Add Qt scraper

assets/javascripts/templates/pages/about_tmpl.coffee

@@ -575,6 +575,11 @@ credits = [
     '2009-2017 Kristopher Michael Kowal',
     'MIT',
     'https://raw.githubusercontent.com/kriskowal/q/master/LICENSE'
+  ], [
+    'Qt',
+    '2012-2018 The Qt Company Ltd',
+    'GFDL',
+    'https://doc.qt.io/qt-5/licensing.html'
   ], [
     'Ramda',
     '2013-2016 Scott Sauyet and Michael Hurley',

assets/stylesheets/application-dark.css.scss

@@ -83,6 +83,7 @@
         'pages/postgres',
         'pages/pug',
         'pages/python',
+        'pages/qt',
         'pages/ramda',
         'pages/rdoc',
         'pages/react_native',

assets/stylesheets/application.css.scss

@@ -83,6 +83,7 @@
         'pages/postgres',
         'pages/pug',
         'pages/python',
+        'pages/qt',
         'pages/ramda',
         'pages/rdoc',
         'pages/react_native',

assets/stylesheets/pages/_qt.scss

@@ -0,0 +1,63 @@
+._qt {
+  @extend %simple;
+
+  h1 {
+    margin-top: 0;
+    @extend %lined-heading;
+  }
+
+  // Home page
+  .landingicons {
+    .icons1of3 {
+      display: flex;
+      flex-wrap: wrap;
+      align-items: center;
+      margin: 1em 0;
+
+      @if $style == 'dark' {
+        &:nth-child(3) img {
+          filter: invert(1) hue-rotate(180deg) brightness(1.5)
+        }
+      }
+    }
+
+    h2 {
+      flex: 1;
+      margin: 0;
+      margin-left: 10px;
+    }
+
+    h2 + p {
+      width: 100%;
+      padding-left: 70px;
+    }
+
+    .centerAlign,
+    .centerAlign img { // Icon
+      width: 60px;
+      height: 60px;
+      margin: 0;
+    }
+  }
+
+  // Syntax highlighting
+  .pre {
+    .operator { @extend .token, .operator; }
+    .number { @extend .token, .number; }
+    .keyword { @extend .token, .keyword; }
+    .type { @extend .token, .function; }
+    .type a { color: inherit; }
+    .pun, .comment { @extend .token, .punctuation; }
+    .kwd, .preprocessor { @extend .token, .keyword; }
+    .str, .string { @extend .token, .string; }
+  }
+
+  // Function headers
+  h3.fn code {
+    @extend %label;
+    float: right;
+    font-size: .8em;
+    padding: 0;
+    margin-right: -.5em;
+  }
+}

lib/docs/filters/qt/clean_html.rb

@@ -0,0 +1,21 @@
+module Docs
+  class Qt
+    class CleanHtmlFilter < Filter
+      def call
+        # Remove unneeded elements
+        css('.copy-notice, .navigationbar, .headerNavi, .footerNavi, .sidebar, .toc, #ec_toggle').remove
+
+        # QML property/method header
+        css('.qmlproto').each do |node|
+          id = node.at_css('tr')['id']
+          node.inner_html = node.at_css('td').inner_html
+          node.name = 'h3'
+          node.add_class '_qml_header'
+          node['id'] = id
+        end
+
+        doc
+      end
+    end
+  end
+end

lib/docs/filters/qt/entries.rb

@@ -0,0 +1,139 @@
+module Docs
+  class Qt
+    class EntriesFilter < Docs::EntriesFilter
+      def get_name
+        header = at_css('h1.title + .small-subtitle a') || at_css('h1.title') || at_css('.context h2')
+        name = header.content
+        name.sub! %r{ Class$}, ' (class)'
+        name.sub! %r{ QML Type$}, ' (QML type)'
+        name.sub! %r{ QML Basic Type$}, ' (QML basic type)'
+
+        # Add '(class)' to the class pages where the subtitle name is used (e.g. qset-const-iterator.html)
+        if at_css('h1.title').content.strip.end_with?(' Class') and !name.include?('(class)')
+          name = "#{name} (class) "
+        end
+
+        name
+      end
+
+      def get_type
+        breadcrumb = css('#main_title_bar + ul li')
+        category = if breadcrumb.length < 3
+          then 'Qt'
+          else breadcrumb.at(1).content
+        end
+
+        if category == 'Qt'
+          return 'Qt Platforms' if name.include? ' for ' or name == 'Qt Platform Abstraction'
+          return 'Qt Quick' if name == 'Qt Quick Test' or name == 'Qt Quick Test Reference Documentation'
+
+          alwaysInQt = ["Qt Configure Options", "Qt Image Formats"]
+          category = name if name.start_with?('Qt ') && !alwaysInQt.include?(name)
+        end
+
+        qtPlatformsTypes = ['Qt Platform Headers', 'Qt Android Extras', 'Qt Mac Extras', 'Qt Windows Extras', 'Qt X11 Extras']
+        return 'Qt Platforms' if qtPlatformsTypes.include? category
+
+        category.sub! ' Manual', ''
+        category
+      end
+
+      def include_default_entry?
+        name != 'All Classes' and name != 'All QML Types'
+      end
+
+      def additional_entries
+        entries = []
+        titles = []
+
+        className = at_css('h1.title').content.strip.sub ' Class', ''
+        displayedClassName = className
+        alternativeClassName = at_css('h1.title + .small-subtitle a')
+        displayedClassName = alternativeClassName.content if alternativeClassName
+
+        # Functions signatures
+        css('h3.fn').each do |node|
+          header = node.clone
+
+          # Skip typenames
+          next if header.content.strip.start_with? 'typename '
+
+          # Remove leading <a name="">
+          header.children.css('a[name]').remove
+
+          # Remove leading <code> tag (virtual/static/… attributes)
+          code = header.children.first
+          code.remove if code.name == 'code'
+
+          # Remove leading ‘const’
+          header.children.first.remove if header.content.strip.start_with? 'const '
+
+          # Remove return type
+          returnType = header.children.first
+          returnType.remove if returnType['class'] == 'type'
+
+          title = header.content.strip
+
+          # Remove leading '&'/'*'
+          title[0] = '' if title[0] == '&' || title[0] == '*'
+
+          # Ignore operator overloads
+          next if title.start_with? 'operator'
+
+          # Remove function parameters
+          title.sub! %r{\(.*\)}, '()'
+
+          # Remove template generics
+          title.sub! %r{^<.*> }, ''
+
+          # Remove ‘const’ at the end
+          title.sub! %r{ const$}, ''
+
+          # Enum/typedef formatting
+          title.sub! %r{(enum|typedef) (.*)}, '\2 (\1)'
+
+          # Remove property type
+          title = "#{displayedClassName}::#{title}" if title.sub! %r{ : .*$}, ''
+
+          # Replace the class name by the alternative class name if available
+          title.sub! className, displayedClassName if alternativeClassName
+
+          unless titles.include? title # Remove duplicates (function overloading)
+            entries << [title, header['id']]
+            titles.push title
+          end
+        end
+
+        # QML properties/functions
+        qmlTypeName = at_css('h1.title').content.sub ' QML Type', ''
+        css('.qmlproto').each do |node|
+          title = node.content.strip
+          id = node.at_css('tr')['id']
+
+          # Remove options
+          title.sub! %r{^\[.*\] }, ''
+
+          # Remove function parameters
+          title.sub! %r{\(.*\)}, '()'
+
+          # Remove property type
+          title.sub! %r{ : .*$}, ''
+
+          # Remove return type
+          title.sub! %r{.* }, ''
+
+          # Remove return type
+          title.sub! %r{.* }, ''
+
+          title = "#{qmlTypeName}.#{title.strip}"
+          unless titles.include? title # Remove duplicates (function overloading)
+            entries << [title, id]
+            titles.push title
+          end
+        end
+
+        entries
+      end
+    end
+  end
+end

lib/docs/scrapers/qt.rb

@@ -0,0 +1,121 @@
+module Docs
+  class Qt < UrlScraper
+    self.name = 'Qt'
+    self.type = 'qt'
+    self.initial_paths = %w(classes.html qmltypes.html)
+    self.root_path = 'index.html'
+    self.links = {
+      home: 'https://www.qt.io',
+      code: 'https://code.qt.io/cgit/'
+    }
+
+    html_filters.push 'qt/entries', 'qt/clean_html'
+
+    options[:container] = '.main'
+
+    options[:skip_patterns] = [
+      # License, copyright attributions
+      /3rdparty/,
+      /attribution/,
+      /license/,
+      /licensing/,
+
+      # Examples, guides, tutorials
+      /example/,
+      /guide$/,
+      /tutorial/,
+      /porting/,
+      /usecase/,
+      /topic/,
+      /^modelview/,
+      /deploy(ing|ment)/,
+      /building/,
+
+      # Old versions, changelog
+      /obsolete/,
+      /compatibility/,
+      /^whatsnew/,
+      /^newclasses/,
+
+      # Deprecated modules
+      /(qtopengl|qgl)/,
+      /qt?script/,
+
+      # Indexes
+      /members/,
+      /module/,
+      /overview/,
+      /^qopenglfunctions/,
+
+      # Tooling
+      /^(qt)?(linguist|assistant|qdbusviewer)/,
+    ]
+
+    options[:skip] = [
+      "qt5-intro.html",
+      "compatmap.html",
+
+      # Indexes
+      "classes.html",
+      "qtmodules.html",
+      "modules-qml.html",
+      "modules-cpp.html",
+      "functions.html",
+      "namespaces.html",
+      "qmltypes.html",
+      "qt3d-qml.html",
+      "qmlbasictypes.html",
+      "guibooks.html",
+      "annotated.html",
+      "overviews-main.html",
+      "reference-overview.html",
+
+      # Tutorials
+      "qtvirtualkeyboard-build.html",
+
+      # Copyright
+      "trademarks.html",
+      "lgpl.html",
+      "bughowto.html",
+
+      # Changelogs
+      "changes.html",
+      "qtlocation-changes.html",
+      "sourcebreaks.html",
+
+      # Best practice guides
+      "accessible.html",
+      "accessible-qtquick.html",
+      "sharedlibrary.html",
+      "exceptionsafety.html",
+      "scalability.html",
+      "session.html",
+      "appicon.html",
+      "accelerators.html",
+
+      # Other
+      "ecmascript.html",
+      "qtremoteobjects-interaction.html",
+    ]
+
+    options[:attribution] = <<-HTML
+      &copy; The Qt Company Ltd<br>
+      Licensed under the GNU Free Documentation License, Version 1.3.
+    HTML
+
+    version '5.11' do
+      self.release = '5.11'
+      self.base_url = 'https://doc.qt.io/qt-5/'
+    end
+
+    version '5.9' do
+      self.release = '5.9'
+      self.base_url = 'https://doc.qt.io/qt-5.9/'
+    end
+
+    version '5.6' do
+      self.release = '5.6'
+      self.base_url = 'https://doc.qt.io/qt-5.6/'
+    end
+  end
+end

public/icons/docs/qt/SOURCE

@@ -0,0 +1 @@
+https://commons.wikimedia.org/wiki/File:Qt_logo_2016.svg