From 0cc7e84a82cd57f58966a41a39f93a59b2fbde09 Mon Sep 17 00:00:00 2001 From: Hongbo Wu Date: Wed, 18 May 2022 15:52:16 +0800 Subject: [PATCH] Fix content not getting parsed by linkedom properly without tag by replacing innerHtml with outerHtml --- packages/content-fetch/fetch-content.js | 4 ++-- packages/puppeteer-parse/index.js | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/packages/content-fetch/fetch-content.js b/packages/content-fetch/fetch-content.js index 8844c2573..1c948fccc 100644 --- a/packages/content-fetch/fetch-content.js +++ b/packages/content-fetch/fetch-content.js @@ -497,7 +497,7 @@ async function retrievePage(url) { logRecord.finalUrl = response.url(); logRecord.contentType = response.headers()['content-type']; - return { context, page, response, finalUrl: finalUrl, contentType: contentType }; + return { context, page, response, finalUrl, contentType }; } catch (error) { if (lastPdfUrl) { return { context, page, finalUrl: lastPdfUrl, contentType: 'application/pdf' }; @@ -609,7 +609,7 @@ async function retrieveHtml(page) { } } }); - return document.documentElement.innerHTML; + return document.documentElement.outerHTML; }, iframes); logRecord.puppeteerSuccess = true; logRecord.timing = { diff --git a/packages/puppeteer-parse/index.js b/packages/puppeteer-parse/index.js index 505b3203e..a2933147e 100644 --- a/packages/puppeteer-parse/index.js +++ b/packages/puppeteer-parse/index.js @@ -678,7 +678,7 @@ async function retrievePage(url) { logRecord.finalUrl = response.url(); logRecord.contentType = response.headers()['content-type']; - return { context, page, response, finalUrl: finalUrl, contentType: contentType }; + return { context, page, response, finalUrl, contentType }; } catch (error) { if (lastPdfUrl) { return { context, page, finalUrl: lastPdfUrl, contentType: 'application/pdf' }; @@ -790,7 +790,7 @@ async function retrieveHtml(page) { } } }); - return document.documentElement.innerHTML; + return document.documentElement.outerHTML; }, iframes); logRecord.puppeteerSuccess = true; logRecord.timing = {