@powersync/service-core-tests 0.17.2 → 0.18.1

package/dist/tests/register-sync-tests.js

@@ -3,9 +3,10 @@ import { JSONBig } from '@powersync/service-jsonbig';
 import path from 'path';
 import * as timers from 'timers/promises';
 import { fileURLToPath } from 'url';
-import { expect, test } from 'vitest';
+import { expect, test, vi } from 'vitest';
 import * as test_utils from '../test-utils/test-utils-index.js';
 import { bucketRequest, METRICS_HELPER } from '../test-utils/test-utils-index.js';
+import { compactActive } from './util.js';
 const __filename = fileURLToPath(import.meta.url);
 const __dirname = path.dirname(__filename);
 const BASIC_SYNC_RULES = `
@@ -205,26 +206,33 @@ streams:
         const lines = await consumeCheckpointLines(stream);
         expect(lines).toMatchSnapshot();
     });
-    test('sync interrupts low-priority buckets on new checkpoints', async () => {
+    test('carries pending low-priority buckets into a new checkpoint', async () => {
         await using f = await factory();
         const syncRules = await updateSyncRules(f, {
             content: `
 bucket_definitions:
-  b0:
+  b0a:
     priority: 2
     data:
-      - SELECT * FROM test WHERE LENGTH(id) <= 5;
+      - SELECT * FROM test WHERE substring(id, 1, 6) = 'first-';
+  b0b:
+    priority: 3
+    data:
+      - SELECT * FROM test WHERE substring(id, 1, 4) = 'low-';
   b1:
     priority: 1
     data:
-      - SELECT * FROM test WHERE LENGTH(id) > 5;
+      - SELECT * FROM test WHERE substring(id, 1, 8) = 'highprio';
     `
         });
         const bucketStorage = f.getInstance(syncRules);
         await using writer = await bucketStorage.createWriter(test_utils.BATCH_OPTIONS);
         const testTable = await test_utils.resolveTestTable(writer, 'test', ['id'], config);
+        const syncRulesContent = syncRules.syncConfigContent[0];
+        const b0aBucket = test_utils.bucketRequest(syncRulesContent, 'b0a[]').bucket;
+        const b0bBucket = test_utils.bucketRequest(syncRulesContent, 'b0b[]').bucket;
+        const b1Bucket = test_utils.bucketRequest(syncRulesContent, 'b1[]').bucket;
         await writer.markAllSnapshotDone('0/1');
-        // Initial data: Add one priority row and 10k low-priority rows.
         await writer.save({
             sourceTable: testTable,
             tag: storage.SaveOperationTag.INSERT,
@@ -234,15 +242,26 @@ bucket_definitions:
             },
             afterReplicaId: 'highprio'
         });
-        for (let i = 0; i < 10_000; i++) {
+        for (let i = 0; i < 999; i++) {
             await writer.save({
                 sourceTable: testTable,
                 tag: storage.SaveOperationTag.INSERT,
                 after: {
-                    id: `${i}`,
-                    description: 'low prio'
+                    id: `first-${i}`,
+                    description: 'first low-priority bucket'
                 },
-                afterReplicaId: `${i}`
+                afterReplicaId: `first-${i}`
+            });
+        }
+        for (let i = 0; i < 9_001; i++) {
+            await writer.save({
+                sourceTable: testTable,
+                tag: storage.SaveOperationTag.INSERT,
+                after: {
+                    id: `low-${i}`,
+                    description: 'last low-priority bucket'
+                },
+                afterReplicaId: `low-${i}`
             });
         }
         await writer.commit('0/1');
@@ -259,17 +278,24 @@ bucket_definitions:
             token: new JwtPayload({ sub: '', exp: Date.now() / 1000 + 10 }),
             isEncodingAsBson: false
         });
-        let sentCheckpoints = 0;
-        let sentRows = 0;
+        const dataLines = [];
+        const partialCheckpoints = [];
+        const checkpoints = [];
+        const checkpointCompletes = [];
+        let afterCheckpointDiff = false;
+        let interruptionTriggered = false;
         for await (let next of stream) {
             if (typeof next == 'string') {
                 next = JSON.parse(next);
             }
             if (typeof next === 'object' && next !== null) {
                 if ('partial_checkpoint_complete' in next) {
-                    if (sentCheckpoints == 1) {
-                        // Save new data to interrupt the low-priority sync.
-                        // Add another high-priority row. This should interrupt the long-running low-priority sync.
+                    partialCheckpoints.push({
+                        priority: next.partial_checkpoint_complete.priority,
+                        afterCheckpointDiff
+                    });
+                    if (next.partial_checkpoint_complete.priority == 2 && !interruptionTriggered) {
+                        interruptionTriggered = true;
                         await writer.save({
                             sourceTable: testTable,
                             tag: storage.SaveOperationTag.INSERT,
@@ -280,27 +306,69 @@ bucket_definitions:
                             afterReplicaId: 'highprio2'
                         });
                         await writer.commit('0/2');
-                    }
-                    else {
-                        // Low-priority sync from the first checkpoint was interrupted. This should not happen before
-                        // 1000 low-priority items were synchronized.
-                        expect(sentCheckpoints).toBe(2);
-                        expect(sentRows).toBeGreaterThan(1000);
+                        // Let the checkpoint watcher observe the update before requesting the next priority.
+                        await timers.setTimeout(50);
                     }
                 }
                 if ('checkpoint' in next || 'checkpoint_diff' in next) {
-                    sentCheckpoints += 1;
+                    checkpoints.push(next);
+                    afterCheckpointDiff = 'checkpoint_diff' in next;
                 }
                 if ('data' in next) {
-                    sentRows += next.data.data.length;
+                    dataLines.push({
+                        bucket: next.data.bucket,
+                        objectIds: next.data.data.flatMap((entry) => (entry.object_id == null ? [] : [entry.object_id])),
+                        afterCheckpointDiff
+                    });
                 }
                 if ('checkpoint_complete' in next) {
+                    checkpointCompletes.push(next);
                     break;
                 }
             }
         }
-        expect(sentCheckpoints).toBe(2);
-        expect(sentRows).toBe(10002);
+        // Expected flow (data may be split into any number of chunks):
+        //
+        // checkpoint
+        // data: b1 contains highprio
+        // partial_checkpoint_complete: priority 1
+        // data: b0a contains all 999 first-* rows (1,000 operations including highprio)
+        // partial_checkpoint_complete: priority 2
+        // ## add highprio2, interrupting before b0b starts
+        // checkpoint_diff
+        // data: b1 contains highprio2
+        // partial_checkpoint_complete: priority 1
+        // data: b0b contains all 9,001 low-* rows carried over from the interrupted checkpoint
+        // checkpoint_complete: only for the new checkpoint
+        expect(interruptionTriggered).toBe(true);
+        expect(checkpoints).toHaveLength(2);
+        expect(checkpoints[0]).toHaveProperty('checkpoint');
+        expect(checkpoints[1]).toHaveProperty('checkpoint_diff');
+        expect(partialCheckpoints).toEqual([
+            { priority: 1, afterCheckpointDiff: false },
+            { priority: 2, afterCheckpointDiff: false },
+            { priority: 1, afterCheckpointDiff: true }
+        ]);
+        expect(checkpointCompletes).toEqual([
+            {
+                checkpoint_complete: {
+                    last_op_id: checkpoints[1].checkpoint_diff.last_op_id
+                }
+            }
+        ]);
+        const objectIdsFor = (bucket, afterDiff) => dataLines
+            .filter((line) => line.bucket == bucket && line.afterCheckpointDiff == afterDiff)
+            .flatMap((line) => line.objectIds);
+        expect(objectIdsFor(b1Bucket, false)).toEqual(['highprio']);
+        const b0aObjectIds = objectIdsFor(b0aBucket, false);
+        expect(new Set(b0aObjectIds)).toEqual(new Set(Array.from({ length: 999 }, (_, i) => `first-${i}`)));
+        expect(b0aObjectIds).toHaveLength(999);
+        expect(objectIdsFor(b0bBucket, false)).toEqual([]);
+        expect(objectIdsFor(b1Bucket, true)).toEqual(['highprio2']);
+        expect(objectIdsFor(b0aBucket, true)).toEqual([]);
+        const b0bObjectIds = objectIdsFor(b0bBucket, true);
+        expect(new Set(b0bObjectIds)).toEqual(new Set(Array.from({ length: 9_001 }, (_, i) => `low-${i}`)));
+        expect(b0bObjectIds).toHaveLength(9_001);
     });
     test('sync interruptions with unrelated data', async () => {
         await using f = await factory();
@@ -420,33 +488,33 @@ bucket_definitions:
         expect(sentCheckpoints).toBe(2);
         expect(sentRows).toBe(10002);
     });
-    test('sync interrupts low-priority buckets on new checkpoints (2)', async () => {
+    test('restarts updated low-priority buckets at a new checkpoint', async () => {
         await using f = await factory();
-        // bucket0a -> send all data
-        // then interrupt checkpoint with new data for all buckets
-        // -> data for all buckets should be sent in the new checkpoint
         const syncRules = await updateSyncRules(f, {
             content: `
 bucket_definitions:
   b0a:
     priority: 2
     data:
-      - SELECT * FROM test WHERE LENGTH(id) <= 5;
+      - SELECT * FROM test WHERE substring(id, 1, 6) = 'first-';
   b0b:
-    priority: 2
+    priority: 3
     data:
-      - SELECT * FROM test WHERE LENGTH(id) <= 5;
+      - SELECT * FROM test WHERE substring(id, 1, 4) = 'low-';
   b1:
     priority: 1
     data:
-      - SELECT * FROM test WHERE LENGTH(id) > 5;
+      - SELECT * FROM test WHERE substring(id, 1, 8) = 'highprio';
     `
         });
         const bucketStorage = f.getInstance(syncRules);
         await using writer = await bucketStorage.createWriter(test_utils.BATCH_OPTIONS);
         const testTable = await test_utils.resolveTestTable(writer, 'test', ['id'], config);
+        const syncRulesContent = syncRules.syncConfigContent[0];
+        const b0aBucket = test_utils.bucketRequest(syncRulesContent, 'b0a[]').bucket;
+        const b0bBucket = test_utils.bucketRequest(syncRulesContent, 'b0b[]').bucket;
+        const b1Bucket = test_utils.bucketRequest(syncRulesContent, 'b1[]').bucket;
         await writer.markAllSnapshotDone('0/1');
-        // Initial data: Add one priority row and 10k low-priority rows.
         await writer.save({
             sourceTable: testTable,
             tag: storage.SaveOperationTag.INSERT,
@@ -456,15 +524,26 @@ bucket_definitions:
             },
             afterReplicaId: 'highprio'
         });
+        for (let i = 0; i < 999; i++) {
+            await writer.save({
+                sourceTable: testTable,
+                tag: storage.SaveOperationTag.INSERT,
+                after: {
+                    id: `first-${i}`,
+                    description: 'first low-priority bucket'
+                },
+                afterReplicaId: `first-${i}`
+            });
+        }
         for (let i = 0; i < 2_000; i++) {
             await writer.save({
                 sourceTable: testTable,
                 tag: storage.SaveOperationTag.INSERT,
                 after: {
-                    id: `${i}`,
+                    id: `low-${i}`,
                     description: 'low prio'
                 },
-                afterReplicaId: `${i}`
+                afterReplicaId: `low-${i}`
             });
         }
         await writer.commit('0/1');
@@ -481,25 +560,24 @@ bucket_definitions:
             token: new JwtPayload({ sub: '', exp: Date.now() / 1000 + 10 }),
             isEncodingAsBson: false
         });
-        let sentRows = 0;
-        let lines = [];
+        const dataLines = [];
+        const partialCheckpoints = [];
+        const checkpoints = [];
+        const checkpointCompletes = [];
+        let afterCheckpointDiff = false;
+        let interruptionTriggered = false;
         for await (let next of stream) {
             if (typeof next == 'string') {
                 next = JSON.parse(next);
             }
             if (typeof next === 'object' && next !== null) {
                 if ('partial_checkpoint_complete' in next) {
-                    lines.push(next);
-                }
-                if ('checkpoint' in next || 'checkpoint_diff' in next) {
-                    lines.push(next);
-                }
-                if ('data' in next) {
-                    lines.push({ data: { ...next.data, data: undefined } });
-                    sentRows += next.data.data.length;
-                    if (sentRows == 1001) {
-                        // Save new data to interrupt the low-priority sync.
-                        // Add another high-priority row. This should interrupt the long-running low-priority sync.
+                    partialCheckpoints.push({
+                        priority: next.partial_checkpoint_complete.priority,
+                        afterCheckpointDiff
+                    });
+                    if (next.partial_checkpoint_complete.priority == 2 && !interruptionTriggered) {
+                        interruptionTriggered = true;
                         await writer.save({
                             sourceTable: testTable,
                             tag: storage.SaveOperationTag.INSERT,
@@ -509,48 +587,79 @@ bucket_definitions:
                             },
                             afterReplicaId: 'highprio2'
                         });
-                        // Also add a low-priority row
                         await writer.save({
                             sourceTable: testTable,
                             tag: storage.SaveOperationTag.INSERT,
                             after: {
-                                id: '2001',
+                                id: 'low-2000',
                                 description: 'Another low-priority row'
                             },
-                            afterReplicaId: '2001'
+                            afterReplicaId: 'low-2000'
                         });
                         await writer.commit('0/2');
-                    }
-                    if (sentRows >= 1000 && sentRows <= 2001) {
-                        // pause for a bit to give the stream time to process interruptions.
-                        // This covers the data batch above and the next one.
+                        // Let the checkpoint watcher observe the update before requesting the next priority.
                         await timers.setTimeout(50);
                     }
                 }
+                if ('checkpoint' in next || 'checkpoint_diff' in next) {
+                    checkpoints.push(next);
+                    afterCheckpointDiff = 'checkpoint_diff' in next;
+                }
+                if ('data' in next) {
+                    dataLines.push({
+                        bucket: next.data.bucket,
+                        objectIds: next.data.data.flatMap((entry) => (entry.object_id == null ? [] : [entry.object_id])),
+                        afterCheckpointDiff
+                    });
+                }
                 if ('checkpoint_complete' in next) {
-                    lines.push(next);
+                    checkpointCompletes.push(next);
                     break;
                 }
             }
         }
-        // Expected lines (full details in snapshot):
+        // Expected flow (data may be split into any number of chunks):
         //
-        // checkpoint (4001)
-        // data (b1[] 0 -> 1)
-        // partial_checkpoint_complete (4001, priority 1)
-        // data (b0a[], 0 -> 2000)
-        // ## adds new data, interrupting the checkpoint
-        // data (b0a[], 2000 -> 4000) # expected - stream is already busy with this by the time it receives the interruption
-        // checkpoint_diff (4004)
-        // data (b1[], 1 -> 4002)
-        // partial_checkpoint_complete (4004, priority 1)
-        // data (b0a[], 4000 -> 4003)
-        // data (b0b[], 0 -> 1999)
-        // data (b0b[], 1999 -> 3999)
-        // data (b0b[], 3999 -> 4004)
-        // checkpoint_complete (4004)
-        expect(lines).toMatchSnapshot();
-        expect(sentRows).toBe(4004);
+        // checkpoint
+        // data: b1 contains highprio
+        // partial_checkpoint_complete: priority 1
+        // data: b0a contains all 999 first-* rows (1,000 operations including highprio)
+        // partial_checkpoint_complete: priority 2
+        // ## add highprio2 and low-2000, interrupting before b0b starts
+        // checkpoint_diff
+        // data: b1 contains highprio2
+        // partial_checkpoint_complete: priority 1
+        // data: b0b contains all 2,001 low-* rows
+        // checkpoint_complete: only for the new checkpoint
+        expect(interruptionTriggered).toBe(true);
+        expect(checkpoints).toHaveLength(2);
+        expect(checkpoints[0]).toHaveProperty('checkpoint');
+        expect(checkpoints[1]).toHaveProperty('checkpoint_diff');
+        expect(partialCheckpoints).toEqual([
+            { priority: 1, afterCheckpointDiff: false },
+            { priority: 2, afterCheckpointDiff: false },
+            { priority: 1, afterCheckpointDiff: true }
+        ]);
+        expect(checkpointCompletes).toEqual([
+            {
+                checkpoint_complete: {
+                    last_op_id: checkpoints[1].checkpoint_diff.last_op_id
+                }
+            }
+        ]);
+        const objectIdsFor = (bucket, afterDiff) => dataLines
+            .filter((line) => line.bucket == bucket && line.afterCheckpointDiff == afterDiff)
+            .flatMap((line) => line.objectIds);
+        expect(objectIdsFor(b1Bucket, false)).toEqual(['highprio']);
+        const b0aObjectIds = objectIdsFor(b0aBucket, false);
+        expect(new Set(b0aObjectIds)).toEqual(new Set(Array.from({ length: 999 }, (_, i) => `first-${i}`)));
+        expect(b0aObjectIds).toHaveLength(999);
+        expect(objectIdsFor(b0bBucket, false)).toEqual([]);
+        expect(objectIdsFor(b1Bucket, true)).toEqual(['highprio2']);
+        expect(objectIdsFor(b0aBucket, true)).toEqual([]);
+        const b0bObjectIds = objectIdsFor(b0bBucket, true);
+        expect(new Set(b0bObjectIds)).toEqual(new Set(Array.from({ length: 2_001 }, (_, i) => `low-${i}`)));
+        expect(b0bObjectIds).toHaveLength(2_001);
     });
     test('sends checkpoint complete line for empty checkpoint', async () => {
         await using f = await factory();
@@ -931,6 +1040,82 @@ bucket_definitions:
         const expLines = await getCheckpointLines(iter);
         expect(expLines).toMatchSnapshot();
     });
+    test('checksum invalidation skips the candidate checkpoint', async (context) => {
+        await using f = await factory();
+        const syncRules = await updateSyncRules(f, {
+            content: BASIC_SYNC_RULES
+        });
+        const bucketStorage = await f.getInstance(syncRules);
+        await using writer = await bucketStorage.createWriter(test_utils.BATCH_OPTIONS);
+        const testTable = await test_utils.resolveTestTable(writer, 'test', ['id'], config);
+        await writer.markAllSnapshotDone('0/1');
+        await writer.save({
+            sourceTable: testTable,
+            tag: storage.SaveOperationTag.INSERT,
+            after: {
+                id: 't1',
+                description: 'Test 1'
+            },
+            afterReplicaId: 't1'
+        });
+        await writer.commit('0/1');
+        let resolveInvalidatedChecksum;
+        const invalidatedChecksum = new Promise((resolve) => {
+            resolveInvalidatedChecksum = resolve;
+        });
+        const checksumSpy = vi.spyOn(bucketStorage, 'getChecksums').mockImplementationOnce(async (checkpoint, buckets) => {
+            resolveInvalidatedChecksum();
+            throw new storage.CheckpointChecksumInvalidatedError(checkpoint.checkpoint, buckets[0].bucket);
+        });
+        const stream = sync.streamResponse({
+            syncContext,
+            bucketStorage,
+            syncRules: bucketStorage.getParsedSyncRules(test_utils.PARSE_OPTIONS),
+            params: {
+                buckets: [],
+                include_checksum: true,
+                raw_data: true
+            },
+            tracker,
+            token: new JwtPayload({ sub: '', exp: Date.now() / 1000 + 10 }),
+            isEncodingAsBson: false
+        });
+        const iter = stream[Symbol.asyncIterator]();
+        context.onTestFinished(() => {
+            iter.return?.();
+        });
+        const linesPromise = getCheckpointLines(iter, { consume: true });
+        await invalidatedChecksum;
+        // The first candidate was discarded before CheckpointLine.advance(). A later
+        // checkpoint must therefore be calculated from the original connection state.
+        await writer.save({
+            sourceTable: testTable,
+            tag: storage.SaveOperationTag.INSERT,
+            after: {
+                id: 't2',
+                description: 'Test 2'
+            },
+            afterReplicaId: 't2'
+        });
+        await writer.commit('0/2');
+        const lines = await linesPromise;
+        expect(checksumSpy).toHaveBeenCalledTimes(2);
+        expect(lines[0]).toEqual({
+            checkpoint: expect.objectContaining({
+                last_op_id: '2'
+            })
+        });
+        expect(lines).not.toContainEqual({
+            checkpoint: expect.objectContaining({
+                last_op_id: '1'
+            })
+        });
+        expect(lines.at(-1)).toEqual({
+            checkpoint_complete: expect.objectContaining({
+                last_op_id: '2'
+            })
+        });
+    });
     test('compacting data - invalidate checkpoint', async (context) => {
         // This tests a case of a compact operation invalidating a checkpoint in the
         // middle of syncing data.
@@ -943,6 +1128,7 @@ bucket_definitions:
         const bucketStorage = await f.getInstance(syncRules);
         await using writer = await bucketStorage.createWriter(test_utils.BATCH_OPTIONS);
         const testTable = await test_utils.resolveTestTable(writer, 'test', ['id'], config);
+        const bucket = bucketRequest(syncRules.syncConfigContent[0], 'mybucket[]').bucket;
         await writer.markAllSnapshotDone('0/1');
         await writer.save({
             sourceTable: testTable,
@@ -1010,10 +1196,7 @@ bucket_definitions:
             afterReplicaId: 't2'
         });
         await writer.commit('0/2');
-        await bucketStorage.compact({
-            minBucketChanges: 1,
-            minChangeRatio: 0
-        });
+        await compactActive(f, { compactBuckets: [bucket] });
         const lines2 = await getCheckpointLines(iter, { consume: true });
         // Snapshot test checks for changes in general.
         // The tests after that documents the specific things we're looking for
@@ -1084,6 +1267,7 @@ bucket_definitions:
         const bucketStorage = await f.getInstance(syncRules);
         await using writer = await bucketStorage.createWriter(test_utils.BATCH_OPTIONS);
         const testTable = await test_utils.resolveTestTable(writer, 'test', ['id'], config);
+        const highPriorityBucket = bucketRequest(syncRules.syncConfigContent[0], 'high_priority[]').bucket;
         await writer.markAllSnapshotDone('0/1');
         await writer.save({
             sourceTable: testTable,
@@ -1160,9 +1344,10 @@ bucket_definitions:
             afterReplicaId: 'high2'
         });
         await writer.commit('0/2');
-        await bucketStorage.compact({
-            minBucketChanges: 1,
-            minChangeRatio: 0
+        await compactActive(f, {
+            // Explicitly compact the high-priority bucket: V3 schedules background
+            // compaction, while this test needs compaction at this exact point.
+            compactBuckets: [highPriorityBucket]
         });
         const lines = await getCheckpointLines(iter, { consume: true });
         const nextCheckpointIndex = lines.findIndex((line) => line?.checkpoint_diff != null);
@@ -1200,7 +1385,7 @@ bucket_definitions:
                 user_id: 'test',
                 heads: { '1': '1/0' }
             }
-        ])).get('test');
+        ])).writeCheckpoints.get('test');
         const params = {
             syncContext,
             bucketStorage,