],
"replyto" => "dansup-too",
"hashtags_with_indices" => [],
"urls_with_indices" => [],
"mentions_with_indices" => [
[
"screen_name" => "dansup-too",
"indices" => [
0,
11,
],
],
],
];
$this->assertEquals($expectedAutolink, $autolink);
$this->assertEquals($expectedEntity, $entities);
}
/** @test **/
public function usernameWithUnderscore()
{
$username = '@dansup_too';
$autolink = Autolink::create()->autolink($username);
$entities = Extractor::create()->extract($username);
$expectedAutolink = '<a class="u-url mention" href="https://pixelfed.dev/dansup_too" rel="external nofollow noopener" target="_blank">@dansup_too</a>';
$expectedEntity = [
"hashtags" => [],
"urls" => [],
"mentions" => [
"dansup_too",
],
"replyto" => "dansup_too",
"hashtags_with_indices" => [],
"urls_with_indices" => [],
"mentions_with_indices" => [
[
"screen_name" => "dansup_too",
"indices" => [
0,
11,
],
],
],
];
$this->assertEquals($expectedAutolink, $autolink);
$this->assertEquals($expectedEntity, $entities);
}
/** @test **/
public function multipleMentions()
{
$text = 'hello @dansup and @pixelfed.team from @username_underscore';
$autolink = Autolink::create()->autolink($text);
$entities = Extractor::create()->extract($text);
$expectedAutolink = 'hello <a class="u-url mention" href="https://pixelfed.dev/dansup" rel="external nofollow noopener" target="_blank">@dansup</a> and <a class="u-url mention" href="https://pixelfed.dev/pixelfed.team" rel="external nofollow noopener" target="_blank">@pixelfed.team</a> from <a class="u-url mention" href="https://pixelfed.dev/username_underscore" rel="external nofollow noopener" target="_blank">@username_underscore</a>';
$expectedEntity = [
"hashtags" => [],
"urls" => [],
"mentions" => [
"dansup",
"pixelfed.team",
"username_underscore",
],
"replyto" => null,
"hashtags_with_indices" => [],
"urls_with_indices" => [],
"mentions_with_indices" => [
[
"screen_name" => "dansup",
"indices" => [
6,
13,
],
],
[
"screen_name" => "pixelfed.team",
"indices" => [
18,
32,
],
],
[
"screen_name" => "username_underscore",
"indices" => [
38,
58,
],
],
],
];
$this->assertEquals($expectedAutolink, $autolink);
$this->assertEquals($expectedEntity, $entities);
}
}