Choose a configuration approach for Aglyph

Release:2.1.1

Aglyph explicitly supports two methods of configuration:

  1. Declarative XML configuration conforming to the Aglyph context DTD
  2. Programmatic configuration via aglyph.binder.Binder

Opinions vary widely on the merits of XML (particularly for configuration), but in fairness there’s also plenty of debate over the merits of “code as configuration” (or “configuration as code” if you prefer) as well. Aglyph strives to not have an opinion one way or the other by supporting either approach.

However, both approaches to Aglyph configuration have strengths and weaknesses that you should understand before choosing one over another.

Declarative XML configuration

An Aglyph context can be defined in an XML document that conforms to the Aglyph context DTD.

The XML context document is parsed by aglyph.context.XMLContext, which is just a subclass of aglyph.context.Context that knows how to populate itself from the parsed XML document. Once populated, this context can then be used to create an aglyph.assembler.Assembler:

>>> from aglyph.assembler import Assembler
>>> from aglyph.context import XMLContext
>>> my_app_context = XMLContext("my-application-context.xml")
>>> assembler = Assembler(my_app_context)

Note

The Aglyph context DTD is provided primarily as a reference. The aglyph.context.XMLContext class uses a non-validating parser by default.

Developers are encouraged to explicitly validate an application’s context XML document during testing.

See also

Use a custom XML parser for XMLContext
This recipe could also be used to force Aglyph to use a validating XML parser.

XMLContext configures mutable builtin objects safely

Consider the following example:

<?xml version="1.0" encoding="utf-8"?>
<context id="cookbook">
    <component id="cookbook.Example">
        <attributes>
            <attribute name="mutable">
               <list>
                  <int>1</int>
                  <int>2</int>
                  <int>3</int>
               </list>
            </attribute>
        </attributes>
    </component>
</context>

Because builtin list objects are mutable, Aglyph will automatically turn the “mutable” attribute above into an aglyph.component.Evaluator (which is very similar to a functools.partial). Whenever the “cookbook.Example” component is assembled, the Evaluator for the “mutable” attribute is called, which will produce a new list object.

Why is this important? Consider a corresponding Binder configuration for the same component:

binder.bind("cookbook.Example").attributes(mutable=[1, 2, 3])

This configuration leads to a (likely) logic error: all objects of the “cookbook.Example” component will share a reference to a single list object. An example illustrates the problem:

>>> example1 = binder.lookup("cookbook.Example")
>>> example1.mutable
[1, 2, 3]
>>> example1.mutable.append(4)
>>> example2 = binder.lookup("cookbook.Example")
>>> example2.mutable
[1, 2, 3, 4]

Uh-oh! That’s almost certainly not what we intended. To guard against this behavior, we would need to modify the binding:

from functools import partial
binder.bind("cookbook.Example").attributes(
    mutable=partial(list, [1, 2, 3])
)

Now we will get a “fresh” list every time the component is assembled, so modifying the list on one instance will not affect the lists of any other instances.

(And what if we were actually specifying a list-of-list, or a tuple-of-list, or a list-of-dict? Now we would need to account for mutability of each member!)

This is an easy thing to forget, and can lead to a great deal of (programmatic) configuration code, which is why aglyph.context.XMLContext handles it automatically for any <list>, <tuple>, and <dict> declared in the XML context document.

XMLContext is Unicode-aware and supports automatic character set conversion

Aglyph properly handles Unicode text and encoded byte data in XML context documents, regardless of Python version.

Aglyph can also provide your application components with byte data encoded to a user-specified character set.

Consider the following example:

<?xml version="1.0" encoding="utf-8"?>
<context id="cookbook">
    <component id="cookbook.TextAndData">
        <attributes>
            <attribute name="text">
                <unicode>ΑΦΔ</unicode>
            </attribute>
            <attribute name="data1">
                <bytes>ΑΦΔ</bytes>
            </attribute>
            <attribute name="data2">
                <bytes encoding="iso-8859-7">ΑΦΔ</bytes>
            </attribute>
        </attributes>
    </component>
</context>

The first thing to notice is that <bytes>ΑΦΔ</bytes> is missing a character encoding. This can be problematic on Python 2, because the default string encoding used by the Unicode implementation is typically ASCII:

$ python2.7
Python 2.7.9 (default, Dec 13 2014, 15:13:49)
[GCC 4.2.1 Compatible Apple LLVM 6.0 (clang-600.0.56)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> import sys
>>> sys.getdefaultencoding()
'ascii'
>>> from aglyph.context import XMLContext
>>> context = XMLContext("cookbook-context.xml")
Traceback (most recent call last):
  ...
UnicodeEncodeError: 'ascii' codec can't encode characters in position 0-2: ordinal not in range(128)

One solution would be to add the encoding= attribute. Alternatively, you can instruct XMLContext to use a different default encoding (it uses the value of sys.getdefaultencoding() by default):

$ python2.7
Python 2.7.9 (default, Dec 13 2014, 15:13:49)
[GCC 4.2.1 Compatible Apple LLVM 6.0 (clang-600.0.56)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> from aglyph.assembler import Assembler
>>> from aglyph.context import XMLContext
>>> context = XMLContext("cookbook-context.xml", default_encoding="UTF-8")
>>> assembler = Assembler(context)
>>> text_and_data = assembler.assemble("cookbook.TextAndData")
>>> text_and_data.text
u'\u0391\u03a6\u0394'
>>> text_and_data.data1
'\xce\x91\xce\xa6\xce\x94'
>>> text_and_data.data2
'\xc1\xd6\xc4'

If we run the same example under Python 3 (which uses “UTF-8” as the default encoding), we still get correct results, but without the need to explicitly set the default encoding on the XMLContext:

$ python3.4
Python 3.4.2 (default, Nov 12 2014, 18:23:59)
[GCC 4.2.1 Compatible Apple LLVM 6.0 (clang-600.0.54)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> from aglyph.assembler import Assembler
>>> from aglyph.context import XMLContext
>>> context = XMLContext("cookbook-context.xml")
>>> assembler = Assembler(context)
>>> text_and_data = assembler.assemble("cookbook.TextAndData")
>>> text_and_data.text
'ΑΦΔ'
>>> text_and_data.data1
b'\xce\x91\xce\xa6\xce\x94'
>>> text_and_data.data2
b'\xc1\xd6\xc4'

One important thing to notice is the difference in the types of the Unicode and byte strings, dependent upon which version of Python is used.

Unicode and character encoding differences between Python 2 and Python 3

The builtin str type has changed significantly between Python 2 and Python 3 (see Text Vs. Data Instead Of Unicode Vs. 8-bit).

In short: str represented encoded byte data up to and including Python 2, but representes Unicode text as of Python 3.0:

Version Unicode text Encoded byte data
Python 2 unicode str
Python 3 str bytes

The Aglyph context DTD defines <bytes>, <str>, and <unicode> elements that correspond to the types in the table above, but treats the element content differently depending on the version of Python under which Aglyph is running:

Version <unicode> content <str> content <bytes> content
Python 2 unicode str str
Python 3 str str bytes

To summarize the above:

  • <unicode> is interpreted as a unicode type in Python 2 and a str type in Python 3
  • <str> is always interpreted as a str type
  • <bytes> is interpreted as a str type in Python 2 and a bytes type in Python 3

Note

For clarity in XML context documents, it is always safe to use <bytes> for encoded byte data and <unicode> for Unicode text (regardless of Python version), avoiding entirely the ambiguity of <str>.

Warning

Althoug the DTD permits an encoding=”...” attribute on <str> elements, the attribute is ignored in Python 3 (a WARNING-level log message is emitted to the aglyph.context.XMLContext channel if it is present).

Programmatic configuration using Binder

New in version 1.1.0.

An aglyph.binder.Binder can be used to both define and assemble components, and can be used in “chained call” fashion to minimize the amount of code necessary to configure Aglyph.

The easiest way to use Binder is to create an instance of it within its own module, configure it, then import the instance from elsewhere in your application to make use of it.

For example, in a cookbook/bindings.py module:

from aglyph.binder import Binder

binder = Binder("my-app-binder")
binder.bind("cookbook.Example").init("sample").attributes(priority=1)
# ... as many bindings as needed ...

The aglyph.binder.Binder.bind() method returns a proxy object that has init and attributes methods which you can use to specify the arguments and attributes/properties/setters, respectively.

Elsewhere in your application, you can assemble a “cookbook.Example” object like this:

from cookbook.bindings import binder

example = binder.lookup("cookbook.Example")

Referring to other components using Binder is a matter of using aglyph.component.Reference to specify argument or attribute values:

from aglyph.binder import Binder
from aglyph.component import Reference

binder = Binder("my-app-binder")
binder.bind("cookbook.Thing").init("value")
binder.bind("cookbook.Transmogrifier").init(Reference("cookbook.Thing"))

Note

The Common usage scenarios cookbook article provides a number of detailed examples of using Binder for Aglyph configuration.

Binder can generate component IDs and References automatically

Binder supports a form of “shorthand” which can be used when a component’s ID will be its importable dotted name.

Let’s rewrite the previous example to demonstrate:

from cookbook import Thing, Transmogrifier
from aglyph.binder import Binder
from aglyph.component import Reference

binder = Binder("my-app-binder")
binder.bind(Thing).init("value")
binder.bind(Transmogrifier).init(Thing)

Notice that we are now passing the Thing and Transmogrifier classes directly to the bind and init methods.

The aglyph.binder.Binder.bind() method’s component_spec argument and to keyword will automatically convert any class, unbound function, or module into an importable dotted name string. Likewise, the init and attributes methods of the proxy object returned by bind() will convert any value that is a class, unbound function, or module into an aglyph.component.Reference to that object.

So in the above example, bind(Thing) actually expands to bind("cookbook.Thing"), bind(Transmogrifier) to bind("cookbook.Transmogrifier"), and init(Thing) to init(Reference("cookbook.Thing")).

Warning

The technique shown here will not work for components that are assigned a user-specified component ID that is not a Python dotted name. In those cases, Reference("user-component-id") must be used to specify a reference.

Custom configuration using Context

Do neither declarative XML nor programmatic configuration suit your fancy?

An aglyph.context.Context is just a dict that maps component ID strings (i.e. aglyph.component.Component.component_id) to aglyph.component.Component instances, so embrace the open source philosophy and “roll your own” configuration mechanism!

Note

Both aglyph.context.XMLContext and aglyph.binder.Binder are simply adaptors that create and populate an internal aglyph.context.Context!