From c9b8c2e1a5b37643349befe17988513ad598df62 Mon Sep 17 00:00:00 2001 From: = Date: Thu, 10 Sep 2020 13:31:57 -0700 Subject: [PATCH] Fiks --- app/crawler/specificCrawlers/olx.js | 6 ++++-- package-lock.json | 4 +--- 2 files changed, 5 insertions(+), 5 deletions(-) diff --git a/app/crawler/specificCrawlers/olx.js b/app/crawler/specificCrawlers/olx.js index a979089..405a613 100644 --- a/app/crawler/specificCrawlers/olx.js +++ b/app/crawler/specificCrawlers/olx.js @@ -178,6 +178,7 @@ class OlxCrawler { this.maxResultsPerPage ); + await this.sleep(this.delayBetweenPages); if (Array.isArray(singlePageResults) && singlePageResults.length > 0) { yield singlePageResults; } else { @@ -227,10 +228,11 @@ class OlxCrawler { const { scrapedData, errors } = await PromisePool - .withConcurrency(3) + .withConcurrency(2) .for(asyncScraping) .process(async data => { const result = await this.scrapeAd(data) + await this.sleep(this.delayBetweenPages); return result; }) @@ -243,7 +245,7 @@ class OlxCrawler { } async scrapeAd(url) { - // console.log("Scraping : ", url); + //console.log("Scraping : ", url); try { const adPageSource = await fetch(url); const body = await adPageSource.text(); diff --git a/package-lock.json b/package-lock.json index f45e4f8..7bee37e 100644 --- a/package-lock.json +++ b/package-lock.json @@ -43,14 +43,12 @@ "@supercharge/goodies": { "version": "1.4.0", "resolved": "https://registry.npmjs.org/@supercharge/goodies/-/goodies-1.4.0.tgz", - "integrity": "sha512-Np6u2qjRwiA3wTgzz4n2yduydIjSXqtJWP5cOnNqjdlCR/EUAK86LAOhEcU+YW211D1ksugns3GqpARJDoXQ7g==", - "dev": true + "integrity": "sha512-Np6u2qjRwiA3wTgzz4n2yduydIjSXqtJWP5cOnNqjdlCR/EUAK86LAOhEcU+YW211D1ksugns3GqpARJDoXQ7g==" }, "@supercharge/promise-pool": { "version": "1.3.0", "resolved": "https://registry.npmjs.org/@supercharge/promise-pool/-/promise-pool-1.3.0.tgz", "integrity": "sha512-9/EVrJevSPEqI4i/gRH8Dt7C+FQT65wRRYuu0MDaGmSLZ2aTel0jOGu8Ae84fPiQ+Ah0B80RPFUxk+K+Cz48DA==", - "dev": true, "requires": { "@supercharge/goodies": "~1.4.0" }