| 1 | //! `/structcopy` command — human-only structural copy (#2033). |
| 2 | //! |
| 3 | //! Copies exactly one bounded, human-selected session object (one transcript |
| 4 | //! item, one tool call+result pair, the current plan snapshot, or one |
| 5 | //! existing Workflow run projection) as deterministic, versioned canonical |
| 6 | //! JSON with a top-level receipt. The default target is the clipboard; an |
| 7 | //! explicit `stdout` argument is the only text-view path. |
| 8 | //! |
| 9 | //! Contract: |
| 10 | //! - Human-only. This is a slash command, never a model-visible tool, event, |
| 11 | //! or authority, and it writes nothing back into App/session/plan/workflow |
| 12 | //! state (see the registry/catalog contract test). |
| 13 | //! - Read-only projection over existing state. Redaction reuses the shared |
| 14 | //! sanitizer seams in `codewhale_sanitize::sanitize` (`redact_json` for |
| 15 | //! values, `sanitize_text` for keys and status labels, which |
| 16 | //! `redact_json` does not reach) plus a strict pass that strips URL |
| 17 | //! userinfo/query/fragment entirely and folds the workspace and home |
| 18 | //! prefixes to labels, removes other absolute paths, and handles generic |
| 19 | //! authority URLs. The workflow object reuses the bounded |
| 20 | //! `WorkflowRunSummary` projection. |
| 21 | //! - Hard caps on final encoded bytes, array items, string bytes, object key |
| 22 | //! bytes, and nesting depth; grapheme-safe truncation; recursively sorted |
| 23 | //! keys; exact full-tree original counts and exact retained counts in the |
| 24 | //! receipt. If receipt metadata alone cannot fit the byte cap, the command |
| 25 | //! fails closed and emits nothing. |
| 26 | //! |
| 27 | //! What this deliberately does **not** claim: |
| 28 | //! - It is not a general PII scrubber. Workspace/home paths retain a useful |
| 29 | //! labelled suffix; other absolute POSIX, drive-letter, and UNC paths are |
| 30 | //! replaced outright. |
| 31 | //! - Redaction is pattern-based (the shared sanitizer's private-key/bearer/ |
| 32 | //! JWT/URL/secret regexes plus this module's strict URL pass). A secret that |
| 33 | //! matches none of those patterns and sits under a non-sensitive key is |
| 34 | //! copied as-is. |
| 35 | //! - Delivery to the clipboard is not confirmed. Terminal-client transports |
| 36 | //! (tmux / OSC 52) are queued on a background writer; the receipt says |
| 37 | //! "queued", not "delivered". |
| 38 | |
| 39 | use std::collections::{BTreeMap, BTreeSet}; |
| 40 | use std::fmt::Write as FmtWrite; |
| 41 | |
| 42 | use serde_json::{Value, json}; |
| 43 | use unicode_segmentation::UnicodeSegmentation; |
| 44 | |
| 45 | use codewhale_command_contract::facets::*; |
| 46 | use codewhale_command_contract::handler::{CommandCapabilities, CommandContexts, CommandHandler}; |
| 47 | use codewhale_command_contract::metadata::{CommandInfo, RegisterCommand}; |
| 48 | use codewhale_command_contract::outcome::StructcopyCommandResult as CommandResult; |
| 49 | use codewhale_sanitize::sanitize::{is_sensitive_key, redact_json, sanitize_text}; |
| 50 | |
| 51 | pub(in crate::commands) const COMMAND_INFO: CommandInfo = CommandInfo { |
| 52 | name: "structcopy", |
| 53 | aliases: &[], |
| 54 | usage: "/structcopy <turn <n>|tool <call-id>|plan|workflow <run-id>> [stdout]", |
| 55 | description_key: "cmd_structcopy_description", |
| 56 | }; |
| 57 | pub(in crate::commands) const CAPABILITIES: CommandCapabilities = |
| 58 | CommandCapabilities::SESSION_STRUCTCOPY.union(CommandCapabilities::PRESENTATION); |
| 59 | |
| 60 | pub(in crate::commands) struct StructcopyCmd; |
| 61 | impl RegisterCommand<CommandResult> for StructcopyCmd { |
| 62 | fn info() -> &'static CommandInfo { |
| 63 | &COMMAND_INFO |
| 64 | } |
| 65 | fn handler() -> CommandHandler<CommandResult> { |
| 66 | CommandHandler::Contextual { |
| 67 | capabilities: CAPABILITIES, |
| 68 | handler: execute_structcopy, |
| 69 | } |
| 70 | } |
| 71 | } |
| 72 | |
| 73 | /// Versioned envelope identity carried in every receipt. |
| 74 | pub(in crate::commands) const SCHEMA_ID: &str = "codewhale/structcopy/v1"; |
| 75 | /// Redaction contract label so consumers can tell which seams ran. |
| 76 | pub(in crate::commands) const REDACTION_CONTRACT: &str = |
| 77 | "export-sanitize/v1+typed-markers/v1+strict-url/v2+path-redact/v2"; |
| 78 | /// Marker substituted for subtrees cut by the depth cap. Structural markers |
| 79 | /// are inserted after bounding and are intentionally exempt from |
| 80 | /// `max_string_bytes`; they are still counted as retained bytes. |
| 81 | pub(in crate::commands) const DEPTH_OMISSION_MARKER: &str = "omitted:depth_cap"; |
| 82 | /// Marker substituted for a URL token that cannot be parsed and therefore |
| 83 | /// cannot be proven free of userinfo/query/fragment. Fail closed. |
| 84 | pub(in crate::commands) const URL_OMISSION_MARKER: &str = "redacted:url"; |
| 85 | /// Marker substituted for an absolute filesystem path outside the labelled |
| 86 | /// workspace/home roots. Paths are privacy-bearing even when they contain no |
| 87 | /// conventional secret token. |
| 88 | pub(in crate::commands) const PATH_OMISSION_MARKER: &str = "redacted:absolute_path"; |
| 89 | pub(in crate::commands) const BEARER_REDACTION_MARKER: &str = "redacted:bearer"; |
| 90 | pub(in crate::commands) const SENSITIVE_VALUE_REDACTION_MARKER: &str = "redacted:sensitive_value"; |
| 91 | |
| 92 | /// Selectors are echoed into the receipt and into status messages, so they |
| 93 | /// get their own tight cap independent of the payload string cap. |
| 94 | pub(in crate::commands) const MAX_SELECTOR_BYTES: usize = 256; |
| 95 | /// Hard caps enforced on every emitted artifact. The byte cap stays well |
| 96 | /// under the OSC 52 clipboard ceiling (100 KiB) so the default clipboard |
| 97 | /// target always fits its weakest transport. |
| 98 | #[derive(Debug, Clone, Copy, PartialEq, Eq)] |
| 99 | pub(in crate::commands) struct Caps { |
| 100 | pub(in crate::commands) max_output_bytes: usize, |
| 101 | pub(in crate::commands) max_array_items: usize, |
| 102 | pub(in crate::commands) max_string_bytes: usize, |
| 103 | pub(in crate::commands) max_depth: usize, |
| 104 | } |
| 105 | |
| 106 | pub(in crate::commands) const DEFAULT_CAPS: Caps = Caps { |
| 107 | max_output_bytes: 48 * 1024, |
| 108 | max_array_items: 64, |
| 109 | max_string_bytes: 2 * 1024, |
| 110 | max_depth: 12, |
| 111 | }; |
| 112 | |
| 113 | /// Object keys are bounded separately from values: they are short by nature, |
| 114 | /// they participate in collision handling, and they are rewritten once during |
| 115 | /// redaction rather than per byte-cap retry. |
| 116 | pub(in crate::commands) const MAX_KEY_BYTES: usize = 256; |
| 117 | |
| 118 | #[derive(Debug, Clone, PartialEq, Eq)] |
| 119 | pub(in crate::commands) enum CopyKind { |
| 120 | Turn(usize), |
| 121 | Tool(String), |
| 122 | Plan, |
| 123 | Workflow(String), |
| 124 | } |
| 125 | |
| 126 | impl CopyKind { |
| 127 | fn display_label( |
| 128 | &self, |
| 129 | presentation: &dyn CommandPresentationContext, |
| 130 | ) -> Result<String, String> { |
| 131 | let key = match self { |
| 132 | CopyKind::Turn(_) => "cmd_structcopy_kind_turn", |
| 133 | CopyKind::Tool(_) => "cmd_structcopy_kind_tool", |
| 134 | CopyKind::Plan => "cmd_structcopy_kind_plan", |
| 135 | CopyKind::Workflow(_) => "cmd_structcopy_kind_workflow", |
| 136 | }; |
| 137 | presentation.translate(key, &[]) |
| 138 | } |
| 139 | } |
| 140 | |
| 141 | #[derive(Debug, Clone, PartialEq, Eq)] |
| 142 | struct CopyRequest { |
| 143 | kind: CopyKind, |
| 144 | stdout: bool, |
| 145 | } |
| 146 | |
| 147 | pub(in crate::commands) fn execute_structcopy( |
| 148 | contexts: CommandContexts<'_>, |
| 149 | arg: Option<&str>, |
| 150 | ) -> CommandResult { |
| 151 | let parts = contexts.into_parts(); |
| 152 | let Some(copy) = parts.structcopy else { |
| 153 | return CommandResult::error("Command capability unavailable: session_structcopy"); |
| 154 | }; |
| 155 | let Some(presentation) = parts.presentation else { |
| 156 | return CommandResult::error("Command capability unavailable: presentation"); |
| 157 | }; |
| 158 | execute_portable(copy, presentation, arg, &DEFAULT_CAPS).unwrap_or_else(CommandResult::error) |
| 159 | } |
| 160 | |
| 161 | fn execute_portable( |
| 162 | copy: &dyn CommandSessionStructcopyContext, |
| 163 | presentation: &dyn CommandPresentationContext, |
| 164 | arg: Option<&str>, |
| 165 | caps: &Caps, |
| 166 | ) -> Result<CommandResult, String> { |
| 167 | let request = match parse_request(arg) { |
| 168 | Ok(request) => request, |
| 169 | Err(()) => { |
| 170 | return Err(presentation.translate( |
| 171 | "cmd_structcopy_usage_error", |
| 172 | &[("usage", COMMAND_INFO.usage)], |
| 173 | )?); |
| 174 | } |
| 175 | }; |
| 176 | let label = request.kind.display_label(presentation)?; |
| 177 | let json = render_copy(copy, presentation, &request.kind, caps)?; |
| 178 | if request.stdout { |
| 179 | return Ok(CommandResult::message(json)); |
| 180 | } |
| 181 | let bytes = json.len().to_string(); |
| 182 | match copy.write_clipboard(&json) { |
| 183 | Ok(transport) => { |
| 184 | let key = match transport { |
| 185 | StructcopyTransport::Native => "cmd_structcopy_clipboard_accepted", |
| 186 | StructcopyTransport::TerminalQueued => "cmd_structcopy_clipboard_queued", |
| 187 | }; |
| 188 | Ok(CommandResult::message( |
| 189 | presentation.translate(key, &[("kind", &label), ("bytes", &bytes)])?, |
| 190 | )) |
| 191 | } |
| 192 | Err(error) => { |
| 193 | Err(presentation.translate("cmd_structcopy_clipboard_failed", &[("error", &error)])?) |
| 194 | } |
| 195 | } |
| 196 | } |
| 197 | |
| 198 | fn parse_request(arg: Option<&str>) -> Result<CopyRequest, ()> { |
| 199 | let raw = arg.unwrap_or("").trim(); |
| 200 | if raw.is_empty() { |
| 201 | return Err(()); |
| 202 | } |
| 203 | let mut tokens: Vec<&str> = raw.split_whitespace().collect(); |
| 204 | let mut stdout = false; |
| 205 | if tokens |
| 206 | .last() |
| 207 | .is_some_and(|last| last.eq_ignore_ascii_case("stdout")) |
| 208 | { |
| 209 | stdout = true; |
| 210 | tokens.pop(); |
| 211 | } |
| 212 | let kind = match tokens.as_slice() { |
| 213 | ["plan"] => CopyKind::Plan, |
| 214 | ["turn", index] => { |
| 215 | let index = index |
| 216 | .parse::<usize>() |
| 217 | .ok() |
| 218 | .filter(|index| *index >= 1) |
| 219 | .ok_or(())?; |
| 220 | CopyKind::Turn(index) |
| 221 | } |
| 222 | ["tool", call_id] => CopyKind::Tool((*call_id).to_string()), |
| 223 | ["workflow", run_id] => CopyKind::Workflow((*run_id).to_string()), |
| 224 | _ => return Err(()), |
| 225 | }; |
| 226 | Ok(CopyRequest { kind, stdout }) |
| 227 | } |
| 228 | |
| 229 | // === Object selection (read-only; unavailable objects are reported, never |
| 230 | // fabricated) === |
| 231 | |
| 232 | fn build_payload( |
| 233 | copy: &dyn CommandSessionStructcopyContext, |
| 234 | kind: &CopyKind, |
| 235 | ) -> Result<(&'static str, Value, Value), StructcopyError> { |
| 236 | match kind { |
| 237 | CopyKind::Turn(index) => Ok(( |
| 238 | "turn", |
| 239 | json!(index), |
| 240 | message_payload(©.transcript_item(*index)?), |
| 241 | )), |
| 242 | CopyKind::Tool(call_id) => { |
| 243 | let pair = copy.tool_pair(call_id)?; |
| 244 | let result = match pair.result { |
| 245 | Some(result) => { |
| 246 | json!({"found":true,"is_error":optional_bool(result.is_error),"content":result.content,"content_blocks":result.content_blocks}) |
| 247 | } |
| 248 | None => json!({"found":false}), |
| 249 | }; |
| 250 | Ok(( |
| 251 | "tool", |
| 252 | json!(call_id), |
| 253 | json!({"call_id":call_id,"name":pair.name,"input":pair.input,"result":result}), |
| 254 | )) |
| 255 | } |
| 256 | CopyKind::Plan => Ok(( |
| 257 | "plan", |
| 258 | Value::Null, |
| 259 | serde_json::to_value(copy.plan_snapshot()?) |
| 260 | .map_err(|error| StructcopyError::Preparation(error.to_string()))?, |
| 261 | )), |
| 262 | CopyKind::Workflow(run_id) => Ok(( |
| 263 | "workflow", |
| 264 | json!(run_id), |
| 265 | serde_json::to_value(copy.workflow_projection(run_id)?) |
| 266 | .map_err(|error| StructcopyError::Preparation(error.to_string()))?, |
| 267 | )), |
| 268 | } |
| 269 | } |
| 270 | |
| 271 | fn message_payload(message: &StructcopyTranscript) -> Value { |
| 272 | match &message.content { |
| 273 | StructcopyContent::InternalContext => { |
| 274 | json!({"index":message.index,"role":message.role,"omission_code":"internal_context"}) |
| 275 | } |
| 276 | StructcopyContent::Visible(blocks) => { |
| 277 | json!({"index":message.index,"role":message.role,"content":blocks.iter().map(block_payload).collect::<Vec<_>>()}) |
| 278 | } |
| 279 | } |
| 280 | } |
| 281 | |
| 282 | pub(in crate::commands) fn optional_bool(value: Option<bool>) -> Value { |
| 283 | value.map_or(Value::Null, Value::Bool) |
| 284 | } |
| 285 | |
| 286 | fn block_payload(block: &StructcopyBlock) -> Value { |
| 287 | match block { |
| 288 | StructcopyBlock::Text(text) => json!({"type":"text","text":text}), |
| 289 | StructcopyBlock::ThinkingOmitted => { |
| 290 | json!({"type":"thinking","omission_code":"internal_reasoning_and_signature"}) |
| 291 | } |
| 292 | StructcopyBlock::ToolUse { |
| 293 | id, |
| 294 | name, |
| 295 | input, |
| 296 | caller_type, |
| 297 | } => json!({"type":"tool_use","id":id,"caller_type":caller_type,"name":name,"input":input}), |
| 298 | StructcopyBlock::ToolResult { |
| 299 | tool_use_id, |
| 300 | result, |
| 301 | } => { |
| 302 | json!({"type":"tool_result","tool_use_id":tool_use_id,"is_error":optional_bool(result.is_error),"content":result.content,"content_blocks":result.content_blocks}) |
| 303 | } |
| 304 | StructcopyBlock::ImageUrl(url) => json!({"type":"image","url":url}), |
| 305 | StructcopyBlock::ImageOmitted => { |
| 306 | json!({"type":"image","omission_code":"inline_or_local_image_payload"}) |
| 307 | } |
| 308 | StructcopyBlock::ServerToolUse { id, name, input } => { |
| 309 | json!({"type":"server_tool_use","id":id,"name":name,"input":input}) |
| 310 | } |
| 311 | StructcopyBlock::ToolSearchToolResult { |
| 312 | tool_use_id, |
| 313 | content, |
| 314 | } => json!({"type":"tool_search_tool_result","tool_use_id":tool_use_id,"content":content}), |
| 315 | StructcopyBlock::CodeExecutionToolResult { |
| 316 | tool_use_id, |
| 317 | content, |
| 318 | } => { |
| 319 | json!({"type":"code_execution_tool_result","tool_use_id":tool_use_id,"content":content}) |
| 320 | } |
| 321 | } |
| 322 | } |
| 323 | |
| 324 | fn selection_error( |
| 325 | presentation: &dyn CommandPresentationContext, |
| 326 | kind: &CopyKind, |
| 327 | error: StructcopyError, |
| 328 | ) -> Result<String, String> { |
| 329 | let label = kind.display_label(presentation)?; |
| 330 | match error { |
| 331 | StructcopyError::Unavailable => { |
| 332 | presentation.translate("cmd_structcopy_unavailable", &[("kind", &label)]) |
| 333 | } |
| 334 | StructcopyError::Busy => presentation.translate("cmd_structcopy_busy", &[("kind", &label)]), |
| 335 | StructcopyError::Preparation(error) => presentation.translate( |
| 336 | "cmd_structcopy_prepare_failed", |
| 337 | &[("kind", &label), ("error", &error)], |
| 338 | ), |
| 339 | } |
| 340 | } |
| 341 | |
| 342 | // === Redaction (composed from existing central seams) === |
| 343 | |
| 344 | /// The strongest existing central redaction, applied before any bounding or |
| 345 | /// serialization and after key normalization. |
| 346 | /// |
| 347 | /// [`redact_json`] replaces values under secret-shaped keys, and runs |
| 348 | /// [`sanitize_text`] over every string *value* — stripping ANSI/control |
| 349 | /// bytes and masking PEM blocks, `Bearer` tokens, JWTs, credential-bearing |
| 350 | /// URLs, and the config layer's known secret patterns. It does **not** touch |
| 351 | /// object *keys*, so this pass runs [`sanitize_text`] over keys as well, |
| 352 | /// then folds workspace/home prefixes to labels and strips URL |
| 353 | /// userinfo/query/fragment outright. |
| 354 | /// |
| 355 | /// Keys are also sorted, bounded, and de-collided here. Original and retained |
| 356 | /// key counts are kept separately so omitted subtrees cannot inflate claims |
| 357 | /// about the emitted object. |
| 358 | fn redact_payload(value: &mut Value, labels: &PathLabels, keys: &mut KeyStats) { |
| 359 | // Normalize keys first so ANSI/control obfuscation cannot hide a |
| 360 | // sensitive-key hint from classification. `strict_strings` classifies |
| 361 | // both the original and normalized key; the shared export pass then runs |
| 362 | // over the normalized tree as defense in depth. |
| 363 | let mut path = Vec::new(); |
| 364 | strict_strings(value, labels, keys, &mut path); |
| 365 | redact_json(value, None); |
| 366 | normalize_redaction_codes(value); |
| 367 | } |
| 368 | |
| 369 | /// Prefix folding for useful filesystem paths. These prefixes are recognised: |
| 370 | /// the workspace root (both as configured and as canonicalized, which differ |
| 371 | /// on macOS where `/var` symlinks to `/private/var`) and `$HOME` / |
| 372 | /// `%USERPROFILE%`. The later strict pass removes every remaining absolute |
| 373 | /// POSIX, drive-letter, or UNC path. |
| 374 | pub(in crate::commands) struct PathLabels { |
| 375 | /// `(prefix, label)` sorted longest-first so that a workspace nested |
| 376 | /// inside `$HOME` folds to `<workspace>` rather than `<home>/…`. |
| 377 | pub(in crate::commands) labels: Vec<(String, &'static str)>, |
| 378 | } |
| 379 | |
| 380 | impl PathLabels { |
| 381 | pub(in crate::commands) fn new(roots: &StructcopyPathRoots) -> Self { |
| 382 | let mut workspace_forms: Vec<String> = Vec::new(); |
| 383 | for form in std::iter::once(&roots.workspace).chain(roots.canonical_workspace.iter()) { |
| 384 | if form.len() > 1 && !workspace_forms.contains(form) { |
| 385 | workspace_forms.push(form.clone()); |
| 386 | } |
| 387 | } |
| 388 | let mut labels: Vec<(String, &'static str)> = workspace_forms |
| 389 | .iter() |
| 390 | .map(|form| (form.clone(), "<workspace>")) |
| 391 | .collect(); |
| 392 | if let Some(home) = &roots.home |
| 393 | && home.len() > 3 |
| 394 | && !workspace_forms.contains(home) |
| 395 | { |
| 396 | labels.push((home.clone(), "<home>")); |
| 397 | } |
| 398 | labels.sort_by(|left, right| { |
| 399 | right |
| 400 | .0 |
| 401 | .len() |
| 402 | .cmp(&left.0.len()) |
| 403 | .then_with(|| left.0.cmp(&right.0)) |
| 404 | }); |
| 405 | Self { labels } |
| 406 | } |
| 407 | |
| 408 | pub(in crate::commands) fn apply(&self, text: &str) -> String { |
| 409 | let mut out = text.to_string(); |
| 410 | for (prefix, label) in &self.labels { |
| 411 | out = replace_path_root(&out, prefix, label); |
| 412 | } |
| 413 | out |
| 414 | } |
| 415 | } |
| 416 | |
| 417 | /// Replace a configured root only when it ends on a path-component boundary. |
| 418 | /// A lexical prefix such as `/opt/app` must not label `/opt/application`; the |
| 419 | /// latter remains foreign and is removed by the absolute-path scrubber. |
| 420 | fn replace_path_root(text: &str, root: &str, label: &str) -> String { |
| 421 | if root.is_empty() { |
| 422 | return text.to_string(); |
| 423 | } |
| 424 | let mut out = String::with_capacity(text.len()); |
| 425 | let mut cursor = 0usize; |
| 426 | while let Some(offset) = text[cursor..].find(root) { |
| 427 | let start = cursor + offset; |
| 428 | let end = start + root.len(); |
| 429 | out.push_str(&text[cursor..start]); |
| 430 | let component_boundary = text[end..] |
| 431 | .chars() |
| 432 | .next() |
| 433 | .is_none_or(|ch| matches!(ch, '/' | '\\')); |
| 434 | if component_boundary { |
| 435 | out.push_str(label); |
| 436 | } else { |
| 437 | out.push_str(root); |
| 438 | } |
| 439 | cursor = end; |
| 440 | } |
| 441 | out.push_str(&text[cursor..]); |
| 442 | out |
| 443 | } |
| 444 | |
| 445 | /// Per-object-key accounting. Computed once during redaction and reported in |
| 446 | /// the receipt so a renamed or truncated key is never silent. |
| 447 | #[derive(Debug, Default, Clone, PartialEq, Eq)] |
| 448 | struct KeyStats { |
| 449 | entries: BTreeMap<Vec<String>, KeyFlags>, |
| 450 | } |
| 451 | |
| 452 | #[derive(Debug, Default, Clone, Copy, PartialEq, Eq)] |
| 453 | struct KeyFlags { |
| 454 | truncated: bool, |
| 455 | deduped: bool, |
| 456 | } |
| 457 | |
| 458 | #[derive(Debug, Default, Clone, Copy, PartialEq, Eq)] |
| 459 | struct RetainedKeyStats { |
| 460 | total: u64, |
| 461 | truncated: u64, |
| 462 | deduped: u64, |
| 463 | } |
| 464 | |
| 465 | impl KeyStats { |
| 466 | fn original_total(&self) -> u64 { |
| 467 | u64::try_from(self.entries.len()).unwrap_or(u64::MAX) |
| 468 | } |
| 469 | } |
| 470 | |
| 471 | fn strict_strings( |
| 472 | value: &mut Value, |
| 473 | labels: &PathLabels, |
| 474 | keys: &mut KeyStats, |
| 475 | path: &mut Vec<String>, |
| 476 | ) { |
| 477 | match value { |
| 478 | Value::String(text) => *text = scrub_string(text, labels), |
| 479 | Value::Array(items) => { |
| 480 | for (index, item) in items.iter_mut().enumerate() { |
| 481 | path.push(format!("i:{index}")); |
| 482 | strict_strings(item, labels, keys, path); |
| 483 | path.pop(); |
| 484 | } |
| 485 | } |
| 486 | Value::Object(map) => { |
| 487 | // Take the map, rewrite each key, and reinsert. Entries are |
| 488 | // processed in sorted original-key order so collision suffixes |
| 489 | // are assigned deterministically regardless of insertion order. |
| 490 | let mut entries: Vec<(String, Value)> = std::mem::take(map).into_iter().collect(); |
| 491 | entries.sort_by(|left, right| left.0.cmp(&right.0)); |
| 492 | for (key, mut item) in entries { |
| 493 | let scrubbed = flatten_ws(&scrub_string(&key, labels)); |
| 494 | let (bounded, was_truncated) = |
| 495 | truncate_string_grapheme_safe(&scrubbed, MAX_KEY_BYTES); |
| 496 | let (unique, collision_truncated) = unique_object_key(map, &bounded); |
| 497 | let sensitive = is_sensitive_key(&key) || is_sensitive_key(&scrubbed); |
| 498 | if sensitive { |
| 499 | item = Value::String("[redacted]".to_string()); |
| 500 | } else { |
| 501 | path.push(key_path_segment(&unique)); |
| 502 | strict_strings(&mut item, labels, keys, path); |
| 503 | path.pop(); |
| 504 | } |
| 505 | path.push(key_path_segment(&unique)); |
| 506 | keys.entries.insert( |
| 507 | path.clone(), |
| 508 | KeyFlags { |
| 509 | truncated: was_truncated || collision_truncated, |
| 510 | deduped: unique != bounded, |
| 511 | }, |
| 512 | ); |
| 513 | path.pop(); |
| 514 | map.insert(unique, item); |
| 515 | } |
| 516 | } |
| 517 | Value::Null | Value::Bool(_) | Value::Number(_) => {} |
| 518 | } |
| 519 | } |
| 520 | |
| 521 | fn key_path_segment(key: &str) -> String { |
| 522 | format!("k:{}:{key}", key.len()) |
| 523 | } |
| 524 | |
| 525 | fn collect_retained_key_stats( |
| 526 | value: &Value, |
| 527 | original: &KeyStats, |
| 528 | path: &mut Vec<String>, |
| 529 | retained: &mut RetainedKeyStats, |
| 530 | ) { |
| 531 | match value { |
| 532 | Value::Array(items) => { |
| 533 | for (index, item) in items.iter().enumerate() { |
| 534 | path.push(format!("i:{index}")); |
| 535 | collect_retained_key_stats(item, original, path, retained); |
| 536 | path.pop(); |
| 537 | } |
| 538 | } |
| 539 | Value::Object(map) => { |
| 540 | for (key, item) in map { |
| 541 | path.push(key_path_segment(key)); |
| 542 | if let Some(flags) = original.entries.get(path) { |
| 543 | retained.total += 1; |
| 544 | if flags.truncated { |
| 545 | retained.truncated += 1; |
| 546 | } |
| 547 | if flags.deduped { |
| 548 | retained.deduped += 1; |
| 549 | } |
| 550 | } |
| 551 | collect_retained_key_stats(item, original, path, retained); |
| 552 | path.pop(); |
| 553 | } |
| 554 | } |
| 555 | Value::Null | Value::Bool(_) | Value::Number(_) | Value::String(_) => {} |
| 556 | } |
| 557 | } |
| 558 | |
| 559 | /// Deterministic collision handling for keys that collapsed onto each other |
| 560 | /// after scrubbing or truncation. |
| 561 | /// |
| 562 | /// Termination is structural rather than hopeful: the numeric reserve is |
| 563 | /// sized for the largest suffix this call can produce, so `base` is fixed and |
| 564 | /// the `map.len() + 1` candidates `base~2 … base~(len+2)` are pairwise |
| 565 | /// distinct. A map holding `len` keys cannot occupy all of them. |
| 566 | /// |
| 567 | /// When `MAX_KEY_BYTES` is smaller than the reserve the suffix still wins: |
| 568 | /// losing a key to a silent overwrite is worse than exceeding a key cap by a |
| 569 | /// few bytes, and the per-key flags record that it happened. |
| 570 | fn unique_object_key(map: &serde_json::Map<String, Value>, requested: &str) -> (String, bool) { |
| 571 | if !map.contains_key(requested) { |
| 572 | return (requested.to_string(), false); |
| 573 | } |
| 574 | let highest = map.len().saturating_add(2); |
| 575 | let reserve = 1 + decimal_width(highest); |
| 576 | let base_cap = MAX_KEY_BYTES.saturating_sub(reserve); |
| 577 | let (base, collision_truncated) = truncate_string_grapheme_safe(requested, base_cap); |
| 578 | for index in 2..=highest { |
| 579 | let candidate = format!("{base}~{index}"); |
| 580 | if !map.contains_key(&candidate) { |
| 581 | return (candidate, collision_truncated); |
| 582 | } |
| 583 | } |
| 584 | unreachable!( |
| 585 | "map of {} keys cannot occupy {} distinct candidates", |
| 586 | map.len(), |
| 587 | highest - 1 |
| 588 | ) |
| 589 | } |
| 590 | |
| 591 | fn decimal_width(mut value: usize) -> usize { |
| 592 | let mut width = 1; |
| 593 | while value >= 10 { |
| 594 | value /= 10; |
| 595 | width += 1; |
| 596 | } |
| 597 | width |
| 598 | } |
| 599 | |
| 600 | /// Collapse every run of whitespace to a single space. Used for object keys, |
| 601 | /// where control layout is a structural hazard rather than data. |
| 602 | fn flatten_ws(text: &str) -> String { |
| 603 | text.split_whitespace().collect::<Vec<_>>().join(" ") |
| 604 | } |
| 605 | |
| 606 | pub(in crate::commands) fn scrub_string(text: &str, labels: &PathLabels) -> String { |
| 607 | // `sanitize_text` first: it strips ANSI and control bytes, so the URL |
| 608 | // scan below cannot be fooled by an escape sequence spliced into a |
| 609 | // scheme. It is idempotent, so re-running it over values that |
| 610 | // `redact_json` already sanitized is safe. |
| 611 | let sanitized = sanitize_text(text); |
| 612 | let bearer_safe = redact_loose_bearers(&sanitized); |
| 613 | let labelled = labels.apply(&bearer_safe); |
| 614 | scrub_paths(&scrub_urls(&labelled)) |
| 615 | } |
| 616 | |
| 617 | /// Convert the prose placeholders owned by the shared sanitizer into stable |
| 618 | /// language-neutral codes. Structural JSON is a machine artifact and must not |
| 619 | /// change with the UI locale. |
| 620 | fn normalize_redaction_codes(value: &mut Value) { |
| 621 | match value { |
| 622 | Value::String(text) => { |
| 623 | *text = text |
| 624 | .replace("[redacted private key]", "redacted:private_key") |
| 625 | .replace("Bearer [redacted]", BEARER_REDACTION_MARKER) |
| 626 | .replace("[redacted token]", "redacted:token") |
| 627 | .replace("[redacted]", SENSITIVE_VALUE_REDACTION_MARKER); |
| 628 | } |
| 629 | Value::Array(items) => { |
| 630 | for item in items { |
| 631 | normalize_redaction_codes(item); |
| 632 | } |
| 633 | } |
| 634 | Value::Object(map) => { |
| 635 | for item in map.values_mut() { |
| 636 | normalize_redaction_codes(item); |
| 637 | } |
| 638 | } |
| 639 | Value::Null | Value::Bool(_) | Value::Number(_) => {} |
| 640 | } |
| 641 | } |
| 642 | |
| 643 | fn redact_loose_bearers(text: &str) -> String { |
| 644 | // Selectors cannot carry the whitespace used by a conventional |
| 645 | // `Bearer <token>` header. Delimiter variants are still secret-shaped; |
| 646 | // redact their entire line tail so token punctuation cannot terminate a |
| 647 | // regex early and expose the remainder. |
| 648 | let lowered = text.to_ascii_lowercase(); |
| 649 | let mut out = String::with_capacity(text.len()); |
| 650 | let mut cursor = 0usize; |
| 651 | while let Some(offset) = ["bearer-", "bearer_", "bearer:", "bearer="] |
| 652 | .iter() |
| 653 | .filter_map(|prefix| lowered[cursor..].find(prefix)) |
| 654 | .min() |
| 655 | { |
| 656 | let start = cursor + offset; |
| 657 | out.push_str(&text[cursor..start]); |
| 658 | let end = text[start..] |
| 659 | .find('\n') |
| 660 | .map(|line_end| start + line_end) |
| 661 | .unwrap_or(text.len()); |
| 662 | out.push_str(BEARER_REDACTION_MARKER); |
| 663 | cursor = end; |
| 664 | } |
| 665 | out.push_str(&text[cursor..]); |
| 666 | out |
| 667 | } |
| 668 | |
| 669 | /// Trailing characters that are punctuation or wrappers around a URL rather |
| 670 | /// than part of it. Trimming generously is safe in both directions: the |
| 671 | /// trimmed tail is re-appended verbatim and can hold no credential, while a |
| 672 | /// tail left attached would be swallowed by the query/fragment strip. |
| 673 | const URL_TRAILING_PUNCTUATION: &[char] = &[ |
| 674 | '.', ',', ';', ':', '!', '?', ')', ']', '}', '>', '"', '\'', '`', '*', '_', '\\', |
| 675 | ]; |
| 676 | |
| 677 | /// Strip URL userinfo, query, and fragment entirely, leaving a |
| 678 | /// `scheme://host[:port]/path` label. |
| 679 | /// |
| 680 | /// The shared sanitizer has already masked credentials in URLs it recognised; |
| 681 | /// this pass enforces the stricter structural-copy contract that no |
| 682 | /// userinfo, query string, or fragment may survive at all — including for |
| 683 | /// URLs that are punctuation-wrapped (`(https://…)`, `<https://…>`, |
| 684 | /// `"https://…"`), embedded mid-token, or uppercased. A token that starts |
| 685 | /// with a syntactically valid `scheme://` prefix but does not parse is replaced outright rather than |
| 686 | /// passed through, because an unparseable URL cannot be proven credential |
| 687 | /// free. |
| 688 | fn scrub_urls(text: &str) -> String { |
| 689 | let mut out = String::with_capacity(text.len()); |
| 690 | let mut cursor = 0usize; |
| 691 | while let Some(offset) = next_url_start(&text[cursor..]) { |
| 692 | let start = cursor + offset; |
| 693 | out.push_str(&text[cursor..start]); |
| 694 | let rest = &text[start..]; |
| 695 | // A scheme prefix contains no whitespace, so `end` is always > 0 and |
| 696 | // the cursor strictly advances. |
| 697 | let end = rest.find(char::is_whitespace).unwrap_or(rest.len()); |
| 698 | out.push_str(&scrub_url_token(&rest[..end])); |
| 699 | cursor = start + end; |
| 700 | } |
| 701 | out.push_str(&text[cursor..]); |
| 702 | out |
| 703 | } |
| 704 | |
| 705 | fn next_url_start(text: &str) -> Option<usize> { |
| 706 | for (separator, _) in text.match_indices("://") { |
| 707 | let before = &text[..separator]; |
| 708 | let start = before |
| 709 | .char_indices() |
| 710 | .rev() |
| 711 | .take_while(|(_, ch)| ch.is_ascii_alphanumeric() || matches!(ch, '+' | '-' | '.')) |
| 712 | .map(|(index, _)| index) |
| 713 | .last() |
| 714 | .unwrap_or(separator); |
| 715 | let scheme = &text[start..separator]; |
| 716 | if scheme |
| 717 | .chars() |
| 718 | .next() |
| 719 | .is_some_and(|ch| ch.is_ascii_alphabetic()) |
| 720 | { |
| 721 | return Some(start); |
| 722 | } |
| 723 | } |
| 724 | None |
| 725 | } |
| 726 | |
| 727 | fn scrub_url_token(token: &str) -> String { |
| 728 | let trimmed = token.trim_end_matches(URL_TRAILING_PUNCTUATION); |
| 729 | let suffix = &token[trimmed.len()..]; |
| 730 | let Ok(mut parsed) = url::Url::parse(trimmed) else { |
| 731 | return format!("{URL_OMISSION_MARKER}{suffix}"); |
| 732 | }; |
| 733 | // `set_username`/`set_password` only fail for cannot-be-a-base URLs. |
| 734 | // Failing closed keeps the "no userinfo survives" claim literally true. |
| 735 | if parsed.set_username("").is_err() || parsed.set_password(None).is_err() { |
| 736 | return format!("{URL_OMISSION_MARKER}{suffix}"); |
| 737 | } |
| 738 | parsed.set_query(None); |
| 739 | parsed.set_fragment(None); |
| 740 | format!("{parsed}{suffix}") |
| 741 | } |
| 742 | |
| 743 | fn scrub_paths(text: &str) -> String { |
| 744 | let mut out = String::with_capacity(text.len()); |
| 745 | let mut cursor = 0usize; |
| 746 | while let Some(start) = next_absolute_path_start(text, cursor) { |
| 747 | out.push_str(&text[cursor..start]); |
| 748 | // An unquoted absolute path can legally contain spaces. Stop at the |
| 749 | // line boundary rather than risk leaking the tail of such a path; |
| 750 | // losing adjacent prose is safer than emitting a customer/user name. |
| 751 | let end = text[start..] |
| 752 | .find('\n') |
| 753 | .map(|offset| start + offset) |
| 754 | .unwrap_or(text.len()); |
| 755 | out.push_str(PATH_OMISSION_MARKER); |
| 756 | cursor = end; |
| 757 | } |
| 758 | out.push_str(&text[cursor..]); |
| 759 | out |
| 760 | } |
| 761 | |
| 762 | fn next_absolute_path_start(text: &str, from: usize) -> Option<usize> { |
| 763 | let bytes = text.as_bytes(); |
| 764 | let mut index = from; |
| 765 | while index < bytes.len() { |
| 766 | let boundary = index == 0 |
| 767 | || text[..index] |
| 768 | .chars() |
| 769 | .next_back() |
| 770 | .is_some_and(|ch| !ch.is_alphanumeric() && !matches!(ch, '_' | '/' | '\\')); |
| 771 | if boundary { |
| 772 | let labelled_root = |
| 773 | text[..index].ends_with("<workspace>") || text[..index].ends_with("<home>"); |
| 774 | let url_separator = index > 0 |
| 775 | && index + 1 < bytes.len() |
| 776 | && bytes[index - 1] == b':' |
| 777 | && bytes[index + 1] == b'/'; |
| 778 | let previous_is_slash = index > 0 && bytes[index - 1] == b'/'; |
| 779 | let posix = |
| 780 | bytes[index] == b'/' && !previous_is_slash && !url_separator && !labelled_root; |
| 781 | let drive = index + 2 < bytes.len() |
| 782 | && bytes[index].is_ascii_alphabetic() |
| 783 | && bytes[index + 1] == b':' |
| 784 | && matches!(bytes[index + 2], b'/' | b'\\'); |
| 785 | let unc = index + 1 < bytes.len() && bytes[index] == b'\\' && bytes[index + 1] == b'\\'; |
| 786 | if posix || drive || unc { |
| 787 | return Some(index); |
| 788 | } |
| 789 | } |
| 790 | index += text[index..].chars().next()?.len_utf8(); |
| 791 | } |
| 792 | None |
| 793 | } |
| 794 | |
| 795 | // === Bounding (hard caps + exact accounting) === |
| 796 | |
| 797 | #[derive(Debug, Default, Clone, Copy, PartialEq, Eq)] |
| 798 | pub(in crate::commands) struct BoundStats { |
| 799 | /// Strings present in the full redacted tree, at every depth. |
| 800 | pub(in crate::commands) strings_total: u64, |
| 801 | /// Strings actually present in the emitted payload, including the |
| 802 | /// structural markers substituted for depth-omitted subtrees. |
| 803 | pub(in crate::commands) strings_retained: u64, |
| 804 | pub(in crate::commands) strings_truncated: u64, |
| 805 | pub(in crate::commands) string_bytes_original: u64, |
| 806 | pub(in crate::commands) string_bytes_retained: u64, |
| 807 | /// Array elements present in the full redacted tree, at every depth — |
| 808 | /// including elements inside subtrees that the depth cap later omits. |
| 809 | pub(in crate::commands) array_items_original: u64, |
| 810 | pub(in crate::commands) array_items_retained: u64, |
| 811 | pub(in crate::commands) depth_omissions: u64, |
| 812 | } |
| 813 | |
| 814 | /// Exact full-tree original counts. Deliberately depth-unbounded: the |
| 815 | /// receipt's `*_original` numbers describe the whole redacted object, so |
| 816 | /// that a subtree removed by the depth cap still shows up in the difference |
| 817 | /// between original and retained. |
| 818 | pub(in crate::commands) fn collect_original_counts(value: &Value, stats: &mut BoundStats) { |
| 819 | match value { |
| 820 | Value::String(text) => { |
| 821 | stats.strings_total += 1; |
| 822 | stats.string_bytes_original += text.len() as u64; |
| 823 | } |
| 824 | Value::Array(items) => { |
| 825 | stats.array_items_original += items.len() as u64; |
| 826 | for item in items { |
| 827 | collect_original_counts(item, stats); |
| 828 | } |
| 829 | } |
| 830 | Value::Object(map) => { |
| 831 | for item in map.values() { |
| 832 | collect_original_counts(item, stats); |
| 833 | } |
| 834 | } |
| 835 | Value::Null | Value::Bool(_) | Value::Number(_) => {} |
| 836 | } |
| 837 | } |
| 838 | |
| 839 | fn bound_value( |
| 840 | value: &mut Value, |
| 841 | caps: &Caps, |
| 842 | stats: &mut BoundStats, |
| 843 | reasons: &mut BTreeSet<&'static str>, |
| 844 | depth: usize, |
| 845 | ) { |
| 846 | match value { |
| 847 | Value::String(text) => { |
| 848 | let (truncated, was_truncated) = |
| 849 | truncate_string_grapheme_safe(text, caps.max_string_bytes); |
| 850 | if was_truncated { |
| 851 | *text = truncated; |
| 852 | stats.strings_truncated += 1; |
| 853 | reasons.insert("string_bytes_cap"); |
| 854 | } |
| 855 | stats.strings_retained += 1; |
| 856 | stats.string_bytes_retained += text.len() as u64; |
| 857 | } |
| 858 | Value::Array(items) => { |
| 859 | if depth >= caps.max_depth { |
| 860 | omit_for_depth(value, stats, reasons); |
| 861 | return; |
| 862 | } |
| 863 | if items.len() > caps.max_array_items { |
| 864 | items.truncate(caps.max_array_items); |
| 865 | reasons.insert("array_items_cap"); |
| 866 | } |
| 867 | stats.array_items_retained += items.len() as u64; |
| 868 | for item in items { |
| 869 | bound_value(item, caps, stats, reasons, depth + 1); |
| 870 | } |
| 871 | } |
| 872 | Value::Object(map) => { |
| 873 | if depth >= caps.max_depth { |
| 874 | omit_for_depth(value, stats, reasons); |
| 875 | return; |
| 876 | } |
| 877 | for item in map.values_mut() { |
| 878 | bound_value(item, caps, stats, reasons, depth + 1); |
| 879 | } |
| 880 | } |
| 881 | Value::Null | Value::Bool(_) | Value::Number(_) => {} |
| 882 | } |
| 883 | } |
| 884 | |
| 885 | /// Replace a too-deep subtree with the structural marker. The marker is a |
| 886 | /// string that really is emitted, so it counts toward the retained totals — |
| 887 | /// otherwise `string_bytes_retained` would understate the artifact it |
| 888 | /// describes. |
| 889 | fn omit_for_depth(value: &mut Value, stats: &mut BoundStats, reasons: &mut BTreeSet<&'static str>) { |
| 890 | stats.depth_omissions += 1; |
| 891 | reasons.insert("depth_cap"); |
| 892 | *value = Value::String(DEPTH_OMISSION_MARKER.to_string()); |
| 893 | stats.strings_retained += 1; |
| 894 | stats.string_bytes_retained += DEPTH_OMISSION_MARKER.len() as u64; |
| 895 | } |
| 896 | |
| 897 | /// UTF-8/grapheme-safe truncation: never splits a grapheme cluster, and the |
| 898 | /// retained bytes (including the ellipsis marker) never exceed the cap. |
| 899 | /// |
| 900 | /// When `max_bytes` is below the ellipsis's own 3 bytes there is no way to |
| 901 | /// emit both content and a truncation marker inside the cap. The honest |
| 902 | /// answer is the empty string plus `true`: the caller records a truncation, |
| 903 | /// and no partial content escapes under a cap it does not fit. |
| 904 | pub(in crate::commands) fn truncate_string_grapheme_safe( |
| 905 | text: &str, |
| 906 | max_bytes: usize, |
| 907 | ) -> (String, bool) { |
| 908 | if text.len() <= max_bytes { |
| 909 | return (text.to_string(), false); |
| 910 | } |
| 911 | if max_bytes < '…'.len_utf8() { |
| 912 | return (String::new(), true); |
| 913 | } |
| 914 | let budget = max_bytes - '…'.len_utf8(); |
| 915 | let mut out = String::new(); |
| 916 | for grapheme in UnicodeSegmentation::graphemes(text, true) { |
| 917 | if out.len() + grapheme.len() > budget { |
| 918 | break; |
| 919 | } |
| 920 | out.push_str(grapheme); |
| 921 | } |
| 922 | out.push('…'); |
| 923 | (out, true) |
| 924 | } |
| 925 | |
| 926 | // === Canonical serialization (deterministic, recursively sorted keys) === |
| 927 | |
| 928 | fn canonical_string(value: &Value) -> String { |
| 929 | let mut out = String::new(); |
| 930 | write_canonical(value, &mut out); |
| 931 | out |
| 932 | } |
| 933 | |
| 934 | fn write_canonical(value: &Value, out: &mut String) { |
| 935 | match value { |
| 936 | Value::Null => out.push_str("null"), |
| 937 | Value::Bool(flag) => out.push_str(if *flag { "true" } else { "false" }), |
| 938 | Value::Number(number) => { |
| 939 | let _ = write!(out, "{number}"); |
| 940 | } |
| 941 | Value::String(text) => { |
| 942 | let encoded = serde_json::to_string(text).unwrap_or_else(|_| "\"\"".to_string()); |
| 943 | out.push_str(&encoded); |
| 944 | } |
| 945 | Value::Array(items) => { |
| 946 | out.push('['); |
| 947 | for (index, item) in items.iter().enumerate() { |
| 948 | if index > 0 { |
| 949 | out.push(','); |
| 950 | } |
| 951 | write_canonical(item, out); |
| 952 | } |
| 953 | out.push(']'); |
| 954 | } |
| 955 | Value::Object(map) => { |
| 956 | let mut entries: Vec<(&String, &Value)> = map.iter().collect(); |
| 957 | entries.sort_by(|left, right| left.0.cmp(right.0)); |
| 958 | out.push('{'); |
| 959 | for (index, (key, item)) in entries.iter().enumerate() { |
| 960 | if index > 0 { |
| 961 | out.push(','); |
| 962 | } |
| 963 | let encoded = serde_json::to_string(key).unwrap_or_else(|_| "\"\"".to_string()); |
| 964 | out.push_str(&encoded); |
| 965 | out.push(':'); |
| 966 | write_canonical(item, out); |
| 967 | } |
| 968 | out.push('}'); |
| 969 | } |
| 970 | } |
| 971 | } |
| 972 | |
| 973 | // === Envelope assembly === |
| 974 | |
| 975 | pub(in crate::commands) fn render_copy( |
| 976 | copy: &dyn CommandSessionStructcopyContext, |
| 977 | presentation: &dyn CommandPresentationContext, |
| 978 | kind: &CopyKind, |
| 979 | caps: &Caps, |
| 980 | ) -> Result<String, String> { |
| 981 | let (kind_label, mut selector, mut payload) = match build_payload(copy, kind) { |
| 982 | Ok(payload) => payload, |
| 983 | Err(error) => return Err(selection_error(presentation, kind, error)?), |
| 984 | }; |
| 985 | let labels = PathLabels::new(©.path_roots()); |
| 986 | |
| 987 | // The selector is echoed verbatim into the receipt, so it goes through |
| 988 | // the same redaction as the payload and gets its own tight byte bound. |
| 989 | let mut selector_keys = KeyStats::default(); |
| 990 | redact_payload(&mut selector, &labels, &mut selector_keys); |
| 991 | bound_selector(&mut selector); |
| 992 | |
| 993 | let mut keys = KeyStats::default(); |
| 994 | redact_payload(&mut payload, &labels, &mut keys); |
| 995 | |
| 996 | // Fit the byte cap by tightening the content caps before ever |
| 997 | // considering a payload omission. |
| 998 | let mut effective = *caps; |
| 999 | for _ in 0..4 { |
| 1000 | let encoded = encode_attempt( |
| 1001 | kind_label, &selector, &payload, &effective, caps, &keys, false, |
| 1002 | ); |
| 1003 | if encoded.len() <= caps.max_output_bytes { |
| 1004 | return Ok(encoded); |
| 1005 | } |
| 1006 | effective.max_string_bytes = (effective.max_string_bytes / 2).max(64); |
| 1007 | effective.max_array_items = (effective.max_array_items / 2).max(1); |
| 1008 | effective.max_depth = effective.max_depth.saturating_sub(2).max(2); |
| 1009 | } |
| 1010 | |
| 1011 | // Last resort: emit receipt metadata only. If even that exceeds the cap, |
| 1012 | // fail closed rather than emit an over-cap artifact. |
| 1013 | let encoded = encode_attempt( |
| 1014 | kind_label, &selector, &payload, &effective, caps, &keys, true, |
| 1015 | ); |
| 1016 | if encoded.len() <= caps.max_output_bytes { |
| 1017 | return Ok(encoded); |
| 1018 | } |
| 1019 | Err(presentation.translate( |
| 1020 | "cmd_structcopy_receipt_too_large", |
| 1021 | &[("bytes", &caps.max_output_bytes.to_string())], |
| 1022 | )?) |
| 1023 | } |
| 1024 | |
| 1025 | /// Bound the selector independently of the payload caps. Selectors are |
| 1026 | /// scalars, so this only has to handle the string case. |
| 1027 | fn bound_selector(selector: &mut Value) { |
| 1028 | if let Value::String(text) = selector { |
| 1029 | let (bounded, _) = truncate_string_grapheme_safe(text, MAX_SELECTOR_BYTES); |
| 1030 | *text = bounded; |
| 1031 | } |
| 1032 | } |
| 1033 | |
| 1034 | fn encode_attempt( |
| 1035 | kind_label: &str, |
| 1036 | selector: &Value, |
| 1037 | payload: &Value, |
| 1038 | effective: &Caps, |
| 1039 | hard: &Caps, |
| 1040 | keys: &KeyStats, |
| 1041 | omit_payload: bool, |
| 1042 | ) -> String { |
| 1043 | let mut candidate = payload.clone(); |
| 1044 | let mut stats = BoundStats::default(); |
| 1045 | let mut reasons = BTreeSet::new(); |
| 1046 | collect_original_counts(&candidate, &mut stats); |
| 1047 | bound_value(&mut candidate, effective, &mut stats, &mut reasons, 0); |
| 1048 | let mut retained_keys = RetainedKeyStats::default(); |
| 1049 | collect_retained_key_stats(&candidate, keys, &mut Vec::new(), &mut retained_keys); |
| 1050 | if effective != hard { |
| 1051 | reasons.insert("caps_tightened_output_bytes_cap"); |
| 1052 | } |
| 1053 | if retained_keys.truncated > 0 { |
| 1054 | reasons.insert("object_key_bytes_cap"); |
| 1055 | } |
| 1056 | if retained_keys.deduped > 0 { |
| 1057 | reasons.insert("object_key_collision"); |
| 1058 | } |
| 1059 | let emitted = if omit_payload { |
| 1060 | // Nothing from the bounding pass was emitted, so every retained |
| 1061 | // counter and every bounding reason would be a claim about an |
| 1062 | // artifact that does not exist. Originals stay; the rest resets. |
| 1063 | reasons.clear(); |
| 1064 | reasons.insert("payload_omitted_output_bytes_cap"); |
| 1065 | stats.strings_retained = 0; |
| 1066 | stats.strings_truncated = 0; |
| 1067 | stats.string_bytes_retained = 0; |
| 1068 | stats.array_items_retained = 0; |
| 1069 | stats.depth_omissions = 0; |
| 1070 | retained_keys = RetainedKeyStats::default(); |
| 1071 | Value::Null |
| 1072 | } else { |
| 1073 | candidate |
| 1074 | }; |
| 1075 | let envelope = assemble_envelope( |
| 1076 | kind_label, |
| 1077 | selector, |
| 1078 | &emitted, |
| 1079 | &stats, |
| 1080 | keys, |
| 1081 | &retained_keys, |
| 1082 | &reasons, |
| 1083 | effective, |
| 1084 | hard, |
| 1085 | ); |
| 1086 | canonical_string(&envelope) |
| 1087 | } |
| 1088 | |
| 1089 | #[allow(clippy::too_many_arguments)] |
| 1090 | fn assemble_envelope( |
| 1091 | kind: &str, |
| 1092 | selector: &Value, |
| 1093 | payload: &Value, |
| 1094 | stats: &BoundStats, |
| 1095 | original_keys: &KeyStats, |
| 1096 | retained_keys: &RetainedKeyStats, |
| 1097 | reasons: &BTreeSet<&'static str>, |
| 1098 | effective: &Caps, |
| 1099 | hard: &Caps, |
| 1100 | ) -> Value { |
| 1101 | json!({ |
| 1102 | "object": payload, |
| 1103 | "receipt": { |
| 1104 | "schema": SCHEMA_ID, |
| 1105 | "human_only": true, |
| 1106 | "kind": kind, |
| 1107 | "selector": selector, |
| 1108 | "redaction": REDACTION_CONTRACT, |
| 1109 | // `caps` is the declared contract; `applied_caps` is what this |
| 1110 | // artifact was actually bounded with. They differ whenever the |
| 1111 | // output-byte cap forced a tightening pass. |
| 1112 | "caps": caps_value(hard), |
| 1113 | "applied_caps": caps_value(effective), |
| 1114 | "counts": { |
| 1115 | "strings_total": stats.strings_total, |
| 1116 | "strings_retained": stats.strings_retained, |
| 1117 | "strings_truncated": stats.strings_truncated, |
| 1118 | "string_bytes_original": stats.string_bytes_original, |
| 1119 | "string_bytes_retained": stats.string_bytes_retained, |
| 1120 | "array_items_original": stats.array_items_original, |
| 1121 | "array_items_retained": stats.array_items_retained, |
| 1122 | "depth_omissions": stats.depth_omissions, |
| 1123 | "object_keys_original": original_keys.original_total(), |
| 1124 | "object_keys_retained": retained_keys.total, |
| 1125 | "object_keys_truncated": retained_keys.truncated, |
| 1126 | "object_keys_deduped": retained_keys.deduped, |
| 1127 | "payload_bytes": canonical_string(payload).len(), |
| 1128 | }, |
| 1129 | "reasons": reasons.iter().copied().collect::<Vec<_>>(), |
| 1130 | } |
| 1131 | }) |
| 1132 | } |
| 1133 | |
| 1134 | fn caps_value(caps: &Caps) -> Value { |
| 1135 | json!({ |
| 1136 | "max_output_bytes": caps.max_output_bytes, |
| 1137 | "max_array_items": caps.max_array_items, |
| 1138 | "max_string_bytes": caps.max_string_bytes, |
| 1139 | "max_key_bytes": MAX_KEY_BYTES, |
| 1140 | "max_depth": caps.max_depth, |
| 1141 | }) |
| 1142 | } |
| 1143 | |
| 1144 | #[cfg(test)] |
| 1145 | mod tests { |
| 1146 | use super::*; |
| 1147 | fn no_labels() -> PathLabels { |
| 1148 | PathLabels { labels: Vec::new() } |
| 1149 | } |
| 1150 | /// `unique_object_key` must terminate and preserve every value even when |
| 1151 | /// the key cap leaves no room at all for a base. |
| 1152 | #[test] |
| 1153 | fn key_dedup_terminates_under_a_degenerate_cap() { |
| 1154 | let mut map = serde_json::Map::new(); |
| 1155 | for _ in 0..12 { |
| 1156 | let (key, _) = unique_object_key(&map, ""); |
| 1157 | assert!(!map.contains_key(&key), "reused key {key:?}"); |
| 1158 | map.insert(key, Value::Null); |
| 1159 | } |
| 1160 | assert_eq!(map.len(), 12, "every insert must survive"); |
| 1161 | |
| 1162 | // Deterministic across runs with the same inputs. |
| 1163 | let mut replay = serde_json::Map::new(); |
| 1164 | for _ in 0..12 { |
| 1165 | let (key, _) = unique_object_key(&replay, ""); |
| 1166 | replay.insert(key, Value::Null); |
| 1167 | } |
| 1168 | let left: Vec<&String> = map.keys().collect(); |
| 1169 | let right: Vec<&String> = replay.keys().collect(); |
| 1170 | assert_eq!(left, right); |
| 1171 | |
| 1172 | assert_eq!(decimal_width(0), 1); |
| 1173 | assert_eq!(decimal_width(9), 1); |
| 1174 | assert_eq!(decimal_width(10), 2); |
| 1175 | assert_eq!(decimal_width(999), 3); |
| 1176 | assert_eq!(decimal_width(1000), 4); |
| 1177 | } |
| 1178 | |
| 1179 | /// URLs do not arrive as tidy whitespace-delimited tokens. Wrapped, |
| 1180 | /// embedded, uppercased, and malformed forms must all lose their |
| 1181 | /// userinfo, query, and fragment. |
| 1182 | #[test] |
| 1183 | fn urls_lose_userinfo_query_and_fragment_in_hostile_shapes() { |
| 1184 | let labels = no_labels(); |
| 1185 | let cases = [ |
| 1186 | "(https://u:p@host.test/a?q=1#f)", |
| 1187 | "<https://u:p@host.test/a?q=1#f>", |
| 1188 | "\"https://u:p@host.test/a?q=1#f\"", |
| 1189 | "'https://u:p@host.test/a?q=1#f'", |
| 1190 | "see https://u:p@host.test/a?q=1#f.", |
| 1191 | "see https://u:p@host.test/a?q=1#f, then", |
| 1192 | "[link](https://u:p@host.test/a?q=1#f)", |
| 1193 | "prefixhttps://u:p@host.test/a?q=1#f", |
| 1194 | "HTTPS://U:P@HOST.TEST/a?q=1#f", |
| 1195 | "ws://u:p@host.test/a?q=1#f", |
| 1196 | "ftp://u:p@host.test/a?q=1#f", |
| 1197 | "postgres://u:p@host.test/db?sslkey=secret#f", |
| 1198 | "mongodb://u:p@host.test/db?authSource=admin#f", |
| 1199 | "redis://u:p@host.test/0?token=secret#f", |
| 1200 | "amqp://u:p@host.test/vhost?token=secret#f", |
| 1201 | "ssh://u:p@host.test/repo?identity=secret#f", |
| 1202 | "socks5://u:p@host.test/path?token=secret#f", |
| 1203 | "trailing`https://u:p@host.test/a?q=1#f`", |
| 1204 | "a=https://u:p@host.test/a?q=1#f&b=2", |
| 1205 | ]; |
| 1206 | for case in cases { |
| 1207 | let scrubbed = scrub_string(case, &labels); |
| 1208 | for forbidden in [ |
| 1209 | "u:p@", |
| 1210 | "q=1", |
| 1211 | "#f", |
| 1212 | "P@HOST", |
| 1213 | "sslkey=secret", |
| 1214 | "authSource=admin", |
| 1215 | "token=secret", |
| 1216 | "identity=secret", |
| 1217 | ] { |
| 1218 | assert!( |
| 1219 | !scrubbed.contains(forbidden), |
| 1220 | "{case:?} kept {forbidden:?}: {scrubbed}" |
| 1221 | ); |
| 1222 | } |
| 1223 | assert!( |
| 1224 | scrubbed.contains("host.test") || scrubbed.contains(URL_OMISSION_MARKER), |
| 1225 | "{case:?} -> {scrubbed}" |
| 1226 | ); |
| 1227 | } |
| 1228 | |
| 1229 | // Two URLs in one string: both are scrubbed, order preserved. |
| 1230 | let both = scrub_string( |
| 1231 | "first https://a:b@one.test/x?y=1#z then https://c:d@two.test/w?v=2#u end", |
| 1232 | &labels, |
| 1233 | ); |
| 1234 | assert!(both.contains("one.test"), "{both}"); |
| 1235 | assert!(both.contains("two.test"), "{both}"); |
| 1236 | assert!(both.starts_with("first "), "{both}"); |
| 1237 | assert!(both.ends_with(" end"), "{both}"); |
| 1238 | for forbidden in ["a:b@", "c:d@", "y=1", "v=2", "#z", "#u"] { |
| 1239 | assert!(!both.contains(forbidden), "kept {forbidden:?}: {both}"); |
| 1240 | } |
| 1241 | |
| 1242 | // Unparseable but scheme-prefixed: fail closed, do not pass through. |
| 1243 | for hostile in [ |
| 1244 | "https://", |
| 1245 | "https://[not-an-ipv6:1]/x?token=leak#f", |
| 1246 | "http://user:pw@:99999/x?token=leak", |
| 1247 | ] { |
| 1248 | let scrubbed = scrub_string(hostile, &labels); |
| 1249 | assert!(!scrubbed.contains("token=leak"), "{hostile} -> {scrubbed}"); |
| 1250 | assert!(!scrubbed.contains("user:pw@"), "{hostile} -> {scrubbed}"); |
| 1251 | } |
| 1252 | |
| 1253 | // An ANSI escape spliced into a scheme must not hide the URL from |
| 1254 | // the scanner: `sanitize_text` runs first. |
| 1255 | let hidden = scrub_string("htt\u{1b}[0mps://u:p@host.test/a?q=1#f", &labels); |
| 1256 | assert!(!hidden.contains("u:p@"), "{hidden}"); |
| 1257 | assert!(!hidden.contains("q=1"), "{hidden}"); |
| 1258 | |
| 1259 | // Text with no URL is untouched. |
| 1260 | assert_eq!( |
| 1261 | scrub_string("plain text, no url", &labels), |
| 1262 | "plain text, no url" |
| 1263 | ); |
| 1264 | } |
| 1265 | |
| 1266 | #[test] |
| 1267 | fn path_labels_require_component_boundaries_and_preserve_repeated_roots() { |
| 1268 | let labels = PathLabels { |
| 1269 | labels: vec![ |
| 1270 | ("/opt/app".to_string(), "<workspace>"), |
| 1271 | ("/Users/alice".to_string(), "<home>"), |
| 1272 | ], |
| 1273 | }; |
| 1274 | |
| 1275 | assert_eq!(labels.apply("/opt/app"), "<workspace>"); |
| 1276 | assert_eq!(labels.apply("/opt/app/src"), "<workspace>/src"); |
| 1277 | assert_eq!(labels.apply(r"/opt/app\src"), r"<workspace>\src"); |
| 1278 | assert_eq!( |
| 1279 | labels.apply("/opt/app/a and /opt/app/b"), |
| 1280 | "<workspace>/a and <workspace>/b" |
| 1281 | ); |
| 1282 | assert_eq!(labels.apply("/Users/alice"), "<home>"); |
| 1283 | assert_eq!( |
| 1284 | labels.apply("/Users/alice/project and /Users/alice/other"), |
| 1285 | "<home>/project and <home>/other" |
| 1286 | ); |
| 1287 | |
| 1288 | for collision in [ |
| 1289 | "/opt/application/customer", |
| 1290 | "/opt/app-old/customer", |
| 1291 | "/Users/alice-old/private", |
| 1292 | "/Users/alice2/private", |
| 1293 | ] { |
| 1294 | assert_eq!( |
| 1295 | labels.apply(collision), |
| 1296 | collision, |
| 1297 | "near-prefix path must not receive a trusted label" |
| 1298 | ); |
| 1299 | assert_eq!( |
| 1300 | scrub_string(collision, &labels), |
| 1301 | PATH_OMISSION_MARKER, |
| 1302 | "near-prefix path must remain foreign and be redacted" |
| 1303 | ); |
| 1304 | } |
| 1305 | } |
| 1306 | |
| 1307 | struct RecordingCopy { |
| 1308 | events: std::cell::RefCell<Vec<String>>, |
| 1309 | error: Option<StructcopyError>, |
| 1310 | transport: Result<StructcopyTransport, String>, |
| 1311 | } |
| 1312 | impl RecordingCopy { |
| 1313 | fn new() -> Self { |
| 1314 | Self { |
| 1315 | events: Default::default(), |
| 1316 | error: None, |
| 1317 | transport: Ok(StructcopyTransport::Native), |
| 1318 | } |
| 1319 | } |
| 1320 | fn observe(&self, event: String) -> Result<(), StructcopyError> { |
| 1321 | self.events.borrow_mut().push(event); |
| 1322 | self.error.clone().map_or(Ok(()), Err) |
| 1323 | } |
| 1324 | } |
| 1325 | impl CommandSessionStructcopyContext for RecordingCopy { |
| 1326 | fn transcript_item(&self, index: usize) -> Result<StructcopyTranscript, StructcopyError> { |
| 1327 | self.observe(format!("turn:{index}"))?; |
| 1328 | Ok(StructcopyTranscript { |
| 1329 | index, |
| 1330 | role: "user".into(), |
| 1331 | content: StructcopyContent::Visible(vec![StructcopyBlock::Text("visible".into())]), |
| 1332 | }) |
| 1333 | } |
| 1334 | fn tool_pair(&self, id: &str) -> Result<StructcopyToolPair, StructcopyError> { |
| 1335 | self.observe(format!("tool:{id}"))?; |
| 1336 | Ok(StructcopyToolPair { |
| 1337 | name: "fetch".into(), |
| 1338 | input: json!({"api_key":"private"}), |
| 1339 | result: None, |
| 1340 | }) |
| 1341 | } |
| 1342 | fn plan_snapshot(&self) -> Result<StructcopyPlan, StructcopyError> { |
| 1343 | self.observe("plan".into())?; |
| 1344 | Ok(StructcopyPlan { |
| 1345 | title: Some("plan".into()), |
| 1346 | ..Default::default() |
| 1347 | }) |
| 1348 | } |
| 1349 | fn workflow_projection(&self, id: &str) -> Result<StructcopyWorkflow, StructcopyError> { |
| 1350 | self.observe(format!("workflow:{id}"))?; |
| 1351 | Err(StructcopyError::Unavailable) |
| 1352 | } |
| 1353 | fn path_roots(&self) -> StructcopyPathRoots { |
| 1354 | self.events.borrow_mut().push("roots".into()); |
| 1355 | StructcopyPathRoots { |
| 1356 | workspace: "/work".into(), |
| 1357 | canonical_workspace: None, |
| 1358 | home: None, |
| 1359 | } |
| 1360 | } |
| 1361 | fn write_clipboard(&self, text: &str) -> Result<StructcopyTransport, String> { |
| 1362 | let payload: Value = |
| 1363 | serde_json::from_str(text).expect("only rendered JSON may reach clipboard"); |
| 1364 | assert_eq!(payload["receipt"]["schema"], SCHEMA_ID); |
| 1365 | assert!(!text.contains("private")); |
| 1366 | self.events.borrow_mut().push("clipboard".into()); |
| 1367 | self.transport.clone() |
| 1368 | } |
| 1369 | } |
| 1370 | struct Labels; |
| 1371 | impl CommandPresentationContext for Labels { |
| 1372 | fn translate(&self, key: &str, _: &[(&str, &str)]) -> Result<String, String> { |
| 1373 | Ok(key.into()) |
| 1374 | } |
| 1375 | } |
| 1376 | |
| 1377 | #[test] |
| 1378 | fn structcopy_portable_selects_one_observation_and_writes_only_after_rendering() { |
| 1379 | for (args, selected) in [ |
| 1380 | ("turn +1", "turn:1"), |
| 1381 | ("tool call-id", "tool:call-id"), |
| 1382 | ("plan", "plan"), |
| 1383 | ] { |
| 1384 | for stdout in [false, true] { |
| 1385 | let copy = RecordingCopy::new(); |
| 1386 | let args = format!("{args}{}", if stdout { " STDOUT" } else { "" }); |
| 1387 | let result = execute_portable(©, &Labels, Some(&args), &DEFAULT_CAPS).unwrap(); |
| 1388 | assert!(!result.is_error); |
| 1389 | assert!(result.action.is_none()); |
| 1390 | let mut expected = vec![selected, "roots"]; |
| 1391 | if stdout { |
| 1392 | assert!(serde_json::from_str::<Value>(&result.message.unwrap()).is_ok()); |
| 1393 | } else { |
| 1394 | expected.push("clipboard"); |
| 1395 | assert_eq!( |
| 1396 | result.message.as_deref(), |
| 1397 | Some("cmd_structcopy_clipboard_accepted") |
| 1398 | ); |
| 1399 | } |
| 1400 | assert_eq!(*copy.events.borrow(), expected); |
| 1401 | } |
| 1402 | } |
| 1403 | } |
| 1404 | |
| 1405 | #[test] |
| 1406 | fn structcopy_portable_rejections_never_write_clipboard() { |
| 1407 | for args in [ |
| 1408 | None, |
| 1409 | Some(""), |
| 1410 | Some("turn 0"), |
| 1411 | Some("TURN 1"), |
| 1412 | Some("plan extra"), |
| 1413 | Some("tool"), |
| 1414 | Some("workflow"), |
| 1415 | Some("stdout plan"), |
| 1416 | ] { |
| 1417 | let copy = RecordingCopy::new(); |
| 1418 | assert!(execute_portable(©, &Labels, args, &DEFAULT_CAPS).is_err()); |
| 1419 | assert!(copy.events.borrow().is_empty()); |
| 1420 | } |
| 1421 | for error in [ |
| 1422 | StructcopyError::Unavailable, |
| 1423 | StructcopyError::Busy, |
| 1424 | StructcopyError::Preparation("broken".into()), |
| 1425 | ] { |
| 1426 | let mut copy = RecordingCopy::new(); |
| 1427 | copy.error = Some(error); |
| 1428 | assert!(execute_portable(©, &Labels, Some("plan"), &DEFAULT_CAPS).is_err()); |
| 1429 | assert_eq!(*copy.events.borrow(), ["plan"]); |
| 1430 | } |
| 1431 | let copy = RecordingCopy::new(); |
| 1432 | assert!(execute_portable(©, &Labels, Some("workflow missing"), &DEFAULT_CAPS).is_err()); |
| 1433 | assert_eq!(*copy.events.borrow(), ["workflow:missing"]); |
| 1434 | let copy = RecordingCopy::new(); |
| 1435 | assert!( |
| 1436 | execute_portable( |
| 1437 | ©, |
| 1438 | &Labels, |
| 1439 | Some("turn 1"), |
| 1440 | &Caps { |
| 1441 | max_output_bytes: 1, |
| 1442 | ..DEFAULT_CAPS |
| 1443 | } |
| 1444 | ) |
| 1445 | .is_err() |
| 1446 | ); |
| 1447 | assert_eq!(*copy.events.borrow(), ["turn:1", "roots"]); |
| 1448 | } |
| 1449 | |
| 1450 | #[test] |
| 1451 | fn structcopy_portable_transport_receipts_and_missing_facets_are_honest() { |
| 1452 | for (transport, key, error) in [ |
| 1453 | ( |
| 1454 | Ok(StructcopyTransport::TerminalQueued), |
| 1455 | "cmd_structcopy_clipboard_queued", |
| 1456 | false, |
| 1457 | ), |
| 1458 | ( |
| 1459 | Err("clipboard unavailable".into()), |
| 1460 | "cmd_structcopy_clipboard_failed", |
| 1461 | true, |
| 1462 | ), |
| 1463 | ] { |
| 1464 | let mut copy = RecordingCopy::new(); |
| 1465 | copy.transport = transport; |
| 1466 | let result = execute_structcopy( |
| 1467 | CommandContexts::empty() |
| 1468 | .with_structcopy(&mut copy) |
| 1469 | .with_presentation(&mut Labels), |
| 1470 | Some("plan"), |
| 1471 | ); |
| 1472 | assert_eq!(result.is_error, error); |
| 1473 | assert!(result.action.is_none()); |
| 1474 | assert_eq!( |
| 1475 | result.message, |
| 1476 | Some(if error { |
| 1477 | format!("Error: {key}") |
| 1478 | } else { |
| 1479 | key.into() |
| 1480 | }) |
| 1481 | ); |
| 1482 | assert_eq!(*copy.events.borrow(), ["plan", "roots", "clipboard"]); |
| 1483 | } |
| 1484 | let mut copy = RecordingCopy::new(); |
| 1485 | assert!(execute_structcopy(CommandContexts::empty(), Some("plan")).is_error); |
| 1486 | assert!( |
| 1487 | execute_structcopy( |
| 1488 | CommandContexts::empty().with_presentation(&mut Labels), |
| 1489 | Some("plan") |
| 1490 | ) |
| 1491 | .is_error |
| 1492 | ); |
| 1493 | assert!( |
| 1494 | execute_structcopy( |
| 1495 | CommandContexts::empty().with_structcopy(&mut copy), |
| 1496 | Some("plan") |
| 1497 | ) |
| 1498 | .is_error |
| 1499 | ); |
| 1500 | assert!(copy.events.borrow().is_empty()); |
| 1501 | } |
| 1502 | } |
| 1503 |