@hatk/hatk 0.0.1-alpha.8 → 0.0.1-alpha.80
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/adapter.d.ts +19 -0
- package/dist/adapter.d.ts.map +1 -0
- package/dist/adapter.js +110 -0
- package/dist/backfill.d.ts +21 -2
- package/dist/backfill.d.ts.map +1 -1
- package/dist/backfill.js +102 -37
- package/dist/car.d.ts +42 -10
- package/dist/car.d.ts.map +1 -1
- package/dist/car.js +154 -14
- package/dist/cli.js +243 -1043
- package/dist/config.d.ts +74 -1
- package/dist/config.d.ts.map +1 -1
- package/dist/config.js +43 -9
- package/dist/database/adapter-factory.d.ts +6 -0
- package/dist/database/adapter-factory.d.ts.map +1 -0
- package/dist/database/adapter-factory.js +20 -0
- package/dist/database/adapters/duckdb-search.d.ts +12 -0
- package/dist/database/adapters/duckdb-search.d.ts.map +1 -0
- package/dist/database/adapters/duckdb-search.js +27 -0
- package/dist/database/adapters/duckdb.d.ts +25 -0
- package/dist/database/adapters/duckdb.d.ts.map +1 -0
- package/dist/database/adapters/duckdb.js +161 -0
- package/dist/database/adapters/sqlite-search.d.ts +23 -0
- package/dist/database/adapters/sqlite-search.d.ts.map +1 -0
- package/dist/database/adapters/sqlite-search.js +74 -0
- package/dist/database/adapters/sqlite.d.ts +18 -0
- package/dist/database/adapters/sqlite.d.ts.map +1 -0
- package/dist/database/adapters/sqlite.js +88 -0
- package/dist/{db.d.ts → database/db.d.ts} +57 -6
- package/dist/database/db.d.ts.map +1 -0
- package/dist/{db.js → database/db.js} +755 -552
- package/dist/database/dialect.d.ts +45 -0
- package/dist/database/dialect.d.ts.map +1 -0
- package/dist/database/dialect.js +72 -0
- package/dist/{fts.d.ts → database/fts.d.ts} +7 -0
- package/dist/database/fts.d.ts.map +1 -0
- package/dist/{fts.js → database/fts.js} +116 -32
- package/dist/database/index.d.ts +7 -0
- package/dist/database/index.d.ts.map +1 -0
- package/dist/database/index.js +6 -0
- package/dist/database/ports.d.ts +50 -0
- package/dist/database/ports.d.ts.map +1 -0
- package/dist/database/ports.js +1 -0
- package/dist/{schema.d.ts → database/schema.d.ts} +14 -3
- package/dist/database/schema.d.ts.map +1 -0
- package/dist/{schema.js → database/schema.js} +81 -41
- package/dist/dev-entry.d.ts +8 -0
- package/dist/dev-entry.d.ts.map +1 -0
- package/dist/dev-entry.js +119 -0
- package/dist/feeds.d.ts +12 -8
- package/dist/feeds.d.ts.map +1 -1
- package/dist/feeds.js +51 -6
- package/dist/hooks.d.ts +85 -0
- package/dist/hooks.d.ts.map +1 -0
- package/dist/hooks.js +163 -0
- package/dist/hydrate.d.ts +7 -6
- package/dist/hydrate.d.ts.map +1 -1
- package/dist/hydrate.js +4 -16
- package/dist/indexer.d.ts +109 -3
- package/dist/indexer.d.ts.map +1 -1
- package/dist/indexer.js +338 -66
- package/dist/jetstream.d.ts +118 -0
- package/dist/jetstream.d.ts.map +1 -0
- package/dist/jetstream.js +225 -0
- package/dist/labels.d.ts +36 -0
- package/dist/labels.d.ts.map +1 -1
- package/dist/labels.js +71 -6
- package/dist/lexicon-resolve.d.ts.map +1 -1
- package/dist/lexicon-resolve.js +27 -112
- package/dist/lexicons/com/atproto/label/defs.json +75 -0
- package/dist/lexicons/com/atproto/moderation/defs.json +30 -0
- package/dist/lexicons/com/atproto/repo/strongRef.json +21 -0
- package/dist/lexicons/dev/hatk/applyWrites.json +87 -0
- package/dist/lexicons/dev/hatk/createRecord.json +40 -0
- package/dist/lexicons/dev/hatk/createReport.json +48 -0
- package/dist/lexicons/dev/hatk/deleteRecord.json +25 -0
- package/dist/lexicons/dev/hatk/describeCollections.json +41 -0
- package/dist/lexicons/dev/hatk/describeFeeds.json +29 -0
- package/dist/lexicons/dev/hatk/describeLabels.json +45 -0
- package/dist/lexicons/dev/hatk/getFeed.json +30 -0
- package/dist/lexicons/dev/hatk/getPreferences.json +19 -0
- package/dist/lexicons/dev/hatk/getRecord.json +26 -0
- package/dist/lexicons/dev/hatk/getRecords.json +32 -0
- package/dist/lexicons/dev/hatk/putPreference.json +28 -0
- package/dist/lexicons/dev/hatk/putRecord.json +41 -0
- package/dist/lexicons/dev/hatk/searchRecords.json +32 -0
- package/dist/lexicons/dev/hatk/uploadBlob.json +23 -0
- package/dist/logger.d.ts +29 -0
- package/dist/logger.d.ts.map +1 -1
- package/dist/logger.js +29 -0
- package/dist/main.js +161 -72
- package/dist/mst.d.ts +18 -1
- package/dist/mst.d.ts.map +1 -1
- package/dist/mst.js +19 -8
- package/dist/oauth/db.d.ts +3 -1
- package/dist/oauth/db.d.ts.map +1 -1
- package/dist/oauth/db.js +56 -19
- package/dist/oauth/discovery.d.ts.map +1 -1
- package/dist/oauth/discovery.js +13 -1
- package/dist/oauth/server.d.ts +92 -1
- package/dist/oauth/server.d.ts.map +1 -1
- package/dist/oauth/server.js +352 -37
- package/dist/oauth/service-describe.d.ts +31 -0
- package/dist/oauth/service-describe.d.ts.map +1 -0
- package/dist/oauth/service-describe.js +61 -0
- package/dist/oauth/session.d.ts +32 -0
- package/dist/oauth/session.d.ts.map +1 -0
- package/dist/oauth/session.js +131 -0
- package/dist/opengraph.d.ts +10 -0
- package/dist/opengraph.d.ts.map +1 -1
- package/dist/opengraph.js +80 -40
- package/dist/pds-proxy.d.ts +82 -0
- package/dist/pds-proxy.d.ts.map +1 -0
- package/dist/pds-proxy.js +356 -0
- package/dist/private-collections.d.ts +8 -0
- package/dist/private-collections.d.ts.map +1 -0
- package/dist/private-collections.js +22 -0
- package/dist/push.d.ts +54 -0
- package/dist/push.d.ts.map +1 -0
- package/dist/push.js +367 -0
- package/dist/renderer.d.ts +27 -0
- package/dist/renderer.d.ts.map +1 -0
- package/dist/renderer.js +46 -0
- package/dist/resolve-hatk.d.ts +6 -0
- package/dist/resolve-hatk.d.ts.map +1 -0
- package/dist/resolve-hatk.js +20 -0
- package/dist/response.d.ts +16 -0
- package/dist/response.d.ts.map +1 -0
- package/dist/response.js +69 -0
- package/dist/scanner.d.ts +21 -0
- package/dist/scanner.d.ts.map +1 -0
- package/dist/scanner.js +88 -0
- package/dist/seed.d.ts +19 -0
- package/dist/seed.d.ts.map +1 -1
- package/dist/seed.js +43 -4
- package/dist/server-init.d.ts +8 -0
- package/dist/server-init.d.ts.map +1 -0
- package/dist/server-init.js +62 -0
- package/dist/server.d.ts +26 -3
- package/dist/server.d.ts.map +1 -1
- package/dist/server.js +832 -638
- package/dist/setup.d.ts +28 -1
- package/dist/setup.d.ts.map +1 -1
- package/dist/setup.js +50 -3
- package/dist/templates/feed.tpl +14 -0
- package/dist/templates/hook.tpl +5 -0
- package/dist/templates/label.tpl +15 -0
- package/dist/templates/og.tpl +17 -0
- package/dist/templates/seed.tpl +11 -0
- package/dist/templates/setup.tpl +5 -0
- package/dist/templates/test-feed.tpl +19 -0
- package/dist/templates/test-xrpc.tpl +19 -0
- package/dist/templates/xrpc.tpl +41 -0
- package/dist/test.d.ts +1 -1
- package/dist/test.d.ts.map +1 -1
- package/dist/test.js +41 -32
- package/dist/views.js +1 -1
- package/dist/vite-plugin.d.ts +1 -1
- package/dist/vite-plugin.d.ts.map +1 -1
- package/dist/vite-plugin.js +254 -66
- package/dist/xrpc.d.ts +82 -11
- package/dist/xrpc.d.ts.map +1 -1
- package/dist/xrpc.js +196 -39
- package/package.json +17 -9
- package/public/admin-auth.js +271 -306
- package/public/admin.html +177 -61
- package/dist/db.d.ts.map +0 -1
- package/dist/fts.d.ts.map +0 -1
- package/dist/oauth/hooks.d.ts +0 -10
- package/dist/oauth/hooks.d.ts.map +0 -1
- package/dist/oauth/hooks.js +0 -40
- package/dist/schema.d.ts.map +0 -1
- package/dist/test-browser.d.ts +0 -14
- package/dist/test-browser.d.ts.map +0 -1
- package/dist/test-browser.js +0 -26
|
@@ -1,136 +1,59 @@
|
|
|
1
|
-
import {
|
|
2
|
-
import {
|
|
3
|
-
import {
|
|
4
|
-
import {
|
|
5
|
-
import {
|
|
6
|
-
let
|
|
7
|
-
let
|
|
8
|
-
let readCon;
|
|
1
|
+
import { toSnakeCase, q } from "./schema.js";
|
|
2
|
+
import { getSearchColumns, stripStopWords, getSearchPort, updateFtsRecord, deleteFtsRecord } from "./fts.js";
|
|
3
|
+
import { emit, timer } from "../logger.js";
|
|
4
|
+
import { OAUTH_DDL } from "../oauth/db.js";
|
|
5
|
+
import { getDialect } from "./dialect.js";
|
|
6
|
+
let port;
|
|
7
|
+
let dialect;
|
|
9
8
|
const schemas = new Map();
|
|
10
|
-
export function
|
|
11
|
-
|
|
12
|
-
readCon?.closeSync();
|
|
13
|
-
}
|
|
14
|
-
catch { }
|
|
15
|
-
try {
|
|
16
|
-
con?.closeSync();
|
|
17
|
-
}
|
|
18
|
-
catch { }
|
|
19
|
-
try {
|
|
20
|
-
instance?.closeSync();
|
|
21
|
-
}
|
|
22
|
-
catch { }
|
|
23
|
-
}
|
|
24
|
-
let writeQueue = Promise.resolve();
|
|
25
|
-
let readQueue = Promise.resolve();
|
|
26
|
-
function enqueue(queue, fn) {
|
|
27
|
-
if (queue === 'write') {
|
|
28
|
-
const p = writeQueue.then(fn);
|
|
29
|
-
writeQueue = p.then(() => { }, () => { });
|
|
30
|
-
return p;
|
|
31
|
-
}
|
|
32
|
-
else {
|
|
33
|
-
const p = readQueue.then(fn);
|
|
34
|
-
readQueue = p.then(() => { }, () => { });
|
|
35
|
-
return p;
|
|
36
|
-
}
|
|
9
|
+
export function getDatabasePort() {
|
|
10
|
+
return port;
|
|
37
11
|
}
|
|
38
|
-
function
|
|
39
|
-
|
|
40
|
-
const idx = i + 1;
|
|
41
|
-
const value = params[i];
|
|
42
|
-
if (value === null || value === undefined) {
|
|
43
|
-
prepared.bindNull(idx);
|
|
44
|
-
}
|
|
45
|
-
else if (typeof value === 'string') {
|
|
46
|
-
prepared.bindVarchar(idx, value);
|
|
47
|
-
}
|
|
48
|
-
else if (typeof value === 'number') {
|
|
49
|
-
if (Number.isInteger(value)) {
|
|
50
|
-
prepared.bindInteger(idx, value);
|
|
51
|
-
}
|
|
52
|
-
else {
|
|
53
|
-
prepared.bindDouble(idx, value);
|
|
54
|
-
}
|
|
55
|
-
}
|
|
56
|
-
else if (typeof value === 'boolean') {
|
|
57
|
-
prepared.bindBoolean(idx, value);
|
|
58
|
-
}
|
|
59
|
-
else if (typeof value === 'bigint') {
|
|
60
|
-
prepared.bindBigInt(idx, value);
|
|
61
|
-
}
|
|
62
|
-
else if (value instanceof Uint8Array) {
|
|
63
|
-
prepared.bindBlob(idx, value);
|
|
64
|
-
}
|
|
65
|
-
else {
|
|
66
|
-
prepared.bindVarchar(idx, String(value));
|
|
67
|
-
}
|
|
68
|
-
}
|
|
12
|
+
export function getSqlDialect() {
|
|
13
|
+
return dialect;
|
|
69
14
|
}
|
|
70
|
-
|
|
71
|
-
|
|
72
|
-
await con.run(sql);
|
|
73
|
-
return;
|
|
74
|
-
}
|
|
75
|
-
const prepared = await con.prepare(sql);
|
|
76
|
-
bindParams(prepared, params);
|
|
77
|
-
await prepared.run();
|
|
15
|
+
export function closeDatabase() {
|
|
16
|
+
port?.close();
|
|
78
17
|
}
|
|
79
|
-
async function run(sql,
|
|
80
|
-
return
|
|
18
|
+
async function run(sql, params = []) {
|
|
19
|
+
return port.execute(sql, params);
|
|
81
20
|
}
|
|
82
21
|
export async function runBatch(operations) {
|
|
83
|
-
|
|
84
|
-
|
|
22
|
+
await port.beginTransaction();
|
|
23
|
+
try {
|
|
85
24
|
for (const op of operations) {
|
|
86
25
|
try {
|
|
87
|
-
|
|
88
|
-
await con.run(op.sql);
|
|
89
|
-
}
|
|
90
|
-
else {
|
|
91
|
-
const prepared = await con.prepare(op.sql);
|
|
92
|
-
bindParams(prepared, op.params);
|
|
93
|
-
await prepared.run();
|
|
94
|
-
}
|
|
26
|
+
await port.execute(op.sql, op.params);
|
|
95
27
|
}
|
|
96
28
|
catch {
|
|
97
29
|
// Skip bad records, continue with rest of batch
|
|
98
30
|
}
|
|
99
31
|
}
|
|
100
|
-
await
|
|
101
|
-
}
|
|
102
|
-
|
|
103
|
-
|
|
104
|
-
if (params.length === 0) {
|
|
105
|
-
const reader = await readCon.runAndReadAll(sql);
|
|
106
|
-
return reader.getRowObjects();
|
|
32
|
+
await port.commit();
|
|
33
|
+
}
|
|
34
|
+
catch {
|
|
35
|
+
await port.rollback();
|
|
107
36
|
}
|
|
108
|
-
const prepared = await readCon.prepare(sql);
|
|
109
|
-
bindParams(prepared, params);
|
|
110
|
-
const reader = await prepared.runAndReadAll();
|
|
111
|
-
return reader.getRowObjects();
|
|
112
37
|
}
|
|
113
|
-
async function all(sql,
|
|
114
|
-
return
|
|
38
|
+
async function all(sql, params = []) {
|
|
39
|
+
return port.query(sql, params);
|
|
115
40
|
}
|
|
116
|
-
export async function initDatabase(dbPath, tableSchemas, ddlStatements) {
|
|
117
|
-
|
|
118
|
-
|
|
119
|
-
|
|
41
|
+
export async function initDatabase(adapter, dbPath, tableSchemas, ddlStatements) {
|
|
42
|
+
port = adapter;
|
|
43
|
+
dialect = getDialect(adapter.dialect);
|
|
44
|
+
await port.open(dbPath);
|
|
120
45
|
for (const schema of tableSchemas) {
|
|
121
46
|
schemas.set(schema.collection, schema);
|
|
122
47
|
}
|
|
123
48
|
for (const ddl of ddlStatements) {
|
|
124
|
-
|
|
125
|
-
await run(statement);
|
|
126
|
-
}
|
|
49
|
+
await port.executeMultiple(ddl);
|
|
127
50
|
}
|
|
128
51
|
// Internal tables for backfill state
|
|
129
52
|
await run(`CREATE TABLE IF NOT EXISTS _repos (
|
|
130
53
|
did TEXT PRIMARY KEY,
|
|
131
54
|
status TEXT NOT NULL DEFAULT 'pending',
|
|
132
55
|
handle TEXT,
|
|
133
|
-
backfilled_at
|
|
56
|
+
backfilled_at ${dialect.timestampType},
|
|
134
57
|
rev TEXT,
|
|
135
58
|
retry_count INTEGER NOT NULL DEFAULT 0,
|
|
136
59
|
retry_after INTEGER NOT NULL DEFAULT 0
|
|
@@ -142,79 +65,350 @@ export async function initDatabase(dbPath, tableSchemas, ddlStatements) {
|
|
|
142
65
|
catch { }
|
|
143
66
|
// Re-queue repos with missing handles so backfill populates them
|
|
144
67
|
await run(`UPDATE _repos SET status = 'pending' WHERE handle IS NULL`);
|
|
68
|
+
// Covering index for the status rollups the admin overview and backfill queue
|
|
69
|
+
// read. Skipped on DuckDB, which scans the column just as fast and would only
|
|
70
|
+
// pay the ART index's insert cost.
|
|
71
|
+
if (port.dialect !== 'duckdb') {
|
|
72
|
+
await run(`CREATE INDEX IF NOT EXISTS idx_repos_status ON _repos(status)`);
|
|
73
|
+
}
|
|
74
|
+
// resolveHandleToDid runs on every feed request that takes an `actor` handle,
|
|
75
|
+
// and without this it scans all of _repos. Not UNIQUE: during a handle change
|
|
76
|
+
// two rows can briefly claim the same handle, and a constraint violation there
|
|
77
|
+
// would fail the indexer write.
|
|
78
|
+
await run(`CREATE INDEX IF NOT EXISTS idx_repos_handle ON _repos(handle)`);
|
|
145
79
|
await run(`CREATE TABLE IF NOT EXISTS _cursor (
|
|
146
80
|
key TEXT PRIMARY KEY,
|
|
147
81
|
value TEXT NOT NULL
|
|
148
82
|
)`);
|
|
149
83
|
// Labels table (atproto-compatible)
|
|
150
|
-
|
|
151
|
-
|
|
152
|
-
|
|
153
|
-
|
|
154
|
-
|
|
155
|
-
|
|
156
|
-
|
|
157
|
-
|
|
158
|
-
|
|
159
|
-
|
|
84
|
+
if (dialect.supportsSequences) {
|
|
85
|
+
await run(`CREATE SEQUENCE IF NOT EXISTS _labels_seq START 1`);
|
|
86
|
+
await run(`CREATE TABLE IF NOT EXISTS _labels (
|
|
87
|
+
id INTEGER PRIMARY KEY DEFAULT nextval('_labels_seq'),
|
|
88
|
+
src TEXT NOT NULL,
|
|
89
|
+
uri TEXT NOT NULL,
|
|
90
|
+
val TEXT NOT NULL,
|
|
91
|
+
neg ${dialect.typeMap.boolean} DEFAULT FALSE,
|
|
92
|
+
cts ${dialect.timestampType} NOT NULL,
|
|
93
|
+
exp ${dialect.timestampType}
|
|
94
|
+
)`);
|
|
95
|
+
}
|
|
96
|
+
else {
|
|
97
|
+
await run(`CREATE TABLE IF NOT EXISTS _labels (
|
|
98
|
+
id INTEGER PRIMARY KEY AUTOINCREMENT,
|
|
99
|
+
src TEXT NOT NULL,
|
|
100
|
+
uri TEXT NOT NULL,
|
|
101
|
+
val TEXT NOT NULL,
|
|
102
|
+
neg INTEGER DEFAULT 0,
|
|
103
|
+
cts TEXT NOT NULL,
|
|
104
|
+
exp TEXT
|
|
105
|
+
)`);
|
|
106
|
+
}
|
|
160
107
|
await run(`CREATE INDEX IF NOT EXISTS idx_labels_uri ON _labels(uri)`);
|
|
161
108
|
await run(`CREATE INDEX IF NOT EXISTS idx_labels_src ON _labels(src)`);
|
|
162
109
|
// Preferences table (generic key-value per user)
|
|
163
110
|
await run(`CREATE TABLE IF NOT EXISTS _preferences (
|
|
164
111
|
did TEXT NOT NULL,
|
|
165
112
|
key TEXT NOT NULL,
|
|
166
|
-
value
|
|
167
|
-
updated_at
|
|
113
|
+
value ${dialect.jsonType} NOT NULL,
|
|
114
|
+
updated_at ${dialect.timestampType} DEFAULT ${dialect.currentTimestamp},
|
|
168
115
|
PRIMARY KEY (did, key)
|
|
116
|
+
)`);
|
|
117
|
+
// Reports table (user-submitted moderation reports)
|
|
118
|
+
if (dialect.supportsSequences) {
|
|
119
|
+
await run(`CREATE SEQUENCE IF NOT EXISTS _reports_seq START 1`);
|
|
120
|
+
await run(`CREATE TABLE IF NOT EXISTS _reports (
|
|
121
|
+
id INTEGER PRIMARY KEY DEFAULT nextval('_reports_seq'),
|
|
122
|
+
subject_uri TEXT NOT NULL,
|
|
123
|
+
subject_did TEXT NOT NULL,
|
|
124
|
+
label TEXT NOT NULL,
|
|
125
|
+
reason TEXT,
|
|
126
|
+
reported_by TEXT NOT NULL,
|
|
127
|
+
status TEXT NOT NULL DEFAULT 'open',
|
|
128
|
+
resolved_by TEXT,
|
|
129
|
+
resolved_at ${dialect.timestampType},
|
|
130
|
+
created_at ${dialect.timestampType} NOT NULL
|
|
131
|
+
)`);
|
|
132
|
+
}
|
|
133
|
+
else {
|
|
134
|
+
await run(`CREATE TABLE IF NOT EXISTS _reports (
|
|
135
|
+
id INTEGER PRIMARY KEY AUTOINCREMENT,
|
|
136
|
+
subject_uri TEXT NOT NULL,
|
|
137
|
+
subject_did TEXT NOT NULL,
|
|
138
|
+
label TEXT NOT NULL,
|
|
139
|
+
reason TEXT,
|
|
140
|
+
reported_by TEXT NOT NULL,
|
|
141
|
+
status TEXT NOT NULL DEFAULT 'open',
|
|
142
|
+
resolved_by TEXT,
|
|
143
|
+
resolved_at TEXT,
|
|
144
|
+
created_at TEXT NOT NULL
|
|
145
|
+
)`);
|
|
146
|
+
}
|
|
147
|
+
await run(`CREATE INDEX IF NOT EXISTS idx_reports_status ON _reports(status)`);
|
|
148
|
+
await run(`CREATE INDEX IF NOT EXISTS idx_reports_subject_uri ON _reports(subject_uri)`);
|
|
149
|
+
// Push notification tokens
|
|
150
|
+
await run(`CREATE TABLE IF NOT EXISTS _push_tokens (
|
|
151
|
+
did TEXT NOT NULL,
|
|
152
|
+
token TEXT NOT NULL,
|
|
153
|
+
platform TEXT NOT NULL,
|
|
154
|
+
created_at TEXT NOT NULL,
|
|
155
|
+
PRIMARY KEY (did, token)
|
|
169
156
|
)`);
|
|
170
157
|
// OAuth tables
|
|
171
|
-
|
|
172
|
-
|
|
158
|
+
await port.executeMultiple(OAUTH_DDL);
|
|
159
|
+
// Migrations: add pds_auth_server to existing sessions tables
|
|
160
|
+
try {
|
|
161
|
+
await run(`ALTER TABLE _oauth_sessions ADD COLUMN pds_auth_server TEXT`);
|
|
162
|
+
}
|
|
163
|
+
catch { }
|
|
164
|
+
}
|
|
165
|
+
/** Normalize SQL type names to handle dialect differences (e.g. VARCHAR → TEXT) */
|
|
166
|
+
function normalizeType(type) {
|
|
167
|
+
const upper = type.toUpperCase();
|
|
168
|
+
if (upper === 'VARCHAR' || upper === 'CHARACTER VARYING')
|
|
169
|
+
return 'TEXT';
|
|
170
|
+
if (upper === 'TIMESTAMP WITH TIME ZONE')
|
|
171
|
+
return 'TIMESTAMPTZ';
|
|
172
|
+
if (upper === 'BOOLEAN' || upper === 'BOOL')
|
|
173
|
+
return 'BOOLEAN';
|
|
174
|
+
if (upper === 'INT' || upper === 'INT4' || upper === 'INT8' || upper === 'BIGINT' || upper === 'SMALLINT')
|
|
175
|
+
return 'INTEGER';
|
|
176
|
+
return upper;
|
|
177
|
+
}
|
|
178
|
+
async function getExistingColumns(tableName) {
|
|
179
|
+
if (!/^[a-zA-Z0-9._]+$/.test(tableName)) {
|
|
180
|
+
throw new Error(`Invalid table name for introspection: ${tableName}`);
|
|
181
|
+
}
|
|
182
|
+
const cols = new Map();
|
|
183
|
+
try {
|
|
184
|
+
const query = dialect.introspectColumnsQuery(tableName);
|
|
185
|
+
const rows = await all(query);
|
|
186
|
+
for (const row of rows) {
|
|
187
|
+
// SQLite PRAGMA returns { name, type }, DuckDB returns { column_name, data_type }
|
|
188
|
+
const name = (row.column_name || row.name);
|
|
189
|
+
const type = normalizeType((row.data_type || row.type || 'TEXT'));
|
|
190
|
+
cols.set(name, type);
|
|
191
|
+
}
|
|
192
|
+
}
|
|
193
|
+
catch {
|
|
194
|
+
// Table doesn't exist yet
|
|
195
|
+
}
|
|
196
|
+
return cols;
|
|
197
|
+
}
|
|
198
|
+
function diffColumns(tableName, existingCols, expectedCols, changes) {
|
|
199
|
+
for (const [colName, colType] of expectedCols) {
|
|
200
|
+
if (!existingCols.has(colName)) {
|
|
201
|
+
changes.push({ table: tableName, action: 'add', column: colName, type: colType });
|
|
202
|
+
}
|
|
203
|
+
}
|
|
204
|
+
for (const [colName] of existingCols) {
|
|
205
|
+
if (!expectedCols.has(colName)) {
|
|
206
|
+
changes.push({ table: tableName, action: 'drop', column: colName });
|
|
207
|
+
}
|
|
208
|
+
}
|
|
209
|
+
for (const [colName, colType] of expectedCols) {
|
|
210
|
+
const existingType = existingCols.get(colName);
|
|
211
|
+
if (existingType && normalizeType(existingType) !== normalizeType(colType)) {
|
|
212
|
+
changes.push({ table: tableName, action: 'retype', column: colName, type: colType });
|
|
213
|
+
}
|
|
214
|
+
}
|
|
215
|
+
}
|
|
216
|
+
/** Build expected columns map for a child/union table */
|
|
217
|
+
function buildChildExpectedCols(columns) {
|
|
218
|
+
const expected = new Map();
|
|
219
|
+
expected.set('parent_uri', 'TEXT');
|
|
220
|
+
expected.set('parent_did', 'TEXT');
|
|
221
|
+
for (const col of columns) {
|
|
222
|
+
expected.set(col.name, normalizeType(col.sqlType));
|
|
223
|
+
}
|
|
224
|
+
return expected;
|
|
225
|
+
}
|
|
226
|
+
export async function migrateSchema(tableSchemas) {
|
|
227
|
+
const changes = [];
|
|
228
|
+
const newCollections = new Set();
|
|
229
|
+
for (const schema of tableSchemas) {
|
|
230
|
+
if (schema.columns.length === 0)
|
|
231
|
+
continue; // generic JSON storage, skip
|
|
232
|
+
const tableName = schema.collection;
|
|
233
|
+
const existingCols = await getExistingColumns(tableName);
|
|
234
|
+
if (existingCols.size === 0) {
|
|
235
|
+
newCollections.add(schema.collection);
|
|
236
|
+
continue; // table just created, nothing to migrate
|
|
237
|
+
}
|
|
238
|
+
// Expected columns: base columns (uri, cid, did, indexed_at) + schema columns
|
|
239
|
+
const expectedCols = new Map();
|
|
240
|
+
expectedCols.set('uri', 'TEXT');
|
|
241
|
+
expectedCols.set('cid', 'TEXT');
|
|
242
|
+
expectedCols.set('did', 'TEXT');
|
|
243
|
+
expectedCols.set('indexed_at', normalizeType(dialect.timestampType));
|
|
244
|
+
for (const col of schema.columns) {
|
|
245
|
+
expectedCols.set(col.name, normalizeType(col.sqlType));
|
|
246
|
+
}
|
|
247
|
+
diffColumns(tableName, existingCols, expectedCols, changes);
|
|
248
|
+
// Diff child tables
|
|
249
|
+
for (const child of schema.children) {
|
|
250
|
+
const childTable = child.tableName.replace(/"/g, '');
|
|
251
|
+
const existingChildCols = await getExistingColumns(childTable);
|
|
252
|
+
if (existingChildCols.size === 0)
|
|
253
|
+
continue;
|
|
254
|
+
diffColumns(childTable, existingChildCols, buildChildExpectedCols(child.columns), changes);
|
|
255
|
+
}
|
|
256
|
+
// Diff union branch tables
|
|
257
|
+
for (const union of schema.unions) {
|
|
258
|
+
for (const branch of union.branches) {
|
|
259
|
+
const branchTable = branch.tableName.replace(/"/g, '');
|
|
260
|
+
const existingBranchCols = await getExistingColumns(branchTable);
|
|
261
|
+
if (existingBranchCols.size === 0)
|
|
262
|
+
continue;
|
|
263
|
+
diffColumns(branchTable, existingBranchCols, buildChildExpectedCols(branch.columns), changes);
|
|
264
|
+
}
|
|
265
|
+
}
|
|
266
|
+
}
|
|
267
|
+
// Detect and drop orphaned child/union tables (query table list once)
|
|
268
|
+
let allTableNames = null;
|
|
269
|
+
try {
|
|
270
|
+
const rows = await all(dialect.listTablesQuery);
|
|
271
|
+
allTableNames = rows.map((r) => r.table_name);
|
|
272
|
+
}
|
|
273
|
+
catch { }
|
|
274
|
+
if (allTableNames) {
|
|
275
|
+
for (const schema of tableSchemas) {
|
|
276
|
+
if (schema.columns.length === 0)
|
|
277
|
+
continue;
|
|
278
|
+
const expectedTables = new Set();
|
|
279
|
+
for (const child of schema.children) {
|
|
280
|
+
expectedTables.add(child.tableName.replace(/"/g, ''));
|
|
281
|
+
}
|
|
282
|
+
for (const union of schema.unions) {
|
|
283
|
+
for (const branch of union.branches) {
|
|
284
|
+
expectedTables.add(branch.tableName.replace(/"/g, ''));
|
|
285
|
+
}
|
|
286
|
+
}
|
|
287
|
+
for (const name of allTableNames) {
|
|
288
|
+
if (name.startsWith(schema.collection + '__') && !expectedTables.has(name)) {
|
|
289
|
+
await run(`DROP TABLE IF EXISTS "${name}"`);
|
|
290
|
+
emit('migration', 'drop_table', { table: name });
|
|
291
|
+
}
|
|
292
|
+
}
|
|
293
|
+
}
|
|
294
|
+
}
|
|
295
|
+
if (changes.length > 0) {
|
|
296
|
+
await applyMigrationChanges(changes);
|
|
297
|
+
}
|
|
298
|
+
// Trigger backfill only for genuinely new collections (tables created this startup)
|
|
299
|
+
// Previously this checked ALL empty tables, which caused infinite resync loops
|
|
300
|
+
// for collections that are legitimately empty (e.g. blocks when nobody has blocked)
|
|
301
|
+
if (newCollections.size > 0) {
|
|
302
|
+
const [hasRepos] = await all(`SELECT 1 FROM _repos LIMIT 1`);
|
|
303
|
+
if (hasRepos) {
|
|
304
|
+
await run(`UPDATE _repos SET status = 'pending' WHERE status = 'active'`);
|
|
305
|
+
for (const collection of newCollections) {
|
|
306
|
+
emit('migration', 'new_collection', { collection });
|
|
307
|
+
}
|
|
308
|
+
}
|
|
309
|
+
}
|
|
310
|
+
return changes;
|
|
311
|
+
}
|
|
312
|
+
async function applyMigrationChanges(changes) {
|
|
313
|
+
for (const change of changes) {
|
|
314
|
+
const quotedTable = `"${change.table}"`;
|
|
315
|
+
const quotedColumn = `"${change.column}"`;
|
|
316
|
+
try {
|
|
317
|
+
switch (change.action) {
|
|
318
|
+
case 'add': {
|
|
319
|
+
await run(`ALTER TABLE ${quotedTable} ADD COLUMN ${quotedColumn} ${change.type}`);
|
|
320
|
+
emit('migration', 'add_column', { table: change.table, column: change.column, type: change.type });
|
|
321
|
+
const schema = schemas.get(change.table);
|
|
322
|
+
if (schema?.refColumns.includes(change.column)) {
|
|
323
|
+
const prefix = change.table.replace(/\./g, '_');
|
|
324
|
+
await run(`CREATE INDEX IF NOT EXISTS idx_${prefix}_${change.column} ON ${quotedTable}(${quotedColumn})`);
|
|
325
|
+
}
|
|
326
|
+
break;
|
|
327
|
+
}
|
|
328
|
+
case 'drop':
|
|
329
|
+
await run(`ALTER TABLE ${quotedTable} DROP COLUMN ${quotedColumn}`);
|
|
330
|
+
emit('migration', 'drop_column', { table: change.table, column: change.column });
|
|
331
|
+
break;
|
|
332
|
+
case 'retype':
|
|
333
|
+
await run(`ALTER TABLE ${quotedTable} DROP COLUMN ${quotedColumn}`);
|
|
334
|
+
await run(`ALTER TABLE ${quotedTable} ADD COLUMN ${quotedColumn} ${change.type}`);
|
|
335
|
+
emit('migration', 'retype_column', { table: change.table, column: change.column, type: change.type });
|
|
336
|
+
break;
|
|
337
|
+
}
|
|
338
|
+
}
|
|
339
|
+
catch (err) {
|
|
340
|
+
console.warn(`[migration] failed to ${change.action} column "${change.column}" on "${change.table}": ${err.message}`);
|
|
341
|
+
emit('migration', 'error', {
|
|
342
|
+
action: change.action,
|
|
343
|
+
table: change.table,
|
|
344
|
+
column: change.column,
|
|
345
|
+
error: err.message,
|
|
346
|
+
});
|
|
347
|
+
}
|
|
173
348
|
}
|
|
174
349
|
}
|
|
175
350
|
export async function getCursor(key) {
|
|
176
|
-
const rows = await all(`SELECT value FROM _cursor WHERE key = $1`, key);
|
|
351
|
+
const rows = await all(`SELECT value FROM _cursor WHERE key = $1`, [key]);
|
|
177
352
|
return rows[0]?.value || null;
|
|
178
353
|
}
|
|
179
354
|
export async function setCursor(key, value) {
|
|
180
|
-
await run(`INSERT OR REPLACE INTO _cursor (key, value) VALUES ($1, $2)`, key, value);
|
|
355
|
+
await run(`INSERT OR REPLACE INTO _cursor (key, value) VALUES ($1, $2)`, [key, value]);
|
|
181
356
|
}
|
|
182
357
|
export async function getRepoStatus(did) {
|
|
183
|
-
const rows = await all(`SELECT status FROM _repos WHERE did = $1`, did);
|
|
358
|
+
const rows = await all(`SELECT status FROM _repos WHERE did = $1`, [did]);
|
|
184
359
|
return rows[0]?.status || null;
|
|
185
360
|
}
|
|
186
361
|
export async function setRepoStatus(did, status, rev, opts) {
|
|
187
362
|
if (status === 'active') {
|
|
188
363
|
// Update existing row preserving handle if not provided
|
|
189
|
-
await run(`UPDATE _repos SET status = $1, handle = COALESCE($2, handle), backfilled_at = $3, rev = COALESCE($4, rev), retry_count = 0, retry_after = 0 WHERE did = $5`, status, opts?.handle || null, new Date().toISOString(), rev || null, did);
|
|
364
|
+
await run(`UPDATE _repos SET status = $1, handle = COALESCE($2, handle), backfilled_at = $3, rev = COALESCE($4, rev), retry_count = 0, retry_after = 0 WHERE did = $5`, [status, opts?.handle || null, new Date().toISOString(), rev || null, did]);
|
|
190
365
|
// Insert if row didn't exist yet
|
|
191
|
-
await run(`INSERT OR IGNORE INTO _repos (did, status, handle, backfilled_at, rev, retry_count, retry_after) VALUES ($1, $2, $3, $4, $5, 0, 0)`, did, status, opts?.handle || null, new Date().toISOString(), rev || null);
|
|
366
|
+
await run(`INSERT OR IGNORE INTO _repos (did, status, handle, backfilled_at, rev, retry_count, retry_after) VALUES ($1, $2, $3, $4, $5, 0, 0)`, [did, status, opts?.handle || null, new Date().toISOString(), rev || null]);
|
|
192
367
|
}
|
|
193
368
|
else if (status === 'failed' && opts) {
|
|
194
|
-
await run(`UPDATE _repos SET status = $1, retry_count = $2, retry_after = $3, handle = COALESCE($4, handle) WHERE did = $5`, status, opts.retryCount ?? 0, opts.retryAfter ?? 0, opts.handle || null, did);
|
|
369
|
+
await run(`UPDATE _repos SET status = $1, retry_count = $2, retry_after = $3, handle = COALESCE($4, handle) WHERE did = $5`, [status, opts.retryCount ?? 0, opts.retryAfter ?? 0, opts.handle || null, did]);
|
|
195
370
|
// If row didn't exist yet, insert it
|
|
196
|
-
await run(`INSERT OR IGNORE INTO _repos (did, status, handle, retry_count, retry_after) VALUES ($1, $2, $3, $4, $5)`, did, status, opts.handle || null, opts.retryCount ?? 0, opts.retryAfter ?? 0);
|
|
371
|
+
await run(`INSERT OR IGNORE INTO _repos (did, status, handle, retry_count, retry_after) VALUES ($1, $2, $3, $4, $5)`, [did, status, opts.handle || null, opts.retryCount ?? 0, opts.retryAfter ?? 0]);
|
|
197
372
|
}
|
|
198
373
|
else {
|
|
199
|
-
await run(`UPDATE _repos SET status = $1 WHERE did = $2`, status, did);
|
|
200
|
-
await run(`INSERT OR IGNORE INTO _repos (did, status) VALUES ($1, $2)`, did, status);
|
|
374
|
+
await run(`UPDATE _repos SET status = $1 WHERE did = $2`, [status, did]);
|
|
375
|
+
await run(`INSERT OR IGNORE INTO _repos (did, status) VALUES ($1, $2)`, [did, status]);
|
|
201
376
|
}
|
|
202
377
|
}
|
|
378
|
+
/** Update the handle for a DID if it exists in _repos. */
|
|
379
|
+
export async function updateRepoHandle(did, handle) {
|
|
380
|
+
await run(`UPDATE _repos SET handle = $1 WHERE did = $2`, [handle, did]);
|
|
381
|
+
}
|
|
382
|
+
export async function getRepoRev(did) {
|
|
383
|
+
const rows = await all(`SELECT rev FROM _repos WHERE did = $1`, [did]);
|
|
384
|
+
return rows[0]?.rev ?? null;
|
|
385
|
+
}
|
|
203
386
|
export async function getRepoRetryInfo(did) {
|
|
204
|
-
const rows = await all(`SELECT retry_count, retry_after FROM _repos WHERE did = $1`, did);
|
|
387
|
+
const rows = await all(`SELECT retry_count, retry_after FROM _repos WHERE did = $1`, [did]);
|
|
205
388
|
if (rows.length === 0)
|
|
206
389
|
return null;
|
|
207
390
|
return { retryCount: Number(rows[0].retry_count), retryAfter: Number(rows[0].retry_after) };
|
|
208
391
|
}
|
|
209
392
|
export async function listRetryEligibleRepos(maxRetries) {
|
|
210
393
|
const now = Math.floor(Date.now() / 1000);
|
|
211
|
-
const rows = await all(`SELECT did FROM _repos WHERE status = 'failed' AND retry_after <= $1 AND retry_count < $2`, now, maxRetries);
|
|
394
|
+
const rows = await all(`SELECT did FROM _repos WHERE status = 'failed' AND retry_after <= $1 AND retry_count < $2`, [now, maxRetries]);
|
|
212
395
|
return rows.map((r) => r.did);
|
|
213
396
|
}
|
|
214
397
|
export async function listPendingRepos() {
|
|
215
398
|
const rows = await all(`SELECT did FROM _repos WHERE status = 'pending'`);
|
|
216
399
|
return rows.map((r) => r.did);
|
|
217
400
|
}
|
|
401
|
+
export async function listActiveRepoDids() {
|
|
402
|
+
const rows = await all(`SELECT did FROM _repos WHERE status = 'active'`);
|
|
403
|
+
return rows.map((r) => r.did);
|
|
404
|
+
}
|
|
405
|
+
export async function removeRepo(did) {
|
|
406
|
+
await run(`DELETE FROM _repos WHERE did = $1`, [did]);
|
|
407
|
+
}
|
|
408
|
+
export async function getRepoHandle(did) {
|
|
409
|
+
const rows = await all(`SELECT handle FROM _repos WHERE did = $1`, [did]);
|
|
410
|
+
return rows[0]?.handle ?? null;
|
|
411
|
+
}
|
|
218
412
|
export async function listAllRepoStatuses() {
|
|
219
413
|
return (await all(`SELECT did, status FROM _repos`));
|
|
220
414
|
}
|
|
@@ -228,27 +422,108 @@ export async function listReposPaginated(opts = {}) {
|
|
|
228
422
|
params.push(status);
|
|
229
423
|
}
|
|
230
424
|
if (q) {
|
|
231
|
-
conditions.push(`(did
|
|
425
|
+
conditions.push(`(did ${dialect.ilike} $${paramIdx} OR handle ${dialect.ilike} $${paramIdx})`);
|
|
232
426
|
params.push(`%${q}%`);
|
|
233
427
|
paramIdx++;
|
|
234
428
|
}
|
|
235
429
|
const where = conditions.length ? ' WHERE ' + conditions.join(' AND ') : '';
|
|
236
|
-
const countRows = await all(`SELECT
|
|
430
|
+
const countRows = await all(`SELECT ${dialect.countAsInteger} as total FROM _repos${where}`, params);
|
|
237
431
|
const total = Number(countRows[0]?.total || 0);
|
|
238
|
-
const rows = await all(`SELECT did, handle, status, backfilled_at, rev FROM _repos${where} ORDER BY backfilled_at
|
|
432
|
+
const rows = await all(`SELECT did, handle, status, backfilled_at, rev FROM _repos${where} ORDER BY CASE WHEN backfilled_at IS NULL THEN 1 ELSE 0 END, backfilled_at DESC, did LIMIT $${paramIdx++} OFFSET $${paramIdx++}`, [...params, limit, offset]);
|
|
239
433
|
return { repos: rows, total };
|
|
240
434
|
}
|
|
241
435
|
export async function getCollectionCounts() {
|
|
436
|
+
const entries = [...schemas];
|
|
242
437
|
const counts = {};
|
|
243
|
-
for (const [collection
|
|
244
|
-
|
|
245
|
-
|
|
246
|
-
|
|
438
|
+
for (const [collection] of entries)
|
|
439
|
+
counts[collection] = 0;
|
|
440
|
+
if (entries.length === 0)
|
|
441
|
+
return counts;
|
|
442
|
+
// One round trip instead of one per collection — reads are serialized on a
|
|
443
|
+
// single connection, so N queries means N waits behind whatever else is
|
|
444
|
+
// reading.
|
|
445
|
+
const sql = entries
|
|
446
|
+
.map(([collection, schema]) => `SELECT '${collection.replace(/'/g, "''")}' as collection, ${dialect.countAsInteger} as count FROM ${schema.tableName}`)
|
|
447
|
+
.join(' UNION ALL ');
|
|
448
|
+
const rows = await all(sql);
|
|
449
|
+
for (const row of rows)
|
|
450
|
+
counts[row.collection] = Number(row.count || 0);
|
|
451
|
+
return counts;
|
|
452
|
+
}
|
|
453
|
+
export async function getRepoStatusCounts() {
|
|
454
|
+
const rows = await all(`SELECT status, ${dialect.countAsInteger} as count FROM _repos GROUP BY status`);
|
|
455
|
+
const counts = {};
|
|
456
|
+
for (const row of rows)
|
|
457
|
+
counts[row.status] = Number(row.count);
|
|
247
458
|
return counts;
|
|
248
459
|
}
|
|
460
|
+
export async function getDatabaseSize() {
|
|
461
|
+
if (dialect.supportsSequences) {
|
|
462
|
+
// DuckDB: pragma_database_size returns pre-formatted strings
|
|
463
|
+
const rows = await all('SELECT database_size, memory_usage, memory_limit FROM pragma_database_size()');
|
|
464
|
+
return rows[0] ?? {};
|
|
465
|
+
}
|
|
466
|
+
// SQLite: compute from page_count * page_size
|
|
467
|
+
const pages = await all('SELECT page_count FROM pragma_page_count()');
|
|
468
|
+
const sizes = await all('SELECT page_size FROM pragma_page_size()');
|
|
469
|
+
const pageCount = Number(pages[0]?.page_count ?? 0);
|
|
470
|
+
const pageSize = Number(sizes[0]?.page_size ?? 0);
|
|
471
|
+
const bytes = pageCount * pageSize;
|
|
472
|
+
const mib = (bytes / 1024 / 1024).toFixed(1);
|
|
473
|
+
return { database_size: `${mib} MiB`, memory_usage: 'N/A', memory_limit: 'N/A' };
|
|
474
|
+
}
|
|
475
|
+
export async function getLabelCount(val) {
|
|
476
|
+
const rows = await all(`SELECT ${dialect.countAsInteger} as count FROM _labels WHERE val = $1`, [
|
|
477
|
+
val,
|
|
478
|
+
]);
|
|
479
|
+
return Number(rows[0]?.count || 0);
|
|
480
|
+
}
|
|
481
|
+
export async function deleteLabels(val) {
|
|
482
|
+
const count = await getLabelCount(val);
|
|
483
|
+
await run(`DELETE FROM _labels WHERE val = $1`, [val]);
|
|
484
|
+
return count;
|
|
485
|
+
}
|
|
486
|
+
export async function getRecentRecords(collection, limit) {
|
|
487
|
+
const schema = schemas.get(collection);
|
|
488
|
+
if (!schema)
|
|
489
|
+
return [];
|
|
490
|
+
const rows = await all(`SELECT t.* FROM ${schema.tableName} t JOIN _repos r ON t.did = r.did WHERE t.indexed_at > r.backfilled_at ORDER BY t.indexed_at DESC LIMIT $1`, [limit]);
|
|
491
|
+
return rows;
|
|
492
|
+
}
|
|
249
493
|
export async function getSchemaDump() {
|
|
250
|
-
|
|
251
|
-
|
|
494
|
+
let rows;
|
|
495
|
+
if (dialect.supportsSequences) {
|
|
496
|
+
// DuckDB: use duckdb_tables() for full DDL
|
|
497
|
+
rows = await all(`SELECT sql FROM duckdb_tables() ORDER BY table_name`);
|
|
498
|
+
}
|
|
499
|
+
else {
|
|
500
|
+
// SQLite: use sqlite_master, skip FTS shadow/internal tables
|
|
501
|
+
rows = await all(`SELECT sql FROM sqlite_master WHERE type='table' AND name NOT LIKE 'sqlite_%' AND name NOT LIKE '_fts_%' AND sql IS NOT NULL ORDER BY name`);
|
|
502
|
+
}
|
|
503
|
+
// Normalize indentation and formatting
|
|
504
|
+
return rows
|
|
505
|
+
.map((r) => {
|
|
506
|
+
let sql = r.sql.trim();
|
|
507
|
+
// Remove quotes around column names (SQLite adds them for some columns)
|
|
508
|
+
sql = sql.replace(/\n\s*"(\w+)"/g, '\n$1');
|
|
509
|
+
// Ensure closing paren is on its own line
|
|
510
|
+
sql = sql.replace(/([^(\s])\)$/, '$1\n)');
|
|
511
|
+
// Normalize leading-comma columns added by ALTER TABLE into trailing commas
|
|
512
|
+
sql = sql.replace(/\n\s*,\s*/g, ',\n');
|
|
513
|
+
// Split into lines and re-indent consistently
|
|
514
|
+
const lines = sql.split('\n').map((l) => l.trim());
|
|
515
|
+
sql = lines
|
|
516
|
+
.map((line, i) => {
|
|
517
|
+
if (i === 0)
|
|
518
|
+
return line; // CREATE TABLE line
|
|
519
|
+
if (line.startsWith(')'))
|
|
520
|
+
return ')'; // closing paren at top level
|
|
521
|
+
return ' ' + line; // indent columns
|
|
522
|
+
})
|
|
523
|
+
.join('\n');
|
|
524
|
+
return sql + ';';
|
|
525
|
+
})
|
|
526
|
+
.join('\n\n');
|
|
252
527
|
}
|
|
253
528
|
export function buildInsertOp(collection, uri, cid, authorDid, record) {
|
|
254
529
|
const schema = schemas.get(collection);
|
|
@@ -267,12 +542,12 @@ export function buildInsertOp(collection, uri, cid, authorDid, record) {
|
|
|
267
542
|
else if (col.originalName.endsWith('__cid') && record[col.originalName.replace('__cid', '')]) {
|
|
268
543
|
rawValue = record[col.originalName.replace('__cid', '')].cid;
|
|
269
544
|
}
|
|
270
|
-
colNames.push(col.name);
|
|
545
|
+
colNames.push(q(col.name));
|
|
271
546
|
placeholders.push(`$${paramIdx++}`);
|
|
272
547
|
if (rawValue === undefined || rawValue === null) {
|
|
273
548
|
values.push(null);
|
|
274
549
|
}
|
|
275
|
-
else if (col.
|
|
550
|
+
else if (col.isJson) {
|
|
276
551
|
values.push(JSON.stringify(rawValue));
|
|
277
552
|
}
|
|
278
553
|
else {
|
|
@@ -287,34 +562,34 @@ export async function insertRecord(collection, uri, cid, authorDid, record) {
|
|
|
287
562
|
if (!schema)
|
|
288
563
|
throw new Error(`Unknown collection: ${collection}`);
|
|
289
564
|
const { sql, params } = buildInsertOp(collection, uri, cid, authorDid, record);
|
|
290
|
-
await run(sql,
|
|
565
|
+
await run(sql, params);
|
|
291
566
|
// Insert child table rows
|
|
292
567
|
for (const child of schema.children) {
|
|
293
568
|
const items = record[child.fieldName];
|
|
294
569
|
if (!Array.isArray(items))
|
|
295
570
|
continue;
|
|
296
571
|
// Delete existing child rows (handles INSERT OR REPLACE on main table)
|
|
297
|
-
await run(`DELETE FROM ${child.tableName} WHERE parent_uri = $1`, uri);
|
|
572
|
+
await run(`DELETE FROM ${child.tableName} WHERE parent_uri = $1`, [uri]);
|
|
298
573
|
for (const item of items) {
|
|
299
574
|
const colNames = ['parent_uri', 'parent_did'];
|
|
300
575
|
const placeholders = ['$1', '$2'];
|
|
301
576
|
const values = [uri, authorDid];
|
|
302
577
|
let idx = 3;
|
|
303
578
|
for (const col of child.columns) {
|
|
304
|
-
colNames.push(col.name);
|
|
579
|
+
colNames.push(q(col.name));
|
|
305
580
|
placeholders.push(`$${idx++}`);
|
|
306
581
|
const raw = item[col.originalName];
|
|
307
582
|
if (raw === undefined || raw === null) {
|
|
308
583
|
values.push(null);
|
|
309
584
|
}
|
|
310
|
-
else if (col.
|
|
585
|
+
else if (col.isJson) {
|
|
311
586
|
values.push(JSON.stringify(raw));
|
|
312
587
|
}
|
|
313
588
|
else {
|
|
314
589
|
values.push(raw);
|
|
315
590
|
}
|
|
316
591
|
}
|
|
317
|
-
await run(`INSERT INTO ${child.tableName} (${colNames.join(', ')}) VALUES (${placeholders.join(', ')})`,
|
|
592
|
+
await run(`INSERT INTO ${child.tableName} (${colNames.join(', ')}) VALUES (${placeholders.join(', ')})`, values);
|
|
318
593
|
}
|
|
319
594
|
}
|
|
320
595
|
// Insert union branch rows
|
|
@@ -327,7 +602,7 @@ export async function insertRecord(collection, uri, cid, authorDid, record) {
|
|
|
327
602
|
continue;
|
|
328
603
|
// Delete existing branch rows (handles INSERT OR REPLACE)
|
|
329
604
|
for (const b of union.branches) {
|
|
330
|
-
await run(`DELETE FROM ${b.tableName} WHERE parent_uri = $1`, uri);
|
|
605
|
+
await run(`DELETE FROM ${b.tableName} WHERE parent_uri = $1`, [uri]);
|
|
331
606
|
}
|
|
332
607
|
if (branch.isArray && branch.arrayField) {
|
|
333
608
|
// Array branch (e.g., embed.images) — insert one row per array item
|
|
@@ -340,20 +615,20 @@ export async function insertRecord(collection, uri, cid, authorDid, record) {
|
|
|
340
615
|
const values = [uri, authorDid];
|
|
341
616
|
let idx = 3;
|
|
342
617
|
for (const col of branch.columns) {
|
|
343
|
-
colNames.push(col.name);
|
|
618
|
+
colNames.push(q(col.name));
|
|
344
619
|
placeholders.push(`$${idx++}`);
|
|
345
620
|
const raw = item[col.originalName];
|
|
346
621
|
if (raw === undefined || raw === null) {
|
|
347
622
|
values.push(null);
|
|
348
623
|
}
|
|
349
|
-
else if (col.
|
|
624
|
+
else if (col.isJson) {
|
|
350
625
|
values.push(JSON.stringify(raw));
|
|
351
626
|
}
|
|
352
627
|
else {
|
|
353
628
|
values.push(raw);
|
|
354
629
|
}
|
|
355
630
|
}
|
|
356
|
-
await run(`INSERT INTO ${branch.tableName} (${colNames.join(', ')}) VALUES (${placeholders.join(', ')})`,
|
|
631
|
+
await run(`INSERT INTO ${branch.tableName} (${colNames.join(', ')}) VALUES (${placeholders.join(', ')})`, values);
|
|
357
632
|
}
|
|
358
633
|
}
|
|
359
634
|
else {
|
|
@@ -364,22 +639,24 @@ export async function insertRecord(collection, uri, cid, authorDid, record) {
|
|
|
364
639
|
const values = [uri, authorDid];
|
|
365
640
|
let idx = 3;
|
|
366
641
|
for (const col of branch.columns) {
|
|
367
|
-
colNames.push(col.name);
|
|
642
|
+
colNames.push(q(col.name));
|
|
368
643
|
placeholders.push(`$${idx++}`);
|
|
369
644
|
const raw = branchData[col.originalName];
|
|
370
645
|
if (raw === undefined || raw === null) {
|
|
371
646
|
values.push(null);
|
|
372
647
|
}
|
|
373
|
-
else if (col.
|
|
648
|
+
else if (col.isJson) {
|
|
374
649
|
values.push(JSON.stringify(raw));
|
|
375
650
|
}
|
|
376
651
|
else {
|
|
377
652
|
values.push(raw);
|
|
378
653
|
}
|
|
379
654
|
}
|
|
380
|
-
await run(`INSERT INTO ${branch.tableName} (${colNames.join(', ')}) VALUES (${placeholders.join(', ')})`,
|
|
655
|
+
await run(`INSERT INTO ${branch.tableName} (${colNames.join(', ')}) VALUES (${placeholders.join(', ')})`, values);
|
|
381
656
|
}
|
|
382
657
|
}
|
|
658
|
+
// Incrementally update FTS index for this record
|
|
659
|
+
await updateFtsRecord(collection, uri);
|
|
383
660
|
}
|
|
384
661
|
/** Extract branch data from a union value, handling wrapper properties */
|
|
385
662
|
function resolveBranchData(unionValue, branch) {
|
|
@@ -394,32 +671,41 @@ export async function deleteRecord(collection, uri) {
|
|
|
394
671
|
const schema = schemas.get(collection);
|
|
395
672
|
if (!schema)
|
|
396
673
|
return;
|
|
674
|
+
// Remove from FTS index before deleting the record data
|
|
675
|
+
await deleteFtsRecord(collection, uri);
|
|
397
676
|
for (const child of schema.children) {
|
|
398
|
-
await run(`DELETE FROM ${child.tableName} WHERE parent_uri = $1`, uri);
|
|
677
|
+
await run(`DELETE FROM ${child.tableName} WHERE parent_uri = $1`, [uri]);
|
|
399
678
|
}
|
|
400
679
|
for (const union of schema.unions) {
|
|
401
680
|
for (const branch of union.branches) {
|
|
402
|
-
await run(`DELETE FROM ${branch.tableName} WHERE parent_uri = $1`, uri);
|
|
681
|
+
await run(`DELETE FROM ${branch.tableName} WHERE parent_uri = $1`, [uri]);
|
|
403
682
|
}
|
|
404
683
|
}
|
|
405
|
-
await run(`DELETE FROM ${schema.tableName} WHERE uri = $1`, uri);
|
|
684
|
+
await run(`DELETE FROM ${schema.tableName} WHERE uri = $1`, [uri]);
|
|
406
685
|
}
|
|
407
686
|
export async function insertLabels(labels) {
|
|
408
687
|
if (labels.length === 0)
|
|
409
688
|
return;
|
|
410
689
|
for (const label of labels) {
|
|
411
|
-
// Skip if an active (non-negated, non-expired, not-superseded-by-negation) label already exists
|
|
412
|
-
const existing = await all(`SELECT 1 FROM _labels l1 WHERE l1.src = $1 AND l1.uri = $2 AND l1.val = $3 AND l1.neg = false AND (l1.exp IS NULL OR l1.exp > CURRENT_TIMESTAMP) AND NOT EXISTS (SELECT 1 FROM _labels l2 WHERE l2.uri = l1.uri AND l2.val = l1.val AND l2.neg = true AND l2.id > l1.id) LIMIT 1`, label.src, label.uri, label.val);
|
|
690
|
+
// Skip if an active (non-negated, non-expired, not-superseded-by-negation) label already exists for this src+uri+val
|
|
691
|
+
const existing = await all(`SELECT 1 FROM _labels l1 WHERE l1.src = $1 AND l1.uri = $2 AND l1.val = $3 AND l1.neg = false AND (l1.exp IS NULL OR l1.exp > CURRENT_TIMESTAMP) AND NOT EXISTS (SELECT 1 FROM _labels l2 WHERE l2.uri = l1.uri AND l2.val = l1.val AND l2.neg = true AND l2.id > l1.id) LIMIT 1`, [label.src, label.uri, label.val]);
|
|
413
692
|
if (!label.neg && existing.length > 0)
|
|
414
693
|
continue;
|
|
415
|
-
await run(`INSERT INTO _labels (src, uri, val, neg, cts, exp) VALUES ($1, $2, $3, $4, $5, $6)`,
|
|
694
|
+
await run(`INSERT INTO _labels (src, uri, val, neg, cts, exp) VALUES ($1, $2, $3, $4, $5, $6)`, [
|
|
695
|
+
label.src,
|
|
696
|
+
label.uri,
|
|
697
|
+
label.val,
|
|
698
|
+
label.neg || false,
|
|
699
|
+
label.cts || new Date().toISOString(),
|
|
700
|
+
label.exp || null,
|
|
701
|
+
]);
|
|
416
702
|
}
|
|
417
703
|
}
|
|
418
704
|
export async function queryLabelsForUris(uris) {
|
|
419
705
|
if (uris.length === 0)
|
|
420
706
|
return new Map();
|
|
421
707
|
const placeholders = uris.map((_, i) => `$${i + 1}`).join(',');
|
|
422
|
-
const rows = await all(`SELECT src, uri, val, neg, cts, exp FROM _labels l1 WHERE uri IN (${placeholders}) AND (exp IS NULL OR exp > CURRENT_TIMESTAMP) AND neg = false AND NOT EXISTS (SELECT 1 FROM _labels l2 WHERE l2.uri = l1.uri AND l2.val = l1.val AND l2.neg = true AND l2.id > l1.id)`,
|
|
708
|
+
const rows = await all(`SELECT src, uri, val, neg, cts, exp FROM _labels l1 WHERE uri IN (${placeholders}) AND (exp IS NULL OR exp > CURRENT_TIMESTAMP) AND neg = false AND NOT EXISTS (SELECT 1 FROM _labels l2 WHERE l2.uri = l1.uri AND l2.val = l1.val AND l2.neg = true AND l2.id > l1.id) GROUP BY uri, val`, uris);
|
|
423
709
|
const result = new Map();
|
|
424
710
|
for (const row of rows) {
|
|
425
711
|
const key = row.uri;
|
|
@@ -429,9 +715,9 @@ export async function queryLabelsForUris(uris) {
|
|
|
429
715
|
src: row.src,
|
|
430
716
|
uri: row.uri,
|
|
431
717
|
val: row.val,
|
|
432
|
-
neg: row.neg,
|
|
718
|
+
neg: !!row.neg,
|
|
433
719
|
cts: normalizeValue(row.cts),
|
|
434
|
-
|
|
720
|
+
...(row.exp ? { exp: String(row.exp) } : {}),
|
|
435
721
|
});
|
|
436
722
|
}
|
|
437
723
|
return result;
|
|
@@ -452,246 +738,209 @@ export async function bulkInsertRecords(records) {
|
|
|
452
738
|
if (!schema)
|
|
453
739
|
continue;
|
|
454
740
|
const stagingTable = `_staging_${collection.replace(/\./g, '_')}`;
|
|
455
|
-
const allCols = ['uri', 'cid', 'did', 'indexed_at', ...schema.columns.map((c) => c.name)];
|
|
741
|
+
const allCols = ['uri', 'cid', 'did', 'indexed_at', ...schema.columns.map((c) => q(c.name))];
|
|
456
742
|
const colDefs = [
|
|
457
743
|
'uri TEXT',
|
|
458
744
|
'cid TEXT',
|
|
459
745
|
'did TEXT',
|
|
460
746
|
'indexed_at TEXT',
|
|
461
|
-
...schema.columns.map((c) =>
|
|
747
|
+
...schema.columns.map((c) => {
|
|
748
|
+
const t = c.sqlType;
|
|
749
|
+
// Use TEXT for timestamp columns in staging (will cast on merge)
|
|
750
|
+
return `${q(c.name)} ${t === 'TIMESTAMP' || t === 'TIMESTAMPTZ' ? 'TEXT' : t}`;
|
|
751
|
+
}),
|
|
462
752
|
];
|
|
463
|
-
|
|
464
|
-
await
|
|
465
|
-
|
|
466
|
-
|
|
467
|
-
|
|
468
|
-
|
|
753
|
+
await port.execute(`DROP TABLE IF EXISTS ${stagingTable}`, []);
|
|
754
|
+
await port.execute(`CREATE TABLE ${stagingTable} (${colDefs.join(', ')})`, []);
|
|
755
|
+
const inserter = await port.createBulkInserter(stagingTable, allCols);
|
|
756
|
+
const now = new Date().toISOString();
|
|
757
|
+
for (const rec of recs) {
|
|
758
|
+
try {
|
|
759
|
+
const values = [rec.uri, rec.cid, rec.did, now];
|
|
760
|
+
for (const col of schema.columns) {
|
|
761
|
+
values.push(resolveColumnValue(col, rec.record));
|
|
762
|
+
}
|
|
763
|
+
inserter.append(values);
|
|
764
|
+
inserted++;
|
|
765
|
+
}
|
|
766
|
+
catch {
|
|
767
|
+
// Skip bad records
|
|
768
|
+
}
|
|
769
|
+
}
|
|
770
|
+
await inserter.close();
|
|
771
|
+
// Merge into target, filtering rows that would violate NOT NULL
|
|
772
|
+
const selectCols = allCols.map((name) => {
|
|
773
|
+
const col = schema.columns.find((c) => q(c.name) === name);
|
|
774
|
+
if (name === 'indexed_at' || (col && (col.sqlType === 'TIMESTAMP' || col.sqlType === 'TIMESTAMPTZ'))) {
|
|
775
|
+
return `${dialect.tryCastTimestamp(name)} AS ${name}`;
|
|
776
|
+
}
|
|
777
|
+
return name;
|
|
778
|
+
});
|
|
779
|
+
const notNullChecks = ['uri IS NOT NULL', 'did IS NOT NULL'];
|
|
780
|
+
for (const col of schema.columns) {
|
|
781
|
+
if (col.notNull) {
|
|
782
|
+
if (col.sqlType === 'TIMESTAMP' || col.sqlType === 'TIMESTAMPTZ') {
|
|
783
|
+
notNullChecks.push(`${dialect.tryCastTimestamp(q(col.name))} IS NOT NULL`);
|
|
784
|
+
}
|
|
785
|
+
else {
|
|
786
|
+
notNullChecks.push(`${q(col.name)} IS NOT NULL`);
|
|
787
|
+
}
|
|
788
|
+
}
|
|
789
|
+
}
|
|
790
|
+
const whereClause = notNullChecks.length ? ` WHERE ${notNullChecks.join(' AND ')}` : '';
|
|
791
|
+
await port.execute(`INSERT OR REPLACE INTO ${schema.tableName} (${allCols.join(', ')}) SELECT ${selectCols.join(', ')} FROM ${stagingTable}${whereClause}`, []);
|
|
792
|
+
await port.execute(`DROP TABLE ${stagingTable}`, []);
|
|
793
|
+
// Populate child tables
|
|
794
|
+
for (const child of schema.children) {
|
|
795
|
+
const childStagingTable = `_staging_${collection.replace(/\./g, '_')}__${child.fieldName}`;
|
|
796
|
+
const childColDefs = [
|
|
797
|
+
'parent_uri TEXT',
|
|
798
|
+
'parent_did TEXT',
|
|
799
|
+
...child.columns.map((c) => {
|
|
800
|
+
const t = c.sqlType;
|
|
801
|
+
return `${q(c.name)} ${t === 'TIMESTAMP' || t === 'TIMESTAMPTZ' ? 'TEXT' : t}`;
|
|
802
|
+
}),
|
|
803
|
+
];
|
|
804
|
+
const childAllCols = ['parent_uri', 'parent_did', ...child.columns.map((c) => q(c.name))];
|
|
805
|
+
await port.execute(`DROP TABLE IF EXISTS ${childStagingTable}`, []);
|
|
806
|
+
await port.execute(`CREATE TABLE ${childStagingTable} (${childColDefs.join(', ')})`, []);
|
|
807
|
+
const childInserter = await port.createBulkInserter(childStagingTable, childAllCols);
|
|
469
808
|
for (const rec of recs) {
|
|
470
|
-
|
|
471
|
-
|
|
472
|
-
|
|
473
|
-
|
|
474
|
-
|
|
475
|
-
|
|
476
|
-
|
|
477
|
-
|
|
478
|
-
rawValue = rawValue.uri;
|
|
479
|
-
}
|
|
480
|
-
else if (col.originalName.endsWith('__cid') && rec.record[col.originalName.replace('__cid', '')]) {
|
|
481
|
-
rawValue = rec.record[col.originalName.replace('__cid', '')].cid;
|
|
482
|
-
}
|
|
483
|
-
if (rawValue === undefined || rawValue === null) {
|
|
484
|
-
appender.appendNull();
|
|
485
|
-
}
|
|
486
|
-
else if (col.duckdbType === 'JSON') {
|
|
487
|
-
appender.appendVarchar(JSON.stringify(rawValue));
|
|
488
|
-
}
|
|
489
|
-
else if (col.duckdbType === 'INTEGER') {
|
|
490
|
-
appender.appendInteger(typeof rawValue === 'number' ? rawValue : parseInt(rawValue));
|
|
491
|
-
}
|
|
492
|
-
else if (col.duckdbType === 'BOOLEAN') {
|
|
493
|
-
appender.appendBoolean(!!rawValue);
|
|
494
|
-
}
|
|
495
|
-
else {
|
|
496
|
-
appender.appendVarchar(String(rawValue));
|
|
809
|
+
const items = rec.record[child.fieldName];
|
|
810
|
+
if (!Array.isArray(items))
|
|
811
|
+
continue;
|
|
812
|
+
for (const item of items) {
|
|
813
|
+
try {
|
|
814
|
+
const values = [rec.uri, rec.did];
|
|
815
|
+
for (const col of child.columns) {
|
|
816
|
+
values.push(resolveRawColumnValue(col, item));
|
|
497
817
|
}
|
|
818
|
+
childInserter.append(values);
|
|
819
|
+
}
|
|
820
|
+
catch {
|
|
821
|
+
// Skip bad items
|
|
498
822
|
}
|
|
499
|
-
appender.endRow();
|
|
500
|
-
inserted++;
|
|
501
|
-
}
|
|
502
|
-
catch {
|
|
503
|
-
// Skip bad records
|
|
504
823
|
}
|
|
505
824
|
}
|
|
506
|
-
|
|
507
|
-
|
|
508
|
-
|
|
509
|
-
|
|
510
|
-
|
|
511
|
-
|
|
512
|
-
|
|
825
|
+
await childInserter.close();
|
|
826
|
+
// Delete existing child rows for these URIs, then merge staging
|
|
827
|
+
const uriPlaceholders = recs.map((_, i) => `$${i + 1}`).join(',');
|
|
828
|
+
await port.execute(`DELETE FROM ${child.tableName} WHERE parent_uri IN (${uriPlaceholders})`, recs.map((r) => r.uri));
|
|
829
|
+
const childSelectCols = childAllCols.map((name) => {
|
|
830
|
+
const col = child.columns.find((c) => q(c.name) === name);
|
|
831
|
+
if (col && (col.sqlType === 'TIMESTAMP' || col.sqlType === 'TIMESTAMPTZ')) {
|
|
832
|
+
return `${dialect.tryCastTimestamp(name)} AS ${name}`;
|
|
513
833
|
}
|
|
514
834
|
return name;
|
|
515
835
|
});
|
|
516
|
-
|
|
517
|
-
|
|
518
|
-
|
|
519
|
-
|
|
520
|
-
|
|
521
|
-
|
|
522
|
-
|
|
523
|
-
|
|
524
|
-
notNullChecks.push(`${col.name} IS NOT NULL`);
|
|
525
|
-
}
|
|
526
|
-
}
|
|
527
|
-
}
|
|
528
|
-
const whereClause = notNullChecks.length ? ` WHERE ${notNullChecks.join(' AND ')}` : '';
|
|
529
|
-
await con.run(`INSERT OR REPLACE INTO ${schema.tableName} (${allCols.join(', ')}) SELECT ${selectCols.join(', ')} FROM ${stagingTable}${whereClause}`);
|
|
530
|
-
await con.run(`DROP TABLE ${stagingTable}`);
|
|
531
|
-
// Populate child tables
|
|
532
|
-
for (const child of schema.children) {
|
|
533
|
-
const childStagingTable = `_staging_${collection.replace(/\./g, '_')}__${child.fieldName}`;
|
|
534
|
-
const childColDefs = [
|
|
836
|
+
await port.execute(`INSERT INTO ${child.tableName} (${childAllCols.join(', ')}) SELECT ${childSelectCols.join(', ')} FROM ${childStagingTable} WHERE parent_uri IS NOT NULL`, []);
|
|
837
|
+
await port.execute(`DROP TABLE ${childStagingTable}`, []);
|
|
838
|
+
}
|
|
839
|
+
// Populate union branch tables
|
|
840
|
+
for (const union of schema.unions) {
|
|
841
|
+
for (const branch of union.branches) {
|
|
842
|
+
const branchStagingTable = `_staging_${collection.replace(/\./g, '_')}__${toSnakeCase(union.fieldName)}_${branch.branchName}`;
|
|
843
|
+
const branchColDefs = [
|
|
535
844
|
'parent_uri TEXT',
|
|
536
845
|
'parent_did TEXT',
|
|
537
|
-
...
|
|
846
|
+
...branch.columns.map((c) => {
|
|
847
|
+
const t = c.sqlType;
|
|
848
|
+
return `${q(c.name)} ${t === 'TIMESTAMP' || t === 'TIMESTAMPTZ' ? 'TEXT' : t}`;
|
|
849
|
+
}),
|
|
538
850
|
];
|
|
539
|
-
const
|
|
540
|
-
await
|
|
541
|
-
await
|
|
542
|
-
const
|
|
851
|
+
const branchAllCols = ['parent_uri', 'parent_did', ...branch.columns.map((c) => q(c.name))];
|
|
852
|
+
await port.execute(`DROP TABLE IF EXISTS ${branchStagingTable}`, []);
|
|
853
|
+
await port.execute(`CREATE TABLE ${branchStagingTable} (${branchColDefs.join(', ')})`, []);
|
|
854
|
+
const branchInserter = await port.createBulkInserter(branchStagingTable, branchAllCols);
|
|
543
855
|
for (const rec of recs) {
|
|
544
|
-
const
|
|
545
|
-
if (!
|
|
856
|
+
const unionValue = rec.record[union.fieldName];
|
|
857
|
+
if (!unionValue || typeof unionValue !== 'object')
|
|
546
858
|
continue;
|
|
547
|
-
|
|
548
|
-
|
|
549
|
-
|
|
550
|
-
|
|
551
|
-
|
|
552
|
-
const rawValue = item[col.originalName];
|
|
553
|
-
if (rawValue === undefined || rawValue === null) {
|
|
554
|
-
childAppender.appendNull();
|
|
555
|
-
}
|
|
556
|
-
else if (col.duckdbType === 'JSON') {
|
|
557
|
-
childAppender.appendVarchar(JSON.stringify(rawValue));
|
|
558
|
-
}
|
|
559
|
-
else if (col.duckdbType === 'INTEGER') {
|
|
560
|
-
childAppender.appendInteger(typeof rawValue === 'number' ? rawValue : parseInt(rawValue));
|
|
561
|
-
}
|
|
562
|
-
else if (col.duckdbType === 'BOOLEAN') {
|
|
563
|
-
childAppender.appendBoolean(!!rawValue);
|
|
564
|
-
}
|
|
565
|
-
else {
|
|
566
|
-
childAppender.appendVarchar(String(rawValue));
|
|
567
|
-
}
|
|
568
|
-
}
|
|
569
|
-
childAppender.endRow();
|
|
570
|
-
}
|
|
571
|
-
catch {
|
|
572
|
-
// Skip bad items
|
|
573
|
-
}
|
|
574
|
-
}
|
|
575
|
-
}
|
|
576
|
-
childAppender.flushSync();
|
|
577
|
-
childAppender.closeSync();
|
|
578
|
-
// Delete existing child rows for these URIs, then merge staging
|
|
579
|
-
const uriPlaceholders = recs.map((_, i) => `$${i + 1}`).join(',');
|
|
580
|
-
const delStmt = await con.prepare(`DELETE FROM ${child.tableName} WHERE parent_uri IN (${uriPlaceholders})`);
|
|
581
|
-
bindParams(delStmt, recs.map((r) => r.uri));
|
|
582
|
-
await delStmt.run();
|
|
583
|
-
const childSelectCols = childAllCols.map((name) => {
|
|
584
|
-
const col = child.columns.find((c) => c.name === name);
|
|
585
|
-
if (col && col.duckdbType === 'TIMESTAMP')
|
|
586
|
-
return `TRY_CAST(${name} AS TIMESTAMP) AS ${name}`;
|
|
587
|
-
return name;
|
|
588
|
-
});
|
|
589
|
-
await con.run(`INSERT INTO ${child.tableName} (${childAllCols.join(', ')}) SELECT ${childSelectCols.join(', ')} FROM ${childStagingTable} WHERE parent_uri IS NOT NULL`);
|
|
590
|
-
await con.run(`DROP TABLE ${childStagingTable}`);
|
|
591
|
-
}
|
|
592
|
-
// Populate union branch tables
|
|
593
|
-
for (const union of schema.unions) {
|
|
594
|
-
for (const branch of union.branches) {
|
|
595
|
-
const branchStagingTable = `_staging_${collection.replace(/\./g, '_')}__${toSnakeCase(union.fieldName)}_${branch.branchName}`;
|
|
596
|
-
const branchColDefs = [
|
|
597
|
-
'parent_uri TEXT',
|
|
598
|
-
'parent_did TEXT',
|
|
599
|
-
...branch.columns.map((c) => `${c.name} ${c.duckdbType === 'TIMESTAMP' ? 'TEXT' : c.duckdbType}`),
|
|
600
|
-
];
|
|
601
|
-
const branchAllCols = ['parent_uri', 'parent_did', ...branch.columns.map((c) => c.name)];
|
|
602
|
-
await con.run(`DROP TABLE IF EXISTS ${branchStagingTable}`);
|
|
603
|
-
await con.run(`CREATE TABLE ${branchStagingTable} (${branchColDefs.join(', ')})`);
|
|
604
|
-
const branchAppender = await con.createAppender(branchStagingTable);
|
|
605
|
-
for (const rec of recs) {
|
|
606
|
-
const unionValue = rec.record[union.fieldName];
|
|
607
|
-
if (!unionValue || typeof unionValue !== 'object')
|
|
608
|
-
continue;
|
|
609
|
-
if (unionValue.$type !== branch.type)
|
|
859
|
+
if (unionValue.$type !== branch.type)
|
|
860
|
+
continue;
|
|
861
|
+
if (branch.isArray && branch.arrayField) {
|
|
862
|
+
const items = resolveBranchData(unionValue, branch)[branch.arrayField];
|
|
863
|
+
if (!Array.isArray(items))
|
|
610
864
|
continue;
|
|
611
|
-
|
|
612
|
-
const items = resolveBranchData(unionValue, branch)[branch.arrayField];
|
|
613
|
-
if (!Array.isArray(items))
|
|
614
|
-
continue;
|
|
615
|
-
for (const item of items) {
|
|
616
|
-
try {
|
|
617
|
-
branchAppender.appendVarchar(rec.uri);
|
|
618
|
-
branchAppender.appendVarchar(rec.did);
|
|
619
|
-
for (const col of branch.columns) {
|
|
620
|
-
const rawValue = item[col.originalName];
|
|
621
|
-
if (rawValue === undefined || rawValue === null) {
|
|
622
|
-
branchAppender.appendNull();
|
|
623
|
-
}
|
|
624
|
-
else if (col.duckdbType === 'JSON') {
|
|
625
|
-
branchAppender.appendVarchar(JSON.stringify(rawValue));
|
|
626
|
-
}
|
|
627
|
-
else if (col.duckdbType === 'INTEGER') {
|
|
628
|
-
branchAppender.appendInteger(typeof rawValue === 'number' ? rawValue : parseInt(rawValue));
|
|
629
|
-
}
|
|
630
|
-
else if (col.duckdbType === 'BOOLEAN') {
|
|
631
|
-
branchAppender.appendBoolean(!!rawValue);
|
|
632
|
-
}
|
|
633
|
-
else {
|
|
634
|
-
branchAppender.appendVarchar(String(rawValue));
|
|
635
|
-
}
|
|
636
|
-
}
|
|
637
|
-
branchAppender.endRow();
|
|
638
|
-
}
|
|
639
|
-
catch {
|
|
640
|
-
// Skip bad items
|
|
641
|
-
}
|
|
642
|
-
}
|
|
643
|
-
}
|
|
644
|
-
else {
|
|
865
|
+
for (const item of items) {
|
|
645
866
|
try {
|
|
646
|
-
const
|
|
647
|
-
branchAppender.appendVarchar(rec.uri);
|
|
648
|
-
branchAppender.appendVarchar(rec.did);
|
|
867
|
+
const values = [rec.uri, rec.did];
|
|
649
868
|
for (const col of branch.columns) {
|
|
650
|
-
|
|
651
|
-
if (rawValue === undefined || rawValue === null) {
|
|
652
|
-
branchAppender.appendNull();
|
|
653
|
-
}
|
|
654
|
-
else if (col.duckdbType === 'JSON') {
|
|
655
|
-
branchAppender.appendVarchar(JSON.stringify(rawValue));
|
|
656
|
-
}
|
|
657
|
-
else if (col.duckdbType === 'INTEGER') {
|
|
658
|
-
branchAppender.appendInteger(typeof rawValue === 'number' ? rawValue : parseInt(rawValue));
|
|
659
|
-
}
|
|
660
|
-
else if (col.duckdbType === 'BOOLEAN') {
|
|
661
|
-
branchAppender.appendBoolean(!!rawValue);
|
|
662
|
-
}
|
|
663
|
-
else {
|
|
664
|
-
branchAppender.appendVarchar(String(rawValue));
|
|
665
|
-
}
|
|
869
|
+
values.push(resolveRawColumnValue(col, item));
|
|
666
870
|
}
|
|
667
|
-
|
|
871
|
+
branchInserter.append(values);
|
|
668
872
|
}
|
|
669
873
|
catch {
|
|
670
|
-
// Skip bad
|
|
874
|
+
// Skip bad items
|
|
671
875
|
}
|
|
672
876
|
}
|
|
673
877
|
}
|
|
674
|
-
|
|
675
|
-
|
|
676
|
-
|
|
677
|
-
|
|
678
|
-
|
|
679
|
-
|
|
680
|
-
|
|
681
|
-
|
|
682
|
-
|
|
683
|
-
|
|
684
|
-
|
|
685
|
-
|
|
686
|
-
}
|
|
687
|
-
await con.run(`INSERT INTO ${branch.tableName} (${branchAllCols.join(', ')}) SELECT ${branchSelectCols.join(', ')} FROM ${branchStagingTable} WHERE parent_uri IS NOT NULL`);
|
|
688
|
-
await con.run(`DROP TABLE ${branchStagingTable}`);
|
|
878
|
+
else {
|
|
879
|
+
try {
|
|
880
|
+
const branchData = resolveBranchData(unionValue, branch);
|
|
881
|
+
const values = [rec.uri, rec.did];
|
|
882
|
+
for (const col of branch.columns) {
|
|
883
|
+
values.push(resolveRawColumnValue(col, branchData));
|
|
884
|
+
}
|
|
885
|
+
branchInserter.append(values);
|
|
886
|
+
}
|
|
887
|
+
catch {
|
|
888
|
+
// Skip bad records
|
|
889
|
+
}
|
|
890
|
+
}
|
|
689
891
|
}
|
|
892
|
+
await branchInserter.close();
|
|
893
|
+
// Delete existing branch rows for these URIs, then merge staging
|
|
894
|
+
const uriPlaceholders = recs.map((_, i) => `$${i + 1}`).join(',');
|
|
895
|
+
await port.execute(`DELETE FROM ${branch.tableName} WHERE parent_uri IN (${uriPlaceholders})`, recs.map((r) => r.uri));
|
|
896
|
+
const branchSelectCols = branchAllCols.map((name) => {
|
|
897
|
+
const col = branch.columns.find((c) => q(c.name) === name);
|
|
898
|
+
if (col && (col.sqlType === 'TIMESTAMP' || col.sqlType === 'TIMESTAMPTZ')) {
|
|
899
|
+
return `${dialect.tryCastTimestamp(name)} AS ${name}`;
|
|
900
|
+
}
|
|
901
|
+
return name;
|
|
902
|
+
});
|
|
903
|
+
await port.execute(`INSERT INTO ${branch.tableName} (${branchAllCols.join(', ')}) SELECT ${branchSelectCols.join(', ')} FROM ${branchStagingTable} WHERE parent_uri IS NOT NULL`, []);
|
|
904
|
+
await port.execute(`DROP TABLE ${branchStagingTable}`, []);
|
|
690
905
|
}
|
|
691
|
-
}
|
|
906
|
+
}
|
|
692
907
|
}
|
|
693
908
|
return inserted;
|
|
694
909
|
}
|
|
910
|
+
/** Extract a column value from a record, handling strongRef expansion and type coercion for bulk insert */
|
|
911
|
+
function resolveColumnValue(col, record) {
|
|
912
|
+
let rawValue = record[col.originalName];
|
|
913
|
+
if (rawValue && typeof rawValue === 'object' && col.name.endsWith('_uri') && col.isRef) {
|
|
914
|
+
rawValue = rawValue.uri;
|
|
915
|
+
}
|
|
916
|
+
else if (col.originalName.endsWith('__cid') && record[col.originalName.replace('__cid', '')]) {
|
|
917
|
+
rawValue = record[col.originalName.replace('__cid', '')].cid;
|
|
918
|
+
}
|
|
919
|
+
return coerceValue(col.sqlType, rawValue);
|
|
920
|
+
}
|
|
921
|
+
/** Extract a raw column value from a data object and coerce for bulk insert */
|
|
922
|
+
function resolveRawColumnValue(col, data) {
|
|
923
|
+
return coerceValue(col.sqlType, data[col.originalName]);
|
|
924
|
+
}
|
|
925
|
+
/** Coerce a value to the appropriate type for insertion */
|
|
926
|
+
function coerceValue(sqlType, rawValue) {
|
|
927
|
+
if (rawValue === undefined || rawValue === null)
|
|
928
|
+
return null;
|
|
929
|
+
// Objects and arrays always need JSON stringification regardless of sqlType
|
|
930
|
+
// (on SQLite, JSON columns map to TEXT but still need stringification)
|
|
931
|
+
if (typeof rawValue === 'object' && !(rawValue instanceof Uint8Array)) {
|
|
932
|
+
return JSON.stringify(rawValue);
|
|
933
|
+
}
|
|
934
|
+
if (sqlType === 'JSON' || sqlType === 'TEXT') {
|
|
935
|
+
return String(rawValue);
|
|
936
|
+
}
|
|
937
|
+
if (sqlType === 'INTEGER' || sqlType === 'BIGINT') {
|
|
938
|
+
return typeof rawValue === 'number' ? rawValue : parseInt(rawValue);
|
|
939
|
+
}
|
|
940
|
+
if (sqlType === 'BOOLEAN')
|
|
941
|
+
return !!rawValue;
|
|
942
|
+
return String(rawValue);
|
|
943
|
+
}
|
|
695
944
|
export async function queryRecords(collection, opts = {}) {
|
|
696
945
|
const schema = schemas.get(collection);
|
|
697
946
|
if (!schema)
|
|
@@ -734,7 +983,7 @@ export async function queryRecords(collection, opts = {}) {
|
|
|
734
983
|
sql += ' AND ' + conditions.join(' AND ');
|
|
735
984
|
sql += ` ORDER BY t.${sortName} ${order.toUpperCase()}, t.cid ${order.toUpperCase()} LIMIT $${paramIdx++}`;
|
|
736
985
|
params.push(limit + 1); // fetch one extra for cursor
|
|
737
|
-
const rows = await all(sql,
|
|
986
|
+
const rows = await all(sql, params);
|
|
738
987
|
const hasMore = rows.length > limit;
|
|
739
988
|
if (hasMore)
|
|
740
989
|
rows.pop();
|
|
@@ -774,7 +1023,7 @@ export async function queryRecords(collection, opts = {}) {
|
|
|
774
1023
|
}
|
|
775
1024
|
export async function getRecordByUri(uri) {
|
|
776
1025
|
for (const [_collection, schema] of schemas) {
|
|
777
|
-
const rows = await all(`SELECT t.*, r.handle FROM ${schema.tableName} t LEFT JOIN _repos r ON t.did = r.did WHERE t.uri = $1 AND (r.status IS NULL OR r.status != 'takendown')`, uri);
|
|
1026
|
+
const rows = await all(`SELECT t.*, r.handle FROM ${schema.tableName} t LEFT JOIN _repos r ON t.did = r.did WHERE t.uri = $1 AND (r.status IS NULL OR r.status != 'takendown')`, [uri]);
|
|
778
1027
|
if (rows.length > 0) {
|
|
779
1028
|
const row = rows[0];
|
|
780
1029
|
if (schema.children.length > 0) {
|
|
@@ -811,7 +1060,7 @@ export async function getRecordsByUris(collection, uris) {
|
|
|
811
1060
|
if (!schema)
|
|
812
1061
|
return [];
|
|
813
1062
|
const placeholders = uris.map((_, i) => `$${i + 1}`).join(',');
|
|
814
|
-
const rows = await all(`SELECT t.*, r.handle FROM ${schema.tableName} t LEFT JOIN _repos r ON t.did = r.did WHERE t.uri IN (${placeholders}) AND (r.status IS NULL OR r.status != 'takendown')`,
|
|
1063
|
+
const rows = await all(`SELECT t.*, r.handle FROM ${schema.tableName} t LEFT JOIN _repos r ON t.did = r.did WHERE t.uri IN (${placeholders}) AND (r.status IS NULL OR r.status != 'takendown')`, uris);
|
|
815
1064
|
// Batch-fetch child rows for all URIs
|
|
816
1065
|
const childData = new Map();
|
|
817
1066
|
for (const child of schema.children) {
|
|
@@ -839,6 +1088,19 @@ export async function getRecordsByUris(collection, uris) {
|
|
|
839
1088
|
const byUri = new Map(rows.map((r) => [r.uri, r]));
|
|
840
1089
|
return uris.map((u) => byUri.get(u)).filter(Boolean);
|
|
841
1090
|
}
|
|
1091
|
+
/** Fetch records by URIs and return as a shaped Map keyed by URI. */
|
|
1092
|
+
export async function getRecordsMap(collection, uris) {
|
|
1093
|
+
if (uris.length === 0)
|
|
1094
|
+
return new Map();
|
|
1095
|
+
const records = await getRecordsByUris(collection, uris);
|
|
1096
|
+
const map = new Map();
|
|
1097
|
+
for (const r of records) {
|
|
1098
|
+
const shaped = reshapeRow(r, r?.__childData, r?.__unionData);
|
|
1099
|
+
if (shaped)
|
|
1100
|
+
map.set(shaped.uri, shaped);
|
|
1101
|
+
}
|
|
1102
|
+
return map;
|
|
1103
|
+
}
|
|
842
1104
|
/**
|
|
843
1105
|
* Multi-phase search across any collection's records.
|
|
844
1106
|
*
|
|
@@ -859,8 +1121,8 @@ export async function searchRecords(collection, query, opts = {}) {
|
|
|
859
1121
|
if (!schema)
|
|
860
1122
|
throw new Error(`Unknown collection: ${collection}`);
|
|
861
1123
|
const elapsed = timer();
|
|
862
|
-
const { limit = 20,
|
|
863
|
-
const textCols = schema.columns.filter((c) => c.
|
|
1124
|
+
const { limit = 20, fuzzy = true } = opts;
|
|
1125
|
+
const textCols = schema.columns.filter((c) => c.sqlType === 'TEXT');
|
|
864
1126
|
// Also check if FTS has indexed any columns (including derived JSON columns)
|
|
865
1127
|
const ftsSearchCols = getSearchColumns(collection);
|
|
866
1128
|
if (textCols.length === 0 && ftsSearchCols.length === 0) {
|
|
@@ -868,149 +1130,67 @@ export async function searchRecords(collection, query, opts = {}) {
|
|
|
868
1130
|
}
|
|
869
1131
|
// FTS shadow table name (dots replaced with underscores)
|
|
870
1132
|
const safeName = '_fts_' + collection.replace(/\./g, '_');
|
|
871
|
-
const ftsSchema = `fts_main_${safeName}`;
|
|
872
1133
|
const phaseErrors = [];
|
|
873
1134
|
const phasesUsed = [];
|
|
874
|
-
// Phase 1: BM25 ranked search
|
|
1135
|
+
// Phase 1: BM25 ranked search via SearchPort
|
|
875
1136
|
let bm25Results = [];
|
|
876
|
-
|
|
877
|
-
|
|
878
|
-
|
|
879
|
-
|
|
880
|
-
|
|
881
|
-
|
|
882
|
-
|
|
883
|
-
|
|
884
|
-
|
|
885
|
-
|
|
886
|
-
|
|
887
|
-
|
|
888
|
-
|
|
889
|
-
|
|
890
|
-
|
|
891
|
-
|
|
892
|
-
|
|
893
|
-
|
|
894
|
-
|
|
895
|
-
|
|
1137
|
+
const sp = getSearchPort();
|
|
1138
|
+
if (sp)
|
|
1139
|
+
try {
|
|
1140
|
+
const ftsQuery = stripStopWords(query);
|
|
1141
|
+
const ftsSearchColNames = getSearchColumns(collection);
|
|
1142
|
+
// Get ranked URIs from the search port
|
|
1143
|
+
const hits = await sp.search(safeName, ftsQuery, ftsSearchColNames, limit + 1, 0);
|
|
1144
|
+
if (hits.length > 0) {
|
|
1145
|
+
const uriList = hits.map((h) => h.uri);
|
|
1146
|
+
const scoreMap = new Map(hits.map((h) => [h.uri, h.score]));
|
|
1147
|
+
// Fetch full records for matched URIs
|
|
1148
|
+
const placeholders = uriList.map((_, i) => `$${i + 1}`).join(', ');
|
|
1149
|
+
const rows = await all(`SELECT m.* FROM ${schema.tableName} m
|
|
1150
|
+
LEFT JOIN _repos r ON m.did = r.did
|
|
1151
|
+
WHERE m.uri IN (${placeholders})
|
|
1152
|
+
AND (r.status IS NULL OR r.status != 'takendown')`, uriList);
|
|
1153
|
+
// Re-attach scores and sort
|
|
1154
|
+
bm25Results = rows
|
|
1155
|
+
.map((r) => ({ ...r, score: scoreMap.get(r.uri) ?? 0 }))
|
|
1156
|
+
.sort((a, b) => b.score - a.score);
|
|
896
1157
|
}
|
|
1158
|
+
phasesUsed.push('bm25');
|
|
1159
|
+
}
|
|
1160
|
+
catch (err) {
|
|
1161
|
+
phaseErrors.push(`bm25: ${err.message}`);
|
|
897
1162
|
}
|
|
898
|
-
sql += ` ORDER BY score, m.cid DESC LIMIT $${paramIdx++}`;
|
|
899
|
-
params.push(limit + 1);
|
|
900
|
-
bm25Results = await all(sql, ...params);
|
|
901
|
-
phasesUsed.push('bm25');
|
|
902
|
-
}
|
|
903
|
-
catch (err) {
|
|
904
|
-
phaseErrors.push(`bm25: ${err.message}`);
|
|
905
|
-
}
|
|
906
1163
|
const bm25Count = bm25Results.length;
|
|
907
1164
|
const hasMore = bm25Results.length > limit;
|
|
908
1165
|
if (hasMore)
|
|
909
1166
|
bm25Results.pop();
|
|
910
|
-
// Phase 2: Exact substring match — boosts phrase matches above BM25 results
|
|
911
|
-
const exactMatchResults = [];
|
|
912
|
-
const bm25Uris = new Set(bm25Results.map((r) => r.uri));
|
|
913
|
-
try {
|
|
914
|
-
const searchParam = `%${query}%`;
|
|
915
|
-
let paramIdx = 1;
|
|
916
|
-
const ilikeConds = [];
|
|
917
|
-
const params = [];
|
|
918
|
-
// TEXT columns — direct ILIKE
|
|
919
|
-
for (const c of textCols) {
|
|
920
|
-
ilikeConds.push(`t.${c.name} ILIKE $${paramIdx++}`);
|
|
921
|
-
params.push(searchParam);
|
|
922
|
-
}
|
|
923
|
-
// JSON columns — cast to text then ILIKE
|
|
924
|
-
const jsonCols = schema.columns.filter((c) => c.duckdbType === 'JSON');
|
|
925
|
-
for (const c of jsonCols) {
|
|
926
|
-
ilikeConds.push(`CAST(t.${c.name} AS TEXT) ILIKE $${paramIdx++}`);
|
|
927
|
-
params.push(searchParam);
|
|
928
|
-
}
|
|
929
|
-
// Handle from _repos table
|
|
930
|
-
ilikeConds.push(`r.handle ILIKE $${paramIdx++}`);
|
|
931
|
-
params.push(searchParam);
|
|
932
|
-
if (ilikeConds.length > 0) {
|
|
933
|
-
const exactSQL = `SELECT t.* FROM ${schema.tableName} t LEFT JOIN _repos r ON t.did = r.did
|
|
934
|
-
WHERE (${ilikeConds.join(' OR ')})
|
|
935
|
-
ORDER BY t.indexed_at DESC
|
|
936
|
-
LIMIT $${paramIdx++}`;
|
|
937
|
-
params.push(limit);
|
|
938
|
-
const rows = await all(exactSQL, ...params);
|
|
939
|
-
phasesUsed.push('exact');
|
|
940
|
-
for (const row of rows) {
|
|
941
|
-
if (!bm25Uris.has(row.uri)) {
|
|
942
|
-
exactMatchResults.push(row);
|
|
943
|
-
bm25Uris.add(row.uri);
|
|
944
|
-
}
|
|
945
|
-
}
|
|
946
|
-
}
|
|
947
|
-
}
|
|
948
|
-
catch (err) {
|
|
949
|
-
phaseErrors.push(`exact: ${err.message}`);
|
|
950
|
-
}
|
|
951
|
-
// Merge: exact matches first, then BM25 results, capped at limit
|
|
952
|
-
const mergedResults = [...exactMatchResults, ...bm25Results].slice(0, limit + (hasMore ? 1 : 0));
|
|
953
|
-
// Replace bm25Results with merged for downstream phases
|
|
954
|
-
bm25Results = mergedResults;
|
|
955
|
-
// Phase 3: ILIKE scan of rows written since last FTS rebuild (immediate searchability)
|
|
956
1167
|
const existingUris = new Set(bm25Results.map((r) => r.uri));
|
|
957
|
-
|
|
958
|
-
|
|
959
|
-
let recentCount = 0;
|
|
960
|
-
if (rebuiltAt && bm25Results.length < limit) {
|
|
961
|
-
const remaining = limit - bm25Results.length;
|
|
962
|
-
const searchParam = `%${query}%`;
|
|
963
|
-
let paramIdx = 1;
|
|
964
|
-
const ilikeParts = textCols.map((c) => `t.${c.name} ILIKE $${paramIdx++}`);
|
|
965
|
-
ilikeParts.push(`r.handle ILIKE $${paramIdx++}`);
|
|
966
|
-
const ilikeConds = ilikeParts.join(' OR ');
|
|
967
|
-
const params = [...textCols.map(() => searchParam), searchParam];
|
|
968
|
-
const recentSQL = `SELECT t.* FROM ${schema.tableName} t LEFT JOIN _repos r ON t.did = r.did
|
|
969
|
-
WHERE t.indexed_at >= $${paramIdx++} AND t.uri NOT IN (SELECT uri FROM ${safeName}) AND (${ilikeConds})
|
|
970
|
-
ORDER BY t.indexed_at DESC
|
|
971
|
-
LIMIT $${paramIdx++}`;
|
|
972
|
-
params.push(rebuiltAt, remaining + existingUris.size);
|
|
973
|
-
try {
|
|
974
|
-
const recentRows = await all(recentSQL, ...params);
|
|
975
|
-
phasesUsed.push('recent');
|
|
976
|
-
for (const row of recentRows) {
|
|
977
|
-
if (bm25Results.length >= limit)
|
|
978
|
-
break;
|
|
979
|
-
if (!existingUris.has(row.uri)) {
|
|
980
|
-
existingUris.add(row.uri);
|
|
981
|
-
bm25Results.push(row);
|
|
982
|
-
recentCount++;
|
|
983
|
-
}
|
|
984
|
-
}
|
|
985
|
-
}
|
|
986
|
-
catch (err) {
|
|
987
|
-
phaseErrors.push(`recent: ${err.message}`);
|
|
988
|
-
}
|
|
989
|
-
}
|
|
990
|
-
// Phase 4: Fuzzy fallback for typo tolerance (if still under limit)
|
|
1168
|
+
// Phase 2: Fuzzy fallback for typo tolerance (if still under limit)
|
|
1169
|
+
// Only available on dialects with jaro_winkler_similarity (DuckDB)
|
|
991
1170
|
let fuzzyCount = 0;
|
|
992
|
-
if (fuzzy && bm25Results.length < limit) {
|
|
1171
|
+
if (fuzzy && dialect.jaroWinklerSimilarity && bm25Results.length < limit) {
|
|
993
1172
|
const remaining = limit - bm25Results.length;
|
|
1173
|
+
const jwFn = dialect.jaroWinklerSimilarity;
|
|
994
1174
|
const simExprs = [
|
|
995
|
-
...textCols.map((c) =>
|
|
996
|
-
|
|
1175
|
+
...textCols.map((c) => `${jwFn}(lower(t.${q(c.name)}), lower($1))`),
|
|
1176
|
+
`${jwFn}(lower(r.handle), lower($1))`,
|
|
997
1177
|
];
|
|
998
1178
|
// Include child table TEXT columns via correlated subquery
|
|
999
1179
|
for (const child of schema.children) {
|
|
1000
1180
|
for (const col of child.columns) {
|
|
1001
|
-
if (col.
|
|
1002
|
-
simExprs.push(`COALESCE((SELECT MAX(
|
|
1181
|
+
if (col.sqlType === 'TEXT') {
|
|
1182
|
+
simExprs.push(`COALESCE((SELECT MAX(${jwFn}(lower(c.${q(col.name)}), lower($1))) FROM ${child.tableName} c WHERE c.parent_uri = t.uri), 0)`);
|
|
1003
1183
|
}
|
|
1004
1184
|
}
|
|
1005
1185
|
}
|
|
1006
|
-
const greatestExpr =
|
|
1186
|
+
const greatestExpr = dialect.greatest(simExprs);
|
|
1007
1187
|
const fuzzySQL = `SELECT t.*, ${greatestExpr} AS fuzzy_score
|
|
1008
1188
|
FROM ${schema.tableName} t LEFT JOIN _repos r ON t.did = r.did
|
|
1009
1189
|
WHERE ${greatestExpr} >= 0.8
|
|
1010
1190
|
ORDER BY fuzzy_score DESC
|
|
1011
1191
|
LIMIT $2`;
|
|
1012
1192
|
try {
|
|
1013
|
-
const fuzzyRows = await all(fuzzySQL, query, remaining + existingUris.size);
|
|
1193
|
+
const fuzzyRows = await all(fuzzySQL, [query, remaining + existingUris.size]);
|
|
1014
1194
|
phasesUsed.push('fuzzy');
|
|
1015
1195
|
for (const row of fuzzyRows) {
|
|
1016
1196
|
if (bm25Results.length >= limit)
|
|
@@ -1025,16 +1205,17 @@ export async function searchRecords(collection, query, opts = {}) {
|
|
|
1025
1205
|
phaseErrors.push(`fuzzy: ${err.message}`);
|
|
1026
1206
|
}
|
|
1027
1207
|
}
|
|
1028
|
-
// Remove score columns
|
|
1029
|
-
const
|
|
1208
|
+
// Remove score columns and reshape into Row<T> with value
|
|
1209
|
+
const rawRecords = bm25Results.map(({ score: _score, fuzzy_score: _fuzzy_score, ...rest }) => rest);
|
|
1210
|
+
const records = rawRecords
|
|
1211
|
+
.map((r) => reshapeRow(r, r?.__childData, r?.__unionData))
|
|
1212
|
+
.filter((r) => r != null);
|
|
1030
1213
|
const lastRow = bm25Results[bm25Results.length - 1];
|
|
1031
1214
|
const nextCursor = hasMore && lastRow?.score != null ? packCursor(lastRow.score, lastRow.cid) : undefined;
|
|
1032
1215
|
emit('search', 'query', {
|
|
1033
1216
|
collection,
|
|
1034
1217
|
query,
|
|
1035
1218
|
bm25_count: bm25Count > limit ? bm25Count - 1 : bm25Count,
|
|
1036
|
-
exact_count: exactMatchResults.length,
|
|
1037
|
-
recent_count: recentCount,
|
|
1038
1219
|
fuzzy_count: fuzzyCount,
|
|
1039
1220
|
total_results: records.length,
|
|
1040
1221
|
duration_ms: elapsed(),
|
|
@@ -1045,10 +1226,13 @@ export async function searchRecords(collection, query, opts = {}) {
|
|
|
1045
1226
|
}
|
|
1046
1227
|
// Raw SQL for script feeds
|
|
1047
1228
|
export async function querySQL(sql, params = []) {
|
|
1048
|
-
return all(sql,
|
|
1229
|
+
return all(sql, params);
|
|
1049
1230
|
}
|
|
1050
|
-
export async function runSQL(sql,
|
|
1051
|
-
return run(sql,
|
|
1231
|
+
export async function runSQL(sql, params = []) {
|
|
1232
|
+
return run(sql, params);
|
|
1233
|
+
}
|
|
1234
|
+
export async function createBulkInserterSQL(table, columns, options) {
|
|
1235
|
+
return port.createBulkInserter(table, columns, options);
|
|
1052
1236
|
}
|
|
1053
1237
|
export function getSchema(collection) {
|
|
1054
1238
|
return schemas.get(collection);
|
|
@@ -1057,7 +1241,9 @@ export async function countByField(collection, field, value) {
|
|
|
1057
1241
|
const schema = schemas.get(collection);
|
|
1058
1242
|
if (!schema)
|
|
1059
1243
|
return 0;
|
|
1060
|
-
const rows = await all(`SELECT COUNT(*) as count FROM ${schema.tableName} WHERE ${field} = $1`,
|
|
1244
|
+
const rows = await all(`SELECT COUNT(*) as count FROM ${schema.tableName} WHERE ${field} = $1`, [
|
|
1245
|
+
value,
|
|
1246
|
+
]);
|
|
1061
1247
|
return Number(rows[0]?.count || 0);
|
|
1062
1248
|
}
|
|
1063
1249
|
export async function countByFieldBatch(collection, field, values) {
|
|
@@ -1067,7 +1253,7 @@ export async function countByFieldBatch(collection, field, values) {
|
|
|
1067
1253
|
if (!schema)
|
|
1068
1254
|
return new Map();
|
|
1069
1255
|
const placeholders = values.map((_, i) => `$${i + 1}`).join(',');
|
|
1070
|
-
const rows = await all(`SELECT ${field}, COUNT(*) as count FROM ${schema.tableName} WHERE ${field} IN (${placeholders}) GROUP BY ${field}`,
|
|
1256
|
+
const rows = await all(`SELECT ${field}, COUNT(*) as count FROM ${schema.tableName} WHERE ${field} IN (${placeholders}) GROUP BY ${field}`, values);
|
|
1071
1257
|
const result = new Map();
|
|
1072
1258
|
for (const row of rows) {
|
|
1073
1259
|
result.set(row[field], Number(row.count));
|
|
@@ -1078,7 +1264,7 @@ export async function findByField(collection, field, value) {
|
|
|
1078
1264
|
const schema = schemas.get(collection);
|
|
1079
1265
|
if (!schema)
|
|
1080
1266
|
return null;
|
|
1081
|
-
const rows = await all(`SELECT * FROM ${schema.tableName} WHERE ${field} = $1 LIMIT 1`, value);
|
|
1267
|
+
const rows = await all(`SELECT * FROM ${schema.tableName} WHERE ${field} = $1 LIMIT 1`, [value]);
|
|
1082
1268
|
return rows[0] || null;
|
|
1083
1269
|
}
|
|
1084
1270
|
export async function findByFieldBatch(collection, field, values) {
|
|
@@ -1088,7 +1274,7 @@ export async function findByFieldBatch(collection, field, values) {
|
|
|
1088
1274
|
if (!schema)
|
|
1089
1275
|
return new Map();
|
|
1090
1276
|
const placeholders = values.map((_, i) => `$${i + 1}`).join(',');
|
|
1091
|
-
const rows = await all(`SELECT t.*, r.handle FROM ${schema.tableName} t LEFT JOIN _repos r ON t.did = r.did WHERE t.${field} IN (${placeholders})`,
|
|
1277
|
+
const rows = await all(`SELECT t.*, r.handle FROM ${schema.tableName} t LEFT JOIN _repos r ON t.did = r.did WHERE t.${field} IN (${placeholders})`, values);
|
|
1092
1278
|
// Attach child data if this collection has decomposed arrays
|
|
1093
1279
|
if (schema.children.length > 0 && rows.length > 0) {
|
|
1094
1280
|
const uris = rows.map((r) => r.uri);
|
|
@@ -1098,7 +1284,6 @@ export async function findByFieldBatch(collection, field, values) {
|
|
|
1098
1284
|
childData.set(child.fieldName, childRows);
|
|
1099
1285
|
}
|
|
1100
1286
|
for (const row of rows) {
|
|
1101
|
-
;
|
|
1102
1287
|
row.__childData = childData;
|
|
1103
1288
|
}
|
|
1104
1289
|
}
|
|
@@ -1129,7 +1314,7 @@ export async function findUriByFields(collection, conditions) {
|
|
|
1129
1314
|
return null;
|
|
1130
1315
|
const where = conditions.map((c, i) => `${c.field} = $${i + 1}`).join(' AND ');
|
|
1131
1316
|
const params = conditions.map((c) => c.value);
|
|
1132
|
-
const rows = await all(`SELECT uri FROM ${schema.tableName} WHERE ${where} LIMIT 1`,
|
|
1317
|
+
const rows = await all(`SELECT uri FROM ${schema.tableName} WHERE ${where} LIMIT 1`, params);
|
|
1133
1318
|
return rows[0]?.uri || null;
|
|
1134
1319
|
}
|
|
1135
1320
|
const ENVELOPE_KEYS = new Set(['uri', 'cid', 'did', 'handle', 'indexed_at']);
|
|
@@ -1145,7 +1330,7 @@ export async function getChildRows(childTableName, parentUris) {
|
|
|
1145
1330
|
if (parentUris.length === 0)
|
|
1146
1331
|
return new Map();
|
|
1147
1332
|
const placeholders = parentUris.map((_, i) => `$${i + 1}`).join(',');
|
|
1148
|
-
const rows = await all(`SELECT * FROM ${childTableName} WHERE parent_uri IN (${placeholders})`,
|
|
1333
|
+
const rows = await all(`SELECT * FROM ${childTableName} WHERE parent_uri IN (${placeholders})`, parentUris);
|
|
1149
1334
|
const result = new Map();
|
|
1150
1335
|
for (const row of rows) {
|
|
1151
1336
|
const key = row.parent_uri;
|
|
@@ -1167,7 +1352,7 @@ export function reshapeRow(row, childData, unionData) {
|
|
|
1167
1352
|
if (schema) {
|
|
1168
1353
|
for (const col of schema.columns) {
|
|
1169
1354
|
nameMap.set(col.name, col.originalName);
|
|
1170
|
-
if (col.
|
|
1355
|
+
if (col.isJson)
|
|
1171
1356
|
jsonCols.add(col.name);
|
|
1172
1357
|
}
|
|
1173
1358
|
}
|
|
@@ -1273,15 +1458,17 @@ export function unpackCursor(cursor) {
|
|
|
1273
1458
|
}
|
|
1274
1459
|
}
|
|
1275
1460
|
export async function queryLabelsByDid(did) {
|
|
1276
|
-
return all(`SELECT * FROM _labels WHERE uri LIKE $1 AND neg = false AND (exp IS NULL OR exp > CURRENT_TIMESTAMP)`,
|
|
1461
|
+
return all(`SELECT * FROM _labels WHERE uri LIKE $1 AND neg = false AND (exp IS NULL OR exp > CURRENT_TIMESTAMP)`, [
|
|
1462
|
+
`at://${did}/%`,
|
|
1463
|
+
]);
|
|
1277
1464
|
}
|
|
1278
1465
|
export async function searchAccounts(query, limit = 20) {
|
|
1279
|
-
return all(`SELECT did, handle, status FROM _repos WHERE did
|
|
1466
|
+
return all(`SELECT did, handle, status FROM _repos WHERE did ${dialect.ilike} $1 OR handle ${dialect.ilike} $1 ORDER BY handle LIMIT $2`, [`%${query}%`, limit]);
|
|
1280
1467
|
}
|
|
1281
1468
|
export async function getAccountRecordCount(did) {
|
|
1282
1469
|
let total = 0;
|
|
1283
1470
|
for (const [, schema] of schemas) {
|
|
1284
|
-
const rows = await all(`SELECT COUNT(*) as count FROM ${schema.tableName} WHERE did = $1`, did);
|
|
1471
|
+
const rows = await all(`SELECT COUNT(*) as count FROM ${schema.tableName} WHERE did = $1`, [did]);
|
|
1285
1472
|
total += Number(rows[0]?.count || 0);
|
|
1286
1473
|
}
|
|
1287
1474
|
return total;
|
|
@@ -1289,17 +1476,17 @@ export async function getAccountRecordCount(did) {
|
|
|
1289
1476
|
export async function getAllRecordUrisForDid(did) {
|
|
1290
1477
|
const uris = [];
|
|
1291
1478
|
for (const [, schema] of schemas) {
|
|
1292
|
-
const rows = await all(`SELECT uri FROM ${schema.tableName} WHERE did = $1`, did);
|
|
1479
|
+
const rows = await all(`SELECT uri FROM ${schema.tableName} WHERE did = $1`, [did]);
|
|
1293
1480
|
uris.push(...rows.map((r) => r.uri));
|
|
1294
1481
|
}
|
|
1295
1482
|
return uris;
|
|
1296
1483
|
}
|
|
1297
1484
|
export async function isTakendownDid(did) {
|
|
1298
|
-
const rows = await all(`SELECT 1 FROM _repos WHERE did = $1 AND status = 'takendown' LIMIT 1`, did);
|
|
1485
|
+
const rows = await all(`SELECT 1 FROM _repos WHERE did = $1 AND status = 'takendown' LIMIT 1`, [did]);
|
|
1299
1486
|
return rows.length > 0;
|
|
1300
1487
|
}
|
|
1301
1488
|
export async function getPreferences(did) {
|
|
1302
|
-
const rows = await all(`SELECT key, value FROM _preferences WHERE did = $1`, did);
|
|
1489
|
+
const rows = await all(`SELECT key, value FROM _preferences WHERE did = $1`, [did]);
|
|
1303
1490
|
const prefs = {};
|
|
1304
1491
|
for (const row of rows) {
|
|
1305
1492
|
try {
|
|
@@ -1312,50 +1499,66 @@ export async function getPreferences(did) {
|
|
|
1312
1499
|
return prefs;
|
|
1313
1500
|
}
|
|
1314
1501
|
export async function putPreference(did, key, value) {
|
|
1315
|
-
await run(`INSERT OR REPLACE INTO _preferences (did, key, value, updated_at) VALUES ($1, $2, $3, $4)`,
|
|
1502
|
+
await run(`INSERT OR REPLACE INTO _preferences (did, key, value, updated_at) VALUES ($1, $2, $3, $4)`, [
|
|
1503
|
+
did,
|
|
1504
|
+
key,
|
|
1505
|
+
JSON.stringify(value),
|
|
1506
|
+
new Date().toISOString(),
|
|
1507
|
+
]);
|
|
1508
|
+
}
|
|
1509
|
+
export async function resolveHandleToDid(handle) {
|
|
1510
|
+
if (handle.startsWith('did:'))
|
|
1511
|
+
return handle;
|
|
1512
|
+
const rows = await all(`SELECT did FROM _repos WHERE handle = $1 LIMIT 1`, [handle]);
|
|
1513
|
+
return rows[0]?.did ?? null;
|
|
1316
1514
|
}
|
|
1317
1515
|
export async function filterTakendownDids(dids) {
|
|
1318
1516
|
if (dids.length === 0)
|
|
1319
1517
|
return new Set();
|
|
1320
1518
|
const placeholders = dids.map((_, i) => `$${i + 1}`).join(',');
|
|
1321
|
-
const rows = await all(`SELECT did FROM _repos WHERE did IN (${placeholders}) AND status = 'takendown'`,
|
|
1519
|
+
const rows = await all(`SELECT did FROM _repos WHERE did IN (${placeholders}) AND status = 'takendown'`, dids);
|
|
1322
1520
|
return new Set(rows.map((r) => r.did));
|
|
1323
1521
|
}
|
|
1324
|
-
export async function
|
|
1325
|
-
|
|
1326
|
-
|
|
1327
|
-
|
|
1328
|
-
|
|
1329
|
-
|
|
1330
|
-
|
|
1331
|
-
|
|
1332
|
-
|
|
1333
|
-
|
|
1334
|
-
|
|
1335
|
-
|
|
1336
|
-
const childColSelects = child.columns
|
|
1337
|
-
.map((c) => `json_extract_string(item.val, '$.${c.originalName}')`)
|
|
1338
|
-
.join(', ');
|
|
1339
|
-
const childColNames = ['parent_uri', 'parent_did', ...child.columns.map((c) => c.name)];
|
|
1340
|
-
const notNullFilters = child.columns
|
|
1341
|
-
.filter((c) => c.notNull)
|
|
1342
|
-
.map((c) => `json_extract_string(item.val, '$.${c.originalName}') IS NOT NULL`);
|
|
1343
|
-
const whereClause = [`p.${snakeField} IS NOT NULL`, ...notNullFilters].join(' AND ');
|
|
1344
|
-
try {
|
|
1345
|
-
await run(`DELETE FROM ${child.tableName}`);
|
|
1346
|
-
await run(`
|
|
1347
|
-
INSERT INTO ${child.tableName} (${childColNames.join(', ')})
|
|
1348
|
-
SELECT p.uri, p.did, ${childColSelects}
|
|
1349
|
-
FROM ${schema.tableName} p,
|
|
1350
|
-
unnest(from_json(p.${snakeField}::JSON, '["json"]')) AS item(val)
|
|
1351
|
-
WHERE ${whereClause}
|
|
1352
|
-
`);
|
|
1353
|
-
const result = await all(`SELECT COUNT(*)::INTEGER as n FROM ${child.tableName}`);
|
|
1354
|
-
console.log(`[db] Backfilled ${child.tableName}: ${result[0]?.n || 0} rows`);
|
|
1355
|
-
}
|
|
1356
|
-
catch (err) {
|
|
1357
|
-
console.warn(`[db] Backfill skipped for ${child.tableName}: ${err.message}`);
|
|
1358
|
-
}
|
|
1359
|
-
}
|
|
1522
|
+
export async function insertReport(report) {
|
|
1523
|
+
const createdAt = new Date().toISOString();
|
|
1524
|
+
const rows = await all(`INSERT INTO _reports (subject_uri, subject_did, label, reason, reported_by, created_at) VALUES ($1, $2, $3, $4, $5, $6) RETURNING id`, [report.subjectUri, report.subjectDid, report.label, report.reason || null, report.reportedBy, createdAt]);
|
|
1525
|
+
return { id: rows[0].id };
|
|
1526
|
+
}
|
|
1527
|
+
export async function queryReports(opts) {
|
|
1528
|
+
const conditions = [];
|
|
1529
|
+
const params = [];
|
|
1530
|
+
let idx = 1;
|
|
1531
|
+
if (opts.status) {
|
|
1532
|
+
conditions.push(`r.status = $${idx++}`);
|
|
1533
|
+
params.push(opts.status);
|
|
1360
1534
|
}
|
|
1535
|
+
if (opts.label) {
|
|
1536
|
+
conditions.push(`r.label = $${idx++}`);
|
|
1537
|
+
params.push(opts.label);
|
|
1538
|
+
}
|
|
1539
|
+
const where = conditions.length > 0 ? `WHERE ${conditions.join(' AND ')}` : '';
|
|
1540
|
+
const limit = opts.limit || 50;
|
|
1541
|
+
const offset = opts.offset || 0;
|
|
1542
|
+
const countRows = await all(`SELECT ${dialect.countAsInteger} as count FROM _reports r ${where}`, params);
|
|
1543
|
+
const total = Number(countRows[0]?.count || 0);
|
|
1544
|
+
const rows = await all(`SELECT r.*, rp.handle as reported_by_handle FROM _reports r LEFT JOIN _repos rp ON r.reported_by = rp.did ${where} ORDER BY r.created_at DESC LIMIT $${idx++} OFFSET $${idx++}`, [...params, limit, offset]);
|
|
1545
|
+
return { reports: rows, total };
|
|
1546
|
+
}
|
|
1547
|
+
export async function resolveReport(id, action, resolvedBy) {
|
|
1548
|
+
const rows = await all(`SELECT subject_uri, label, status FROM _reports WHERE id = $1`, [id]);
|
|
1549
|
+
if (!rows.length)
|
|
1550
|
+
return null;
|
|
1551
|
+
if (rows[0].status !== 'open')
|
|
1552
|
+
return null;
|
|
1553
|
+
await run(`UPDATE _reports SET status = $1, resolved_by = $2, resolved_at = $3 WHERE id = $4`, [
|
|
1554
|
+
action,
|
|
1555
|
+
resolvedBy,
|
|
1556
|
+
new Date().toISOString(),
|
|
1557
|
+
id,
|
|
1558
|
+
]);
|
|
1559
|
+
return { subjectUri: rows[0].subject_uri, label: rows[0].label };
|
|
1560
|
+
}
|
|
1561
|
+
export async function getOpenReportCount() {
|
|
1562
|
+
const rows = await all(`SELECT ${dialect.countAsInteger} as count FROM _reports WHERE status = 'open'`);
|
|
1563
|
+
return Number(rows[0]?.count || 0);
|
|
1361
1564
|
}
|