Fixed arstechnica content-script

- the body selector is now div.post-content
- removed the now useless replaceStrings
- give priority to {Article,NewsArticle,Web}.description in JSON-LD,
  before *.description

See https://community.readeck.org/d/113-some-webpages-content-not-being-extracted

pkg/extract/contentscripts/assets/scripts/0-config.js

@@ -18,11 +18,11 @@ exports.isActive = function () {
  */
 exports.setConfig = function (config) {
   switch (true) {
+    case $.domain == "arstechnica.co.uk":
+      $.overrideConfig(config, "https://arstechnica.com/")
     case $.domain == "arstechnica.com":
-      config.replaceStrings = [
-        ['" data-src="', '"><img src="'],
-        ['" data-responsive="', '" /><span data-responsive="'],
-        ['<figure style="', '</span><figure data-style="'],
+      config.bodySelectors = [
+        "//div[contains(concat(' ',normalize-space(@class),' '),' post-content ')]",
       ]
       break
 

pkg/extract/meta/meta.go

@@ -58,7 +58,7 @@ func ExtractMeta(m *extract.ProcessMessage, next extract.Processor) extract.Proc
 		if headline, ok := md.getProp("Article.name", "*.headline", "{Movie,VideObject}.name").(string); ok {
 			d.Title = headline
 		}
-		if description, ok := md.getProp("*.description").(string); ok {
+		if description, ok := md.getProp("{Article,NewsArticle,WebPage}.description", "*.description").(string); ok {
 			d.Description = description
 		}
 		if image, ok := md.getProp("*.{image,image.url,thumbnailUrl}").(string); ok {