Add Hashgraph Online docs scraper (standards-sdk + registry-broker)

lib/docs/filters/hol/clean_html.rb

@@ -0,0 +1,42 @@
+module Docs
+  class Hol
+    class CleanHtmlFilter < Filter
+      def call
+        @doc = at_css('article .theme-doc-markdown') || at_css('article')
+        return doc if @doc.nil?
+
+        css(
+          '.theme-doc-breadcrumbs',
+          '.theme-doc-toc-mobile',
+          '.theme-doc-footer',
+          '.theme-doc-version-badge',
+          '.pagination-nav',
+          '.theme-edit-this-page',
+          '.hash-link',
+          '.anchor-link',
+          'button.copyButtonIcon_Lhsm',
+          'button.clean-btn',
+        ).remove
+
+        css('pre').each do |node|
+          lines = node.css('.token-line')
+          node.content = lines.map(&:content).join("\n") if lines.any?
+          node.remove_attribute('style')
+
+          language = node['class'].to_s[/language-([a-z0-9_-]+)/i, 1]
+          language ||= node.at_css('code')&.[]('class').to_s[/language-([a-z0-9_-]+)/i, 1]
+          node['data-language'] = language if language
+
+          wrapper = node.ancestors('.theme-code-block').first
+          wrapper.replace(node) if wrapper
+        end
+
+        css('.table-of-contents').remove
+        css('*[class]').remove_attribute('class')
+        css('*[style]').remove_attribute('style')
+
+        doc
+      end
+    end
+  end
+end

lib/docs/filters/hol/entries.rb

@@ -0,0 +1,77 @@
+module Docs
+  class Hol
+    class EntriesFilter < Docs::EntriesFilter
+      def include_default_entry?
+        !root_page?
+      end
+
+      def get_name
+        heading = at_css('h1')
+        return super if heading.nil?
+        normalized_heading_text(heading)
+      end
+
+      def get_type
+        return standards_sdk_type if standards_sdk_doc?
+        return registry_broker_type if registry_broker_doc?
+        nil
+      end
+
+      def additional_entries
+        return [] if root_page?
+
+        css('h2[id], h3[id]').each_with_object([]) do |node, entries|
+          section_name = normalized_heading_text(node)
+          next if section_name.empty?
+          next if section_name == name
+          entries << ["#{name}: #{section_name}", node['id']]
+        end
+      end
+
+      private
+
+      def normalized_heading_text(node)
+        fragment = node.dup
+        fragment.css('a').remove
+        fragment.content.gsub(/\s+/, ' ').strip
+      end
+
+      def standards_sdk_doc?
+        slug.start_with?('libraries/standards-sdk/')
+      end
+
+      def registry_broker_doc?
+        slug.start_with?('registry-broker/')
+      end
+
+      def standards_sdk_type
+        sdk_slug = slug.delete_prefix('libraries/standards-sdk/').split('/').first
+        return 'Standards SDK' if sdk_slug.nil? || sdk_slug.empty?
+        return sdk_slug.upcase if sdk_slug.match?(/\Ahcs-\d+\z/)
+
+        case sdk_slug
+        when 'registry-broker-client'
+          'SDK Registry Broker Client'
+        when 'utils-services'
+          'SDK Utilities & Services'
+        else
+          "SDK #{sdk_slug.tr('-', ' ').split.map(&:capitalize).join(' ')}"
+        end
+      end
+
+      def registry_broker_type
+        broker_slug = slug.delete_prefix('registry-broker/').split('/').first
+        return 'Registry Broker' if broker_slug.nil? || broker_slug.empty?
+
+        case broker_slug
+        when 'api'
+          'Registry Broker API'
+        when 'chat', 'encrypted-chat', 'multi-protocol-chat'
+          'Registry Broker Chat'
+        else
+          "Registry Broker #{broker_slug.tr('-', ' ').split.map(&:capitalize).join(' ')}"
+        end
+      end
+    end
+  end
+end

lib/docs/scrapers/hol.rb

@@ -0,0 +1,39 @@
+module Docs
+  class Hol < UrlScraper
+    self.name = 'Hashgraph Online'
+    self.slug = 'hol'
+    self.type = 'simple'
+    self.release = '0.1.161'
+    self.base_url = 'https://hol.org/docs/'
+    self.root_path = 'libraries/standards-sdk/'
+    self.initial_paths = %w(
+      libraries/standards-sdk/
+      registry-broker/
+    )
+    self.links = {
+      home: 'https://hol.org/',
+      code: 'https://github.com/hashgraph-online/standards-sdk'
+    }
+
+    html_filters.push 'hol/entries', 'hol/clean_html'
+
+    options[:trailing_slash] = true
+    options[:only_patterns] = [
+      %r{\A(?:libraries/standards-sdk|registry-broker)(?:/|$)},
+    ]
+    options[:skip_patterns] = [
+      %r{\Aregistry-broker/examples/(?:chat-demo|ping-agent-demo)/?},
+      %r{\Aregistry-broker/getting-started/(?:faq|first-registration|installation|quick-start)/?},
+      %r{\Aregistry-broker/(?:feature-your-agent|moltbook|partner-program|skills-upload-discovery|updating-agents)/?},
+    ]
+
+    options[:attribution] = <<-HTML
+      Copyright &copy; 2025 Hashgraph Online DAO.<br>
+      Licensed under the Apache License 2.0.
+    HTML
+
+    def get_latest_version(opts)
+      get_npm_version('@hashgraphonline/standards-sdk', opts)
+    end
+  end
+end

public/icons/docs/hol/SOURCE

@@ -0,0 +1 @@
+https://hol.org/logo.png