[devdocsgjs/main: 34/64] Add nix and nixpkgs

[Andy Holmes <andyholmes src gnome org>]

commit f1dd2acc0851267b280504d971d8bd37e97681f1
Author: Thomas Dy <thatsmydoing gmail com>
Date:   Sun Apr 24 18:28:13 2022 +0900

    Add nix and nixpkgs
    
    This only adds a subset of nix and nixpkgs functions. Notably
    derivation, stdenv and language-specific functions are omitted for now
    as their formats differ significantly and will require more cleanup.

 .../javascripts/templates/pages/about_tmpl.coffee  |   5 +
 lib/docs/filters/nix/clean_html.rb                 | 113 +++++++++++++++++++++
 lib/docs/filters/nix/entries.rb                    |  22 ++++
 lib/docs/scrapers/nix.rb                           |  31 ++++++
 public/icons/docs/nix/16.png                       | Bin 0 -> 1220 bytes
 public/icons/docs/nix/16 2x png                    | Bin 0 -> 1936 bytes
 public/icons/docs/nix/SOURCE                       |   1 +
 7 files changed, 172 insertions(+)
---
diff --git a/assets/javascripts/templates/pages/about_tmpl.coffee 
b/assets/javascripts/templates/pages/about_tmpl.coffee
index a366449a..6fdba573 100644
--- a/assets/javascripts/templates/pages/about_tmpl.coffee
+++ b/assets/javascripts/templates/pages/about_tmpl.coffee
@@ -587,6 +587,11 @@ credits = [
     '2006-2021 Andreas Rumpf',
     'MIT',
     'https://github.com/nim-lang/Nim#license'
+  ], [
+    'Nix',
+    '2022 NixOS Contributors',
+    'LGPLv2.1',
+    'https://github.com/NixOS/nix#license'
   ], [
     'Node.js',
     'Joyent, Inc. and other Node contributors<br>Node.js is a trademark of Joyent, Inc.',
diff --git a/lib/docs/filters/nix/clean_html.rb b/lib/docs/filters/nix/clean_html.rb
new file mode 100644
index 00000000..731fc0cf
--- /dev/null
+++ b/lib/docs/filters/nix/clean_html.rb
@@ -0,0 +1,113 @@
+module Docs
+  class Nix
+    class CleanHtmlFilter < Filter
+      def call
+        if subpath == 'nixpkgs/stable/index.html'
+          new_root = Nokogiri::XML::Node.new 'div', doc.document
+
+          # lib functions
+          lib_sections = xpath("//*[@id='sec-functions-library']/ancestor::section[1]/section/section")
+          lib_sections.css('.titlepage .title').each do |title|
+            title.name = 'h2'
+            strip_section_number(title.children)
+            title['id'] = title.content.gsub(/[^a-zA-Z0-9]/, '-')
+          end
+          lib_sections.css('.example .title strong').each do |title|
+            title.content = title.content.sub(/^Example ([0-9.]+)/, 'Example: ')
+          end
+          new_root.add_child(lib_sections)
+
+          # fetchers
+          fetcher_sections = 
xpath("//*[@id='chap-pkgs-fetchers']/ancestor::section[1]/section[position()>1]")
+          fetcher_sections.css('.titlepage .title').each do |title|
+            strip_section_number(title.children)
+            prefix_with(title, 'pkgs') if title.name == 'h2'
+          end
+          new_root.add_child(fetcher_sections)
+
+          # trivial builders
+          trivial_sections = xpath("//*[@id='chap-trivial-builders']/ancestor::section[1]/section")
+          trivial_sections.css('.titlepage .title').each do |title|
+            strip_section_number(title.children)
+            prefix_with(title, 'pkgs') if title.name == 'h2'
+          end
+          new_root.add_child(trivial_sections)
+
+          # special builders
+          special_sections = xpath("//*[@id='chap-special']/ancestor::section[1]/section")
+          special_sections.css('.titlepage .title').each do |title|
+            strip_section_number(title.children)
+            if title.name == 'h2'
+              title.children[0].wrap('<code>')
+              prefix_with(title, 'pkgs')
+            end
+          end
+          new_root.add_child(special_sections)
+
+          # image builders
+          image_sections = xpath("//*[@id='chap-images']/ancestor::section[1]/section")
+          image_sections.css('.titlepage .title').each do |title|
+            strip_section_number(title.children)
+            title.children[0].wrap('<code>')
+          end
+          image_sections.each do |section|
+            prefix = section.at_xpath('*[@class="titlepage"]//*[@class="title"]').content
+            next unless ["pkgs.dockerTools", "pkgs.ociTools"].include?(prefix)
+            section.xpath('section/*[@class="titlepage"]//*[@class="title"]').each do |title|
+              prefix_with(title, prefix)
+              title['data-add-to-index'] = ''
+            end
+          end
+          new_root.add_child(image_sections)
+
+          new_root
+        elsif subpath == 'nix/stable/expressions/builtins.html'
+          @doc = doc.at_css('main dl')
+
+          # strip out the first entry, `derivation`, the actual documentation
+          # exists in a separate page
+          derivation_dt = doc.children.at_css('dt')
+          if derivation_dt.content.starts_with?('derivation')
+            derivation_dt.remove
+            doc.children.at_css('dd').remove
+          else
+            raise RuntimeError.new('First entry is not derivation, update the scraper')
+          end
+
+          doc.css('dt').each do |title|
+            title.name = 'h2'
+            unwrap(title.at_css('a'))
+            title.children[0].children.before('builtins.')
+          end
+          doc.css('dd').each do |description|
+            description.name = 'div'
+          end
+
+          doc
+        else
+          doc
+        end
+      end
+
+      def strip_section_number(title_children)
+        while title_children.first.content == ''
+          title_children.shift.remove
+        end
+        first_text = title_children.first
+        return unless first_text.text?
+        first_text.content = first_text.content.sub(/[0-9.]+ ?/, '')
+        first_text.remove if first_text.blank?
+      end
+
+      def prefix_with(title_node, text)
+        title_node.css('code').each do |code|
+          code.content = "#{text}.#{code.content}" unless code.content.starts_with?("#{text}.")
+        end
+      end
+
+      def unwrap(node)
+        node.replace(node.inner_html)
+      end
+    end
+  end
+end
diff --git a/lib/docs/filters/nix/entries.rb b/lib/docs/filters/nix/entries.rb
new file mode 100644
index 00000000..f4ff093a
--- /dev/null
+++ b/lib/docs/filters/nix/entries.rb
@@ -0,0 +1,22 @@
+module Docs
+  class Nix
+    class EntriesFilter < Docs::EntriesFilter
+      def include_default_entry?
+        false
+      end
+
+      def additional_entries
+        css('h2, h3[data-add-to-index]').flat_map do |node|
+          node.css('code').map do |code|
+            title = code.content
+            index = title.rindex('.')
+            type = title[0...index]
+            name = title.match(/^[^\s]+/)[0]
+
+            [name, node['id'], type]
+          end
+        end
+      end
+    end
+  end
+end
diff --git a/lib/docs/scrapers/nix.rb b/lib/docs/scrapers/nix.rb
new file mode 100644
index 00000000..ae549818
--- /dev/null
+++ b/lib/docs/scrapers/nix.rb
@@ -0,0 +1,31 @@
+module Docs
+  class Nix < UrlScraper
+    self.type = 'simple'
+    self.release = '2.8.0'
+    self.base_url = 'https://nixos.org/manual/'
+    self.root_path = 'nix/stable/expressions/builtins.html'
+    self.initial_paths = %w(
+      nix/stable/expressions/builtins.html
+      nixpkgs/stable/index.html)
+    self.links = {
+      home: 'https://nixos.org/',
+      code: 'https://github.com/NixOS/nix'
+    }
+
+    html_filters.push 'nix/clean_html', 'nix/entries'
+
+    options[:skip_links] = true
+
+    options[:attribution] = <<-HTML
+      &copy; 2022 NixOS Contributors<br>
+      Licensed under the LGPL License.
+    HTML
+
+    def get_latest_version(opts)
+      doc = fetch_doc('https://nixos.org/manual/nix/stable/', opts)
+      json = JSON.parse(doc.at_css('body')['data-nix-channels'])
+      channel = json.find { |c| c['channel'] == 'stable' }
+      channel['version']
+    end
+  end
+end
diff --git a/public/icons/docs/nix/16.png b/public/icons/docs/nix/16.png
new file mode 100644
index 00000000..13ad550b
Binary files /dev/null and b/public/icons/docs/nix/16.png differ
diff --git a/public/icons/docs/nix/16 2x png b/public/icons/docs/nix/16 2x png
new file mode 100644
index 00000000..5c1d1368
Binary files /dev/null and b/public/icons/docs/nix/16 2x png differ
diff --git a/public/icons/docs/nix/SOURCE b/public/icons/docs/nix/SOURCE
new file mode 100644
index 00000000..c2f6c583
--- /dev/null
+++ b/public/icons/docs/nix/SOURCE
@@ -0,0 +1 @@
+https://github.com/NixOS/nixos-homepage/blob/master/logo/nixos-logo-only-hires.png