Featherweight, dependencyless, object oriented HTML generator.
Switch branches/tags
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


Objective Elements

This is a tiny gem that builds nicely formatted HTML using sane, readable Ruby. I use it for jekyll plugins, but you can use it anywhere. It's ~100 lines, tested with rspec, and has no dependencies.

It doesn't actually know any HTML, just how to format it.

This is meant to be less involved and more flexible than nokogiri's XML/HTML generator. There's no DSL to learn, and no cleverness to wrap your mind around. Its specialty is taking fragmented, disjointed information and condensing it into a string of properly formatted HTML. It's as agnostic as possible on the input, while being extremely consistent with its output.

How it works:

  • Instantiate a SingleTag or DoubleTag

  • Add attributes & content. Nest tags infinitely.

  • Render it with .to_s


Have you ever tried to build HTML with string concatenation and interpolation? It starts out simply enough, but once you account for all the what-ifs, the indentation, the closing tags, and the spaces you only need sometimes, it turns into a horrible mess.

The problem, of course, is that building long, complex, varying blocks of text with string concatenation and interpolation is fragile, unreadable, and painful. You know this, but you're not going to write an entirely new class or pull in some big new dependency just for 10 lines of HTML, so instead you hammer through it and end up with code like this:

picture_tag = "<picture>\n"\
              "#{markdown_escape * 4}"\
              "<img src=\"#{url}#{instance['source_default'][:generated_src]}\" "\
              "#{html_attr_string}>\n"\"#{markdown_escape * 2}</picture>\n"

or this:

    def build_li(this_item_data, icon_location, label)
      li = "  <li#{@attributes['li']}>"
      if this_item_data && this_item_data['url']
        li << "<a href=\"#{this_item_data['url']}\"#{@attributes['a']}>"
      li << build_image_tag(icon_location)
      li << label
      li << '</a>' if this_item_data['url']
      li << "</li>\n"

Which is why I sat down and wrote this gem. It's super simple, you probably could have written it too, but hey! Now you don't have to. Here's a demo:


p = DoubleTag.new 'p'
# <p>
# </p>

# Add attributes as a hash. keys can be strings or symbols, values can be arrays or strings:
p.add_attributes class: 'stumpy grumpy', 'id' => 'the-ugly-one'

# Add attributes as a string!
p.add_attributes 'class="slimy" data-awesomeness="11"'

# Add content. It can be anything, or an array of anythings.
p.add_content "Icky"

# <p class="stumpy grumpy slimy" id="the-ugly-one" data-awesomeness="11">
#   Icky
# </p>

# Want a oneliner?
p.oneline = true
# <p class="stumpy grumpy slimy" id="the-ugly-one" data-awesomeness="11">Icky</p>
p.oneline = false

# Build a tag all at once:
p.add_content DoubleTag.new(
  content: 'Link!',
  attributes: {href: 'awesome-possum.com'},
  oneline: true

# Add a parent tag:
div = p.add_parent DoubleTag.new 'div'

# Ruby implicitly calls .to_s on things when you try to perform string functions with them, so
# this works:
# <div>
#   <p class="stumpy mopey grumpy slimy" id="the-ugly-one" data-awesomeness="11">
#     Icky
#     <a href="awesome-possum.com">Link!</a>
#   </p>
# </div>

For complete example usage, see jekyll_icon_list, or this pull request to jekyll-picture-tag.


# Gemfile

gem 'objective_elements', '~> 0.2.0'
# Wherever you need to use it:

require 'objective_elements'


So we're on the same page, here's the terminology I'm using:

<p class="stumpy">Hello</p>
|a|       b      |  c  | d |
  • a - element
  • b - attributes
  • a+b - opening tag
  • c - content
  • d - closing tag


There are 2 classes: SingleTag is the base class, and DoubleTag inherits from it. A SingleTag is a self-closing tag, meaning it has no content and no closing tag. A DoubleTag is the other kind.

SingleTag Properties:


  • String
  • Mandatory
  • Which type of tag it is, such as 'hr' or 'img'


  • Hash
  • Optional
  • Keys are stored as symbols, values are stored as arrays of strings: {class: ['stumpy', 'slimy']}
  • add them with .add_attributes, which can accept a few different formats.

SingleTag Methods (that you care about)

SingleTag.new(element, attributes: {})

.to_s - The big one. Returns your HTML as a string, nondestructively.

.add_attributes(new) - The strongly recommended way to add new attributes. Can accept a hash (keys can be either symbols or strings, values can be either arrays or strings), or a string in the standard HTML syntax (attribute="value" attribute2="value2 value3"). Returns self.

.reset_attributes(new) - Deletes all attributes, calls add_attributes on supplied argument if given.

.delete_attributes(keys) - Accepts a single attribute, or an array of attributes (keys or strings), and deletes it.

.rewrite_attributes - Accepts anything add_attributes understands, but replaces existing attributes instead of appending to them.

.add_parent(DoubleTag) - returns supplied DoubleTag, with self added as a child.

attr_reader :attributes attr_accessor :element

DoubleTag Properties:

DoubleTag Inherits all of SingleTag's properties and methods, and adds content and a closing tag.


  • Array
  • Optional
  • Contains anything (but probably just strings and tags. Anything else will be turned into a string with .to_s, which is an alias for .inspect most of the time).
  • Each element in the array corresponds to at least one line of HTML
  • Multiline child tags will get as many lines as they need (like you'd expect).
  • Child elements are not rendered until the parent is rendered, meaning you can access and modify them after defining a parent.
  • add with .add_content, or modify the content array directly.


  • Boolean
  • optional, defaults to false.
  • When true, the entire element and its content will be rendered as a single line. Useful for anchor tags and list items.

DoubleTag Methods (that you care about)

DoubleTag.new(element, attributes: {}, oneline: false, content: []) - You can initialize it with content.

add_content(anything) - Smart enough to handle both arrays and not-arrays without getting dorked up. When given an array, its elements will be appended to the content array. When given a single item, that item will be inserted at the end of the array. (Remember each element in the content array gets at least one line!)

attr_accessor: content - You can modify the content array directly if you like. If you're just adding items, you should use .add_content

.to_a - Mostly used internally, but if you want an array of strings, each element a line with appropriate indentation applied, this is how you can get it.


Indentation is defined by the indent method on the DoubleTag class, which is two markdown-escaped spaces by default ("\ \ "). If you'd like to change it:

  1. Make a new class, inherit from DoubleTag.
  2. Override indent with whatever you want.
  3. Use your new class instead of DoubleTag.


require 'ojbective_elements'

class MyDoubleTag < DoubleTag
 def indent
   # 4 escaped spaces:
   "\ \ \ \ "

MyDoubleTag.new('p', content: 'hello').to_s
# <p>
#     hello
# </p>


  • It doesn't know a single HTML element on its own, so it does nothing to ensure your HTML is valid.

  • A parent tag can't put siblings on the same line. You can either do this (with oneline: true on the strong tag):

      Here is some

    or this (default behavior):

      Here is some

    But you can't do this without string interpolation or something:

    Here is some <strong>strong</strong> text. 

    This doesn't affect how the browser will render it, but it might bug you if you're particular about source code layout.

  • If you set 'oneline: true' on a parent DoubleTag, but not all its children DoubleTags, the output will not be pretty. I advise against it. Handling this situation is on the TODO list.

  • It doesn't wrap long lines of text, and it doesn't indent text with newlines embedded. It's on the TODO list.


For code style, I've been using rubocop with the default settings and would appreciate if you did the same.

If you add new functionality, or change existing functionality, please update the rspec tests to reflect it.


contact: robert@robert-buchberger.com


The gem is available as open source under the terms of the MIT License.