Overhaul of Page Generation

2014-12-02

 

When migrating from jekyl to gulp.js I was basically following jekyls lead, purely to ease the migration pains.

Some point later Daniel Naab adapted my aproach. Although he build ontop of my ideas, he played the ball out of the park. What really stuck me as ingenious is the integration of prose.io. The combination of travis-ci and prose.io, means that novice users can author the website without knowing what git and a text editor is. The result is something close to a high end CMS with static generated HTML pages.

So I took some inspiration back from Daniel Naab. So here is a rundown from my enhanced glup.js powered website, this thime with wercker and prose.io support.

The basic layout of the website is as follows:

+ assets
  favicon.ico
  + css
  + js
  + fonts
+ content
  + media
  + pages
  + posts
+ templates
wercker.yml
_prose.yml
package.json
site.json
gulp.js

The assets folder contains the bits that make up the website's design. These are copied verbatim to the output. The content folder contains the actual raw content, this is the folder where day to day editing is done. The content is seperated into three types, the media folder contains unstructured fils, such as pictures, the page folder contains pages, that is timeless content and the posts folder conttains journal entries. The template folder contains the templates used to render the pages and posts. Finally the root folder contains all the configuration for the different services, such as the wercker.yml.

The gulpfile still works mostly like the original as described in my jekyl to gulp.js post, just the paths are cleaned up. You can see the entiere file in my github repository.

The interseting bit that changes is that now the journal backlog paginates:

function posts(basename, count) {
  var stream = through.obj(function(file, enc, cb) {
        this.push(file);
        cb();
    });

  if (site.posts)
  {
    var c     = 0;
    var page  = 0;
    var posts = [];
    site.posts.forEach(function (post) {
      posts.push(post);
      c++;
      if (c == count) {        
        var file = new gutil.File({
          path: basename + (page == 0 ? '' : page) + '.html',
          contents: new Buffer('')
        });
        console.log('page=' + page + ' c=' + c + ' posts.length=' + site.posts.length);
        file.page = {
          posts: posts, 
          prevPage: page != 0 ? basename + ((page-1) == 0 ? '' : page-1) + '.html' : null,
          nextPage: (page+1) * count < site.posts.length ? basename + (page+1) + '.html' : null,
          };
        stream.write(file);

        c = 0;
        posts = [];
        page++;
      }
    });   

    if (posts.length != 0) {
      var file = new gutil.File({
        path: basename + (page == 0 ? '' : page) + '.html',
        contents: new Buffer('')
      });
      file.page = {
        posts: posts, 
        prevPage: page != 0 ? basename + ((page-1) == 0 ? '' : page) + '.html' : null,
        nextPage: null,
        };
      stream.write(file);
    }
  }

  stream.end();
  stream.emit("end");

  return stream;
}

gulp.task('archive', ['posts'], function () {
    return posts('journal', 10)
        .pipe(applyTemplate('templates/journal.html'))
        .pipe(gulp.dest('build/'));
});

Also the special pages index.html and atom.xml are generated through a template:

function dummy(file) {
  var stream = through.obj(function(file, enc, cb) {
        this.push(file);
        cb();
    });

  if (site)
  {
    var file = new gutil.File({
      path: file,
      contents: new Buffer('')
    });
    file.page = {}        
    stream.write(file);        
  }

  stream.end();
  stream.emit("end");

  return stream;
}

gulp.task('index', ['posts'], function () {
    return dummy('index.html')
        .pipe(applyTemplate('templates/index.html'))
        .pipe(gulp.dest('build/'));
});

gulp.task('rss', ['posts'], function () {
  return dummy('atom.xml')
    .pipe(applyTemplate('templates/atom.xml'))
    .pipe(gulp.dest('build/'));
});

The site build and deployment is handled through wercker. The configuration for this process is:

box: wercker/[email protected]
build:
  steps:
    - wercker/[email protected]
    - yeehaa/[email protected]
deploy:
  steps:
    - s3sync:
        key_id: $KEY
        key_secret: $SECRET
        bucket_url: $URL
        source_dir: build

This does basically what you think it does. It runns npm install followed by a gulp call. The built site is then auto deployed to a AWS S3 bucket using s3sync.

The prose.io configuration is also quite simple:

prose:
  rooturl: 'content'
  siteurl: 'http://www.rioki.org'
  media: 'content/media'
  metadata:
    content/posts:      
      - name: "title"
        field:
          element: "text"
          label: "Title
          help: "The blog post title"
          placeholder: "Enter title"
      - name: "tags"
        field:
          element: "multiselect"
          label: "Tags"
          placeholder: "Choose Tags"
          alterable: True
          options:
            - name: "Meta"
              value: "meta"
            - name: "C++"
              value: "cpp"
            - name: "openGL"
              value: "opengl"
            - name: "Web"
              value: "web"

The only thing I need to do is populate the tags with more values to choose from. But the alterable flag means I can add some in a pinch.

BTW this post was written with prose.io, let's hope it works well.