saas: refactor proxy provider (#1186)

.github/workflows/cd.yml

@@ -84,6 +84,6 @@ jobs:
       - name: Deploy SEARCH-EU with Tag
         run: |
           gcloud beta run deploy search-eu --image us-docker.pkg.dev/reader-6b7dc/jina-reader/reader@${{steps.container.outputs.imageid}} --tag ${{ env.RELEASE_VERSION }} --command '' --args build/stand-alone/search.js --region europe-west1 --async --min-instances 0 --deploy-health-check --use-http2
-      - name: Deploy SERP-JP with Tag
+      - name: Deploy SERP-HK with Tag
         run: |
-          gcloud beta run deploy serp-jp --image us-docker.pkg.dev/reader-6b7dc/jina-reader/reader@${{steps.container.outputs.imageid}} --tag ${{ env.RELEASE_VERSION }} --command '' --args build/stand-alone/serp.js --region asia-northeast1 --async --min-instances 0 --deploy-health-check --use-http2
+          gcloud beta run deploy serp-hk --image us-docker.pkg.dev/reader-6b7dc/jina-reader/reader@${{steps.container.outputs.imageid}} --tag ${{ env.RELEASE_VERSION }} --command '' --args build/stand-alone/serp.js --region asia-east2 --async --min-instances 0 --deploy-health-check --use-http2

src/api/crawler.ts

@@ -41,13 +41,13 @@ import {
 } from '../services/errors';
 
 import { countGPTToken as estimateToken } from '../shared/utils/openai';
-import { ProxyProvider } from '../shared/services/proxy-provider';
+import { ProxyProviderService } from '../shared/services/proxy-provider';
 import { FirebaseStorageBucketControl } from '../shared/services/firebase-storage-bucket';
 import { JinaEmbeddingsAuthDTO } from '../dto/jina-embeddings-auth';
 import { RobotsTxtService } from '../services/robots-text';
 import { TempFileManager } from '../services/temp-file';
 import { MiscService } from '../services/misc';
-import { HTTPServiceError } from 'civkit';
+import { HTTPServiceError } from 'civkit/http';
 import { GeoIPService } from '../services/geoip';
 
 export interface ExtraScrappingOptions extends ScrappingOptions {
@@ -87,7 +87,7 @@ export class CrawlerHost extends RPCHost {
         protected puppeteerControl: PuppeteerControl,
         protected curlControl: CurlControl,
         protected cfBrowserRendering: CFBrowserRendering,
-        protected proxyProvider: ProxyProvider,
+        protected proxyProvider: ProxyProviderService,
         protected lmControl: LmControl,
         protected jsdomControl: JSDomControl,
         protected snapshotFormatter: SnapshotFormatter,
@@ -1232,6 +1232,7 @@ export class CrawlerHost extends RPCHost {
         };
     }
 
+    proxyIterMap = new WeakMap<ExtraScrappingOptions, ReturnType<ProxyProviderService['iterAlloc']>>();
     @retryWith((err) => {
         if (err instanceof ServiceBadApproachError) {
             return false;
@@ -1250,8 +1251,17 @@ export class CrawlerHost extends RPCHost {
         if (opts?.allocProxy === 'none') {
             return this.curlControl.sideLoad(url, opts);
         }
+        let proxy;
+        if (opts) {
+            let it = this.proxyIterMap.get(opts);
+            if (!it) {
+                it = this.proxyProvider.iterAlloc(this.figureOutBestProxyCountry(opts));
+                this.proxyIterMap.set(opts, it);
+            }
+            proxy = (await it.next()).value;
+        }
 
-        const proxy = await this.proxyProvider.alloc(this.figureOutBestProxyCountry(opts));
+        proxy ??= await this.proxyProvider.alloc(this.figureOutBestProxyCountry(opts));
         this.logger.debug(`Proxy allocated`, { proxy: proxy.href });
         const r = await this.curlControl.sideLoad(url, {
             ...opts,

src/services/serp/google.ts

@@ -12,7 +12,7 @@ import { ApplicationError } from 'civkit/civ-rpc';
 import { ServiceBadApproachError, ServiceBadAttemptError } from '../errors';
 import { parseJSONText } from 'civkit/vectorize';
 import { retryWith } from 'civkit/decorators';
-import { ProxyProvider } from '../../shared/services/proxy-provider';
+import { ProxyProviderService } from '../../shared/services/proxy-provider';
 
 @singleton()
 export class GoogleSERP extends AsyncService {
@@ -24,7 +24,7 @@ export class GoogleSERP extends AsyncService {
         protected puppeteerControl: SERPSpecializedPuppeteerControl,
         protected jsDomControl: JSDomControl,
         protected curlControl: CurlControl,
-        protected proxyProvider: ProxyProvider,
+        protected proxyProvider: ProxyProviderService,
     ) {
         const filteredDeps = isMainThread ? arguments : _.without(arguments, puppeteerControl);
         super(...filteredDeps);

thinapps-shared

@@ -1 +1 @@
-Subproject commit 424f50ca8b6277d74185e16aa67ff2b366d9f727
+Subproject commit 6fac86977536a7b7440edba8d4cf2a1f0e769e8c