@gscdump/engine-sqlite 1.4.0 → 1.4.1
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/index.d.mts +8 -4402
- package/dist/index.mjs +8 -561
- package/dist/metrics.d.mts +8 -0
- package/dist/metrics.mjs +14 -0
- package/dist/query-source.d.mts +11 -0
- package/dist/query-source.mjs +15 -0
- package/dist/r2-manifest-schema.d.mts +775 -0
- package/dist/r2-manifest-schema.mjs +81 -0
- package/dist/r2-manifest-store.d.mts +11 -0
- package/dist/r2-manifest-store.mjs +282 -0
- package/dist/resolver-adapter.d.mts +27 -0
- package/dist/resolver-adapter.mjs +41 -0
- package/dist/runner.d.mts +42 -0
- package/dist/runner.mjs +30 -0
- package/dist/schema.d.mts +3542 -0
- package/dist/schema.mjs +112 -0
- package/package.json +2 -2
|
@@ -0,0 +1,81 @@
|
|
|
1
|
+
import { sql } from "drizzle-orm";
|
|
2
|
+
import { index, integer, primaryKey, sqliteTable, text, unique } from "drizzle-orm/sqlite-core";
|
|
3
|
+
const r2Manifest = sqliteTable("r2_manifest", {
|
|
4
|
+
id: text("id").primaryKey(),
|
|
5
|
+
userId: integer("user_id").notNull(),
|
|
6
|
+
siteId: text("site_id"),
|
|
7
|
+
table: text("table").notNull(),
|
|
8
|
+
partition: text("partition").notNull(),
|
|
9
|
+
objectKey: text("object_key").notNull(),
|
|
10
|
+
rowCount: integer("row_count").notNull().default(0),
|
|
11
|
+
bytes: integer("bytes").notNull().default(0),
|
|
12
|
+
createdAt: integer("created_at").notNull(),
|
|
13
|
+
retiredAt: integer("retired_at"),
|
|
14
|
+
tier: text("tier"),
|
|
15
|
+
searchType: text("search_type"),
|
|
16
|
+
schemaVersion: integer("schema_version")
|
|
17
|
+
}, (t) => [
|
|
18
|
+
index("idx_r2_manifest_live").on(t.userId, t.siteId, t.table, t.partition, t.retiredAt),
|
|
19
|
+
index("idx_r2_manifest_lookup").on(t.userId, t.siteId, t.table, t.searchType, t.tier, t.partition, t.retiredAt),
|
|
20
|
+
index("idx_r2_manifest_retired").on(t.retiredAt),
|
|
21
|
+
index("idx_r2_manifest_tier").on(t.userId, t.siteId, t.table, t.tier, t.retiredAt),
|
|
22
|
+
unique("r2_manifest_object_key_unique").on(t.objectKey)
|
|
23
|
+
]);
|
|
24
|
+
const r2WriteErrors = sqliteTable("r2_write_errors", {
|
|
25
|
+
id: text("id").primaryKey(),
|
|
26
|
+
userId: integer("user_id").notNull(),
|
|
27
|
+
siteId: text("site_id"),
|
|
28
|
+
table: text("table"),
|
|
29
|
+
date: text("date"),
|
|
30
|
+
error: text("error").notNull(),
|
|
31
|
+
createdAt: integer("created_at").notNull().default(sql`(unixepoch())`)
|
|
32
|
+
}, (t) => [index("idx_r2_write_errors_user").on(t.userId, t.createdAt), index("idx_r2_write_errors_created").on(t.createdAt)]);
|
|
33
|
+
const r2ShadowDiffs = sqliteTable("r2_shadow_diffs", {
|
|
34
|
+
id: text("id").primaryKey(),
|
|
35
|
+
userId: integer("user_id").notNull(),
|
|
36
|
+
siteId: text("site_id"),
|
|
37
|
+
endpoint: text("endpoint").notNull(),
|
|
38
|
+
diff: text("diff").notNull(),
|
|
39
|
+
createdAt: integer("created_at").notNull().default(sql`(unixepoch())`)
|
|
40
|
+
}, (t) => [index("idx_r2_shadow_diffs_user").on(t.userId, t.createdAt), index("idx_r2_shadow_diffs_endpoint").on(t.endpoint, t.createdAt)]);
|
|
41
|
+
const r2Locks = sqliteTable("r2_locks", {
|
|
42
|
+
scope: text("scope").primaryKey(),
|
|
43
|
+
holderId: text("holder_id").notNull(),
|
|
44
|
+
acquiredAt: integer("acquired_at").notNull(),
|
|
45
|
+
expiresAt: integer("expires_at").notNull()
|
|
46
|
+
}, (t) => [index("idx_r2_locks_expires").on(t.expiresAt)]);
|
|
47
|
+
const r2Watermarks = sqliteTable("r2_watermarks", {
|
|
48
|
+
userId: integer("user_id").notNull(),
|
|
49
|
+
siteId: text("site_id").notNull().default(""),
|
|
50
|
+
table: text("table").notNull(),
|
|
51
|
+
newestDateSynced: text("newest_date_synced").notNull(),
|
|
52
|
+
oldestDateSynced: text("oldest_date_synced").notNull(),
|
|
53
|
+
lastSyncAt: integer("last_sync_at").notNull()
|
|
54
|
+
}, (t) => [primaryKey({ columns: [
|
|
55
|
+
t.userId,
|
|
56
|
+
t.siteId,
|
|
57
|
+
t.table
|
|
58
|
+
] })]);
|
|
59
|
+
const r2SyncStates = sqliteTable("r2_sync_states", {
|
|
60
|
+
userId: integer("user_id").notNull(),
|
|
61
|
+
siteId: text("site_id").notNull().default(""),
|
|
62
|
+
table: text("table").notNull(),
|
|
63
|
+
date: text("date").notNull(),
|
|
64
|
+
searchType: text("search_type").notNull().default(""),
|
|
65
|
+
state: text("state", { enum: [
|
|
66
|
+
"pending",
|
|
67
|
+
"inflight",
|
|
68
|
+
"done",
|
|
69
|
+
"failed"
|
|
70
|
+
] }).notNull(),
|
|
71
|
+
updatedAt: integer("updated_at").notNull(),
|
|
72
|
+
attempts: integer("attempts").notNull().default(0),
|
|
73
|
+
error: text("error")
|
|
74
|
+
}, (t) => [primaryKey({ columns: [
|
|
75
|
+
t.userId,
|
|
76
|
+
t.siteId,
|
|
77
|
+
t.table,
|
|
78
|
+
t.date,
|
|
79
|
+
t.searchType
|
|
80
|
+
] }), index("idx_r2_sync_states_state").on(t.state)]);
|
|
81
|
+
export { r2Locks, r2Manifest, r2ShadowDiffs, r2SyncStates, r2Watermarks, r2WriteErrors };
|
|
@@ -0,0 +1,11 @@
|
|
|
1
|
+
import { r2Locks, r2Manifest, r2SyncStates, r2Watermarks } from "./r2-manifest-schema.mjs";
|
|
2
|
+
import { ManifestStore } from "@gscdump/engine";
|
|
3
|
+
import { DrizzleD1Database } from "drizzle-orm/d1";
|
|
4
|
+
type AnalyticsManifestDb = DrizzleD1Database<{
|
|
5
|
+
r2Manifest: typeof r2Manifest;
|
|
6
|
+
r2Locks: typeof r2Locks;
|
|
7
|
+
r2SyncStates: typeof r2SyncStates;
|
|
8
|
+
r2Watermarks: typeof r2Watermarks;
|
|
9
|
+
}>;
|
|
10
|
+
declare function createD1ManifestStore(db: AnalyticsManifestDb): ManifestStore;
|
|
11
|
+
export { AnalyticsManifestDb, createD1ManifestStore };
|
|
@@ -0,0 +1,282 @@
|
|
|
1
|
+
import { r2Locks, r2Manifest, r2SyncStates, r2Watermarks } from "./r2-manifest-schema.mjs";
|
|
2
|
+
import { and, eq, inArray, isNotNull, isNull, lt, lte, or, sql } from "drizzle-orm";
|
|
3
|
+
import { inferSearchType } from "@gscdump/engine";
|
|
4
|
+
import { engineErrorToException, engineErrors } from "@gscdump/engine/errors";
|
|
5
|
+
function toRow(e) {
|
|
6
|
+
return {
|
|
7
|
+
id: e.objectKey,
|
|
8
|
+
userId: Number(e.userId),
|
|
9
|
+
siteId: e.siteId ?? null,
|
|
10
|
+
table: e.table,
|
|
11
|
+
partition: e.partition,
|
|
12
|
+
objectKey: e.objectKey,
|
|
13
|
+
rowCount: e.rowCount,
|
|
14
|
+
bytes: e.bytes,
|
|
15
|
+
createdAt: e.createdAt,
|
|
16
|
+
retiredAt: e.retiredAt ?? null,
|
|
17
|
+
tier: e.tier ?? null,
|
|
18
|
+
searchType: e.searchType ?? null,
|
|
19
|
+
schemaVersion: e.schemaVersion ?? null
|
|
20
|
+
};
|
|
21
|
+
}
|
|
22
|
+
function fromRow(r) {
|
|
23
|
+
return {
|
|
24
|
+
userId: String(r.userId),
|
|
25
|
+
siteId: r.siteId ?? void 0,
|
|
26
|
+
table: r.table,
|
|
27
|
+
partition: r.partition,
|
|
28
|
+
objectKey: r.objectKey,
|
|
29
|
+
rowCount: r.rowCount,
|
|
30
|
+
bytes: r.bytes,
|
|
31
|
+
createdAt: r.createdAt,
|
|
32
|
+
retiredAt: r.retiredAt ?? void 0,
|
|
33
|
+
...r.tier !== null ? { tier: r.tier } : {},
|
|
34
|
+
...r.searchType !== null ? { searchType: r.searchType } : {},
|
|
35
|
+
...r.schemaVersion !== null ? { schemaVersion: r.schemaVersion } : {}
|
|
36
|
+
};
|
|
37
|
+
}
|
|
38
|
+
const siteIdOf = (s) => s ?? "";
|
|
39
|
+
const searchTypeOf = (s) => s === void 0 || s === "web" ? "" : s;
|
|
40
|
+
const LOCK_TTL_MS = 3e4;
|
|
41
|
+
const LOCK_ACQUIRE_TIMEOUT_MS = 5e3;
|
|
42
|
+
const LOCK_RETRY_MIN_MS = 25;
|
|
43
|
+
const LOCK_RETRY_MAX_MS = 150;
|
|
44
|
+
const D1_BATCH_LIMIT = 95;
|
|
45
|
+
const D1_IN_CLAUSE_CHUNK = 90;
|
|
46
|
+
function lockScopeKey(scope) {
|
|
47
|
+
return `${scope.userId}|${siteIdOf(scope.siteId)}|${scope.table}|${scope.partition}`;
|
|
48
|
+
}
|
|
49
|
+
function jitterDelay() {
|
|
50
|
+
return LOCK_RETRY_MIN_MS + Math.floor(Math.random() * (LOCK_RETRY_MAX_MS - LOCK_RETRY_MIN_MS));
|
|
51
|
+
}
|
|
52
|
+
function tierMatchCond(target) {
|
|
53
|
+
const explicit = eq(r2Manifest.tier, target);
|
|
54
|
+
if (target === "raw") return or(explicit, and(isNull(r2Manifest.tier), sql`${r2Manifest.partition} LIKE 'daily/%'`));
|
|
55
|
+
if (target === "d30") return or(explicit, and(isNull(r2Manifest.tier), sql`${r2Manifest.partition} LIKE 'monthly/%'`));
|
|
56
|
+
return explicit;
|
|
57
|
+
}
|
|
58
|
+
function createD1ManifestStore(db) {
|
|
59
|
+
async function listByFilter(filter, liveOnly) {
|
|
60
|
+
const baseConds = [eq(r2Manifest.userId, Number(filter.userId))];
|
|
61
|
+
if (liveOnly) baseConds.push(isNull(r2Manifest.retiredAt));
|
|
62
|
+
if (filter.siteId !== void 0) baseConds.push(eq(r2Manifest.siteId, filter.siteId));
|
|
63
|
+
if (filter.table !== void 0) baseConds.push(eq(r2Manifest.table, filter.table));
|
|
64
|
+
if (filter.tier !== void 0) {
|
|
65
|
+
const cond = tierMatchCond(filter.tier);
|
|
66
|
+
if (cond) baseConds.push(cond);
|
|
67
|
+
}
|
|
68
|
+
if (filter.searchType !== void 0) baseConds.push(filter.searchType === "web" ? or(eq(r2Manifest.searchType, "web"), isNull(r2Manifest.searchType)) : eq(r2Manifest.searchType, filter.searchType));
|
|
69
|
+
const PARTITION_CHUNK = 80;
|
|
70
|
+
if (filter.partitions) {
|
|
71
|
+
if (filter.partitions.length === 0) return [];
|
|
72
|
+
const statements = [];
|
|
73
|
+
for (let i = 0; i < filter.partitions.length; i += PARTITION_CHUNK) {
|
|
74
|
+
const slice = filter.partitions.slice(i, i + PARTITION_CHUNK);
|
|
75
|
+
const conds = [...baseConds, inArray(r2Manifest.partition, slice)];
|
|
76
|
+
statements.push(db.select().from(r2Manifest).where(and(...conds)));
|
|
77
|
+
}
|
|
78
|
+
return (await db.batch(statements)).flat().map((r) => fromRow(r));
|
|
79
|
+
}
|
|
80
|
+
return (await db.select().from(r2Manifest).where(and(...baseConds))).map(fromRow);
|
|
81
|
+
}
|
|
82
|
+
const listLive = (filter) => listByFilter(filter, true);
|
|
83
|
+
const listAll = (filter) => listByFilter(filter, false);
|
|
84
|
+
async function registerVersions(newEntries, superseding) {
|
|
85
|
+
const supersededAt = newEntries[0]?.createdAt ?? Date.now();
|
|
86
|
+
const statements = [];
|
|
87
|
+
if (superseding && superseding.length > 0) {
|
|
88
|
+
const keys = superseding.map((s) => s.objectKey);
|
|
89
|
+
for (let i = 0; i < keys.length; i += D1_IN_CLAUSE_CHUNK) {
|
|
90
|
+
const slice = keys.slice(i, i + D1_IN_CLAUSE_CHUNK);
|
|
91
|
+
statements.push(db.update(r2Manifest).set({ retiredAt: supersededAt }).where(and(inArray(r2Manifest.objectKey, slice), isNull(r2Manifest.retiredAt))));
|
|
92
|
+
}
|
|
93
|
+
}
|
|
94
|
+
for (const e of newEntries) statements.push(db.insert(r2Manifest).values(toRow(e)).onConflictDoUpdate({
|
|
95
|
+
target: r2Manifest.objectKey,
|
|
96
|
+
set: {
|
|
97
|
+
userId: sql`excluded.user_id`,
|
|
98
|
+
siteId: sql`excluded.site_id`,
|
|
99
|
+
table: sql`excluded."table"`,
|
|
100
|
+
partition: sql`excluded.partition`,
|
|
101
|
+
rowCount: sql`excluded.row_count`,
|
|
102
|
+
bytes: sql`excluded.bytes`,
|
|
103
|
+
createdAt: sql`excluded.created_at`,
|
|
104
|
+
retiredAt: sql`excluded.retired_at`,
|
|
105
|
+
tier: sql`excluded.tier`,
|
|
106
|
+
searchType: sql`excluded.search_type`,
|
|
107
|
+
schemaVersion: sql`excluded.schema_version`
|
|
108
|
+
}
|
|
109
|
+
}));
|
|
110
|
+
if (statements.length === 0) return;
|
|
111
|
+
for (let i = 0; i < statements.length; i += D1_BATCH_LIMIT) {
|
|
112
|
+
const chunk = statements.slice(i, i + D1_BATCH_LIMIT);
|
|
113
|
+
await db.batch(chunk);
|
|
114
|
+
}
|
|
115
|
+
}
|
|
116
|
+
async function listRetired(olderThan) {
|
|
117
|
+
return (await db.select().from(r2Manifest).where(and(isNotNull(r2Manifest.retiredAt), lte(r2Manifest.retiredAt, olderThan)))).map(fromRow);
|
|
118
|
+
}
|
|
119
|
+
async function deleteEntries(entries) {
|
|
120
|
+
if (entries.length === 0) return;
|
|
121
|
+
const keys = entries.map((e) => e.objectKey);
|
|
122
|
+
const statements = [];
|
|
123
|
+
for (let i = 0; i < keys.length; i += D1_IN_CLAUSE_CHUNK) statements.push(db.delete(r2Manifest).where(inArray(r2Manifest.objectKey, keys.slice(i, i + D1_IN_CLAUSE_CHUNK))));
|
|
124
|
+
for (let i = 0; i < statements.length; i += D1_BATCH_LIMIT) await db.batch(statements.slice(i, i + D1_BATCH_LIMIT));
|
|
125
|
+
}
|
|
126
|
+
async function getWatermarks(filter) {
|
|
127
|
+
const conds = [eq(r2Watermarks.userId, Number(filter.userId))];
|
|
128
|
+
if (filter.siteId !== void 0) conds.push(eq(r2Watermarks.siteId, filter.siteId));
|
|
129
|
+
if (filter.table !== void 0) conds.push(eq(r2Watermarks.table, filter.table));
|
|
130
|
+
return (await db.select().from(r2Watermarks).where(and(...conds))).map((r) => ({
|
|
131
|
+
userId: String(r.userId),
|
|
132
|
+
siteId: r.siteId === "" ? void 0 : r.siteId,
|
|
133
|
+
table: r.table,
|
|
134
|
+
newestDateSynced: r.newestDateSynced,
|
|
135
|
+
oldestDateSynced: r.oldestDateSynced,
|
|
136
|
+
lastSyncAt: r.lastSyncAt
|
|
137
|
+
}));
|
|
138
|
+
}
|
|
139
|
+
async function bumpWatermark(scope, date, at) {
|
|
140
|
+
const lastSyncAt = at ?? Date.now();
|
|
141
|
+
await db.insert(r2Watermarks).values({
|
|
142
|
+
userId: Number(scope.userId),
|
|
143
|
+
siteId: siteIdOf(scope.siteId),
|
|
144
|
+
table: scope.table,
|
|
145
|
+
newestDateSynced: date,
|
|
146
|
+
oldestDateSynced: date,
|
|
147
|
+
lastSyncAt
|
|
148
|
+
}).onConflictDoUpdate({
|
|
149
|
+
target: [
|
|
150
|
+
r2Watermarks.userId,
|
|
151
|
+
r2Watermarks.siteId,
|
|
152
|
+
r2Watermarks.table
|
|
153
|
+
],
|
|
154
|
+
set: {
|
|
155
|
+
newestDateSynced: sql`CASE WHEN excluded.newest_date_synced > newest_date_synced THEN excluded.newest_date_synced ELSE newest_date_synced END`,
|
|
156
|
+
oldestDateSynced: sql`CASE WHEN excluded.oldest_date_synced < oldest_date_synced THEN excluded.oldest_date_synced ELSE oldest_date_synced END`,
|
|
157
|
+
lastSyncAt: sql`CASE WHEN excluded.last_sync_at > last_sync_at THEN excluded.last_sync_at ELSE last_sync_at END`
|
|
158
|
+
}
|
|
159
|
+
}).run();
|
|
160
|
+
}
|
|
161
|
+
async function getSyncStates(filter) {
|
|
162
|
+
const conds = [eq(r2SyncStates.userId, Number(filter.userId))];
|
|
163
|
+
if (filter.siteId !== void 0) conds.push(eq(r2SyncStates.siteId, filter.siteId));
|
|
164
|
+
if (filter.table !== void 0) conds.push(eq(r2SyncStates.table, filter.table));
|
|
165
|
+
if (filter.state !== void 0) conds.push(eq(r2SyncStates.state, filter.state));
|
|
166
|
+
if (filter.searchType !== void 0) conds.push(eq(r2SyncStates.searchType, searchTypeOf(filter.searchType)));
|
|
167
|
+
return (await db.select().from(r2SyncStates).where(and(...conds))).map((r) => ({
|
|
168
|
+
userId: String(r.userId),
|
|
169
|
+
siteId: r.siteId === "" ? void 0 : r.siteId,
|
|
170
|
+
table: r.table,
|
|
171
|
+
date: r.date,
|
|
172
|
+
searchType: inferSearchType({ searchType: r.searchType === "" ? void 0 : r.searchType }),
|
|
173
|
+
state: r.state,
|
|
174
|
+
updatedAt: r.updatedAt,
|
|
175
|
+
attempts: r.attempts,
|
|
176
|
+
error: r.error ?? void 0
|
|
177
|
+
}));
|
|
178
|
+
}
|
|
179
|
+
async function setSyncState(scope, state, detail) {
|
|
180
|
+
const updatedAt = detail?.at ?? Date.now();
|
|
181
|
+
const errorText = detail?.error ?? null;
|
|
182
|
+
await db.insert(r2SyncStates).values({
|
|
183
|
+
userId: Number(scope.userId),
|
|
184
|
+
siteId: siteIdOf(scope.siteId),
|
|
185
|
+
table: scope.table,
|
|
186
|
+
date: scope.date,
|
|
187
|
+
searchType: searchTypeOf(scope.searchType),
|
|
188
|
+
state,
|
|
189
|
+
updatedAt,
|
|
190
|
+
attempts: state === "inflight" ? 1 : 0,
|
|
191
|
+
error: errorText
|
|
192
|
+
}).onConflictDoUpdate({
|
|
193
|
+
target: [
|
|
194
|
+
r2SyncStates.userId,
|
|
195
|
+
r2SyncStates.siteId,
|
|
196
|
+
r2SyncStates.table,
|
|
197
|
+
r2SyncStates.date,
|
|
198
|
+
r2SyncStates.searchType
|
|
199
|
+
],
|
|
200
|
+
set: {
|
|
201
|
+
state: sql`excluded.state`,
|
|
202
|
+
updatedAt: sql`excluded.updated_at`,
|
|
203
|
+
attempts: sql`CASE WHEN excluded.state = 'inflight' THEN attempts + 1 ELSE attempts END`,
|
|
204
|
+
error: sql`CASE
|
|
205
|
+
WHEN excluded.state = 'done' THEN NULL
|
|
206
|
+
WHEN excluded.state = 'inflight' THEN error
|
|
207
|
+
ELSE excluded.error
|
|
208
|
+
END`
|
|
209
|
+
}
|
|
210
|
+
}).run();
|
|
211
|
+
}
|
|
212
|
+
async function withLock(scope, fn) {
|
|
213
|
+
const key = lockScopeKey(scope);
|
|
214
|
+
const holderId = crypto.randomUUID();
|
|
215
|
+
const deadline = Date.now() + LOCK_ACQUIRE_TIMEOUT_MS;
|
|
216
|
+
while (true) {
|
|
217
|
+
const now = Date.now();
|
|
218
|
+
const expiresAt = now + LOCK_TTL_MS;
|
|
219
|
+
await db.insert(r2Locks).values({
|
|
220
|
+
scope: key,
|
|
221
|
+
holderId,
|
|
222
|
+
acquiredAt: now,
|
|
223
|
+
expiresAt
|
|
224
|
+
}).onConflictDoUpdate({
|
|
225
|
+
target: r2Locks.scope,
|
|
226
|
+
set: {
|
|
227
|
+
holderId: sql`excluded.holder_id`,
|
|
228
|
+
acquiredAt: sql`excluded.acquired_at`,
|
|
229
|
+
expiresAt: sql`excluded.expires_at`
|
|
230
|
+
},
|
|
231
|
+
setWhere: lt(r2Locks.expiresAt, now)
|
|
232
|
+
}).run();
|
|
233
|
+
if ((await db.select({ holderId: r2Locks.holderId }).from(r2Locks).where(eq(r2Locks.scope, key)).get())?.holderId === holderId) break;
|
|
234
|
+
if (Date.now() >= deadline) throw engineErrorToException(engineErrors.lockAcquireTimeout(key, LOCK_ACQUIRE_TIMEOUT_MS));
|
|
235
|
+
await new Promise((resolve) => setTimeout(resolve, jitterDelay()));
|
|
236
|
+
}
|
|
237
|
+
return await fn().finally(async () => {
|
|
238
|
+
await Promise.resolve(db.delete(r2Locks).where(and(eq(r2Locks.scope, key), eq(r2Locks.holderId, holderId))).run()).catch((error) => {
|
|
239
|
+
console.warn(`[gscdump/engine-sqlite] failed to release manifest lock ${key}; lease will expire`, error);
|
|
240
|
+
});
|
|
241
|
+
});
|
|
242
|
+
}
|
|
243
|
+
async function purgeTenant(filter) {
|
|
244
|
+
const userIdNum = Number(filter.userId);
|
|
245
|
+
const entriesCond = filter.siteId !== void 0 ? and(eq(r2Manifest.userId, userIdNum), eq(r2Manifest.siteId, filter.siteId)) : eq(r2Manifest.userId, userIdNum);
|
|
246
|
+
const watermarksCond = filter.siteId !== void 0 ? and(eq(r2Watermarks.userId, userIdNum), eq(r2Watermarks.siteId, filter.siteId)) : eq(r2Watermarks.userId, userIdNum);
|
|
247
|
+
const syncStatesCond = filter.siteId !== void 0 ? and(eq(r2SyncStates.userId, userIdNum), eq(r2SyncStates.siteId, filter.siteId)) : eq(r2SyncStates.userId, userIdNum);
|
|
248
|
+
const [entriesRows, watermarkRows, syncStateRows] = await Promise.all([
|
|
249
|
+
db.select({ n: sql`count(*)` }).from(r2Manifest).where(entriesCond).all(),
|
|
250
|
+
db.select({ n: sql`count(*)` }).from(r2Watermarks).where(watermarksCond).all(),
|
|
251
|
+
db.select({ n: sql`count(*)` }).from(r2SyncStates).where(syncStatesCond).all()
|
|
252
|
+
]);
|
|
253
|
+
const entriesRemoved = Number(entriesRows[0]?.n ?? 0);
|
|
254
|
+
const watermarksRemoved = Number(watermarkRows[0]?.n ?? 0);
|
|
255
|
+
const syncStatesRemoved = Number(syncStateRows[0]?.n ?? 0);
|
|
256
|
+
await db.batch([
|
|
257
|
+
db.delete(r2Manifest).where(entriesCond),
|
|
258
|
+
db.delete(r2Watermarks).where(watermarksCond),
|
|
259
|
+
db.delete(r2SyncStates).where(syncStatesCond)
|
|
260
|
+
]);
|
|
261
|
+
return {
|
|
262
|
+
entriesRemoved,
|
|
263
|
+
watermarksRemoved,
|
|
264
|
+
syncStatesRemoved
|
|
265
|
+
};
|
|
266
|
+
}
|
|
267
|
+
return {
|
|
268
|
+
listLive,
|
|
269
|
+
listAll,
|
|
270
|
+
registerVersion: (entry, superseding) => registerVersions([entry], superseding),
|
|
271
|
+
registerVersions,
|
|
272
|
+
listRetired,
|
|
273
|
+
delete: deleteEntries,
|
|
274
|
+
getWatermarks,
|
|
275
|
+
bumpWatermark,
|
|
276
|
+
getSyncStates,
|
|
277
|
+
setSyncState,
|
|
278
|
+
withLock,
|
|
279
|
+
purgeTenant
|
|
280
|
+
};
|
|
281
|
+
}
|
|
282
|
+
export { createD1ManifestStore };
|
|
@@ -0,0 +1,27 @@
|
|
|
1
|
+
import { schema } from "./schema.mjs";
|
|
2
|
+
import { SqliteRowExecutor } from "./runner.mjs";
|
|
3
|
+
import { ResolverAdapter } from "@gscdump/engine/resolver";
|
|
4
|
+
type TableKey = keyof typeof schema;
|
|
5
|
+
interface CreateSqliteResolverAdapterOptions {
|
|
6
|
+
regex?: boolean;
|
|
7
|
+
}
|
|
8
|
+
declare function createSqliteResolverAdapter(options?: CreateSqliteResolverAdapterOptions): ResolverAdapter<TableKey>;
|
|
9
|
+
declare const sqliteResolverAdapter: ResolverAdapter<"gsc_countries" | "gsc_devices" | "gsc_hourly_pages" | "gsc_keywords" | "gsc_page_keywords" | "gsc_pages" | "gsc_query_dim" | "gsc_search_appearance" | "gsc_search_appearance_page_queries" | "gsc_search_appearance_pages" | "gsc_search_appearance_queries">;
|
|
10
|
+
/**
|
|
11
|
+
* One-shot probe that detects whether the host SQLite has a working REGEXP
|
|
12
|
+
* function. SQLite ships REGEXP as a no-op stub by default; D1, libsql, and
|
|
13
|
+
* sqlite3 with the regexp extension register a real implementation. Probe
|
|
14
|
+
* once at adapter construction so per-query plan-time gating is honest.
|
|
15
|
+
*/
|
|
16
|
+
declare function probeSqliteRegex(executor: SqliteRowExecutor): Promise<boolean>;
|
|
17
|
+
interface CreateSqliteResolverAdapterFromExecutorOptions {
|
|
18
|
+
executor: SqliteRowExecutor;
|
|
19
|
+
}
|
|
20
|
+
/**
|
|
21
|
+
* Async factory that probes REGEXP availability on `executor` and returns
|
|
22
|
+
* a resolver adapter with `capabilities.regex` set accordingly. Prefer over
|
|
23
|
+
* the manual `createSqliteResolverAdapter({ regex })` whenever you have
|
|
24
|
+
* an executor handy.
|
|
25
|
+
*/
|
|
26
|
+
declare function createSqliteResolverAdapterFromExecutor(options: CreateSqliteResolverAdapterFromExecutorOptions): Promise<ResolverAdapter<TableKey>>;
|
|
27
|
+
export { CreateSqliteResolverAdapterFromExecutorOptions, CreateSqliteResolverAdapterOptions, TableKey, createSqliteResolverAdapter, createSqliteResolverAdapterFromExecutor, probeSqliteRegex, sqliteResolverAdapter };
|
|
@@ -0,0 +1,41 @@
|
|
|
1
|
+
import { schema } from "./schema.mjs";
|
|
2
|
+
import { compileSqlite } from "./runner.mjs";
|
|
3
|
+
import { sql } from "drizzle-orm";
|
|
4
|
+
import { createResolverAdapter } from "@gscdump/engine/resolver";
|
|
5
|
+
function createSqliteResolverAdapter(options = {}) {
|
|
6
|
+
return createResolverAdapter({
|
|
7
|
+
schema,
|
|
8
|
+
datasetToTableKey: {
|
|
9
|
+
pages: "gsc_pages",
|
|
10
|
+
queries: "gsc_keywords",
|
|
11
|
+
page_queries: "gsc_page_keywords",
|
|
12
|
+
countries: "gsc_countries",
|
|
13
|
+
dates: "gsc_devices",
|
|
14
|
+
search_appearance: "gsc_search_appearance",
|
|
15
|
+
search_appearance_pages: "gsc_search_appearance_pages",
|
|
16
|
+
search_appearance_queries: "gsc_search_appearance_queries",
|
|
17
|
+
search_appearance_page_queries: "gsc_search_appearance_page_queries",
|
|
18
|
+
hourly_pages: "gsc_hourly_pages"
|
|
19
|
+
},
|
|
20
|
+
metricCast: "REAL",
|
|
21
|
+
regexPredicate: (expr, pattern, negate) => negate ? sql`NOT (${expr} REGEXP ${pattern})` : sql`${expr} REGEXP ${pattern}`,
|
|
22
|
+
tableLabel: "sqlite/resolver-adapter",
|
|
23
|
+
includeSiteId: true,
|
|
24
|
+
queryDimTableRef: () => sql.raw("\"gsc_query_dim\" AS \"query_dim\""),
|
|
25
|
+
queryDimSiteScoped: true,
|
|
26
|
+
compile: compileSqlite,
|
|
27
|
+
capabilities: {
|
|
28
|
+
regex: options.regex ?? false,
|
|
29
|
+
comparisonJoin: true,
|
|
30
|
+
windowTotals: true
|
|
31
|
+
}
|
|
32
|
+
});
|
|
33
|
+
}
|
|
34
|
+
const sqliteResolverAdapter = createSqliteResolverAdapter();
|
|
35
|
+
async function probeSqliteRegex(executor) {
|
|
36
|
+
return executor(`SELECT 'a' REGEXP 'a' AS r`, [], "all").then(({ rows }) => rows.length > 0).catch(() => false);
|
|
37
|
+
}
|
|
38
|
+
async function createSqliteResolverAdapterFromExecutor(options) {
|
|
39
|
+
return createSqliteResolverAdapter({ regex: await probeSqliteRegex(options.executor) });
|
|
40
|
+
}
|
|
41
|
+
export { createSqliteResolverAdapter, createSqliteResolverAdapterFromExecutor, probeSqliteRegex, sqliteResolverAdapter };
|
|
@@ -0,0 +1,42 @@
|
|
|
1
|
+
import { Schema } from "./schema.mjs";
|
|
2
|
+
import { SQL } from "drizzle-orm";
|
|
3
|
+
import { ScopedRunnerOptions, TableScope } from "@gscdump/engine/scope";
|
|
4
|
+
import { SqliteRemoteDatabase } from "drizzle-orm/sqlite-proxy";
|
|
5
|
+
declare function compileSqlite(query: SQL): {
|
|
6
|
+
sql: string;
|
|
7
|
+
params: unknown[];
|
|
8
|
+
};
|
|
9
|
+
type SqliteRowExecutor = (sql: string, params: unknown[], method: 'run' | 'all' | 'values' | 'get') => Promise<{
|
|
10
|
+
rows: unknown[];
|
|
11
|
+
}>;
|
|
12
|
+
interface SqliteInsightRunnerOptions<TSchema extends Record<string, unknown> = Schema> {
|
|
13
|
+
executor: SqliteRowExecutor;
|
|
14
|
+
logger?: boolean;
|
|
15
|
+
/**
|
|
16
|
+
* Convert object rows to positional arrays before handing to drizzle.
|
|
17
|
+
* D1 HTTP + most REST-style drivers return objects; set true when the
|
|
18
|
+
* executor yields `{ col: value }` rows.
|
|
19
|
+
*/
|
|
20
|
+
rowsAsArrays?: boolean;
|
|
21
|
+
/**
|
|
22
|
+
* Override the bundled gsc_* schema. Pass an extended schema (superset
|
|
23
|
+
* of the upstream tables) when the consumer DB carries additional tables
|
|
24
|
+
* like sitemaps or indexing state. Defaults to the bundled schema.
|
|
25
|
+
*/
|
|
26
|
+
schema?: TSchema;
|
|
27
|
+
}
|
|
28
|
+
interface SqliteInsightRunner<TSchema extends Record<string, unknown> = Schema> {
|
|
29
|
+
db: SqliteRemoteDatabase<TSchema>;
|
|
30
|
+
}
|
|
31
|
+
declare function createSqliteInsightRunner<TSchema extends Record<string, unknown> = Schema>(opts: SqliteInsightRunnerOptions<TSchema>): SqliteInsightRunner<TSchema>;
|
|
32
|
+
/**
|
|
33
|
+
* Per-table predicate set from {siteId, window}. Unlike the /browser
|
|
34
|
+
* variant, every sqlite table has `site_id` so the siteId predicate is
|
|
35
|
+
* always emitted when supplied.
|
|
36
|
+
*
|
|
37
|
+
* Date bounds come from either a resolved window or bare startDate/endDate
|
|
38
|
+
* (one or both). This lets callers pass unbounded query-string params
|
|
39
|
+
* through without first normalizing them.
|
|
40
|
+
*/
|
|
41
|
+
declare const scopeFor: (table: "gsc_countries" | "gsc_devices" | "gsc_hourly_pages" | "gsc_keywords" | "gsc_page_keywords" | "gsc_pages" | "gsc_query_dim" | "gsc_search_appearance" | "gsc_search_appearance_page_queries" | "gsc_search_appearance_pages" | "gsc_search_appearance_queries", opts: ScopedRunnerOptions) => TableScope, mergeScope: typeof import("@gscdump/engine/scope").mergeScope;
|
|
42
|
+
export { type ScopedRunnerOptions, SqliteInsightRunner, SqliteInsightRunnerOptions, SqliteRowExecutor, type TableScope, compileSqlite, createSqliteInsightRunner, mergeScope, scopeFor };
|
package/dist/runner.mjs
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
1
|
+
import { schema } from "./schema.mjs";
|
|
2
|
+
import { createScopedHelpers } from "@gscdump/engine/scope";
|
|
3
|
+
import { SQLiteAsyncDialect } from "drizzle-orm/sqlite-core";
|
|
4
|
+
import { drizzle } from "drizzle-orm/sqlite-proxy";
|
|
5
|
+
const sqliteDialect = new SQLiteAsyncDialect();
|
|
6
|
+
function compileSqlite(query) {
|
|
7
|
+
const compiled = sqliteDialect.sqlToQuery(query);
|
|
8
|
+
return {
|
|
9
|
+
sql: compiled.sql,
|
|
10
|
+
params: compiled.params
|
|
11
|
+
};
|
|
12
|
+
}
|
|
13
|
+
function createSqliteInsightRunner(opts) {
|
|
14
|
+
const { executor, logger, rowsAsArrays, schema: schemaOverride } = opts;
|
|
15
|
+
const callback = async (sql, params, method) => {
|
|
16
|
+
const result = await executor(sql, params, method);
|
|
17
|
+
if (!rowsAsArrays) return { rows: result.rows };
|
|
18
|
+
return { rows: result.rows.map((r) => {
|
|
19
|
+
if (Array.isArray(r)) return r;
|
|
20
|
+
if (r && typeof r === "object") return Object.values(r);
|
|
21
|
+
return r;
|
|
22
|
+
}) };
|
|
23
|
+
};
|
|
24
|
+
return { db: drizzle(callback, {
|
|
25
|
+
schema: schemaOverride ?? schema,
|
|
26
|
+
logger
|
|
27
|
+
}) };
|
|
28
|
+
}
|
|
29
|
+
const { scopeFor, mergeScope } = createScopedHelpers(schema);
|
|
30
|
+
export { compileSqlite, createSqliteInsightRunner, mergeScope, scopeFor };
|