Permalink
Find file
Fetching contributors…
Cannot retrieve contributors at this time
93 lines (66 sloc) 3.65 KB
Writing new processors
----------------------
First and foremost: the filename and unique_filename processor modules are
intended as simple examples that you can look at to learn how to write
processors. The other processor modules are also simple (most processors will
be simple!), but are somewhat larger and therefor more difficult to use to
grok the possibilities of processors.
I've copy/pasted the unique_filename.Mtime processor below and annotated it
to indicate the helper methods and automagic features the Processor class
provides:
# This imports the Processor class and exceptions it can raise.
from processor import *
# Additional imports.
import stat
import shutil
# The Mtime processor class is a subclass of Processor.
class Mtime(Processor):
"""gives the file a unique filename based on its mtime"""
# This variable is used by the arbitrator to detect if this file should be
# processed on a per-server basis or if the same file can be processed
# once and then synced to all servers. The default is False and does not
# have to be set explicitly as was done here for demonstrative purposes.
# Set it to True to enable per-server processing.
different_per_server = False
# This variable is picked up automatically by the Processor base class and
# is used to validate the input file. The empty tuple can be used to allow
# any extension, otherwise use tuples like (".css", ".js").
valid_extensions = () # Any extension is valid.
# Each processor must have a run method, which will perform the actual
# processing. Before this method is called, basic validation is performed
# automatically: it is verified that the input file exists and that the
# extension of the input file is in the list of valid extensions.
def run(self):
# Processor.get_path_parts() is one of the helper methods provided.
# E.g. Processor.get_path_parts("/foo/bar/baz.jpg") yields:
# ("/foo/bar", "baz.jpg", "baz", "jpg")
# Get the parts of the input file.
(path, basename, name, extension) = Processor.get_path_parts(self, self.input_file)
# Set the output file base name.
mtime = os.stat(self.input_file)[stat.ST_MTIME]
# set_output_file_basename() is another helper method and should be
# used to set the output file. It ensures that the path has not been
# changed, which is not allowed. Only changes in the filename are
# allowed.
self.set_output_file_basename(name + "_" + str(mtime) + extension)
# Copy the input file to the output file.
shutil.copyfile(self.input_file, self.output_file)
# Each processor should always return the output file. This makes
# perfect sense, as we've received an input file and are now returning
# the output file, which will then be used as the input file for the
# next processor, and so on.
return self.output_file
Writing new transporters
------------------------
Transporters are (very) thin threaded wrappers around Django custom storage
systems. Django custom storage systems are subclasses of the Django Storage
class.
So step one is writing a new Django custom storage system. You can learn how
to do that here:
http://docs.djangoproject.com/en/1.0/howto/custom-file-storage/
But please make sure it's not already been written:
http://code.welldev.org/django-storages/wiki/Home
After that is finished, you must write the thin wrapper class that will be
your Transporter. Look at the FTP and S3 Transporters as an example. Actually
writing a tutorial on Transporters would make it seem more complex than it is,
so just read those 50 or so lines of code :)