| 1 | use std::fs; |
| 2 | |
| 3 | use serde_json::{Value, json}; |
| 4 | use tempfile::tempdir; |
| 5 | use tokio_util::sync::CancellationToken; |
| 6 | |
| 7 | use crate::tools::spec::{ApprovalRequirement, ToolContext, ToolSpec}; |
| 8 | |
| 9 | use super::{GrepFilesTool, matches_glob}; |
| 10 | |
| 11 | #[test] |
| 12 | fn grep_description_matches_default_exclusion_behavior() { |
| 13 | let description = GrepFilesTool.description(); |
| 14 | |
| 15 | assert!(description.contains("skips common non-code directories")); |
| 16 | assert!(!description.contains("respects `.gitignore`")); |
| 17 | } |
| 18 | |
| 19 | /// Representative of the ~150 shared `optional_*` call sites outside the |
| 20 | /// three that motivated the change: a wrong type is refused by name, an |
| 21 | /// absent or null field still takes its default. |
| 22 | #[tokio::test] |
| 23 | async fn grep_refuses_mistyped_optional_parameters_by_name() { |
| 24 | let tmp = tempdir().expect("tempdir"); |
| 25 | fs::write(tmp.path().join("a.txt"), "needle\n").expect("write"); |
| 26 | let ctx = ToolContext::new(tmp.path()); |
| 27 | |
| 28 | for (field, input) in [ |
| 29 | ( |
| 30 | "max_results", |
| 31 | json!({"pattern": "needle", "max_results": "10"}), |
| 32 | ), |
| 33 | ( |
| 34 | "case_insensitive", |
| 35 | json!({"pattern": "needle", "case_insensitive": "true"}), |
| 36 | ), |
| 37 | ( |
| 38 | "context_lines", |
| 39 | json!({"pattern": "needle", "context_lines": 2.5}), |
| 40 | ), |
| 41 | ("path", json!({"pattern": "needle", "path": ["."]})), |
| 42 | ] { |
| 43 | let err = GrepFilesTool |
| 44 | .execute(input, &ctx) |
| 45 | .await |
| 46 | .expect_err("a mistyped optional parameter must be refused"); |
| 47 | let err = err.to_string(); |
| 48 | assert!(err.contains(field), "error must name '{field}': {err}"); |
| 49 | } |
| 50 | |
| 51 | GrepFilesTool |
| 52 | .execute( |
| 53 | json!({"pattern": "needle", "max_results": Value::Null, "path": Value::Null}), |
| 54 | &ctx, |
| 55 | ) |
| 56 | .await |
| 57 | .expect("explicit nulls read as absent"); |
| 58 | } |
| 59 | |
| 60 | /// The model supplies the pattern, so it compiles through the bounded |
| 61 | /// regex cache: an over-long pattern is refused, not compiled. |
| 62 | #[tokio::test] |
| 63 | async fn grep_refuses_a_pattern_past_the_user_regex_length_cap() { |
| 64 | let tmp = tempdir().expect("tempdir"); |
| 65 | fs::write(tmp.path().join("a.txt"), "needle\n").expect("write"); |
| 66 | let ctx = ToolContext::new(tmp.path()); |
| 67 | let long = "a".repeat(crate::regex_cache::MAX_USER_REGEX_PATTERN_BYTES + 1); |
| 68 | let err = GrepFilesTool |
| 69 | .execute(json!({ "pattern": long }), &ctx) |
| 70 | .await |
| 71 | .expect_err("an over-long pattern must be refused"); |
| 72 | assert!(err.to_string().contains("limit"), "{err}"); |
| 73 | } |
| 74 | |
| 75 | #[test] |
| 76 | fn test_matches_glob_star() { |
| 77 | assert!(matches_glob("test.rs", "*.rs")); |
| 78 | assert!(matches_glob("foo.rs", "*.rs")); |
| 79 | assert!(!matches_glob("test.ts", "*.rs")); |
| 80 | assert!(!matches_glob("test.rs.bak", "*.rs")); |
| 81 | } |
| 82 | |
| 83 | #[test] |
| 84 | fn test_matches_glob_question() { |
| 85 | assert!(matches_glob("test.rs", "test.??")); |
| 86 | assert!(!matches_glob("test.rs", "test.?")); |
| 87 | } |
| 88 | |
| 89 | #[test] |
| 90 | fn test_matches_glob_double_star() { |
| 91 | assert!(matches_glob("src/main.rs", "src/**")); |
| 92 | assert!(matches_glob("src/lib/mod.rs", "src/**")); |
| 93 | assert!(matches_glob("node_modules/pkg/index.js", "node_modules/*")); |
| 94 | } |
| 95 | |
| 96 | #[test] |
| 97 | fn test_matches_glob_path() { |
| 98 | assert!(matches_glob("src/main.rs", "src/*.rs")); |
| 99 | assert!(!matches_glob("lib/main.rs", "src/*.rs")); |
| 100 | } |
| 101 | |
| 102 | /// Regression for #249: byte-index slicing panics on multi-byte |
| 103 | /// characters inside filenames like `dialogue_line__冰糖.mp3`. |
| 104 | #[test] |
| 105 | fn test_matches_glob_unicode_filename() { |
| 106 | let filename = "dialogue_line__冰糖.mp3"; |
| 107 | // The filename should match *.mp3 without panicking. |
| 108 | assert!(matches_glob(filename, "*.mp3")); |
| 109 | // Asterisk matching against multi-byte characters must succeed. |
| 110 | assert!(matches_glob(filename, "dialogue_line__*")); |
| 111 | // Literal multi-byte characters inside the pattern must match. |
| 112 | assert!(matches_glob(filename, "*冰糖*")); |
| 113 | // Non-matching pattern must not panic either. |
| 114 | assert!(!matches_glob(filename, "nonexistent*")); |
| 115 | } |
| 116 | |
| 117 | #[tokio::test] |
| 118 | async fn test_grep_files_basic() { |
| 119 | let tmp = tempdir().expect("tempdir"); |
| 120 | let ctx = ToolContext::new(tmp.path().to_path_buf()); |
| 121 | |
| 122 | // Create test files |
| 123 | fs::write( |
| 124 | tmp.path().join("test.rs"), |
| 125 | "fn main() {\n println!(\"hello\");\n}\n", |
| 126 | ) |
| 127 | .expect("write"); |
| 128 | fs::write( |
| 129 | tmp.path().join("lib.rs"), |
| 130 | "pub fn hello() {}\npub fn world() {}\n", |
| 131 | ) |
| 132 | .expect("write"); |
| 133 | |
| 134 | let tool = GrepFilesTool; |
| 135 | let result = tool |
| 136 | .execute(json!({"pattern": "fn"}), &ctx) |
| 137 | .await |
| 138 | .expect("execute"); |
| 139 | |
| 140 | assert!(result.success); |
| 141 | assert!(result.content.contains("main")); |
| 142 | assert!(result.content.contains("hello")); |
| 143 | } |
| 144 | |
| 145 | #[tokio::test] |
| 146 | async fn test_grep_files_with_context() { |
| 147 | let tmp = tempdir().expect("tempdir"); |
| 148 | let ctx = ToolContext::new(tmp.path().to_path_buf()); |
| 149 | |
| 150 | fs::write( |
| 151 | tmp.path().join("test.txt"), |
| 152 | "line1\nline2\nMATCH\nline4\nline5\n", |
| 153 | ) |
| 154 | .expect("write"); |
| 155 | |
| 156 | let tool = GrepFilesTool; |
| 157 | let result = tool |
| 158 | .execute(json!({"pattern": "MATCH", "context_lines": 1}), &ctx) |
| 159 | .await |
| 160 | .expect("execute"); |
| 161 | |
| 162 | assert!(result.success); |
| 163 | assert!(result.content.contains("line2")); // context before |
| 164 | assert!(result.content.contains("line4")); // context after |
| 165 | |
| 166 | let parsed: Value = serde_json::from_str(&result.content).unwrap(); |
| 167 | let matches = parsed["matches"].as_array().unwrap(); |
| 168 | assert_eq!(matches.len(), 1); |
| 169 | assert_eq!(matches[0]["context_before"], "line2"); |
| 170 | assert_eq!(matches[0]["context_after"], "line4"); |
| 171 | assert!(matches[0]["context_before"].is_string()); |
| 172 | assert!(matches[0]["context_after"].is_string()); |
| 173 | } |
| 174 | |
| 175 | #[tokio::test] |
| 176 | async fn test_grep_files_multi_line_context_remains_arrays() { |
| 177 | let tmp = tempdir().expect("tempdir"); |
| 178 | let ctx = ToolContext::new(tmp.path().to_path_buf()); |
| 179 | |
| 180 | fs::write(tmp.path().join("test.txt"), "a\nb\nMATCH\nd\ne\n").expect("write"); |
| 181 | |
| 182 | let tool = GrepFilesTool; |
| 183 | let result = tool |
| 184 | .execute(json!({"pattern": "MATCH", "context_lines": 2}), &ctx) |
| 185 | .await |
| 186 | .expect("execute"); |
| 187 | |
| 188 | let parsed: Value = serde_json::from_str(&result.content).unwrap(); |
| 189 | let matches = parsed["matches"].as_array().unwrap(); |
| 190 | assert_eq!(matches.len(), 1); |
| 191 | assert_eq!(matches[0]["context_before"], json!(["a", "b"])); |
| 192 | assert_eq!(matches[0]["context_after"], json!(["d", "e"])); |
| 193 | } |
| 194 | |
| 195 | #[tokio::test] |
| 196 | async fn test_grep_files_case_insensitive() { |
| 197 | let tmp = tempdir().expect("tempdir"); |
| 198 | let ctx = ToolContext::new(tmp.path().to_path_buf()); |
| 199 | |
| 200 | fs::write( |
| 201 | tmp.path().join("test.txt"), |
| 202 | "Hello World\nHELLO WORLD\nhello world\n", |
| 203 | ) |
| 204 | .expect("write"); |
| 205 | |
| 206 | let tool = GrepFilesTool; |
| 207 | let result = tool |
| 208 | .execute(json!({"pattern": "hello", "case_insensitive": true}), &ctx) |
| 209 | .await |
| 210 | .expect("execute"); |
| 211 | |
| 212 | assert!(result.success); |
| 213 | // Should find all 3 lines |
| 214 | let parsed: Value = serde_json::from_str(&result.content).unwrap(); |
| 215 | assert_eq!(parsed["total_matches"].as_u64().unwrap(), 3); |
| 216 | } |
| 217 | |
| 218 | #[tokio::test] |
| 219 | async fn test_grep_files_include_filter() { |
| 220 | let tmp = tempdir().expect("tempdir"); |
| 221 | let ctx = ToolContext::new(tmp.path().to_path_buf()); |
| 222 | |
| 223 | fs::write(tmp.path().join("test.rs"), "fn test() {}\n").expect("write"); |
| 224 | fs::write(tmp.path().join("test.js"), "function test() {}\n").expect("write"); |
| 225 | |
| 226 | let tool = GrepFilesTool; |
| 227 | let result = tool |
| 228 | .execute(json!({"pattern": "test", "include": ["*.rs"]}), &ctx) |
| 229 | .await |
| 230 | .expect("execute"); |
| 231 | |
| 232 | assert!(result.success); |
| 233 | // Should only match .rs file |
| 234 | let parsed: Value = serde_json::from_str(&result.content).unwrap(); |
| 235 | let matches = parsed["matches"].as_array().unwrap(); |
| 236 | assert_eq!(matches.len(), 1); |
| 237 | let file = matches[0]["file"].as_str().unwrap(); |
| 238 | assert!( |
| 239 | file.rsplit('.') |
| 240 | .next() |
| 241 | .is_some_and(|ext| ext.eq_ignore_ascii_case("rs")) |
| 242 | ); |
| 243 | } |
| 244 | |
| 245 | #[tokio::test] |
| 246 | #[cfg(unix)] |
| 247 | async fn test_grep_files_does_not_follow_symlinked_files() { |
| 248 | let tmp = tempdir().expect("tempdir"); |
| 249 | let root = tmp.path().join("workspace"); |
| 250 | let outside = tmp.path().join("outside"); |
| 251 | std::fs::create_dir_all(&root).expect("mkdir workspace"); |
| 252 | std::fs::create_dir_all(&outside).expect("mkdir outside"); |
| 253 | let outside_file = outside.join("secret.txt"); |
| 254 | fs::write(&outside_file, "NEEDLE\n").expect("write outside"); |
| 255 | std::os::unix::fs::symlink(&outside_file, root.join("secret.txt")).expect("symlink"); |
| 256 | |
| 257 | let ctx = ToolContext::new(root); |
| 258 | let tool = GrepFilesTool; |
| 259 | let result = tool |
| 260 | .execute(json!({"pattern": "NEEDLE"}), &ctx) |
| 261 | .await |
| 262 | .expect("execute"); |
| 263 | |
| 264 | assert!(result.success); |
| 265 | let parsed: Value = serde_json::from_str(&result.content).unwrap(); |
| 266 | assert_eq!(parsed["total_matches"].as_u64().unwrap(), 0); |
| 267 | assert_eq!(parsed["files_searched"].as_u64().unwrap(), 0); |
| 268 | } |
| 269 | |
| 270 | #[tokio::test] |
| 271 | #[cfg(unix)] |
| 272 | async fn test_grep_files_default_mode_skips_symlinked_directories_but_keeps_real_files() { |
| 273 | let tmp = tempdir().expect("tempdir"); |
| 274 | let workspace = tmp.path().join("workspace"); |
| 275 | let real_dir = workspace.join("real"); |
| 276 | std::fs::create_dir_all(&real_dir).expect("mkdir workspace"); |
| 277 | fs::write(real_dir.join("needle.txt"), "NEEDLE\n").expect("write real file"); |
| 278 | std::os::unix::fs::symlink(&workspace, real_dir.join("loop")).expect("symlink loop"); |
| 279 | |
| 280 | let ctx = ToolContext::new(workspace); |
| 281 | let tool = GrepFilesTool; |
| 282 | let result = tool |
| 283 | .execute(json!({"pattern": "NEEDLE"}), &ctx) |
| 284 | .await |
| 285 | .expect("execute"); |
| 286 | |
| 287 | assert!(result.success); |
| 288 | let parsed: Value = serde_json::from_str(&result.content).unwrap(); |
| 289 | assert_eq!(parsed["total_matches"].as_u64().unwrap(), 1); |
| 290 | assert_eq!(parsed["files_searched"].as_u64().unwrap(), 1); |
| 291 | let matches = parsed["matches"].as_array().unwrap(); |
| 292 | assert_eq!(matches.len(), 1); |
| 293 | assert!( |
| 294 | matches[0]["file"] |
| 295 | .as_str() |
| 296 | .unwrap() |
| 297 | .ends_with("real/needle.txt") |
| 298 | ); |
| 299 | } |
| 300 | |
| 301 | #[tokio::test] |
| 302 | #[cfg(unix)] |
| 303 | async fn test_grep_files_follow_symlinks_avoids_directory_cycles() { |
| 304 | let tmp = tempdir().expect("tempdir"); |
| 305 | let workspace = tmp.path().join("workspace"); |
| 306 | let real_dir = workspace.join("real"); |
| 307 | fs::create_dir_all(&real_dir).expect("mkdir"); |
| 308 | fs::write(real_dir.join("needle.txt"), "NEEDLE\n").expect("write"); |
| 309 | std::os::unix::fs::symlink(&workspace, real_dir.join("loop")).expect("symlink loop"); |
| 310 | |
| 311 | let ctx = ToolContext::new(workspace).with_follow_symlinks(true); |
| 312 | let tool = GrepFilesTool; |
| 313 | let result = tool |
| 314 | .execute(json!({"pattern": "NEEDLE"}), &ctx) |
| 315 | .await |
| 316 | .expect("execute"); |
| 317 | |
| 318 | assert!(result.success); |
| 319 | let parsed: Value = serde_json::from_str(&result.content).unwrap(); |
| 320 | assert_eq!(parsed["total_matches"].as_u64().unwrap(), 1); |
| 321 | assert_eq!(parsed["files_searched"].as_u64().unwrap(), 1); |
| 322 | let matches = parsed["matches"].as_array().unwrap(); |
| 323 | assert!(matches[0]["file"].as_str().unwrap().ends_with("needle.txt")); |
| 324 | } |
| 325 | |
| 326 | #[tokio::test] |
| 327 | async fn test_grep_files_invalid_regex() { |
| 328 | let tmp = tempdir().expect("tempdir"); |
| 329 | let ctx = ToolContext::new(tmp.path().to_path_buf()); |
| 330 | |
| 331 | let tool = GrepFilesTool; |
| 332 | let result = tool.execute(json!({"pattern": "[invalid"}), &ctx).await; |
| 333 | |
| 334 | assert!(result.is_err()); |
| 335 | } |
| 336 | |
| 337 | #[tokio::test] |
| 338 | async fn test_grep_files_respects_cancel_token() { |
| 339 | let tmp = tempdir().expect("tempdir"); |
| 340 | fs::write(tmp.path().join("test.txt"), "needle\n").expect("write"); |
| 341 | let cancel_token = CancellationToken::new(); |
| 342 | cancel_token.cancel(); |
| 343 | let ctx = ToolContext::new(tmp.path().to_path_buf()).with_cancel_token(cancel_token); |
| 344 | |
| 345 | let tool = GrepFilesTool; |
| 346 | let err = tool |
| 347 | .execute(json!({"pattern": "needle"}), &ctx) |
| 348 | .await |
| 349 | .expect_err("cancelled grep should return an error"); |
| 350 | |
| 351 | assert!( |
| 352 | format!("{err:?}").contains("cancelled"), |
| 353 | "unexpected error: {err:?}" |
| 354 | ); |
| 355 | } |
| 356 | |
| 357 | #[tokio::test] |
| 358 | async fn test_grep_files_streaming_stops_at_max_results() { |
| 359 | let tmp = tempdir().expect("tempdir"); |
| 360 | let ctx = ToolContext::new(tmp.path().to_path_buf()); |
| 361 | |
| 362 | // Two files with many matches each; the walk must stop once the |
| 363 | // budget is exhausted without dropping context for the last match. |
| 364 | for name in ["a.txt", "b.txt"] { |
| 365 | let body: String = (1..=20).map(|n| format!("needle {n}\n")).collect(); |
| 366 | fs::write(tmp.path().join(name), body).expect("write"); |
| 367 | } |
| 368 | |
| 369 | let tool = GrepFilesTool; |
| 370 | let result = tool |
| 371 | .execute(json!({"pattern": "needle", "max_results": 5}), &ctx) |
| 372 | .await |
| 373 | .expect("execute"); |
| 374 | |
| 375 | assert!(result.success); |
| 376 | let parsed: Value = serde_json::from_str(&result.content).unwrap(); |
| 377 | let matches = parsed["matches"].as_array().unwrap(); |
| 378 | assert_eq!(matches.len(), 5); |
| 379 | assert_eq!(parsed["total_matches"].as_u64().unwrap(), 5); |
| 380 | // All five matches must come from the first file walked, in file |
| 381 | // order (streaming preserves walk order). |
| 382 | let first_file = matches[0]["file"].as_str().unwrap().to_string(); |
| 383 | for m in matches { |
| 384 | assert_eq!(m["file"].as_str().unwrap(), first_file); |
| 385 | } |
| 386 | // The final in-budget match still gets its full after-context even |
| 387 | // though the match budget was exhausted on it. |
| 388 | assert_eq!( |
| 389 | matches[4]["context_after"], |
| 390 | json!(["needle 6", "needle 7"]), |
| 391 | "last match must keep after-context lines" |
| 392 | ); |
| 393 | } |
| 394 | |
| 395 | #[tokio::test] |
| 396 | async fn test_grep_files_ring_buffer_context_matches_full_read() { |
| 397 | let tmp = tempdir().expect("tempdir"); |
| 398 | let ctx = ToolContext::new(tmp.path().to_path_buf()); |
| 399 | |
| 400 | // Matches at the start, middle, and end of the file exercise the |
| 401 | // partial before-context (ring not yet full) and truncated |
| 402 | // after-context (EOF) paths. |
| 403 | fs::write( |
| 404 | tmp.path().join("ctx.txt"), |
| 405 | "MATCH first\nb1\nb2\nb3\nMATCH mid\na1\na2\na3\nMATCH last\n", |
| 406 | ) |
| 407 | .expect("write"); |
| 408 | |
| 409 | let tool = GrepFilesTool; |
| 410 | let result = tool |
| 411 | .execute(json!({"pattern": "MATCH", "context_lines": 2}), &ctx) |
| 412 | .await |
| 413 | .expect("execute"); |
| 414 | |
| 415 | let parsed: Value = serde_json::from_str(&result.content).unwrap(); |
| 416 | let matches = parsed["matches"].as_array().unwrap(); |
| 417 | assert_eq!(matches.len(), 3); |
| 418 | assert_eq!(matches[0]["context_before"], json!([])); |
| 419 | assert_eq!(matches[0]["context_after"], json!(["b1", "b2"])); |
| 420 | assert_eq!(matches[1]["context_before"], json!(["b2", "b3"])); |
| 421 | assert_eq!(matches[1]["context_after"], json!(["a1", "a2"])); |
| 422 | assert_eq!(matches[2]["context_before"], json!(["a2", "a3"])); |
| 423 | assert_eq!(matches[2]["context_after"], json!([])); |
| 424 | assert_eq!(matches[2]["line_number"].as_u64().unwrap(), 9); |
| 425 | } |
| 426 | |
| 427 | #[tokio::test] |
| 428 | async fn test_grep_files_streaming_skips_invalid_utf8_files() { |
| 429 | let tmp = tempdir().expect("tempdir"); |
| 430 | let ctx = ToolContext::new(tmp.path().to_path_buf()); |
| 431 | |
| 432 | // Invalid UTF-8 after a matching line: the whole file must be |
| 433 | // skipped, matching the historical read_to_string behavior. |
| 434 | fs::write( |
| 435 | tmp.path().join("binary.txt"), |
| 436 | [b"needle\n".as_slice(), &[0xFF, 0xFE, 0x00]].concat(), |
| 437 | ) |
| 438 | .expect("write"); |
| 439 | fs::write(tmp.path().join("clean.txt"), "needle\n").expect("write"); |
| 440 | |
| 441 | let tool = GrepFilesTool; |
| 442 | let result = tool |
| 443 | .execute(json!({"pattern": "needle"}), &ctx) |
| 444 | .await |
| 445 | .expect("execute"); |
| 446 | |
| 447 | let parsed: Value = serde_json::from_str(&result.content).unwrap(); |
| 448 | assert_eq!(parsed["total_matches"].as_u64().unwrap(), 1); |
| 449 | assert_eq!(parsed["files_searched"].as_u64().unwrap(), 1); |
| 450 | let matches = parsed["matches"].as_array().unwrap(); |
| 451 | assert!(matches[0]["file"].as_str().unwrap().ends_with("clean.txt")); |
| 452 | } |
| 453 | |
| 454 | #[test] |
| 455 | fn test_grep_files_tool_properties() { |
| 456 | let tool = GrepFilesTool; |
| 457 | assert_eq!(tool.name(), "grep_files"); |
| 458 | assert!(tool.is_read_only()); |
| 459 | assert!(tool.is_sandboxable()); |
| 460 | assert_eq!(tool.approval_requirement(), ApprovalRequirement::Auto); |
| 461 | } |
| 462 | |
| 463 | #[test] |
| 464 | fn test_parallel_support_flags() { |
| 465 | let tool = GrepFilesTool; |
| 466 | assert!(tool.supports_parallel()); |
| 467 | } |
| 468 | |
| 469 | #[test] |
| 470 | fn double_star_exclude_prefix_stops_at_a_path_component() { |
| 471 | // Directory excludes still cover the directory and everything under it. |
| 472 | assert!(matches_glob("build", "build/**")); |
| 473 | assert!(matches_glob("build/out.o", "build/**")); |
| 474 | assert!(matches_glob(".git/config", ".git/**")); |
| 475 | // A same-prefix sibling file or directory is a different path. |
| 476 | assert!(!matches_glob("build.rs", "build/**")); |
| 477 | assert!(!matches_glob("build.gradle", "build/**")); |
| 478 | assert!(!matches_glob("builders/x.rs", "build/**")); |
| 479 | assert!(!matches_glob(".github/workflows/ci.yml", ".git/**")); |
| 480 | assert!(!matches_glob(".gitignore", ".git/**")); |
| 481 | assert!(!matches_glob("targets/a.rs", "target/**")); |
| 482 | } |
| 483 | |
| 484 | #[tokio::test] |
| 485 | async fn grep_files_reports_truncation_when_matches_exceed_the_cap() { |
| 486 | let tmp = tempdir().expect("tempdir"); |
| 487 | let ctx = ToolContext::new(tmp.path().to_path_buf()); |
| 488 | // Three files, two matches each: six matches against a cap of 3. |
| 489 | for name in ["a.txt", "b.txt", "c.txt"] { |
| 490 | fs::write(tmp.path().join(name), "needle\nneedle\n").expect("write"); |
| 491 | } |
| 492 | |
| 493 | let capped = GrepFilesTool |
| 494 | .execute(json!({"pattern": "needle", "max_results": 3}), &ctx) |
| 495 | .await |
| 496 | .expect("execute"); |
| 497 | let capped: Value = serde_json::from_str(&capped.content).unwrap(); |
| 498 | assert_eq!(capped["matches"].as_array().unwrap().len(), 3); |
| 499 | assert_eq!(capped["truncated"], json!(true), "{capped}"); |
| 500 | |
| 501 | // Exactly at the cap with nothing more is complete, not truncated. |
| 502 | let exact = GrepFilesTool |
| 503 | .execute(json!({"pattern": "needle", "max_results": 6}), &ctx) |
| 504 | .await |
| 505 | .expect("execute"); |
| 506 | let exact: Value = serde_json::from_str(&exact.content).unwrap(); |
| 507 | assert_eq!(exact["matches"].as_array().unwrap().len(), 6); |
| 508 | assert_eq!(exact["truncated"], json!(false), "{exact}"); |
| 509 | } |
| 510 | |
| 511 | #[cfg(unix)] |
| 512 | #[tokio::test] |
| 513 | async fn grep_files_skips_an_unreadable_subdirectory() { |
| 514 | use std::os::unix::fs::PermissionsExt; |
| 515 | |
| 516 | let tmp = tempdir().expect("tempdir"); |
| 517 | let ctx = ToolContext::new(tmp.path().to_path_buf()); |
| 518 | fs::write(tmp.path().join("found.txt"), "needle\n").expect("write"); |
| 519 | let locked = tmp.path().join("locked"); |
| 520 | fs::create_dir(&locked).expect("mkdir"); |
| 521 | fs::write(locked.join("hidden.txt"), "needle\n").expect("write"); |
| 522 | fs::set_permissions(&locked, fs::Permissions::from_mode(0o000)).expect("chmod"); |
| 523 | if fs::read_dir(&locked).is_ok() { |
| 524 | // Running with privileges that ignore directory modes; nothing to prove. |
| 525 | fs::set_permissions(&locked, fs::Permissions::from_mode(0o755)).expect("chmod"); |
| 526 | return; |
| 527 | } |
| 528 | |
| 529 | let result = GrepFilesTool |
| 530 | .execute(json!({"pattern": "needle"}), &ctx) |
| 531 | .await; |
| 532 | fs::set_permissions(&locked, fs::Permissions::from_mode(0o755)).expect("chmod"); |
| 533 | |
| 534 | let result = result.expect("an unreadable subdirectory must not fail the search"); |
| 535 | let parsed: Value = serde_json::from_str(&result.content).unwrap(); |
| 536 | let matches = parsed["matches"].as_array().unwrap(); |
| 537 | assert_eq!(matches.len(), 1, "{parsed}"); |
| 538 | assert!( |
| 539 | matches[0]["file"] |
| 540 | .as_str() |
| 541 | .is_some_and(|file| file.ends_with("found.txt")), |
| 542 | "{parsed}" |
| 543 | ); |
| 544 | assert_eq!(parsed["unreadable_paths_skipped"], json!(1), "{parsed}"); |
| 545 | } |
| 546 | |
| 547 | #[test] |
| 548 | fn double_star_without_a_directory_boundary_is_a_string_prefix() { |
| 549 | assert!(matches_glob("src/test_foo.rs", "src/test_**")); |
| 550 | assert!(matches_glob("src/test_dir/a.rs", "src/test_**")); |
| 551 | assert!(!matches_glob("src/other.rs", "src/test_**")); |
| 552 | } |
| 553 | |
| 554 | #[tokio::test] |
| 555 | async fn grep_files_include_directory_glob_matches_nested_files() { |
| 556 | let tmp = tempdir().expect("tempdir"); |
| 557 | let ctx = ToolContext::new(tmp.path().to_path_buf()); |
| 558 | fs::create_dir_all(tmp.path().join("src").join("inner")).expect("mkdir"); |
| 559 | fs::create_dir_all(tmp.path().join("srcs")).expect("mkdir"); |
| 560 | fs::write( |
| 561 | tmp.path().join("src").join("inner").join("a.rs"), |
| 562 | "needle\n", |
| 563 | ) |
| 564 | .expect("write"); |
| 565 | fs::write(tmp.path().join("srcs").join("b.rs"), "needle\n").expect("write"); |
| 566 | |
| 567 | let result = GrepFilesTool |
| 568 | .execute(json!({"pattern": "needle", "include": ["src/**"]}), &ctx) |
| 569 | .await |
| 570 | .expect("execute"); |
| 571 | let parsed: Value = serde_json::from_str(&result.content).unwrap(); |
| 572 | let files: Vec<&str> = parsed["matches"] |
| 573 | .as_array() |
| 574 | .unwrap() |
| 575 | .iter() |
| 576 | .filter_map(|m| m["file"].as_str()) |
| 577 | .collect(); |
| 578 | assert_eq!(files.len(), 1, "{parsed}"); |
| 579 | assert!( |
| 580 | files[0].replace('\\', "/").ends_with("src/inner/a.rs"), |
| 581 | "{parsed}" |
| 582 | ); |
| 583 | } |
| 584 | |
| 585 | #[tokio::test] |
| 586 | async fn grep_files_stops_at_the_cap_without_scanning_further_files() { |
| 587 | let tmp = tempdir().expect("tempdir"); |
| 588 | let ctx = ToolContext::new(tmp.path().to_path_buf()); |
| 589 | for name in ["a.txt", "b.txt", "c.txt", "d.txt", "e.txt"] { |
| 590 | fs::write(tmp.path().join(name), "needle\n").expect("write"); |
| 591 | } |
| 592 | |
| 593 | let result = GrepFilesTool |
| 594 | .execute(json!({"pattern": "needle", "max_results": 2}), &ctx) |
| 595 | .await |
| 596 | .expect("execute"); |
| 597 | let parsed: Value = serde_json::from_str(&result.content).unwrap(); |
| 598 | assert_eq!(parsed["matches"].as_array().unwrap().len(), 2, "{parsed}"); |
| 599 | // The budget fills exactly at the end of the second file; the walk |
| 600 | // must stop there rather than scan on looking for one more match. |
| 601 | assert_eq!(parsed["files_searched"], json!(2), "{parsed}"); |
| 602 | assert_eq!(parsed["truncated"], json!(true), "{parsed}"); |
| 603 | } |
| 604 | |
| 605 | #[cfg(unix)] |
| 606 | #[tokio::test] |
| 607 | async fn grep_files_counts_an_unreadable_file() { |
| 608 | use std::os::unix::fs::PermissionsExt; |
| 609 | |
| 610 | let tmp = tempdir().expect("tempdir"); |
| 611 | let ctx = ToolContext::new(tmp.path().to_path_buf()); |
| 612 | fs::write(tmp.path().join("found.txt"), "needle\n").expect("write"); |
| 613 | let locked = tmp.path().join("locked.txt"); |
| 614 | fs::write(&locked, "needle\n").expect("write"); |
| 615 | fs::set_permissions(&locked, fs::Permissions::from_mode(0o000)).expect("chmod"); |
| 616 | if fs::File::open(&locked).is_ok() { |
| 617 | // Running with privileges that ignore file modes; nothing to prove. |
| 618 | fs::set_permissions(&locked, fs::Permissions::from_mode(0o644)).expect("chmod"); |
| 619 | return; |
| 620 | } |
| 621 | |
| 622 | let result = GrepFilesTool |
| 623 | .execute(json!({"pattern": "needle"}), &ctx) |
| 624 | .await; |
| 625 | fs::set_permissions(&locked, fs::Permissions::from_mode(0o644)).expect("chmod"); |
| 626 | |
| 627 | let parsed: Value = serde_json::from_str(&result.expect("execute").content).unwrap(); |
| 628 | assert_eq!(parsed["matches"].as_array().unwrap().len(), 1, "{parsed}"); |
| 629 | assert_eq!(parsed["unreadable_paths_skipped"], json!(1), "{parsed}"); |
| 630 | } |
| 631 |