reflex-search 2.3.0

A local-first, structure-aware code search engine for AI agents
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
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
//! Snapshot engine: point-in-time capture of structural index state
//!
//! Snapshots capture the structural facts needed for diffing: file metadata,
//! dependency edges, and aggregate metrics. Each snapshot is a standalone SQLite
//! database under `.reflex/snapshots/`.

use anyhow::{Context, Result};
use chrono::{Datelike, Local};
use rusqlite::Connection;
use serde::{Deserialize, Serialize};
use std::path::{Path, PathBuf};

use crate::cache::CacheManager;
use crate::git;

/// Information about a snapshot
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct SnapshotInfo {
    /// Unique identifier (timestamp-based filename without extension)
    pub id: String,
    /// Full path to the snapshot database
    pub path: PathBuf,
    /// When the snapshot was created
    pub timestamp: String,
    /// Git branch at snapshot time (if available)
    pub git_branch: Option<String>,
    /// Git commit SHA at snapshot time (if available)
    pub git_commit: Option<String>,
    /// Reflex version that created the snapshot
    pub reflex_version: String,
    /// Number of files in the snapshot
    pub file_count: usize,
    /// Total line count across all files
    pub total_lines: usize,
    /// Number of dependency edges
    pub edge_count: usize,
    /// Database file size in bytes
    pub size_bytes: u64,
    /// blake3 hash of all (path, content_hash) pairs — used for change detection
    #[serde(skip_serializing_if = "Option::is_none")]
    pub content_fingerprint: Option<String>,
}

/// Report from garbage collection
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct GcReport {
    pub snapshots_before: usize,
    pub snapshots_after: usize,
    pub removed: usize,
    pub corrupted_removed: usize,
    pub space_freed_bytes: u64,
}

/// Result of ensuring a snapshot exists
#[derive(Debug)]
pub enum EnsureSnapshotResult {
    /// A new snapshot was created because the index changed (or no snapshot existed)
    Created(SnapshotInfo),
    /// The latest snapshot's fingerprint matches the current index
    Reused(SnapshotInfo),
}

/// Get the snapshots directory for a cache
pub fn get_snapshots_dir(cache: &CacheManager) -> PathBuf {
    cache.path().join("snapshots")
}

/// Compute a deterministic fingerprint of the current index state.
///
/// Hashes all (path, content_hash) pairs from meta.db's files table (the last
/// indexed tree), sorted by path. The streaming blake3 hasher avoids allocating all
/// pairs in memory.
pub fn compute_index_fingerprint(cache: &CacheManager) -> Result<String> {
    let meta_db_path = cache.path().join("meta.db");
    if !meta_db_path.exists() {
        anyhow::bail!("No index found. Run `rfx index` first.");
    }

    let conn = crate::cache::open_meta_db(&meta_db_path)
        .context("Failed to open meta.db for fingerprint")?;

    let mut stmt = conn.prepare(
        "SELECT f.path, f.hash
         FROM files f
         ORDER BY f.path",
    )?;

    let mut hasher = blake3::Hasher::new();
    let mut rows = stmt.query([])?;
    while let Some(row) = rows.next()? {
        let path: String = row.get(0)?;
        let hash: String = row.get(1)?;
        hasher.update(path.as_bytes());
        hasher.update(b":");
        hasher.update(hash.as_bytes());
        hasher.update(b"\n");
    }

    Ok(hasher.finalize().to_hex().to_string())
}

/// Ensure a current snapshot exists.
///
/// Creates a new snapshot if:
/// - No snapshots exist yet
/// - The latest snapshot's fingerprint doesn't match the current index
///
/// Runs GC after creating a new snapshot.
pub fn ensure_snapshot(
    cache: &CacheManager,
    retention: &super::config::RetentionConfig,
) -> Result<EnsureSnapshotResult> {
    let current_fingerprint = compute_index_fingerprint(cache)?;
    let snapshots = list_snapshots(cache)?;

    if let Some(latest) = snapshots.first()
        && latest.content_fingerprint.as_deref() == Some(&current_fingerprint)
    {
        return Ok(EnsureSnapshotResult::Reused(latest.clone()));
    }

    let info = create_snapshot(cache)?;
    run_gc(cache, retention)?;
    Ok(EnsureSnapshotResult::Created(info))
}

/// Create a snapshot of the current index state
///
/// Copies structural data from meta.db into a standalone snapshot database.
/// Returns info about the created snapshot.
pub fn create_snapshot(cache: &CacheManager) -> Result<SnapshotInfo> {
    let snapshots_dir = get_snapshots_dir(cache);
    std::fs::create_dir_all(&snapshots_dir).context("Failed to create snapshots directory")?;

    let now = Local::now();
    let timestamp = now.format("%Y%m%d_%H%M%S").to_string();
    let snapshot_path = snapshots_dir.join(format!("{}.db", timestamp));

    // Check meta.db exists
    let meta_db_path = cache.path().join("meta.db");
    if !meta_db_path.exists() {
        anyhow::bail!("No index found. Run `rfx index` first.");
    }

    // Create snapshot database
    let conn = Connection::open(&snapshot_path).context("Failed to create snapshot database")?;

    // Enable WAL mode for better concurrent access
    conn.execute_batch("PRAGMA journal_mode=WAL;")?;

    // Create snapshot schema. `walk_seq`, `vendored`, `dep_id`, `package`,
    // `package_members` and `import_edges` are what the `DependencyIndex` graph
    // queries read (hotspots, cycles, islands); see `upgrade_graph_schema`.
    conn.execute_batch(
        "CREATE TABLE files (
            id INTEGER PRIMARY KEY,
            path TEXT NOT NULL,
            language TEXT,
            line_count INTEGER DEFAULT 0,
            walk_seq INTEGER NOT NULL DEFAULT 0,
            vendored INTEGER NOT NULL DEFAULT 0
        );

        CREATE TABLE dependency_edges (
            source_file_id INTEGER NOT NULL,
            target_file_id INTEGER NOT NULL,
            import_type TEXT NOT NULL,
            dep_id INTEGER,
            package TEXT
        );

        CREATE TABLE package_members (
            package TEXT NOT NULL,
            member TEXT NOT NULL,
            file_id INTEGER NOT NULL,
            PRIMARY KEY (package, member, file_id)
        ) WITHOUT ROWID;

        -- The edges as meta.db's `import_edges` view names them (already package-
        -- expanded and free of vendored files when copied)
        CREATE VIEW import_edges AS
        SELECT dep_id, source_file_id AS src, target_file_id AS dst, import_type, package
        FROM dependency_edges;

        CREATE TABLE metrics (
            module_path TEXT PRIMARY KEY,
            file_count INTEGER NOT NULL,
            total_lines INTEGER NOT NULL
        );

        CREATE TABLE metadata (
            key TEXT PRIMARY KEY,
            value TEXT NOT NULL
        );

        -- Compatibility view so DependencyIndex methods work transparently
        CREATE VIEW file_dependencies AS
        SELECT source_file_id AS file_id,
               '' AS imported_path,
               target_file_id AS resolved_file_id,
               import_type,
               0 AS line_number,
               NULL AS imported_symbols
        FROM dependency_edges;

        -- Empty exports view for schema compatibility
        CREATE VIEW file_exports AS
        SELECT 0 AS id, 0 AS file_id, NULL AS exported_symbol,
               '' AS source_path, NULL AS resolved_source_id, 0 AS line_number
        WHERE 0;

        CREATE INDEX idx_dep_edges_source ON dependency_edges(source_file_id);
        CREATE INDEX idx_dep_edges_target ON dependency_edges(target_file_id);
        CREATE INDEX idx_files_path ON files(path);",
    )?;

    // Attach meta.db and copy data
    conn.execute(
        "ATTACH DATABASE ?1 AS source",
        [meta_db_path.to_str().unwrap()],
    )?;

    // Copy files
    conn.execute(
        "INSERT INTO files (id, path, language, line_count, walk_seq, vendored)
         SELECT id, path, language, line_count, walk_seq, vendored FROM source.files",
        [],
    )?;

    // Copy dependency edges (file-resolved and package-expanded, `import_edges`)
    conn.execute(
        "INSERT INTO dependency_edges (source_file_id, target_file_id, import_type, dep_id, package)
         SELECT src, dst, import_type, dep_id, package
         FROM source.import_edges",
        [],
    )?;

    // Package members: islands group the files of one Go package
    conn.execute(
        "INSERT INTO package_members (package, member, file_id)
         SELECT package, member, file_id FROM source.package_members",
        [],
    )?;

    // Compute module metrics by aggregating files by top-level directory
    conn.execute(
        "INSERT INTO metrics (module_path, file_count, total_lines)
         SELECT
             CASE
                 WHEN INSTR(path, '/') > 0 THEN SUBSTR(path, 1, INSTR(path, '/') - 1)
                 ELSE '.'
             END AS module_path,
             COUNT(*) AS file_count,
             COALESCE(SUM(line_count), 0) AS total_lines
         FROM files
         GROUP BY module_path",
        [],
    )?;

    // Detach source
    conn.execute("DETACH DATABASE source", [])?;

    // Write metadata
    let git_state = git::get_git_state_optional(".").unwrap_or(None);

    let metadata = vec![
        ("timestamp", now.to_rfc3339()),
        ("reflex_version", env!("CARGO_PKG_VERSION").to_string()),
        ("schema_version", SCHEMA_VERSION.to_string()),
    ];

    for (key, value) in &metadata {
        conn.execute(
            "INSERT INTO metadata (key, value) VALUES (?1, ?2)",
            rusqlite::params![key, value],
        )?;
    }

    if let Some(ref state) = git_state {
        conn.execute(
            "INSERT INTO metadata (key, value) VALUES ('git_branch', ?1)",
            [&state.branch],
        )?;
        conn.execute(
            "INSERT INTO metadata (key, value) VALUES ('git_commit', ?1)",
            [&state.commit],
        )?;
    }

    // Store content fingerprint for change detection
    let fingerprint = compute_index_fingerprint(cache)?;
    conn.execute(
        "INSERT INTO metadata (key, value) VALUES ('content_fingerprint', ?1)",
        [&fingerprint],
    )?;

    // Gather stats for the returned info
    let file_count: usize = conn.query_row("SELECT COUNT(*) FROM files", [], |row| row.get(0))?;
    let total_lines: usize = conn.query_row(
        "SELECT COALESCE(SUM(line_count), 0) FROM files",
        [],
        |row| row.get(0),
    )?;
    let edge_count: usize = conn.query_row("SELECT COUNT(*) FROM dependency_edges", [], |row| {
        row.get(0)
    })?;

    // Close connection before getting file size
    drop(conn);

    let size_bytes = std::fs::metadata(&snapshot_path)
        .map(|m| m.len())
        .unwrap_or(0);

    Ok(SnapshotInfo {
        id: timestamp.clone(),
        path: snapshot_path,
        timestamp: now.to_rfc3339(),
        git_branch: git_state.as_ref().map(|s| s.branch.clone()),
        git_commit: git_state.as_ref().map(|s| s.commit.clone()),
        reflex_version: env!("CARGO_PKG_VERSION").to_string(),
        file_count,
        total_lines,
        edge_count,
        size_bytes,
        content_fingerprint: Some(fingerprint),
    })
}

/// List all available snapshots, sorted by timestamp (newest first)
pub fn list_snapshots(cache: &CacheManager) -> Result<Vec<SnapshotInfo>> {
    let snapshots_dir = get_snapshots_dir(cache);
    if !snapshots_dir.exists() {
        return Ok(Vec::new());
    }

    let mut snapshots = Vec::new();

    for entry in std::fs::read_dir(&snapshots_dir)? {
        let entry = entry?;
        let path = entry.path();

        if path.extension().is_some_and(|ext| ext == "db") {
            match read_snapshot_info(&path) {
                Ok(info) => snapshots.push(info),
                Err(e) => {
                    log::warn!("Skipping corrupted snapshot {:?}: {}", path, e);
                }
            }
        }
    }

    // Sort newest first
    snapshots.sort_by(|a, b| b.id.cmp(&a.id));

    Ok(snapshots)
}

/// Get info about a specific snapshot by ID
pub fn get_snapshot(cache: &CacheManager, id: &str) -> Result<SnapshotInfo> {
    let snapshot_path = get_snapshots_dir(cache).join(format!("{}.db", id));
    if !snapshot_path.exists() {
        anyhow::bail!("Snapshot '{}' not found", id);
    }
    read_snapshot_info(&snapshot_path)
}

/// Delete a specific snapshot
pub fn delete_snapshot(cache: &CacheManager, id: &str) -> Result<()> {
    let snapshot_path = get_snapshots_dir(cache).join(format!("{}.db", id));
    if snapshot_path.exists() {
        std::fs::remove_file(&snapshot_path).context("Failed to delete snapshot")?;
    }
    Ok(())
}

/// The snapshot schema `create_snapshot` writes. Version 1 (through 2.2.0) had no
/// `walk_seq`, `vendored`, `package_members` or `import_edges`, so every graph query
/// on it failed, and `rfx snapshot diff` showed no hotspots, cycles or islands.
const SCHEMA_VERSION: u32 = 2;

/// Bring a version 1 snapshot up to the schema the graph queries read, in place.
/// What version 1 did not record is filled in: walk order is id order, no file is
/// vendored, and there are no package groups. A version 2 snapshot is left as it is.
pub fn upgrade_graph_schema(path: &Path) -> Result<()> {
    // The pragma helper, as `DependencyIndex::from_db_path` uses on the same file
    let conn = crate::cache::open_meta_db(path).context("Failed to open snapshot database")?;
    let has_walk_seq: bool = conn.query_row(
        "SELECT COUNT(*) FROM pragma_table_info('files') WHERE name = 'walk_seq'",
        [],
        |row| row.get::<_, i64>(0).map(|n| n > 0),
    )?;
    if has_walk_seq {
        return Ok(());
    }
    conn.execute_batch(&format!(
        "BEGIN;
         ALTER TABLE files ADD COLUMN walk_seq INTEGER NOT NULL DEFAULT 0;
         UPDATE files SET walk_seq = id;
         ALTER TABLE files ADD COLUMN vendored INTEGER NOT NULL DEFAULT 0;
         CREATE TABLE IF NOT EXISTS package_members (
             package TEXT NOT NULL,
             member TEXT NOT NULL,
             file_id INTEGER NOT NULL,
             PRIMARY KEY (package, member, file_id)
         ) WITHOUT ROWID;
         CREATE VIEW IF NOT EXISTS import_edges AS
         SELECT rowid AS dep_id, source_file_id AS src, target_file_id AS dst, import_type,
                NULL AS package
         FROM dependency_edges;
         INSERT OR REPLACE INTO metadata (key, value) VALUES ('schema_version', '{SCHEMA_VERSION}');
         COMMIT;"
    ))
    .with_context(|| format!("Failed to upgrade snapshot {}", path.display()))?;
    Ok(())
}

/// Validate a snapshot database integrity
pub fn validate_snapshot(path: &Path) -> Result<bool> {
    let conn = Connection::open(path)?;
    let result: String = conn.query_row("PRAGMA integrity_check", [], |row| row.get(0))?;
    Ok(result == "ok")
}

/// Run garbage collection based on retention policy
///
/// Keeps the most recent snapshots according to the retention tiers:
/// - `daily`: Keep N most recent daily snapshots
/// - `weekly`: Keep N most recent weekly snapshots (one per week)
/// - `monthly`: Keep N most recent monthly snapshots (one per month)
pub fn run_gc(cache: &CacheManager, config: &super::config::RetentionConfig) -> Result<GcReport> {
    let snapshots = list_snapshots(cache)?;
    let snapshots_before = snapshots.len();
    let mut to_keep: std::collections::HashSet<String> = std::collections::HashSet::new();
    let mut corrupted_removed = 0u64;

    // First pass: remove corrupted snapshots
    for snapshot in &snapshots {
        match validate_snapshot(&snapshot.path) {
            Ok(true) => {}
            _ => {
                log::warn!("Removing corrupted snapshot: {}", snapshot.id);
                let _ = std::fs::remove_file(&snapshot.path);
                corrupted_removed += 1;
                continue;
            }
        }
    }

    // Collect valid snapshots with parsed timestamps
    let valid_snapshots: Vec<&SnapshotInfo> = snapshots
        .iter()
        .filter(|s| validate_snapshot(&s.path).unwrap_or(false))
        .collect();

    // Already sorted newest first from list_snapshots
    // Keep daily tier: the N most recent
    for snapshot in valid_snapshots.iter().take(config.daily) {
        to_keep.insert(snapshot.id.clone());
    }

    // Keep weekly tier: one per calendar week, most recent N weeks
    let mut weeks_seen: std::collections::HashSet<String> = std::collections::HashSet::new();
    let mut weekly_kept = 0;
    for snapshot in &valid_snapshots {
        if weekly_kept >= config.weekly {
            break;
        }
        let week_key = snapshot_to_week_key(&snapshot.id);
        if weeks_seen.insert(week_key) {
            to_keep.insert(snapshot.id.clone());
            weekly_kept += 1;
        }
    }

    // Keep monthly tier: one per calendar month, most recent N months
    let mut months_seen: std::collections::HashSet<String> = std::collections::HashSet::new();
    let mut monthly_kept = 0;
    for snapshot in &valid_snapshots {
        if monthly_kept >= config.monthly {
            break;
        }
        let month_key = snapshot_to_month_key(&snapshot.id);
        if months_seen.insert(month_key) {
            to_keep.insert(snapshot.id.clone());
            monthly_kept += 1;
        }
    }

    // Remove snapshots not in any keep tier
    let mut space_freed: u64 = 0;
    let mut removed = 0usize;

    for snapshot in &valid_snapshots {
        if !to_keep.contains(&snapshot.id) {
            space_freed += snapshot.size_bytes;
            let _ = std::fs::remove_file(&snapshot.path);
            removed += 1;
        }
    }

    Ok(GcReport {
        snapshots_before,
        snapshots_after: snapshots_before - removed - corrupted_removed as usize,
        removed: removed + corrupted_removed as usize,
        corrupted_removed: corrupted_removed as usize,
        space_freed_bytes: space_freed,
    })
}

/// Read snapshot info from a database file
fn read_snapshot_info(path: &Path) -> Result<SnapshotInfo> {
    let conn = Connection::open(path).context("Failed to open snapshot database")?;

    let id = path
        .file_stem()
        .and_then(|s| s.to_str())
        .unwrap_or("unknown")
        .to_string();

    // Read metadata
    let get_meta = |key: &str| -> Option<String> {
        conn.query_row("SELECT value FROM metadata WHERE key = ?1", [key], |row| {
            row.get(0)
        })
        .ok()
    };

    let timestamp = get_meta("timestamp").unwrap_or_else(|| id.clone());
    let git_branch = get_meta("git_branch");
    let git_commit = get_meta("git_commit");
    let reflex_version = get_meta("reflex_version").unwrap_or_else(|| "unknown".to_string());
    let content_fingerprint = get_meta("content_fingerprint");

    // Gather stats
    let file_count: usize = conn
        .query_row("SELECT COUNT(*) FROM files", [], |row| row.get(0))
        .unwrap_or(0);
    let total_lines: usize = conn
        .query_row(
            "SELECT COALESCE(SUM(line_count), 0) FROM files",
            [],
            |row| row.get(0),
        )
        .unwrap_or(0);
    let edge_count: usize = conn
        .query_row("SELECT COUNT(*) FROM dependency_edges", [], |row| {
            row.get(0)
        })
        .unwrap_or(0);

    let size_bytes = std::fs::metadata(path).map(|m| m.len()).unwrap_or(0);

    Ok(SnapshotInfo {
        id,
        path: path.to_path_buf(),
        timestamp,
        git_branch,
        git_commit,
        reflex_version,
        file_count,
        total_lines,
        edge_count,
        size_bytes,
        content_fingerprint,
    })
}

/// Extract week key from snapshot ID (YYYYMMDD_HHMMSS -> YYYY-WXX)
fn snapshot_to_week_key(id: &str) -> String {
    // Parse YYYYMMDD from ID
    if id.len() >= 8 {
        let year: i32 = id[0..4].parse().unwrap_or(2000);
        let month: u32 = id[4..6].parse().unwrap_or(1);
        let day: u32 = id[6..8].parse().unwrap_or(1);

        if let Some(date) = chrono::NaiveDate::from_ymd_opt(year, month, day) {
            return format!("{}-W{:02}", year, date.iso_week().week());
        }
    }
    id.to_string()
}

/// Extract month key from snapshot ID (YYYYMMDD_HHMMSS -> YYYY-MM)
fn snapshot_to_month_key(id: &str) -> String {
    if id.len() >= 6 {
        format!("{}-{}", &id[0..4], &id[4..6])
    } else {
        id.to_string()
    }
}

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

    #[test]
    fn test_week_key() {
        assert_eq!(snapshot_to_week_key("20260407_120000"), "2026-W15");
    }

    #[test]
    fn test_month_key() {
        assert_eq!(snapshot_to_month_key("20260407_120000"), "2026-04");
        assert_eq!(snapshot_to_month_key("20261231_235959"), "2026-12");
    }

    #[test]
    fn test_snapshot_info_serialization() {
        let info = SnapshotInfo {
            id: "20260407_120000".to_string(),
            path: PathBuf::from("/tmp/test.db"),
            timestamp: "2026-04-07T12:00:00+00:00".to_string(),
            git_branch: Some("main".to_string()),
            git_commit: Some("abc123".to_string()),
            reflex_version: "1.0.5".to_string(),
            file_count: 100,
            total_lines: 10000,
            edge_count: 50,
            size_bytes: 1024,
            content_fingerprint: None,
        };
        let json = serde_json::to_string(&info).unwrap();
        assert!(json.contains("20260407_120000"));
    }
}