+1

Cargo.lock

··· 6305 "p384", 6306 "rand 0.8.5", 6307 "redis", 0 6308 "reqwest", 6309 "serde", 6310 "serde_bytes",

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

+1

Cargo.toml

··· 30 multibase = "0.9.1" 31 multihash = "0.19.3" 32 rand = "0.8.5" 0 33 reqwest = { version = "0.12.28", features = ["json"] } 34 serde = { version = "1.0.228", features = ["derive"] } 35 serde_bytes = "0.11.14"

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

+1 -1

frontend/src/lib/api.ts

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

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

+9 -1

src/api/identity/account.rs

··· 194 .into_response(); 195 } 196 } 197 - input.handle.to_lowercase() 0 0 0 0 0 0 0 0 198 }; 199 let email: Option<String> = input 200 .email

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

+7

src/api/identity/did.rs

··· 582 ) 583 .into_response(); 584 } 0 0 0 0 0 0 0 585 let hostname = std::env::var("PDS_HOSTNAME").unwrap_or_else(|_| "localhost".to_string()); 586 let suffix = format!(".{}", hostname); 587 let is_service_domain = crate::handle::is_service_domain_handle(new_handle, &hostname);

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

+5 -3

src/api/repo/record/batch.rs

··· 1 - use super::validation::validate_record; 2 use super::write::has_verified_comms_channel; 3 use crate::api::repo::record::utils::{CommitParams, RecordOp, commit_and_log, extract_blob_cids}; 4 use crate::delegation::{self, DelegationActionType}; ··· 304 value, 305 } => { 306 if input.validate.unwrap_or(true) 307 - && let Err(err_response) = validate_record(value, collection) 0 308 { 309 return *err_response; 310 } ··· 357 value, 358 } => { 359 if input.validate.unwrap_or(true) 360 - && let Err(err_response) = validate_record(value, collection) 0 361 { 362 return *err_response; 363 }

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

+13 -1

src/api/repo/record/validation.rs

··· 7 use serde_json::json; 8 9 pub fn validate_record(record: &serde_json::Value, collection: &str) -> Result<(), Box<Response>> { 0 0 0 0 0 0 0 0 10 let validator = RecordValidator::new(); 11 - match validator.validate(record, collection) { 12 Ok(_) => Ok(()), 13 Err(ValidationError::MissingType) => Err(Box::new(( 14 StatusCode::BAD_REQUEST, ··· 29 Err(ValidationError::InvalidDatetime { path }) => Err(Box::new(( 30 StatusCode::BAD_REQUEST, 31 Json(json!({"error": "InvalidRecord", "message": format!("Invalid datetime format at '{}'", path)})), 0 0 0 0 32 ).into_response())), 33 Err(e) => Err(Box::new(( 34 StatusCode::BAD_REQUEST,

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

+5 -3

src/api/repo/record/write.rs

··· 1 - use super::validation::validate_record; 2 use crate::api::repo::record::utils::{CommitParams, RecordOp, commit_and_log, extract_blob_cids}; 3 use crate::delegation::{self, DelegationActionType}; 4 use crate::repo::tracking::TrackingBlockStore; ··· 257 } 258 }; 259 if input.validate.unwrap_or(true) 260 - && let Err(err_response) = validate_record(&input.record, &input.collection) 0 261 { 262 return *err_response; 263 } ··· 480 }; 481 let key = format!("{}/{}", collection_nsid, input.rkey); 482 if input.validate.unwrap_or(true) 483 - && let Err(err_response) = validate_record(&input.record, &input.collection) 0 484 { 485 return *err_response; 486 }

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

+6

src/api/validation.rs

··· 16 StartsWithInvalidChar, 17 EndsWithInvalidChar, 18 ContainsSpaces, 0 19 } 20 21 impl std::fmt::Display for HandleValidationError { ··· 41 } 42 Self::EndsWithInvalidChar => write!(f, "Handle cannot end with a hyphen or underscore"), 43 Self::ContainsSpaces => write!(f, "Handle cannot contain spaces"), 0 44 } 45 } 46 } ··· 80 if !c.is_ascii_alphanumeric() && c != '-' && c != '_' { 81 return Err(HandleValidationError::InvalidCharacters); 82 } 0 0 0 0 83 } 84 85 Ok(handle.to_lowercase())

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

+1

src/lib.rs

··· 10 pub mod handle; 11 pub mod image; 12 pub mod metrics; 0 13 pub mod oauth; 14 pub mod plc; 15 pub mod rate_limit;

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

+262

src/moderation/mod.rs

··· 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

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

+127 -22

src/validation/mod.rs

··· 17 InvalidRecord(String), 18 #[error("Unknown record type: {0}")] 19 UnknownType(String), 0 0 20 } 21 22 #[derive(Debug, Clone, Copy, PartialEq, Eq)] ··· 53 record: &Value, 54 collection: &str, 55 ) -> Result<ValidationStatus, ValidationError> { 0 0 0 0 0 0 0 0 0 56 let obj = record.as_object().ok_or_else(|| { 57 ValidationError::InvalidRecord("Record must be an object".to_string()) 58 })?; ··· 78 "app.bsky.graph.block" => self.validate_block(obj)?, 79 "app.bsky.graph.list" => self.validate_list(obj)?, 80 "app.bsky.graph.listitem" => self.validate_list_item(obj)?, 81 - "app.bsky.feed.generator" => self.validate_feed_generator(obj)?, 82 "app.bsky.feed.threadgate" => self.validate_threadgate(obj)?, 83 "app.bsky.labeler.service" => self.validate_labeler_service(obj)?, 0 84 _ => { 85 if self.require_lexicon { 86 return Err(ValidationError::UnknownType(record_type.to_string())); ··· 126 }); 127 } 128 for (i, tag) in tags.iter().enumerate() { 129 - if let Some(tag_str) = tag.as_str() 130 - && tag_str.len() > 640 131 - { 132 - return Err(ValidationError::InvalidField { 133 - path: format!("tags/{}", i), 134 - message: "Tag exceeds maximum length of 640 bytes".to_string(), 135 - }); 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 136 } 137 } 138 } ··· 154 ), 155 }); 156 } 0 0 0 0 0 157 } 158 if let Some(description) = obj.get("description").and_then(|v| v.as_str()) { 159 let grapheme_count = description.chars().count(); ··· 164 "Description exceeds maximum length of 2560 characters (got {})", 165 grapheme_count 166 ), 0 0 0 0 0 167 }); 168 } 169 } ··· 238 if !obj.contains_key("createdAt") { 239 return Err(ValidationError::MissingField("createdAt".to_string())); 240 } 241 - if let Some(name) = obj.get("name").and_then(|v| v.as_str()) 242 - && (name.is_empty() || name.len() > 64) 243 - { 244 - return Err(ValidationError::InvalidField { 245 - path: "name".to_string(), 246 - message: "Name must be 1-64 characters".to_string(), 247 - }); 0 0 0 0 0 248 } 249 Ok(()) 250 } ··· 268 fn validate_feed_generator( 269 &self, 270 obj: &serde_json::Map<String, Value>, 0 271 ) -> Result<(), ValidationError> { 272 if !obj.contains_key("did") { 273 return Err(ValidationError::MissingField("did".to_string())); ··· 278 if !obj.contains_key("createdAt") { 279 return Err(ValidationError::MissingField("createdAt".to_string())); 280 } 281 - if let Some(display_name) = obj.get("displayName").and_then(|v| v.as_str()) 282 - && (display_name.is_empty() || display_name.len() > 240) 283 - { 284 - return Err(ValidationError::InvalidField { 285 - path: "displayName".to_string(), 286 - message: "displayName must be 1-240 characters".to_string(), 287 - }); 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 288 } 289 Ok(()) 290 }

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

+196

tests/banned_words.rs

··· 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

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