| 1 | //! `@`-mention parsing, completion, and expansion for the composer. |
| 2 | //! |
| 3 | //! Two responsibilities live here: |
| 4 | //! |
| 5 | //! 1. **Tab-completion** at the cursor — `try_autocomplete_file_mention` is |
| 6 | //! called by the composer's Tab handler. Walks the workspace, ranks |
| 7 | //! candidates by prefix-then-substring match, and either splices the |
| 8 | //! completion in directly (single match), extends to a shared prefix, or |
| 9 | //! surfaces options in the status line. |
| 10 | //! 2. **Expansion before send** — when the user hits Enter on a message that |
| 11 | //! contains `@<path>` references, `user_request_with_file_mentions` |
| 12 | //! appends a "Local context from @mentions" block with the file contents |
| 13 | //! (or directory listings, or media-attachment hints) so the model can see |
| 14 | //! what the user pointed at. Capped per-message and per-file. |
| 15 | //! |
| 16 | //! The module is deliberately self-contained: nothing inside reaches into UI |
| 17 | //! widgets or rendering, so it stays unit-testable from `ui/tests.rs` and |
| 18 | //! from its own module-level tests. |
| 19 | //! |
| 20 | //! Pulled out of `ui.rs` to shrink the 5,500-line monolith and to give the |
| 21 | //! mention logic a single home that future maintainers can find without |
| 22 | //! grepping for `@` across half the codebase. |
| 23 | |
| 24 | use std::fmt::Write; |
| 25 | use std::io::Read; |
| 26 | use std::path::{Path, PathBuf}; |
| 27 | |
| 28 | use serde::{Deserialize, Serialize}; |
| 29 | |
| 30 | use crate::tui::app::{App, MentionCompletionCache}; |
| 31 | use crate::tui::git_mention::{self, GitMentionCache, GitMentionKind}; |
| 32 | use crate::tui::mention_completion::{MentionDiscoveryBehavior, MentionDiscoveryKey}; |
| 33 | use crate::working_set::Workspace; |
| 34 | |
| 35 | /// Maximum number of `@`-mentions whose contents are inlined into one user |
| 36 | /// message. Beyond this we stop appending blocks but the raw `@token` text |
| 37 | /// remains in the message. |
| 38 | pub const MAX_FILE_MENTIONS_PER_MESSAGE: usize = 8; |
| 39 | /// Per-file byte ceiling when inlining mention contents. |
| 40 | pub const MAX_MENTION_FILE_BYTES: u64 = 128 * 1024; |
| 41 | /// Per-directory entry ceiling when inlining a directory listing. |
| 42 | pub const MAX_DIRECTORY_MENTION_ENTRIES: usize = 80; |
| 43 | |
| 44 | /// Maximum file-mention completion candidates to consider per keypress. Caps |
| 45 | /// the cost of walking large workspaces; subsequent keystrokes narrow further. |
| 46 | const FILE_MENTION_COMPLETION_LIMIT: usize = 64; |
| 47 | |
| 48 | /// Compact composer preview row for local context. `included=false` also |
| 49 | /// covers lexical `@` mentions whose exact inclusion is resolved on send. |
| 50 | #[derive(Debug, Clone, PartialEq, Eq)] |
| 51 | pub struct FileMentionPreview { |
| 52 | pub kind: String, |
| 53 | pub label: String, |
| 54 | pub detail: Option<String>, |
| 55 | pub included: bool, |
| 56 | pub removable: bool, |
| 57 | } |
| 58 | |
| 59 | /// Durable, compact metadata for a user-visible context reference. |
| 60 | /// |
| 61 | /// The transcript keeps the user's compact text (`@path` or `[Attached ...]`) |
| 62 | /// readable. This record preserves the exact target and inclusion state for |
| 63 | /// the context inspector and for session resume without leaking raw metadata |
| 64 | /// into the visible history cell. |
| 65 | #[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)] |
| 66 | pub struct ContextReference { |
| 67 | pub kind: ContextReferenceKind, |
| 68 | pub source: ContextReferenceSource, |
| 69 | /// Short badge for terminal display, e.g. `file`, `dir`, `image`. |
| 70 | pub badge: String, |
| 71 | /// Compact display label from the transcript, without the leading `@`. |
| 72 | pub label: String, |
| 73 | /// Resolved target path or URI-equivalent string. |
| 74 | pub target: String, |
| 75 | pub included: bool, |
| 76 | pub expanded: bool, |
| 77 | #[serde(default, skip_serializing_if = "Option::is_none")] |
| 78 | pub detail: Option<String>, |
| 79 | } |
| 80 | |
| 81 | #[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)] |
| 82 | #[serde(rename_all = "snake_case")] |
| 83 | pub enum ContextReferenceKind { |
| 84 | File, |
| 85 | Directory, |
| 86 | Missing, |
| 87 | Unsupported, |
| 88 | MediaMention, |
| 89 | MediaAttachment, |
| 90 | /// `@git` / `@diff` — curated git context rather than a path (#4067). |
| 91 | GitContext, |
| 92 | } |
| 93 | |
| 94 | #[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)] |
| 95 | #[serde(rename_all = "snake_case")] |
| 96 | pub enum ContextReferenceSource { |
| 97 | AtMention, |
| 98 | Attachment, |
| 99 | } |
| 100 | |
| 101 | // --------------------------------------------------------------------------- |
| 102 | // Tab-completion |
| 103 | // --------------------------------------------------------------------------- |
| 104 | |
| 105 | /// If the cursor sits inside a `@<partial>` token in the input, return the |
| 106 | /// byte offset where the `@` starts (so we can splice in a completion) and |
| 107 | /// the partial path the user has typed so far. The token stops at whitespace |
| 108 | /// or the end of input. Returns `None` when the cursor is outside any mention |
| 109 | /// or the token is empty (`@` with nothing after it). |
| 110 | pub fn partial_file_mention_at_cursor(input: &str, cursor_chars: usize) -> Option<(usize, String)> { |
| 111 | let chars: Vec<char> = input.chars().collect(); |
| 112 | if cursor_chars > chars.len() { |
| 113 | return None; |
| 114 | } |
| 115 | // Walk left from the cursor until we find an `@` or a whitespace; if |
| 116 | // whitespace comes first the cursor isn't inside a mention. |
| 117 | let mut start_chars = cursor_chars; |
| 118 | while start_chars > 0 { |
| 119 | let prev = chars[start_chars - 1]; |
| 120 | if prev == '@' { |
| 121 | start_chars -= 1; |
| 122 | break; |
| 123 | } |
| 124 | if prev.is_whitespace() { |
| 125 | return None; |
| 126 | } |
| 127 | start_chars -= 1; |
| 128 | } |
| 129 | if start_chars == cursor_chars || chars.get(start_chars) != Some(&'@') { |
| 130 | return None; |
| 131 | } |
| 132 | // Confirm the `@` itself is at a valid mention boundary. |
| 133 | if !is_file_mention_start(&chars, start_chars) { |
| 134 | return None; |
| 135 | } |
| 136 | // Consume from the `@` to the next whitespace (the end of the token). |
| 137 | let mut end_chars = start_chars + 1; |
| 138 | while end_chars < chars.len() && !chars[end_chars].is_whitespace() { |
| 139 | end_chars += 1; |
| 140 | } |
| 141 | let partial: String = chars[start_chars + 1..end_chars].iter().collect(); |
| 142 | let byte_start: usize = chars[..start_chars].iter().map(|c| c.len_utf8()).sum(); |
| 143 | Some((byte_start, partial)) |
| 144 | } |
| 145 | |
| 146 | /// Cwd-aware completion entry point. Shares its walker with the future |
| 147 | /// Ctrl+P fuzzy picker (#97); see [`Workspace::completions`] for the |
| 148 | /// ranking + display rules. |
| 149 | #[cfg(test)] |
| 150 | pub fn find_file_mention_completions( |
| 151 | workspace: &Workspace, |
| 152 | partial: &str, |
| 153 | limit: usize, |
| 154 | ) -> Vec<String> { |
| 155 | let entries = workspace.completions(partial, limit); |
| 156 | // #441: re-rank by frecency so files the user mentions a lot float up. |
| 157 | // Never-mentioned candidates fall back to the workspace ranker's order. |
| 158 | let entries = super::file_frecency::rerank_by_frecency(entries); |
| 159 | tracing::debug!( |
| 160 | target: "codewhale_tui::file_mention", |
| 161 | partial = %partial, |
| 162 | workspace = %workspace.root.display(), |
| 163 | cwd = ?std::env::current_dir().ok(), |
| 164 | match_count = entries.len(), |
| 165 | "file mention completion walk", |
| 166 | ); |
| 167 | entries |
| 168 | } |
| 169 | |
| 170 | /// Resolve the `@`-mention completion popup contents for the current |
| 171 | /// composer state. Returns an empty `Vec` when: |
| 172 | /// |
| 173 | /// - The popup is suppressed (`app.mention_menu_hidden`). |
| 174 | /// - The cursor is not inside an `@<partial>` token. |
| 175 | /// - The workspace walk produced no candidates. |
| 176 | /// |
| 177 | /// Mirrors `visible_slash_menu_entries` so the composer widget can treat |
| 178 | /// both menus identically (one `Vec<String>` of entries, one selected index). |
| 179 | /// |
| 180 | /// Once the composer widget is extended to render this as a popup, it will |
| 181 | /// pair with `apply_mention_menu_selection` for the Up/Down/Enter flow. |
| 182 | #[must_use] |
| 183 | pub fn visible_mention_menu_entries(app: &mut App, limit: usize) -> Vec<String> { |
| 184 | if app.mention_menu_hidden { |
| 185 | app.composer.mention_discovery.cancel(); |
| 186 | return Vec::new(); |
| 187 | } |
| 188 | let Some((_byte_start, partial)) = |
| 189 | partial_file_mention_at_cursor(&app.input, app.cursor_position) |
| 190 | else { |
| 191 | app.composer.mention_discovery.cancel(); |
| 192 | return Vec::new(); |
| 193 | }; |
| 194 | if limit == 0 { |
| 195 | app.composer.mention_discovery.cancel(); |
| 196 | return Vec::new(); |
| 197 | } |
| 198 | |
| 199 | mention_menu_entries(app, &partial, limit).0 |
| 200 | } |
| 201 | |
| 202 | /// Drain a completed discovery result without waiting. The event loop calls |
| 203 | /// this once per tick so a finished scan repaints the popup even when the user |
| 204 | /// has stopped typing. |
| 205 | pub(crate) fn poll_background_mention_discovery(app: &mut App) -> bool { |
| 206 | if app.composer.mention_discovery.poll() { |
| 207 | app.composer.mention_completion_cache = None; |
| 208 | return true; |
| 209 | } |
| 210 | false |
| 211 | } |
| 212 | |
| 213 | /// Return `(entries, ready)`. `ready = false` means discovery is still in the |
| 214 | /// background; callers must not misreport that temporary empty result as "no |
| 215 | /// matches". |
| 216 | fn mention_menu_entries(app: &mut App, partial: &str, limit: usize) -> (Vec<String>, bool) { |
| 217 | if poll_background_mention_discovery(app) { |
| 218 | app.needs_redraw = true; |
| 219 | } |
| 220 | |
| 221 | let workspace = app.workspace.clone(); |
| 222 | let cwd = app.composer.mention_cwd.clone(); |
| 223 | let walk_depth = app.mention_walk_depth; |
| 224 | let behavior = app.mention_menu_behavior.clone(); |
| 225 | let follow_links = app.workspace_follow_symlinks; |
| 226 | let discovery_key = if behavior == "browser" { |
| 227 | MentionDiscoveryKey::browser( |
| 228 | workspace.clone(), |
| 229 | cwd.clone(), |
| 230 | walk_depth, |
| 231 | follow_links, |
| 232 | partial.to_string(), |
| 233 | ) |
| 234 | } else { |
| 235 | MentionDiscoveryKey::fuzzy(workspace.clone(), cwd.clone(), walk_depth, follow_links) |
| 236 | }; |
| 237 | app.composer |
| 238 | .mention_discovery |
| 239 | .ensure_requested(discovery_key.clone()); |
| 240 | |
| 241 | if let Some(ref cache) = app.composer.mention_completion_cache |
| 242 | && cache.workspace == workspace |
| 243 | && cache.cwd == cwd |
| 244 | && cache.partial == partial |
| 245 | && cache.limit == limit |
| 246 | && cache.walk_depth == walk_depth |
| 247 | && cache.behavior == behavior |
| 248 | && cache.follow_links == follow_links |
| 249 | { |
| 250 | return (cache.entries.clone(), true); |
| 251 | } |
| 252 | |
| 253 | let Some(candidates) = app |
| 254 | .composer |
| 255 | .mention_discovery |
| 256 | .cached_entries(&discovery_key) |
| 257 | else { |
| 258 | return (Vec::new(), false); |
| 259 | }; |
| 260 | let entries = match &discovery_key.behavior { |
| 261 | MentionDiscoveryBehavior::Fuzzy => { |
| 262 | let ranked = crate::working_set::rank_completion_candidates(candidates, partial, limit); |
| 263 | super::file_frecency::rerank_by_frecency(ranked) |
| 264 | } |
| 265 | MentionDiscoveryBehavior::Browser { .. } => { |
| 266 | candidates.iter().take(limit).cloned().collect() |
| 267 | } |
| 268 | }; |
| 269 | |
| 270 | let entries = with_git_mention_entries(entries, partial, limit); |
| 271 | |
| 272 | app.composer.mention_completion_cache = Some(MentionCompletionCache { |
| 273 | workspace, |
| 274 | cwd, |
| 275 | partial: partial.to_string(), |
| 276 | limit, |
| 277 | walk_depth, |
| 278 | behavior, |
| 279 | follow_links, |
| 280 | entries: entries.clone(), |
| 281 | }); |
| 282 | |
| 283 | (entries, true) |
| 284 | } |
| 285 | |
| 286 | /// Prepend the `@git` / `@diff` tokens that prefix-match `partial` to the path |
| 287 | /// completions, so curated git context is discoverable from the same menu as |
| 288 | /// files (#4067). They lead because a two-entry prefix match is what the user |
| 289 | /// meant when they typed `gi` or `di`, and paths still fill the rest. |
| 290 | /// |
| 291 | /// A bare `@` is deliberately left alone: that menu is the file picker, and |
| 292 | /// pushing two fixed tokens above every path would cost a slot on every |
| 293 | /// mention the user makes. The tokens appear as soon as a matching character |
| 294 | /// is typed. |
| 295 | fn with_git_mention_entries(entries: Vec<String>, partial: &str, limit: usize) -> Vec<String> { |
| 296 | // `mention_menu_limit = 0` is a documented way to disable the popup |
| 297 | // entirely. The git tokens are menu entries like any other and must |
| 298 | // respect the same cap, or setting 0 would still pop a one-entry menu. |
| 299 | if limit == 0 { |
| 300 | return Vec::new(); |
| 301 | } |
| 302 | let needle = partial.trim().to_lowercase(); |
| 303 | if needle.is_empty() { |
| 304 | return entries; |
| 305 | } |
| 306 | let matching: Vec<String> = GitMentionKind::iter_all() |
| 307 | .filter(|kind| kind.token().starts_with(&needle)) |
| 308 | .map(|kind| kind.token().to_string()) |
| 309 | .collect(); |
| 310 | if matching.is_empty() { |
| 311 | return entries; |
| 312 | } |
| 313 | let mut combined = matching; |
| 314 | combined.truncate(limit); |
| 315 | for entry in entries { |
| 316 | if combined.len() >= limit { |
| 317 | break; |
| 318 | } |
| 319 | if !combined.contains(&entry) { |
| 320 | combined.push(entry); |
| 321 | } |
| 322 | } |
| 323 | combined |
| 324 | } |
| 325 | |
| 326 | /// Apply the currently selected `@`-mention popup entry to the composer |
| 327 | /// input, splicing it in place of the `@<partial>` token at the cursor. |
| 328 | /// Returns `true` if a substitution occurred. |
| 329 | /// |
| 330 | /// Designed to be invoked by the same keybinding that drives |
| 331 | /// `apply_slash_menu_selection` (Enter / Tab); the caller is responsible |
| 332 | /// for choosing which menu is "active" based on cursor context. |
| 333 | pub fn apply_mention_menu_selection(app: &mut App, entries: &[String]) -> bool { |
| 334 | if entries.is_empty() { |
| 335 | return false; |
| 336 | } |
| 337 | let Some((byte_start, partial)) = |
| 338 | partial_file_mention_at_cursor(&app.input, app.cursor_position) |
| 339 | else { |
| 340 | return false; |
| 341 | }; |
| 342 | let selected_idx = app |
| 343 | .mention_menu_selected |
| 344 | .min(entries.len().saturating_sub(1)); |
| 345 | let replacement = &entries[selected_idx]; |
| 346 | // #441: bump this path's frecency before we splice it in. The store |
| 347 | // persists asynchronously, so this never blocks input handling. |
| 348 | super::file_frecency::record_mention(replacement); |
| 349 | replace_file_mention(app, byte_start, &partial, replacement); |
| 350 | app.mention_menu_hidden = false; |
| 351 | app.status_message = Some(format!("Attached @{replacement}")); |
| 352 | true |
| 353 | } |
| 354 | |
| 355 | /// Tab-completion handler for `@file` mentions. Mirrors the slash-command |
| 356 | /// flow: a single match is applied directly; multiple matches with a longer |
| 357 | /// shared prefix extend the partial; otherwise the first few candidates are |
| 358 | /// surfaced via the status line. Returns true when the input was modified or |
| 359 | /// a suggestion was offered, so the caller can short-circuit other handlers. |
| 360 | pub fn try_autocomplete_file_mention(app: &mut App) -> bool { |
| 361 | let Some((byte_start, partial)) = |
| 362 | partial_file_mention_at_cursor(&app.input, app.cursor_position) |
| 363 | else { |
| 364 | return false; |
| 365 | }; |
| 366 | let (candidates, ready) = mention_menu_entries(app, &partial, FILE_MENTION_COMPLETION_LIMIT); |
| 367 | if !ready { |
| 368 | return true; |
| 369 | } |
| 370 | if candidates.is_empty() { |
| 371 | app.status_message = Some(no_file_mention_matches_status( |
| 372 | &partial, |
| 373 | app.mention_walk_depth, |
| 374 | )); |
| 375 | return true; |
| 376 | } |
| 377 | if candidates.len() == 1 { |
| 378 | // #441: a unique-match completion is also a "mention" for ranking. |
| 379 | super::file_frecency::record_mention(&candidates[0]); |
| 380 | replace_file_mention(app, byte_start, &partial, &candidates[0]); |
| 381 | app.status_message = Some(format!("Attached @{}", candidates[0])); |
| 382 | return true; |
| 383 | } |
| 384 | let candidate_refs: Vec<&str> = candidates.iter().map(String::as_str).collect(); |
| 385 | let shared = longest_common_prefix(&candidate_refs); |
| 386 | if shared.len() > partial.len() { |
| 387 | replace_file_mention(app, byte_start, &partial, shared); |
| 388 | app.status_message = Some(format!("@{shared}…")); |
| 389 | return true; |
| 390 | } |
| 391 | let preview = candidates |
| 392 | .iter() |
| 393 | .take(5) |
| 394 | .map(|c| format!("@{c}")) |
| 395 | .collect::<Vec<_>>() |
| 396 | .join(", "); |
| 397 | app.status_message = Some(format!("Matches: {preview}")); |
| 398 | true |
| 399 | } |
| 400 | |
| 401 | fn no_file_mention_matches_status(partial: &str, walk_depth: usize) -> String { |
| 402 | if path_partial_reaches_walk_depth(partial, walk_depth) { |
| 403 | format!( |
| 404 | "No files match @{partial} (mention_walk_depth={walk_depth}; use /config set mention_walk_depth 0 to search deeper)" |
| 405 | ) |
| 406 | } else { |
| 407 | format!("No files match @{partial}") |
| 408 | } |
| 409 | } |
| 410 | |
| 411 | fn path_partial_reaches_walk_depth(partial: &str, walk_depth: usize) -> bool { |
| 412 | if walk_depth == 0 { |
| 413 | return false; |
| 414 | } |
| 415 | let component_count = partial |
| 416 | .split(['/', '\\']) |
| 417 | .filter(|component| !component.is_empty()) |
| 418 | .count(); |
| 419 | component_count >= walk_depth |
| 420 | } |
| 421 | |
| 422 | /// Splice a completion into the input, replacing the `@<partial>` token at |
| 423 | /// `byte_start` with `@<replacement>`. Cursor moves to the end of the new |
| 424 | /// token so further keystrokes extend (or escape via space) naturally. |
| 425 | fn replace_file_mention(app: &mut App, byte_start: usize, partial: &str, replacement: &str) { |
| 426 | let original_token_len = '@'.len_utf8() + partial.len(); |
| 427 | let original_token_end = byte_start + original_token_len; |
| 428 | let mut new_input = |
| 429 | String::with_capacity(app.input.len() - original_token_len + 1 + replacement.len()); |
| 430 | new_input.push_str(&app.input[..byte_start]); |
| 431 | new_input.push('@'); |
| 432 | new_input.push_str(replacement); |
| 433 | if original_token_end < app.input.len() { |
| 434 | new_input.push_str(&app.input[original_token_end..]); |
| 435 | } |
| 436 | let new_cursor_chars = |
| 437 | app.input[..byte_start].chars().count() + 1 + replacement.chars().count(); |
| 438 | app.input = new_input; |
| 439 | app.cursor_position = new_cursor_chars; |
| 440 | } |
| 441 | |
| 442 | pub fn longest_common_prefix<'a>(values: &[&'a str]) -> &'a str { |
| 443 | let Some(first) = values.first().copied() else { |
| 444 | return ""; |
| 445 | }; |
| 446 | let mut end = first.len(); |
| 447 | |
| 448 | for value in values.iter().skip(1) { |
| 449 | while end > 0 && !value.starts_with(&first[..end]) { |
| 450 | end -= 1; |
| 451 | // Ensure we land on a valid UTF-8 char boundary. |
| 452 | while end > 0 && !first.is_char_boundary(end) { |
| 453 | end -= 1; |
| 454 | } |
| 455 | } |
| 456 | if end == 0 { |
| 457 | return ""; |
| 458 | } |
| 459 | } |
| 460 | |
| 461 | &first[..end] |
| 462 | } |
| 463 | |
| 464 | // --------------------------------------------------------------------------- |
| 465 | // Expansion at send-time |
| 466 | // --------------------------------------------------------------------------- |
| 467 | |
| 468 | /// Append a "Local context from @mentions" block to the user's message when |
| 469 | /// any `@path` references are present. Returns the input unchanged when |
| 470 | /// there are none. |
| 471 | /// |
| 472 | /// `cwd` carries the user's launch directory and drives the second |
| 473 | /// resolution pass (issue #101): relative `@<path>` mentions resolve under |
| 474 | /// `cwd` when `workspace.join(path)` doesn't exist, so the user's mental |
| 475 | /// anchor (their shell's pwd) wins when it diverges from `--workspace`. |
| 476 | /// Pass `None` to disable the cwd pass entirely (workspace-only). |
| 477 | /// |
| 478 | /// Resolution here never walks the tree on submit (#4365). A miss on the |
| 479 | /// exact two-pass lookup falls back to a bounded, unique-match-only search of |
| 480 | /// the composer's already-built background completion index |
| 481 | /// (`completion_index`, when the caller has one cached); the winning |
| 482 | /// candidate must still exist on disk. Ambiguous, stale, or absent matches |
| 483 | /// stay an honest `<missing-file>` that names only what the user typed — |
| 484 | /// never a fabricated workspace-root path. |
| 485 | /// |
| 486 | /// Convenience wrapper that allocates a throwaway cache. Test-only: the real |
| 487 | /// send paths share one cache across the references and payload passes. |
| 488 | #[cfg(test)] |
| 489 | pub fn user_request_with_file_mentions( |
| 490 | input: &str, |
| 491 | workspace: &Path, |
| 492 | cwd: Option<PathBuf>, |
| 493 | ) -> String { |
| 494 | user_request_with_file_mentions_cached( |
| 495 | input, |
| 496 | workspace, |
| 497 | cwd, |
| 498 | &mut GitMentionCache::default(), |
| 499 | None, |
| 500 | ) |
| 501 | } |
| 502 | |
| 503 | pub fn user_request_with_file_mentions_cached( |
| 504 | input: &str, |
| 505 | workspace: &Path, |
| 506 | cwd: Option<PathBuf>, |
| 507 | git_cache: &mut GitMentionCache, |
| 508 | completion_index: Option<&[String]>, |
| 509 | ) -> String { |
| 510 | let Some(context) = |
| 511 | local_context_from_file_mentions(input, workspace, cwd, git_cache, completion_index) |
| 512 | else { |
| 513 | return input.to_string(); |
| 514 | }; |
| 515 | format!("{input}\n\n---\n\nLocal context from @mentions:\n{context}") |
| 516 | } |
| 517 | |
| 518 | #[must_use] |
| 519 | pub fn pending_context_previews(input: &str) -> Vec<FileMentionPreview> { |
| 520 | let mut previews = Vec::new(); |
| 521 | let mut seen = std::collections::HashSet::new(); |
| 522 | for mention in extract_file_mentions(input) |
| 523 | .into_iter() |
| 524 | .take(MAX_FILE_MENTIONS_PER_MESSAGE) |
| 525 | { |
| 526 | if !seen.insert(mention.clone()) { |
| 527 | continue; |
| 528 | } |
| 529 | // Composer previews stay lexical (no git subprocess from the render |
| 530 | // loop, same rule as #4365 for path stats); the payload is resolved |
| 531 | // once at submit time. |
| 532 | if let Some(kind) = git_mention::git_mention_kind(&mention) { |
| 533 | previews.push(FileMentionPreview { |
| 534 | kind: "git".to_string(), |
| 535 | label: kind.label().to_string(), |
| 536 | detail: Some("resolved on send".to_string()), |
| 537 | included: false, |
| 538 | removable: false, |
| 539 | }); |
| 540 | continue; |
| 541 | } |
| 542 | let media = is_media_path(Path::new(&mention)); |
| 543 | previews.push(FileMentionPreview { |
| 544 | kind: if media { "media" } else { "mention" }.to_string(), |
| 545 | label: mention, |
| 546 | detail: Some(if media { |
| 547 | "use /attach for media bytes".to_string() |
| 548 | } else { |
| 549 | "resolved on send".to_string() |
| 550 | }), |
| 551 | // Lexical preview deliberately does not stat the path while the |
| 552 | // user types. Exact inclusion/missing metadata is resolved once, |
| 553 | // at submit time, rather than from the render loop (#4365). |
| 554 | included: false, |
| 555 | removable: false, |
| 556 | }); |
| 557 | } |
| 558 | |
| 559 | for attachment in extract_media_attachment_references(input) { |
| 560 | previews.push(FileMentionPreview { |
| 561 | kind: attachment.kind, |
| 562 | label: attachment.path, |
| 563 | detail: Some("attached media".to_string()), |
| 564 | included: true, |
| 565 | removable: true, |
| 566 | }); |
| 567 | } |
| 568 | previews |
| 569 | } |
| 570 | |
| 571 | /// Convenience wrapper that allocates a throwaway cache. Test-only, as above. |
| 572 | #[cfg(test)] |
| 573 | #[must_use] |
| 574 | pub fn context_references_from_input( |
| 575 | input: &str, |
| 576 | workspace: &Path, |
| 577 | cwd: Option<PathBuf>, |
| 578 | ) -> Vec<ContextReference> { |
| 579 | context_references_from_input_cached( |
| 580 | input, |
| 581 | workspace, |
| 582 | cwd, |
| 583 | &mut GitMentionCache::default(), |
| 584 | None, |
| 585 | ) |
| 586 | } |
| 587 | |
| 588 | #[must_use] |
| 589 | pub fn context_references_from_input_cached( |
| 590 | input: &str, |
| 591 | workspace: &Path, |
| 592 | cwd: Option<PathBuf>, |
| 593 | git_cache: &mut GitMentionCache, |
| 594 | completion_index: Option<&[String]>, |
| 595 | ) -> Vec<ContextReference> { |
| 596 | let mut references = Vec::new(); |
| 597 | let mut seen = std::collections::HashSet::new(); |
| 598 | let ws = Workspace::with_cwd(workspace.to_path_buf(), cwd); |
| 599 | |
| 600 | for mention in extract_file_mentions(input) |
| 601 | .into_iter() |
| 602 | .take(MAX_FILE_MENTIONS_PER_MESSAGE) |
| 603 | { |
| 604 | // Git mentions resolve against the working tree, not the path index, |
| 605 | // so the inspector reports their real size and budget (#4067). |
| 606 | if let Some(kind) = git_mention::git_mention_kind(&mention) { |
| 607 | let payload = git_cache.resolve(kind, workspace).clone(); |
| 608 | let detail = match payload.unavailable_reason.as_deref() { |
| 609 | Some(reason) => format!("{}, {reason}", kind.label()), |
| 610 | None if payload.truncated => format!( |
| 611 | "{}, {} bytes truncated at {} budget", |
| 612 | kind.label(), |
| 613 | payload.bytes, |
| 614 | kind.byte_budget() |
| 615 | ), |
| 616 | None => format!("{}, {} bytes", kind.label(), payload.bytes), |
| 617 | }; |
| 618 | let reference = ContextReference { |
| 619 | kind: ContextReferenceKind::GitContext, |
| 620 | source: ContextReferenceSource::AtMention, |
| 621 | badge: "git".to_string(), |
| 622 | label: kind.token().to_string(), |
| 623 | target: workspace.display().to_string(), |
| 624 | included: payload.unavailable_reason.is_none(), |
| 625 | expanded: false, |
| 626 | detail: Some(detail), |
| 627 | }; |
| 628 | if seen.insert(format!("git-mention:{}", kind.token())) { |
| 629 | references.push(reference); |
| 630 | } |
| 631 | continue; |
| 632 | } |
| 633 | |
| 634 | let (path, display_path, exists) = |
| 635 | resolve_mention_for_send(&ws, &mention, completion_index); |
| 636 | let reference = context_reference_for_mention(&mention, &path, &display_path, exists); |
| 637 | if !seen.insert(format!( |
| 638 | "{:?}:{:?}:{}:{}", |
| 639 | reference.source, reference.kind, reference.target, reference.label |
| 640 | )) { |
| 641 | continue; |
| 642 | } |
| 643 | references.push(reference); |
| 644 | } |
| 645 | |
| 646 | for reference in extract_media_attachment_references(input) { |
| 647 | let context_reference = ContextReference { |
| 648 | kind: ContextReferenceKind::MediaAttachment, |
| 649 | source: ContextReferenceSource::Attachment, |
| 650 | badge: reference.kind, |
| 651 | label: reference.path.clone(), |
| 652 | target: reference.path, |
| 653 | included: true, |
| 654 | expanded: false, |
| 655 | detail: Some("attached media".to_string()), |
| 656 | }; |
| 657 | if !seen.insert(format!( |
| 658 | "{:?}:{:?}:{}:{}", |
| 659 | context_reference.source, |
| 660 | context_reference.kind, |
| 661 | context_reference.target, |
| 662 | context_reference.label |
| 663 | )) { |
| 664 | continue; |
| 665 | } |
| 666 | references.push(context_reference); |
| 667 | } |
| 668 | |
| 669 | references |
| 670 | } |
| 671 | |
| 672 | fn context_reference_for_mention( |
| 673 | raw: &str, |
| 674 | path: &Path, |
| 675 | display_path: &str, |
| 676 | exists: bool, |
| 677 | ) -> ContextReference { |
| 678 | if !exists { |
| 679 | return ContextReference { |
| 680 | kind: ContextReferenceKind::Missing, |
| 681 | source: ContextReferenceSource::AtMention, |
| 682 | badge: "missing".to_string(), |
| 683 | label: raw.to_string(), |
| 684 | // No resolved target exists; naming the workspace-root guess here |
| 685 | // would present a path we already know is wrong. |
| 686 | target: raw.to_string(), |
| 687 | included: false, |
| 688 | expanded: false, |
| 689 | detail: Some("not found".to_string()), |
| 690 | }; |
| 691 | } |
| 692 | if path.is_dir() { |
| 693 | return ContextReference { |
| 694 | kind: ContextReferenceKind::Directory, |
| 695 | source: ContextReferenceSource::AtMention, |
| 696 | badge: "dir".to_string(), |
| 697 | label: raw.to_string(), |
| 698 | target: display_path.to_string(), |
| 699 | included: true, |
| 700 | expanded: true, |
| 701 | detail: Some("directory listing".to_string()), |
| 702 | }; |
| 703 | } |
| 704 | if !path.is_file() { |
| 705 | return ContextReference { |
| 706 | kind: ContextReferenceKind::Unsupported, |
| 707 | source: ContextReferenceSource::AtMention, |
| 708 | badge: "skipped".to_string(), |
| 709 | label: raw.to_string(), |
| 710 | target: display_path.to_string(), |
| 711 | included: false, |
| 712 | expanded: false, |
| 713 | detail: Some("unsupported path".to_string()), |
| 714 | }; |
| 715 | } |
| 716 | if is_media_path(path) { |
| 717 | return ContextReference { |
| 718 | kind: ContextReferenceKind::MediaMention, |
| 719 | source: ContextReferenceSource::AtMention, |
| 720 | badge: "media".to_string(), |
| 721 | label: raw.to_string(), |
| 722 | target: display_path.to_string(), |
| 723 | included: false, |
| 724 | expanded: false, |
| 725 | detail: Some("use /attach for media bytes".to_string()), |
| 726 | }; |
| 727 | } |
| 728 | |
| 729 | let detail = match std::fs::metadata(path) { |
| 730 | Ok(metadata) if metadata.len() > MAX_MENTION_FILE_BYTES => { |
| 731 | Some("included truncated".to_string()) |
| 732 | } |
| 733 | Ok(_) => Some("included".to_string()), |
| 734 | Err(err) => Some(format!("metadata: {err}")), |
| 735 | }; |
| 736 | |
| 737 | ContextReference { |
| 738 | kind: ContextReferenceKind::File, |
| 739 | source: ContextReferenceSource::AtMention, |
| 740 | badge: "file".to_string(), |
| 741 | label: raw.to_string(), |
| 742 | target: display_path.to_string(), |
| 743 | included: true, |
| 744 | expanded: true, |
| 745 | detail: detail.or_else(|| Some(display_path.to_string())), |
| 746 | } |
| 747 | } |
| 748 | |
| 749 | #[derive(Debug, Clone, PartialEq, Eq)] |
| 750 | pub struct MediaAttachmentReference { |
| 751 | pub kind: String, |
| 752 | pub path: String, |
| 753 | pub start_byte: usize, |
| 754 | pub end_byte: usize, |
| 755 | } |
| 756 | |
| 757 | pub fn media_attachment_references(input: &str) -> Vec<MediaAttachmentReference> { |
| 758 | let mut out = Vec::new(); |
| 759 | let mut offset = 0usize; |
| 760 | for line in input.split_inclusive('\n') { |
| 761 | let start_byte = offset; |
| 762 | let end_byte = offset + line.len(); |
| 763 | offset = end_byte; |
| 764 | let trimmed = line.trim(); |
| 765 | let Some(body) = trimmed |
| 766 | .strip_prefix("[Attached ") |
| 767 | .and_then(|value| value.strip_suffix(']')) |
| 768 | else { |
| 769 | continue; |
| 770 | }; |
| 771 | let Some((kind, rest)) = body.split_once(": ") else { |
| 772 | continue; |
| 773 | }; |
| 774 | let path = rest |
| 775 | .rsplit_once(" at ") |
| 776 | .map_or(rest, |(_, path)| path) |
| 777 | .trim(); |
| 778 | if !path.is_empty() { |
| 779 | out.push(MediaAttachmentReference { |
| 780 | kind: kind.trim().to_string(), |
| 781 | path: path.to_string(), |
| 782 | start_byte, |
| 783 | end_byte, |
| 784 | }); |
| 785 | } |
| 786 | } |
| 787 | out |
| 788 | } |
| 789 | |
| 790 | fn extract_media_attachment_references(input: &str) -> Vec<MediaAttachmentReference> { |
| 791 | media_attachment_references(input) |
| 792 | } |
| 793 | |
| 794 | fn local_context_from_file_mentions( |
| 795 | input: &str, |
| 796 | workspace: &Path, |
| 797 | cwd: Option<PathBuf>, |
| 798 | git_cache: &mut GitMentionCache, |
| 799 | completion_index: Option<&[String]>, |
| 800 | ) -> Option<String> { |
| 801 | let mentions = extract_file_mentions(input); |
| 802 | if mentions.is_empty() { |
| 803 | return None; |
| 804 | } |
| 805 | |
| 806 | let mut blocks = Vec::new(); |
| 807 | let mut seen = std::collections::HashSet::new(); |
| 808 | let ws = Workspace::with_cwd(workspace.to_path_buf(), cwd); |
| 809 | |
| 810 | for mention in mentions.into_iter().take(MAX_FILE_MENTIONS_PER_MESSAGE) { |
| 811 | // `@git` / `@diff` resolve to curated git context, not to a path, so |
| 812 | // they short-circuit before any workspace path resolution (#4067). |
| 813 | if let Some(kind) = git_mention::git_mention_kind(&mention) { |
| 814 | if !seen.insert(format!("git-mention:{}", kind.token())) { |
| 815 | continue; |
| 816 | } |
| 817 | blocks.push(git_cache.resolve(kind, workspace).block.clone()); |
| 818 | continue; |
| 819 | } |
| 820 | |
| 821 | // `Workspace::resolve_exact` already returns absolute paths when the root |
| 822 | // is absolute (TUI always runs from an absolute workspace), so we |
| 823 | // skip `canonicalize()` here — it's per-mention I/O on the |
| 824 | // message-send hot path. Accept the rare symlink-aliasing dedup |
| 825 | // miss as the cost of avoiding a syscall (Gemini code-review). |
| 826 | let (path, display_path, exists) = |
| 827 | resolve_mention_for_send(&ws, &mention, completion_index); |
| 828 | tracing::debug!( |
| 829 | target: "codewhale_tui::file_mention", |
| 830 | raw_typed = %mention, |
| 831 | workspace = %workspace.display(), |
| 832 | cwd = ?std::env::current_dir().ok(), |
| 833 | resolved = %display_path, |
| 834 | exists, |
| 835 | "file mention resolution", |
| 836 | ); |
| 837 | |
| 838 | // Gate every block — including <missing-file> — through the dedup |
| 839 | // set so a user typing the same non-existent file twice doesn't |
| 840 | // waste tokens on duplicate missing-file blocks (Devin code-review). |
| 841 | // Missing mentions dedup on the typed token: there is no resolved |
| 842 | // path to key on, and the workspace-root guess must not become one. |
| 843 | let dedup_key = if exists { |
| 844 | display_path.clone() |
| 845 | } else { |
| 846 | format!("missing:{mention}") |
| 847 | }; |
| 848 | if !seen.insert(dedup_key) { |
| 849 | continue; |
| 850 | } |
| 851 | |
| 852 | if exists { |
| 853 | blocks.push(render_file_mention_context(&mention, &path, &display_path)); |
| 854 | } else { |
| 855 | // Honest miss: name only what the user typed. Emitting the |
| 856 | // workspace-root join as `path=` presented a non-existent file |
| 857 | // as if it were the resolved target. |
| 858 | blocks.push(format!("<missing-file mention=\"@{mention}\" />")); |
| 859 | } |
| 860 | } |
| 861 | |
| 862 | if blocks.is_empty() { |
| 863 | None |
| 864 | } else { |
| 865 | Some(blocks.join("\n\n")) |
| 866 | } |
| 867 | } |
| 868 | |
| 869 | /// Send-time mention resolution: exact two-pass lookup first (workspace root, |
| 870 | /// then launch cwd), then a bounded fallback against the composer's cached |
| 871 | /// background completion index. Returns the path, its display form, and |
| 872 | /// whether it exists. On a miss the returned path is the workspace-root guess |
| 873 | /// — callers must not present it as resolved when `exists` is false. |
| 874 | fn resolve_mention_for_send( |
| 875 | ws: &Workspace, |
| 876 | mention: &str, |
| 877 | completion_index: Option<&[String]>, |
| 878 | ) -> (PathBuf, String, bool) { |
| 879 | let guess = match ws.resolve_exact(mention) { |
| 880 | Ok(path) => { |
| 881 | let display = path.display().to_string(); |
| 882 | return (path, display, true); |
| 883 | } |
| 884 | Err(guess) => guess, |
| 885 | }; |
| 886 | if let Some(resolved) = completion_index |
| 887 | .and_then(|candidates| resolve_mention_in_completion_index(mention, candidates, ws)) |
| 888 | { |
| 889 | let display = resolved.display().to_string(); |
| 890 | return (resolved, display, true); |
| 891 | } |
| 892 | let display = guess.display().to_string(); |
| 893 | (guess, display, false) |
| 894 | } |
| 895 | |
| 896 | /// Bounded send-time fallback for `@`-mention misses. |
| 897 | /// |
| 898 | /// #4365 keeps filesystem walks off the submit path, so instead of walking we |
| 899 | /// match the typed token against the composer's already-built background |
| 900 | /// completion index (workspace- or cwd-relative display strings). A candidate |
| 901 | /// wins only when it is the *unique* path-suffix or basename match and the |
| 902 | /// winning path still resolves on disk — the index may be a few seconds |
| 903 | /// stale. Anything else (no hit, ambiguous hit, stale hit) returns `None` so |
| 904 | /// the caller emits an honest `<missing-file>` instead of attaching an |
| 905 | /// arbitrary same-name file from a nested directory. |
| 906 | fn resolve_mention_in_completion_index( |
| 907 | mention: &str, |
| 908 | candidates: &[String], |
| 909 | ws: &Workspace, |
| 910 | ) -> Option<PathBuf> { |
| 911 | // Absolute and home-anchored mentions name an exact location; a basename |
| 912 | // lookalike elsewhere in the tree would be a different file, not a fix-up. |
| 913 | // A leading separator is rooted on every platform, but `Path::is_absolute` |
| 914 | // is false for `/foo` on Windows (no drive prefix), which would otherwise |
| 915 | // let a rooted miss fall through to the index and attach an unrelated |
| 916 | // same-name file. Test the root marker directly so the guard holds there. |
| 917 | // `\` is a root marker only on Windows; on Unix it is an ordinary |
| 918 | // (if unusual) leading filename character, so leave that case alone. |
| 919 | let rooted = mention.starts_with('/') || (cfg!(windows) && mention.starts_with('\\')); |
| 920 | if mention.starts_with('~') || rooted || Path::new(mention).is_absolute() { |
| 921 | return None; |
| 922 | } |
| 923 | let needle = mention.replace('\\', "/"); |
| 924 | let needle = needle.trim_matches('/'); |
| 925 | if needle.is_empty() { |
| 926 | return None; |
| 927 | } |
| 928 | let needle_lower = needle.to_lowercase(); |
| 929 | let basename_lower = needle_lower.rsplit('/').next()?; |
| 930 | |
| 931 | let normalized = |candidate: &str| candidate.replace('\\', "/"); |
| 932 | let suffix_match = |candidate: &str| { |
| 933 | let cand = normalized(candidate); |
| 934 | let cand = cand.trim_end_matches('/').to_lowercase(); |
| 935 | !cand.is_empty() && (cand == needle_lower || cand.ends_with(&format!("/{needle_lower}"))) |
| 936 | }; |
| 937 | let basename_match = |candidate: &str| { |
| 938 | let cand = normalized(candidate); |
| 939 | let cand = cand.trim_end_matches('/').to_lowercase(); |
| 940 | !cand.is_empty() && cand.rsplit('/').next() == Some(basename_lower) |
| 941 | }; |
| 942 | |
| 943 | // Prefer path-suffix hits: they carry the user's typed directory context. |
| 944 | // Fall back to basename hits only when no suffix hit exists. Either way, |
| 945 | // more than one distinct winner is ambiguous and resolves nothing. |
| 946 | let predicates: [&dyn Fn(&str) -> bool; 2] = [&suffix_match, &basename_match]; |
| 947 | for predicate in predicates { |
| 948 | let mut winner: Option<&str> = None; |
| 949 | for candidate in candidates { |
| 950 | if !predicate(candidate) { |
| 951 | continue; |
| 952 | } |
| 953 | match winner { |
| 954 | None => winner = Some(candidate.as_str()), |
| 955 | Some(existing) if existing == candidate.as_str() => {} |
| 956 | Some(_) => return None, |
| 957 | } |
| 958 | } |
| 959 | if let Some(winner) = winner { |
| 960 | // The index can be stale; only a path that still resolves exists. |
| 961 | // Display strings are workspace- or cwd-relative, which |
| 962 | // `resolve_exact` re-anchors exactly like a typed path. |
| 963 | // |
| 964 | // Rejoin the components with this platform's separator first. |
| 965 | // Index strings are `/`-separated, and `root.join("ops/f.md")` |
| 966 | // keeps that slash verbatim on Windows, yielding a mixed |
| 967 | // `C:\ws\ops/f.md` that we then hand to the model and print in |
| 968 | // the context inspector. Same path, inconsistent rendering. |
| 969 | let native: PathBuf = winner.split('/').filter(|part| !part.is_empty()).collect(); |
| 970 | return ws.resolve_exact(&native.to_string_lossy()).ok(); |
| 971 | } |
| 972 | } |
| 973 | None |
| 974 | } |
| 975 | |
| 976 | fn extract_file_mentions(input: &str) -> Vec<String> { |
| 977 | let chars: Vec<char> = input.chars().collect(); |
| 978 | let mut mentions = Vec::new(); |
| 979 | let mut idx = 0; |
| 980 | |
| 981 | while idx < chars.len() { |
| 982 | if chars[idx] != '@' || !is_file_mention_start(&chars, idx) { |
| 983 | idx += 1; |
| 984 | continue; |
| 985 | } |
| 986 | |
| 987 | let Some(next) = chars.get(idx + 1).copied() else { |
| 988 | break; |
| 989 | }; |
| 990 | if next.is_whitespace() { |
| 991 | idx += 1; |
| 992 | continue; |
| 993 | } |
| 994 | |
| 995 | if matches!(next, '"' | '\'') { |
| 996 | let quote = next; |
| 997 | let mut end = idx + 2; |
| 998 | let mut raw = String::new(); |
| 999 | while end < chars.len() && chars[end] != quote { |
| 1000 | raw.push(chars[end]); |
| 1001 | end += 1; |
| 1002 | } |
| 1003 | if !raw.trim().is_empty() { |
| 1004 | mentions.push(raw.trim().to_string()); |
| 1005 | } |
| 1006 | idx = end.saturating_add(1); |
| 1007 | continue; |
| 1008 | } |
| 1009 | |
| 1010 | let mut end = idx + 1; |
| 1011 | let mut raw = String::new(); |
| 1012 | while end < chars.len() && !chars[end].is_whitespace() { |
| 1013 | raw.push(chars[end]); |
| 1014 | end += 1; |
| 1015 | } |
| 1016 | let trimmed = trim_unquoted_mention(&raw); |
| 1017 | if !trimmed.is_empty() { |
| 1018 | mentions.push(trimmed.to_string()); |
| 1019 | } |
| 1020 | idx = end; |
| 1021 | } |
| 1022 | |
| 1023 | mentions |
| 1024 | } |
| 1025 | |
| 1026 | fn is_file_mention_start(chars: &[char], idx: usize) -> bool { |
| 1027 | if idx == 0 { |
| 1028 | return true; |
| 1029 | } |
| 1030 | chars |
| 1031 | .get(idx.saturating_sub(1)) |
| 1032 | .is_some_and(|ch| ch.is_whitespace() || matches!(ch, '(' | '[' | '{' | '<' | '"' | '\'')) |
| 1033 | } |
| 1034 | |
| 1035 | fn trim_unquoted_mention(raw: &str) -> &str { |
| 1036 | let mut trimmed = raw.trim(); |
| 1037 | while trimmed.chars().count() > 1 |
| 1038 | && trimmed |
| 1039 | .chars() |
| 1040 | .last() |
| 1041 | .is_some_and(|ch| matches!(ch, ',' | ';' | ':' | '!' | '?' | ')' | ']' | '}')) |
| 1042 | { |
| 1043 | trimmed = &trimmed[..trimmed.len() - trimmed.chars().last().unwrap().len_utf8()]; |
| 1044 | } |
| 1045 | trimmed |
| 1046 | } |
| 1047 | |
| 1048 | fn render_file_mention_context(raw: &str, path: &Path, display_path: &str) -> String { |
| 1049 | if !path.exists() { |
| 1050 | return format!("<missing-file mention=\"@{raw}\" path=\"{display_path}\" />"); |
| 1051 | } |
| 1052 | if path.is_dir() { |
| 1053 | return render_directory_mention_context(raw, path, display_path); |
| 1054 | } |
| 1055 | if !path.is_file() { |
| 1056 | return format!("<unsupported-path mention=\"@{raw}\" path=\"{display_path}\" />"); |
| 1057 | } |
| 1058 | if is_media_path(path) { |
| 1059 | return format!( |
| 1060 | "<media-file mention=\"@{raw}\" path=\"{display_path}\">\nUse /attach {raw} when the intent is to attach this image or video to the next message.\n</media-file>" |
| 1061 | ); |
| 1062 | } |
| 1063 | |
| 1064 | match read_text_prefix(path) { |
| 1065 | Ok((text, truncated)) => { |
| 1066 | let truncated_attr = if truncated { " truncated=\"true\"" } else { "" }; |
| 1067 | format!( |
| 1068 | "<file mention=\"@{raw}\" path=\"{display_path}\"{truncated_attr}>\n{text}\n</file>" |
| 1069 | ) |
| 1070 | } |
| 1071 | Err(err) => { |
| 1072 | format!( |
| 1073 | "<unreadable-file mention=\"@{raw}\" path=\"{display_path}\">\n{err}\n</unreadable-file>" |
| 1074 | ) |
| 1075 | } |
| 1076 | } |
| 1077 | } |
| 1078 | |
| 1079 | fn render_directory_mention_context(raw: &str, path: &Path, display_path: &str) -> String { |
| 1080 | let entries = match std::fs::read_dir(path) { |
| 1081 | Ok(entries) => entries, |
| 1082 | Err(err) => { |
| 1083 | return format!( |
| 1084 | "<unreadable-directory mention=\"@{raw}\" path=\"{display_path}\">\n{err}\n</unreadable-directory>" |
| 1085 | ); |
| 1086 | } |
| 1087 | }; |
| 1088 | |
| 1089 | let mut names = entries |
| 1090 | .filter_map(|entry| entry.ok()) |
| 1091 | .map(|entry| { |
| 1092 | let marker = entry |
| 1093 | .file_type() |
| 1094 | .ok() |
| 1095 | .filter(|ty| ty.is_dir()) |
| 1096 | .map_or("", |_| "/"); |
| 1097 | format!("{}{}", entry.file_name().to_string_lossy(), marker) |
| 1098 | }) |
| 1099 | .collect::<Vec<_>>(); |
| 1100 | names.sort(); |
| 1101 | let total = names.len(); |
| 1102 | names.truncate(MAX_DIRECTORY_MENTION_ENTRIES); |
| 1103 | let mut body = names.join("\n"); |
| 1104 | if total > MAX_DIRECTORY_MENTION_ENTRIES { |
| 1105 | let omitted = total - MAX_DIRECTORY_MENTION_ENTRIES; |
| 1106 | let _ = write!(body, "\n... {omitted} more entries"); |
| 1107 | } |
| 1108 | format!("<directory mention=\"@{raw}\" path=\"{display_path}\">\n{body}\n</directory>") |
| 1109 | } |
| 1110 | |
| 1111 | fn read_text_prefix(path: &Path) -> std::io::Result<(String, bool)> { |
| 1112 | let mut file = std::fs::File::open(path)?; |
| 1113 | let mut buffer = Vec::new(); |
| 1114 | file.by_ref() |
| 1115 | .take(MAX_MENTION_FILE_BYTES + 1) |
| 1116 | .read_to_end(&mut buffer)?; |
| 1117 | let truncated = buffer.len() as u64 > MAX_MENTION_FILE_BYTES; |
| 1118 | if truncated { |
| 1119 | buffer.truncate(MAX_MENTION_FILE_BYTES as usize); |
| 1120 | // Round down to the nearest valid UTF-8 character boundary so a |
| 1121 | // multi-byte sequence (CJK, emoji, etc.) is never split at the cut point. |
| 1122 | // Only adjust when error_len() is None — that means truncation landed |
| 1123 | // mid-sequence (incomplete tail). A Some(_) error_len means the file |
| 1124 | // genuinely contains invalid UTF-8 bytes; leave the buffer intact so |
| 1125 | // the from_utf8 call below returns the correct "file is not UTF-8" error. |
| 1126 | if let Err(e) = std::str::from_utf8(&buffer) |
| 1127 | && e.error_len().is_none() |
| 1128 | { |
| 1129 | buffer.truncate(e.valid_up_to()); |
| 1130 | } |
| 1131 | } |
| 1132 | if buffer.contains(&0) { |
| 1133 | return Err(std::io::Error::new( |
| 1134 | std::io::ErrorKind::InvalidData, |
| 1135 | "file appears to be binary", |
| 1136 | )); |
| 1137 | } |
| 1138 | let text = std::str::from_utf8(&buffer) |
| 1139 | .map_err(|_| std::io::Error::new(std::io::ErrorKind::InvalidData, "file is not UTF-8"))? |
| 1140 | .to_string(); |
| 1141 | Ok((text, truncated)) |
| 1142 | } |
| 1143 | |
| 1144 | fn is_media_path(path: &Path) -> bool { |
| 1145 | let Some(ext) = path.extension().and_then(|ext| ext.to_str()) else { |
| 1146 | return false; |
| 1147 | }; |
| 1148 | matches!( |
| 1149 | ext.to_ascii_lowercase().as_str(), |
| 1150 | "png" |
| 1151 | | "jpg" |
| 1152 | | "jpeg" |
| 1153 | | "gif" |
| 1154 | | "webp" |
| 1155 | | "bmp" |
| 1156 | | "tif" |
| 1157 | | "tiff" |
| 1158 | | "ppm" |
| 1159 | | "mp4" |
| 1160 | | "mov" |
| 1161 | | "m4v" |
| 1162 | | "webm" |
| 1163 | | "avi" |
| 1164 | | "mkv" |
| 1165 | ) |
| 1166 | } |
| 1167 | |
| 1168 | // --------------------------------------------------------------------------- |
| 1169 | // #101 regression repros |
| 1170 | // --------------------------------------------------------------------------- |
| 1171 | // |
| 1172 | // The bug being guarded: typing `@<some/file>` resolved under `--workspace`, |
| 1173 | // not the user's launch CWD. When the two diverged (the canonical case is |
| 1174 | // `--workspace=/repo` with `pwd=/repo/sub`), every relative `@` token routed |
| 1175 | // to the wrong root and the prompt got `<missing-file>` blocks. |
| 1176 | #[cfg(test)] |
| 1177 | mod tests { |
| 1178 | use super::*; |
| 1179 | use tempfile::TempDir; |
| 1180 | |
| 1181 | /// #101 regression — workspace-vs-cwd divergence: `@bar.txt` typed from |
| 1182 | /// the cwd `<root>/sub` MUST resolve to `<root>/sub/bar.txt`, never to |
| 1183 | /// `<root>/bar.txt` (which doesn't exist). |
| 1184 | #[test] |
| 1185 | fn cwd_pass_resolves_when_workspace_pass_misses() { |
| 1186 | let tmp = TempDir::new().expect("tempdir"); |
| 1187 | let sub = tmp.path().join("sub"); |
| 1188 | std::fs::create_dir_all(&sub).expect("mkdir"); |
| 1189 | let bar = sub.join("bar.txt"); |
| 1190 | std::fs::write(&bar, "hello bar").expect("write bar"); |
| 1191 | |
| 1192 | let content = |
| 1193 | user_request_with_file_mentions("look at @bar.txt", tmp.path(), Some(sub.clone())); |
| 1194 | |
| 1195 | // The block must reference the cwd-rooted path with the file's body — |
| 1196 | // and crucially it must NOT collapse to <missing-file>. |
| 1197 | assert!( |
| 1198 | content.contains("hello bar"), |
| 1199 | "expected file body to be inlined; got: {content}", |
| 1200 | ); |
| 1201 | assert!( |
| 1202 | !content.contains("<missing-file"), |
| 1203 | "must not surface <missing-file> for a path that exists under cwd; got: {content}", |
| 1204 | ); |
| 1205 | let bar_disp = bar.display().to_string(); |
| 1206 | assert!( |
| 1207 | content.contains(&bar_disp), |
| 1208 | "expected resolved path {bar_disp} in content; got: {content}", |
| 1209 | ); |
| 1210 | // Belt-and-suspenders: the workspace-rooted path doesn't exist and |
| 1211 | // must not appear in the rendered <file path="..."> attribute. |
| 1212 | let wrong = tmp.path().join("bar.txt").display().to_string(); |
| 1213 | assert!( |
| 1214 | !content.contains(&format!("path=\"{wrong}\"")), |
| 1215 | "should NOT have routed to {wrong}; got: {content}", |
| 1216 | ); |
| 1217 | } |
| 1218 | |
| 1219 | /// #101 regression — nested workspace path: `@nested/deep/file.md` with |
| 1220 | /// the file at workspace root resolves through the workspace pass. |
| 1221 | #[test] |
| 1222 | fn workspace_pass_resolves_nested_path() { |
| 1223 | let tmp = TempDir::new().expect("tempdir"); |
| 1224 | let nested = tmp.path().join("nested/deep"); |
| 1225 | std::fs::create_dir_all(&nested).expect("mkdir"); |
| 1226 | let file_md = nested.join("file.md"); |
| 1227 | std::fs::write(&file_md, "# nested deep").expect("write file_md"); |
| 1228 | |
| 1229 | // Cwd is irrelevant; an unrelated tempdir would do. Pass `None` so we |
| 1230 | // are unambiguously testing the workspace-pass path. |
| 1231 | let content = user_request_with_file_mentions("see @nested/deep/file.md", tmp.path(), None); |
| 1232 | |
| 1233 | assert!(content.contains("# nested deep"), "got: {content}"); |
| 1234 | assert!(!content.contains("<missing-file"), "got: {content}"); |
| 1235 | // Path-separator-portable check: the resolved path's filename is the |
| 1236 | // most reliable cross-platform anchor (Windows mixes `/` and `\` when |
| 1237 | // join() preserves user-typed separators). |
| 1238 | let basename = file_md |
| 1239 | .file_name() |
| 1240 | .and_then(|n| n.to_str()) |
| 1241 | .expect("file_name utf-8"); |
| 1242 | assert!( |
| 1243 | content.contains(basename), |
| 1244 | "basename {basename} not in path; got: {content}", |
| 1245 | ); |
| 1246 | } |
| 1247 | |
| 1248 | /// Snapshot-style check: the rendered `<file>` block for a resolvable |
| 1249 | /// mention must include the expected attributes and contents, and must |
| 1250 | /// NOT contain `<missing-file>`. |
| 1251 | #[test] |
| 1252 | fn resolvable_mention_renders_file_block_not_missing_file() { |
| 1253 | let tmp = TempDir::new().expect("tempdir"); |
| 1254 | std::fs::write(tmp.path().join("guide.md"), "# Guide\nUse the fast path.\n") |
| 1255 | .expect("write"); |
| 1256 | |
| 1257 | let content = user_request_with_file_mentions("read @guide.md", tmp.path(), None); |
| 1258 | |
| 1259 | // Header + tag presence. |
| 1260 | assert!(content.contains("Local context from @mentions:")); |
| 1261 | assert!(content.contains("<file mention=\"@guide.md\"")); |
| 1262 | assert!(content.contains("# Guide\nUse the fast path.")); |
| 1263 | assert!(content.ends_with("</file>"), "got: {content}"); |
| 1264 | // The bug fingerprint MUST be absent. |
| 1265 | assert!(!content.contains("<missing-file"), "got: {content}"); |
| 1266 | } |
| 1267 | |
| 1268 | /// Negative test: a truly missing path still produces `<missing-file>` |
| 1269 | /// so the user gets an explicit signal instead of silent failure. |
| 1270 | #[test] |
| 1271 | fn truly_missing_mention_still_renders_missing_file() { |
| 1272 | let tmp = TempDir::new().expect("tempdir"); |
| 1273 | |
| 1274 | let content = user_request_with_file_mentions( |
| 1275 | "huh @does/not/exist.txt", |
| 1276 | tmp.path(), |
| 1277 | Some(tmp.path().to_path_buf()), |
| 1278 | ); |
| 1279 | |
| 1280 | assert!( |
| 1281 | content.contains("<missing-file mention=\"@does/not/exist.txt\""), |
| 1282 | "got: {content}", |
| 1283 | ); |
| 1284 | } |
| 1285 | |
| 1286 | #[test] |
| 1287 | fn pending_context_preview_is_lexical_and_does_not_probe_paths() { |
| 1288 | let previews = pending_context_previews("read @guide.md and @missing.md"); |
| 1289 | |
| 1290 | assert_eq!(previews.len(), 2); |
| 1291 | assert_eq!(previews[0].kind, "mention"); |
| 1292 | assert_eq!(previews[0].label, "guide.md"); |
| 1293 | assert!(!previews[0].included); |
| 1294 | assert_eq!(previews[0].detail.as_deref(), Some("resolved on send")); |
| 1295 | assert_eq!(previews[1].kind, "mention"); |
| 1296 | assert_eq!(previews[1].label, "missing.md"); |
| 1297 | assert!(!previews[1].included); |
| 1298 | assert_eq!(previews[1].detail.as_deref(), Some("resolved on send")); |
| 1299 | } |
| 1300 | |
| 1301 | #[test] |
| 1302 | fn pending_context_preview_distinguishes_attach_media_from_at_media() { |
| 1303 | let tmp = TempDir::new().expect("tempdir"); |
| 1304 | std::fs::write(tmp.path().join("photo.png"), b"png").expect("write"); |
| 1305 | let attached = tmp.path().join("photo.png").display().to_string(); |
| 1306 | let input = format!("inspect @photo.png\n[Attached image: {attached}]"); |
| 1307 | |
| 1308 | let previews = pending_context_previews(&input); |
| 1309 | |
| 1310 | assert!( |
| 1311 | previews |
| 1312 | .iter() |
| 1313 | .any(|item| item.kind == "media" && !item.included), |
| 1314 | "at-mention media should be hint-only: {previews:?}" |
| 1315 | ); |
| 1316 | assert!( |
| 1317 | previews |
| 1318 | .iter() |
| 1319 | .any(|item| item.kind == "image" && item.included), |
| 1320 | "/attach media should be included: {previews:?}" |
| 1321 | ); |
| 1322 | } |
| 1323 | |
| 1324 | #[test] |
| 1325 | fn manually_typed_basename_does_not_fuzzy_attach_nested_file() { |
| 1326 | let tmp = TempDir::new().expect("tempdir"); |
| 1327 | let nested = tmp.path().join("nested"); |
| 1328 | std::fs::create_dir_all(&nested).expect("mkdir"); |
| 1329 | std::fs::write(nested.join("guide.md"), "nested secret").expect("write"); |
| 1330 | |
| 1331 | // With no completion index on hand there is no send-time fallback: |
| 1332 | // the miss stays an explicit <missing-file> rather than attaching an |
| 1333 | // arbitrary same-name file from a nested directory (#4365). |
| 1334 | let content = user_request_with_file_mentions("read @guide.md", tmp.path(), None); |
| 1335 | |
| 1336 | assert!( |
| 1337 | content.contains("<missing-file mention=\"@guide.md\""), |
| 1338 | "a manually typed basename should remain exact: {content}", |
| 1339 | ); |
| 1340 | assert!( |
| 1341 | !content.contains("nested secret"), |
| 1342 | "exact resolution must not silently attach a fuzzy nested match: {content}", |
| 1343 | ); |
| 1344 | } |
| 1345 | |
| 1346 | // --------------------------------------------------------------------- |
| 1347 | // Send-time completion-index fallback |
| 1348 | // --------------------------------------------------------------------- |
| 1349 | // |
| 1350 | // The dogfood failure this guards: `@FINISH-0.9.4.md` typed at the |
| 1351 | // workspace root resolved "not found" and injected a <missing-file> block |
| 1352 | // carrying the wrong (workspace-root) path, even though the file sat one |
| 1353 | // directory down. Misses now fall back to a bounded unique-match search |
| 1354 | // of the composer's background completion index; unresolvable misses emit |
| 1355 | // an honest block that names only what the user typed. |
| 1356 | |
| 1357 | fn expand_with_index( |
| 1358 | input: &str, |
| 1359 | workspace: &Path, |
| 1360 | cwd: Option<PathBuf>, |
| 1361 | index: &[String], |
| 1362 | ) -> String { |
| 1363 | user_request_with_file_mentions_cached( |
| 1364 | input, |
| 1365 | workspace, |
| 1366 | cwd, |
| 1367 | &mut GitMentionCache::default(), |
| 1368 | Some(index), |
| 1369 | ) |
| 1370 | } |
| 1371 | |
| 1372 | /// A unique basename hit in the completion index resolves a nested file |
| 1373 | /// and injects its real path. |
| 1374 | #[test] |
| 1375 | fn mention_miss_resolves_via_unique_index_basename() { |
| 1376 | let tmp = TempDir::new().expect("tempdir"); |
| 1377 | let nested = tmp.path().join("ops"); |
| 1378 | std::fs::create_dir_all(&nested).expect("mkdir"); |
| 1379 | std::fs::write(nested.join("FINISH-0.9.4.md"), "ship list").expect("write"); |
| 1380 | |
| 1381 | let index = vec!["ops/FINISH-0.9.4.md".to_string(), "README.md".to_string()]; |
| 1382 | let content = expand_with_index("finish @FINISH-0.9.4.md", tmp.path(), None, &index); |
| 1383 | |
| 1384 | assert!(content.contains("ship list"), "got: {content}"); |
| 1385 | assert!(!content.contains("<missing-file"), "got: {content}"); |
| 1386 | let real = nested.join("FINISH-0.9.4.md").display().to_string(); |
| 1387 | assert!( |
| 1388 | content.contains(&real), |
| 1389 | "expected resolved path {real} in content; got: {content}", |
| 1390 | ); |
| 1391 | } |
| 1392 | |
| 1393 | /// A typed partial path resolves through a unique path-suffix hit. |
| 1394 | #[test] |
| 1395 | fn mention_miss_resolves_via_unique_index_suffix() { |
| 1396 | let tmp = TempDir::new().expect("tempdir"); |
| 1397 | let nested = tmp.path().join("nested/deep"); |
| 1398 | std::fs::create_dir_all(&nested).expect("mkdir"); |
| 1399 | std::fs::write(nested.join("file.md"), "deep body").expect("write"); |
| 1400 | // A same-basename file elsewhere must not make the suffix hit |
| 1401 | // ambiguous: the typed directory context disambiguates. |
| 1402 | let other = tmp.path().join("other"); |
| 1403 | std::fs::create_dir_all(&other).expect("mkdir"); |
| 1404 | std::fs::write(other.join("file.md"), "other body").expect("write"); |
| 1405 | |
| 1406 | let index = vec![ |
| 1407 | "nested/deep/file.md".to_string(), |
| 1408 | "other/file.md".to_string(), |
| 1409 | ]; |
| 1410 | let content = expand_with_index("see @deep/file.md", tmp.path(), None, &index); |
| 1411 | |
| 1412 | assert!(content.contains("deep body"), "got: {content}"); |
| 1413 | assert!(!content.contains("other body"), "got: {content}"); |
| 1414 | assert!(!content.contains("<missing-file"), "got: {content}"); |
| 1415 | } |
| 1416 | |
| 1417 | /// Two same-basename candidates with no typed directory context are |
| 1418 | /// ambiguous: nothing is attached and the miss stays explicit. |
| 1419 | #[test] |
| 1420 | fn ambiguous_index_basename_stays_missing() { |
| 1421 | let tmp = TempDir::new().expect("tempdir"); |
| 1422 | for dir in ["a", "b"] { |
| 1423 | std::fs::create_dir_all(tmp.path().join(dir)).expect("mkdir"); |
| 1424 | std::fs::write(tmp.path().join(dir).join("guide.md"), format!("body {dir}")) |
| 1425 | .expect("write"); |
| 1426 | } |
| 1427 | |
| 1428 | let index = vec!["a/guide.md".to_string(), "b/guide.md".to_string()]; |
| 1429 | let content = expand_with_index("read @guide.md", tmp.path(), None, &index); |
| 1430 | |
| 1431 | assert!( |
| 1432 | content.contains("<missing-file mention=\"@guide.md\""), |
| 1433 | "an ambiguous basename must not attach an arbitrary winner: {content}", |
| 1434 | ); |
| 1435 | assert!(!content.contains("body a"), "got: {content}"); |
| 1436 | assert!(!content.contains("body b"), "got: {content}"); |
| 1437 | } |
| 1438 | |
| 1439 | /// A stale index entry (file deleted after the scan) must not attach. |
| 1440 | #[test] |
| 1441 | fn stale_index_entry_stays_missing() { |
| 1442 | let tmp = TempDir::new().expect("tempdir"); |
| 1443 | |
| 1444 | let index = vec!["ghost.md".to_string()]; |
| 1445 | let content = expand_with_index("boo @ghost.md", tmp.path(), None, &index); |
| 1446 | |
| 1447 | assert!( |
| 1448 | content.contains("<missing-file mention=\"@ghost.md\" />"), |
| 1449 | "got: {content}", |
| 1450 | ); |
| 1451 | } |
| 1452 | |
| 1453 | /// Absolute mentions name an exact location; the index must never |
| 1454 | /// substitute a same-basename file from inside the workspace. |
| 1455 | #[test] |
| 1456 | fn absolute_mention_miss_never_uses_index() { |
| 1457 | let tmp = TempDir::new().expect("tempdir"); |
| 1458 | std::fs::write(tmp.path().join("guide.md"), "workspace guide").expect("write"); |
| 1459 | |
| 1460 | let index = vec!["guide.md".to_string()]; |
| 1461 | let content = expand_with_index( |
| 1462 | "read @/definitely/absent/guide.md", |
| 1463 | tmp.path(), |
| 1464 | None, |
| 1465 | &index, |
| 1466 | ); |
| 1467 | |
| 1468 | assert!( |
| 1469 | content.contains("<missing-file mention=\"@/definitely/absent/guide.md\" />"), |
| 1470 | "got: {content}", |
| 1471 | ); |
| 1472 | assert!(!content.contains("workspace guide"), "got: {content}"); |
| 1473 | } |
| 1474 | |
| 1475 | /// The honest miss format: the block names only the typed mention and |
| 1476 | /// never the non-existent workspace-root join. |
| 1477 | #[test] |
| 1478 | fn missing_file_block_names_only_the_typed_mention() { |
| 1479 | let tmp = TempDir::new().expect("tempdir"); |
| 1480 | |
| 1481 | let content = user_request_with_file_mentions( |
| 1482 | "huh @does/not/exist.txt", |
| 1483 | tmp.path(), |
| 1484 | Some(tmp.path().to_path_buf()), |
| 1485 | ); |
| 1486 | |
| 1487 | assert!( |
| 1488 | content.contains("<missing-file mention=\"@does/not/exist.txt\" />"), |
| 1489 | "got: {content}", |
| 1490 | ); |
| 1491 | let wrong = tmp.path().join("does/not/exist.txt").display().to_string(); |
| 1492 | assert!( |
| 1493 | !content.contains(&wrong), |
| 1494 | "must not inject the wrong workspace-root path {wrong}; got: {content}", |
| 1495 | ); |
| 1496 | } |
| 1497 | |
| 1498 | /// The context inspector mirrors the payload: index-resolved mentions |
| 1499 | /// report their real path, unresolved ones report the typed token. |
| 1500 | #[test] |
| 1501 | fn context_references_reflect_index_resolution() { |
| 1502 | let tmp = TempDir::new().expect("tempdir"); |
| 1503 | let nested = tmp.path().join("ops"); |
| 1504 | std::fs::create_dir_all(&nested).expect("mkdir"); |
| 1505 | std::fs::write(nested.join("runbook.md"), "steps").expect("write"); |
| 1506 | |
| 1507 | let index = vec!["ops/runbook.md".to_string()]; |
| 1508 | let references = context_references_from_input_cached( |
| 1509 | "read @runbook.md and @absent.md", |
| 1510 | tmp.path(), |
| 1511 | None, |
| 1512 | &mut GitMentionCache::default(), |
| 1513 | Some(&index), |
| 1514 | ); |
| 1515 | |
| 1516 | let resolved = references |
| 1517 | .iter() |
| 1518 | .find(|r| r.label == "runbook.md") |
| 1519 | .expect("runbook reference"); |
| 1520 | assert_eq!(resolved.kind, ContextReferenceKind::File); |
| 1521 | assert!(resolved.included); |
| 1522 | let real = nested.join("runbook.md").display().to_string(); |
| 1523 | assert_eq!(resolved.target, real, "{resolved:?}"); |
| 1524 | |
| 1525 | let missing = references |
| 1526 | .iter() |
| 1527 | .find(|r| r.label == "absent.md") |
| 1528 | .expect("absent reference"); |
| 1529 | assert_eq!(missing.kind, ContextReferenceKind::Missing); |
| 1530 | assert!(!missing.included); |
| 1531 | assert_eq!( |
| 1532 | missing.target, "absent.md", |
| 1533 | "a missing mention must not report the workspace-root guess as its target: {missing:?}", |
| 1534 | ); |
| 1535 | } |
| 1536 | |
| 1537 | #[test] |
| 1538 | fn media_attachment_references_include_removable_line_ranges() { |
| 1539 | let input = "before\n[Attached image: 8x4 PNG at /tmp/pasted.png]\nafter"; |
| 1540 | |
| 1541 | let references = media_attachment_references(input); |
| 1542 | |
| 1543 | assert_eq!(references.len(), 1); |
| 1544 | let reference = &references[0]; |
| 1545 | assert_eq!(reference.kind, "image"); |
| 1546 | assert_eq!(reference.path, "/tmp/pasted.png"); |
| 1547 | assert_eq!( |
| 1548 | &input[reference.start_byte..reference.end_byte], |
| 1549 | "[Attached image: 8x4 PNG at /tmp/pasted.png]\n" |
| 1550 | ); |
| 1551 | } |
| 1552 | |
| 1553 | #[test] |
| 1554 | fn context_references_preserve_exact_targets_and_roundtrip() { |
| 1555 | let tmp = TempDir::new().expect("tempdir"); |
| 1556 | std::fs::create_dir_all(tmp.path().join("src")).expect("mkdir"); |
| 1557 | std::fs::write(tmp.path().join("src/main.rs"), "fn main() {}").expect("write"); |
| 1558 | let input = "read @src/main.rs"; |
| 1559 | |
| 1560 | let references = |
| 1561 | context_references_from_input(input, tmp.path(), Some(tmp.path().to_path_buf())); |
| 1562 | |
| 1563 | assert_eq!(references.len(), 1); |
| 1564 | let reference = &references[0]; |
| 1565 | assert_eq!(reference.kind, ContextReferenceKind::File); |
| 1566 | assert_eq!(reference.source, ContextReferenceSource::AtMention); |
| 1567 | assert_eq!(reference.label, "src/main.rs"); |
| 1568 | assert!(reference.target.ends_with("src/main.rs")); |
| 1569 | assert!(reference.included); |
| 1570 | assert!(reference.expanded); |
| 1571 | |
| 1572 | let encoded = serde_json::to_string(reference).expect("serialize"); |
| 1573 | let decoded: ContextReference = serde_json::from_str(&encoded).expect("deserialize"); |
| 1574 | assert_eq!(&decoded, reference); |
| 1575 | } |
| 1576 | |
| 1577 | /// Regression test for #1441: truncating at MAX_MENTION_FILE_BYTES must not |
| 1578 | /// split a multi-byte UTF-8 sequence, which previously produced U+FFFD |
| 1579 | /// replacement characters in the TUI output. |
| 1580 | #[test] |
| 1581 | fn read_text_prefix_truncation_respects_utf8_char_boundary() { |
| 1582 | use std::io::Write; |
| 1583 | |
| 1584 | // Build a file that is MAX_MENTION_FILE_BYTES - 1 ASCII bytes followed |
| 1585 | // by a 3-byte CJK character (U+4E2D, '中'). The naive truncate at |
| 1586 | // MAX_MENTION_FILE_BYTES cuts after the first byte of '中', producing |
| 1587 | // an invalid sequence. |
| 1588 | let tmp = TempDir::new().expect("tempdir"); |
| 1589 | let path = tmp.path().join("cjk.txt"); |
| 1590 | let mut f = std::fs::File::create(&path).expect("create"); |
| 1591 | let padding = vec![b'a'; MAX_MENTION_FILE_BYTES as usize - 1]; |
| 1592 | f.write_all(&padding).expect("write padding"); |
| 1593 | f.write_all("中".as_bytes()).expect("write CJK"); |
| 1594 | |
| 1595 | let (text, truncated) = read_text_prefix(&path).expect("should succeed"); |
| 1596 | assert!( |
| 1597 | truncated, |
| 1598 | "file exceeds limit so should be marked truncated" |
| 1599 | ); |
| 1600 | assert!( |
| 1601 | !text.contains('\u{FFFD}'), |
| 1602 | "truncated text must not contain replacement characters; got: {text:?}", |
| 1603 | ); |
| 1604 | } |
| 1605 | // --------------------------------------------------------------------- |
| 1606 | // #4067 — @git / @diff composer mentions |
| 1607 | // --------------------------------------------------------------------- |
| 1608 | |
| 1609 | fn init_test_repo(dir: &Path) { |
| 1610 | for args in [ |
| 1611 | vec!["init", "--initial-branch=main"], |
| 1612 | vec!["config", "user.email", "test@example.com"], |
| 1613 | vec!["config", "user.name", "Test"], |
| 1614 | ] { |
| 1615 | let out = std::process::Command::new("git") |
| 1616 | .args(&args) |
| 1617 | .current_dir(dir) |
| 1618 | .output() |
| 1619 | .expect("git available in tests"); |
| 1620 | assert!(out.status.success(), "git {args:?} failed"); |
| 1621 | } |
| 1622 | } |
| 1623 | |
| 1624 | fn commit_test_repo(dir: &Path) { |
| 1625 | for args in [vec!["add", "-A"], vec!["commit", "-m", "initial"]] { |
| 1626 | std::process::Command::new("git") |
| 1627 | .args(&args) |
| 1628 | .current_dir(dir) |
| 1629 | .output() |
| 1630 | .expect("git available in tests"); |
| 1631 | } |
| 1632 | } |
| 1633 | |
| 1634 | #[test] |
| 1635 | fn git_and_diff_mentions_inline_curated_context_not_paths() { |
| 1636 | let tmp = TempDir::new().expect("tempdir"); |
| 1637 | init_test_repo(tmp.path()); |
| 1638 | std::fs::write(tmp.path().join("a.txt"), "one\n").expect("write"); |
| 1639 | commit_test_repo(tmp.path()); |
| 1640 | std::fs::write(tmp.path().join("a.txt"), "two\n").expect("write"); |
| 1641 | |
| 1642 | let expanded = user_request_with_file_mentions( |
| 1643 | "look at @git and @diff", |
| 1644 | tmp.path(), |
| 1645 | Some(tmp.path().to_path_buf()), |
| 1646 | ); |
| 1647 | |
| 1648 | assert!(expanded.contains("<git-status"), "{expanded}"); |
| 1649 | assert!(expanded.contains("<git-diff"), "{expanded}"); |
| 1650 | assert!(expanded.contains("a.txt"), "{expanded}"); |
| 1651 | // The tokens are not treated as paths, so no missing-file block. |
| 1652 | assert!(!expanded.contains("<missing-file"), "{expanded}"); |
| 1653 | } |
| 1654 | |
| 1655 | #[test] |
| 1656 | fn git_mentions_outside_a_repository_say_so_explicitly() { |
| 1657 | let tmp = TempDir::new().expect("tempdir"); |
| 1658 | let expanded = user_request_with_file_mentions( |
| 1659 | "status? @git", |
| 1660 | tmp.path(), |
| 1661 | Some(tmp.path().to_path_buf()), |
| 1662 | ); |
| 1663 | assert!(expanded.contains("<git-unavailable"), "{expanded}"); |
| 1664 | assert!(expanded.contains("not a git repository"), "{expanded}"); |
| 1665 | } |
| 1666 | |
| 1667 | #[test] |
| 1668 | fn git_mention_is_deduplicated_within_one_message() { |
| 1669 | let tmp = TempDir::new().expect("tempdir"); |
| 1670 | init_test_repo(tmp.path()); |
| 1671 | std::fs::write(tmp.path().join("a.txt"), "one\n").expect("write"); |
| 1672 | commit_test_repo(tmp.path()); |
| 1673 | std::fs::write(tmp.path().join("a.txt"), "two\n").expect("write"); |
| 1674 | |
| 1675 | let expanded = user_request_with_file_mentions( |
| 1676 | "@diff and again @diff", |
| 1677 | tmp.path(), |
| 1678 | Some(tmp.path().to_path_buf()), |
| 1679 | ); |
| 1680 | assert_eq!(expanded.matches("<git-diff").count(), 1, "{expanded}"); |
| 1681 | } |
| 1682 | |
| 1683 | #[test] |
| 1684 | fn paths_that_merely_start_with_git_stay_file_mentions() { |
| 1685 | let tmp = TempDir::new().expect("tempdir"); |
| 1686 | std::fs::write(tmp.path().join("diff.txt"), "plain file").expect("write"); |
| 1687 | |
| 1688 | let expanded = user_request_with_file_mentions( |
| 1689 | "see @diff.txt", |
| 1690 | tmp.path(), |
| 1691 | Some(tmp.path().to_path_buf()), |
| 1692 | ); |
| 1693 | assert!(expanded.contains("plain file"), "{expanded}"); |
| 1694 | assert!(!expanded.contains("<git-diff"), "{expanded}"); |
| 1695 | } |
| 1696 | |
| 1697 | #[test] |
| 1698 | fn large_diff_is_truncated_and_the_inspector_reports_the_budget() { |
| 1699 | let tmp = TempDir::new().expect("tempdir"); |
| 1700 | init_test_repo(tmp.path()); |
| 1701 | std::fs::write(tmp.path().join("big.txt"), "seed\n").expect("write"); |
| 1702 | commit_test_repo(tmp.path()); |
| 1703 | let bulk: String = (0..40_000).map(|i| format!("line {i}\n")).collect(); |
| 1704 | std::fs::write(tmp.path().join("big.txt"), bulk).expect("write"); |
| 1705 | |
| 1706 | let expanded = |
| 1707 | user_request_with_file_mentions("@diff", tmp.path(), Some(tmp.path().to_path_buf())); |
| 1708 | assert!( |
| 1709 | expanded.contains("truncated=\"true\""), |
| 1710 | "expected truncation marker" |
| 1711 | ); |
| 1712 | |
| 1713 | let references = |
| 1714 | context_references_from_input("@diff", tmp.path(), Some(tmp.path().to_path_buf())); |
| 1715 | let git_ref = references |
| 1716 | .iter() |
| 1717 | .find(|r| r.kind == ContextReferenceKind::GitContext) |
| 1718 | .expect("git reference present in the inspector"); |
| 1719 | assert_eq!(git_ref.label, "diff"); |
| 1720 | assert!(git_ref.included); |
| 1721 | let detail = git_ref.detail.clone().unwrap_or_default(); |
| 1722 | assert!(detail.contains("truncated at"), "{detail}"); |
| 1723 | assert!( |
| 1724 | detail.contains(&crate::tui::git_mention::MAX_GIT_DIFF_MENTION_BYTES.to_string()), |
| 1725 | "{detail}" |
| 1726 | ); |
| 1727 | } |
| 1728 | |
| 1729 | #[test] |
| 1730 | fn empty_repository_reference_is_visible_but_not_included() { |
| 1731 | let tmp = TempDir::new().expect("tempdir"); |
| 1732 | init_test_repo(tmp.path()); |
| 1733 | std::fs::write(tmp.path().join("a.txt"), "one\n").expect("write"); |
| 1734 | commit_test_repo(tmp.path()); |
| 1735 | |
| 1736 | let references = |
| 1737 | context_references_from_input("@diff", tmp.path(), Some(tmp.path().to_path_buf())); |
| 1738 | let git_ref = references |
| 1739 | .iter() |
| 1740 | .find(|r| r.kind == ContextReferenceKind::GitContext) |
| 1741 | .expect("git reference present even when there is nothing to show"); |
| 1742 | assert!(!git_ref.included); |
| 1743 | assert!( |
| 1744 | git_ref |
| 1745 | .detail |
| 1746 | .as_deref() |
| 1747 | .is_some_and(|d| d.contains("no working-tree changes")), |
| 1748 | "{:?}", |
| 1749 | git_ref.detail |
| 1750 | ); |
| 1751 | } |
| 1752 | |
| 1753 | #[test] |
| 1754 | fn composer_preview_lists_git_mentions_without_running_git() { |
| 1755 | let previews = pending_context_previews("@git @diff"); |
| 1756 | let kinds: Vec<&str> = previews.iter().map(|p| p.kind.as_str()).collect(); |
| 1757 | assert_eq!(kinds, vec!["git", "git"]); |
| 1758 | assert!(previews.iter().all(|p| !p.included)); |
| 1759 | } |
| 1760 | |
| 1761 | /// #4067 review follow-up: `mention_menu_limit = 0` is a documented way to |
| 1762 | /// disable the popup. The git tokens are menu entries like any other and |
| 1763 | /// must respect the same cap — otherwise setting 0 still pops a one-entry |
| 1764 | /// menu the moment the user types `@g`. |
| 1765 | #[test] |
| 1766 | fn git_mention_entries_respect_a_zero_menu_limit() { |
| 1767 | let paths = vec!["src/main.rs".to_string()]; |
| 1768 | assert!(with_git_mention_entries(paths.clone(), "g", 0).is_empty()); |
| 1769 | assert!(with_git_mention_entries(paths.clone(), "d", 0).is_empty()); |
| 1770 | assert!(with_git_mention_entries(paths.clone(), "", 0).is_empty()); |
| 1771 | assert!(with_git_mention_entries(Vec::new(), "gi", 0).is_empty()); |
| 1772 | } |
| 1773 | |
| 1774 | /// A small non-zero limit must cap the token list this function builds. |
| 1775 | /// |
| 1776 | /// The empty-partial branch is pass-through by design — those entries were |
| 1777 | /// already capped upstream by `rank_completion_candidates`, and shrinking |
| 1778 | /// them here would silently drop paths the caller asked for. |
| 1779 | #[test] |
| 1780 | fn git_mention_entries_never_exceed_the_menu_limit() { |
| 1781 | let paths = vec!["a.rs".to_string(), "b.rs".to_string()]; |
| 1782 | for limit in 1..=4 { |
| 1783 | let matched = with_git_mention_entries(paths.clone(), "d", limit); |
| 1784 | assert!(matched.len() <= limit, "limit {limit}: {matched:?}"); |
| 1785 | // Both tokens match a bare prefix that hits `git` and `diff` |
| 1786 | // through separate entries; the cap still holds. |
| 1787 | let both = with_git_mention_entries(Vec::new(), "", limit); |
| 1788 | assert!(both.len() <= limit, "limit {limit}: {both:?}"); |
| 1789 | } |
| 1790 | // Pass-through: the caller's already-capped paths survive untouched. |
| 1791 | assert_eq!(with_git_mention_entries(paths.clone(), "", 1), paths); |
| 1792 | } |
| 1793 | |
| 1794 | /// #4067 review follow-up: one submit resolves a git mention once, not |
| 1795 | /// once per surface. `@diff` makes git compute the whole working-tree diff |
| 1796 | /// before the byte budget applies, so a repeat is real wasted work. |
| 1797 | #[test] |
| 1798 | fn a_submit_resolves_each_git_mention_only_once() { |
| 1799 | let tmp = TempDir::new().expect("tempdir"); |
| 1800 | init_test_repo(tmp.path()); |
| 1801 | std::fs::write(tmp.path().join("a.txt"), "one\n").expect("write"); |
| 1802 | commit_test_repo(tmp.path()); |
| 1803 | std::fs::write(tmp.path().join("a.txt"), "two\n").expect("write"); |
| 1804 | |
| 1805 | let mut cache = crate::tui::git_mention::GitMentionCache::default(); |
| 1806 | let references = context_references_from_input_cached( |
| 1807 | "@diff", |
| 1808 | tmp.path(), |
| 1809 | Some(tmp.path().to_path_buf()), |
| 1810 | &mut cache, |
| 1811 | None, |
| 1812 | ); |
| 1813 | let expanded = user_request_with_file_mentions_cached( |
| 1814 | "@diff", |
| 1815 | tmp.path(), |
| 1816 | Some(tmp.path().to_path_buf()), |
| 1817 | &mut cache, |
| 1818 | None, |
| 1819 | ); |
| 1820 | |
| 1821 | // Both surfaces describe the same resolution. |
| 1822 | let git_ref = references |
| 1823 | .iter() |
| 1824 | .find(|r| r.kind == ContextReferenceKind::GitContext) |
| 1825 | .expect("git reference"); |
| 1826 | assert!(git_ref.included); |
| 1827 | assert!(expanded.contains("<git-diff"), "{expanded}"); |
| 1828 | |
| 1829 | // And the shared cache holds exactly one entry for it. |
| 1830 | assert_eq!(cache.len(), 1, "the diff must be resolved once per submit"); |
| 1831 | } |
| 1832 | |
| 1833 | #[test] |
| 1834 | fn completion_offers_git_and_diff_alongside_paths() { |
| 1835 | let paths = vec!["src/main.rs".to_string(), "docs/guide.md".to_string()]; |
| 1836 | // A bare `@` stays the file picker. |
| 1837 | assert_eq!(with_git_mention_entries(paths.clone(), "", 8), paths); |
| 1838 | |
| 1839 | let narrowed = with_git_mention_entries(paths.clone(), "di", 8); |
| 1840 | assert_eq!(narrowed.first().map(String::as_str), Some("diff")); |
| 1841 | assert!(narrowed.contains(&"src/main.rs".to_string())); |
| 1842 | |
| 1843 | let git_only = with_git_mention_entries(paths.clone(), "g", 8); |
| 1844 | assert_eq!(git_only.first().map(String::as_str), Some("git")); |
| 1845 | |
| 1846 | // A partial that matches no token leaves path completion untouched. |
| 1847 | assert_eq!(with_git_mention_entries(paths.clone(), "src", 8), paths); |
| 1848 | } |
| 1849 | } |
| 1850 |