Unless you wish to use DevDocs offline or contribute to the code, I recommend using the hosted version at [devdocs.io](http://devdocs.io). It's up-to-date and requires no setup.
DevDocs is made of two separate pieces: a Ruby scraper responsible for generating the documentation and index files, and a JavaScript front-end powered by a small Sinatra app.
DevDocs requires Ruby 2.2.0 and a JavaScript runtime supported by [ExecJS](https://github.com/sstephenson/execjs#readme) (included in OS X and Windows; [Node.js](http://nodejs.org/) on Linux). Once you have these installed, run the following commands:
Finally, point your browser at [localhost:9292](http://localhost:9292) (the first request will take a few seconds to compile the assets). You're all set.
The `thor docs:download` command is used to download/update individual documentations (e.g. `thor docs:download html css`), or all at the same time (using the `--all` option). You can see the list of available documentations by running `thor docs:list`.
**Note:** there is currently no update mechanism other than `git pull origin master` to update the code and `thor docs:download` to download the latest version of the docs. To stay informed about new releases, be sure to watch this repository and/or subscribe to the [newsletter](http://eepurl.com/HnLUz).
The app's main goals are to: keep boot and load times as short as possible; improve the quality, speed, and order of search results; maximize the use of caching and other performance optimizations; maintain a clean, readable user interface; support full keyboard navigation; reduce “context switch” by using a consistent typography and design across all documentations; reduce clutter by focusing on a specific category of content (API/reference) and by indexing only the minimum useful to most developers.
**Note:** DevDocs is neither a programming guide nor a search engine. All content is pulled from third-party sources and the project doesn't intend to compete with full-text search engines. Its backbone is metadata: each piece of content must be identified by a unique, obvious and short string. Tutorials, guides and other content that don't fit this requirement are outside the scope of the project.
The web app is all JavaScript, written in [CoffeeScript](http://coffeescript.org), and powered by a small [Sinatra](http://www.sinatrarb.com)/[Sprockets](https://github.com/sstephenson/sprockets) application. It relies on files generated by the [scraper](#scraper).
Many of the code's design decisions were driven by the fact that the app uses XHR to load content directly into the main frame. This includes stripping the original documents of most of their HTML markup (e.g. scripts and stylesheets) to avoid polluting the main frame, and prefixing all CSS class names with an underscore to prevent conflicts.
Another driving factor is performance and the fact that everything happens in the browser. `applicationCache` (which comes with its own set of constraints) and `localStorage` are used to speed up the boot time, while memory consumption is kept in check by allowing the user to pick his/her own set of documentations. On the other hand, the search algorithm is currently not very sophisticated because it needs to be fast even searching through 100k entries.
This allows the code to take advantage of the latest DOM and HTML5 APIs and make developing DevDocs a lot more fun!
## Scraper
The scraper is responsible for generating the documentation and index files (metadata) used by the [app](#app). It's written in Ruby under the `Docs` module.
There are currently two kinds of scrapers: `UrlScraper` which downloads files via HTTP and `FileScraper` which reads them from the local filesystem. They both make copies of HTML documents, recursively following links that match a set of rules and applying all sorts of modifications along the way, in addition to building an index of the files and their metadata. Documents are parsed using [Nokogiri](http://nokogiri.org).
These modifications are applied via a set of filters using the [HTML::Pipeline](https://github.com/jch/html-pipeline) library. Each scraper includes filters specific to its documentation, one of which is tasked with figuring out the pages' metadata.
The end result is a set of normalized HTML partials and a JSON index file. Because the index files are loaded separately by the [app](#app) following the user's preferences, the code also creates a JSON manifest file containing information about the documentations currently available on the system (such as their name, version, update date, etc.).
This software is licensed under the terms of the Mozilla Public License v2.0. See the [COPYRIGHT](https://github.com/Thibaut/devdocs/blob/master/COPYRIGHT) and [LICENSE](https://github.com/Thibaut/devdocs/blob/master/LICENSE) files.
**Note:** I consider _DevDocs_ to be a trademark. You may not use the name to endorse or promote products derived from this software without my permission, except as may be necessary to comply with the notice/attribution requirements.
**Additionally**, I wish that any documentation file generated using this software be attributed to DevDocs. Let's be fair to all contributors by not stealing their hard work.
