datasphere 0.1.1

Background daemon that distills knowledge from Claude Code sessions into a searchable graph
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
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
use chrono::{Local, Utc};
use clap::{Parser, Subcommand};
use datasphere::{
    chunk_text, discover_sessions, discover_sessions_in_dir, embed, extract_knowledge,
    list_all_projects, read_transcript, AllProjectsWatcher, Job, JobStatus, Node, Processed,
    Queue, SessionInfo, SourceType, Store,
};
use std::path::PathBuf;
use std::time::{Duration, Instant};
use uuid::Uuid;

#[derive(Parser)]
#[command(name = "ds")]
#[command(about = "Datasphere - distills knowledge from Claude Code sessions")]
struct Cli {
    #[command(subcommand)]
    command: Commands,
}

#[derive(Subcommand)]
enum Commands {
    /// Show database statistics
    Stats,

    /// Show stored nodes
    Show {
        /// Maximum number of nodes to show
        #[arg(short, long, default_value = "10")]
        limit: usize,
    },

    /// Scan and distill transcripts (one-shot)
    Scan {
        /// Maximum number of transcripts to process (newest first)
        #[arg(short, long)]
        limit: Option<usize>,

        /// Project path to scan (defaults to current directory)
        #[arg(short, long)]
        project: Option<PathBuf>,
    },

    /// Start the daemon (watches all projects)
    Start {
        /// Run in foreground instead of daemonizing
        #[arg(short, long)]
        foreground: bool,
    },

    /// Stop the running daemon
    Stop,

    /// Show daemon status
    Status,

    /// Show or manage the job queue
    Queue {
        #[command(subcommand)]
        action: Option<QueueAction>,
    },

    /// Add a text file to the knowledge graph (no LLM distillation)
    Add {
        /// Path to the file to add
        file: PathBuf,
    },

    /// Search the knowledge graph for relevant nodes
    Query {
        /// Search query text
        query: String,

        /// Maximum number of results
        #[arg(short, long, default_value = "5")]
        limit: usize,

        /// Output format (text or json)
        #[arg(short, long, default_value = "text")]
        format: String,
    },

    /// Delete everything (database + queue) and start fresh
    Reset,

    /// Find nodes similar to a given node
    Related {
        /// Node ID (UUID) to find related nodes for
        node_id: String,

        /// Maximum number of results
        #[arg(short, long, default_value = "5")]
        limit: usize,

        /// Output format (text or json)
        #[arg(short, long, default_value = "text")]
        format: String,
    },
}

#[derive(Subcommand)]
enum QueueAction {
    /// Show queue counts (default)
    Status,
    /// List pending jobs
    Pending,
    /// Clear completed jobs
    Clear,
    /// Delete entire queue (all jobs, all statuses)
    Nuke,
}

/// Get the default database path
fn default_db_path() -> PathBuf {
    dirs::home_dir()
        .unwrap_or_else(|| PathBuf::from("."))
        .join(".datasphere")
        .join("db")
}

/// Get the daemon PID file path
fn daemon_pid_path() -> PathBuf {
    dirs::home_dir()
        .unwrap_or_else(|| PathBuf::from("."))
        .join(".datasphere")
        .join("daemon.pid")
}

/// Get the daemon log file path
fn daemon_log_path() -> PathBuf {
    dirs::home_dir()
        .unwrap_or_else(|| PathBuf::from("."))
        .join(".datasphere")
        .join("daemon.log")
}

/// Check if daemon is running, returns PID if running
fn is_daemon_running() -> Option<u32> {
    let pid_path = daemon_pid_path();
    if !pid_path.exists() {
        return None;
    }

    let pid_str = std::fs::read_to_string(&pid_path).ok()?;
    let pid: u32 = pid_str.trim().parse().ok()?;

    // Check if process is actually running
    #[cfg(unix)]
    {
        // kill(pid, 0) checks if process exists without sending a signal
        let result = unsafe { libc::kill(pid as i32, 0) };
        if result == 0 {
            return Some(pid);
        }
    }

    // PID file exists but process is dead - clean up
    let _ = std::fs::remove_file(&pid_path);
    None
}

/// Format bytes as human-readable size
fn format_size(bytes: u64) -> String {
    const KB: u64 = 1024;
    const MB: u64 = KB * 1024;
    const GB: u64 = MB * 1024;

    if bytes >= GB {
        format!("{:.2} GB", bytes as f64 / GB as f64)
    } else if bytes >= MB {
        format!("{:.2} MB", bytes as f64 / MB as f64)
    } else if bytes >= KB {
        format!("{:.2} KB", bytes as f64 / KB as f64)
    } else {
        format!("{} bytes", bytes)
    }
}

/// Calculate total size of a directory recursively
fn dir_size(path: &PathBuf) -> u64 {
    if !path.exists() {
        return 0;
    }

    walkdir::WalkDir::new(path)
        .into_iter()
        .filter_map(|e| e.ok())
        .filter(|e| e.file_type().is_file())
        .filter_map(|e| e.metadata().ok())
        .map(|m| m.len())
        .sum()
}

/// Hamming distance threshold for considering a session "changed"
/// AIDEV-NOTE: 10 bits out of 64 (~15%) means meaningful content change
const SIMHASH_CHANGE_THRESHOLD: u32 = 10;

/// Quick check if a transcript has any meaningful content (messages/summaries)
/// Returns false for empty or content-free transcripts to avoid queueing them
fn transcript_has_content(path: &std::path::Path) -> bool {
    match read_transcript(path) {
        Ok(entries) => entries.iter().any(|e| e.is_message() || e.is_summary()),
        Err(_) => false,
    }
}

/// Process a single session transcript
/// Returns (nodes_created, skipped) tuple
async fn process_session(
    store: &Store,
    session: &SessionInfo,
) -> Result<(usize, bool), Box<dyn std::error::Error>> {
    println!("  Reading transcript...");

    // Parse transcript
    let entries = read_transcript(&session.transcript_path)?;
    println!("  Parsed {} entries", entries.len());

    if entries.is_empty() {
        println!("  Empty transcript, skipping");
        return Ok((0, true));
    }

    // Get all messages for context
    let messages: Vec<_> = entries
        .iter()
        .filter(|e| e.is_message() || e.is_summary())
        .collect();

    if messages.is_empty() {
        println!("  No messages found, skipping");
        return Ok((0, true));
    }

    // Count message types
    let user_count = messages.iter().filter(|e| e.is_user()).count();
    let assistant_count = messages.iter().filter(|e| e.is_assistant()).count();
    let summary_count = messages.iter().filter(|e| e.is_summary()).count();

    println!(
        "  Collected {} items ({} user, {} assistant, {} summaries)",
        messages.len(),
        user_count,
        assistant_count,
        summary_count
    );

    // Format context for LLM
    let context = datasphere::format_context(&messages);
    if context.trim().is_empty() {
        println!("  Empty context after formatting, skipping");
        return Ok((0, true));
    }

    println!("  Context size: {} chars", context.len());

    // Compute SimHash of context
    let current_simhash = simhash::simhash(&context) as i64;

    // Check if already processed
    if let Some(existing) = store.get_processed(&session.session_id).await? {
        let hamming = simhash::hamming_distance(existing.simhash as u64, current_simhash as u64);

        if hamming <= SIMHASH_CHANGE_THRESHOLD {
            println!(
                "  Unchanged (simhash distance: {} bits, threshold: {})",
                hamming, SIMHASH_CHANGE_THRESHOLD
            );
            return Ok((0, true));
        }

        println!(
            "  Session changed (simhash distance: {} bits), re-distilling...",
            hamming
        );

        // Delete old nodes, then processed record
        let node_ids: Vec<Uuid> = existing.node_ids
            .iter()
            .filter_map(|id| id.parse::<Uuid>().ok())
            .collect();
        if !node_ids.is_empty() {
            store.delete_nodes(&node_ids).await?;
            println!("  Deleted {} old node(s)", node_ids.len());
        }
        store.delete_processed(&session.session_id).await?;
    }

    // Distill knowledge via LLM
    println!("  Distilling via LLM...");
    let distill_start = Instant::now();
    let extraction = match extract_knowledge(&context).await {
        Ok(result) => result,
        Err(e) => {
            eprintln!("  LLM extraction failed: {}", e);
            return Err(e.into());
        }
    };
    let distill_elapsed = distill_start.elapsed();

    // Log chunking info if used
    if extraction.chunks_used > 1 {
        println!("  Distilled {} chunks in {:.1}s", extraction.chunks_used, distill_elapsed.as_secs_f32());
    } else {
        println!("  Distilled in {:.1}s", distill_elapsed.as_secs_f32());
    }

    if extraction.insights.is_empty() {
        println!("  No substantive knowledge found");
        // Still record as processed
        let record = Processed {
            source_id: session.session_id.clone(),
            source_type: "session".to_string(),
            simhash: current_simhash,
            processed_at: Utc::now(),
            node_count: 0,
            node_ids: Vec::new(),
        };
        store.insert_processed(&record).await?;
        return Ok((0, false));
    }

    println!("  Extracted {} insight(s)", extraction.insights.len());

    // Embed and store each insight as a separate node
    let total_insights = extraction.insights.len();
    let mut node_ids = Vec::new();
    for (i, insight) in extraction.insights.into_iter().enumerate() {
        let embed_start = Instant::now();
        let embedding = embed(&insight.content).await?;
        println!("  Embedded {}/{} in {:.1}s", i + 1, total_insights, embed_start.elapsed().as_secs_f32());

        let node = insight.into_node(
            session.session_id.clone(),
            SourceType::Session,
            embedding,
        );
        let node_id = node.id.to_string();

        store.insert_node(&node).await?;

        node_ids.push(node_id);
    }

    // Record as processed
    let record = Processed {
        source_id: session.session_id.clone(),
        source_type: "session".to_string(),
        simhash: current_simhash,
        processed_at: Utc::now(),
        node_count: node_ids.len() as i32,
        node_ids,
    };
    store.insert_processed(&record).await?;

    println!("  Done! Created {} node(s)", record.node_count);
    Ok((record.node_count as usize, false))
}

/// Process a single text file (no LLM distillation, direct embedding)
/// Returns number of nodes created
async fn process_file(
    store: &Store,
    file_path: &PathBuf,
) -> Result<usize, Box<dyn std::error::Error>> {
    // Canonicalize path to avoid duplicates
    let canonical_path = file_path.canonicalize()
        .map_err(|e| format!("Failed to canonicalize path: {}", e))?;
    let source_id = canonical_path.to_string_lossy().to_string();

    println!("Processing file: {}", canonical_path.display());

    // Read file content
    let content = std::fs::read_to_string(&canonical_path)
        .map_err(|e| format!("Failed to read file: {}", e))?;

    if content.trim().is_empty() {
        println!("  Empty file, skipping");
        return Ok(0);
    }

    println!("  File size: {} chars", content.len());

    // Compute SimHash
    let current_simhash = simhash::simhash(&content) as i64;

    // Check if already processed
    if let Some(existing) = store.get_processed(&source_id).await? {
        let hamming = simhash::hamming_distance(existing.simhash as u64, current_simhash as u64);

        if hamming <= SIMHASH_CHANGE_THRESHOLD {
            println!(
                "  Unchanged (simhash distance: {} bits, threshold: {})",
                hamming, SIMHASH_CHANGE_THRESHOLD
            );
            return Ok(0);
        }

        println!(
            "  File changed (simhash distance: {} bits), re-embedding...",
            hamming
        );

        // Delete old nodes, then processed record
        let node_ids: Vec<Uuid> = existing.node_ids
            .iter()
            .filter_map(|id| id.parse::<Uuid>().ok())
            .collect();
        if !node_ids.is_empty() {
            store.delete_nodes(&node_ids).await?;
            println!("  Deleted {} old node(s)", node_ids.len());
        }
        store.delete_processed(&source_id).await?;
    }

    // Chunk content if needed
    let chunks = chunk_text(&content);
    println!("  Chunks: {}", chunks.len());

    // Embed each chunk and create nodes
    let mut node_ids = Vec::new();
    for (i, chunk) in chunks.iter().enumerate() {
        let embed_start = Instant::now();
        let embedding = embed(chunk).await?;
        println!("  Embedded {}/{} in {:.1}s", i + 1, chunks.len(), embed_start.elapsed().as_secs_f32());

        let node = Node::new(
            chunk.clone(),
            source_id.clone(),
            SourceType::File,
            embedding,
            1.0, // Full confidence for raw file content
        );
        let node_id = node.id.to_string();
        store.insert_node(&node).await?;

        node_ids.push(node_id);
    }

    // Record as processed
    let record = Processed {
        source_id: source_id.clone(),
        source_type: "file".to_string(),
        simhash: current_simhash,
        processed_at: Utc::now(),
        node_count: node_ids.len() as i32,
        node_ids,
    };
    store.insert_processed(&record).await?;

    println!("  Done! Created {} node(s)", record.node_count);
    Ok(record.node_count as usize)
}

/// Run scan command - one-shot distillation
async fn run_scan(
    project: Option<PathBuf>,
    limit: Option<usize>,
) -> Result<(), Box<dyn std::error::Error>> {
    let project_path = project.unwrap_or_else(|| {
        std::env::current_dir().unwrap_or_else(|_| PathBuf::from("."))
    });

    println!("ds scan");
    println!("===========");
    println!("Project: {}", project_path.display());

    // Discover sessions
    println!("\nDiscovering sessions...");
    let sessions = match discover_sessions(&project_path) {
        Ok(s) => s,
        Err(e) => {
            eprintln!("Failed to discover sessions: {}", e);
            return Ok(());
        }
    };

    if sessions.is_empty() {
        println!("No sessions found for this project.");
        return Ok(());
    }

    // Apply limit
    let sessions: Vec<SessionInfo> = match limit {
        Some(n) => sessions.into_iter().take(n).collect(),
        None => sessions,
    };

    println!("Found {} session(s) to process", sessions.len());
    if let Some(n) = limit {
        println!("(limited to {} newest)", n);
    }

    // Open store
    let db_path = default_db_path();
    println!("\nDatabase: {}", db_path.display());
    let store = Store::open(db_path.to_str().unwrap()).await?;

    // Process each session
    let mut total_nodes = 0;
    let mut skipped = 0;
    let mut failed = 0;

    for (i, session) in sessions.iter().enumerate() {
        println!(
            "\n[{}/{}] Session: {} ({})",
            i + 1,
            sessions.len(),
            &session.session_id[..8],
            format_size(session.size_bytes)
        );

        match process_session(&store, session).await {
            Ok((nodes, was_skipped)) => {
                total_nodes += nodes;
                if was_skipped {
                    skipped += 1;
                }
            }
            Err(e) => {
                eprintln!("  Error: {}", e);
                failed += 1;
            }
        }
    }

    // Summary
    println!("\n-----------");
    println!("Scan complete!");
    println!("  Processed: {} sessions", sessions.len() - skipped - failed);
    println!("  Skipped:   {} (already processed)", skipped);
    println!("  Failed:    {}", failed);
    println!("  Nodes:     {} created", total_nodes);

    Ok(())
}

/// Delay between processing jobs (rate limiting)
const JOB_DELAY_MS: u64 = 500;

/// Start daemon in background
fn run_start_daemon() -> Result<(), Box<dyn std::error::Error>> {
    // Check if already running
    if let Some(pid) = is_daemon_running() {
        println!("Daemon already running (PID {})", pid);
        return Ok(());
    }

    // Get path to current executable
    let exe = std::env::current_exe()?;
    let log_path = daemon_log_path();

    // Ensure .datasphere directory exists
    if let Some(parent) = log_path.parent() {
        std::fs::create_dir_all(parent)?;
    }

    // Open log file (truncate on start)
    let log_file = std::fs::File::create(&log_path)?;

    // Spawn daemon process with stdout/stderr redirected to log file
    let child = std::process::Command::new(&exe)
        .arg("start")
        .arg("--foreground")
        .stdout(log_file.try_clone()?)
        .stderr(log_file)
        .stdin(std::process::Stdio::null())
        .spawn()?;

    // Write PID file
    let pid = child.id();
    std::fs::write(daemon_pid_path(), pid.to_string())?;

    println!("Daemon started (PID {})", pid);
    println!("Log: {}", log_path.display());

    Ok(())
}

/// Stop running daemon
fn run_stop() -> Result<(), Box<dyn std::error::Error>> {
    match is_daemon_running() {
        Some(pid) => {
            #[cfg(unix)]
            {
                // Send SIGTERM
                let result = unsafe { libc::kill(pid as i32, libc::SIGTERM) };
                if result == 0 {
                    println!("Stopping daemon (PID {})", pid);

                    // Wait briefly for graceful shutdown
                    std::thread::sleep(std::time::Duration::from_millis(500));

                    // Check if still running
                    if is_daemon_running().is_some() {
                        println!("Daemon still running, sending SIGKILL...");
                        unsafe { libc::kill(pid as i32, libc::SIGKILL) };
                    }

                    // Clean up PID file
                    let _ = std::fs::remove_file(daemon_pid_path());
                    println!("Daemon stopped");
                } else {
                    eprintln!("Failed to stop daemon: {}", std::io::Error::last_os_error());
                }
            }

            #[cfg(not(unix))]
            {
                eprintln!("Stop not supported on this platform");
            }
        }
        None => {
            println!("Daemon not running");
        }
    }
    Ok(())
}

/// Show daemon status
fn run_status() -> Result<(), Box<dyn std::error::Error>> {
    match is_daemon_running() {
        Some(pid) => {
            println!("Daemon running (PID {})", pid);
            println!("Log: {}", daemon_log_path().display());
        }
        None => {
            println!("Daemon not running");
        }
    }
    Ok(())
}

/// Run start command - daemon mode watching all projects (foreground)
async fn run_start_foreground() -> Result<(), Box<dyn std::error::Error>> {
    println!("ds start");
    println!("============");

    // Open store
    let db_path = default_db_path();
    println!("Database: {}", db_path.display());
    let store = Store::open(db_path.to_str().unwrap()).await?;

    // Open queue
    let queue = Queue::open_default().map_err(|e| format!("Failed to open queue: {}", e))?;

    // Resume any pending/processing jobs from previous run
    let (pending, processing, _, _) = queue.counts().unwrap_or((0, 0, 0, 0));
    if pending > 0 || processing > 0 {
        println!("Resuming {} pending, {} processing jobs from previous run", pending, processing);
    }

    // Scan existing sessions and queue unprocessed ones
    println!("\nScanning existing sessions...");
    let projects = list_all_projects().unwrap_or_default();
    let mut queued_initial = 0;

    for project in &projects {
        if let Ok(sessions) = discover_sessions_in_dir(&project.project_dir) {
            for session in sessions {
                // Check if already processed
                if store.get_processed(&session.session_id).await?.is_some() {
                    continue;
                }

                // Skip empty transcripts
                if !transcript_has_content(&session.transcript_path) {
                    continue;
                }

                let job = Job {
                    source_id: session.session_id.clone(),
                    source_type: "session".to_string(),
                    project_id: project.project_id.clone(),
                    transcript_path: session.transcript_path.to_string_lossy().to_string(),
                    queued_at: Utc::now(),
                    status: JobStatus::Pending,
                    error: None,
                };

                if queue.add(job).is_ok() {
                    queued_initial += 1;
                }
            }
        }
    }

    if queued_initial > 0 {
        println!("Queued {} unprocessed session(s)", queued_initial);
    } else {
        println!("All sessions already processed");
    }

    // Create all-projects watcher
    println!("\nStarting all-projects watcher...");
    let watcher = match AllProjectsWatcher::new() {
        Ok(w) => w,
        Err(e) => {
            eprintln!("Failed to create watcher: {}", e);
            return Ok(());
        }
    };
    println!("Watching: {}", watcher.projects_dir().display());

    // Stats
    let mut queued_count = 0;
    let mut processed_count = 0;
    let mut total_nodes = 0;

    println!("\nDaemon running (Ctrl+C to stop)...\n");

    // Set up signal handling for graceful shutdown
    #[cfg(unix)]
    let mut sigterm = tokio::signal::unix::signal(tokio::signal::unix::SignalKind::terminate())?;
    let mut shutdown = false;

    // Main loop: poll for watcher events, process queue
    while !shutdown {
        // Check for shutdown signals
        #[cfg(unix)]
        {
            use tokio::time::timeout;
            // Non-blocking check for SIGTERM
            if let Ok(Some(())) = timeout(Duration::from_millis(0), sigterm.recv()).await {
                println!("\n[SHUTDOWN] Received SIGTERM, shutting down...");
                shutdown = true;
                continue;
            }
        }
        // Check for Ctrl+C (cross-platform)
        {
            use tokio::time::timeout;
            if let Ok(Ok(())) = timeout(Duration::from_millis(0), tokio::signal::ctrl_c()).await {
                println!("\n[SHUTDOWN] Received Ctrl+C, shutting down...");
                shutdown = true;
                continue;
            }
        }
        // Drain all pending watcher events into queue
        while let Some(event) = watcher.try_recv() {
            // Skip empty transcripts
            if !transcript_has_content(&event.session.transcript_path) {
                continue;
            }

            let job = Job {
                source_id: event.session.session_id.clone(),
                source_type: "session".to_string(),
                project_id: event.project_id.clone(),
                transcript_path: event.session.transcript_path.to_string_lossy().to_string(),
                queued_at: Utc::now(),
                status: JobStatus::Pending,
                error: None,
            };

            match queue.add(job) {
                Ok(true) => {
                    queued_count += 1;
                    println!(
                        "[QUEUE] {} ({}) from {}",
                        &event.session.session_id[..8.min(event.session.session_id.len())],
                        if event.is_new { "new" } else { "modified" },
                        &event.project_id
                    );
                }
                Ok(false) => {} // Duplicate, already queued
                Err(e) => eprintln!("Failed to queue job: {}", e),
            }
        }

        // Process one job from queue
        if let Ok(Some(job)) = queue.pop_pending() {
            println!(
                "[{}] [PROCESS] {} from {}",
                Local::now().format("%H:%M:%S"),
                &job.source_id[..8.min(job.source_id.len())],
                &job.project_id
            );

            // Build SessionInfo from job
            let session = SessionInfo {
                session_id: job.source_id.clone(),
                transcript_path: PathBuf::from(&job.transcript_path),
                modified_at: job.queued_at, // Use queued time as proxy
                size_bytes: std::fs::metadata(&job.transcript_path)
                    .map(|m| m.len())
                    .unwrap_or(0),
            };

            match process_session(&store, &session).await {
                Ok((nodes, was_skipped)) => {
                    if let Err(e) = queue.mark_done(&job.source_id) {
                        eprintln!("  Failed to mark done: {}", e);
                    }
                    if !was_skipped {
                        processed_count += 1;
                        total_nodes += nodes;
                    }
                }
                Err(e) => {
                    eprintln!("  Error: {}", e);
                    if let Err(e2) = queue.mark_failed(&job.source_id, &e.to_string()) {
                        eprintln!("  Failed to mark failed: {}", e2);
                    }
                }
            }

            // Rate limit
            tokio::time::sleep(Duration::from_millis(JOB_DELAY_MS)).await;
        } else {
            // No pending jobs, sleep briefly before checking again
            tokio::time::sleep(Duration::from_millis(100)).await;
        }
    }

    // Clean up PID file on graceful shutdown
    let _ = std::fs::remove_file(daemon_pid_path());

    println!("\nDaemon stopped.");
    println!("  Queued:    {} sessions", queued_count);
    println!("  Processed: {} sessions", processed_count);
    println!("  Nodes:     {} created", total_nodes);
    Ok(())
}

/// Run queue command - show/manage job queue
fn run_queue(action: Option<QueueAction>) -> Result<(), Box<dyn std::error::Error>> {
    let queue = Queue::open_default().map_err(|e| format!("Failed to open queue: {}", e))?;

    match action.unwrap_or(QueueAction::Status) {
        QueueAction::Status => {
            let (pending, processing, done, failed) = queue.counts()?;
            println!("ds queue");
            println!("============");
            println!("Pending:    {}", pending);
            println!("Processing: {}", processing);
            println!("Done:       {}", done);
            println!("Failed:     {}", failed);
        }

        QueueAction::Pending => {
            let jobs = queue.list_pending()?;
            if jobs.is_empty() {
                println!("No pending jobs.");
            } else {
                println!("Pending jobs ({}):", jobs.len());
                for job in jobs {
                    println!(
                        "  {} ({})",
                        &job.source_id[..8.min(job.source_id.len())],
                        &job.project_id
                    );
                }
            }
        }

        QueueAction::Clear => {
            let cleared = queue.clear_done()?;
            println!("Cleared {} completed jobs.", cleared);
        }

        QueueAction::Nuke => {
            let nuked = queue.nuke()?;
            println!("Nuked {} jobs.", nuked);
        }
    }

    Ok(())
}

/// Run query command - search knowledge graph
async fn run_query(
    query: &str,
    limit: usize,
    format: &str,
) -> Result<(), Box<dyn std::error::Error>> {
    let db_path = default_db_path();

    if !db_path.exists() {
        eprintln!("Database not found at: {}", db_path.display());
        return Ok(());
    }

    let store = Store::open(db_path.to_str().unwrap()).await?;

    // Embed the query
    let embedding = embed(query).await?;

    // Search for similar nodes
    let results = store.search_similar_with_scores(&embedding, limit).await?;

    if results.is_empty() {
        if format == "json" {
            println!("[]");
        } else {
            println!("No relevant results found.");
        }
        return Ok(());
    }

    if format == "json" {
        // JSON output for MCP consumption
        let json_results: Vec<serde_json::Value> = results
            .iter()
            .map(|(node, score)| {
                serde_json::json!({
                    "id": node.id.to_string(),
                    "content": node.content,
                    "source": node.source,
                    "similarity": score,
                    "timestamp": node.timestamp.to_rfc3339(),
                })
            })
            .collect();
        println!("{}", serde_json::to_string_pretty(&json_results)?);
    } else {
        // Human-readable text output
        for (i, (node, score)) in results.iter().enumerate() {
            println!("─── Result {} (similarity: {:.2}) ───", i + 1, score);
            println!("Source: {}", node.source);
            println!("Time:   {}", node.timestamp.format("%Y-%m-%d %H:%M"));
            println!("{}", node.content);
            println!();
        }
    }

    Ok(())
}

/// Run related command - find nodes similar to a given node
async fn run_related(
    node_id: &str,
    limit: usize,
    format: &str,
) -> Result<(), Box<dyn std::error::Error>> {
    let db_path = default_db_path();

    if !db_path.exists() {
        eprintln!("Database not found at: {}", db_path.display());
        return Ok(());
    }

    // Parse node ID as UUID
    let uuid = node_id.parse::<Uuid>().map_err(|_| {
        format!("Invalid node ID: {}. Expected a UUID.", node_id)
    })?;

    let store = Store::open(db_path.to_str().unwrap()).await?;

    // Get the source node
    let node = store.get_node(uuid).await?.ok_or_else(|| {
        format!("Node not found: {}", node_id)
    })?;

    // Search for similar nodes (request one extra to filter out self)
    let results = store.search_similar_with_scores(&node.embedding, limit + 1).await?;

    // Filter out the source node itself
    let results: Vec<_> = results
        .into_iter()
        .filter(|(n, _)| n.id != uuid)
        .take(limit)
        .collect();

    if results.is_empty() {
        if format == "json" {
            println!("[]");
        } else {
            println!("No related nodes found.");
        }
        return Ok(());
    }

    if format == "json" {
        let json_results: Vec<serde_json::Value> = results
            .iter()
            .map(|(n, score)| {
                serde_json::json!({
                    "id": n.id.to_string(),
                    "content": n.content,
                    "source": n.source,
                    "similarity": score,
                    "timestamp": n.timestamp.to_rfc3339(),
                })
            })
            .collect();
        println!("{}", serde_json::to_string_pretty(&json_results)?);
    } else {
        println!("Nodes related to {}:", &node_id[..8.min(node_id.len())]);
        println!();
        for (i, (n, score)) in results.iter().enumerate() {
            println!("─── {} (similarity: {:.2}) ───", i + 1, score);
            println!("ID:     {}", n.id);
            println!("Source: {}", n.source);
            println!("Time:   {}", n.timestamp.format("%Y-%m-%d %H:%M"));
            println!("{}", n.content);
            println!();
        }
    }

    Ok(())
}

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let cli = Cli::parse();

    match cli.command {
        Commands::Stats => {
            let db_path = default_db_path();

            if !db_path.exists() {
                println!("Database not found at: {}", db_path.display());
                println!("Run 'ds scan' to create the database.");
                return Ok(());
            }

            let store = Store::open(db_path.to_str().unwrap()).await?;

            let nodes = store.count_nodes().await?;
            let processed = store.count_processed().await?;
            let db_size = dir_size(&db_path);

            println!("ds stats");
            println!("============");
            println!("Database:   {}", db_path.display());
            println!("Size:       {}", format_size(db_size));
            println!();
            println!("Nodes:      {}", nodes);
            println!("Processed:  {} transcripts", processed);
        }

        Commands::Show { limit } => {
            let db_path = default_db_path();

            if !db_path.exists() {
                println!("Database not found at: {}", db_path.display());
                return Ok(());
            }

            let store = Store::open(db_path.to_str().unwrap()).await?;
            let nodes = store.list_nodes(limit).await?;

            if nodes.is_empty() {
                println!("No nodes stored yet.");
                return Ok(());
            }

            for (i, node) in nodes.iter().enumerate() {
                println!("─── Node {} ───", i + 1);
                println!("ID:      {}", node.id);
                println!("Source:  {}", node.source);
                println!("Time:    {}", node.timestamp.format("%Y-%m-%d %H:%M"));
                println!("Content:\n{}", node.content);
                println!();
            }
        }

        Commands::Scan { limit, project } => {
            run_scan(project, limit).await?;
        }

        Commands::Start { foreground } => {
            if foreground {
                run_start_foreground().await?;
            } else {
                run_start_daemon()?;
            }
        }

        Commands::Stop => {
            run_stop()?;
        }

        Commands::Status => {
            run_status()?;
        }

        Commands::Queue { action } => {
            run_queue(action)?;
        }

        Commands::Add { file } => {
            if !file.exists() {
                eprintln!("File not found: {}", file.display());
                return Ok(());
            }

            let db_path = default_db_path();
            let store = Store::open(db_path.to_str().unwrap()).await?;

            println!("ds add");
            println!("==========");

            match process_file(&store, &file).await {
                Ok(nodes) => {
                    if nodes > 0 {
                        println!("\nCreated {} node(s)", nodes);
                    }
                }
                Err(e) => {
                    eprintln!("Error: {}", e);
                }
            }
        }

        Commands::Query { query, limit, format } => {
            run_query(&query, limit, &format).await?;
        }

        Commands::Related { node_id, limit, format } => {
            run_related(&node_id, limit, &format).await?;
        }

        Commands::Reset => {
            println!("ds reset");
            println!("============");

            // Delete database
            let db_path = default_db_path();
            if db_path.exists() {
                std::fs::remove_dir_all(&db_path)
                    .map_err(|e| format!("Failed to delete database: {}", e))?;
                println!("Deleted database: {}", db_path.display());
            } else {
                println!("Database not found (already clean)");
            }

            // Nuke queue
            let queue = Queue::open_default()?;
            let nuked = queue.nuke()?;
            if nuked > 0 {
                println!("Nuked {} queued jobs", nuked);
            } else {
                println!("Queue was empty");
            }

            println!("\nReset complete. Run 'ds start' to begin fresh.");
        }
    }

    Ok(())
}