Spaces:
Build error
Build error
fix: firebase limit on document size causing cache failures
Browse files
backend/functions/src/cloud-functions/crawler.ts
CHANGED
|
@@ -13,7 +13,7 @@ import normalizeUrl from "@esm2cjs/normalize-url";
|
|
| 13 |
import { AltTextService } from '../services/alt-text';
|
| 14 |
import TurndownService from 'turndown';
|
| 15 |
import { parseString as parseSetCookieString } from 'set-cookie-parser';
|
| 16 |
-
import { CookieParam } from 'puppeteer';
|
| 17 |
import { Crawled } from '../db/crawled';
|
| 18 |
import { tidyMarkdown } from '../utils/markdown';
|
| 19 |
import { cleanAttribute } from '../utils/misc';
|
|
@@ -408,28 +408,44 @@ ${this.content}
|
|
| 408 |
|
| 409 |
const cache = (await Crawled.fromFirestoreQuery(Crawled.COLLECTION.where('urlPathDigest', '==', digest).orderBy('createdAt', 'desc').limit(1)))?.[0];
|
| 410 |
|
| 411 |
-
if (cache) {
|
| 412 |
-
|
| 413 |
-
|
| 414 |
-
this.logger.info(`${stale ? 'Only stale ' : ''}Cache exists for ${urlToCrawl}, normalized digest: ${digest}, ${age}ms old`, {
|
| 415 |
-
url: urlToCrawl, digest, age, stale
|
| 416 |
-
});
|
| 417 |
|
| 418 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 419 |
|
| 420 |
-
|
| 421 |
-
|
| 422 |
-
|
| 423 |
-
|
| 424 |
-
|
| 425 |
-
|
| 426 |
-
|
| 427 |
-
|
| 428 |
-
|
| 429 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 430 |
}
|
| 431 |
|
| 432 |
-
return
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 433 |
}
|
| 434 |
|
| 435 |
async setToCache(urlToCrawl: URL, snapshot: PageSnapshot) {
|
|
@@ -444,10 +460,24 @@ ${this.content}
|
|
| 444 |
createdAt: nowDate,
|
| 445 |
expireAt: new Date(nowDate.valueOf() + this.cacheRetentionMs),
|
| 446 |
urlPathDigest: digest,
|
| 447 |
-
|
| 448 |
-
|
| 449 |
-
|
| 450 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 451 |
});
|
| 452 |
|
| 453 |
if (snapshot.screenshot) {
|
|
@@ -458,6 +488,7 @@ ${this.content}
|
|
| 458 |
});
|
| 459 |
cache.screenshotAvailable = true;
|
| 460 |
}
|
|
|
|
| 461 |
const r = await Crawled.save(cache.degradeForFireStore()).catch((err) => {
|
| 462 |
this.logger.error(`Failed to save cache for ${urlToCrawl}`, { err: marshalErrorLike(err) });
|
| 463 |
|
|
|
|
| 13 |
import { AltTextService } from '../services/alt-text';
|
| 14 |
import TurndownService from 'turndown';
|
| 15 |
import { parseString as parseSetCookieString } from 'set-cookie-parser';
|
| 16 |
+
import type { CookieParam } from 'puppeteer';
|
| 17 |
import { Crawled } from '../db/crawled';
|
| 18 |
import { tidyMarkdown } from '../utils/markdown';
|
| 19 |
import { cleanAttribute } from '../utils/misc';
|
|
|
|
| 408 |
|
| 409 |
const cache = (await Crawled.fromFirestoreQuery(Crawled.COLLECTION.where('urlPathDigest', '==', digest).orderBy('createdAt', 'desc').limit(1)))?.[0];
|
| 410 |
|
| 411 |
+
if (!cache) {
|
| 412 |
+
return undefined;
|
| 413 |
+
}
|
|
|
|
|
|
|
|
|
|
| 414 |
|
| 415 |
+
const age = Date.now() - cache.createdAt.valueOf();
|
| 416 |
+
const stale = cache.createdAt.valueOf() < (Date.now() - this.cacheValidMs);
|
| 417 |
+
this.logger.info(`${stale ? 'Stale cache exists' : 'Cache hit'} for ${urlToCrawl}, normalized digest: ${digest}, ${age}ms old`, {
|
| 418 |
+
url: urlToCrawl, digest, age, stale
|
| 419 |
+
});
|
| 420 |
|
| 421 |
+
let snapshot: PageSnapshot | undefined;
|
| 422 |
+
let screenshotUrl: string | undefined;
|
| 423 |
+
const preparations = [
|
| 424 |
+
this.firebaseObjectStorage.downloadFile(`snapshots/${cache._id}`).then((r) => {
|
| 425 |
+
snapshot = JSON.parse(r.toString('utf-8'));
|
| 426 |
+
}),
|
| 427 |
+
cache.screenshotAvailable ?
|
| 428 |
+
this.firebaseObjectStorage.signDownloadUrl(`screenshots/${cache._id}`, Date.now() + this.urlValidMs).then((r) => {
|
| 429 |
+
screenshotUrl = r;
|
| 430 |
+
}) :
|
| 431 |
+
Promise.resolve(undefined)
|
| 432 |
+
];
|
| 433 |
+
try {
|
| 434 |
+
await Promise.all(preparations);
|
| 435 |
+
} catch (_err) {
|
| 436 |
+
// Swallow cache errors.
|
| 437 |
+
return undefined;
|
| 438 |
}
|
| 439 |
|
| 440 |
+
return {
|
| 441 |
+
isFresh: !stale,
|
| 442 |
+
...cache,
|
| 443 |
+
snapshot: {
|
| 444 |
+
...snapshot,
|
| 445 |
+
screenshot: undefined,
|
| 446 |
+
screenshotUrl,
|
| 447 |
+
} as PageSnapshot & { screenshotUrl?: string; }
|
| 448 |
+
};
|
| 449 |
}
|
| 450 |
|
| 451 |
async setToCache(urlToCrawl: URL, snapshot: PageSnapshot) {
|
|
|
|
| 460 |
createdAt: nowDate,
|
| 461 |
expireAt: new Date(nowDate.valueOf() + this.cacheRetentionMs),
|
| 462 |
urlPathDigest: digest,
|
| 463 |
+
});
|
| 464 |
+
|
| 465 |
+
const savingOfSnapshot = this.firebaseObjectStorage.saveFile(`snapshots/${cache._id}`,
|
| 466 |
+
Buffer.from(
|
| 467 |
+
JSON.stringify({
|
| 468 |
+
...snapshot,
|
| 469 |
+
screenshot: undefined
|
| 470 |
+
}),
|
| 471 |
+
'utf-8'
|
| 472 |
+
),
|
| 473 |
+
{
|
| 474 |
+
metadata: {
|
| 475 |
+
contentType: 'application/json',
|
| 476 |
+
}
|
| 477 |
+
}
|
| 478 |
+
).then((r) => {
|
| 479 |
+
cache.snapshotAvailable = true;
|
| 480 |
+
return r;
|
| 481 |
});
|
| 482 |
|
| 483 |
if (snapshot.screenshot) {
|
|
|
|
| 488 |
});
|
| 489 |
cache.screenshotAvailable = true;
|
| 490 |
}
|
| 491 |
+
await savingOfSnapshot;
|
| 492 |
const r = await Crawled.save(cache.degradeForFireStore()).catch((err) => {
|
| 493 |
this.logger.error(`Failed to save cache for ${urlToCrawl}`, { err: marshalErrorLike(err) });
|
| 494 |
|
backend/functions/src/db/crawled.ts
CHANGED
|
@@ -22,11 +22,14 @@ export class Crawled extends FirestoreRecord {
|
|
| 22 |
urlPathDigest!: string;
|
| 23 |
|
| 24 |
@Prop()
|
| 25 |
-
snapshot
|
| 26 |
|
| 27 |
@Prop()
|
| 28 |
screenshotAvailable?: boolean;
|
| 29 |
|
|
|
|
|
|
|
|
|
|
| 30 |
@Prop()
|
| 31 |
createdAt!: Date;
|
| 32 |
|
|
|
|
| 22 |
urlPathDigest!: string;
|
| 23 |
|
| 24 |
@Prop()
|
| 25 |
+
snapshot?: PageSnapshot & { screenshot: never; };
|
| 26 |
|
| 27 |
@Prop()
|
| 28 |
screenshotAvailable?: boolean;
|
| 29 |
|
| 30 |
+
@Prop()
|
| 31 |
+
snapshotAvailable?: boolean;
|
| 32 |
+
|
| 33 |
@Prop()
|
| 34 |
createdAt!: Date;
|
| 35 |
|
thinapps-shared
CHANGED
|
@@ -1 +1 @@
|
|
| 1 |
-
Subproject commit
|
|
|
|
| 1 |
+
Subproject commit 64157bc57ef9ce2cec69f37b5f55fccb71742b6f
|