A swiftype sync extension for middleman.
This extension adds a middleman command swiftype
. You can run it by executing middleman swiftype
.
This command pushes your content to the swiftype search api.
In contrast to the swiftype crawler it only pushes content you want to be searched. So no indexing of menu items, headers or footers but only usable content.
This extension also supports the swiftype sections. So you can use the tagged categories/tags as keywords for the search. I, for example, add tags to every post but I don't show them on my blog. However, those can be used for searching this way.
A lot of this extension can be configured by using lambda's. This is the example config that comes with swiftype and is based on my own configuration:
# Configuration of the swiftype extension
activate :swiftype do |swiftype|
swiftype.api_key = 'MY_SECRET_API_KEY'
swiftype.engine_slug = 'my_awesome_blog'
swiftype.pages_selector = lambda { |p| p.path.match(/\.html/) && p.metadata[:options][:layout] == nil }
swiftype.title_selector = lamda { |mm_instance, p| '...' }
swiftype.process_html = lambda { |f| f.search('.//div[@class="linenodiv"]').remove }
swiftype.generate_sections = lambda { |p| (p.metadata[:page]['tags'] ||= []) + (p.metadata[:page]['categories'] ||= []) }
swiftype.generate_info = lambda { |f| TruncateHTML.truncate_html(strip_img(f.to_s), blog.options.summary_length, '...') }
swiftype.should_index = lamda { |p, title| '...' }
end
swiftype.api_key
and swiftype.engine_slug
are required. The rest of the options are optional.
The api key and engine slug can be found in the swiftype dashboard.
The pages_selector
can be used to filter the pages that are searchable. If this option is not used all pages will be searched. So this will include any rss or atom feeds generated.
The title_selector
can be used to look up a page's title (for each page). For example, maybe you store the titles in a customized table of contents file.
process_html
can be used for transforming the html content that will be send to swiftype. In my example I'm using this to remove line numbers in code blocks: I don't want them to be searchable by swiftype.
generate_sections
can be used for search keywords you want to use but are not in the main content. I base mine on the categories & tags for a post.
generate_info
is an option that can be used for anything. I use it for storing the summary of each post.
should_index
can be used to filter out pages that you don't want to index. Return true to index in swiftype, false to skip.
This project is tested with RSpec. To run the tests, simply type:
$ bundle exec rspec