lean-ctx 3.6.25

Context Runtime for AI Agents with CCP. 63 MCP tools, 10 read modes, 60+ compression patterns, cross-session memory (CCP), persistent AI knowledge with temporal facts + contradiction detection, multi-agent context sharing, LITM-aware positioning, AAAK compact format, adaptive compression with Thompson Sampling bandits. Supports 24+ AI tools. Reduces LLM token consumption by up to 99%.
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
use std::collections::HashSet;
use std::path::Path;
use std::path::PathBuf;

use ignore::WalkBuilder;
use regex::RegexBuilder;

use crate::core::protocol;
use crate::core::symbol_map::{self, SymbolMap};
use crate::core::tokens::count_tokens;
use crate::tools::CrpMode;

pub(crate) const MAX_FILE_SIZE: u64 = 512_000;
pub(crate) const MAX_WALK_DEPTH: usize = 20;
const MAX_MATCH_LINE_WIDTH: usize = 150;

/// Searches files for a regex pattern with compressed output and monorepo scope hints.
pub fn handle(
    pattern: &str,
    dir: &str,
    ext_filter: Option<&str>,
    max_results: usize,
    _crp_mode: CrpMode,
    respect_gitignore: bool,
    allow_secret_paths: bool,
) -> (String, usize) {
    let ext_filter = ext_filter.map(|e| e.strip_prefix('.').unwrap_or(e));
    const MAX_PATTERN_LEN: usize = 1024;
    const MAX_REGEX_SIZE: usize = 1 << 20; // 1 MiB DFA limit

    let redact = crate::core::redaction::redaction_enabled_for_active_role();
    if pattern.len() > MAX_PATTERN_LEN {
        return (
            format!(
                "ERROR: pattern too long ({} > {MAX_PATTERN_LEN} chars)",
                pattern.len()
            ),
            0,
        );
    }
    let re = match RegexBuilder::new(pattern)
        .size_limit(MAX_REGEX_SIZE)
        .dfa_size_limit(MAX_REGEX_SIZE)
        .build()
    {
        Ok(r) => r,
        Err(e) => return (format!("ERROR: invalid regex: {e}"), 0),
    };

    let root = Path::new(dir);
    if !root.exists() {
        return (format!("ERROR: {dir} does not exist"), 0);
    }

    let mut files: Vec<PathBuf> = Vec::new();
    let mut matches = Vec::new();
    let mut raw_tokens_accum: usize = 0;
    let mut files_searched = 0u32;
    let mut files_skipped_size = 0u32;
    let mut files_skipped_encoding = 0u32;
    let mut files_skipped_boundary = 0u32;

    // Fast path: a warm resident trigram index narrows the candidate files in
    // memory, eliminating the per-call directory walk + full-corpus read. The
    // index covers the exact same file universe as the walk below, and matches
    // are still verified line-by-line with the same regex — so results are
    // identical. Missing/stale index → returns None and triggers a background
    // (re)build; this call uses the walk fallback.
    let used_index = if let Some(idx) =
        crate::core::search_index::get_fresh(dir, respect_gitignore, allow_secret_paths)
    {
        files = idx.candidate_paths(pattern, ext_filter).into_paths();
        true
    } else {
        false
    };

    if !used_index {
        let walker = WalkBuilder::new(root)
            .hidden(true)
            .max_depth(Some(MAX_WALK_DEPTH))
            .git_ignore(respect_gitignore)
            .git_global(respect_gitignore)
            .git_exclude(respect_gitignore)
            .build();

        for entry in walker.filter_map(std::result::Result::ok) {
            if entry.file_type().is_none_or(|ft| ft.is_dir()) {
                continue;
            }

            if entry.file_type().is_some_and(|ft| ft.is_symlink()) {
                continue;
            }

            let path = entry.path();

            if is_binary_ext(path) || is_generated_file(path) {
                continue;
            }

            if !allow_secret_paths && crate::core::io_boundary::is_secret_like(path).is_some() {
                files_skipped_boundary += 1;
                continue;
            }

            if let Some(ext) = ext_filter {
                let file_ext = path.extension().and_then(|e| e.to_str()).unwrap_or("");
                if file_ext != ext {
                    continue;
                }
            }

            if let Ok(meta) = std::fs::metadata(path) {
                if meta.len() > MAX_FILE_SIZE {
                    files_skipped_size += 1;
                    continue;
                }
            }

            files.push(path.to_path_buf());
        }
    }

    // Deterministic search: stable file ordering makes max_results truncation reproducible.
    files.sort_unstable_by(|a, b| a.as_os_str().cmp(b.as_os_str()));

    let root_str = root.to_string_lossy();
    for path in &files {
        if matches.len() >= max_results {
            break;
        }

        let Ok(content) = std::fs::read_to_string(path) else {
            files_skipped_encoding += 1;
            continue;
        };

        files_searched += 1;

        for (i, line) in content.lines().enumerate() {
            if re.is_match(line) {
                let short_path =
                    protocol::shorten_path_relative(&path.to_string_lossy(), &root_str);
                // Count raw tokens incrementally (avoids separate Vec + join)
                raw_tokens_accum += count_tokens(line.trim()) + 2;
                let mut shown = if redact {
                    crate::core::redaction::redact_text(line.trim())
                } else {
                    line.trim().to_string()
                };
                if shown.len() > MAX_MATCH_LINE_WIDTH {
                    shown.truncate(shown.floor_char_boundary(MAX_MATCH_LINE_WIDTH));
                    shown.push_str("...");
                }
                matches.push(format!("{short_path}:{} {}", i + 1, shown));
                if matches.len() >= max_results {
                    break;
                }
            }
        }
    }

    if matches.is_empty() {
        let mut msg = format!("0 matches for '{pattern}' in {files_searched} files");
        if files_skipped_size > 0 {
            msg.push_str(&format!(" ({files_skipped_size} large files skipped)"));
        }
        if files_skipped_encoding > 0 {
            msg.push_str(&format!(
                " ({files_skipped_encoding} files skipped: binary/encoding)"
            ));
        }
        if files_skipped_boundary > 0 {
            msg.push_str(&format!(
                " ({files_skipped_boundary} secret-like files skipped by boundary policy)"
            ));
        }
        return (msg, 0);
    }

    // Prefix-cache-friendly: structural file list before per-query match content
    let matched_files: Vec<&str> = {
        let mut seen = HashSet::new();
        matches
            .iter()
            .filter_map(|m| {
                let file = extract_file_from_match(m);
                if seen.insert(file) {
                    Some(file)
                } else {
                    None
                }
            })
            .collect()
    };

    let mut result = format!("{} matches in {} files", matches.len(), files_searched);
    if matched_files.len() > 1 {
        if matched_files.len() <= 10 {
            result.push_str(" [");
            result.push_str(&matched_files.join(", "));
            result.push(']');
        } else {
            let shown: Vec<&str> = matched_files.iter().take(8).copied().collect();
            result.push_str(&format!(
                " [{}, +{} more]",
                shown.join(", "),
                matched_files.len() - 8
            ));
        }
    }
    result.push_str(":\n");
    result.push_str(&matches.join("\n"));

    if files_skipped_size > 0 {
        result.push_str(&format!("\n({files_skipped_size} files >512KB skipped)"));
    }
    if files_skipped_encoding > 0 {
        result.push_str(&format!(
            "\n({files_skipped_encoding} files skipped: binary/encoding)"
        ));
    }
    if files_skipped_boundary > 0 {
        result.push_str(&format!(
            "\n({files_skipped_boundary} secret-like files skipped by boundary policy)"
        ));
    }

    let scope_hint = {
        static SHOWN: std::sync::atomic::AtomicBool = std::sync::atomic::AtomicBool::new(false);
        if SHOWN.load(std::sync::atomic::Ordering::Relaxed) {
            None
        } else {
            let hint = monorepo_scope_hint(&matches, dir);
            if hint.is_some() {
                SHOWN.store(true, std::sync::atomic::Ordering::Relaxed);
            }
            hint
        }
    };

    if let Some(delta) = crate::core::search_delta::compute_delta(pattern, &matches) {
        let native_estimate = (raw_tokens_accum as f64 * 2.5).ceil() as usize;
        let original = native_estimate.max(raw_tokens_accum);
        return (delta, original);
    }

    if symbol_map::substitution_enabled() {
        let file_ext = ext_filter.unwrap_or("rs");
        let mut sym = SymbolMap::new();
        let idents = symbol_map::extract_identifiers(&result, file_ext);
        for ident in &idents {
            sym.register(ident);
        }
        if sym.len() >= 3 {
            let sym_table = sym.format_table();
            let compressed = sym.apply(&result);
            let original_tok = count_tokens(&result);
            let compressed_tok = count_tokens(&compressed) + count_tokens(&sym_table);
            let net_saving = original_tok.saturating_sub(compressed_tok);
            if original_tok > 0 && net_saving * 100 / original_tok >= 5 {
                result = format!("{compressed}{sym_table}");
            }
        }
    }

    if let Some(hint) = scope_hint {
        result.push_str(&hint);
    }

    let native_estimate = (raw_tokens_accum as f64 * 2.5).ceil() as usize;
    let original = native_estimate.max(raw_tokens_accum);

    (result, original)
}

pub(crate) fn is_binary_ext(path: &Path) -> bool {
    let ext = path.extension().and_then(|e| e.to_str()).unwrap_or("");
    matches!(
        ext,
        "png"
            | "jpg"
            | "jpeg"
            | "gif"
            | "webp"
            | "ico"
            | "svg"
            | "woff"
            | "woff2"
            | "ttf"
            | "eot"
            | "pdf"
            | "zip"
            | "tar"
            | "gz"
            | "br"
            | "zst"
            | "bz2"
            | "xz"
            | "mp3"
            | "mp4"
            | "webm"
            | "ogg"
            | "wasm"
            | "so"
            | "dylib"
            | "dll"
            | "exe"
            | "lock"
            | "map"
            | "snap"
            | "patch"
            | "db"
            | "sqlite"
            | "parquet"
            | "arrow"
            | "bin"
            | "o"
            | "a"
            | "class"
            | "pyc"
            | "pyo"
    )
}

pub(crate) fn is_generated_file(path: &Path) -> bool {
    let name = path.file_name().and_then(|n| n.to_str()).unwrap_or("");
    name.ends_with(".min.js")
        || name.ends_with(".min.css")
        || name.ends_with(".bundle.js")
        || name.ends_with(".chunk.js")
        || name.ends_with(".d.ts")
        || name.ends_with(".js.map")
        || name.ends_with(".css.map")
}

/// Extract file path from a grep match line, handling Windows drive letters (e.g. "C:").
fn extract_file_from_match(line: &str) -> &str {
    let start = if line.len() >= 2
        && line.as_bytes().first().is_some_and(u8::is_ascii_alphabetic)
        && line.as_bytes().get(1) == Some(&b':')
    {
        2
    } else {
        0
    };
    match line[start..].find(':') {
        Some(pos) => &line[..start + pos],
        None => line,
    }
}

fn monorepo_scope_hint(matches: &[String], search_dir: &str) -> Option<String> {
    let top_dirs: HashSet<&str> = matches
        .iter()
        .filter_map(|m| {
            let path = extract_file_from_match(m);
            let relative = path.strip_prefix("./").unwrap_or(path);
            let relative = relative.strip_prefix(search_dir).unwrap_or(relative);
            let relative = relative.strip_prefix('/').unwrap_or(relative);
            relative.split('/').next()
        })
        .collect();

    if top_dirs.len() > 3 {
        let mut dirs: Vec<&&str> = top_dirs.iter().collect();
        dirs.sort();
        let dir_list: Vec<String> = dirs.iter().take(6).map(|d| format!("'{d}'")).collect();
        let extra = if top_dirs.len() > 6 {
            format!(", +{} more", top_dirs.len() - 6)
        } else {
            String::new()
        };
        Some(format!(
            "\n\nResults span {} directories ({}{}). \
             Use the 'path' parameter to scope to a specific service, \
             e.g. path=\"{}/\".",
            top_dirs.len(),
            dir_list.join(", "),
            extra,
            dirs[0]
        ))
    } else {
        None
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::tools::CrpMode;

    #[test]
    fn search_results_are_deterministically_ordered_by_path() {
        let dir = tempfile::tempdir().unwrap();
        let a = dir.path().join("a.txt");
        let b = dir.path().join("b.txt");
        std::fs::write(&b, "match\n").unwrap();
        std::fs::write(&a, "match\n").unwrap();

        let (out, _orig) = handle(
            "match",
            dir.path().to_string_lossy().as_ref(),
            Some("txt"),
            10,
            CrpMode::Off,
            true,
            true,
        );

        let mut match_lines: Vec<&str> = out
            .lines()
            .filter(|l| l.contains(".txt:") && l.contains("match"))
            .collect();
        // Expect exactly the 2 match lines, ordered a.txt then b.txt.
        match_lines.truncate(2);
        assert_eq!(match_lines.len(), 2);
        assert!(
            match_lines[0].contains("a.txt:"),
            "first match should come from a.txt, got: {}",
            match_lines[0]
        );
        assert!(
            match_lines[1].contains("b.txt:"),
            "second match should come from b.txt, got: {}",
            match_lines[1]
        );
    }

    #[test]
    fn symbol_substitution_is_off_by_default() {
        let _lock = crate::core::data_dir::test_env_lock();
        std::env::remove_var("LEAN_CTX_SYMBOL_MAP");
        let dir = tempfile::tempdir().unwrap();
        let f = dir.path().join("a.rs");
        std::fs::write(
            &f,
            "fn longIdentifierAlpha() {}\nfn longIdentifierBeta() {}\nfn longIdentifierGamma() {}\n",
        )
        .unwrap();

        let (out, _orig) = handle(
            "longIdentifier",
            dir.path().to_string_lossy().as_ref(),
            Some("rs"),
            10,
            CrpMode::Off,
            true,
            true,
        );

        assert!(
            !out.contains("§MAP"),
            "default agent-facing output must not carry a §MAP table: {out}"
        );
        assert!(
            !out.contains('α'),
            "default agent-facing output must not carry α-symbols: {out}"
        );
        assert!(
            out.contains("longIdentifierAlpha"),
            "identifiers should appear raw by default: {out}"
        );
    }

    #[test]
    fn secret_like_files_are_skipped_by_default() {
        let dir = tempfile::tempdir().unwrap();
        let secret = dir.path().join("key.pem");
        let ok = dir.path().join("ok.txt");
        std::fs::write(&secret, "match\n").unwrap();
        std::fs::write(&ok, "match\n").unwrap();

        let (out, _orig) = handle(
            "match",
            dir.path().to_string_lossy().as_ref(),
            None,
            10,
            CrpMode::Off,
            true,
            false,
        );

        assert!(out.contains("ok.txt:"), "expected ok.txt match, got: {out}");
        assert!(
            !out.contains("key.pem:"),
            "secret-like file should be skipped, got: {out}"
        );
        assert!(
            out.contains("secret-like files skipped"),
            "expected boundary skip note, got: {out}"
        );
    }
}