WebOb File-Serving Example
==========================
This document shows how you can make a static-file-serving application
using WebOb. We'll quickly build this up from minimal functionality
to a high-quality file serving application.
.. note:: Starting from 1.2b4, WebOb ships with a :mod:`webob.static` module
which implements a :class:`webob.static.FileApp` WSGI application similar to the
one described below.
This document stays as a didactic example how to serve files with WebOb, but
you should consider using applications from :mod:`webob.static` in
production.
.. comment:
>>> import webob, os
>>> base_dir = os.path.dirname(os.path.dirname(webob.__file__))
>>> doc_dir = os.path.join(base_dir, 'docs')
>>> from doctest import ELLIPSIS
First we'll setup a really simple shim around our application, which
we can use as we improve our application:
.. code-block:: python
>>> from webob import Request, Response
>>> import os
>>> class FileApp(object):
... def __init__(self, filename):
... self.filename = filename
... def __call__(self, environ, start_response):
... res = make_response(self.filename)
... return res(environ, start_response)
>>> import mimetypes
>>> def get_mimetype(filename):
... type, encoding = mimetypes.guess_type(filename)
... # We'll ignore encoding, even though we shouldn't really
... return type or 'application/octet-stream'
Now we can make different definitions of ``make_response``. The
simplest version:
.. code-block:: python
>>> def make_response(filename):
... res = Response(content_type=get_mimetype(filename))
... res.body = open(filename, 'rb').read()
... return res
Let's give it a go. We'll test it out with a file ``test-file.txt``
in the WebOb doc directory:
.. code-block:: python
>>> fn = os.path.join(doc_dir, 'test-file.txt')
>>> open(fn).read()
'This is a test. Hello test people!'
>>> app = FileApp(fn)
>>> req = Request.blank('/')
>>> print req.get_response(app)
200 OK
Content-Type: text/plain; charset=UTF-8
Content-Length: 35
<BLANKLINE>
This is a test. Hello test people!
Well, that worked. But it's not a very fancy object. First, it reads
everything into memory, and that's bad. We'll create an iterator instead:
.. code-block:: python
>>> class FileIterable(object):
... def __init__(self, filename):
... self.filename = filename
... def __iter__(self):
... return FileIterator(self.filename)
>>> class FileIterator(object):
... chunk_size = 4096
... def __init__(self, filename):
... self.filename = filename
... self.fileobj = open(self.filename, 'rb')
... def __iter__(self):
... return self
... def next(self):
... chunk = self.fileobj.read(self.chunk_size)
... if not chunk:
... raise StopIteration
... return chunk
... __next__ = next # py3 compat
>>> def make_response(filename):
... res = Response(content_type=get_mimetype(filename))
... res.app_iter = FileIterable(filename)
... res.content_length = os.path.getsize(filename)
... return res
And testing:
.. code-block:: python
>>> req = Request.blank('/')
>>> print req.get_response(app)
200 OK
Content-Type: text/plain; charset=UTF-8
Content-Length: 35
<BLANKLINE>
This is a test. Hello test people!
Well, that doesn't *look* different, but lets *imagine* that it's
different because we know we changed some code. Now to add some basic
metadata to the response:
.. code-block:: python
>>> def make_response(filename):
... res = Response(content_type=get_mimetype(filename),
... conditional_response=True)
... res.app_iter = FileIterable(filename)
... res.content_length = os.path.getsize(filename)
... res.last_modified = os.path.getmtime(filename)
... res.etag = '%s-%s-%s' % (os.path.getmtime(filename),
... os.path.getsize(filename), hash(filename))
... return res
Now, with ``conditional_response`` on, and with ``last_modified`` and
``etag`` set, we can do conditional requests:
.. code-block:: python
>>> req = Request.blank('/')
>>> res = req.get_response(app)
>>> print res
200 OK
Content-Type: text/plain; charset=UTF-8
Content-Length: 35
Last-Modified: ... GMT
ETag: ...-...
<BLANKLINE>
This is a test. Hello test people!
>>> req2 = Request.blank('/')
>>> req2.if_none_match = res.etag
>>> req2.get_response(app)
<Response ... 304 Not Modified>
>>> req3 = Request.blank('/')
>>> req3.if_modified_since = res.last_modified
>>> req3.get_response(app)
<Response ... 304 Not Modified>
We can even do Range requests, but it will currently involve iterating
through the file unnecessarily. When there's a range request (and you
set ``conditional_response=True``) the application will satisfy that
request. But with an arbitrary iterator the only way to do that is to
run through the beginning of the iterator until you get to the chunk
that the client asked for. We can do better because we can use
``fileobj.seek(pos)`` to move around the file much more efficiently.
So we'll add an extra method, ``app_iter_range``, that ``Response``
looks for:
.. code-block:: python
>>> class FileIterable(object):
... def __init__(self, filename, start=None, stop=None):
... self.filename = filename
... self.start = start
... self.stop = stop
... def __iter__(self):
... return FileIterator(self.filename, self.start, self.stop)
... def app_iter_range(self, start, stop):
... return self.__class__(self.filename, start, stop)
>>> class FileIterator(object):
... chunk_size = 4096
... def __init__(self, filename, start, stop):
... self.filename = filename
... self.fileobj = open(self.filename, 'rb')
... if start:
... self.fileobj.seek(start)
... if stop is not None:
... self.length = stop - start
... else:
... self.length = None
... def __iter__(self):
... return self
... def next(self):
... if self.length is not None and self.length <= 0:
... raise StopIteration
... chunk = self.fileobj.read(self.chunk_size)
... if not chunk:
... raise StopIteration
... if self.length is not None:
... self.length -= len(chunk)
... if self.length < 0:
... # Chop off the extra:
... chunk = chunk[:self.length]
... return chunk
... __next__ = next # py3 compat
Now we'll test it out:
.. code-block:: python
>>> req = Request.blank('/')
>>> res = req.get_response(app)
>>> req2 = Request.blank('/')
>>> # Re-fetch the first 5 bytes:
>>> req2.range = (0, 5)
>>> res2 = req2.get_response(app)
>>> res2
<Response ... 206 Partial Content>
>>> # Let's check it's our custom class:
>>> res2.app_iter
<FileIterable object at ...>
>>> res2.body
'This '
>>> # Now, conditional range support:
>>> req3 = Request.blank('/')
>>> req3.if_range = res.etag
>>> req3.range = (0, 5)
>>> req3.get_response(app)
<Response ... 206 Partial Content>
>>> req3.if_range = 'invalid-etag'
>>> req3.get_response(app)
<Response ... 200 OK>
