From: Dessalines Date: Mon, 30 Dec 2019 01:29:07 +0000 (-0500) Subject: Adding case insensitivity to slur filter. X-Git-Url: http://these/git/%22https:/join-lemmy.org/%22%7B%7D/static/%7BelementUrl%7D?a=commitdiff_plain;h=4ad836d6902f6cd662b3da7f410ef2b0d42b8cb2;p=lemmy.git Adding case insensitivity to slur filter. - Fixes #388 --- diff --git a/server/src/lib.rs b/server/src/lib.rs index dd0097f1..cddd5b86 100644 --- a/server/src/lib.rs +++ b/server/src/lib.rs @@ -42,7 +42,7 @@ use lettre::{SmtpClient, Transport}; use lettre_email::Email; use rand::distributions::Alphanumeric; use rand::{thread_rng, Rng}; -use regex::Regex; +use regex::{Regex, RegexBuilder}; pub fn to_datetime_utc(ndt: NaiveDateTime) -> DateTime { DateTime::::from_utc(ndt, Utc) @@ -138,11 +138,11 @@ mod tests { #[test] fn test_slur_filter() { let test = - "coons test dindu ladyboy tranny retardeds. This is a bunch of other safe text.".to_string(); + "coons test dindu ladyboy tranny retardeds. Capitalized Nigger. This is a bunch of other safe text.".to_string(); let slur_free = "No slurs here"; assert_eq!( remove_slurs(&test), - "*removed* test *removed* *removed* *removed* *removed*. This is a bunch of other safe text." + "*removed* test *removed* *removed* *removed* *removed*. Capitalized *removed*. This is a bunch of other safe text." .to_string() ); assert!(has_slurs(&test)); @@ -165,6 +165,6 @@ mod tests { lazy_static! { static ref EMAIL_REGEX: Regex = Regex::new(r"^[a-zA-Z0-9.!#$%&’*+/=?^_`{|}~-]+@[a-zA-Z0-9-]+(?:\.[a-zA-Z0-9-]+)*$").unwrap(); - static ref SLUR_REGEX: Regex = Regex::new(r"(fag(g|got|tard)?|maricos?|cock\s?sucker(s|ing)?|nig(\b|g?(a|er)?s?)\b|dindu(s?)|mudslime?s?|kikes?|mongoloids?|towel\s*heads?|\bspi(c|k)s?\b|\bchinks?|niglets?|beaners?|\bnips?\b|\bcoons?\b|jungle\s*bunn(y|ies?)|jigg?aboo?s?|\bpakis?\b|rag\s*heads?|gooks?|cunts?|bitch(es|ing|y)?|puss(y|ies?)|twats?|feminazis?|whor(es?|ing)|\bslut(s|t?y)?|\btrann?(y|ies?)|ladyboy(s?)|\b(b|re|r)tard(ed)?s?)").unwrap(); + static ref SLUR_REGEX: Regex = RegexBuilder::new(r"(fag(g|got|tard)?|maricos?|cock\s?sucker(s|ing)?|nig(\b|g?(a|er)?s?)\b|dindu(s?)|mudslime?s?|kikes?|mongoloids?|towel\s*heads?|\bspi(c|k)s?\b|\bchinks?|niglets?|beaners?|\bnips?\b|\bcoons?\b|jungle\s*bunn(y|ies?)|jigg?aboo?s?|\bpakis?\b|rag\s*heads?|gooks?|cunts?|bitch(es|ing|y)?|puss(y|ies?)|twats?|feminazis?|whor(es?|ing)|\bslut(s|t?y)?|\btrann?(y|ies?)|ladyboy(s?)|\b(b|re|r)tard(ed)?s?)").case_insensitive(true).build().unwrap(); static ref USERNAME_MATCHES_REGEX: Regex = Regex::new(r"/u/[a-zA-Z][0-9a-zA-Z_]*").unwrap(); }