+1

Cargo.lock

reviewed

··· 6305 6305 "p384", 6306 6306 "rand 0.8.5", 6307 6307 "redis", 6308 6308 + "regex", 6308 6309 "reqwest", 6309 6310 "serde", 6310 6311 "serde_bytes",

+1

Cargo.toml

reviewed

··· 30 30 multibase = "0.9.1" 31 31 multihash = "0.19.3" 32 32 rand = "0.8.5" 33 33 + regex = "1" 33 34 reqwest = { version = "0.12.28", features = ["json"] } 34 35 serde = { version = "1.0.228", features = ["derive"] } 35 36 serde_bytes = "0.11.14"

+1 -1

frontend/src/lib/api.ts

reviewed

··· 175 175 }); 176 176 const data = await response.json(); 177 177 if (!response.ok) { 178 178 - throw new ApiError(data.error, data.message, response.status); 178 178 + throw new ApiError(response.status, data.error, data.message); 179 179 } 180 180 return data; 181 181 },

+9 -1

src/api/identity/account.rs

reviewed

··· 194 194 .into_response(); 195 195 } 196 196 } 197 197 - input.handle.to_lowercase() 197 197 + let handle_lower = input.handle.to_lowercase(); 198 198 + if crate::moderation::has_explicit_slur(&handle_lower) { 199 199 + return ( 200 200 + StatusCode::BAD_REQUEST, 201 201 + Json(json!({"error": "InvalidHandle", "message": "Inappropriate language in handle"})), 202 202 + ) 203 203 + .into_response(); 204 204 + } 205 205 + handle_lower 198 206 }; 199 207 let email: Option<String> = input 200 208 .email

+7

src/api/identity/did.rs

reviewed

··· 582 582 ) 583 583 .into_response(); 584 584 } 585 585 + if crate::moderation::has_explicit_slur(new_handle) { 586 586 + return ( 587 587 + StatusCode::BAD_REQUEST, 588 588 + Json(json!({"error": "InvalidHandle", "message": "Inappropriate language in handle"})), 589 589 + ) 590 590 + .into_response(); 591 591 + } 585 592 let hostname = std::env::var("PDS_HOSTNAME").unwrap_or_else(|_| "localhost".to_string()); 586 593 let suffix = format!(".{}", hostname); 587 594 let is_service_domain = crate::handle::is_service_domain_handle(new_handle, &hostname);

+5 -3

src/api/repo/record/batch.rs

reviewed

··· 1 1 - use super::validation::validate_record; 1 1 + use super::validation::validate_record_with_rkey; 2 2 use super::write::has_verified_comms_channel; 3 3 use crate::api::repo::record::utils::{CommitParams, RecordOp, commit_and_log, extract_blob_cids}; 4 4 use crate::delegation::{self, DelegationActionType}; ··· 304 304 value, 305 305 } => { 306 306 if input.validate.unwrap_or(true) 307 307 - && let Err(err_response) = validate_record(value, collection) 307 307 + && let Err(err_response) = 308 308 + validate_record_with_rkey(value, collection, rkey.as_deref()) 308 309 { 309 310 return *err_response; 310 311 } ··· 357 358 value, 358 359 } => { 359 360 if input.validate.unwrap_or(true) 360 360 - && let Err(err_response) = validate_record(value, collection) 361 361 + && let Err(err_response) = 362 362 + validate_record_with_rkey(value, collection, Some(rkey)) 361 363 { 362 364 return *err_response; 363 365 }

+13 -1

src/api/repo/record/validation.rs

reviewed

··· 7 7 use serde_json::json; 8 8 9 9 pub fn validate_record(record: &serde_json::Value, collection: &str) -> Result<(), Box<Response>> { 10 10 + validate_record_with_rkey(record, collection, None) 11 11 + } 12 12 + 13 13 + pub fn validate_record_with_rkey( 14 14 + record: &serde_json::Value, 15 15 + collection: &str, 16 16 + rkey: Option<&str>, 17 17 + ) -> Result<(), Box<Response>> { 10 18 let validator = RecordValidator::new(); 11 11 - match validator.validate(record, collection) { 19 19 + match validator.validate_with_rkey(record, collection, rkey) { 12 20 Ok(_) => Ok(()), 13 21 Err(ValidationError::MissingType) => Err(Box::new(( 14 22 StatusCode::BAD_REQUEST, ··· 29 37 Err(ValidationError::InvalidDatetime { path }) => Err(Box::new(( 30 38 StatusCode::BAD_REQUEST, 31 39 Json(json!({"error": "InvalidRecord", "message": format!("Invalid datetime format at '{}'", path)})), 40 40 + ).into_response())), 41 41 + Err(ValidationError::BannedContent { path }) => Err(Box::new(( 42 42 + StatusCode::BAD_REQUEST, 43 43 + Json(json!({"error": "InvalidRecord", "message": format!("Unacceptable slur in record at '{}'", path)})), 32 44 ).into_response())), 33 45 Err(e) => Err(Box::new(( 34 46 StatusCode::BAD_REQUEST,

+5 -3

src/api/repo/record/write.rs

reviewed

··· 1 1 - use super::validation::validate_record; 1 1 + use super::validation::validate_record_with_rkey; 2 2 use crate::api::repo::record::utils::{CommitParams, RecordOp, commit_and_log, extract_blob_cids}; 3 3 use crate::delegation::{self, DelegationActionType}; 4 4 use crate::repo::tracking::TrackingBlockStore; ··· 257 257 } 258 258 }; 259 259 if input.validate.unwrap_or(true) 260 260 - && let Err(err_response) = validate_record(&input.record, &input.collection) 260 260 + && let Err(err_response) = 261 261 + validate_record_with_rkey(&input.record, &input.collection, input.rkey.as_deref()) 261 262 { 262 263 return *err_response; 263 264 } ··· 480 481 }; 481 482 let key = format!("{}/{}", collection_nsid, input.rkey); 482 483 if input.validate.unwrap_or(true) 483 483 - && let Err(err_response) = validate_record(&input.record, &input.collection) 484 484 + && let Err(err_response) = 485 485 + validate_record_with_rkey(&input.record, &input.collection, Some(&input.rkey)) 484 486 { 485 487 return *err_response; 486 488 }

+6

src/api/validation.rs

reviewed

··· 16 16 StartsWithInvalidChar, 17 17 EndsWithInvalidChar, 18 18 ContainsSpaces, 19 19 + BannedWord, 19 20 } 20 21 21 22 impl std::fmt::Display for HandleValidationError { ··· 41 42 } 42 43 Self::EndsWithInvalidChar => write!(f, "Handle cannot end with a hyphen or underscore"), 43 44 Self::ContainsSpaces => write!(f, "Handle cannot contain spaces"), 45 45 + Self::BannedWord => write!(f, "Inappropriate language in handle"), 44 46 } 45 47 } 46 48 } ··· 80 82 if !c.is_ascii_alphanumeric() && c != '-' && c != '_' { 81 83 return Err(HandleValidationError::InvalidCharacters); 82 84 } 85 85 + } 86 86 + 87 87 + if crate::moderation::has_explicit_slur(handle) { 88 88 + return Err(HandleValidationError::BannedWord); 83 89 } 84 90 85 91 Ok(handle.to_lowercase())

+1

src/lib.rs

reviewed

··· 10 10 pub mod handle; 11 11 pub mod image; 12 12 pub mod metrics; 13 13 + pub mod moderation; 13 14 pub mod oauth; 14 15 pub mod plc; 15 16 pub mod rate_limit;

+262

src/moderation/mod.rs

reviewed

··· 1 1 + /* 2 2 + * CONTENT WARNING 3 3 + * 4 4 + * This file contains explicit slurs and hateful language. We're sorry you have to see them. 5 5 + * 6 6 + * These words exist here for one reason: to ensure our moderation system correctly blocks them. 7 7 + * We can't verify the filter catches the n-word without testing against the actual word. 8 8 + * Euphemisms wouldn't prove the protection works. 9 9 + * 10 10 + * If reading this file has caused you distress, please know: 11 11 + * - you are valued and welcome in this community 12 12 + * - these words do not reflect the views of this project or its contributors 13 13 + * - we maintain this code precisely because we believe everyone deserves an experience on the web that is free from this kinda language 14 14 + */ 15 15 + 16 16 + use regex::Regex; 17 17 + use std::sync::OnceLock; 18 18 + 19 19 + static SLUR_REGEXES: OnceLock<Vec<Regex>> = OnceLock::new(); 20 20 + static EXTRA_BANNED_WORDS: OnceLock<Vec<String>> = OnceLock::new(); 21 21 + 22 22 + fn get_slur_regexes() -> &'static Vec<Regex> { 23 23 + SLUR_REGEXES.get_or_init(|| { 24 24 + vec![ 25 25 + Regex::new(r"\b[cĆćĈĉČčĊċÇçḈḉȻȼꞒꞓꟄꞔƇƈɕ][hĤĥȞȟḦḧḢḣḨḩḤḥḪḫH̱ẖĦħⱧⱨꞪɦꞕΗНн][iÍíi̇́Ììi̇̀ĬĭÎîǏǐÏïḮḯĨĩi̇̃ĮįĮ́į̇́Į̃į̇̃ĪīĪ̀ī̀ỈỉȈȉI̋i̋ȊȋỊịꞼꞽḬḭƗɨᶖİiIıＩｉ1lĺľļḷḹl̃ḽḻłŀƚꝉⱡɫɬꞎꬷꬸꬹᶅɭȴＬｌ][nŃńǸǹŇňÑñṄṅŅņṆṇṊṋṈṉN̈n̈ƝɲŊŋꞐꞑꞤꞥᵰᶇɳȵꬻꬼИиПпＮｎ][kḰḱǨǩĶķḲḳḴḵƘƙⱩⱪᶄꝀꝁꝂꝃꝄꝅꞢꞣ][sŚśṤṥŜŝŠšṦṧṠṡŞşṢṣṨṩȘșS̩s̩ꞨꞩⱾȿꟅʂᶊᵴ]?\b").unwrap(), 26 26 + Regex::new(r"\b[cĆćĈĉČčĊċÇçḈḉȻȼꞒꞓꟄꞔƇƈɕ][ÓóÒòŎŏÔôỐốỒồỖỗỔổǑǒÖöȪȫŐőÕõṌṍṎṏȬȭȮȯO͘o͘ȰȱØøǾǿǪǫǬǭŌōṒṓṐṑỎỏȌȍȎȏƠơỚớỜờỠỡỞởỢợỌọỘộO̩o̩Ò̩ò̩Ó̩ó̩ƟɵꝊꝋꝌꝍⱺＯｏ0]{2}[nŃńǸǹŇňÑñṄṅŅņṆṇṊṋṈṉN̈n̈ƝɲŊŋꞐꞑꞤꞥᵰᶇɳȵꬻꬼИиПпＮｎ][sŚśṤṥŜŝŠšṦṧṠṡŞşṢṣṨṩȘșS̩s̩ꞨꞩⱾȿꟅʂᶊᵴ]?\b").unwrap(), 27 27 + Regex::new(r"\b[fḞḟƑƒꞘꞙᵮᶂ][aÁáÀàĂăẮắẰằẴẵẲẳÂâẤấẦầẪẫẨẩǍǎÅåǺǻÄäǞǟÃãȦȧǠǡĄąĄ́ą́Ą̃ą̃ĀāĀ̀ā̀ẢảȀȁA̋a̋ȂȃẠạẶặẬậḀḁȺⱥꞺꞻᶏẚＡａ@4][gǴǵĞğĜĝǦǧĠġG̃g̃ĢģḠḡǤǥꞠꞡƓɠᶃꬶＧｇ]{1,2}([ÓóÒòŎŏÔôỐốỒồỖỗỔổǑǒÖöȪȫŐőÕõṌṍṎṏȬȭȮȯO͘o͘ȰȱØøǾǿǪǫǬǭŌōṒṓṐṑỎỏȌȍȎȏƠơỚớỜờỠỡỞởỢợỌọỘộO̩o̩Ò̩ò̩Ó̩ó̩ƟɵꝊꝋꝌꝍⱺＯｏ0e3ЄєЕеÉéÈèĔĕÊêẾếỀềỄễỂểÊ̄ê̄Ê̌ê̌ĚěËëẼẽĖėĖ́ė́Ė̃ė̃ȨȩḜḝĘęĘ́ę́Ę̃ę̃ĒēḖḗḔḕẺẻȄȅE̋e̋ȆȇẸẹỆệḘḙḚḛɆɇE̩e̩È̩è̩É̩é̩ᶒⱸꬴꬳＥｅiÍíi̇́Ììi̇̀ĬĭÎîǏǐÏïḮḯĨĩi̇̃ĮįĮ́į̇́Į̃į̇̃ĪīĪ̀ī̀ỈỉȈȉI̋i̋ȊȋỊịꞼꞽḬḭƗɨᶖİiIıＩｉ1lĺľļḷḹl̃ḽḻłŀƚꝉⱡɫɬꞎꬷꬸꬹᶅɭȴＬｌ][tŤťṪṫŢţṬṭȚțṰṱṮṯŦŧȾⱦƬƭƮʈT̈ẗᵵƫȶ]{1,2}([rŔŕŘřṘṙŖŗȐȑȒȓṚṛṜṝṞṟR̃r̃ɌɍꞦꞧⱤɽᵲᶉꭉ][yÝýỲỳŶŷY̊ẙŸÿỸỹẎẏȲȳỶỷỴỵɎɏƳƴỾỿ]|[rŔŕŘřṘṙŖŗȐȑȒȓṚṛṜṝṞṟR̃r̃ɌɍꞦꞧⱤɽᵲᶉꭉ][iÍíi̇́Ììi̇̀ĬĭÎîǏǐÏïḮḯĨĩi̇̃ĮįĮ́į̇́Į̃į̇̃ĪīĪ̀ī̀ỈỉȈȉI̋i̋ȊȋỊịꞼꞽḬḭƗɨᶖİiIıＩｉ1lĺľļḷḹl̃ḽḻłŀƚꝉⱡɫɬꞎꬷꬸꬹᶅɭȴＬｌ][e3ЄєЕеÉéÈèĔĕÊêẾếỀềỄễỂểÊ̄ê̄Ê̌ê̌ĚěËëẼẽĖėĖ́ė́Ė̃ė̃ȨȩḜḝĘęĘ́ę́Ę̃ę̃ĒēḖḗḔḕẺẻȄȅE̋e̋ȆȇẸẹỆệḘḙḚḛɆɇE̩e̩È̩è̩É̩é̩ᶒⱸꬴꬳＥｅ])?)?[sŚśṤṥŜŝŠšṦṧṠṡŞşṢṣṨṩȘșS̩s̩ꞨꞩⱾȿꟅʂᶊᵴ]?\b").unwrap(), 28 28 + Regex::new(r"\b[kḰḱǨǩĶķḲḳḴḵƘƙⱩⱪᶄꝀꝁꝂꝃꝄꝅꞢꞣ][iÍíi̇́Ììi̇̀ĬĭÎîǏǐÏïḮḯĨĩi̇̃ĮįĮ́į̇́Į̃į̇̃ĪīĪ̀ī̀ỈỉȈȉI̋i̋ȊȋỊịꞼꞽḬḭƗɨᶖİiIıＩｉ1lĺľļḷḹl̃ḽḻłŀƚꝉⱡɫɬꞎꬷꬸꬹᶅɭȴＬｌyÝýỲỳŶŷY̊ẙŸÿỸỹẎẏȲȳỶỷỴỵɎɏƳƴỾỿ][kḰḱǨǩĶķḲḳḴḵƘƙⱩⱪᶄꝀꝁꝂꝃꝄꝅꞢꞣ][e3ЄєЕеÉéÈèĔĕÊêẾếỀềỄễỂểÊ̄ê̄Ê̌ê̌ĚěËëẼẽĖėĖ́ė́Ė̃ė̃ȨȩḜḝĘęĘ́ę́Ę̃ę̃ĒēḖḗḔḕẺẻȄȅE̋e̋ȆȇẸẹỆệḘḙḚḛɆɇE̩e̩È̩è̩É̩é̩ᶒⱸꬴꬳＥｅ]([rŔŕŘřṘṙŖŗȐȑȒȓṚṛṜṝṞṟR̃r̃ɌɍꞦꞧⱤɽᵲᶉꭉ][yÝýỲỳŶŷY̊ẙŸÿỸỹẎẏȲȳỶỷỴỵɎɏƳƴỾỿ]|[rŔŕŘřṘṙŖŗȐȑȒȓṚṛṜṝṞṟR̃r̃ɌɍꞦꞧⱤɽᵲᶉꭉ][iÍíi̇́Ììi̇̀ĬĭÎîǏǐÏïḮḯĨĩi̇̃ĮįĮ́į̇́Į̃į̇̃ĪīĪ̀ī̀ỈỉȈȉI̋i̋ȊȋỊịꞼꞽḬḭƗɨᶖİiIıＩｉ1lĺľļḷḹl̃ḽḻłŀƚꝉⱡɫɬꞎꬷꬸꬹᶅɭȴＬｌ][e3ЄєЕеÉéÈèĔĕÊêẾếỀềỄễỂểÊ̄ê̄Ê̌ê̌ĚěËëẼẽĖėĖ́ė́Ė̃ė̃ȨȩḜḝĘęĘ́ę́Ę̃ę̃ĒēḖḗḔḕẺẻȄȅE̋e̋ȆȇẸẹỆệḘḙḚḛɆɇE̩e̩È̩è̩É̩é̩ᶒⱸꬴꬳＥｅ])?[sŚśṤṥŜŝŠšṦṧṠṡŞşṢṣṨṩȘșS̩s̩ꞨꞩⱾȿꟅʂᶊᵴ]*\b").unwrap(), 29 29 + Regex::new(r"\b[nŃńǸǹŇňÑñṄṅŅņṆṇṊṋṈṉN̈n̈ƝɲŊŋꞐꞑꞤꞥᵰᶇɳȵꬻꬼИиПпＮｎ][iÍíi̇́Ììi̇̀ĬĭÎîǏǐÏïḮḯĨĩi̇̃ĮįĮ́į̇́Į̃į̇̃ĪīĪ̀ī̀ỈỉȈȉI̋i̋ȊȋỊịꞼꞽḬḭƗɨᶖİiIıＩｉ1lĺľļḷḹl̃ḽḻłŀƚꝉⱡɫɬꞎꬷꬸꬹᶅɭȴＬｌoÓóÒòŎŏÔôỐốỒồỖỗỔổǑǒÖöȪȫŐőÕõṌṍṎṏȬȭȮȯO͘o͘ȰȱØøǾǿǪǫǬǭŌōṒṓṐṑỎỏȌȍȎȏƠơỚớỜờỠỡỞởỢợỌọỘộO̩o̩Ò̩ò̩Ó̩ó̩ƟɵꝊꝋꝌꝍⱺＯｏІіa4ÁáÀàĂăẮắẰằẴẵẲẳÂâẤấẦầẪẫẨẩǍǎÅåǺǻÄäǞǟÃãȦȧǠǡĄąĄ́ą́Ą̃ą̃ĀāĀ̀ā̀ẢảȀȁA̋a̋ȂȃẠạẶặẬậḀḁȺⱥꞺꞻᶏẚＡａ][gǴǵĞğĜĝǦǧĠġG̃g̃ĢģḠḡǤǥꞠꞡƓɠᶃꬶＧｇqꝖꝗꝘꝙɋʠ]{2}(l[e3ЄєЕеÉéÈèĔĕÊêẾếỀềỄễỂểÊ̄ê̄Ê̌ê̌ĚěËëẼẽĖėĖ́ė́Ė̃ė̃ȨȩḜḝĘęĘ́ę́Ę̃ę̃ĒēḖḗḔḕẺẻȄȅE̋e̋ȆȇẸẹỆệḘḙḚḛɆɇE̩e̩È̩è̩É̩é̩ᶒⱸꬴꬳＥｅ]t|[e3ЄєЕеÉéÈèĔĕÊêẾếỀềỄễỂểÊ̄ê̄Ê̌ê̌ĚěËëẼẽĖėĖ́ė́Ė̃ė̃ȨȩḜḝĘęĘ́ę́Ę̃ę̃ĒēḖḗḔḕẺẻȄȅE̋e̋ȆȇẸẹỆệḘḙḚḛɆɇE̩e̩È̩è̩É̩é̩ᶒⱸꬴꬳＥｅaÁáÀàĂăẮắẰằẴẵẲẳÂâẤấẦầẪẫẨẩǍǎÅåǺǻÄäǞǟÃãȦȧǠǡĄąĄ́ą́Ą̃ą̃ĀāĀ̀ā̀ẢảȀȁA̋a̋ȂȃẠạẶặẬậḀḁȺⱥꞺꞻᶏẚＡａ][rŔŕŘřṘṙŖŗȐȑȒȓṚṛṜṝṞṟR̃r̃ɌɍꞦꞧⱤɽᵲᶉꭉ]?|n[ÓóÒòŎŏÔôỐốỒồỖỗỔổǑǒÖöȪȫŐőÕõṌṍṎṏȬȭȮȯO͘o͘ȰȱØøǾǿǪǫǬǭŌōṒṓṐṑỎỏȌȍȎȏƠơỚớỜờỠỡỞởỢợỌọỘộO̩o̩Ò̩ò̩Ó̩ó̩ƟɵꝊꝋꝌꝍⱺＯｏ0][gǴǵĞğĜĝǦǧĠġG̃g̃ĢģḠḡǤǥꞠꞡƓɠᶃꬶＧｇqꝖꝗꝘꝙɋʠ]|[a4ÁáÀàĂăẮắẰằẴẵẲẳÂâẤấẦầẪẫẨẩǍǎÅåǺǻÄäǞǟÃãȦȧǠǡĄąĄ́ą́Ą̃ą̃ĀāĀ̀ā̀ẢảȀȁA̋a̋ȂȃẠạẶặẬậḀḁȺⱥꞺꞻᶏẚＡａ]?)?[sŚśṤṥŜŝŠšṦṧṠṡŞşṢṣṨṩȘșS̩s̩ꞨꞩⱾȿꟅʂᶊᵴ]?\b").unwrap(), 30 30 + Regex::new(r"[nŃńǸǹŇňÑñṄṅŅņṆṇṊṋṈṉN̈n̈ƝɲŊŋꞐꞑꞤꞥᵰᶇɳȵꬻꬼИиПпＮｎ][iÍíi̇́Ììi̇̀ĬĭÎîǏǐÏïḮḯĨĩi̇̃ĮįĮ́į̇́Į̃į̇̃ĪīĪ̀ī̀ỈỉȈȉI̋i̋ȊȋỊịꞼꞽḬḭƗɨᶖİiIıＩｉ1lĺľļḷḹl̃ḽḻłŀƚꝉⱡɫɬꞎꬷꬸꬹᶅɭȴＬｌoÓóÒòŎŏÔôỐốỒồỖỗỔổǑǒÖöȪȫŐőÕõṌṍṎṏȬȭȮȯO͘o͘ȰȱØøǾǿǪǫǬǭŌōṒṓṐṑỎỏȌȍȎȏƠơỚớỜờỠỡỞởỢợỌọỘộO̩o̩Ò̩ò̩Ó̩ó̩ƟɵꝊꝋꝌꝍⱺＯｏІіa4ÁáÀàĂăẮắẰằẴẵẲẳÂâẤấẦầẪẫẨẩǍǎÅåǺǻÄäǞǟÃãȦȧǠǡĄąĄ́ą́Ą̃ą̃ĀāĀ̀ā̀ẢảȀȁA̋a̋ȂȃẠạẶặẬậḀḁȺⱥꞺꞻᶏẚＡａ][gǴǵĞğĜĝǦǧĠġG̃g̃ĢģḠḡǤǥꞠꞡƓɠᶃꬶＧｇqꝖꝗꝘꝙɋʠ]{2}(l[e3ЄєЕеÉéÈèĔĕÊêẾếỀềỄễỂểÊ̄ê̄Ê̌ê̌ĚěËëẼẽĖėĖ́ė́Ė̃ė̃ȨȩḜḝĘęĘ́ę́Ę̃ę̃ĒēḖḗḔḕẺẻȄȅE̋e̋ȆȇẸẹỆệḘḙḚḛɆɇE̩e̩È̩è̩É̩é̩ᶒⱸꬴꬳＥｅ]t|[e3ЄєЕеÉéÈèĔĕÊêẾếỀềỄễỂểÊ̄ê̄Ê̌ê̌ĚěËëẼẽĖėĖ́ė́Ė̃ė̃ȨȩḜḝĘęĘ́ę́Ę̃ę̃ĒēḖḗḔḕẺẻȄȅE̋e̋ȆȇẸẹỆệḘḙḚḛɆɇE̩e̩È̩è̩É̩é̩ᶒⱸꬴꬳＥｅ][rŔŕŘřṘṙŖŗȐȑȒȓṚṛṜṝṞṟR̃r̃ɌɍꞦꞧⱤɽᵲᶉꭉ])[sŚśṤṥŜŝŠšṦṧṠṡŞşṢṣṨṩȘșS̩s̩ꞨꞩⱾȿꟅʂᶊᵴ]?").unwrap(), 31 31 + Regex::new(r"\b[tŤťṪṫŢţṬṭȚțṰṱṮṯŦŧȾⱦƬƭƮʈT̈ẗᵵƫȶ][rŔŕŘřṘṙŖŗȐȑȒȓṚṛṜṝṞṟR̃r̃ɌɍꞦꞧⱤɽᵲᶉꭉ][aÁáÀàĂăẮắẰằẴẵẲẳÂâẤấẦầẪẫẨẩǍǎÅåǺǻÄäǞǟÃãȦȧǠǡĄąĄ́ą́Ą̃ą̃ĀāĀ̀ā̀ẢảȀȁA̋a̋ȂȃẠạẶặẬậḀḁȺⱥꞺꞻᶏẚＡａ4]+[nŃńǸǹŇňÑñṄṅŅņṆṇṊṋṈṉN̈n̈ƝɲŊŋꞐꞑꞤꞥᵰᶇɳȵꬻꬼИиПпＮｎ]{1,2}([iÍíi̇́Ììi̇̀ĬĭÎîǏǐÏïḮḯĨĩi̇̃ĮįĮ́į̇́Į̃į̇̃ĪīĪ̀ī̀ỈỉȈȉI̋i̋ȊȋỊịꞼꞽḬḭƗɨᶖİiIıＩｉ1lĺľļḷḹl̃ḽḻłŀƚꝉⱡɫɬꞎꬷꬸꬹᶅɭȴＬｌ][e3ЄєЕеÉéÈèĔĕÊêẾếỀềỄễỂểÊ̄ê̄Ê̌ê̌ĚěËëẼẽĖėĖ́ė́Ė̃ė̃ȨȩḜḝĘęĘ́ę́Ę̃ę̃ĒēḖḗḔḕẺẻȄȅE̋e̋ȆȇẸẹỆệḘḙḚḛɆɇE̩e̩È̩è̩É̩é̩ᶒⱸꬴꬳＥｅ]|[yÝýỲỳŶŷY̊ẙŸÿỸỹẎẏȲȳỶỷỴỵɎɏƳƴỾỿ]|[e3ЄєЕеÉéÈèĔĕÊêẾếỀềỄễỂểÊ̄ê̄Ê̌ê̌ĚěËëẼẽĖėĖ́ė́Ė̃ė̃ȨȩḜḝĘęĘ́ę́Ę̃ę̃ĒēḖḗḔḕẺẻȄȅE̋e̋ȆȇẸẹỆệḘḙḚḛɆɇE̩e̩È̩è̩É̩é̩ᶒⱸꬴꬳＥｅ][rŔŕŘřṘṙŖŗȐȑȒȓṚṛṜṝṞṟR̃r̃ɌɍꞦꞧⱤɽᵲᶉꭉ])[sŚśṤṥŜŝŠšṦṧṠṡŞşṢṣṨṩȘșS̩s̩ꞨꞩⱾȿꟅʂᶊᵴ]?\b").unwrap(), 32 32 + ] 33 33 + }) 34 34 + } 35 35 + 36 36 + fn get_extra_banned_words() -> &'static Vec<String> { 37 37 + EXTRA_BANNED_WORDS.get_or_init(|| { 38 38 + std::env::var("PDS_BANNED_WORDS") 39 39 + .unwrap_or_default() 40 40 + .split(',') 41 41 + .map(|s| s.trim().to_lowercase()) 42 42 + .filter(|s| !s.is_empty()) 43 43 + .collect() 44 44 + }) 45 45 + } 46 46 + 47 47 + fn strip_trailing_digits(s: &str) -> &str { 48 48 + s.trim_end_matches(|c: char| c.is_ascii_digit()) 49 49 + } 50 50 + 51 51 + fn normalize_leetspeak(s: &str) -> String { 52 52 + s.chars() 53 53 + .map(|c| match c { 54 54 + '4' | '@' => 'a', 55 55 + '3' => 'e', 56 56 + '1' | '!' | '|' => 'i', 57 57 + '0' => 'o', 58 58 + '5' | '$' => 's', 59 59 + '7' => 't', 60 60 + '8' => 'b', 61 61 + '9' => 'g', 62 62 + _ => c, 63 63 + }) 64 64 + .collect() 65 65 + } 66 66 + 67 67 + pub fn has_explicit_slur(text: &str) -> bool { 68 68 + has_explicit_slur_with_extra_words(text, get_extra_banned_words()) 69 69 + } 70 70 + 71 71 + fn has_explicit_slur_with_extra_words(text: &str, extra_words: &[String]) -> bool { 72 72 + let text_lower = text.to_lowercase(); 73 73 + let normalized = text_lower.replace('.', "").replace('-', "").replace('_', ""); 74 74 + let stripped = strip_trailing_digits(&text_lower); 75 75 + let normalized_stripped = strip_trailing_digits(&normalized); 76 76 + 77 77 + let regexes = get_slur_regexes(); 78 78 + if regexes.iter().any(|r| { 79 79 + r.is_match(&text_lower) 80 80 + || r.is_match(&normalized) 81 81 + || r.is_match(stripped) 82 82 + || r.is_match(normalized_stripped) 83 83 + }) { 84 84 + return true; 85 85 + } 86 86 + 87 87 + if !extra_words.is_empty() { 88 88 + let leet_normalized = normalize_leetspeak(&normalized); 89 89 + let leet_stripped = normalize_leetspeak(strip_trailing_digits(&leet_normalized)); 90 90 + if extra_words.iter().any(|w| { 91 91 + text_lower.contains(w) 92 92 + || normalized.contains(w) 93 93 + || stripped.contains(w) 94 94 + || normalized_stripped.contains(w) 95 95 + || leet_normalized.contains(w) 96 96 + || leet_stripped.contains(w) 97 97 + }) { 98 98 + return true; 99 99 + } 100 100 + } 101 101 + false 102 102 + } 103 103 + 104 104 + #[cfg(test)] 105 105 + mod tests { 106 106 + use super::*; 107 107 + 108 108 + #[test] 109 109 + fn test_chink_pattern() { 110 110 + assert!(has_explicit_slur("chink")); 111 111 + assert!(has_explicit_slur("chinks")); 112 112 + assert!(has_explicit_slur("CHINK")); 113 113 + assert!(has_explicit_slur("Chinks")); 114 114 + } 115 115 + 116 116 + #[test] 117 117 + fn test_coon_pattern() { 118 118 + assert!(has_explicit_slur("coon")); 119 119 + assert!(has_explicit_slur("coons")); 120 120 + assert!(has_explicit_slur("COON")); 121 121 + } 122 122 + 123 123 + #[test] 124 124 + fn test_fag_pattern() { 125 125 + assert!(has_explicit_slur("fag")); 126 126 + assert!(has_explicit_slur("fags")); 127 127 + assert!(has_explicit_slur("faggot")); 128 128 + assert!(has_explicit_slur("faggots")); 129 129 + assert!(has_explicit_slur("faggotry")); 130 130 + } 131 131 + 132 132 + #[test] 133 133 + fn test_kike_pattern() { 134 134 + assert!(has_explicit_slur("kike")); 135 135 + assert!(has_explicit_slur("kikes")); 136 136 + assert!(has_explicit_slur("KIKE")); 137 137 + assert!(has_explicit_slur("kikery")); 138 138 + } 139 139 + 140 140 + #[test] 141 141 + fn test_nigger_pattern() { 142 142 + assert!(has_explicit_slur("nigger")); 143 143 + assert!(has_explicit_slur("niggers")); 144 144 + assert!(has_explicit_slur("NIGGER")); 145 145 + assert!(has_explicit_slur("nigga")); 146 146 + assert!(has_explicit_slur("niggas")); 147 147 + } 148 148 + 149 149 + #[test] 150 150 + fn test_tranny_pattern() { 151 151 + assert!(has_explicit_slur("tranny")); 152 152 + assert!(has_explicit_slur("trannies")); 153 153 + assert!(has_explicit_slur("TRANNY")); 154 154 + } 155 155 + 156 156 + #[test] 157 157 + fn test_normalization_bypass() { 158 158 + assert!(has_explicit_slur("n.i.g.g.e.r")); 159 159 + assert!(has_explicit_slur("n-i-g-g-e-r")); 160 160 + assert!(has_explicit_slur("n_i_g_g_e_r")); 161 161 + assert!(has_explicit_slur("f.a.g")); 162 162 + assert!(has_explicit_slur("f-a-g")); 163 163 + assert!(has_explicit_slur("c.h.i.n.k")); 164 164 + assert!(has_explicit_slur("k_i_k_e")); 165 165 + } 166 166 + 167 167 + #[test] 168 168 + fn test_trailing_digits_bypass() { 169 169 + assert!(has_explicit_slur("faggot123")); 170 170 + assert!(has_explicit_slur("nigger69")); 171 171 + assert!(has_explicit_slur("chink420")); 172 172 + assert!(has_explicit_slur("fag1")); 173 173 + assert!(has_explicit_slur("kike2024")); 174 174 + assert!(has_explicit_slur("n_i_g_g_e_r123")); 175 175 + } 176 176 + 177 177 + #[test] 178 178 + fn test_embedded_in_sentence() { 179 179 + assert!(has_explicit_slur("you are a faggot")); 180 180 + assert!(has_explicit_slur("stupid nigger")); 181 181 + assert!(has_explicit_slur("go away chink")); 182 182 + } 183 183 + 184 184 + #[test] 185 185 + fn test_safe_words_not_matched() { 186 186 + assert!(!has_explicit_slur("hello")); 187 187 + assert!(!has_explicit_slur("world")); 188 188 + assert!(!has_explicit_slur("bluesky")); 189 189 + assert!(!has_explicit_slur("tranquil")); 190 190 + assert!(!has_explicit_slur("programmer")); 191 191 + assert!(!has_explicit_slur("trigger")); 192 192 + assert!(!has_explicit_slur("bigger")); 193 193 + assert!(!has_explicit_slur("digger")); 194 194 + assert!(!has_explicit_slur("figure")); 195 195 + assert!(!has_explicit_slur("configure")); 196 196 + } 197 197 + 198 198 + #[test] 199 199 + fn test_similar_but_safe_words() { 200 200 + assert!(!has_explicit_slur("niggardly")); 201 201 + assert!(!has_explicit_slur("raccoon")); 202 202 + } 203 203 + 204 204 + #[test] 205 205 + fn test_empty_and_whitespace() { 206 206 + assert!(!has_explicit_slur("")); 207 207 + assert!(!has_explicit_slur(" ")); 208 208 + assert!(!has_explicit_slur("\t\n")); 209 209 + } 210 210 + 211 211 + #[test] 212 212 + fn test_case_insensitive() { 213 213 + assert!(has_explicit_slur("NIGGER")); 214 214 + assert!(has_explicit_slur("Nigger")); 215 215 + assert!(has_explicit_slur("NiGgEr")); 216 216 + assert!(has_explicit_slur("FAGGOT")); 217 217 + assert!(has_explicit_slur("Faggot")); 218 218 + } 219 219 + 220 220 + #[test] 221 221 + fn test_leetspeak_bypass() { 222 222 + assert!(has_explicit_slur("f4ggot")); 223 223 + assert!(has_explicit_slur("f4gg0t")); 224 224 + assert!(has_explicit_slur("n1gger")); 225 225 + assert!(has_explicit_slur("n1gg3r")); 226 226 + assert!(has_explicit_slur("k1ke")); 227 227 + assert!(has_explicit_slur("ch1nk")); 228 228 + assert!(has_explicit_slur("tr4nny")); 229 229 + } 230 230 + 231 231 + #[test] 232 232 + fn test_normalize_leetspeak() { 233 233 + assert_eq!(normalize_leetspeak("h3llo"), "hello"); 234 234 + assert_eq!(normalize_leetspeak("w0rld"), "world"); 235 235 + assert_eq!(normalize_leetspeak("t3$t"), "test"); 236 236 + assert_eq!(normalize_leetspeak("b4dw0rd"), "badword"); 237 237 + assert_eq!(normalize_leetspeak("l33t5p34k"), "leetspeak"); 238 238 + assert_eq!(normalize_leetspeak("@ss"), "ass"); 239 239 + assert_eq!(normalize_leetspeak("sh!t"), "shit"); 240 240 + assert_eq!(normalize_leetspeak("normal"), "normal"); 241 241 + } 242 242 + 243 243 + #[test] 244 244 + fn test_extra_banned_words() { 245 245 + let extra = vec!["badword".to_string(), "offensive".to_string()]; 246 246 + 247 247 + assert!(has_explicit_slur_with_extra_words("badword", &extra)); 248 248 + assert!(has_explicit_slur_with_extra_words("BADWORD", &extra)); 249 249 + assert!(has_explicit_slur_with_extra_words("b.a.d.w.o.r.d", &extra)); 250 250 + assert!(has_explicit_slur_with_extra_words("b-a-d-w-o-r-d", &extra)); 251 251 + assert!(has_explicit_slur_with_extra_words("b_a_d_w_o_r_d", &extra)); 252 252 + assert!(has_explicit_slur_with_extra_words("badword123", &extra)); 253 253 + assert!(has_explicit_slur_with_extra_words("b4dw0rd", &extra)); 254 254 + assert!(has_explicit_slur_with_extra_words("b4dw0rd789", &extra)); 255 255 + assert!(has_explicit_slur_with_extra_words("b.4.d.w.0.r.d", &extra)); 256 256 + assert!(has_explicit_slur_with_extra_words("this contains badword here", &extra)); 257 257 + assert!(has_explicit_slur_with_extra_words("0ff3n$1v3", &extra)); 258 258 + 259 259 + assert!(!has_explicit_slur_with_extra_words("goodword", &extra)); 260 260 + assert!(!has_explicit_slur_with_extra_words("hello world", &extra)); 261 261 + } 262 262 + }

+127 -22

src/validation/mod.rs

reviewed

··· 17 17 InvalidRecord(String), 18 18 #[error("Unknown record type: {0}")] 19 19 UnknownType(String), 20 20 + #[error("Unacceptable slur in record at {path}")] 21 21 + BannedContent { path: String }, 20 22 } 21 23 22 24 #[derive(Debug, Clone, Copy, PartialEq, Eq)] ··· 53 55 record: &Value, 54 56 collection: &str, 55 57 ) -> Result<ValidationStatus, ValidationError> { 58 58 + self.validate_with_rkey(record, collection, None) 59 59 + } 60 60 + 61 61 + pub fn validate_with_rkey( 62 62 + &self, 63 63 + record: &Value, 64 64 + collection: &str, 65 65 + rkey: Option<&str>, 66 66 + ) -> Result<ValidationStatus, ValidationError> { 56 67 let obj = record.as_object().ok_or_else(|| { 57 68 ValidationError::InvalidRecord("Record must be an object".to_string()) 58 69 })?; ··· 78 89 "app.bsky.graph.block" => self.validate_block(obj)?, 79 90 "app.bsky.graph.list" => self.validate_list(obj)?, 80 91 "app.bsky.graph.listitem" => self.validate_list_item(obj)?, 81 81 - "app.bsky.feed.generator" => self.validate_feed_generator(obj)?, 92 92 + "app.bsky.feed.generator" => self.validate_feed_generator(obj, rkey)?, 82 93 "app.bsky.feed.threadgate" => self.validate_threadgate(obj)?, 83 94 "app.bsky.labeler.service" => self.validate_labeler_service(obj)?, 95 95 + "app.bsky.graph.starterpack" => self.validate_starterpack(obj)?, 84 96 _ => { 85 97 if self.require_lexicon { 86 98 return Err(ValidationError::UnknownType(record_type.to_string())); ··· 126 138 }); 127 139 } 128 140 for (i, tag) in tags.iter().enumerate() { 129 129 - if let Some(tag_str) = tag.as_str() 130 130 - && tag_str.len() > 640 131 131 - { 132 132 - return Err(ValidationError::InvalidField { 133 133 - path: format!("tags/{}", i), 134 134 - message: "Tag exceeds maximum length of 640 bytes".to_string(), 135 135 - }); 141 141 + if let Some(tag_str) = tag.as_str() { 142 142 + if tag_str.len() > 640 { 143 143 + return Err(ValidationError::InvalidField { 144 144 + path: format!("tags/{}", i), 145 145 + message: "Tag exceeds maximum length of 640 bytes".to_string(), 146 146 + }); 147 147 + } 148 148 + if crate::moderation::has_explicit_slur(tag_str) { 149 149 + return Err(ValidationError::BannedContent { 150 150 + path: format!("tags/{}", i), 151 151 + }); 152 152 + } 153 153 + } 154 154 + } 155 155 + } 156 156 + if let Some(facets) = obj.get("facets").and_then(|v| v.as_array()) { 157 157 + for (i, facet) in facets.iter().enumerate() { 158 158 + if let Some(features) = facet.get("features").and_then(|v| v.as_array()) { 159 159 + for (j, feature) in features.iter().enumerate() { 160 160 + let is_tag = feature 161 161 + .get("$type") 162 162 + .and_then(|v| v.as_str()) 163 163 + .is_some_and(|t| t == "app.bsky.richtext.facet#tag"); 164 164 + if is_tag { 165 165 + if let Some(tag) = feature.get("tag").and_then(|v| v.as_str()) { 166 166 + if crate::moderation::has_explicit_slur(tag) { 167 167 + return Err(ValidationError::BannedContent { 168 168 + path: format!("facets/{}/features/{}/tag", i, j), 169 169 + }); 170 170 + } 171 171 + } 172 172 + } 173 173 + } 136 174 } 137 175 } 138 176 } ··· 154 192 ), 155 193 }); 156 194 } 195 195 + if crate::moderation::has_explicit_slur(display_name) { 196 196 + return Err(ValidationError::BannedContent { 197 197 + path: "displayName".to_string(), 198 198 + }); 199 199 + } 157 200 } 158 201 if let Some(description) = obj.get("description").and_then(|v| v.as_str()) { 159 202 let grapheme_count = description.chars().count(); ··· 164 207 "Description exceeds maximum length of 2560 characters (got {})", 165 208 grapheme_count 166 209 ), 210 210 + }); 211 211 + } 212 212 + if crate::moderation::has_explicit_slur(description) { 213 213 + return Err(ValidationError::BannedContent { 214 214 + path: "description".to_string(), 167 215 }); 168 216 } 169 217 } ··· 238 286 if !obj.contains_key("createdAt") { 239 287 return Err(ValidationError::MissingField("createdAt".to_string())); 240 288 } 241 241 - if let Some(name) = obj.get("name").and_then(|v| v.as_str()) 242 242 - && (name.is_empty() || name.len() > 64) 243 243 - { 244 244 - return Err(ValidationError::InvalidField { 245 245 - path: "name".to_string(), 246 246 - message: "Name must be 1-64 characters".to_string(), 247 247 - }); 289 289 + if let Some(name) = obj.get("name").and_then(|v| v.as_str()) { 290 290 + if name.is_empty() || name.len() > 64 { 291 291 + return Err(ValidationError::InvalidField { 292 292 + path: "name".to_string(), 293 293 + message: "Name must be 1-64 characters".to_string(), 294 294 + }); 295 295 + } 296 296 + if crate::moderation::has_explicit_slur(name) { 297 297 + return Err(ValidationError::BannedContent { 298 298 + path: "name".to_string(), 299 299 + }); 300 300 + } 248 301 } 249 302 Ok(()) 250 303 } ··· 268 321 fn validate_feed_generator( 269 322 &self, 270 323 obj: &serde_json::Map<String, Value>, 324 324 + rkey: Option<&str>, 271 325 ) -> Result<(), ValidationError> { 272 326 if !obj.contains_key("did") { 273 327 return Err(ValidationError::MissingField("did".to_string())); ··· 278 332 if !obj.contains_key("createdAt") { 279 333 return Err(ValidationError::MissingField("createdAt".to_string())); 280 334 } 281 281 - if let Some(display_name) = obj.get("displayName").and_then(|v| v.as_str()) 282 282 - && (display_name.is_empty() || display_name.len() > 240) 283 283 - { 284 284 - return Err(ValidationError::InvalidField { 285 285 - path: "displayName".to_string(), 286 286 - message: "displayName must be 1-240 characters".to_string(), 287 287 - }); 335 335 + if let Some(rkey) = rkey { 336 336 + if crate::moderation::has_explicit_slur(rkey) { 337 337 + return Err(ValidationError::BannedContent { 338 338 + path: "rkey".to_string(), 339 339 + }); 340 340 + } 341 341 + } 342 342 + if let Some(display_name) = obj.get("displayName").and_then(|v| v.as_str()) { 343 343 + if display_name.is_empty() || display_name.len() > 240 { 344 344 + return Err(ValidationError::InvalidField { 345 345 + path: "displayName".to_string(), 346 346 + message: "displayName must be 1-240 characters".to_string(), 347 347 + }); 348 348 + } 349 349 + if crate::moderation::has_explicit_slur(display_name) { 350 350 + return Err(ValidationError::BannedContent { 351 351 + path: "displayName".to_string(), 352 352 + }); 353 353 + } 354 354 + } 355 355 + Ok(()) 356 356 + } 357 357 + 358 358 + fn validate_starterpack( 359 359 + &self, 360 360 + obj: &serde_json::Map<String, Value>, 361 361 + ) -> Result<(), ValidationError> { 362 362 + if !obj.contains_key("name") { 363 363 + return Err(ValidationError::MissingField("name".to_string())); 364 364 + } 365 365 + if !obj.contains_key("createdAt") { 366 366 + return Err(ValidationError::MissingField("createdAt".to_string())); 367 367 + } 368 368 + if let Some(name) = obj.get("name").and_then(|v| v.as_str()) { 369 369 + if name.is_empty() || name.len() > 500 { 370 370 + return Err(ValidationError::InvalidField { 371 371 + path: "name".to_string(), 372 372 + message: "name must be 1-500 characters".to_string(), 373 373 + }); 374 374 + } 375 375 + if crate::moderation::has_explicit_slur(name) { 376 376 + return Err(ValidationError::BannedContent { 377 377 + path: "name".to_string(), 378 378 + }); 379 379 + } 380 380 + } 381 381 + if let Some(description) = obj.get("description").and_then(|v| v.as_str()) { 382 382 + if description.len() > 3000 { 383 383 + return Err(ValidationError::InvalidField { 384 384 + path: "description".to_string(), 385 385 + message: "description must be at most 3000 characters".to_string(), 386 386 + }); 387 387 + } 388 388 + if crate::moderation::has_explicit_slur(description) { 389 389 + return Err(ValidationError::BannedContent { 390 390 + path: "description".to_string(), 391 391 + }); 392 392 + } 288 393 } 289 394 Ok(()) 290 395 }

+196

tests/banned_words.rs

reviewed

··· 1 1 + /* 2 2 + * CONTENT WARNING 3 3 + * 4 4 + * This file contains explicit slurs and hateful language. We're sorry you have to see them. 5 5 + * 6 6 + * These words exist here for one reason: to ensure our moderation system correctly blocks them. 7 7 + * We can't verify the filter catches the n-word without testing against the actual word. 8 8 + * Euphemisms wouldn't prove the protection works. 9 9 + * 10 10 + * If reading this file has caused you distress, please know: 11 11 + * - you are valued and welcome in this community 12 12 + * - these words do not reflect the views of this project or its contributors 13 13 + * - we maintain this code precisely because we believe everyone deserves an experience on the web that is free from this kinda language 14 14 + */ 15 15 + 16 16 + mod common; 17 17 + mod helpers; 18 18 + use common::*; 19 19 + use helpers::*; 20 20 + use reqwest::StatusCode; 21 21 + use serde_json::json; 22 22 + 23 23 + #[tokio::test] 24 24 + async fn test_handle_with_slur_rejected() { 25 25 + let client = client(); 26 26 + let timestamp = chrono::Utc::now().timestamp_millis(); 27 27 + let offensive_handle = format!("nigger{}", timestamp); 28 28 + 29 29 + let create_payload = json!({ 30 30 + "handle": offensive_handle, 31 31 + "email": format!("test{}@example.com", timestamp), 32 32 + "password": "TestPassword123!" 33 33 + }); 34 34 + 35 35 + let res = client 36 36 + .post(format!( 37 37 + "{}/xrpc/com.atproto.server.createAccount", 38 38 + base_url().await 39 39 + )) 40 40 + .json(&create_payload) 41 41 + .send() 42 42 + .await 43 43 + .expect("Request failed"); 44 44 + 45 45 + assert_eq!(res.status(), StatusCode::BAD_REQUEST); 46 46 + let body: serde_json::Value = res.json().await.unwrap(); 47 47 + assert_eq!(body["error"], "InvalidHandle"); 48 48 + assert!(body["message"] 49 49 + .as_str() 50 50 + .unwrap_or("") 51 51 + .contains("Inappropriate language")); 52 52 + } 53 53 + 54 54 + #[tokio::test] 55 55 + async fn test_handle_with_normalized_slur_rejected() { 56 56 + let client = client(); 57 57 + let timestamp = chrono::Utc::now().timestamp_millis(); 58 58 + let offensive_handle = format!("n-i-g-g-e-r{}", timestamp); 59 59 + 60 60 + let create_payload = json!({ 61 61 + "handle": offensive_handle, 62 62 + "email": format!("test{}@example.com", timestamp), 63 63 + "password": "TestPassword123!" 64 64 + }); 65 65 + 66 66 + let res = client 67 67 + .post(format!( 68 68 + "{}/xrpc/com.atproto.server.createAccount", 69 69 + base_url().await 70 70 + )) 71 71 + .json(&create_payload) 72 72 + .send() 73 73 + .await 74 74 + .expect("Request failed"); 75 75 + 76 76 + assert_eq!(res.status(), StatusCode::BAD_REQUEST); 77 77 + let body: serde_json::Value = res.json().await.unwrap(); 78 78 + assert_eq!(body["error"], "InvalidHandle"); 79 79 + } 80 80 + 81 81 + #[tokio::test] 82 82 + async fn test_handle_update_with_slur_rejected() { 83 83 + let client = client(); 84 84 + let (_, jwt) = setup_new_user("handleupdate").await; 85 85 + 86 86 + let update_payload = json!({ 87 87 + "handle": "faggots" 88 88 + }); 89 89 + 90 90 + let res = client 91 91 + .post(format!( 92 92 + "{}/xrpc/com.atproto.identity.updateHandle", 93 93 + base_url().await 94 94 + )) 95 95 + .bearer_auth(&jwt) 96 96 + .json(&update_payload) 97 97 + .send() 98 98 + .await 99 99 + .expect("Request failed"); 100 100 + 101 101 + assert_eq!(res.status(), StatusCode::BAD_REQUEST); 102 102 + let body: serde_json::Value = res.json().await.unwrap(); 103 103 + assert_eq!(body["error"], "InvalidHandle"); 104 104 + } 105 105 + 106 106 + #[tokio::test] 107 107 + async fn test_profile_displayname_with_slur_rejected() { 108 108 + let client = client(); 109 109 + let (did, jwt) = setup_new_user("profileslur").await; 110 110 + 111 111 + let profile = json!({ 112 112 + "repo": did, 113 113 + "collection": "app.bsky.actor.profile", 114 114 + "rkey": "self", 115 115 + "record": { 116 116 + "$type": "app.bsky.actor.profile", 117 117 + "displayName": "I am a kike" 118 118 + } 119 119 + }); 120 120 + 121 121 + let res = client 122 122 + .post(format!( 123 123 + "{}/xrpc/com.atproto.repo.putRecord", 124 124 + base_url().await 125 125 + )) 126 126 + .bearer_auth(&jwt) 127 127 + .json(&profile) 128 128 + .send() 129 129 + .await 130 130 + .expect("Request failed"); 131 131 + 132 132 + assert_eq!(res.status(), StatusCode::BAD_REQUEST); 133 133 + let body: serde_json::Value = res.json().await.unwrap(); 134 134 + assert_eq!(body["error"], "InvalidRecord"); 135 135 + } 136 136 + 137 137 + #[tokio::test] 138 138 + async fn test_profile_description_with_slur_rejected() { 139 139 + let client = client(); 140 140 + let (did, jwt) = setup_new_user("profiledesc").await; 141 141 + 142 142 + let profile = json!({ 143 143 + "repo": did, 144 144 + "collection": "app.bsky.actor.profile", 145 145 + "rkey": "self", 146 146 + "record": { 147 147 + "$type": "app.bsky.actor.profile", 148 148 + "displayName": "Normal Name", 149 149 + "description": "I hate all chinks" 150 150 + } 151 151 + }); 152 152 + 153 153 + let res = client 154 154 + .post(format!( 155 155 + "{}/xrpc/com.atproto.repo.putRecord", 156 156 + base_url().await 157 157 + )) 158 158 + .bearer_auth(&jwt) 159 159 + .json(&profile) 160 160 + .send() 161 161 + .await 162 162 + .expect("Request failed"); 163 163 + 164 164 + assert_eq!(res.status(), StatusCode::BAD_REQUEST); 165 165 + let body: serde_json::Value = res.json().await.unwrap(); 166 166 + assert_eq!(body["error"], "InvalidRecord"); 167 167 + } 168 168 + 169 169 + #[tokio::test] 170 170 + async fn test_clean_content_allowed() { 171 171 + let client = client(); 172 172 + let (did, jwt) = setup_new_user("cleanpost").await; 173 173 + 174 174 + let post = json!({ 175 175 + "repo": did, 176 176 + "collection": "app.bsky.feed.post", 177 177 + "record": { 178 178 + "$type": "app.bsky.feed.post", 179 179 + "text": "This is a perfectly normal post about coding and technology!", 180 180 + "createdAt": chrono::Utc::now().to_rfc3339() 181 181 + } 182 182 + }); 183 183 + 184 184 + let res = client 185 185 + .post(format!( 186 186 + "{}/xrpc/com.atproto.repo.createRecord", 187 187 + base_url().await 188 188 + )) 189 189 + .bearer_auth(&jwt) 190 190 + .json(&post) 191 191 + .send() 192 192 + .await 193 193 + .expect("Request failed"); 194 194 + 195 195 + assert_eq!(res.status(), StatusCode::OK); 196 196 + }