Enaml Web

A web component toolkit for enaml that let's you build websites in python declaratively.

Short intro

To use enaml web, you simply replace html tags with the enaml component (the capitalized tag name). For example:

from web.components.api import *

enamldef Index(Html):
    Head:
        Title:
            text = "Hello world"
    Body:
        H1:
            text = "Hello world"

Calling render() on an instance of this enaml view then generates the html from the view. This is shown in the simple case of a static site generator:

import enaml
from web.core.app import WebApplication

app = WebApplication()

# Import Index from index.enaml
with enaml.imports():
    from index import Index

# Render the Index.enaml to index.html
view = Index()
with open('index.html', 'w') as f:
    f.write(view.render())

You can also use it in a request handler with your favorite web framework. You can pass in objects to populate dynamic views. For example with tornado web you can do something like this:

import enaml
import tornado.web
import tornado.ioloop
from web.core.app import WebApplication

# Import Index from index.enaml
with enaml.imports():
    from index import Index

class IndexHandler(tornado.web.RequestHandler):
    view = Index()
    def get(self, request):
        return self.view.render(request=request)

class Application(tornado.web.Application):
    def __init__(self):
        super(Application, self).__init__([
                (r'/',IndexHandler) 
           ],
        )
        
if __name__ == "__main__":
    web_app = WebApplication()
    app = Application()
    app.listen(8888)
    tornado.ioloop.IOLoop.current().start()

How it works

enaml-web generates a dom of lxml elements. You can use this to create any html page.

Inhernetly secure

Since an lxml dom is generated it means that your code is inherently secure from injection as it automatically escapes all attributes. Also a closing tag cannot be accidentally missed.

The atom framework provides additional security by enforcing runtime type checking and optional validation.

Extendable via templates and blocks

Like other template engines, enaml-web provides a "Block" node that allows you to define a part of a template that can be overridden or extended.

Enaml also provides pattern nodes for handling conditional statments, loops, dynamic nodes based on lists or models, and nodes generated from more complex templates (ex automatic form generation).

No template tags needed

Many templating engines require the use of "template tags" wrapped in {% %} or similar to allow the use of python code to transform variables.

Since enaml is an extension to python, you can use any python code directly in your enaml components and templates. You don't need any template tags. You can, import and use tag functions from other frameworks if you need.

You can "render" raw html source into nodes such as wysiwyg content from a database or other sources. Components for rendering markdown and highlighted code blocks are also provided.

Component based

Since enaml views are like python classes, you can "subclass" and extend any component and extend it's functionality. This enables you to quickly build reusable components.

I'm working on components for several common css frameworks so they can simply be installed and used.

materialize-ui
semantic-ui (coming soon)
bootstrap (coming soon)

Data binding

Because enaml-web is generating a dom, you can use websockets and some js to manipulate the dom to do data binding between the client to server.

Each node as a unique identifier and can be modified using change events. An example of this is in the examples folder.

You can also have the client trigger events on the server and have the server trigger JS events on the client.

To use:

Include enaml.js in your page
Observe the modified event of an Html node and pass these changes to the client via websockets.
Enamljs will send events back to the server, update the dom accordingly.

Data models

Forms can automatically be generated and populated using enaml's DynamicTemplate nodes. An implementation of the AutoForm using the materalize css framework is available on my personal repo. With this, we can take a model like:

from atom.api import Atom, Unicode, Bool, Enum

class Message(Atom):
    name = Unicode()
    email = Unicode()
    message = Unicode()
    options = Enum("Email","Phone","Text")
    sign_up = Bool(True)

Then use the AutoForm node and pass in either a new or populated instance of the model to render the form.

from templates import Base 
from web.components.api import *
from web.core.api import Block


enamldef AddMessageView(Base): page:
    attr message
    Block:
        block = page.content
        AutoForm:
            model << message

Simple ORM with MongoDB

Using Atom makes it easy to build a simple ORM. One is now provided that handles serialization to and from MongoDB (or anything that takes json). It's used simlar to django. It assumes you're using motor but should also work with txmongo as it simply proxies calls to the provided MongoDB collection.

For example:

from atom.api import Unicode, Int, Instance, List
from web.core.db import Model


class Group(Model):
    name = Unicode()

class User(Model):
    name = Unicode()
    age = Int()
    groups = List(Group)

Then we can create an instance and save it. It will perform an upsert or replace the existing entry.

admins = Group(name="Admins")
await admins.save()

# It will save admins using it's ObjectID 
bob = User(name="Bob", age=32, groups=[admins])
await bob.save()

tom = User(name="Tom", age=34, groups=[admins])
await tom.save()

To fetch from the DB each model has a ModelManager called objects that will simply return the collection for the model type. For example.

# Fetch from db, you can use any MongoDB queries here
state = await User.objects.find_one({'name': "James"})
if state:
    james = await User.restore(state)
    
# etc...

Restoring is async because it will automatically fetch any related objects (ex the groups in this case). It saves objects using the ObjectID when present.

And finally you can either delete using queries on the manager directly or call delete on the object.

You can exclude members from being saved to the DB by tagging them with .tag(store=False).

Raw, Markdown, and Code nodes

TheRaw node parses text into dom nodes (using lxml's html parser). Similarly Markdown and Code nodes parse markdown and highlight code respectively.

For example, you can use wagtal's richtext tag to render to a dom via:

from web.components.api import *
from web.core.api import *
from wagtail.core.templatetags.wagtailcore_tags import richtext
from myapp.views.base import Page

enamldef BlogPage(Page):
    body.cls = 'template-blogpage'
    Block:
        block = parent.content
        Raw:
            source << richtext(page.body)

This let's you use web wysiwyg editors to insert content into the etree.

Block node

You can define a base template, then overwrite parts using the Block node.

In one file put:

from web.components.api import *
from web.core.api import Block

enamldef Base(Html):
    attr user
    attr site
    attr request
    alias content
    Head:
        Title:
            text << site.title
    Body:
        Header:
            text = "Header"
        Block: content:
            pass
        Footer:
            text = "Footer"

Then you can import that view and extend the template and override the block's content.

from templates import Base 
from web.components.api import *
from web.core.api import Block

enamldef Page(Base): page:
    Block:
        block = page.content
        P:
            text = "Content inserted between Header and Footer"

Blocks let you either replace, append, or prepend to the content.

Custom Components

Probably the best part, with enaml you can easily create reusable components and share them through the views as you would any python class.

For instance, to create a materalize breadcrumbs component that automatically follows the current request path, simply include the required css/scripts in your base template, define the component as shown below:

from web.components.api import *
from web.core.api import Looper

enamldef Breadcrumbs(Nav): nav:
    attr path # ex. pass in a tornado request.path
    attr color = ""
    attr breadcrumbs << path[1:-1].split("/")
    tag = 'nav'
    Div:
        cls = 'nav-wrapper {}'.format(nav.color)
        Div:
            cls = 'container'
            Div:
                cls = 'col s12'
                Looper:
                    iterable << breadcrumbs
                    A:
                        href = "/{}/".format("/".join(breadcrumbs[:loop_index+1]))
                        cls = "breadcrumb"
                        text = loop_item.title()

then use it it as follows

# in your template add
Breadcrumbs:
    path << request.path

Gotachas

Text and tail nodes

Lxml uses text and tail properties to set text before and after child nodes, which can be confusing.

For instance in html you can do

<p>This is a sentence <a href="#">click here</a> then keep going</p>

To make this with enaml you need to do this:

P:
    text = "This is a sentence"
    A:
        href = "#"
        text = "click here"
        tail = "then keep going"

Notice how tail is set on the A NOT the P.
See lxml etree documentation for more details.

Tag attribute

In the current implementation the xml tag used is the lowercase of the class name. When you subclass a component you must explicity set the tag attribute to the desired tag name. For example:

enamldef Icon(I):
    tag = 'i' # Force tag to be 'i' instead of 'icon' since 'icon' is not a valid html element
    cls = 'material-icons'

Examples

My website uses it

www.codelv.com - Built entirely using enaml-web

vahndi/enaml-web