AI/reader
1
0
Fork 0
mirror of https://git.mirrors.martin98.com/https://github.com/jina-ai/reader synced 2025-08-16 12:56:00 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix

Browse Source
This commit is contained in:
Yanlong Wang 2024-07-25 14:56:19 +08:00
parent 09dbbd3b0f
commit 9bcde30f11
No known key found for this signature in database
GPG Key ID: C0A623C0BADF9F37
3 changed files with 27 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
backend/functions/src/cloud-functions/crawler.ts
View File

@ -776,7 +776,9 @@ ${suffixMixins.length ? `\n${suffixMixins.join('\n\n')}\n` : ''}`;
getUrlDigest(urlToCrawl: URL) { getUrlDigest(urlToCrawl: URL) {
const normalizedURL = new URL(urlToCrawl); const normalizedURL = new URL(urlToCrawl);
normalizedURL.hash = ''; if (!normalizedURL.hash.startsWith('/')) {
normalizedURL.hash = '';
}
const normalizedUrl = normalizedURL.toString().toLowerCase(); const normalizedUrl = normalizedURL.toString().toLowerCase();
const digest = md5Hasher.hash(normalizedUrl.toString()); const digest = md5Hasher.hash(normalizedUrl.toString());

14
backend/functions/src/cloud-functions/searcher.ts
View File

@ -5,7 +5,7 @@ import {
objHashMd5B64Of, objHashMd5B64Of,
} from 'civkit'; } from 'civkit';
import { singleton } from 'tsyringe'; import { singleton } from 'tsyringe';
import { AsyncContext, CloudHTTPv2, Ctx, InsufficientBalanceError, Logger, OutputServerEventStream, RPCReflect } from '../shared'; import { AsyncContext, CloudHTTPv2, Ctx, InsufficientBalanceError, Logger, OutputServerEventStream, Param, RPCReflect } from '../shared';
import { RateLimitControl, RateLimitDesc } from '../shared/services/rate-limit'; import { RateLimitControl, RateLimitDesc } from '../shared/services/rate-limit';
import _ from 'lodash'; import _ from 'lodash';
import { Request, Response } from 'express'; import { Request, Response } from 'express';
@ -83,6 +83,8 @@ export class SearcherHost extends RPCHost {
res: Response, res: Response,
}, },
auth: JinaEmbeddingsAuthDTO, auth: JinaEmbeddingsAuthDTO,
@Param('count', { default: 5, validate: (v) => v >= 3 && v <= 10 })
count: number,
crawlerOptions: CrawlerOptions, crawlerOptions: CrawlerOptions,
braveSearchExplicitOperators: BraveSearchExplicitOperatorsDto, braveSearchExplicitOperators: BraveSearchExplicitOperatorsDto,
) { ) {
@ -157,7 +159,7 @@ export class SearcherHost extends RPCHost {
const searchQuery = braveSearchExplicitOperators.addTo(ctx.req.path.slice(1)); const searchQuery = braveSearchExplicitOperators.addTo(ctx.req.path.slice(1));
const r = await this.cachedWebSearch({ const r = await this.cachedWebSearch({
q: searchQuery, q: searchQuery,
count: 10 count: Math.floor(count * 2)
}, crawlerOptions.noCache); }, crawlerOptions.noCache);
if (!r.web?.results.length) { if (!r.web?.results.length) {
@ -226,7 +228,7 @@ export class SearcherHost extends RPCHost {
if (_.some(scrapped, (x) => this.pageQualified(x))) { if (_.some(scrapped, (x) => this.pageQualified(x))) {
setEarlyReturnTimer(); setEarlyReturnTimer();
} }
if (!this.searchResultsQualified(scrapped)) { if (!this.searchResultsQualified(scrapped, count)) {
continue; continue;
} }
if (earlyReturnTimer) { if (earlyReturnTimer) {
@ -274,7 +276,7 @@ export class SearcherHost extends RPCHost {
setEarlyReturnTimer(); setEarlyReturnTimer();
} }
if (!this.searchResultsQualified(scrapped)) { if (!this.searchResultsQualified(scrapped, count)) {
continue; continue;
} }
@ -425,8 +427,8 @@ ${suffixMixins.length ? `\n${suffixMixins.join('\n')}\n` : ''}`;
formattedPage.html; formattedPage.html;
} }
searchResultsQualified(results: FormattedPage[]) { searchResultsQualified(results: FormattedPage[], targetResultCount = this.targetResultCount) {
return _.every(results, (x) => this.pageQualified(x)) && results.length >= this.targetResultCount; return _.every(results, (x) => this.pageQualified(x)) && results.length >= targetResultCount;
} }
async cachedWebSearch(query: WebSearchQueryParams, noCache: boolean = false) { async cachedWebSearch(query: WebSearchQueryParams, noCache: boolean = false) {

24
backend/functions/src/services/brave-search.ts
View File

@ -1,4 +1,4 @@
import { AsyncService, AutoCastable, DownstreamServiceFailureError, Prop, RPC_CALL_ENVIRONMENT, marshalErrorLike, retry } from 'civkit'; import { AsyncService, AutoCastable, DownstreamServiceFailureError, Prop, RPC_CALL_ENVIRONMENT, delay, marshalErrorLike } from 'civkit';
import { singleton } from 'tsyringe'; import { singleton } from 'tsyringe';
import { Logger } from '../shared/services/logger'; import { Logger } from '../shared/services/logger';
import { SecretExposer } from '../shared/services/secrets'; import { SecretExposer } from '../shared/services/secrets';
@ -31,7 +31,6 @@ export class BraveSearchService extends AsyncService {
this.braveSearchHTTP = new BraveSearchHTTP(this.secretExposer.BRAVE_SEARCH_API_KEY); this.braveSearchHTTP = new BraveSearchHTTP(this.secretExposer.BRAVE_SEARCH_API_KEY);
} }
@retry(3, Math.ceil(500 + 500 * Math.random()))
async webSearch(query: WebSearchQueryParams) { async webSearch(query: WebSearchQueryParams) {
const ip = this.threadLocal.get('ip'); const ip = this.threadLocal.get('ip');
const extraHeaders: WebSearchOptionalHeaderOptions = {}; const extraHeaders: WebSearchOptionalHeaderOptions = {};
@ -65,16 +64,25 @@ export class BraveSearchService extends AsyncService {
encoded.q = (Buffer.from(encoded.q).toString('ascii') === encoded.q) ? encoded.q : encodeURIComponent(encoded.q); encoded.q = (Buffer.from(encoded.q).toString('ascii') === encoded.q) ? encoded.q : encodeURIComponent(encoded.q);
} }
try { let maxTries = 11;
const r = await this.braveSearchHTTP.webSearch(encoded, { headers: extraHeaders as Record<string, string> });
return r.parsed; while (maxTries--) {
} catch (err: any) { try {
this.logger.error(`Web search failed: ${err?.message}`, { err: marshalErrorLike(err) }); const r = await this.braveSearchHTTP.webSearch(encoded, { headers: extraHeaders as Record<string, string> });
throw new DownstreamServiceFailureError({ message: `Search failed` }); return r.parsed;
} catch (err: any) {
this.logger.error(`Web search failed: ${err?.message}`, { err: marshalErrorLike(err) });
if (err?.status === 429) {
await delay(500 + 1000 * Math.random());
continue;
}
throw new DownstreamServiceFailureError({ message: `Search failed` });
}
} }
throw new DownstreamServiceFailureError({ message: `Search failed` });
} }
} }