This repository contains the source code for the Crosswalk website. The live website is generated from this code.

Any bugs for the website should be logged on the Crosswalk Jira, under the Website component.

Pull requests for the website should be submitted via github.

Note that this document gives an overview of the site and explains how to setup and release for staging and live servers. If you need to edit the content of the site, the HACKING file explains how to set the site up locally.

The Crosswalk website consists of the following functional areas:

1. Main landing page. Mostly static content. Some Javascript is used to compensate for style areas that were difficult to code in a pure CSS model. Javascript is also used for dynamically changing the version strings in the top overview section.

2. Documentation and Contribute. Statically generated from content checked into the crosswalk-website.git project. These sections are hand-picked selections from the wiki which are particularly useful or frequently viewed.

   The menu content on the left is generated via xwalk.js when the page is loaded, based on the content of menus.js. The content in the contribute/ and documentation/ directories are cached .html files generated from the markdown sources. This occurs during the site.sh mklive script execution described later.

3. Wiki. This content is a dynamic proxy to the Wiki content hosted in the crosswalk-website.wiki.git on GitHub. Whenever a commit is made to the wiki, the Gollum Event Webhook on GitHub invokes the regen.php page on the live site; this in turn recreates the content viewed in the Pages and History pages of the Wiki.

   Fetches of actual Wiki content are proxied on the server to GitHub via php (XHR on the client is blocked due to Access-Control-Allow-Origin; currently PHP doesn't cache the fetched content - it should). The wiki content is served to the client in the format output by the GitHub wiki system (Gollum). The client-side javascript (see xwalk.js content_response and generate_wiki_page) creates pages using content from the wiki: it pulls out the wiki DOM element, performs some URL rewrites, and then injects the resulting content into the Crosswalk website page requested by the browser.

The content on crosswalk-project.org relates to the current stable release of Crosswalk.

Documentation for previous versions stays on the wiki.

The content under the #wiki/ path on crosswalk-project.org is maintained in the crosswalk-website wiki. This can either be edited via github or by cloning the wiki github repo (the latter is required if you want to add images).

See https://crosswalk-project.org/#wiki/Editing-the-Wiki for instructions.

Other content on crosswalk-project.org is managed through this project (crosswalk-website). The process for editing this content is as follows:

• Fork the crosswalk-website repo.

• Make your edits (see HACKING.md).

• Make a pull request via github asking for your changes to be merged into the master branch of the crosswalk-website repo.

The general work flow for releasing a new version of the Crosswalk website is as follows:

1. Develop on a local machine (see HACKING.md)
2. Create a 'live snapshot' via the command ./site.sh mklive
3. Push latest 'live snapshot' to the staging server via the script: ./site.sh push
4. Test the results on the staging server at http://stg.crosswalk-project.org/
5. Push the staging version to the live server via the command ./site.sh push live

To push to the staging or live servers, you will need to be given access by the Crosswalk infrastructure team.

NOTE: You can determine the version of the website that is active by fetching the file REVISION. The part before the colon is the branch name, the part after the colon is the commit-id. This file is queried by several of the functions defined in scripts/common.inc.

There are several pieces of content that are generated during the site development. These include:

• xwalk.css <= generated from xwalk.scss
• markdown.css <= generated from markdown.scss
• menus.js <= generated from menus.js.php
• wiki/pages.md.html <= generated from regen.php via gfm.php
• wiki/history.md.html <= generated from regen.php via gfm.php
• contribute/*.html <= generated from the markdown files in contribute/*.md via gfm.php
• documentation/*.html <= generated from the markdown files in contribute/*.md via gfm.php

On your local development server, all of the above are regenerated dynamically when any source file changes (via .htaccess and gfm.php).

On the staging and live servers, no content is generated on the fly: the mklive script generates a one-off snapshot from the source files (*.md, *.scss etc.). This snapshot is then pushed to the server as static HTML/CSS/JS. See ./site.sh --help mklive for details.

This is a list of one time setup instructions for hosting crosswalk-website. These steps have already been carried out on the live and production servers, and do not have to be repeated to make a new release of the site. However, they are provided for reference in case the setup needs to be done again on a different server, or if the existing setup needs to be modified or debugged.

Running the live version requires the rewrite module in Apache2. This is used in the wiki subsystem to map requested URLs through a PHP script which will then perform smart matching of leaf names.

Enable the rewrite module via:

sudo a2enmod rewrite
sudo service apache2 restart

The rest of the content is served from static files that are generated as part of the development cycle, described in the Workflow section (above).

In addition, you will need to enable the cURL extension for PHP. This is used by the gfm.php script to perform HTTP requests for pages from the Crosswalk wiki. Enable it by editing the php.ini file for the PHP installation and adding this line (there should be several other extension lines in php.ini already, so add it after those):

extension=php_curl.so

To host the Crosswalk website, the following needs to be done on the server:

# Initialize the site content into the docroot directory
sudo git clone https://github.com/crosswalk-project/crosswalk-website.git docroot
cd docroot

# make a clone of the wiki content
sudo git clone --bare https://github.com/crosswalk-project/crosswalk-website.wiki.git wiki.git

# everything should be owned by drush:users...
sudo chown -R drush:users * .git*

# ...except htaccess, and wiki and cache directories
# (which Apache needs to write to)
sudo chown -R wwwrun:www .htaccess wiki.git wiki cache

# create a configuration file with github credentials (see below)
sudo cp site-config.php.template site-config.php
sudo vim site-config.php

# Switch to the latest live branch
. scripts/common.inc
branch=$(get_remote_live_info)
echo ${branch} > REVISION
git checkout -f ${branch}

At this point, the site is now initialized and set to the latest live branch. The scripts/ directory is not part of the live version of the site - once you checkout ${branch}, you won't be able to run any of the scripts on the live site.

The chown command is necessary as regen.php writes to wiki.git and wiki, which also executes as the Apache user. On Ubuntu this user is wwwrun, but if you're using a different operating system, or a different Apache distribution, the user may be someone else; for example, XAMPP uses nobody as the user.

The site-config.php file should be created on the server in the root directory, using the site-config.php.template file as the template.

site-config.php requires credentials for accessing the github API (see the github.php script for how the github API proxy is implemented). The proxy is used by the Downloads and Channels Viewer pages, to populate the Crosswalk version numbers, download URLs, commit SHAs etc.

The credentials required are a Client ID and Client Secret, which can be created by registering a new application against a github account.

The crosswalk-project.org website needs to be set up in tandem with a github hook to ensure that any changes to the wiki are immediately reflected on the website. The hook should invoke the regen.php script for any gollum events triggered by changes to the crosswalk project's wiki.

However, this hook has to be manually added to the github project for crosswalk-website: the built-in web hooks available via github Settings cannot be set up to respond to gollum wiki events.

The JSON data to required for configuring the hook looks like this:

{
  "name": "web",
  "active": true,
  "events": [
    "gollum"
  ],
  "config": {
    "url": "http://crosswalk-project.org/regen.php",
    "content_type": "json"
  }
}

Paste this into a file and post it to the github API using an HTTP client tool; for example, if the JSON file was called config.json and you were posting it as the foo user via curl, you would do:

curl -k -u foo -d @config.json \
  https://api.github.com/repos/crosswalk-project/crosswalk-website/hooks

You should get a response from the API which indicates whether the request was successful. From now on, any time the wiki for crosswalk-website changes, the script at http://crosswalk-project.org/regen.php will be invoked with details of the pages which changed.