scrapely 1.2.21 → 1.2.23
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/actor.d.ts +13 -1
- package/dist/actor.js +16 -3
- package/dist/charging.d.ts +39 -0
- package/dist/charging.js +189 -3
- package/dist/index.d.ts +2 -2
- package/dist/index.js +4 -2
- package/package.json +3 -1
package/dist/actor.d.ts
CHANGED
|
@@ -19,6 +19,11 @@ import { Configuration, type ConfigurationOptions } from './configuration.js';
|
|
|
19
19
|
import type { StorageClient } from '@crawlee/types';
|
|
20
20
|
import { type ApiStorageConfig } from './storage/api-storage-client.js';
|
|
21
21
|
import { ChargingManager, type ChargeOptions as ChargingChargeOptions, type ChargeResult as ChargingChargeResult } from './charging.js';
|
|
22
|
+
/** Options for a multi-event charged push (`Actor.pushData(row, { charges })`). */
|
|
23
|
+
export interface PushDataChargeOptions {
|
|
24
|
+
/** Events per row, e.g. `{ result: 1, page: 3 }`. */
|
|
25
|
+
charges: Record<string, number>;
|
|
26
|
+
}
|
|
22
27
|
import { createProxyConfiguration, type CreateProxyConfigurationOptions } from './proxy-configuration.js';
|
|
23
28
|
import { type BrowserConfiguration, type BrowserConfigurationOptions } from './browser-configuration.js';
|
|
24
29
|
import { type Clearance, type SolveOptions, type SolverVendorStatus } from './solver.js';
|
|
@@ -405,7 +410,14 @@ export declare class Actor {
|
|
|
405
410
|
* @returns ChargeResult indicating if the user's spending limit was reached
|
|
406
411
|
*/
|
|
407
412
|
static pushData<Data extends Record<string, any>>(item: Data | Data[], eventName: string): Promise<ChargeResult>;
|
|
408
|
-
|
|
413
|
+
/**
|
|
414
|
+
* Stores rows that each cost several pay-per-event events, charged atomically:
|
|
415
|
+
* `pushData(row, { charges: { result: 1, page: 3 } })` charges 1 `result` and 3 `page`
|
|
416
|
+
* events per row. A row is stored only when all of its events were charged; rows the
|
|
417
|
+
* spending limit cannot pay for in full are not stored.
|
|
418
|
+
*/
|
|
419
|
+
static pushData<Data extends Record<string, any>>(item: Data | Data[], options: PushDataChargeOptions): Promise<ChargeResult>;
|
|
420
|
+
pushData<Data extends Record<string, any>>(item: Data | Data[], eventNameOrOptions?: string | PushDataChargeOptions): Promise<void | ChargeResult>;
|
|
409
421
|
private usesPushDataInterception;
|
|
410
422
|
private pushDataViaInterceptedClient;
|
|
411
423
|
private pushDataWithExplicitCharging;
|
package/dist/actor.js
CHANGED
|
@@ -29,6 +29,7 @@ const events_client_js_1 = require("./events/events-client.js");
|
|
|
29
29
|
const scrapely_event_manager_js_1 = require("./events/scrapely-event-manager.js");
|
|
30
30
|
const secrets_js_1 = require("./crypto/secrets.js");
|
|
31
31
|
const charging_js_1 = require("./charging.js");
|
|
32
|
+
const charging_js_2 = require("./charging.js");
|
|
32
33
|
const charging_client_js_1 = require("./charging_client.js");
|
|
33
34
|
const proxy_configuration_js_1 = require("./proxy-configuration.js");
|
|
34
35
|
const browser_configuration_js_1 = require("./browser-configuration.js");
|
|
@@ -513,11 +514,23 @@ class Actor {
|
|
|
513
514
|
}
|
|
514
515
|
return cachedInput ?? undefined;
|
|
515
516
|
}
|
|
516
|
-
static async pushData(item,
|
|
517
|
-
return Actor.getDefaultInstance().pushData(item,
|
|
517
|
+
static async pushData(item, eventNameOrOptions) {
|
|
518
|
+
return Actor.getDefaultInstance().pushData(item, eventNameOrOptions);
|
|
518
519
|
}
|
|
519
|
-
async pushData(item,
|
|
520
|
+
async pushData(item, eventNameOrOptions) {
|
|
520
521
|
Actor._ensureActorInit('pushData');
|
|
522
|
+
if (eventNameOrOptions !== undefined && typeof eventNameOrOptions === 'object') {
|
|
523
|
+
const { Dataset } = await import('crawlee');
|
|
524
|
+
const dataset = await Dataset.open();
|
|
525
|
+
return (0, charging_js_2.pushDataWithCharges)({
|
|
526
|
+
chargingManager: this.chargingManager,
|
|
527
|
+
items: item,
|
|
528
|
+
charges: eventNameOrOptions.charges,
|
|
529
|
+
isDefaultDataset: dataset.id === this.config.get('defaultDatasetId'),
|
|
530
|
+
pushFn: async (rows) => { await dataset.pushData(rows); },
|
|
531
|
+
});
|
|
532
|
+
}
|
|
533
|
+
const eventName = eventNameOrOptions;
|
|
521
534
|
// Exact synthetic names only — prefix would block real priced events like apify-foo.
|
|
522
535
|
if (eventName === 'apify-default-dataset-item' ||
|
|
523
536
|
eventName === 'scrapely-default-dataset-item') {
|
package/dist/charging.d.ts
CHANGED
|
@@ -86,6 +86,19 @@ export declare class ChargingManager {
|
|
|
86
86
|
* Actors never go positive, so they get nothing extra for free.
|
|
87
87
|
*/
|
|
88
88
|
private prepaidRowBalance;
|
|
89
|
+
/**
|
|
90
|
+
* Run spend the platform counts against maxTotalChargeUsd that this SDK does
|
|
91
|
+
* not charge itself (compute/proxy/browser usage, server-billed events).
|
|
92
|
+
* Learnt from the charge endpoint's remainingBudgetNd; only ever grows.
|
|
93
|
+
*/
|
|
94
|
+
private platformOtherUsd;
|
|
95
|
+
private lastRemainingBudgetNd;
|
|
96
|
+
/** maxTotalChargeUsd minus this SDK's charges and the platform's other run spend. */
|
|
97
|
+
private remainingBudgetUsd;
|
|
98
|
+
/** Fold the last charge response's remainingBudgetNd into the local budget (monotonic). */
|
|
99
|
+
private syncPlatformBudget;
|
|
100
|
+
/** @internal Platform spend outside this SDK's charges (tests). */
|
|
101
|
+
_platformOtherUsd(): number;
|
|
89
102
|
/** @internal Called by Actor shutdown before the final flush. */
|
|
90
103
|
close(): void;
|
|
91
104
|
/**
|
|
@@ -110,6 +123,15 @@ export declare class ChargingManager {
|
|
|
110
123
|
_notePrepaidCharge(eventName: string, chargedCount: number): void;
|
|
111
124
|
/** @internal */
|
|
112
125
|
_isPayPerEvent(): boolean;
|
|
126
|
+
/**
|
|
127
|
+
* @internal The priced part of a multi-event row (`charges`), plus the
|
|
128
|
+
* default-dataset-item event on the default dataset when it is priced, and
|
|
129
|
+
* how many rows the remaining budget pays for in full.
|
|
130
|
+
*/
|
|
131
|
+
_rowCharges(charges: Record<string, number>, isDefaultDataset: boolean, rowCount: number): {
|
|
132
|
+
perRow: Record<string, number>;
|
|
133
|
+
rows: number;
|
|
134
|
+
};
|
|
113
135
|
/** @internal Current prepaid-row balance (tests). */
|
|
114
136
|
_prepaidRowBalance(): number;
|
|
115
137
|
/**
|
|
@@ -252,6 +274,23 @@ export declare function pushDataAndCharge<T>({ chargingManager, items, eventName
|
|
|
252
274
|
isDefaultDataset: boolean;
|
|
253
275
|
pushFn: (limitedItems: T | T[]) => Promise<void>;
|
|
254
276
|
}): Promise<ChargeResult>;
|
|
277
|
+
/** Validate `{ eventName: eventsPerRow }` for a multi-event push. */
|
|
278
|
+
export declare function validateRowCharges(charges: unknown): Record<string, number>;
|
|
279
|
+
/**
|
|
280
|
+
* Push rows that each cost several events, atomically per row
|
|
281
|
+
* (`charges = { result: 1, page: 3 }`). A row is stored only when every one of
|
|
282
|
+
* its events was charged: rows the budget cannot pay for in full are not
|
|
283
|
+
* stored, and an operation that came back short is voided and re-charged for
|
|
284
|
+
* the rows that are kept, so no row is stored part-paid and no event is billed
|
|
285
|
+
* for a row that was not stored. Unpriced events are ignored.
|
|
286
|
+
*/
|
|
287
|
+
export declare function pushDataWithCharges<T>({ chargingManager, items, charges, isDefaultDataset, pushFn, }: {
|
|
288
|
+
chargingManager: ChargingManager;
|
|
289
|
+
items: T | T[];
|
|
290
|
+
charges: Record<string, number>;
|
|
291
|
+
isDefaultDataset: boolean;
|
|
292
|
+
pushFn: (rows: T | T[]) => Promise<void>;
|
|
293
|
+
}): Promise<ChargeResult>;
|
|
255
294
|
/**
|
|
256
295
|
* @internal True when a dataset push certainly stored nothing: the storage API
|
|
257
296
|
* answered with an error status, or the request never reached it. Timeouts and
|
package/dist/charging.js
CHANGED
|
@@ -9,6 +9,8 @@ exports.resolveDefaultDatasetItemEventName = resolveDefaultDatasetItemEventName;
|
|
|
9
9
|
exports.hasDefaultDatasetItemEventInPricing = hasDefaultDatasetItemEventInPricing;
|
|
10
10
|
exports.mergeChargeResults = mergeChargeResults;
|
|
11
11
|
exports.pushDataAndCharge = pushDataAndCharge;
|
|
12
|
+
exports.validateRowCharges = validateRowCharges;
|
|
13
|
+
exports.pushDataWithCharges = pushDataWithCharges;
|
|
12
14
|
exports.pushCertainlyFailed = pushCertainlyFailed;
|
|
13
15
|
// SPDX-License-Identifier: Apache-2.0
|
|
14
16
|
// Derived from apify-sdk-js (src/charging.ts),
|
|
@@ -163,6 +165,31 @@ class ChargingManager {
|
|
|
163
165
|
* Actors never go positive, so they get nothing extra for free.
|
|
164
166
|
*/
|
|
165
167
|
prepaidRowBalance = 0;
|
|
168
|
+
/**
|
|
169
|
+
* Run spend the platform counts against maxTotalChargeUsd that this SDK does
|
|
170
|
+
* not charge itself (compute/proxy/browser usage, server-billed events).
|
|
171
|
+
* Learnt from the charge endpoint's remainingBudgetNd; only ever grows.
|
|
172
|
+
*/
|
|
173
|
+
platformOtherUsd = 0;
|
|
174
|
+
lastRemainingBudgetNd = null;
|
|
175
|
+
/** maxTotalChargeUsd minus this SDK's charges and the platform's other run spend. */
|
|
176
|
+
remainingBudgetUsd() {
|
|
177
|
+
return this.maxTotalChargeUsd - this.calculateTotalChargedAmount() - this.platformOtherUsd;
|
|
178
|
+
}
|
|
179
|
+
/** Fold the last charge response's remainingBudgetNd into the local budget (monotonic). */
|
|
180
|
+
syncPlatformBudget() {
|
|
181
|
+
const nd = this.lastRemainingBudgetNd;
|
|
182
|
+
this.lastRemainingBudgetNd = null;
|
|
183
|
+
if (nd === null || nd < 0 || !Number.isFinite(this.maxTotalChargeUsd))
|
|
184
|
+
return;
|
|
185
|
+
const other = Number((this.maxTotalChargeUsd - nd / 1e9 - this.calculateTotalChargedAmount()).toFixed(9));
|
|
186
|
+
if (other > this.platformOtherUsd)
|
|
187
|
+
this.platformOtherUsd = other;
|
|
188
|
+
}
|
|
189
|
+
/** @internal Platform spend outside this SDK's charges (tests). */
|
|
190
|
+
_platformOtherUsd() {
|
|
191
|
+
return this.platformOtherUsd;
|
|
192
|
+
}
|
|
166
193
|
/** @internal Called by Actor shutdown before the final flush. */
|
|
167
194
|
close() {
|
|
168
195
|
this._closed = true;
|
|
@@ -234,6 +261,30 @@ class ChargingManager {
|
|
|
234
261
|
_isPayPerEvent() {
|
|
235
262
|
return this.chargingState !== undefined && this.isPayPerEvent;
|
|
236
263
|
}
|
|
264
|
+
/**
|
|
265
|
+
* @internal The priced part of a multi-event row (`charges`), plus the
|
|
266
|
+
* default-dataset-item event on the default dataset when it is priced, and
|
|
267
|
+
* how many rows the remaining budget pays for in full.
|
|
268
|
+
*/
|
|
269
|
+
_rowCharges(charges, isDefaultDataset, rowCount) {
|
|
270
|
+
const perRow = {};
|
|
271
|
+
for (const [name, n] of Object.entries(charges)) {
|
|
272
|
+
if ((this.calculateEventPrice(name) ?? 0) > 0)
|
|
273
|
+
perRow[name] = n;
|
|
274
|
+
}
|
|
275
|
+
if (isDefaultDataset) {
|
|
276
|
+
const def = resolveDefaultDatasetItemEventName(this.pricingInfo);
|
|
277
|
+
if (!(def in perRow) && (this.calculateEventPrice(def) ?? 0) > 0)
|
|
278
|
+
perRow[def] = 1;
|
|
279
|
+
}
|
|
280
|
+
const rowCost = Object.entries(perRow).reduce((a, [name, n]) => a + (this.calculateEventPrice(name) ?? 0) * n, 0);
|
|
281
|
+
let rows = rowCount;
|
|
282
|
+
if (rowCost > 0 && Number.isFinite(this.maxTotalChargeUsd)) {
|
|
283
|
+
const fit = Math.floor(Number((this.remainingBudgetUsd() / rowCost).toFixed(4)));
|
|
284
|
+
rows = Math.min(rows, Math.max(0, fit));
|
|
285
|
+
}
|
|
286
|
+
return { perRow, rows };
|
|
287
|
+
}
|
|
237
288
|
/** @internal Current prepaid-row balance (tests). */
|
|
238
289
|
_prepaidRowBalance() {
|
|
239
290
|
return this.prepaidRowBalance;
|
|
@@ -517,7 +568,7 @@ class ChargingManager {
|
|
|
517
568
|
return eventName;
|
|
518
569
|
}
|
|
519
570
|
calculateMaxChargesByPrice(price) {
|
|
520
|
-
const unroundedResult =
|
|
571
|
+
const unroundedResult = this.remainingBudgetUsd() / price;
|
|
521
572
|
const roundedResult = Math.floor(Number(unroundedResult.toFixed(4)));
|
|
522
573
|
return Math.max(0, roundedResult);
|
|
523
574
|
}
|
|
@@ -625,6 +676,8 @@ class ChargingManager {
|
|
|
625
676
|
});
|
|
626
677
|
const body = extractChargeResponseBody(response);
|
|
627
678
|
const accepted = readAcceptedCount(body, intent.count);
|
|
679
|
+
const remaining = body.remainingBudgetNd;
|
|
680
|
+
this.lastRemainingBudgetNd = typeof remaining === 'number' && Number.isFinite(remaining) && remaining >= 0 ? remaining : null;
|
|
628
681
|
return accepted;
|
|
629
682
|
}
|
|
630
683
|
catch (err) {
|
|
@@ -930,6 +983,7 @@ class ChargingManager {
|
|
|
930
983
|
if (accepted < chargedCount)
|
|
931
984
|
this.markChargeLimitHit();
|
|
932
985
|
chargedCount = this.applyAcceptedCount(normalizedEventName, chargedCount, accepted, eventPricing.price);
|
|
986
|
+
this.syncPlatformBudget();
|
|
933
987
|
}
|
|
934
988
|
else {
|
|
935
989
|
throw new ChargeUncertainError(`Charge for event '${normalizedEventName}' (count=${chargedCount}) was not confirmed by the platform; operationId=${intent.idempotencyKey}`, {
|
|
@@ -1042,8 +1096,9 @@ async function pushDataAndCharge({ chargingManager, items, eventName, isDefaultD
|
|
|
1042
1096
|
let unpaidPastLimit = false;
|
|
1043
1097
|
if (isDefaultDataset &&
|
|
1044
1098
|
requested > 0 &&
|
|
1045
|
-
|
|
1046
|
-
|
|
1099
|
+
// Optional calls: partial managers (tests, older shims) skip the net, as before.
|
|
1100
|
+
chargingManager._isPayPerEvent?.() === true &&
|
|
1101
|
+
!chargingManager._isPaidPerItem?.(eventName, isDefaultDataset)) {
|
|
1047
1102
|
unpaidPastLimit = chargingManager.isChargeLimitReached?.() ?? false;
|
|
1048
1103
|
const allowed = chargingManager._takeUnpaidRows(requested);
|
|
1049
1104
|
if (allowed < requested) {
|
|
@@ -1236,6 +1291,137 @@ async function pushDataAndCharge({ chargingManager, items, eventName, isDefaultD
|
|
|
1236
1291
|
};
|
|
1237
1292
|
});
|
|
1238
1293
|
}
|
|
1294
|
+
/** Validate `{ eventName: eventsPerRow }` for a multi-event push. */
|
|
1295
|
+
function validateRowCharges(charges) {
|
|
1296
|
+
if (!charges || typeof charges !== 'object' || Array.isArray(charges) || Object.keys(charges).length === 0) {
|
|
1297
|
+
throw new Error('charges must be a non-empty object of { eventName: countPerRow }');
|
|
1298
|
+
}
|
|
1299
|
+
const out = {};
|
|
1300
|
+
for (const [name, count] of Object.entries(charges)) {
|
|
1301
|
+
if (!name)
|
|
1302
|
+
throw new Error('charges keys must be event names');
|
|
1303
|
+
if (isSyntheticEvent(name))
|
|
1304
|
+
throw new Error(`Cannot charge for synthetic event '${name}' manually`);
|
|
1305
|
+
if (typeof count !== 'number' || !Number.isInteger(count) || count < 1) {
|
|
1306
|
+
throw new Error(`charges['${name}'] must be a positive integer (events per row)`);
|
|
1307
|
+
}
|
|
1308
|
+
out[name] = count;
|
|
1309
|
+
}
|
|
1310
|
+
return out;
|
|
1311
|
+
}
|
|
1312
|
+
/**
|
|
1313
|
+
* Push rows that each cost several events, atomically per row
|
|
1314
|
+
* (`charges = { result: 1, page: 3 }`). A row is stored only when every one of
|
|
1315
|
+
* its events was charged: rows the budget cannot pay for in full are not
|
|
1316
|
+
* stored, and an operation that came back short is voided and re-charged for
|
|
1317
|
+
* the rows that are kept, so no row is stored part-paid and no event is billed
|
|
1318
|
+
* for a row that was not stored. Unpriced events are ignored.
|
|
1319
|
+
*/
|
|
1320
|
+
async function pushDataWithCharges({ chargingManager, items, charges, isDefaultDataset, pushFn, }) {
|
|
1321
|
+
const perRowIn = validateRowCharges(charges);
|
|
1322
|
+
const rowsIn = Array.isArray(items) ? items : [items];
|
|
1323
|
+
const single = !Array.isArray(items);
|
|
1324
|
+
const store = (rows) => chargeGatedPush.run(true, () => pushFn(single ? rows[0] : rows));
|
|
1325
|
+
if (rowsIn.length === 0)
|
|
1326
|
+
return { eventChargeLimitReached: false, chargedCount: 0, chargeableWithinLimit: {} };
|
|
1327
|
+
return chargingManager._withOperationLock(async () => {
|
|
1328
|
+
if (chargingManager.isClosed()) {
|
|
1329
|
+
throw new Error('ChargingManager is closed (actor is shutting down); charge rejected');
|
|
1330
|
+
}
|
|
1331
|
+
const voidsConfirmed = (await chargingManager._retryPendingVoids?.()) !== false;
|
|
1332
|
+
if (!chargingManager._isPayPerEvent()) {
|
|
1333
|
+
await store(rowsIn);
|
|
1334
|
+
return { eventChargeLimitReached: false, chargedCount: 0, chargeableWithinLimit: {} };
|
|
1335
|
+
}
|
|
1336
|
+
let { perRow, rows } = chargingManager._rowCharges(perRowIn, isDefaultDataset, rowsIn.length);
|
|
1337
|
+
if (Object.keys(perRow).length === 0) {
|
|
1338
|
+
// Only free events: an unpaid push (prepaid-row rules apply).
|
|
1339
|
+
const pastLimit = chargingManager.isChargeLimitReached();
|
|
1340
|
+
const allowed = isDefaultDataset ? chargingManager._takeUnpaidRows(rowsIn.length) : rowsIn.length;
|
|
1341
|
+
if (allowed < rowsIn.length)
|
|
1342
|
+
chargingManager._noteRefusedDefaultDatasetItems(rowsIn.length - allowed);
|
|
1343
|
+
if (allowed > 0)
|
|
1344
|
+
await store(rowsIn.slice(0, allowed));
|
|
1345
|
+
return { eventChargeLimitReached: pastLimit, chargedCount: 0, chargeableWithinLimit: chargingManager._chargeableWithinLimit() };
|
|
1346
|
+
}
|
|
1347
|
+
if (!voidsConfirmed && Object.keys(perRow).some((n) => !isSyntheticEvent(n))) {
|
|
1348
|
+
throw new Error('[Scrapely] An earlier charge for a failed dataset push could not be voided yet; retry the push');
|
|
1349
|
+
}
|
|
1350
|
+
const undo = async (ops) => {
|
|
1351
|
+
for (const op of ops) {
|
|
1352
|
+
if (op.charged <= 0)
|
|
1353
|
+
continue;
|
|
1354
|
+
if (!op.operationId)
|
|
1355
|
+
chargingManager.applyAcceptedCount(op.name, op.charged, 0);
|
|
1356
|
+
else if (op.uncertain)
|
|
1357
|
+
await chargingManager._cancelUncertainCharge(op.name, op.operationId, op.charged);
|
|
1358
|
+
else
|
|
1359
|
+
await chargingManager._voidCharge(op.name, op.operationId, op.charged);
|
|
1360
|
+
}
|
|
1361
|
+
};
|
|
1362
|
+
const order = Object.entries(perRow).sort(([a], [b]) => Number(isSyntheticEvent(a)) - Number(isSyntheticEvent(b)) || a.localeCompare(b));
|
|
1363
|
+
let ops = [];
|
|
1364
|
+
for (let attempt = 0; attempt < 4 && rows > 0; attempt++) {
|
|
1365
|
+
ops = [];
|
|
1366
|
+
let shortTo = null;
|
|
1367
|
+
for (const [name, n] of order) {
|
|
1368
|
+
const want = n * rows;
|
|
1369
|
+
chargingManager._takeLastAcceptedOperationId?.();
|
|
1370
|
+
let got;
|
|
1371
|
+
try {
|
|
1372
|
+
const r = await chargingManager._chargeExclusive({ eventName: name, count: want });
|
|
1373
|
+
got = r.chargedCount;
|
|
1374
|
+
ops.push({ name, charged: got, operationId: chargingManager._takeLastAcceptedOperationId?.(), uncertain: false });
|
|
1375
|
+
}
|
|
1376
|
+
catch (err) {
|
|
1377
|
+
if (!(err instanceof ChargeUncertainError)) {
|
|
1378
|
+
await undo(ops);
|
|
1379
|
+
throw err;
|
|
1380
|
+
}
|
|
1381
|
+
// Not confirmed: stays pending and is re-posted at exit, so it counts as accepted.
|
|
1382
|
+
got = err.count;
|
|
1383
|
+
ops.push({ name, charged: got, operationId: err.operationId, uncertain: true });
|
|
1384
|
+
}
|
|
1385
|
+
if (got < want) {
|
|
1386
|
+
shortTo = Math.floor(got / n);
|
|
1387
|
+
break;
|
|
1388
|
+
}
|
|
1389
|
+
}
|
|
1390
|
+
if (shortTo === null)
|
|
1391
|
+
break;
|
|
1392
|
+
// An event came back short: void this attempt and retry with the rows it pays for.
|
|
1393
|
+
await undo(ops);
|
|
1394
|
+
ops = [];
|
|
1395
|
+
rows = shortTo;
|
|
1396
|
+
}
|
|
1397
|
+
if (rows <= 0 || ops.length === 0) {
|
|
1398
|
+
await undo(ops);
|
|
1399
|
+
chargingManager._markChargeLimitHit();
|
|
1400
|
+
if (isDefaultDataset)
|
|
1401
|
+
chargingManager._noteRefusedDefaultDatasetItems(rowsIn.length);
|
|
1402
|
+
return { eventChargeLimitReached: true, chargedCount: 0, chargeableWithinLimit: chargingManager._chargeableWithinLimit() };
|
|
1403
|
+
}
|
|
1404
|
+
if (rows < rowsIn.length) {
|
|
1405
|
+
chargingManager._markChargeLimitHit();
|
|
1406
|
+
if (isDefaultDataset)
|
|
1407
|
+
chargingManager._noteRefusedDefaultDatasetItems(rowsIn.length - rows);
|
|
1408
|
+
}
|
|
1409
|
+
try {
|
|
1410
|
+
await store(rowsIn.slice(0, rows));
|
|
1411
|
+
}
|
|
1412
|
+
catch (pushErr) {
|
|
1413
|
+
console.error(`[Scrapely] Dataset push failed after charges accepted (rows=${rows}): ${pushErr.message}; voiding the charges`);
|
|
1414
|
+
if (pushCertainlyFailed(pushErr))
|
|
1415
|
+
await undo(ops);
|
|
1416
|
+
throw pushErr;
|
|
1417
|
+
}
|
|
1418
|
+
return {
|
|
1419
|
+
eventChargeLimitReached: rows < rowsIn.length || chargingManager.isChargeLimitReached(),
|
|
1420
|
+
chargedCount: rows,
|
|
1421
|
+
chargeableWithinLimit: chargingManager._chargeableWithinLimit(),
|
|
1422
|
+
};
|
|
1423
|
+
});
|
|
1424
|
+
}
|
|
1239
1425
|
/** HTTP status of an API/client error, if the server answered. */
|
|
1240
1426
|
function httpStatusOf(err) {
|
|
1241
1427
|
const e = err;
|
package/dist/index.d.ts
CHANGED
|
@@ -1,12 +1,12 @@
|
|
|
1
1
|
import './log-env.js';
|
|
2
2
|
export { setCdpUrl } from './patch-browser-pool.js';
|
|
3
|
-
export { Actor, ACTOR_PERMISSION_LEVEL, type ActorInput, type InitOptions, type MainOptions, type ExitOptions, type RebootOptions, type AbortOptions, type MetamorphOptions, type RecordOptions, type OpenStorageOptions, type CallOptions, type CallTaskOptions, type StartOptions, type SetStatusMessageOptions, type WebhookOptions, type ApifyEnv, type UserFunc, type Token, type Timeout, EXIT_CODES, } from './actor';
|
|
3
|
+
export { Actor, ACTOR_PERMISSION_LEVEL, type ActorInput, type InitOptions, type MainOptions, type ExitOptions, type RebootOptions, type AbortOptions, type PushDataChargeOptions, type MetamorphOptions, type RecordOptions, type OpenStorageOptions, type CallOptions, type CallTaskOptions, type StartOptions, type SetStatusMessageOptions, type WebhookOptions, type ApifyEnv, type UserFunc, type Token, type Timeout, EXIT_CODES, } from './actor';
|
|
4
4
|
export type { StorageAlias, StorageId, StorageName, StorageIdentifier, StorageIdentifierWithoutAlias, } from './storage-identifiers.js';
|
|
5
5
|
export { ApifyClient, type ApifyClientOptions, type ActorRun } from 'scrapely-client';
|
|
6
6
|
export { Configuration, type ConfigurationOptions } from './configuration.js';
|
|
7
7
|
export { DEGRADED_STATUS_MESSAGE, IpBanned, SolveAttemptError, SolverAuthError, SolverRateLimited, VendorUnavailable, degradedStatusMessage, detectVendor, isBlocked, reportSolveOutcome, solve, solverConfigured, solverStatus, stepAllowed, type Clearance, type SolveCookie, type SolveOptions, type SolverStepRequest, type SolverVendorStatus, } from './solver.js';
|
|
8
8
|
export { ScrapelyEventManager as PlatformEventManager } from './events/scrapely-event-manager.js';
|
|
9
|
-
export { ChargingManager, ChargeUncertainError, ChargeRejectedError, DEFAULT_DATASET_ITEM_EVENT, APIFY_DEFAULT_DATASET_ITEM_EVENT, SCRAPELY_DEFAULT_DATASET_ITEM_EVENT, hasDefaultDatasetItemEventInPricing, isDefaultDatasetItemEvent, resolveDefaultDatasetItemEventName, mergeChargeResults, pushDataAndCharge, type ChargeOptions, type ChargeResult, type ActorPricingInfo, } from './charging.js';
|
|
9
|
+
export { ChargingManager, ChargeUncertainError, ChargeRejectedError, DEFAULT_DATASET_ITEM_EVENT, APIFY_DEFAULT_DATASET_ITEM_EVENT, SCRAPELY_DEFAULT_DATASET_ITEM_EVENT, hasDefaultDatasetItemEventInPricing, isDefaultDatasetItemEvent, resolveDefaultDatasetItemEventName, mergeChargeResults, pushDataAndCharge, pushDataWithCharges, validateRowCharges, type ChargeOptions, type ChargeResult, type ActorPricingInfo, } from './charging.js';
|
|
10
10
|
export { createChargingClient, pushDataChargingContext, USES_PUSH_DATA_INTERCEPTION, type PpeAwarePushDataContext, } from './charging_client.js';
|
|
11
11
|
export { BillingTracker, type BillingClient } from './storage/billing-tracker.js';
|
|
12
12
|
export { ApiStorageClientFactory, type ApiStorageConfig } from './storage/api-storage-client.js';
|
package/dist/index.js
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
"use strict";
|
|
2
2
|
Object.defineProperty(exports, "__esModule", { value: true });
|
|
3
|
-
exports.
|
|
4
|
-
exports.puppeteerUtils = exports.playwrightUtils = exports.LinkeDOMCrawler = exports.JSDOMCrawler = exports.BrowserCrawler = exports.BasicCrawler = exports.HttpCrawler = exports.CheerioCrawler = exports.PuppeteerCrawler = exports.PlaywrightCrawler = exports.ScrapelyRequestList = exports.enqueueLinks = exports.utils = exports.LoggerText = exports.LoggerJson = exports.Logger = void 0;
|
|
3
|
+
exports.log = exports.RequestQueue = exports.KeyValueStore = exports.Dataset = exports.createBrowserConfiguration = exports.BrowserConfiguration = exports.getPlatformProxyEnv = exports.createProxyConfiguration = exports.validateProxySessionId = exports.isPlatformProxySentinel = exports.ProxyConfiguration = exports.ApiStorageClientFactory = exports.BillingTracker = exports.USES_PUSH_DATA_INTERCEPTION = exports.pushDataChargingContext = exports.createChargingClient = exports.validateRowCharges = exports.pushDataWithCharges = exports.pushDataAndCharge = exports.mergeChargeResults = exports.resolveDefaultDatasetItemEventName = exports.isDefaultDatasetItemEvent = exports.hasDefaultDatasetItemEventInPricing = exports.SCRAPELY_DEFAULT_DATASET_ITEM_EVENT = exports.APIFY_DEFAULT_DATASET_ITEM_EVENT = exports.DEFAULT_DATASET_ITEM_EVENT = exports.ChargeRejectedError = exports.ChargeUncertainError = exports.ChargingManager = exports.PlatformEventManager = exports.stepAllowed = exports.solverStatus = exports.solverConfigured = exports.solve = exports.reportSolveOutcome = exports.isBlocked = exports.detectVendor = exports.degradedStatusMessage = exports.VendorUnavailable = exports.SolverRateLimited = exports.SolverAuthError = exports.SolveAttemptError = exports.IpBanned = exports.DEGRADED_STATUS_MESSAGE = exports.Configuration = exports.ApifyClient = exports.EXIT_CODES = exports.ACTOR_PERMISSION_LEVEL = exports.Actor = exports.setCdpUrl = void 0;
|
|
4
|
+
exports.puppeteerUtils = exports.playwrightUtils = exports.LinkeDOMCrawler = exports.JSDOMCrawler = exports.BrowserCrawler = exports.BasicCrawler = exports.HttpCrawler = exports.CheerioCrawler = exports.PuppeteerCrawler = exports.PlaywrightCrawler = exports.ScrapelyRequestList = exports.enqueueLinks = exports.utils = exports.LoggerText = exports.LoggerJson = exports.Logger = exports.LogLevel = exports.Log = void 0;
|
|
5
5
|
require("./log-env.js");
|
|
6
6
|
var patch_browser_pool_js_1 = require("./patch-browser-pool.js");
|
|
7
7
|
Object.defineProperty(exports, "setCdpUrl", { enumerable: true, get: function () { return patch_browser_pool_js_1.setCdpUrl; } });
|
|
@@ -43,6 +43,8 @@ Object.defineProperty(exports, "isDefaultDatasetItemEvent", { enumerable: true,
|
|
|
43
43
|
Object.defineProperty(exports, "resolveDefaultDatasetItemEventName", { enumerable: true, get: function () { return charging_js_1.resolveDefaultDatasetItemEventName; } });
|
|
44
44
|
Object.defineProperty(exports, "mergeChargeResults", { enumerable: true, get: function () { return charging_js_1.mergeChargeResults; } });
|
|
45
45
|
Object.defineProperty(exports, "pushDataAndCharge", { enumerable: true, get: function () { return charging_js_1.pushDataAndCharge; } });
|
|
46
|
+
Object.defineProperty(exports, "pushDataWithCharges", { enumerable: true, get: function () { return charging_js_1.pushDataWithCharges; } });
|
|
47
|
+
Object.defineProperty(exports, "validateRowCharges", { enumerable: true, get: function () { return charging_js_1.validateRowCharges; } });
|
|
46
48
|
var charging_client_js_1 = require("./charging_client.js");
|
|
47
49
|
Object.defineProperty(exports, "createChargingClient", { enumerable: true, get: function () { return charging_client_js_1.createChargingClient; } });
|
|
48
50
|
Object.defineProperty(exports, "pushDataChargingContext", { enumerable: true, get: function () { return charging_client_js_1.pushDataChargingContext; } });
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "scrapely",
|
|
3
|
-
"version": "1.2.
|
|
3
|
+
"version": "1.2.23",
|
|
4
4
|
"description": "Scrapely SDK for building JavaScript and TypeScript Actors",
|
|
5
5
|
"main": "dist/index.js",
|
|
6
6
|
"types": "dist/index.d.ts",
|
|
@@ -23,6 +23,8 @@
|
|
|
23
23
|
"test": "npx tsx tests/test-mvp.ts",
|
|
24
24
|
"test:charging": "npx tsx --test tests/charging-pending.test.ts",
|
|
25
25
|
"test:charge-limit-guard": "npx tsx --test tests/charge-limit-guard.test.ts",
|
|
26
|
+
"test:charge-multi-event": "npx tsx --test tests/charge-multi-event.test.ts",
|
|
27
|
+
"test:charge-platform-budget": "npx tsx --test tests/charge-platform-budget.test.ts",
|
|
26
28
|
"test:storage-retries": "npx tsx --test tests/storage-retries.test.ts",
|
|
27
29
|
"test:push-idempotency": "npx tsx --test tests/push-idempotency.test.ts",
|
|
28
30
|
"test:stdio-drain": "npx tsx --test tests/stdio-drain.test.ts",
|