This is a pre-production deployment of Warehouse. Changes made here affect the production instance of PyPI (
Help us improve Python packaging - Donate today!

A framework for allowing customizing templates

Project Description



This package contains an infrastructure and API for the customization of templates. The only template languages supported by this system are “pure-push” languages which do not call into arbitrary Python code while executing. Examples of such languages are json-template (supported out of the box) and XSLT. The advantage of such languages is that they are reasonably secure to expose through-the-web customization without an elaborate security infrastructure.

Let’s go through the use cases that this system must support:

  • templates exist on the filesystem, and those are used by default.
  • templates can be customized.
  • this customization can be stored in another database (ZODB, filesystem, a relational database, etc); this is up to the person integrating hurry.custom.
  • update template automatically if it is changed in the database.
  • it is possible to retrieve the template source (for display in a UI or for later use within for instance a web-browser for client-side rendering).
  • support server-side rendering of templates (producing HTML or an email message or whatever). Input is particular to template language (but should be considered immutable).
  • provide (static) input samples (such as JSON or XML files) to make it easier to edit and test templates. These input samples can be added both to the filesystem as well as to the database.
  • round-trip support. The customized templates and samples can be retrieved from the database and exported back to the filesystem. This is useful when templates need to be taken back under version control after a period of customization by end users.

The package is agnostic about (these things are pluggable):

  • the database used for storing customizations of templates or their samples.
  • the particular push-only template language used.

What this package does not do is provide a user interface. It only provides the API that lets you construct such user interfaces.

Creating and registering a template language

In order to register a new push-only template we need to provide a factory that takes the template text (which could be compiled down further). Instantiating the factory should result in a callable that takes the input data (in whatever format is native to the template language). The ITemplate interface defines such an object:

>>> from hurry.custom.interfaces import ITemplate, CompileError, RenderError

For the purposes of demonstrating the functionality in this package, we supply a very simplistic push-only templating language, based on template strings as provided by the Python string module:

>>> import string
>>> from zope.interface import implements
>>> class StringTemplate(object):
...    implements(ITemplate)
...    def __init__(self, text):
...        if '&' in text:
...            raise CompileError("& in template!")
...        self.source = text
...        self.template = string.Template(text)
...    def __call__(self, input):
...        try:
...            return self.template.substitute(input)
...        except KeyError, e:
...            raise RenderError(unicode(e))

Let’s demonstrate it. To render the template, simply call it with the data as an argument:

>>> template = StringTemplate('Hello $thing')
>>> template({'thing': 'world'})
'Hello world'

Note we have put some special logic in the __init__ that triggers a CompileError error if the string & is found in the template. This is so we can easily demonstrate templates that are broken - treat a template with & as a template with a syntax (compilation) error. Let’s try it:

>>> template = StringTemplate('Hello & bye')
Traceback (most recent call last):
CompileError: & in template!

We have also made sure we catch a possible runtime error (a KeyError when a key is missing in the input dictionary in this case) and raise this as a RenderError:

>>> template = StringTemplate('Hello $thing')
>>> template({'thang': 'world'})
Traceback (most recent call last):
RenderError: 'thing'

The template class defines a template language. Let’s register the template language so the system is aware of it and treats .st files on the filesystem as a string template:

>>> from hurry import custom
>>> custom.register_language(StringTemplate, extension='.st')

Loading a template from the filesystem

hurry.custom assumes that any templates that can be customized reside on the filesystem primarily and are shipped along with an application’s source code. They form collections. A collection is simply a directory (with possible sub-directories) that contains templates.

Let’s create a collection of templates on the filesystem:

>>> import tempfile, os
>>> templates_path = tempfile.mkdtemp(prefix='hurry.custom')

We create a single template, for now:

>>> test1_path = os.path.join(templates_path, '')
>>> f = open(test1_path, 'w')
>>> f.write('Hello $thing')
>>> f.close()

We also create an extra template:

>>> test2_path = os.path.join(templates_path, '')
>>> f = open(test2_path, 'w')
>>> f.write("It's full of $thing")
>>> f.close()

In order for the system to work, we need to register this collection of templates on the filesystem. We need to supply a globally unique collection id, the templates path, and (optionally) a title:

>>> custom.register_collection(id='templates', path=templates_path)

We can now render the template:

>>> custom.render('templates', '', {'thing': 'world'})
u'Hello world'

We’ll try another template:

>>> custom.render('templates', '', {'thing': 'stars'})
u"It's full of stars"

We can also look up the template object:

>>> template = custom.lookup('templates', '')

We got our proper template:

>>> template({'thing': 'world'})
u'Hello world'

The templat also has a source attribute:

>>> template.source
u'Hello $thing'

The source text of the template was interpreted as a UTF-8 string. The template source should always be in unicode format (or in plain ASCII).

The underlying template will not be reloaded unless it is changed on the filesystem:

>>> orig = template.template

When we trigger a potential reload nothing happens - the template did not change on the filesystem:

>>> template.source
u'Hello $thing'
>>> template.template is orig

It will however automatically reload the template when it has changed on the filesystem. We will demonstrate that by modifying the file:

>>> f = open(test1_path, 'w')
>>> f.write('Bye $thing')
>>> f.close()

Unfortunately this won’t work in the tests as the modification time of files has a second-granularity on some platforms, way too long to delay the tests for. We will therefore manually update the last updated time as a hack:

>>> template._last_updated -= 1

Now the template will have changed:

>>> template.source
u'Bye $thing'

>>> template({'thing': 'world'})
u'Bye world'

Customization database

So far all our work was done in the root (filesystem) database. We can get it now:

>>> root_db = custom.root_collection('templates')

Before any customization database was registered we could also have gotten it using custom.collection, which gets the collection in context:

>>> custom.collection('templates') is root_db

Let’s now register a customization database for our collection, in a particular site. This means in such a site, the new customized template database will be used (with a fallback on the original one if no customization can be found or if there is an error in the use of a customization).

Let’s create a site first:

>>> site1 = DummySite(id=1)

We register a customization database for our collection named templates. For the purposes of testing we will use an in-memory database:

>>> mem_db = custom.InMemoryTemplateDatabase('templates', 'Templates')
>>> from hurry.custom.interfaces import ITemplateDatabase
>>> sm1 = site1.getSiteManager()
>>> sm1.registerUtility(mem_db, provided=ITemplateDatabase,
...   name='templates')

We go into this site:

>>> setSite(site1)

We can now find this collection using custom.collection:

>>> custom.collection('templates') is mem_db

The collection below it is the root collection:

>>> custom.next_collection('templates', mem_db) is root_db

Below this, there is no collection and we’ll get a lookup error:

>>> custom.next_collection('templates', root_db)
Traceback (most recent call last):
ComponentLookupError: No collection available for: templates

We haven’t placed any customization in the customization database yet, so we’ll see the same thing as before when we look up the template:

>>> custom.render('templates', '', {'thing': "universe"})
u'Bye universe'

Customization of a template

Now that we have a locally set up customization database, we can customize the template.

In this customization we change ‘Bye’ to ‘Goodbye’:

>>> source = root_db.get_source('')
>>> source = source.replace('Bye', 'Goodbye')

We now need to update the database so that it has this customized version of the template. We do this by calling the update method on the database with the template id and the new source.

This update operation is not supported on the default filesystem database:

>>> root_db.update('', source)
Traceback (most recent call last):
NotSupported: Cannot update templates in FilesystemTemplateDatabase.

It is supported on the site-local in-memory database we’ve just installed though:

>>> mem_db.update('', source)

All you need to do to hook in your own database is to implement the ITemplateDatabase interface and register it (either globally or locally in a site).

Let’s see whether we get the customized template now:

>>> custom.render('templates', '', {'thing': 'planet'})
u'Goodbye planet'

Broken custom template

If a custom template cannot be compiled, the system falls back on the filesystem template instead. We construct a broken custom template by adding & to it:

>>> original_source = root_db.get_source('')
>>> source = original_source.replace('full of', 'filled with &')
>>> mem_db.update('', source)

We try to render this template, but instead we’ll see the original template:

>>> custom.render('templates', '', {'thing': 'planets'})
u"It's full of planets"

It could also be the case that the custom template can be compiled but instead cannot be rendered. Let’s construct one that expects thang instead of thing:

>>> source = original_source.replace('$thing', '$thang')
>>> mem_db.update('', source)

When rendering the system will notice the RenderError and fall back on the original uncustomized template for rendering:

>>> custom.render('templates', '', {'thing': 'planets'})
u"It's full of planets"

Checking which template languages are recognized

We can check which template languages are recognized:

>>> languages = custom.recognized_languages()
>>> sorted(languages)
[(u'.st', <class 'StringTemplate'>)]

When we register another language:

>>> class StringTemplate2(StringTemplate):
...   pass
>>> custom.register_language(StringTemplate2, extension='.st2')

It will show up too:

>>> languages = custom.recognized_languages()
>>> sorted(languages)
[(u'.st', <class 'StringTemplate'>), (u'.st2', <class 'StringTemplate2'>)]

Retrieving which templates can be customized

For the filesystem-level templates it is possible to get a data structure that indicates which templates can be customized. This is useful when constructing a UI. This data structure is designed to be easily useful as JSON so that a client-side UI can be constructed.

Let’s retrieve the customization database for our collection:

>>> l = custom.structure('templates')
>>> from pprint import pprint
>>> pprint(l)
[{'extension': '.st',
  'name': 'test1',
  'path': '',
  'template': ''},
 {'extension': '.st',
  'name': 'test2',
  'path': '',
  'template': ''}]


In a customization user interface it is useful to be able to test the template. Sometimes this can be done with live data coming from the software, but in other cases it is more convenient to try it on some representative sample data. This sample data needs to be in the format as expected as the argument when calling the template.

Just like a template language is stored as plain text on the filesystem, sample data can also be stored as plain text on the file system. The format of this plain text is its data language. Examples of data languages are JSON and XML.

For the purposes of demonstration, we’ll define a simle data language that can turn into a dictionary a data file with key value pairs like this:

>>> data = """\
... a: b
... c: d
... e: f
... """

Now we define a function that can parse this data into a dictionary:

>>> def parse_dict_data(data):
...    result = {}
...    for line in data.splitlines():
...        key, value = line.split(':')
...        key = key.strip()
...        value = value.strip()
...        result[key] = value
...    return result
>>> d = parse_dict_data(data)
>>> sorted(d.items())
[('a', 'b'), ('c', 'd'), ('e', 'f')]

The idea is that we can ask a particular template for those sample inputs that are available for it. Let’s for instance check for sample inputs available for

>>> root_db.get_samples('')

There’s nothing yet.

In order to get samples to work, we first need to register the data language:

>>> custom.register_data_language(parse_dict_data, '.d')

Files with the extension .d can now be recognized as containing sample data.

We still need to tell the system that StringTemplate templates in particular can be expected to find sample data with this extension. In order to express this, we need to register the StringTemplate language again with an extra argument that indicates this (sample_extension):

>>> custom.register_language(StringTemplate,
...    extension='.st', sample_extension='.d')

Now we can actually look for samples. Of course there still aren’t any as we haven’t created any .d files yet:

>>> root_db.get_samples('')

We need a pattern to associate a sample data file with a template file. The convention used is that a sample data file is in the same directory as the template file, and starts with the name of the template followed by a dash (-). Following the dash should be the name of the sample itself. Finally, the extension should be the sample extension. Here we create a sample file for the template:

>>> test1_path = os.path.join(templates_path, 'test1-sample1.d')
>>> f = open(test1_path, 'w')
>>> f.write('thing: galaxy')
>>> f.close()

Now when we ask for the samples available for our test1 template, we should see sample1:

>>> r = root_db.get_samples('')
>>> r
{'sample1': {'thing': 'galaxy'}}

By definition, we can use the sample data for a template and pass it to the template itself:

>>> template = custom.lookup('templates', '')
>>> template(r['sample1'])
u'Goodbye galaxy'

Testing a template

In a user interface it can be useful to be able to test whether the template compiles and renders. hurry.custom therefore implements a check function that does so. This function raises an error (CompileError or RenderError), and passes silently if there is no problem.

Let’s first try it with a broken template:

>>> custom.check('templates', '', 'foo & bar')
Traceback (most recent call last):
CompileError: & in template!

We’ll now try it with a template that does compile but doesn’t work with sample1, as no something is supplied:

>>> custom.check('templates', '', 'hello $something')
Traceback (most recent call last):
RenderError: 'something'

Error handling

Let’s try to render a template in a collection that doesn’t exist. We get a message that the template database could not be found:

>>> custom.render('nonexistent', '', {})
Traceback (most recent call last):
ComponentLookupError: (<InterfaceClass hurry.custom.interfaces.ITemplateDatabase>, 'nonexistent')

Let’s render a non-existent template in an existing database. We get the lookup error of the deepest database, which is assumed to be the filesystem:

>>> custom.render('templates', '', {})
Traceback (most recent call last):
IOError: [Errno 2] No such file or directory: '.../'

Let’s render a template with an unrecognized extension:

>>> custom.render('templates', 'dummy.unrecognized', {})
Traceback (most recent call last):
ComponentLookupError: (<InterfaceClass hurry.custom.interfaces.ITemplate>, '.unrecognized')

The template language .unrecognized could not be found. Let’s make the file exist; we should get the same result:

>>> unrecognized = os.path.join(templates_path, 'dummy.unrecognized')
>>> f = open(unrecognized, 'w')
>>> f.write('Some weird template language')
>>> f.close()

Now let’s look at it again:

>>> template = custom.render('templates', 'dummy.unrecognized', {})
Traceback (most recent call last):
ComponentLookupError: (<InterfaceClass hurry.custom.interfaces.ITemplate>, '.unrecognized')

If we try to look up a template in the root collection with a CompileError in it, we’ll get a CompileError:

>>> compile_error = os.path.join(templates_path, '')
>>> f = open(compile_error, 'w')
>>> f.write('A & compile error')
>>> f.close()
>>> compile_error_template = custom.lookup('templates', '')
Traceback (most recent call last):
CompileError: & in template!

The same applies to trying to render it:

>>> custom.render('templates', '', {})
Traceback (most recent call last):
CompileError: & in template!

If we try to render a template in the root collection we get a RenderError:

>>> render_error = os.path.join(templates_path, '')
>>> f = open(render_error, 'w')
>>> f.write('A $thang')
>>> f.close()
>>> custom.render('templates', '', {'thing': 'thing'})
Traceback (most recent call last):
RenderError: u'thang'

We’ll get a ComponentLookupError if we look for a collection with an unknown id:

>>> custom.collection('unknown_id')
Traceback (most recent call last):
ComponentLookupError: (<InterfaceClass hurry.custom.interfaces.ITemplateDatabase>, 'unknown_id')

We also can’t look for a next collection if the id we specify is unknown:

>>> custom.next_collection('unknown_id', mem_db)
Traceback (most recent call last):
ComponentLookupError: No more utilities for <InterfaceClass hurry.custom.interfaces.ITemplateDatabase>, 'unknown_id' have been found.

Similarly we can’t get a root collection if the id is unknown:

>>> custom.root_collection('unknown_id')
Traceback (most recent call last):
ComponentLookupError: (<InterfaceClass hurry.custom.interfaces.ITemplateDatabase>, 'unknown_id')


0.6.2 (2009-06-15)

  • Both RenderError and CompileError subclass from a common Error base class.

0.6.1 (2009-06-15)

  • structure functionality now skips directories and files start that with a period.
  • If data language is not known, don’t return any samples.

0.6 (2009-06-10)

  • Introduce the notion of CompileError and RenderError. A CompileError should be raised by a template if the template cannot be parsed or compiled. A RenderError should be raised if there is any run-time error during template rendering.
  • Introduce render in the API and de-emphasize the use of lookup. Normally templates are rendered by calling render.
  • When a template is looked up and there is a CompileError during its creation, fall back on original template.
  • When a template is rendered using the top-level render function and there is a RenderError during the rendering process, fall back on the original template.
  • Remove original_source and samples methods from IManagedTemplate interface. These are better handled by directly using the ITemplateDatabase API.
  • Some fixes in the interfaces, bringing them more inline with the code.
  • Expose collection, next_collection and root_collection functions.

0.5 (2009-05-22)

  • Initial public release.


Release History

Release History

This version
History Node


History Node


History Node


History Node


Download Files

Download Files

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

File Name & Checksum SHA256 Checksum Help Version File Type Upload Date
hurry.custom-0.6.2.tar.gz (26.3 kB) Copy SHA256 Checksum SHA256 Source Jun 15, 2009

Supported By

WebFaction WebFaction Technical Writing Elastic Elastic Search Pingdom Pingdom Monitoring Dyn Dyn DNS Sentry Sentry Error Logging CloudAMQP CloudAMQP RabbitMQ Heroku Heroku PaaS Kabu Creative Kabu Creative UX & Design Fastly Fastly CDN DigiCert DigiCert EV Certificate Rackspace Rackspace Cloud Servers DreamHost DreamHost Log Hosting