Sophie

Sophie

distrib > Mandriva > 2010.0 > i586 > media > contrib-release > by-pkgid > 48787808f0aa8c66c98ca8aead901311 > files > 78

pyblosxom-1.4.3-2mdv2010.0.noarch.rpm

===============
Writing Plugins
===============

:Author: PyBlosxom Development Team
:Version: $Id: dev_writing_plugins.txt 1147 2007-11-30 17:37:47Z willhelm $
:Copyright: This document is distributed under the MIT license.

.. contents::


Summary
=======

This chapter covers a bunch of useful things to know when writing
PyBlosxom plugins.  This chapter, moreso than the rest of this manual,
is very much a work in progress.

FIXME - absorb from wiki, Will's tips, and tutorials


Things that all plugins should have
===================================

All plugins should have a docstring at the top of the file that describes
in detail:

1. what the plugin does
2. how to install it
3. how to configure it
4. the license the plugin is distributed under
5. and any copyright information you have
6. any notes about requirements (e.g. "Requires Python 2.3 or greater")

For example, this is at the top of Will's wbgpager plugin::

   """
   Quickly written plugin for paging long index pages.  

   PyBlosxom uses the num_entries configuration variable to prevent
   more than num_entries being rendered by cutting the list down
   to num_entries entries.  So if your num_entries is set to 20, you
   will only see the first 20 entries rendered.

   The wbgpager overrides this functionality and allows for paging.
   It does some dirty stuff so that PyBlosxom doesn't cut the list down
   and then wbgpager cuts it down in the prepare callback later down
   the line.

   To install wbgpager, do the following:

     1. add "wbgpager" to your load_plugins list variable in your
        config.py file--make sure it's the first thing listed so
        that it has a chance to operate on the entry list before
        other plugins.
     2. add the $page_navigation variable to your head or foot
        (or both) templates.  this is where the page navigation
        HTML will appear.


   Here are some additional configuration variables to adjust the 
   behavior:

     wbgpager_count_from
       datatype:       int
       default value:  0
       description:    Some folks like their paging to start at 1--this
                       enables you to do that.

     wbgpager_previous_text
       datatype:       string
       default value:  <<
       description:    Allows you to change the text for the prev link.

     wbgpager_next_text
       datatype:       string
       default value:  >>
       description:    Allows you to change the text for the next link.

     wbgpager_linkstyle
       datatype:       integer
       default value:  0
       description:    This allows you to change the link style of the paging.
                       style 0:  [1] 2 3 4 5 6 7 8 9 ... >>
                       style 1:  Page 1 of 4 >>


   That should be it!


   Note: This plugin doesn't work particularly well with static rendering.
   The problem is that it relies on the querystring to figure out which
   page to show and when you're static rendering, only the first page
   is rendered.  This will require a lot of thought to fix.  If you are
   someone who is passionate about fixing this issue, let me know.


   Permission is hereby granted, free of charge, to any person
   obtaining a copy of this software and associated documentation
   files (the "Software"), to deal in the Software without restriction,
   including without limitation the rights to use, copy, modify,
   merge, publish, distribute, sublicense, and/or sell copies of the
   Software, and to permit persons to whom the Software is furnished
   to do so, subject to the following conditions:

   The above copyright notice and this permission notice shall be
   included in all copies or substantial portions of the Software.

   THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
   EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
   OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
   NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
   BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
   ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
   CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
   SOFTWARE.

   Copyright 2004, 2005, 2006 Will Guaraldi
   """


All plugins should have the following module-level variables 
defined in them just after the docstring:

* ``__author__`` - This holds your name and email address
  so that people can contact you when they have problems.

* ``__version__`` - This holds the version number and release
  date so that people know what version of the plugin they're looking 
  at.

* ``__url__`` - This holds the url where people can find information
  about your plugin and documentation and download new versions of your
  plugin.

* ``__description__`` - This is a one-sentence description of what your 
  plugin does.


For example::

   __author__      = "Will Guaraldi - willg at bluesock dot org"
   __version__     = "version 1.5 2006-01-15"
   __url__         = "http://www.bluesock.org/~willg/pyblosxom/"
   __description__ = "Splits long indexes into multiple pages."


After that, you should have a verify_installation section that
verifies that the plugin is configured correctly.  As of PyBlosxom 0.9, 
the pyblosxom.cgi is able to test your PyBlosxom installation.  It 
verifies certain items in your config.py file and also loads all the 
plugins and lets them verify their configuration as well.
First it tells you your Python version, OS name, and then proceeds to verify
your config properties (did you specify a valid datadir?  does it
exist?...) and then initializes all your plugins and executes
verify_installation(request) on every plugin you have installed that 
has the function.

As a plugin developer, you should add a verify_installation function
to your plugin module.  Something like this (taken from pycategories)::

   def verify_installation(request):
       config = request.getConfiguration()

       if not config.has_key("category_flavour"):
           print "missing optional config property 'category_flavour' "
           print "which allows you to specify the flavour for the category "
           print "link.  refer to pycategory plugin documentation for more "
           print "details."
       return 1


This gives you (the plugin developer) the opportunity to walk the user 
through configuring your highly complex, quantum-charged, turbo plugin 
in small baby steps without having to hunt for where their logs might be.

So check the things you need to check, print out error messages
(informative ones), and then return a 1 if the plugin is configured 
correctly or a 0 if it's not configured correctly.

This is not a substitute for the user to read the installation instructions.  
It should be a really easy way to catch a lot of potential problems
without involving the web server's error logs and debugging information
being sent to a web-browser and things of that nature.

Here's another example of verify_installation from Will's wbgpager
plugin::

   def verify_installation(request):
       config = request.getConfiguration()
       if config.get("num_entries", 0) == 0:
           print "missing config property 'num_entries'.  wbgpager won't do "
           print "anything without num_entries set.  either set num_entries "
           print "to a positive integer, or disable the wbgpager plugin."
           print "see the documentation at the top of the wbgpager plugin "
           print "code file for more details."
           return 0

       return 1



How to log messages to a log file
=================================

First you need to get the logger instance.  After that, you can call 
debug, info, warning, error and critical on the logger instance.  For 
example::

   from pyblosxom import tools

   def cb_prepare(args):
      ...
      logger = tools.getLogger()
      logger.info("blah blah blah...")

      try:
         ...
      except Exception, e:
         logger.error(e)



How to store plugin state between callbacks
===========================================

The easiest way to store state between callbacks is to store the data
in the data dict of the Request object.  For example::

   STATE_KEY = "myplugin_state"

   def cb_date_head(args):
      request = args["request"]
      data = request.getData()

      if data.has_key(STATE_KEY) and data[STATE_KEY]["blah"] == "blahblah":
         ...


   def cb_filelist(args):
      request = args["request"]
      data = request.getData()

      data[STATE_KEY] = {}
      data[STATE_KEY]["blah"] = "blahblah"



How to implement a callback
===========================

If you want to implement a callback, you add a function corresponding
to the callback name to your plugin module.  For example, if you wanted
to modify the Request object just before rendering, you'd implement
cb_prepare like this::

   def cb_prepare(args):
       pass


Obviously, since we have ``pass`` we're not actually doing anything
here, but when the user sends a request and PyBlosxom handles it, 
this function in your plugin will get called when PyBlosxom runs the 
prepare callback.

Each callback passes in arguments through a single dictionary.  Each
callback passes in different arguments and expects different return
values.  Check the architecture chapter for a list of all the callbacks
that are available, their arguments, and return values.



Writing an entryparser
======================

Entry parsing functions take in a filename and the Request object.
They then open the file and parse it out.  The can call cb_preformat and 
cb_postformat as they see fit.  They should return a dict containing at 
least "title" and "story" keys.  The "title" should be a single string.  
The "story" should be a list of strings (with \n at the end).

Here's an example code that reads .plain files which have the title as 
the first line, metadata lines that start with # and then after all the 
metadata the body of the entry::

   import os

   def cb_entryparser(entryparsingdict):
       """
       Register self as plain file handler
       """
       entryparsingdict['plain'] = parse
       return entryparsingdict

   def parse(filename, request):
       """
       We just read everything off the file here, using the filename as
       title
       """
       entrydata = {}

       f = open(filename, "r")
       lines = f.readlines()
       f.close()

       # strip off the first line and use that as the title.
       title = lines.pop(0).strip()
       entrydata['title'] = title

       # absorb meta data lines which begin with a # and consist
       # of a name and a value
       while lines and lines[0].startswith("#"):
           meta = lines.pop(0)
           meta = meta[1:].strip()     # remove the hash
           meta = meta.split(" ", 1)
           entrydata[meta[0].strip()] = meta[1].strip()

       # join the rest of the lines as the story
       story = ''.join(lines)
       entrydata["story"] = "".join(lines)

       return entrydata



Writing a preformatter plugin
=============================

FIXME - need more about preformatters here

A typical preformatter plugin looks like this::

   def cb_preformat(args):
       if args['parser'] == 'linebreaks':
           return parse(''.join(args['story']))

   def parse(text):
       # A preformatter to convert linebreak to its HTML counterpart
       text = re.sub('\\n\\n+','</p><p>',text)
       text = re.sub('\\n','<br />',text)
       return '<p>%s</p>' % text