| 1 | //! Structural, offline-by-default diagnostics shared by doctor renderers. |
| 2 | |
| 3 | use std::path::{Path, PathBuf}; |
| 4 | |
| 5 | use anyhow::{Context, Result}; |
| 6 | use serde::Serialize; |
| 7 | |
| 8 | /// Canonical user-scoped paths reported by both human and JSON doctor output. |
| 9 | /// |
| 10 | /// Resolution is read-only: this type does not construct managers, create |
| 11 | /// directories, or trigger legacy migration. |
| 12 | #[derive(Debug, Clone, PartialEq, Eq, Serialize)] |
| 13 | pub(crate) struct DoctorPathReport { |
| 14 | pub(crate) home: PathBuf, |
| 15 | pub(crate) config: PathBuf, |
| 16 | pub(crate) settings: PathBuf, |
| 17 | pub(crate) sessions: PathBuf, |
| 18 | pub(crate) logs: PathBuf, |
| 19 | pub(crate) automations: PathBuf, |
| 20 | pub(crate) task_manager_root: PathBuf, |
| 21 | pub(crate) task_manager_tasks: PathBuf, |
| 22 | pub(crate) task_manager_artifacts: PathBuf, |
| 23 | pub(crate) runtime_store: PathBuf, |
| 24 | pub(crate) runtime_events: PathBuf, |
| 25 | pub(crate) personal_fleet_definitions: PathBuf, |
| 26 | pub(crate) personal_fleet_agents: PathBuf, |
| 27 | pub(crate) secrets: PathBuf, |
| 28 | } |
| 29 | |
| 30 | impl DoctorPathReport { |
| 31 | pub(crate) fn resolve(config_override: Option<&Path>) -> Result<Self> { |
| 32 | let home = codewhale_paths::codewhale_home() |
| 33 | .map_err(anyhow::Error::new)? |
| 34 | .context("could not resolve the canonical Codewhale state root")?; |
| 35 | let config = match config_override { |
| 36 | Some(path) => codewhale_config::resolve_config_path(Some(path.to_path_buf())) |
| 37 | .context("could not normalize the explicit config path")?, |
| 38 | None => codewhale_config::resolve_config_path(None) |
| 39 | .unwrap_or_else(|_| home.join(codewhale_config::CONFIG_FILE_NAME)), |
| 40 | }; |
| 41 | let settings = crate::settings::Settings::path() |
| 42 | .context("could not resolve the canonical settings path")?; |
| 43 | let sessions = codewhale_config::resolve_state_dir("sessions") |
| 44 | .context("could not resolve the sessions path")?; |
| 45 | let logs = crate::runtime_log::log_directory() |
| 46 | .context("could not resolve the runtime log directory")?; |
| 47 | let automations = crate::automation_manager::default_automations_dir(); |
| 48 | let task_manager_root = crate::task_manager::default_tasks_dir(); |
| 49 | let task_manager_tasks = task_manager_root.join("tasks"); |
| 50 | let task_manager_artifacts = task_manager_root.join("artifacts"); |
| 51 | let runtime_config = crate::runtime_threads::RuntimeThreadManagerConfig::from_task_data_dir( |
| 52 | task_manager_root.clone(), |
| 53 | ); |
| 54 | let runtime_store = runtime_config.data_dir; |
| 55 | let runtime_events = runtime_store.join("events"); |
| 56 | let personal_fleet_definitions = crate::fleet::exact::personal_fleet_definitions_dir() |
| 57 | .context("could not resolve the personal Fleet definitions directory")?; |
| 58 | let personal_fleet_agents = crate::fleet::profile::personal_agent_profile_dir() |
| 59 | .context("could not resolve the personal Fleet agent directory")?; |
| 60 | let (secrets, _) = codewhale_secrets::FileKeyringStore::default_paths_read_only() |
| 61 | .context("could not resolve the file secret backend path")?; |
| 62 | Ok(Self { |
| 63 | home, |
| 64 | config, |
| 65 | settings, |
| 66 | sessions, |
| 67 | logs, |
| 68 | automations, |
| 69 | task_manager_root, |
| 70 | task_manager_tasks, |
| 71 | task_manager_artifacts, |
| 72 | runtime_store, |
| 73 | runtime_events, |
| 74 | personal_fleet_definitions, |
| 75 | personal_fleet_agents, |
| 76 | secrets, |
| 77 | }) |
| 78 | } |
| 79 | |
| 80 | pub(crate) fn entries(&self) -> [(&'static str, &Path); 14] { |
| 81 | [ |
| 82 | ("home", self.home.as_path()), |
| 83 | ("config", self.config.as_path()), |
| 84 | ("settings", self.settings.as_path()), |
| 85 | ("sessions", self.sessions.as_path()), |
| 86 | ("logs", self.logs.as_path()), |
| 87 | ("automations", self.automations.as_path()), |
| 88 | ("task_manager_root", self.task_manager_root.as_path()), |
| 89 | ("task_manager_tasks", self.task_manager_tasks.as_path()), |
| 90 | ( |
| 91 | "task_manager_artifacts", |
| 92 | self.task_manager_artifacts.as_path(), |
| 93 | ), |
| 94 | ("runtime_store", self.runtime_store.as_path()), |
| 95 | ("runtime_events", self.runtime_events.as_path()), |
| 96 | ( |
| 97 | "personal_fleet_definitions", |
| 98 | self.personal_fleet_definitions.as_path(), |
| 99 | ), |
| 100 | ( |
| 101 | "personal_fleet_agents", |
| 102 | self.personal_fleet_agents.as_path(), |
| 103 | ), |
| 104 | ("secrets", self.secrets.as_path()), |
| 105 | ] |
| 106 | } |
| 107 | } |
| 108 | |
| 109 | /// Render structural secret-backend facts for the human doctor report. |
| 110 | /// |
| 111 | /// The input type cannot carry secret values, keeping this renderer safe by |
| 112 | /// construction. |
| 113 | pub(crate) fn secret_backend_human_lines( |
| 114 | diagnostic: &codewhale_secrets::SecretBackendDiagnostic, |
| 115 | ) -> Vec<String> { |
| 116 | use codewhale_secrets::{ |
| 117 | SecretBackendDiagnosticKind, SecretBackendInspection, SecretBackendPresence, |
| 118 | }; |
| 119 | |
| 120 | let presence = |value| match value { |
| 121 | SecretBackendPresence::Present => "present", |
| 122 | SecretBackendPresence::Absent => "absent", |
| 123 | SecretBackendPresence::Unknown => "unknown", |
| 124 | }; |
| 125 | let inspection = match diagnostic.inspection { |
| 126 | SecretBackendInspection::MetadataOnly => "metadata_only", |
| 127 | SecretBackendInspection::NotProbed => "not_probed", |
| 128 | }; |
| 129 | let mut lines = match diagnostic.backend { |
| 130 | SecretBackendDiagnosticKind::File => vec![ |
| 131 | "backend: file".to_string(), |
| 132 | format!("presence: {} ({inspection})", presence(diagnostic.presence)), |
| 133 | ], |
| 134 | SecretBackendDiagnosticKind::System => vec![ |
| 135 | "backend: system".to_string(), |
| 136 | "status: unknown (not_probed)".to_string(), |
| 137 | ], |
| 138 | SecretBackendDiagnosticKind::Unknown => vec![ |
| 139 | "backend: unknown".to_string(), |
| 140 | "status: unknown (not_probed; unsupported configuration)".to_string(), |
| 141 | ], |
| 142 | }; |
| 143 | if let Some(path) = diagnostic.path.as_deref() { |
| 144 | lines.push(format!("path: {}", path.display())); |
| 145 | } |
| 146 | if let Some(path) = diagnostic.legacy_path.as_deref() { |
| 147 | lines.push(format!( |
| 148 | "legacy_path: {} ({}, {inspection})", |
| 149 | path.display(), |
| 150 | presence(diagnostic.legacy_presence) |
| 151 | )); |
| 152 | } |
| 153 | lines.push("No credential-store values were read or printed by this check.".to_string()); |
| 154 | lines |
| 155 | } |
| 156 | |
| 157 | /// Report key names — never values — for config entries whose value is |
| 158 | /// shaped like a bearer credential. `config.toml` is plain text, not a |
| 159 | /// secret store; doctor warns so tokens migrate to the secret backend |
| 160 | /// (morning-report issue: a plaintext OAuth token sat beside a `[redacted]` |
| 161 | /// sibling entry). |
| 162 | pub(crate) fn config_credential_shaped_keys(raw: &str) -> Vec<String> { |
| 163 | fn strong_shape(value: &str) -> bool { |
| 164 | const PREFIXES: [&str; 9] = [ |
| 165 | "sk-", |
| 166 | "sk_", |
| 167 | "xai-", |
| 168 | "ghp_", |
| 169 | "gho_", |
| 170 | "github_pat_", |
| 171 | "xoxb-", |
| 172 | "xoxp-", |
| 173 | "eyJ", |
| 174 | ]; |
| 175 | value.len() >= 20 && PREFIXES.iter().any(|prefix| value.starts_with(prefix)) |
| 176 | } |
| 177 | fn suspect_key(key: &str) -> bool { |
| 178 | let key = key.to_ascii_lowercase(); |
| 179 | [ |
| 180 | "token", |
| 181 | "secret", |
| 182 | "password", |
| 183 | "credential", |
| 184 | "api_key", |
| 185 | "apikey", |
| 186 | "access_key", |
| 187 | ] |
| 188 | .iter() |
| 189 | .any(|needle| key.contains(needle)) |
| 190 | } |
| 191 | fn random_shape(value: &str) -> bool { |
| 192 | value.len() >= 24 |
| 193 | && value |
| 194 | .chars() |
| 195 | .all(|ch| ch.is_ascii_alphanumeric() || matches!(ch, '-' | '_' | '.')) |
| 196 | && value.chars().any(|ch| ch.is_ascii_digit()) |
| 197 | && value.chars().any(|ch| ch.is_ascii_alphabetic()) |
| 198 | } |
| 199 | |
| 200 | let mut flagged: Vec<String> = Vec::new(); |
| 201 | for line in raw.lines() { |
| 202 | let line = line.trim(); |
| 203 | if line.starts_with('#') { |
| 204 | continue; |
| 205 | } |
| 206 | let Some((key, value)) = line.split_once('=') else { |
| 207 | continue; |
| 208 | }; |
| 209 | let key = key.trim().trim_matches('"'); |
| 210 | let value = value.trim(); |
| 211 | let Some(value) = value |
| 212 | .strip_prefix('"') |
| 213 | .and_then(|value| value.strip_suffix('"')) |
| 214 | else { |
| 215 | continue; |
| 216 | }; |
| 217 | if value.is_empty() || value.eq_ignore_ascii_case("[redacted]") { |
| 218 | continue; |
| 219 | } |
| 220 | if (strong_shape(value) || (suspect_key(key) && random_shape(value))) |
| 221 | && !flagged.iter().any(|existing| existing == key) |
| 222 | { |
| 223 | flagged.push(key.to_string()); |
| 224 | } |
| 225 | } |
| 226 | flagged |
| 227 | } |
| 228 | |
| 229 | /// Return only the non-secret network authority of a configured URL. |
| 230 | /// |
| 231 | /// Userinfo, path, query keys and values, and fragments are all omitted because |
| 232 | /// every one of those components can carry credentials. Parse failures also |
| 233 | /// omit the original input rather than echoing an attacker-controlled value. |
| 234 | pub(crate) fn structural_url_authority(url: &str) -> String { |
| 235 | let Some(parsed) = reqwest::Url::parse(url).ok() else { |
| 236 | return "unparseable (configured value omitted)".to_string(); |
| 237 | }; |
| 238 | let Some(host) = parsed.host_str() else { |
| 239 | return "unparseable (configured value omitted)".to_string(); |
| 240 | }; |
| 241 | let host = if host.contains(':') { |
| 242 | format!("[{host}]") |
| 243 | } else { |
| 244 | host.to_string() |
| 245 | }; |
| 246 | let mut authority = format!("{}://{host}", parsed.scheme()); |
| 247 | if let Some(port) = parsed.port() { |
| 248 | authority.push(':'); |
| 249 | authority.push_str(&port.to_string()); |
| 250 | } |
| 251 | authority |
| 252 | } |
| 253 | |
| 254 | /// Explicit live operations requested for one doctor invocation. |
| 255 | #[derive(Debug, Clone, Copy, Default, PartialEq, Eq)] |
| 256 | pub(crate) struct DoctorProbeRequest { |
| 257 | pub(crate) check_updates: bool, |
| 258 | pub(crate) probe_api: bool, |
| 259 | pub(crate) probe_local: bool, |
| 260 | pub(crate) probe_mcp: bool, |
| 261 | } |
| 262 | |
| 263 | impl DoctorProbeRequest { |
| 264 | /// Whether a release service may be contacted. |
| 265 | pub(crate) fn should_check_updates(self) -> bool { |
| 266 | self.check_updates |
| 267 | } |
| 268 | |
| 269 | /// Whether the configured provider endpoint may be contacted. |
| 270 | /// |
| 271 | /// Hosted and local endpoints have separate opt-ins because a local probe |
| 272 | /// can wake a desktop-managed daemon. |
| 273 | pub(crate) fn should_probe_api(self, endpoint_is_local: bool) -> bool { |
| 274 | if endpoint_is_local { |
| 275 | self.probe_local |
| 276 | } else { |
| 277 | self.probe_api |
| 278 | } |
| 279 | } |
| 280 | |
| 281 | /// Whether configured MCP processes may be started and contacted. |
| 282 | pub(crate) fn should_probe_mcp(self) -> bool { |
| 283 | self.probe_mcp |
| 284 | } |
| 285 | } |
| 286 | |
| 287 | #[derive(Debug, Clone, PartialEq, Eq)] |
| 288 | enum DoctorUpdateReport { |
| 289 | NotChecked, |
| 290 | UpdateAvailable { latest: String }, |
| 291 | UpToDate { latest: String }, |
| 292 | CurrentNewer { latest: String }, |
| 293 | ReleaseMetadataInvalid, |
| 294 | ReleaseCheckFailed, |
| 295 | } |
| 296 | |
| 297 | fn doctor_update_report<E>( |
| 298 | current_version: &str, |
| 299 | latest_result: Result<String, E>, |
| 300 | ) -> DoctorUpdateReport { |
| 301 | let Ok(raw_latest) = latest_result else { |
| 302 | return DoctorUpdateReport::ReleaseCheckFailed; |
| 303 | }; |
| 304 | let Some(latest) = doctor_safe_release_tag(&raw_latest) else { |
| 305 | return DoctorUpdateReport::ReleaseMetadataInvalid; |
| 306 | }; |
| 307 | match codewhale_release::compare_release_versions(current_version, &latest) { |
| 308 | Ok(std::cmp::Ordering::Less) => DoctorUpdateReport::UpdateAvailable { latest }, |
| 309 | Ok(std::cmp::Ordering::Equal) => DoctorUpdateReport::UpToDate { latest }, |
| 310 | Ok(std::cmp::Ordering::Greater) => DoctorUpdateReport::CurrentNewer { latest }, |
| 311 | Err(_) => DoctorUpdateReport::ReleaseMetadataInvalid, |
| 312 | } |
| 313 | } |
| 314 | |
| 315 | /// Canonicalize a release tag before any doctor renderer sees it. A release |
| 316 | /// server response is untrusted input: it may not become an error echo or a |
| 317 | /// terminal control sequence merely because the user opted into an update |
| 318 | /// check. |
| 319 | fn doctor_safe_release_tag(raw: &str) -> Option<String> { |
| 320 | let version = raw.trim().strip_prefix('v').unwrap_or(raw.trim()); |
| 321 | semver::Version::parse(version) |
| 322 | .ok() |
| 323 | .map(|version| format!("v{version}")) |
| 324 | } |
| 325 | |
| 326 | fn doctor_update_report_lines(report: &DoctorUpdateReport) -> Vec<String> { |
| 327 | match report { |
| 328 | DoctorUpdateReport::NotChecked => vec![ |
| 329 | "latest: unknown (not checked; offline default)".to_string(), |
| 330 | "Run `codewhale doctor --check-updates` to opt in.".to_string(), |
| 331 | ], |
| 332 | DoctorUpdateReport::UpdateAvailable { latest } => vec![ |
| 333 | format!("latest: {latest}"), |
| 334 | "Update available. Run `codewhale update` to install.".to_string(), |
| 335 | ], |
| 336 | DoctorUpdateReport::UpToDate { latest } => { |
| 337 | vec![ |
| 338 | format!("latest: {latest}"), |
| 339 | "Already up to date.".to_string(), |
| 340 | ] |
| 341 | } |
| 342 | DoctorUpdateReport::CurrentNewer { latest } => vec![ |
| 343 | format!("latest: {latest}"), |
| 344 | "Current build is newer than the latest published release.".to_string(), |
| 345 | ], |
| 346 | DoctorUpdateReport::ReleaseMetadataInvalid => vec![ |
| 347 | "latest: unknown (release metadata invalid; details omitted)".to_string(), |
| 348 | "Run `codewhale update --check` to retry.".to_string(), |
| 349 | ], |
| 350 | DoctorUpdateReport::ReleaseCheckFailed => vec![ |
| 351 | "latest: unknown (release check failed; details omitted)".to_string(), |
| 352 | "Run `codewhale update --check` to retry.".to_string(), |
| 353 | ], |
| 354 | } |
| 355 | } |
| 356 | |
| 357 | /// Print the update portion of the human doctor report. |
| 358 | /// |
| 359 | /// The release service is contacted only when `--check-updates` populated the |
| 360 | /// explicit request bit. The default branch returns before constructing an |
| 361 | /// HTTP request. Failure details are deliberately typed and generic because |
| 362 | /// transport errors and release metadata are untrusted strings. |
| 363 | pub(crate) async fn print_update_report(probes: DoctorProbeRequest) { |
| 364 | let current_version = env!("CARGO_PKG_VERSION"); |
| 365 | println!(" · current: v{current_version}"); |
| 366 | let report = if probes.should_check_updates() { |
| 367 | doctor_update_report( |
| 368 | current_version, |
| 369 | codewhale_release::latest_release_tag_async(codewhale_release::ReleaseChannel::Stable) |
| 370 | .await, |
| 371 | ) |
| 372 | } else { |
| 373 | DoctorUpdateReport::NotChecked |
| 374 | }; |
| 375 | for (index, line) in doctor_update_report_lines(&report).into_iter().enumerate() { |
| 376 | let indent = if index == 0 { " ·" } else { " " }; |
| 377 | println!("{indent} {line}"); |
| 378 | } |
| 379 | } |
| 380 | |
| 381 | #[cfg(test)] |
| 382 | #[path = "doctor/tests.rs"] |
| 383 | mod tests; |
| 384 |