+ while div = at_css('h1 + div')
+ div.before(div.children)
+ div.remove
+ end
+
+ # Remove dead links (e.g. ngRepeat)
+ css('a.type-hint').each do |node|
+ node.name = 'code'
+ node.remove_attribute 'href'
+ end
+
+ # Remove some
elements
+ css('h1 > code', 'pre > code', 'h6 > code').each do |node|
+ node.before(node.content).remove
+ end
+
+ # Fix code indentation
+ css('code', 'pre').each do |node|
+ node.inner_html = node.inner_html.strip_heredoc.strip
+ end
+
+ # Make elements
+ css('.in-javascript', '.in-html-template-binding').each do |node|
+ node.name = 'pre'
+ node.content = node.content
+ end
+
+ css('ul.methods', 'ul.properties', 'ul.events').add_class('defs')
+
+ # Remove ng-* attributes
+ css('*').each do |node|
+ node.attributes.each_key do |attribute|
+ node.remove_attribute(attribute) if attribute.start_with? 'ng-'
+ end
+ end
+ end
+ end
+ end
+end
diff --git a/lib/docs/filters/angular/clean_urls.rb b/lib/docs/filters/angular/clean_urls.rb
new file mode 100644
index 00000000..13fc3ad9
--- /dev/null
+++ b/lib/docs/filters/angular/clean_urls.rb
@@ -0,0 +1,11 @@
+module Docs
+ class Angular
+ class CleanUrlsFilter < Filter
+ def call
+ html.gsub! 'angularjs.org/partials/api/', 'angularjs.org/api/'
+ html.gsub! %r{angularjs.org/api/(.+?)\.html}, 'angularjs.org/api/\1'
+ html
+ end
+ end
+ end
+end
diff --git a/lib/docs/filters/angular/entries.rb b/lib/docs/filters/angular/entries.rb
new file mode 100644
index 00000000..f8aec38c
--- /dev/null
+++ b/lib/docs/filters/angular/entries.rb
@@ -0,0 +1,40 @@
+module Docs
+ class Angular
+ class EntriesFilter < Docs::EntriesFilter
+ def get_name
+ name = slug.split(':').last
+ name.sub! %r{\Ang\.}, ''
+ name << " (#{subtype})" if subtype == 'directive' || subtype == 'filter'
+ name
+ end
+
+ def get_type
+ type = slug.split('.').first
+ type << " #{subtype}s" if type == 'ng' && subtype
+ type
+ end
+
+ def subtype
+ return @subtype if defined? @subtype
+ node = at_css 'h1'
+ data = node.content.match %r{\((.+) in module} if node
+ @subtype = data && data[1]
+ end
+
+ def additional_entries
+ entries = []
+
+ css('ul.defs').each do |list|
+ list.css('> li > h3:first-child').each do |node|
+ name = node.content.strip
+ name.sub! %r{\(.+\)}, '()'
+ name.prepend "#{self.name.split.first}."
+ entries << [name, node['id']]
+ end
+ end
+
+ entries
+ end
+ end
+ end
+end
diff --git a/lib/docs/scrapers/angular.rb b/lib/docs/scrapers/angular.rb
index f20a588a..c92e4cc7 100644
--- a/lib/docs/scrapers/angular.rb
+++ b/lib/docs/scrapers/angular.rb
@@ -1,19 +1,60 @@
module Docs
class Angular < UrlScraper
- # This scraper is currently broken; the problem being that Angular's
- # documentation isn't available as static pages. I will try to restore it
- # once Angular 1.2.0 is released.
- #
- # In the past it used static-ng-doc by Sal Lara (github.com/natchiketa/static-ng-doc)
- # to scrape the doc's HTML partials (e.g. docs.angularjs.org/partials/api/ng.html).
- #
- # If you want to help this is what I need: a static page with links to each
- # HTML partial. Or better yet, a static version of Angular's documentation.
-
self.name = 'Angular.js'
self.slug = 'angular'
self.type = 'angular'
- self.version = '1.0.7'
- self.base_url = ''
+ self.version = '1.2.10'
+ self.base_url = 'http://docs.angularjs.org/partials/api/'
+
+ html_filters.insert_before 'normalize_paths', 'angular/clean_html'
+ html_filters.push 'angular/entries', 'title'
+ text_filters.push 'angular/clean_urls'
+
+ options[:title] = false
+ options[:root_title] = 'Angular.js'
+
+ options[:fix_urls] = ->(url) do
+ url.sub! '/partials/api/api/', '/partials/api/'
+ url.sub! '/partials/api/guide/', '/guide/'
+ url.sub! %r{/partials/api/(.+?)(?
+ Licensed under the Creative Commons Attribution License 3.0.
+ HTML
+
+ private
+
+ def request_one(url)
+ stub_root_page if url == root_url.to_s
+ super
+ end
+
+ def request_all(urls, &block)
+ stub_root_page
+ super
+ end
+
+ def stub_root_page
+ response = Typhoeus::Response.new(
+ effective_url: root_url.to_s,
+ code: 200,
+ headers: { 'Content-Type' => 'text/html' },
+ body: get_root_page_body)
+
+ Typhoeus.stub(root_url.to_s).and_return(response)
+ end
+
+ def get_root_page_body
+ require 'capybara'
+ Capybara.current_driver = :selenium
+ Capybara.visit('http://docs.angularjs.org/api/')
+ Capybara.find('.side-navigation')['innerHTML']
+ end
end
end