Add playwright documentation

assets/javascripts/news.json

@@ -1,4 +1,8 @@
 [
+  [
+    "2024-01-24",
+    "New documentation: <a href=\"/playwright/\">Playwright</a>"
+  ],
   [
     "2024-01-20",
     "New documentation: <a href=\"/htmx/\">htmx</a>"

docs/file-scrapers.md

@@ -167,8 +167,8 @@ download it, extract it with `dpkg -x $PACKAGE ./` and move `./usr/share/doc/ope
 to `path/to/devdocs/docs/openjdk~$VERSION`
 
 ```sh
-curl -O http://ftp.at.debian.org/debian/pool/main/o/openjdk-21/openjdk-21-doc_21.0.1+12-3_all.deb
-tar xf openjdk-21-doc_21.0.1+12-3_all.deb
+curl -O http://ftp.at.debian.org/debian/pool/main/o/openjdk-21/openjdk-21-doc_21.0.2+13-2_all.deb
+tar xf openjdk-21-doc_21.0.2+13-2_all.deb
 tar xf data.tar.xz
 mv ./usr/share/doc/openjdk-21-jre-headless/api/ docs/openjdk~$VERSION
 ```

lib/docs/filters/openjdk/clean_html_new.rb

@@ -12,6 +12,7 @@ def call
         end
 
         css('.header .sub-title', 'hr', '.table-tabs').remove
+        css('.copy').remove
 
         # fix ul section that contains summaries or tables
         css('ul').each do |node|

lib/docs/filters/playwright/clean_html.rb

@@ -0,0 +1,29 @@
+module Docs
+  class Playwright
+    class CleanHtmlFilter < Filter
+      def call
+        @doc = at_css('.markdown')
+
+        css('x-search').remove
+        css('hr').remove
+        css('font:contains("Added in")').remove
+        css('.list-anchor').remove
+
+        css('.alert').each do |node|
+          node.name = 'blockquote'
+        end
+
+        css('pre').each do |node|
+          node.content = node.css('.token-line').map(&:content).join("\n")
+          node.remove_attribute('style')
+          node['data-language'] = node.content =~ /\A\s*</ ? 'html' : 'javascript'
+          node.ancestors('.theme-code-block').first.replace(node)
+        end
+
+        css('*[class]').remove_attribute('class')
+
+        doc
+      end
+    end
+  end
+end

lib/docs/filters/playwright/entries.rb

@@ -0,0 +1,23 @@
+module Docs
+  class Playwright
+    class EntriesFilter < Docs::EntriesFilter
+      def get_name
+        at_css('h1').children.select(&:text?).map(&:content).join.strip
+      end
+
+      def type
+        type = at_css('.menu__link--active').content
+        return "#{type}: #{name}" if slug.starts_with?('api/')
+        type
+      end
+
+      def additional_entries
+        css('x-search').each_with_object [] do |node, entries|
+          prev = node.previous_element
+          prev = prev.previous_element until prev['id']
+          entries << [node.text, prev['id']]
+        end
+      end
+    end
+  end
+end

lib/docs/scrapers/playwright.rb

@@ -0,0 +1,27 @@
+module Docs
+  class Playwright < UrlScraper
+    self.name = 'Playwright'
+    self.type = 'simple'
+    self.release = '1.41.1'
+    self.base_url = 'https://playwright.dev/docs/'
+    self.root_path = 'intro'
+    self.links = {
+      home: 'https://playwright.dev/',
+      code: 'https://github.com/microsoft/playwright'
+    }
+
+    # Docusaurus like react_native
+    html_filters.push 'playwright/entries', 'playwright/clean_html'
+    options[:download_images] = false
+
+	# https://github.com/microsoft/playwright/blob/main/LICENSE
+    options[:attribution] = <<-HTML
+      &copy; 2024 Microsoft<br>
+	  Licensed under the Apache License, Version 2.0.
+    HTML
+
+    def get_latest_version(opts)
+      get_npm_version('@playwright/test', opts)
+      end
+  end
+end

public/icons/docs/playwright/SOURCE

@@ -0,0 +1 @@
+https://playwright.dev/img/playwright-logo.svg