@bluefields/db 0.2.0
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/LICENSE +202 -0
- package/dist/client.d.ts +6 -0
- package/dist/client.js +13 -0
- package/dist/client.js.map +1 -0
- package/dist/index.d.ts +19 -0
- package/dist/index.js +20 -0
- package/dist/index.js.map +1 -0
- package/dist/migrate.d.ts +11 -0
- package/dist/migrate.js +48 -0
- package/dist/migrate.js.map +1 -0
- package/dist/schema/api-keys.d.ts +145 -0
- package/dist/schema/api-keys.js +18 -0
- package/dist/schema/api-keys.js.map +1 -0
- package/dist/schema/audit-log.d.ts +228 -0
- package/dist/schema/audit-log.js +49 -0
- package/dist/schema/audit-log.js.map +1 -0
- package/dist/schema/cost-records.d.ts +190 -0
- package/dist/schema/cost-records.js +55 -0
- package/dist/schema/cost-records.js.map +1 -0
- package/dist/schema/crawl-job-pages.d.ts +186 -0
- package/dist/schema/crawl-job-pages.js +31 -0
- package/dist/schema/crawl-job-pages.js.map +1 -0
- package/dist/schema/crawl-jobs.d.ts +271 -0
- package/dist/schema/crawl-jobs.js +40 -0
- package/dist/schema/crawl-jobs.js.map +1 -0
- package/dist/schema/customers.d.ts +145 -0
- package/dist/schema/customers.js +19 -0
- package/dist/schema/customers.js.map +1 -0
- package/dist/schema/event-feedback.d.ts +170 -0
- package/dist/schema/event-feedback.js +35 -0
- package/dist/schema/event-feedback.js.map +1 -0
- package/dist/schema/events.d.ts +291 -0
- package/dist/schema/events.js +58 -0
- package/dist/schema/events.js.map +1 -0
- package/dist/schema/extractions.d.ts +327 -0
- package/dist/schema/extractions.js +56 -0
- package/dist/schema/extractions.js.map +1 -0
- package/dist/schema/host-extractors.d.ts +131 -0
- package/dist/schema/host-extractors.js +18 -0
- package/dist/schema/host-extractors.js.map +1 -0
- package/dist/schema/host-records.d.ts +237 -0
- package/dist/schema/host-records.js +42 -0
- package/dist/schema/host-records.js.map +1 -0
- package/dist/schema/notification-preferences.d.ts +195 -0
- package/dist/schema/notification-preferences.js +49 -0
- package/dist/schema/notification-preferences.js.map +1 -0
- package/dist/schema/scrape-cache.d.ts +136 -0
- package/dist/schema/scrape-cache.js +23 -0
- package/dist/schema/scrape-cache.js.map +1 -0
- package/dist/schema/sessions.d.ts +153 -0
- package/dist/schema/sessions.js +27 -0
- package/dist/schema/sessions.js.map +1 -0
- package/dist/schema/snapshots.d.ts +382 -0
- package/dist/schema/snapshots.js +60 -0
- package/dist/schema/snapshots.js.map +1 -0
- package/dist/schema/subscriptions.d.ts +604 -0
- package/dist/schema/subscriptions.js +79 -0
- package/dist/schema/subscriptions.js.map +1 -0
- package/dist/schema/usage-periods.d.ts +446 -0
- package/dist/schema/usage-periods.js +71 -0
- package/dist/schema/usage-periods.js.map +1 -0
- package/dist/tsconfig.tsbuildinfo +1 -0
- package/package.json +54 -0
- package/src/migrations/0001_initial.sql +398 -0
- package/src/migrations/0002_pg_cron.sql +109 -0
- package/src/migrations/0003_webhook_secret_cleanup.sql +55 -0
- package/src/migrations/0004_events_pg_notify.sql +39 -0
- package/src/migrations/0005_subscription_consecutive_failures.sql +19 -0
- package/src/migrations/0006_notification_preferences.sql +27 -0
- package/src/migrations/0007_crawl_jobs.sql +65 -0
- package/src/migrations/0008_sessions.sql +38 -0
- package/src/migrations/0009_scrape_cache.sql +34 -0
- package/src/migrations/0010_neon_defaults.sql +99 -0
- package/src/migrations/0011_credits_consumed_numeric.sql +12 -0
- package/src/migrations/0012_crawl_page_attestation.sql +10 -0
- package/src/migrations/0013_customer_deleted_at.sql +5 -0
|
@@ -0,0 +1,186 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* crawl_job_pages — per-page results within a crawl job.
|
|
3
|
+
*
|
|
4
|
+
* Composite PK on (job_id, url_canonical) gives us BFS dedup at the DB
|
|
5
|
+
* level — even if the worker double-discovers a URL, we can't insert it
|
|
6
|
+
* twice.
|
|
7
|
+
*/
|
|
8
|
+
export declare const crawlJobPages: import("drizzle-orm/pg-core").PgTableWithColumns<{
|
|
9
|
+
name: "crawl_job_pages";
|
|
10
|
+
schema: undefined;
|
|
11
|
+
columns: {
|
|
12
|
+
jobId: import("drizzle-orm/pg-core").PgColumn<{
|
|
13
|
+
name: "job_id";
|
|
14
|
+
tableName: "crawl_job_pages";
|
|
15
|
+
dataType: "string";
|
|
16
|
+
columnType: "PgUUID";
|
|
17
|
+
data: string;
|
|
18
|
+
driverParam: string;
|
|
19
|
+
notNull: true;
|
|
20
|
+
hasDefault: false;
|
|
21
|
+
isPrimaryKey: false;
|
|
22
|
+
isAutoincrement: false;
|
|
23
|
+
hasRuntimeDefault: false;
|
|
24
|
+
enumValues: undefined;
|
|
25
|
+
baseColumn: never;
|
|
26
|
+
identity: undefined;
|
|
27
|
+
generated: undefined;
|
|
28
|
+
}, {}, {}>;
|
|
29
|
+
urlCanonical: import("drizzle-orm/pg-core").PgColumn<{
|
|
30
|
+
name: "url_canonical";
|
|
31
|
+
tableName: "crawl_job_pages";
|
|
32
|
+
dataType: "string";
|
|
33
|
+
columnType: "PgText";
|
|
34
|
+
data: string;
|
|
35
|
+
driverParam: string;
|
|
36
|
+
notNull: true;
|
|
37
|
+
hasDefault: false;
|
|
38
|
+
isPrimaryKey: false;
|
|
39
|
+
isAutoincrement: false;
|
|
40
|
+
hasRuntimeDefault: false;
|
|
41
|
+
enumValues: [string, ...string[]];
|
|
42
|
+
baseColumn: never;
|
|
43
|
+
identity: undefined;
|
|
44
|
+
generated: undefined;
|
|
45
|
+
}, {}, {}>;
|
|
46
|
+
depth: import("drizzle-orm/pg-core").PgColumn<{
|
|
47
|
+
name: "depth";
|
|
48
|
+
tableName: "crawl_job_pages";
|
|
49
|
+
dataType: "number";
|
|
50
|
+
columnType: "PgInteger";
|
|
51
|
+
data: number;
|
|
52
|
+
driverParam: string | number;
|
|
53
|
+
notNull: false;
|
|
54
|
+
hasDefault: false;
|
|
55
|
+
isPrimaryKey: false;
|
|
56
|
+
isAutoincrement: false;
|
|
57
|
+
hasRuntimeDefault: false;
|
|
58
|
+
enumValues: undefined;
|
|
59
|
+
baseColumn: never;
|
|
60
|
+
identity: undefined;
|
|
61
|
+
generated: undefined;
|
|
62
|
+
}, {}, {}>;
|
|
63
|
+
contentHash: import("drizzle-orm/pg-core").PgColumn<{
|
|
64
|
+
name: "content_hash";
|
|
65
|
+
tableName: "crawl_job_pages";
|
|
66
|
+
dataType: "string";
|
|
67
|
+
columnType: "PgText";
|
|
68
|
+
data: string;
|
|
69
|
+
driverParam: string;
|
|
70
|
+
notNull: false;
|
|
71
|
+
hasDefault: false;
|
|
72
|
+
isPrimaryKey: false;
|
|
73
|
+
isAutoincrement: false;
|
|
74
|
+
hasRuntimeDefault: false;
|
|
75
|
+
enumValues: [string, ...string[]];
|
|
76
|
+
baseColumn: never;
|
|
77
|
+
identity: undefined;
|
|
78
|
+
generated: undefined;
|
|
79
|
+
}, {}, {}>;
|
|
80
|
+
responseStatus: import("drizzle-orm/pg-core").PgColumn<{
|
|
81
|
+
name: "response_status";
|
|
82
|
+
tableName: "crawl_job_pages";
|
|
83
|
+
dataType: "number";
|
|
84
|
+
columnType: "PgInteger";
|
|
85
|
+
data: number;
|
|
86
|
+
driverParam: string | number;
|
|
87
|
+
notNull: false;
|
|
88
|
+
hasDefault: false;
|
|
89
|
+
isPrimaryKey: false;
|
|
90
|
+
isAutoincrement: false;
|
|
91
|
+
hasRuntimeDefault: false;
|
|
92
|
+
enumValues: undefined;
|
|
93
|
+
baseColumn: never;
|
|
94
|
+
identity: undefined;
|
|
95
|
+
generated: undefined;
|
|
96
|
+
}, {}, {}>;
|
|
97
|
+
fetchedAt: import("drizzle-orm/pg-core").PgColumn<{
|
|
98
|
+
name: "fetched_at";
|
|
99
|
+
tableName: "crawl_job_pages";
|
|
100
|
+
dataType: "date";
|
|
101
|
+
columnType: "PgTimestamp";
|
|
102
|
+
data: Date;
|
|
103
|
+
driverParam: string;
|
|
104
|
+
notNull: true;
|
|
105
|
+
hasDefault: true;
|
|
106
|
+
isPrimaryKey: false;
|
|
107
|
+
isAutoincrement: false;
|
|
108
|
+
hasRuntimeDefault: false;
|
|
109
|
+
enumValues: undefined;
|
|
110
|
+
baseColumn: never;
|
|
111
|
+
identity: undefined;
|
|
112
|
+
generated: undefined;
|
|
113
|
+
}, {}, {}>;
|
|
114
|
+
extractionId: import("drizzle-orm/pg-core").PgColumn<{
|
|
115
|
+
name: "extraction_id";
|
|
116
|
+
tableName: "crawl_job_pages";
|
|
117
|
+
dataType: "string";
|
|
118
|
+
columnType: "PgUUID";
|
|
119
|
+
data: string;
|
|
120
|
+
driverParam: string;
|
|
121
|
+
notNull: false;
|
|
122
|
+
hasDefault: false;
|
|
123
|
+
isPrimaryKey: false;
|
|
124
|
+
isAutoincrement: false;
|
|
125
|
+
hasRuntimeDefault: false;
|
|
126
|
+
enumValues: undefined;
|
|
127
|
+
baseColumn: never;
|
|
128
|
+
identity: undefined;
|
|
129
|
+
generated: undefined;
|
|
130
|
+
}, {}, {}>;
|
|
131
|
+
status: import("drizzle-orm/pg-core").PgColumn<{
|
|
132
|
+
name: "status";
|
|
133
|
+
tableName: "crawl_job_pages";
|
|
134
|
+
dataType: "string";
|
|
135
|
+
columnType: "PgText";
|
|
136
|
+
data: string;
|
|
137
|
+
driverParam: string;
|
|
138
|
+
notNull: true;
|
|
139
|
+
hasDefault: true;
|
|
140
|
+
isPrimaryKey: false;
|
|
141
|
+
isAutoincrement: false;
|
|
142
|
+
hasRuntimeDefault: false;
|
|
143
|
+
enumValues: [string, ...string[]];
|
|
144
|
+
baseColumn: never;
|
|
145
|
+
identity: undefined;
|
|
146
|
+
generated: undefined;
|
|
147
|
+
}, {}, {}>;
|
|
148
|
+
statusDetail: import("drizzle-orm/pg-core").PgColumn<{
|
|
149
|
+
name: "status_detail";
|
|
150
|
+
tableName: "crawl_job_pages";
|
|
151
|
+
dataType: "string";
|
|
152
|
+
columnType: "PgText";
|
|
153
|
+
data: string;
|
|
154
|
+
driverParam: string;
|
|
155
|
+
notNull: false;
|
|
156
|
+
hasDefault: false;
|
|
157
|
+
isPrimaryKey: false;
|
|
158
|
+
isAutoincrement: false;
|
|
159
|
+
hasRuntimeDefault: false;
|
|
160
|
+
enumValues: [string, ...string[]];
|
|
161
|
+
baseColumn: never;
|
|
162
|
+
identity: undefined;
|
|
163
|
+
generated: undefined;
|
|
164
|
+
}, {}, {}>;
|
|
165
|
+
attestation: import("drizzle-orm/pg-core").PgColumn<{
|
|
166
|
+
name: "attestation";
|
|
167
|
+
tableName: "crawl_job_pages";
|
|
168
|
+
dataType: "json";
|
|
169
|
+
columnType: "PgJsonb";
|
|
170
|
+
data: unknown;
|
|
171
|
+
driverParam: unknown;
|
|
172
|
+
notNull: false;
|
|
173
|
+
hasDefault: false;
|
|
174
|
+
isPrimaryKey: false;
|
|
175
|
+
isAutoincrement: false;
|
|
176
|
+
hasRuntimeDefault: false;
|
|
177
|
+
enumValues: undefined;
|
|
178
|
+
baseColumn: never;
|
|
179
|
+
identity: undefined;
|
|
180
|
+
generated: undefined;
|
|
181
|
+
}, {}, {}>;
|
|
182
|
+
};
|
|
183
|
+
dialect: "pg";
|
|
184
|
+
}>;
|
|
185
|
+
export type CrawlJobPage = typeof crawlJobPages.$inferSelect;
|
|
186
|
+
export type NewCrawlJobPage = typeof crawlJobPages.$inferInsert;
|
|
@@ -0,0 +1,31 @@
|
|
|
1
|
+
// SPDX-License-Identifier: Apache-2.0
|
|
2
|
+
import { integer, jsonb, pgTable, primaryKey, text, timestamp, uuid } from 'drizzle-orm/pg-core';
|
|
3
|
+
import { crawlJobs } from './crawl-jobs.js';
|
|
4
|
+
/**
|
|
5
|
+
* crawl_job_pages — per-page results within a crawl job.
|
|
6
|
+
*
|
|
7
|
+
* Composite PK on (job_id, url_canonical) gives us BFS dedup at the DB
|
|
8
|
+
* level — even if the worker double-discovers a URL, we can't insert it
|
|
9
|
+
* twice.
|
|
10
|
+
*/
|
|
11
|
+
export const crawlJobPages = pgTable('crawl_job_pages', {
|
|
12
|
+
jobId: uuid('job_id')
|
|
13
|
+
.notNull()
|
|
14
|
+
.references(() => crawlJobs.id, { onDelete: 'cascade' }),
|
|
15
|
+
urlCanonical: text('url_canonical').notNull(),
|
|
16
|
+
depth: integer('depth'),
|
|
17
|
+
contentHash: text('content_hash'),
|
|
18
|
+
responseStatus: integer('response_status'),
|
|
19
|
+
fetchedAt: timestamp('fetched_at', { withTimezone: true }).notNull().defaultNow(),
|
|
20
|
+
extractionId: uuid('extraction_id'),
|
|
21
|
+
/** 'ok' | 'fetch_failed' | 'extract_failed' | 'excluded' | 'depth_limit' */
|
|
22
|
+
status: text('status').notNull().default('ok'),
|
|
23
|
+
statusDetail: text('status_detail'),
|
|
24
|
+
/**
|
|
25
|
+
* Signed v1 attestation manifest for this page (`{ manifest, signature }`),
|
|
26
|
+
* or null. Populated only when ATTEST_SYNC_ROUTES is enabled at fetch time;
|
|
27
|
+
* nullable + additive so existing rows and the default-off path are unaffected.
|
|
28
|
+
*/
|
|
29
|
+
attestation: jsonb('attestation'),
|
|
30
|
+
}, (table) => [primaryKey({ columns: [table.jobId, table.urlCanonical] })]);
|
|
31
|
+
//# sourceMappingURL=crawl-job-pages.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"crawl-job-pages.js","sourceRoot":"","sources":["../../src/schema/crawl-job-pages.ts"],"names":[],"mappings":"AAAA,sCAAsC;AACtC,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,OAAO,EAAE,UAAU,EAAE,IAAI,EAAE,SAAS,EAAE,IAAI,EAAE,MAAM,qBAAqB,CAAC;AAEjG,OAAO,EAAE,SAAS,EAAE,MAAM,iBAAiB,CAAC;AAE5C;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,aAAa,GAAG,OAAO,CAClC,iBAAiB,EACjB;IACE,KAAK,EAAE,IAAI,CAAC,QAAQ,CAAC;SAClB,OAAO,EAAE;SACT,UAAU,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,EAAE,EAAE,EAAE,QAAQ,EAAE,SAAS,EAAE,CAAC;IAC1D,YAAY,EAAE,IAAI,CAAC,eAAe,CAAC,CAAC,OAAO,EAAE;IAC7C,KAAK,EAAE,OAAO,CAAC,OAAO,CAAC;IACvB,WAAW,EAAE,IAAI,CAAC,cAAc,CAAC;IACjC,cAAc,EAAE,OAAO,CAAC,iBAAiB,CAAC;IAC1C,SAAS,EAAE,SAAS,CAAC,YAAY,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,UAAU,EAAE;IACjF,YAAY,EAAE,IAAI,CAAC,eAAe,CAAC;IACnC,4EAA4E;IAC5E,MAAM,EAAE,IAAI,CAAC,QAAQ,CAAC,CAAC,OAAO,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC;IAC9C,YAAY,EAAE,IAAI,CAAC,eAAe,CAAC;IACnC;;;;OAIG;IACH,WAAW,EAAE,KAAK,CAAC,aAAa,CAAC;CAClC,EACD,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,UAAU,CAAC,EAAE,OAAO,EAAE,CAAC,KAAK,CAAC,KAAK,EAAE,KAAK,CAAC,YAAY,CAAC,EAAE,CAAC,CAAC,CACxE,CAAC"}
|
|
@@ -0,0 +1,271 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* crawl_jobs — long-running jobs (Phase C + D).
|
|
3
|
+
*
|
|
4
|
+
* Shared by /crawl, /batch/scrape, /extract. job_kind discriminates.
|
|
5
|
+
* One table keeps polling, streaming, webhook-on-completion uniform
|
|
6
|
+
* across all three endpoints.
|
|
7
|
+
*/
|
|
8
|
+
export declare const crawlJobs: import("drizzle-orm/pg-core").PgTableWithColumns<{
|
|
9
|
+
name: "crawl_jobs";
|
|
10
|
+
schema: undefined;
|
|
11
|
+
columns: {
|
|
12
|
+
id: import("drizzle-orm/pg-core").PgColumn<{
|
|
13
|
+
name: "id";
|
|
14
|
+
tableName: "crawl_jobs";
|
|
15
|
+
dataType: "string";
|
|
16
|
+
columnType: "PgUUID";
|
|
17
|
+
data: string;
|
|
18
|
+
driverParam: string;
|
|
19
|
+
notNull: true;
|
|
20
|
+
hasDefault: true;
|
|
21
|
+
isPrimaryKey: true;
|
|
22
|
+
isAutoincrement: false;
|
|
23
|
+
hasRuntimeDefault: false;
|
|
24
|
+
enumValues: undefined;
|
|
25
|
+
baseColumn: never;
|
|
26
|
+
identity: undefined;
|
|
27
|
+
generated: undefined;
|
|
28
|
+
}, {}, {}>;
|
|
29
|
+
customerId: import("drizzle-orm/pg-core").PgColumn<{
|
|
30
|
+
name: "customer_id";
|
|
31
|
+
tableName: "crawl_jobs";
|
|
32
|
+
dataType: "string";
|
|
33
|
+
columnType: "PgUUID";
|
|
34
|
+
data: string;
|
|
35
|
+
driverParam: string;
|
|
36
|
+
notNull: true;
|
|
37
|
+
hasDefault: false;
|
|
38
|
+
isPrimaryKey: false;
|
|
39
|
+
isAutoincrement: false;
|
|
40
|
+
hasRuntimeDefault: false;
|
|
41
|
+
enumValues: undefined;
|
|
42
|
+
baseColumn: never;
|
|
43
|
+
identity: undefined;
|
|
44
|
+
generated: undefined;
|
|
45
|
+
}, {}, {}>;
|
|
46
|
+
jobKind: import("drizzle-orm/pg-core").PgColumn<{
|
|
47
|
+
name: "job_kind";
|
|
48
|
+
tableName: "crawl_jobs";
|
|
49
|
+
dataType: "string";
|
|
50
|
+
columnType: "PgText";
|
|
51
|
+
data: string;
|
|
52
|
+
driverParam: string;
|
|
53
|
+
notNull: true;
|
|
54
|
+
hasDefault: false;
|
|
55
|
+
isPrimaryKey: false;
|
|
56
|
+
isAutoincrement: false;
|
|
57
|
+
hasRuntimeDefault: false;
|
|
58
|
+
enumValues: [string, ...string[]];
|
|
59
|
+
baseColumn: never;
|
|
60
|
+
identity: undefined;
|
|
61
|
+
generated: undefined;
|
|
62
|
+
}, {}, {}>;
|
|
63
|
+
startUrl: import("drizzle-orm/pg-core").PgColumn<{
|
|
64
|
+
name: "start_url";
|
|
65
|
+
tableName: "crawl_jobs";
|
|
66
|
+
dataType: "string";
|
|
67
|
+
columnType: "PgText";
|
|
68
|
+
data: string;
|
|
69
|
+
driverParam: string;
|
|
70
|
+
notNull: false;
|
|
71
|
+
hasDefault: false;
|
|
72
|
+
isPrimaryKey: false;
|
|
73
|
+
isAutoincrement: false;
|
|
74
|
+
hasRuntimeDefault: false;
|
|
75
|
+
enumValues: [string, ...string[]];
|
|
76
|
+
baseColumn: never;
|
|
77
|
+
identity: undefined;
|
|
78
|
+
generated: undefined;
|
|
79
|
+
}, {}, {}>;
|
|
80
|
+
urlCanonical: import("drizzle-orm/pg-core").PgColumn<{
|
|
81
|
+
name: "url_canonical";
|
|
82
|
+
tableName: "crawl_jobs";
|
|
83
|
+
dataType: "string";
|
|
84
|
+
columnType: "PgText";
|
|
85
|
+
data: string;
|
|
86
|
+
driverParam: string;
|
|
87
|
+
notNull: false;
|
|
88
|
+
hasDefault: false;
|
|
89
|
+
isPrimaryKey: false;
|
|
90
|
+
isAutoincrement: false;
|
|
91
|
+
hasRuntimeDefault: false;
|
|
92
|
+
enumValues: [string, ...string[]];
|
|
93
|
+
baseColumn: never;
|
|
94
|
+
identity: undefined;
|
|
95
|
+
generated: undefined;
|
|
96
|
+
}, {}, {}>;
|
|
97
|
+
status: import("drizzle-orm/pg-core").PgColumn<{
|
|
98
|
+
name: "status";
|
|
99
|
+
tableName: "crawl_jobs";
|
|
100
|
+
dataType: "string";
|
|
101
|
+
columnType: "PgText";
|
|
102
|
+
data: string;
|
|
103
|
+
driverParam: string;
|
|
104
|
+
notNull: true;
|
|
105
|
+
hasDefault: true;
|
|
106
|
+
isPrimaryKey: false;
|
|
107
|
+
isAutoincrement: false;
|
|
108
|
+
hasRuntimeDefault: false;
|
|
109
|
+
enumValues: [string, ...string[]];
|
|
110
|
+
baseColumn: never;
|
|
111
|
+
identity: undefined;
|
|
112
|
+
generated: undefined;
|
|
113
|
+
}, {}, {}>;
|
|
114
|
+
params: import("drizzle-orm/pg-core").PgColumn<{
|
|
115
|
+
name: "params";
|
|
116
|
+
tableName: "crawl_jobs";
|
|
117
|
+
dataType: "json";
|
|
118
|
+
columnType: "PgJsonb";
|
|
119
|
+
data: unknown;
|
|
120
|
+
driverParam: unknown;
|
|
121
|
+
notNull: true;
|
|
122
|
+
hasDefault: true;
|
|
123
|
+
isPrimaryKey: false;
|
|
124
|
+
isAutoincrement: false;
|
|
125
|
+
hasRuntimeDefault: false;
|
|
126
|
+
enumValues: undefined;
|
|
127
|
+
baseColumn: never;
|
|
128
|
+
identity: undefined;
|
|
129
|
+
generated: undefined;
|
|
130
|
+
}, {}, {}>;
|
|
131
|
+
totalPagesDiscovered: import("drizzle-orm/pg-core").PgColumn<{
|
|
132
|
+
name: "total_pages_discovered";
|
|
133
|
+
tableName: "crawl_jobs";
|
|
134
|
+
dataType: "number";
|
|
135
|
+
columnType: "PgInteger";
|
|
136
|
+
data: number;
|
|
137
|
+
driverParam: string | number;
|
|
138
|
+
notNull: true;
|
|
139
|
+
hasDefault: true;
|
|
140
|
+
isPrimaryKey: false;
|
|
141
|
+
isAutoincrement: false;
|
|
142
|
+
hasRuntimeDefault: false;
|
|
143
|
+
enumValues: undefined;
|
|
144
|
+
baseColumn: never;
|
|
145
|
+
identity: undefined;
|
|
146
|
+
generated: undefined;
|
|
147
|
+
}, {}, {}>;
|
|
148
|
+
totalPagesScraped: import("drizzle-orm/pg-core").PgColumn<{
|
|
149
|
+
name: "total_pages_scraped";
|
|
150
|
+
tableName: "crawl_jobs";
|
|
151
|
+
dataType: "number";
|
|
152
|
+
columnType: "PgInteger";
|
|
153
|
+
data: number;
|
|
154
|
+
driverParam: string | number;
|
|
155
|
+
notNull: true;
|
|
156
|
+
hasDefault: true;
|
|
157
|
+
isPrimaryKey: false;
|
|
158
|
+
isAutoincrement: false;
|
|
159
|
+
hasRuntimeDefault: false;
|
|
160
|
+
enumValues: undefined;
|
|
161
|
+
baseColumn: never;
|
|
162
|
+
identity: undefined;
|
|
163
|
+
generated: undefined;
|
|
164
|
+
}, {}, {}>;
|
|
165
|
+
webhookUrl: import("drizzle-orm/pg-core").PgColumn<{
|
|
166
|
+
name: "webhook_url";
|
|
167
|
+
tableName: "crawl_jobs";
|
|
168
|
+
dataType: "string";
|
|
169
|
+
columnType: "PgText";
|
|
170
|
+
data: string;
|
|
171
|
+
driverParam: string;
|
|
172
|
+
notNull: false;
|
|
173
|
+
hasDefault: false;
|
|
174
|
+
isPrimaryKey: false;
|
|
175
|
+
isAutoincrement: false;
|
|
176
|
+
hasRuntimeDefault: false;
|
|
177
|
+
enumValues: [string, ...string[]];
|
|
178
|
+
baseColumn: never;
|
|
179
|
+
identity: undefined;
|
|
180
|
+
generated: undefined;
|
|
181
|
+
}, {}, {}>;
|
|
182
|
+
webhookSecret: import("drizzle-orm/pg-core").PgColumn<{
|
|
183
|
+
name: "webhook_secret";
|
|
184
|
+
tableName: "crawl_jobs";
|
|
185
|
+
dataType: "string";
|
|
186
|
+
columnType: "PgText";
|
|
187
|
+
data: string;
|
|
188
|
+
driverParam: string;
|
|
189
|
+
notNull: false;
|
|
190
|
+
hasDefault: false;
|
|
191
|
+
isPrimaryKey: false;
|
|
192
|
+
isAutoincrement: false;
|
|
193
|
+
hasRuntimeDefault: false;
|
|
194
|
+
enumValues: [string, ...string[]];
|
|
195
|
+
baseColumn: never;
|
|
196
|
+
identity: undefined;
|
|
197
|
+
generated: undefined;
|
|
198
|
+
}, {}, {}>;
|
|
199
|
+
createdAt: import("drizzle-orm/pg-core").PgColumn<{
|
|
200
|
+
name: "created_at";
|
|
201
|
+
tableName: "crawl_jobs";
|
|
202
|
+
dataType: "date";
|
|
203
|
+
columnType: "PgTimestamp";
|
|
204
|
+
data: Date;
|
|
205
|
+
driverParam: string;
|
|
206
|
+
notNull: true;
|
|
207
|
+
hasDefault: true;
|
|
208
|
+
isPrimaryKey: false;
|
|
209
|
+
isAutoincrement: false;
|
|
210
|
+
hasRuntimeDefault: false;
|
|
211
|
+
enumValues: undefined;
|
|
212
|
+
baseColumn: never;
|
|
213
|
+
identity: undefined;
|
|
214
|
+
generated: undefined;
|
|
215
|
+
}, {}, {}>;
|
|
216
|
+
startedAt: import("drizzle-orm/pg-core").PgColumn<{
|
|
217
|
+
name: "started_at";
|
|
218
|
+
tableName: "crawl_jobs";
|
|
219
|
+
dataType: "date";
|
|
220
|
+
columnType: "PgTimestamp";
|
|
221
|
+
data: Date;
|
|
222
|
+
driverParam: string;
|
|
223
|
+
notNull: false;
|
|
224
|
+
hasDefault: false;
|
|
225
|
+
isPrimaryKey: false;
|
|
226
|
+
isAutoincrement: false;
|
|
227
|
+
hasRuntimeDefault: false;
|
|
228
|
+
enumValues: undefined;
|
|
229
|
+
baseColumn: never;
|
|
230
|
+
identity: undefined;
|
|
231
|
+
generated: undefined;
|
|
232
|
+
}, {}, {}>;
|
|
233
|
+
completedAt: import("drizzle-orm/pg-core").PgColumn<{
|
|
234
|
+
name: "completed_at";
|
|
235
|
+
tableName: "crawl_jobs";
|
|
236
|
+
dataType: "date";
|
|
237
|
+
columnType: "PgTimestamp";
|
|
238
|
+
data: Date;
|
|
239
|
+
driverParam: string;
|
|
240
|
+
notNull: false;
|
|
241
|
+
hasDefault: false;
|
|
242
|
+
isPrimaryKey: false;
|
|
243
|
+
isAutoincrement: false;
|
|
244
|
+
hasRuntimeDefault: false;
|
|
245
|
+
enumValues: undefined;
|
|
246
|
+
baseColumn: never;
|
|
247
|
+
identity: undefined;
|
|
248
|
+
generated: undefined;
|
|
249
|
+
}, {}, {}>;
|
|
250
|
+
error: import("drizzle-orm/pg-core").PgColumn<{
|
|
251
|
+
name: "error";
|
|
252
|
+
tableName: "crawl_jobs";
|
|
253
|
+
dataType: "string";
|
|
254
|
+
columnType: "PgText";
|
|
255
|
+
data: string;
|
|
256
|
+
driverParam: string;
|
|
257
|
+
notNull: false;
|
|
258
|
+
hasDefault: false;
|
|
259
|
+
isPrimaryKey: false;
|
|
260
|
+
isAutoincrement: false;
|
|
261
|
+
hasRuntimeDefault: false;
|
|
262
|
+
enumValues: [string, ...string[]];
|
|
263
|
+
baseColumn: never;
|
|
264
|
+
identity: undefined;
|
|
265
|
+
generated: undefined;
|
|
266
|
+
}, {}, {}>;
|
|
267
|
+
};
|
|
268
|
+
dialect: "pg";
|
|
269
|
+
}>;
|
|
270
|
+
export type CrawlJob = typeof crawlJobs.$inferSelect;
|
|
271
|
+
export type NewCrawlJob = typeof crawlJobs.$inferInsert;
|
|
@@ -0,0 +1,40 @@
|
|
|
1
|
+
// SPDX-License-Identifier: Apache-2.0
|
|
2
|
+
import { sql } from 'drizzle-orm';
|
|
3
|
+
import { index, integer, jsonb, pgTable, text, timestamp, uuid } from 'drizzle-orm/pg-core';
|
|
4
|
+
import { customers } from './customers.js';
|
|
5
|
+
/**
|
|
6
|
+
* crawl_jobs — long-running jobs (Phase C + D).
|
|
7
|
+
*
|
|
8
|
+
* Shared by /crawl, /batch/scrape, /extract. job_kind discriminates.
|
|
9
|
+
* One table keeps polling, streaming, webhook-on-completion uniform
|
|
10
|
+
* across all three endpoints.
|
|
11
|
+
*/
|
|
12
|
+
export const crawlJobs = pgTable('crawl_jobs', {
|
|
13
|
+
id: uuid('id').primaryKey().defaultRandom(),
|
|
14
|
+
customerId: uuid('customer_id')
|
|
15
|
+
.notNull()
|
|
16
|
+
.references(() => customers.id),
|
|
17
|
+
/** 'crawl' | 'batch_scrape' | 'extract' */
|
|
18
|
+
jobKind: text('job_kind').notNull(),
|
|
19
|
+
/** Seed URL for crawl; null for batch_scrape/extract. */
|
|
20
|
+
startUrl: text('start_url'),
|
|
21
|
+
urlCanonical: text('url_canonical'),
|
|
22
|
+
/** 'pending' | 'running' | 'succeeded' | 'failed' | 'cancelled' */
|
|
23
|
+
status: text('status').notNull().default('pending'),
|
|
24
|
+
/** Opaque JSON of request params (maxDepth, urls, schema, etc.). */
|
|
25
|
+
params: jsonb('params').notNull().default(sql `'{}'::jsonb`),
|
|
26
|
+
totalPagesDiscovered: integer('total_pages_discovered').notNull().default(0),
|
|
27
|
+
totalPagesScraped: integer('total_pages_scraped').notNull().default(0),
|
|
28
|
+
webhookUrl: text('webhook_url'),
|
|
29
|
+
webhookSecret: text('webhook_secret'),
|
|
30
|
+
createdAt: timestamp('created_at', { withTimezone: true }).notNull().defaultNow(),
|
|
31
|
+
startedAt: timestamp('started_at', { withTimezone: true }),
|
|
32
|
+
completedAt: timestamp('completed_at', { withTimezone: true }),
|
|
33
|
+
error: text('error'),
|
|
34
|
+
}, (table) => [
|
|
35
|
+
index('idx_crawl_jobs_customer_time').on(table.customerId, table.createdAt.desc()),
|
|
36
|
+
index('idx_crawl_jobs_status_pending')
|
|
37
|
+
.on(table.status, table.createdAt)
|
|
38
|
+
.where(sql `${table.status} IN ('pending', 'running')`),
|
|
39
|
+
]);
|
|
40
|
+
//# sourceMappingURL=crawl-jobs.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"crawl-jobs.js","sourceRoot":"","sources":["../../src/schema/crawl-jobs.ts"],"names":[],"mappings":"AAAA,sCAAsC;AACtC,OAAO,EAAE,GAAG,EAAE,MAAM,aAAa,CAAC;AAClC,OAAO,EAAE,KAAK,EAAE,OAAO,EAAE,KAAK,EAAE,OAAO,EAAE,IAAI,EAAE,SAAS,EAAE,IAAI,EAAE,MAAM,qBAAqB,CAAC;AAE5F,OAAO,EAAE,SAAS,EAAE,MAAM,gBAAgB,CAAC;AAE3C;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,SAAS,GAAG,OAAO,CAC9B,YAAY,EACZ;IACE,EAAE,EAAE,IAAI,CAAC,IAAI,CAAC,CAAC,UAAU,EAAE,CAAC,aAAa,EAAE;IAC3C,UAAU,EAAE,IAAI,CAAC,aAAa,CAAC;SAC5B,OAAO,EAAE;SACT,UAAU,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,EAAE,CAAC;IACjC,2CAA2C;IAC3C,OAAO,EAAE,IAAI,CAAC,UAAU,CAAC,CAAC,OAAO,EAAE;IACnC,yDAAyD;IACzD,QAAQ,EAAE,IAAI,CAAC,WAAW,CAAC;IAC3B,YAAY,EAAE,IAAI,CAAC,eAAe,CAAC;IACnC,mEAAmE;IACnE,MAAM,EAAE,IAAI,CAAC,QAAQ,CAAC,CAAC,OAAO,EAAE,CAAC,OAAO,CAAC,SAAS,CAAC;IACnD,oEAAoE;IACpE,MAAM,EAAE,KAAK,CAAC,QAAQ,CAAC,CAAC,OAAO,EAAE,CAAC,OAAO,CAAC,GAAG,CAAA,aAAa,CAAC;IAC3D,oBAAoB,EAAE,OAAO,CAAC,wBAAwB,CAAC,CAAC,OAAO,EAAE,CAAC,OAAO,CAAC,CAAC,CAAC;IAC5E,iBAAiB,EAAE,OAAO,CAAC,qBAAqB,CAAC,CAAC,OAAO,EAAE,CAAC,OAAO,CAAC,CAAC,CAAC;IACtE,UAAU,EAAE,IAAI,CAAC,aAAa,CAAC;IAC/B,aAAa,EAAE,IAAI,CAAC,gBAAgB,CAAC;IACrC,SAAS,EAAE,SAAS,CAAC,YAAY,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,CAAC,CAAC,OAAO,EAAE,CAAC,UAAU,EAAE;IACjF,SAAS,EAAE,SAAS,CAAC,YAAY,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,CAAC;IAC1D,WAAW,EAAE,SAAS,CAAC,cAAc,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,CAAC;IAC9D,KAAK,EAAE,IAAI,CAAC,OAAO,CAAC;CACrB,EACD,CAAC,KAAK,EAAE,EAAE,CAAC;IACT,KAAK,CAAC,8BAA8B,CAAC,CAAC,EAAE,CAAC,KAAK,CAAC,UAAU,EAAE,KAAK,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC;IAClF,KAAK,CAAC,+BAA+B,CAAC;SACnC,EAAE,CAAC,KAAK,CAAC,MAAM,EAAE,KAAK,CAAC,SAAS,CAAC;SACjC,KAAK,CAAC,GAAG,CAAA,GAAG,KAAK,CAAC,MAAM,4BAA4B,CAAC;CACzD,CACF,CAAC"}
|
|
@@ -0,0 +1,145 @@
|
|
|
1
|
+
export declare const customers: import("drizzle-orm/pg-core").PgTableWithColumns<{
|
|
2
|
+
name: "customers";
|
|
3
|
+
schema: undefined;
|
|
4
|
+
columns: {
|
|
5
|
+
id: import("drizzle-orm/pg-core").PgColumn<{
|
|
6
|
+
name: "id";
|
|
7
|
+
tableName: "customers";
|
|
8
|
+
dataType: "string";
|
|
9
|
+
columnType: "PgUUID";
|
|
10
|
+
data: string;
|
|
11
|
+
driverParam: string;
|
|
12
|
+
notNull: true;
|
|
13
|
+
hasDefault: true;
|
|
14
|
+
isPrimaryKey: true;
|
|
15
|
+
isAutoincrement: false;
|
|
16
|
+
hasRuntimeDefault: false;
|
|
17
|
+
enumValues: undefined;
|
|
18
|
+
baseColumn: never;
|
|
19
|
+
identity: undefined;
|
|
20
|
+
generated: undefined;
|
|
21
|
+
}, {}, {}>;
|
|
22
|
+
email: import("drizzle-orm/pg-core").PgColumn<{
|
|
23
|
+
name: "email";
|
|
24
|
+
tableName: "customers";
|
|
25
|
+
dataType: "string";
|
|
26
|
+
columnType: "PgText";
|
|
27
|
+
data: string;
|
|
28
|
+
driverParam: string;
|
|
29
|
+
notNull: true;
|
|
30
|
+
hasDefault: false;
|
|
31
|
+
isPrimaryKey: false;
|
|
32
|
+
isAutoincrement: false;
|
|
33
|
+
hasRuntimeDefault: false;
|
|
34
|
+
enumValues: [string, ...string[]];
|
|
35
|
+
baseColumn: never;
|
|
36
|
+
identity: undefined;
|
|
37
|
+
generated: undefined;
|
|
38
|
+
}, {}, {}>;
|
|
39
|
+
orgName: import("drizzle-orm/pg-core").PgColumn<{
|
|
40
|
+
name: "org_name";
|
|
41
|
+
tableName: "customers";
|
|
42
|
+
dataType: "string";
|
|
43
|
+
columnType: "PgText";
|
|
44
|
+
data: string;
|
|
45
|
+
driverParam: string;
|
|
46
|
+
notNull: false;
|
|
47
|
+
hasDefault: false;
|
|
48
|
+
isPrimaryKey: false;
|
|
49
|
+
isAutoincrement: false;
|
|
50
|
+
hasRuntimeDefault: false;
|
|
51
|
+
enumValues: [string, ...string[]];
|
|
52
|
+
baseColumn: never;
|
|
53
|
+
identity: undefined;
|
|
54
|
+
generated: undefined;
|
|
55
|
+
}, {}, {}>;
|
|
56
|
+
plan: import("drizzle-orm/pg-core").PgColumn<{
|
|
57
|
+
name: "plan";
|
|
58
|
+
tableName: "customers";
|
|
59
|
+
dataType: "string";
|
|
60
|
+
columnType: "PgText";
|
|
61
|
+
data: string;
|
|
62
|
+
driverParam: string;
|
|
63
|
+
notNull: true;
|
|
64
|
+
hasDefault: true;
|
|
65
|
+
isPrimaryKey: false;
|
|
66
|
+
isAutoincrement: false;
|
|
67
|
+
hasRuntimeDefault: false;
|
|
68
|
+
enumValues: [string, ...string[]];
|
|
69
|
+
baseColumn: never;
|
|
70
|
+
identity: undefined;
|
|
71
|
+
generated: undefined;
|
|
72
|
+
}, {}, {}>;
|
|
73
|
+
stripeCustomerId: import("drizzle-orm/pg-core").PgColumn<{
|
|
74
|
+
name: "stripe_customer_id";
|
|
75
|
+
tableName: "customers";
|
|
76
|
+
dataType: "string";
|
|
77
|
+
columnType: "PgText";
|
|
78
|
+
data: string;
|
|
79
|
+
driverParam: string;
|
|
80
|
+
notNull: false;
|
|
81
|
+
hasDefault: false;
|
|
82
|
+
isPrimaryKey: false;
|
|
83
|
+
isAutoincrement: false;
|
|
84
|
+
hasRuntimeDefault: false;
|
|
85
|
+
enumValues: [string, ...string[]];
|
|
86
|
+
baseColumn: never;
|
|
87
|
+
identity: undefined;
|
|
88
|
+
generated: undefined;
|
|
89
|
+
}, {}, {}>;
|
|
90
|
+
createdAt: import("drizzle-orm/pg-core").PgColumn<{
|
|
91
|
+
name: "created_at";
|
|
92
|
+
tableName: "customers";
|
|
93
|
+
dataType: "date";
|
|
94
|
+
columnType: "PgTimestamp";
|
|
95
|
+
data: Date;
|
|
96
|
+
driverParam: string;
|
|
97
|
+
notNull: true;
|
|
98
|
+
hasDefault: true;
|
|
99
|
+
isPrimaryKey: false;
|
|
100
|
+
isAutoincrement: false;
|
|
101
|
+
hasRuntimeDefault: false;
|
|
102
|
+
enumValues: undefined;
|
|
103
|
+
baseColumn: never;
|
|
104
|
+
identity: undefined;
|
|
105
|
+
generated: undefined;
|
|
106
|
+
}, {}, {}>;
|
|
107
|
+
updatedAt: import("drizzle-orm/pg-core").PgColumn<{
|
|
108
|
+
name: "updated_at";
|
|
109
|
+
tableName: "customers";
|
|
110
|
+
dataType: "date";
|
|
111
|
+
columnType: "PgTimestamp";
|
|
112
|
+
data: Date;
|
|
113
|
+
driverParam: string;
|
|
114
|
+
notNull: true;
|
|
115
|
+
hasDefault: true;
|
|
116
|
+
isPrimaryKey: false;
|
|
117
|
+
isAutoincrement: false;
|
|
118
|
+
hasRuntimeDefault: false;
|
|
119
|
+
enumValues: undefined;
|
|
120
|
+
baseColumn: never;
|
|
121
|
+
identity: undefined;
|
|
122
|
+
generated: undefined;
|
|
123
|
+
}, {}, {}>;
|
|
124
|
+
deletedAt: import("drizzle-orm/pg-core").PgColumn<{
|
|
125
|
+
name: "deleted_at";
|
|
126
|
+
tableName: "customers";
|
|
127
|
+
dataType: "date";
|
|
128
|
+
columnType: "PgTimestamp";
|
|
129
|
+
data: Date;
|
|
130
|
+
driverParam: string;
|
|
131
|
+
notNull: false;
|
|
132
|
+
hasDefault: false;
|
|
133
|
+
isPrimaryKey: false;
|
|
134
|
+
isAutoincrement: false;
|
|
135
|
+
hasRuntimeDefault: false;
|
|
136
|
+
enumValues: undefined;
|
|
137
|
+
baseColumn: never;
|
|
138
|
+
identity: undefined;
|
|
139
|
+
generated: undefined;
|
|
140
|
+
}, {}, {}>;
|
|
141
|
+
};
|
|
142
|
+
dialect: "pg";
|
|
143
|
+
}>;
|
|
144
|
+
export type Customer = typeof customers.$inferSelect;
|
|
145
|
+
export type NewCustomer = typeof customers.$inferInsert;
|