From cbc9239850c956c8b95ec915eee998735d959c87 Mon Sep 17 00:00:00 2001
From: Allen Kinzalow <allen.kinzalow@gmail.com>
Date: Fri, 22 Sep 2023 15:51:50 -0500
Subject: [PATCH] [#53] Fix css selector for news post content container

---
 .changeset/metal-jokes-poke.md                     | 5 +++++
 package.json                                       | 2 +-
 src/scrapers/news/news.ts                          | 2 +-
 src/scrapers/news/transformers/breakTransformer.ts | 5 ++++-
 yarn.lock                                          | 8 ++++----
 5 files changed, 15 insertions(+), 7 deletions(-)
 create mode 100644 .changeset/metal-jokes-poke.md

diff --git a/.changeset/metal-jokes-poke.md b/.changeset/metal-jokes-poke.md
new file mode 100644
index 0000000..06dfbff
--- /dev/null
+++ b/.changeset/metal-jokes-poke.md
@@ -0,0 +1,5 @@
+---
+"osrs-web-scraper": patch
+---
+
+Fix css selector for news post content container
diff --git a/package.json b/package.json
index 65d8746..37052b6 100644
--- a/package.json
+++ b/package.json
@@ -29,7 +29,7 @@
     "date-fns": "^2.30.0",
     "dotenv": "^16.0.3",
     "image-size": "^1.0.2",
-    "node-html-parser": "^6.1.5",
+    "node-html-parser": "^6.1.10",
     "puppeteer": "^20.8.0",
     "rss-parser": "^3.13.0",
     "tsconfig-paths": "^4.2.0",
diff --git a/src/scrapers/news/news.ts b/src/scrapers/news/news.ts
index b0165e8..c41d035 100644
--- a/src/scrapers/news/news.ts
+++ b/src/scrapers/news/news.ts
@@ -20,7 +20,7 @@ const news: ScrapingService<MediaWikiBuilder> = {
         const $ = window.$;
         const title = $(".news-article-header__title").html();
         const headerHtml = $(".news-article-header").html();
-        const contentHtml = $(".news-article-content #article-top").html();
+        const contentHtml = $(".news-article-content").html();
         return {
           title,
           header: headerHtml,
diff --git a/src/scrapers/news/transformers/breakTransformer.ts b/src/scrapers/news/transformers/breakTransformer.ts
index ece10a3..87daf9c 100644
--- a/src/scrapers/news/transformers/breakTransformer.ts
+++ b/src/scrapers/news/transformers/breakTransformer.ts
@@ -22,7 +22,10 @@ class NewsBreakTransformer extends MediaWikiTransformer {
           after instanceof MediaWikiBreak
         ) {
           index++;
-        } else if (before instanceof MediaWikiHeader) {
+        } else if (
+          before instanceof MediaWikiHeader &&
+          after instanceof MediaWikiBreak
+        ) {
           continue;
         } else {
           transformedContent.push(current);
diff --git a/yarn.lock b/yarn.lock
index 45c410b..036582f 100644
--- a/yarn.lock
+++ b/yarn.lock
@@ -4161,10 +4161,10 @@ node-fetch@^2.6.12:
   dependencies:
     whatwg-url "^5.0.0"
 
-node-html-parser@^6.1.5:
-  version "6.1.5"
-  resolved "https://registry.yarnpkg.com/node-html-parser/-/node-html-parser-6.1.5.tgz#c819dceb13a10a7642ff92f94f870b4f77968097"
-  integrity sha512-fAaM511feX++/Chnhe475a0NHD8M7AxDInsqQpz6x63GRF7xYNdS8Vo5dKsIVPgsOvG7eioRRTZQnWBrhDHBSg==
+node-html-parser@^6.1.10:
+  version "6.1.10"
+  resolved "https://registry.yarnpkg.com/node-html-parser/-/node-html-parser-6.1.10.tgz#5db11eac3ccbea6fc1b04a22c8a0e3a0774cfae6"
+  integrity sha512-6/uWdWxjQWQ7tMcFK2wWlrflsQUzh1HsEzlIf2j5+TtzfhT2yUvg3DwZYAmjEHeR3uX74ko7exjHW69J0tOzIg==
   dependencies:
     css-select "^5.1.0"
     he "1.2.0"