Skip to main content

A template language grammar inspired by the Python code aesthetic

Project description

A python-inspired templating language.


import app.url
import request
from plywood.plugin import compress
# looks like python so far?

doctype(5)  # or doctype('strict') doctype('xhtml'), etc.
html:  # this'll start looking a lot like jade, but with quotes and colons
  # even though 'html' is a function call, the parentheses are optional.
      if self.title:  # context variables are available on 'self'
        # docstrings are stripped of preceding whitespace and the first and
        # last newline is removed.
        {{self.title}} |
        """  # string interpolation uses plywood in 'inline' mode.  Each line
             # will be joined with a space.
      'Welcome'  # string literals require quotes
      # passing values to tag attributes are escaped (html-entitized) automatically
      # if you want to escape using xml, pass {'format': 'xml'} in your options.
      link(rel='stylesheet', type='text/css', href=url.static('css/reset.css'))
      link(rel='stylesheet', type='text/css', href=url.static('css/welcome.css'))
    script(src="//", type="text/javascript")
      script(src=url.static("js/underscore.js"), type="text/javascript")
      script(src=url.static("js/backbone.js"), type="text/javascript")
    ieif 'lt IE 9':
      script(src="//", type="text/javascript")
      link(rel='stylesheet', type='text/css', href=url.static('css/ie.css'))
    # blocks? block inheritance?  of course!
    div(class="wrapper", id="main-header")
    div.wrapper(id="main-header"):  # class shorthand - only HTML plugins have this trick

    # I struggled long and hard on what to do about the #id shorthands.
    # in the end, I couldn't in good conscience call this a "python
    # inspired" language if '#' was not the comment delimiter.  So the id
    # shorthand is "@" instead:
    div.wrapper@main-header:  # and yes, that '-' is part of the
         # 'main-header' token.  to support xml/html tag and attribute name,
         # I had to allow ':' and '-' in variable name.

      # for xml usage, the token parsing will accept some gnarly-looking elements in
      # argument lists, and this uses the html-plugin constructor, so that
      # you don't have to create a bunch of plugins for your XML documents.
      # (you still need commas between)
      <book xmlns='',
          <isbn:number>: 1568491379
          # inlining is easy
          p(class="logo"): 'logo'
          # more complicated inlining
          p: a(href=url.reverse("login")): 'Login'
            if self.user:
              'Welcome, '{}'
        if not self.user:
            a(href=url.reverse("login")): 'Log In'
            a(href=url.reverse("logout")): 'Log Out'


          if messages:
              for message in self.messages:
                li(class=message.tags):  message
          # code literals, so that savvy editors can color the source code

            var fade_out = _(function() {

            setTimeout(fade_out, 5000);
            $("ul.messages").bind("click", fade_out);

          '&copy;{now(%Y)} colinta'


$ pip install plywood
$ ply < in.ply > out.html


Each line starts with a statement, which can either be a function (div, block) a literal (', '''), or a control statement (if, else, for).

Functions get called with the arguments and a “block”:

# arguments are ((), {}), block is Block()
# arguments are ((), {'class': 'divvy'}), block is Block()
# arguments are (('autofocus'), {'id': 'bio'}), block is Block(Literal('This is my bio'),)
textarea("autofocus", id="bio"): 'This is my bio'

Even if there is no “block”, you’ll get at the least at empty block object that you can call block.get_value() on. It will be “falsey”, though, so you can check for the existence of a block. The minimum “truthy” block is an empty string. That means div: '' will have a “truthy” block, but div will have a “falsey” block.

You can extend the crap out of plywood, because div, if, block, the whole lot, are all written as plywood extensions. Without the builtin extensions, the language couldn’t actually do anything, because it is at its core just a language grammar.


The main reason: I envisioned an HTML templating language that had python-like syntax, and the options that are out there now (Haml, Coffekup, Jade) don’t hit the mark.

Plain-Jane HTML? Sure, if you want. That is, I think, the best alternative to plywood! For that, use Jinja2.

The template languages that take an HTML-agnostic view (jinja2, django) is HTML made nastier by inserting additional markup. I looked at Jade and Haml as “yeah, you’re getting there”, but they didn’t nail it. Plus, have you tried writing extensions for those systems? Ooof. Nasty stuff. Writing a plugin for plywood is much easier, and since you can take some part in the parsing and runtime process, you can write some pretty hefty plugins!

I’m unapologettically a DIY-er. I think that sometimes wheels just need re-inventing!



Colin Thomas-Arnold


2012 Colin Thomas-Arnold <>

Copyright (c) 2012, Colin Thomas-Arnold All rights reserved.

See LICENSE for more details (it’s a simplified BSD license).

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

plywood-1.2.2.tar.gz (33.9 kB view hashes)

Uploaded Source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page