Skip to main content

aven_core/sync/
persistence.rs

1use std::collections::HashSet;
2use std::path::Path;
3use std::time::Instant;
4
5use anyhow::{Context, Result, bail};
6use sqlx::{QueryBuilder, Sqlite, SqliteConnection};
7
8use super::apply::apply_remote_change;
9use super::wire::{ChangeRow, ChangeWire, PushAck, SyncRequest, SyncResponse};
10use crate::change_log::op_type;
11use crate::db::{Database, begin_immediate, get_meta, set_meta};
12
13#[derive(Debug, Clone, Default, PartialEq, Eq)]
14pub struct SyncPersistenceStatus {
15    pub pinned_server: Option<String>,
16    pub pending_changes: i64,
17    pub conflicts: i64,
18    pub sync_cursor: Option<String>,
19    pub local_sequence: Option<String>,
20    pub last_attempt: Option<String>,
21    pub last_success: Option<String>,
22    pub last_error: Option<String>,
23    pub last_pushed: Option<String>,
24    pub last_pulled: Option<String>,
25    pub last_cursor: Option<String>,
26}
27
28#[derive(Debug, Clone)]
29pub struct ClientSyncPage {
30    pub request: SyncRequest,
31    pub pending: usize,
32}
33
34#[derive(Debug)]
35pub struct ApplySyncPage {
36    pub request: SyncRequest,
37    pub response: SyncResponse,
38    pub attempted_at: String,
39    pub previous_pushed: i64,
40    pub previous_pulled: usize,
41}
42
43#[derive(Debug)]
44pub struct ServerSyncPage {
45    pub request: SyncRequest,
46}
47
48#[derive(Debug)]
49pub struct ServerSyncResult {
50    pub accepted_count: i64,
51    pub push_acks: Vec<PushAck>,
52    pub changes: Vec<ChangeWire>,
53    pub has_more: bool,
54    pub assign_ms: u128,
55    pub pull_query_ms: u128,
56}
57
58impl Database {
59    pub async fn sync_persistence_status(&self) -> Result<SyncPersistenceStatus> {
60        let mut conn = self.acquire().await?;
61        let pending_changes =
62            sqlx::query_scalar("SELECT count(*) FROM changes WHERE server_seq IS NULL")
63                .fetch_one(&mut *conn)
64                .await?;
65        let conflicts = sqlx::query_scalar("SELECT count(*) FROM conflicts WHERE resolved = 0")
66            .fetch_one(&mut *conn)
67            .await?;
68        Ok(SyncPersistenceStatus {
69            pinned_server: get_meta(&mut conn, "sync_server_url").await?,
70            pending_changes,
71            conflicts,
72            sync_cursor: get_meta(&mut conn, "sync_cursor").await?,
73            local_sequence: get_meta(&mut conn, "local_seq").await?,
74            last_attempt: get_meta(&mut conn, "sync_last_attempt_at").await?,
75            last_success: get_meta(&mut conn, "sync_last_success_at").await?,
76            last_error: get_meta(&mut conn, "sync_last_error").await?,
77            last_pushed: get_meta(&mut conn, "sync_last_pushed").await?,
78            last_pulled: get_meta(&mut conn, "sync_last_pulled").await?,
79            last_cursor: get_meta(&mut conn, "sync_last_cursor").await?,
80        })
81    }
82
83    pub async fn begin_sync_attempt(&self, attempted_at: String) -> Result<()> {
84        let mut conn = self.acquire().await?;
85        set_meta(&mut conn, "sync_last_attempt_at", &attempted_at).await
86    }
87
88    pub async fn record_sync_error(&self, error: String) -> Result<()> {
89        let mut conn = self.acquire().await?;
90        set_meta(&mut conn, "sync_last_error", &error).await
91    }
92
93    pub(super) async fn pending_sync_change_count(&self) -> Result<i64> {
94        let mut conn = self.acquire().await?;
95        Ok(
96            sqlx::query_scalar("SELECT COUNT(*) FROM changes WHERE server_seq IS NULL")
97                .fetch_one(&mut *conn)
98                .await?,
99        )
100    }
101
102    pub(super) async fn pending_blob_contracts(
103        &self,
104    ) -> Result<Vec<super::wire::BlobUploadContract>> {
105        let mut conn = self.acquire().await?;
106        let changes = load_unsynced_changes(&mut conn, i64::MAX as usize).await?;
107        super::blob::unique_blob_contracts(&changes)
108    }
109
110    pub async fn prepare_client_sync_page(
111        &self,
112        server: String,
113        push_limit: usize,
114        pull_limit: u32,
115    ) -> Result<ClientSyncPage> {
116        let mut conn = self.acquire().await?;
117        validate_sync_server(&mut conn, &server).await?;
118        let client_id = get_meta(&mut conn, "client_id")
119            .await?
120            .context("missing client id")?;
121        let after = sync_cursor(&mut conn).await?;
122        let changes = load_unsynced_changes(&mut conn, push_limit).await?;
123        let pending = changes.len();
124        Ok(ClientSyncPage {
125            request: SyncRequest {
126                protocol_version: Some(super::wire::SYNC_PROTOCOL_VERSION),
127                client_id,
128                after,
129                pull_limit: Some(pull_limit),
130                changes,
131            },
132            pending,
133        })
134    }
135
136    pub async fn apply_client_sync_page(&self, page: ApplySyncPage) -> Result<usize> {
137        let envelope = super::wire::validate_sync_request_envelope(&page.request)?;
138        let request_change_ids = page
139            .request
140            .changes
141            .iter()
142            .map(|change| change.change_id.clone())
143            .collect::<Vec<_>>();
144        super::wire::validate_sync_response_for_request(
145            envelope.after,
146            envelope.pull_limit,
147            &request_change_ids,
148            &page.response,
149        )?;
150        let mut conn = self.acquire().await?;
151        apply_sync_response(&mut conn, page).await
152    }
153
154    pub async fn persist_server_sync_page(&self, page: ServerSyncPage) -> Result<ServerSyncResult> {
155        self.persist_server_sync_page_inner(page, None).await
156    }
157
158    pub async fn persist_server_sync_page_with_blobs(
159        &self,
160        page: ServerSyncPage,
161        blob_dir: &Path,
162    ) -> Result<ServerSyncResult> {
163        self.persist_server_sync_page_inner(page, Some(blob_dir))
164            .await
165    }
166
167    async fn persist_server_sync_page_inner(
168        &self,
169        page: ServerSyncPage,
170        blob_dir: Option<&Path>,
171    ) -> Result<ServerSyncResult> {
172        let envelope = super::wire::validate_sync_request_envelope(&page.request)?;
173        for change in &page.request.changes {
174            super::wire::validate_pushed_change(change)?;
175        }
176        let mut conn = self.acquire().await?;
177        let assign_started = Instant::now();
178        let (accepted_count, push_acks) =
179            assign_server_sequences(&mut conn, page.request.changes, blob_dir).await?;
180        let assign_ms = assign_started.elapsed().as_millis();
181        let pull_query_started = Instant::now();
182        let (changes, has_more) =
183            load_server_changes_after(&mut conn, envelope.after, envelope.pull_limit).await?;
184        let pull_query_ms = pull_query_started.elapsed().as_millis();
185        Ok(ServerSyncResult {
186            accepted_count,
187            push_acks,
188            changes,
189            has_more,
190            assign_ms,
191            pull_query_ms,
192        })
193    }
194}
195
196async fn sync_cursor(conn: &mut SqliteConnection) -> Result<i64> {
197    Ok(get_meta(conn, "sync_cursor")
198        .await?
199        .unwrap_or_else(|| "0".to_string())
200        .parse::<i64>()?)
201}
202
203async fn validate_sync_server(conn: &mut SqliteConnection, server: &str) -> Result<()> {
204    let normalized = server.trim_end_matches('/');
205    if let Some(existing) = get_meta(conn, "sync_server_url").await? {
206        if existing != normalized {
207            bail!(
208                "error sync-server-changed existing={} requested={} hint=\"use a fresh database for a different sync server\"",
209                existing,
210                normalized
211            );
212        }
213    } else {
214        set_meta(conn, "sync_server_url", normalized).await?;
215    }
216    Ok(())
217}
218
219async fn load_unsynced_changes(
220    conn: &mut SqliteConnection,
221    limit: usize,
222) -> Result<Vec<ChangeWire>> {
223    let limit = limit as i64;
224    let rows = sqlx::query_as!(
225        ChangeRow,
226        r#"SELECT change_id AS "change_id!: String", client_id AS "client_id!: String",
227         local_seq AS "local_seq!: i64", entity_type AS "entity_type!: String",
228         entity_id AS "entity_id!: String", field, op_type AS "op_type!: String",
229         payload AS "payload!: String", base_version, created_at AS "created_at!: String",
230         server_seq
231         FROM changes WHERE server_seq IS NULL ORDER BY local_seq, created_at LIMIT ?"#,
232        limit,
233    )
234    .fetch_all(&mut *conn)
235    .await?;
236    Ok(rows.into_iter().map(ChangeRow::into_wire).collect())
237}
238
239async fn apply_sync_response(conn: &mut SqliteConnection, page: ApplySyncPage) -> Result<usize> {
240    let mut applied = 0;
241    let mut tx = begin_immediate(conn).await?;
242    let current_cursor = sync_cursor(&mut tx).await?;
243    if current_cursor != page.request.after {
244        bail!(
245            "error stale-sync-page expected_cursor={} request_cursor={}",
246            current_cursor,
247            page.request.after
248        );
249    }
250    update_change_server_seqs_if_missing(&mut tx, &page.response.push_acks).await?;
251    let existing_change_ids = load_existing_change_ids(&mut tx, &page.response.changes).await?;
252    for change in &page.response.changes {
253        if existing_change_ids.contains(change.change_id.as_str()) {
254            update_change_server_seq(&mut tx, &change.change_id, change.server_seq).await?;
255            continue;
256        }
257        apply_remote_change(&mut tx, change).await?;
258        insert_wire_change(&mut tx, change).await?;
259        applied += 1;
260    }
261    let pushed = page.previous_pushed + page.response.push_acks.len() as i64;
262    let pulled = page.previous_pulled + applied;
263    set_meta(&mut tx, "sync_cursor", &page.response.cursor.to_string()).await?;
264    set_meta(&mut tx, "sync_last_success_at", &page.attempted_at).await?;
265    set_meta(&mut tx, "sync_last_error", "").await?;
266    set_meta(&mut tx, "sync_last_pushed", &pushed.to_string()).await?;
267    set_meta(&mut tx, "sync_last_pulled", &pulled.to_string()).await?;
268    set_meta(
269        &mut tx,
270        "sync_last_cursor",
271        &page.response.cursor.to_string(),
272    )
273    .await?;
274    tx.commit().await?;
275    Ok(applied)
276}
277
278async fn load_existing_change_ids(
279    conn: &mut SqliteConnection,
280    changes: &[ChangeWire],
281) -> Result<HashSet<String>> {
282    if changes.is_empty() {
283        return Ok(HashSet::new());
284    }
285    let mut query_builder =
286        QueryBuilder::<Sqlite>::new("SELECT change_id FROM changes WHERE change_id IN (");
287    let mut separated = query_builder.separated(", ");
288    for change in changes {
289        separated.push_bind(&change.change_id);
290    }
291    separated.push_unseparated(")");
292    Ok(query_builder
293        .build_query_scalar::<String>()
294        .fetch_all(&mut *conn)
295        .await?
296        .into_iter()
297        .collect())
298}
299
300async fn update_change_server_seq(
301    conn: &mut SqliteConnection,
302    change_id: &str,
303    server_seq: Option<i64>,
304) -> Result<()> {
305    if let Some(server_seq) = server_seq {
306        sqlx::query!(
307            "UPDATE changes SET server_seq = ? WHERE change_id = ? AND server_seq IS NULL",
308            server_seq,
309            change_id,
310        )
311        .execute(&mut *conn)
312        .await?;
313    }
314    Ok(())
315}
316
317async fn update_change_server_seqs_if_missing(
318    conn: &mut SqliteConnection,
319    push_acks: &[PushAck],
320) -> Result<()> {
321    if push_acks.is_empty() {
322        return Ok(());
323    }
324    let mut query_builder = QueryBuilder::<Sqlite>::new("WITH updates(change_id, server_seq) AS (");
325    query_builder.push_values(push_acks, |mut row, ack| {
326        row.push_bind(&ack.change_id).push_bind(ack.server_seq);
327    });
328    query_builder.push(
329        ") UPDATE changes
330         SET server_seq = (
331             SELECT updates.server_seq
332             FROM updates
333             WHERE updates.change_id = changes.change_id
334         )
335         WHERE server_seq IS NULL
336           AND change_id IN (SELECT change_id FROM updates)",
337    );
338    query_builder.build().execute(&mut *conn).await?;
339    Ok(())
340}
341
342async fn insert_wire_change(conn: &mut SqliteConnection, change: &ChangeWire) -> Result<()> {
343    let payload = change.payload.to_string();
344    sqlx::query!(
345        "INSERT OR IGNORE INTO changes(change_id, client_id, local_seq, entity_type, entity_id, field,
346         op_type, payload, base_version, created_at, server_seq)
347         VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)",
348        change.change_id,
349        change.client_id,
350        change.local_seq,
351        change.entity_type,
352        change.entity_id,
353        change.field,
354        change.op_type,
355        payload,
356        change.base_version,
357        change.created_at,
358        change.server_seq,
359    )
360    .execute(&mut *conn)
361    .await?;
362    Ok(())
363}
364
365async fn assign_server_sequences(
366    conn: &mut SqliteConnection,
367    changes: Vec<ChangeWire>,
368    blob_dir: Option<&Path>,
369) -> Result<(i64, Vec<PushAck>)> {
370    if changes.is_empty() {
371        return Ok((0, Vec::new()));
372    }
373    let mut tx = begin_immediate(conn).await?;
374    let existing_change_ids = load_existing_change_ids(&mut tx, &changes).await?;
375    let unassigned_changes = changes
376        .iter()
377        .filter(|change| !existing_change_ids.contains(&change.change_id))
378        .cloned()
379        .collect::<Vec<_>>();
380    if unassigned_changes
381        .iter()
382        .any(|change| change.op_type == op_type::ATTACHMENT_ADD)
383        && blob_dir.is_none()
384    {
385        bail!("error attachment-blob-storage-required");
386    }
387    if let Some(blob_dir) = blob_dir {
388        ensure_attachment_blobs_present(&mut tx, blob_dir, &unassigned_changes).await?;
389    }
390    let mut next_server_seq = next_available_server_seq(&mut tx).await?;
391    let mut accepted_count = 0_i64;
392    let mut push_acks = Vec::with_capacity(changes.len());
393    for change in changes {
394        let existing_server_seq = sqlx::query_scalar::<_, Option<i64>>(
395            "SELECT server_seq FROM changes WHERE change_id = ?",
396        )
397        .bind(&change.change_id)
398        .fetch_optional(&mut *tx)
399        .await?
400        .flatten();
401        let server_seq = if let Some(server_seq) = existing_server_seq {
402            server_seq
403        } else {
404            let server_seq = next_server_seq;
405            next_server_seq += 1;
406            let payload = change.payload.to_string();
407            sqlx::query(
408                "INSERT INTO changes(change_id, client_id, local_seq, entity_type, entity_id, field,
409                 op_type, payload, base_version, created_at, server_seq)
410                 VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)",
411            )
412            .bind(&change.change_id)
413            .bind(&change.client_id)
414            .bind(change.local_seq)
415            .bind(&change.entity_type)
416            .bind(&change.entity_id)
417            .bind(&change.field)
418            .bind(&change.op_type)
419            .bind(payload)
420            .bind(&change.base_version)
421            .bind(&change.created_at)
422            .bind(server_seq)
423            .execute(&mut *tx)
424            .await?;
425            apply_server_blob_reference(&mut tx, &change).await?;
426            accepted_count += 1;
427            server_seq
428        };
429        push_acks.push(PushAck {
430            change_id: change.change_id,
431            server_seq,
432        });
433    }
434    tx.commit().await?;
435    Ok((accepted_count, push_acks))
436}
437
438async fn next_available_server_seq(conn: &mut SqliteConnection) -> Result<i64> {
439    Ok(sqlx::query_scalar!(
440        r#"SELECT COALESCE(MAX(server_seq), 0) + 1 AS "seq!: i64" FROM changes"#
441    )
442    .fetch_one(&mut *conn)
443    .await?)
444}
445
446async fn apply_server_blob_reference(
447    conn: &mut SqliteConnection,
448    change: &ChangeWire,
449) -> Result<()> {
450    match change.op_type.as_str() {
451        op_type::ATTACHMENT_ADD => {
452            let workspace_id = change.payload["workspace_id"]
453                .as_str()
454                .context("payload missing workspace_id")?;
455            let attachment_id = change.payload["attachment_id"]
456                .as_str()
457                .context("payload missing attachment_id")?;
458            let sha256 = change.payload["sha256"]
459                .as_str()
460                .context("payload missing sha256")?;
461            let byte_size = change.payload["byte_size"]
462                .as_i64()
463                .context("payload missing byte_size")?;
464            sqlx::query(
465                "INSERT INTO server_blob_references(
466                   workspace_id, attachment_id, task_id, sha256, byte_size, deleted
467                 ) VALUES (?, ?, ?, ?, ?, 0)
468                 ON CONFLICT(workspace_id, attachment_id) DO UPDATE SET
469                   task_id = excluded.task_id, sha256 = excluded.sha256,
470                   byte_size = excluded.byte_size, deleted = 0",
471            )
472            .bind(workspace_id)
473            .bind(attachment_id)
474            .bind(&change.entity_id)
475            .bind(sha256)
476            .bind(byte_size)
477            .execute(&mut *conn)
478            .await?;
479            sqlx::query(
480                "DELETE FROM blob_upload_reservations WHERE workspace_id = ? AND sha256 = ?",
481            )
482            .bind(workspace_id)
483            .bind(sha256)
484            .execute(&mut *conn)
485            .await?;
486        }
487        op_type::ATTACHMENT_DELETE => {
488            sqlx::query(
489                "UPDATE server_blob_references SET deleted = 1
490                 WHERE workspace_id = ? AND attachment_id = ?",
491            )
492            .bind(
493                change.payload["workspace_id"]
494                    .as_str()
495                    .context("payload missing workspace_id")?,
496            )
497            .bind(
498                change.payload["attachment_id"]
499                    .as_str()
500                    .context("payload missing attachment_id")?,
501            )
502            .execute(&mut *conn)
503            .await?;
504        }
505        op_type::SET_FIELD if change.field.as_deref() == Some("deleted") => {
506            let deleted = change.payload["value"]
507                .as_str()
508                .is_some_and(|value| value == "1");
509            sqlx::query(
510                "INSERT INTO server_task_tombstones(workspace_id, task_id, deleted)
511                 VALUES (?, ?, ?)
512                 ON CONFLICT(workspace_id, task_id) DO UPDATE SET deleted = excluded.deleted",
513            )
514            .bind(
515                change.payload["workspace_id"]
516                    .as_str()
517                    .context("payload missing workspace_id")?,
518            )
519            .bind(&change.entity_id)
520            .bind(i64::from(deleted))
521            .execute(&mut *conn)
522            .await?;
523        }
524        _ => {}
525    }
526    Ok(())
527}
528
529async fn ensure_attachment_blobs_present(
530    conn: &mut SqliteConnection,
531    blob_dir: &Path,
532    changes: &[ChangeWire],
533) -> Result<()> {
534    for change in changes {
535        if change.op_type != op_type::ATTACHMENT_ADD {
536            continue;
537        }
538        let contract = super::blob::attachment_blob_contract(change)?
539            .context("error attachment-blob-missing")?;
540        let Some(row) =
541            crate::attachments::storage::blob_inventory_row(conn, &contract.sha256).await?
542        else {
543            bail!("error attachment-blob-missing");
544        };
545        if !row.available || !crate::attachments::object_path(blob_dir, &contract.sha256)?.exists()
546        {
547            bail!("error attachment-blob-missing");
548        }
549        if row.byte_size != contract.byte_size || row.media_type != contract.media_type {
550            bail!("error blob-inventory-metadata-mismatch");
551        }
552        let bytes =
553            tokio::fs::read(crate::attachments::object_path(blob_dir, &contract.sha256)?).await?;
554        if i64::try_from(bytes.len()).ok() != Some(contract.byte_size)
555            || crate::attachments::storage::sha256_hex(&bytes) != contract.sha256
556        {
557            bail!("error attachment-blob-content-mismatch");
558        }
559        let validated =
560            crate::attachments::decode::validate_image(bytes, Some(contract.media_type.clone()))
561                .await?;
562        if (validated.facts.width, validated.facts.height) != (contract.width, contract.height) {
563            bail!("error blob-inventory-metadata-mismatch");
564        }
565        let admitted: bool = sqlx::query_scalar(
566            "SELECT EXISTS(
567               SELECT 1 FROM server_blob_references sbr
568               LEFT JOIN server_task_tombstones st
569                 ON st.workspace_id = sbr.workspace_id AND st.task_id = sbr.task_id
570               WHERE sbr.workspace_id = ? AND sbr.sha256 = ? AND sbr.deleted = 0
571                 AND COALESCE(st.deleted, 0) = 0
572             ) OR EXISTS(
573               SELECT 1 FROM blob_upload_reservations
574               WHERE workspace_id = ? AND sha256 = ? AND byte_size = ? AND expires_at > ?
575             )",
576        )
577        .bind(&contract.workspace_id)
578        .bind(&contract.sha256)
579        .bind(&contract.workspace_id)
580        .bind(&contract.sha256)
581        .bind(contract.byte_size)
582        .bind(crate::ids::now())
583        .fetch_one(&mut *conn)
584        .await?;
585        if !admitted {
586            bail!("error attachment-blob-unreserved");
587        }
588    }
589    Ok(())
590}
591
592async fn load_server_changes_after(
593    conn: &mut SqliteConnection,
594    after: i64,
595    pull_limit: u32,
596) -> Result<(Vec<ChangeWire>, bool)> {
597    let fetch_limit = i64::from(pull_limit) + 1;
598    let rows = sqlx::query_as!(
599        ChangeRow,
600        r#"SELECT change_id AS "change_id!: String", client_id AS "client_id!: String",
601         local_seq AS "local_seq!: i64", entity_type AS "entity_type!: String",
602         entity_id AS "entity_id!: String", field, op_type AS "op_type!: String",
603         payload AS "payload!: String", base_version, created_at AS "created_at!: String",
604         server_seq
605         FROM changes WHERE server_seq > ? ORDER BY server_seq LIMIT ?"#,
606        after,
607        fetch_limit,
608    )
609    .fetch_all(&mut *conn)
610    .await?;
611    let has_more = rows.len() > pull_limit as usize;
612    let changes = rows
613        .into_iter()
614        .take(pull_limit as usize)
615        .map(ChangeRow::into_wire)
616        .collect();
617    Ok((changes, has_more))
618}