changelog

Python Development Environment (Python IDE plugin for Eclipse). Features editor, code completion, re

2001-12-24  Phil Dawes  <[e-mail]>

	* bike/parsing/addtokens.py: Reworked addtokens scheme to 'get all
	tokens for node, and then pass them to children to take their
	nodes'. This makes for a much easier time processing tokens.

2001-12-16  Phil Dawes  <[e-mail]>

	* bike/parsing/tokenhandler.py: Changed Getattr rename
	functionality to work when attribute token isn't second token of
	node tokenlist

	* bike/refactor/test_renameMethod.py: refactored renameMethod
	tests into generic abstract base class. These are now used by the
	ui package (rather than vice versa)

	* bike/ui/bikecli.py: Added code to tell you how many changes have
	been made, and advice on what might be wrong if this number is 0.

2001-12-15  Phil Dawes  <[e-mail]>

	* bike/parsing/tokenhandler.py: Added getNodeTokens()
	implementation to miss out any preceeding cruft not picked up by
	addtoken functionality on previous nodes.

2001-12-14  Phil Dawes  <[e-mail]>

	* bike/refactor/renameMethod.py: Changed renameMethodDefinition to
	work with nested methods (utilising the new typeinfo.fqn stuff).
	Removed findclass and findmethod.

2001-12-12  Phil Dawes  <[e-mail]>

	* bike/ui/bikecli.py: Modified prompt message a bit. Added actual
	standard io stuff so that it works with a user input as well as
	through the test harness.

	* bike/refactor/renameMethod.py: Modified RenameMethodReferences
	to create a stack of scopes. The code uses the typeinfo from each
	scope to determine the type of references, and if it can't find
	the type there, it checks its own references deduced from the
	classname passed in. I.e. it assumes that the classname is a
	class, even if it can't find the definition, and renames instances
	derived from it.

	Also ammended the 'prompt user callback' code to only ask the user
	if the type engine doesnt know the type of the instance
	expression. (before now it was prompting if the type was not the
	same as the input classname)

2001-12-11  Phil Dawes  <[e-mail]>

	* bike/ui/bikecli.py: Added code to handle prompting the user for
	whether the method reference should be renamed.

	* bike/refactor/renameMethod.py: Added code to prompt user when
	type of object instance isn't known

2001-12-08  Phil Dawes  <[e-mail]>

	* bike/refactor.py: now takes advantage of typeinfo stuff to
	deduce types of references
	
2001-12-03  Phil Dawes  <[e-mail]>

	* bike/parsing/addtypeinfo.py: New classes to deduce the types of
	references. Adds a typeinfo attribute to each scope ast node,
	which maps reference names to their types (currently as strings).

2001-11-26  Phil Dawes  <[e-mail]>

	* bike/refactor/renameMethod.py: Added stack to maintain
	references to object instances while visiting package structure

2001-11-25  Phil Dawes  <[e-mail]>

	* bike/refactor/renameMethod.py: Added support for 'from foo
	import *'
	
	* bike/ui/test_bikecli.py: Refactored tests so that all the single
	file tests are exercised in a package hierarchy context.

2001-11-24  Phil Dawes  <[e-mail]>

	* bike/refactor/renameMethod.py: Added support to rename methods
	based on imported classes

2001-11-23  Phil Dawes  <[e-mail]>

	* bike/refactor/renameMethod.py: Refactored code (took out
	RenameMethod class, since wasnt using instance state). Added
	support for compound references (a.b.c.foo)

2001-11-21  Phil Dawes  <[e-mail]>

	* bike/parsing/load_application.py: removed. (see load.py instead)

	* bike/refactor/renameMethod.py: Added support for fully qualified
	classnames in _findClass. e.g. a.b.bah.foo. This means that the ui
	now supports renaming package nested methods (but not their
	references).

2001-11-20  Phil Dawes  <[e-mail]>

	* bike/parsing: Added fixes to parser module to reflect changes in
	python 2.2 compiler module

2001-11-16  Phil Dawes  <[e-mail]>

	* bike/parsing/output.py: new function 'save' saves an entire ast
	tree

	* bike/refactor/renameMethod.py: Adapted to take an ast rather
	than source file

2001-11-14  Phil Dawes  <[e-mail]>

	* bike/parsing/load.py: Replacement for load_application.py using
	Juergen's pythius loader code. load() hierarchically loads a bunch
	of packages and modules into a big AST.

	* bike/parsing/extended_ast.py: Fleshed out Package and Top nodes
	to support hierarchical module ast


2001-11-08  Phil Dawes  <[e-mail]>

	* bike/parsing/load_application.py: Fixed import visitor code to
	handle comma seperated imports. (i.e. import foo,bar). Added code
	to recognise imports via a from clause (i.e. from foo import bar)

	Removed Imports.py. Integrated code into load_application
	

2001-11-06  Phil Dawes  <[e-mail]>

	* bike/parsing/load_application.py: Added load_application and
	Imports modules back into build. Started writing tests for them.

	
2001-11-05  Phil Dawes  <[e-mail]>

	* bike/parsing/brmtransformer.py : Rewrote this to override the
	methods which create nodes at the root of the parse. This means
	that a few methods adding parser nodelists to ast nodes should
	cater for practically all the AST nodes.

	* bike/parsing/tokenhandler.py (GetattrTokenHander.renameAttr):
	added renameAttr method which renames the attribute in the object
	state, and in the corresponding tokens.  This is a very brittle
	implementation, but will be fixed as new functionality is
	added. (it's the simplist thing right now)

	* bike/parsing/addtokens.py (AddTokens.visitGetattr): added
	token support for getattr

	* bike/parsing/tokenutils.py: Added getTokensPreceedingNodes,
	which takes a parsetree of nodes and returns all the tokens before
	the node represented by the parsetree. This is important because
	until there is an token-handling implementation of all the ast
	nodes, some tokens will be missed. By eating the tokens in between
	the implemented ast nodes, this ensures all the tokens will be
	outputted at the end.

	
2001-10-30  Phil Dawes  <[e-mail]>

	Created new CVS module, which makes use of distutils much more
	uniform.
	
	* setup.py: Created distutils setup stript: 

2001-10-25  Phil Dawes  <[e-mail]>

	* parsing/output.py: Code to handle spacing in inline comments

	* refactor/renameMethod.py: Started code to handle references when
	object is created in same scope


2001-10-24  Phil Dawes  <[e-mail]>

	* parsing/output.py: spaces in conditionals (a == b etc..)

	* parsing/tokenutils.py: Refactored code in TokenList to handle
	rearranging indent/dedent tokens so that whole-line comments are
	indented with the things they comment. The transformations are now
	done after the initial scan (slower, but easier to follow)
	
	
2001-10-23  Phil Dawes  <[e-mail]>

	* parsing/output.py: ensured a.b not seperated by spaces. 

	
2001-10-22  Phil Dawes  <[e-mail]>
	
	* parsing/output.py: Code to seperate a=b with spaces (a = b)

	* ui/test_cli.py: Merged cli tests with renameMethod tests to
	ensure all tests work through the cli

2001-10-19  Phil Dawes  <[e-mail]>

	* */testall.py: unified testing scripts so that all tests can be
	run from one place
	
2001-10-15  Phil Dawes  <[e-mail]>

	* ui/cli.py: Added simple command line interface.

2001-10-14  Phil Dawes  <[e-mail]>

	* parsing/output.py: Finished simple ast->string functionality
	
2001-10-13  Phil Dawes  <[e-mail]>

	* testall.py: added master test script

	* parsing/tokenhandler.py (FunctionTokenHandler.rename): added
	rename a function and keep tokens in sync functionality


2001-10-12  Phil Dawes  <[e-mail]>

	* parsing/output.py: started tokens -> string functionality

2001-10-07  Phil Dawes  <[e-mail]>
	* refactor/renameMethod.py: very basic findClass, findMethod
	functionality. Needs refactoring into analysis package, and
	merging with existing code
	
2001-10-05  Phil Dawes  <[e-mail]>

	* refactor/simplediff.py: wrote a simple diff utility to support
	writing top down tests. (i.e. do a renameMethod, then assert that
	the diff is correct)

	
2001-10-04  Phil Dawes  <[e-mail]>

	* refactor/renameMethod.py: Got fedup with trying to secondguess
	what the parser module should do. Decided to do some top-down
	proper XP style development. Started with implementing the
	renameMethod story.

2001-10-01  Phil Dawes  <[e-mail]>

	* parsing/test_addtokens.py: refactored tests so that they use
	brmtransformer to get a node, rather than mocking it.

2001-09-29  Phil Dawes  <[e-mail]>

	* parsing/brmtransformer.py: Added nodelist to pass_stmt
	
2001-09-28  Phil Dawes  <[e-mail]>

	* parsing/addtokens.py (AddTokens.visitModule): finished method

	* parsing/tokenutils.py
	(TokenConverter.getTokensUptoLastNewlineBeforeCode): method added

2001-09-27  Phil Dawes  <[e-mail]>
	
	* parsing/addtokens.py: started visitmodule() impl 

	* parsing/extended_ast.py: Added tokens support to Module

2001-09-26  Phil Dawes  <[e-mail]>

	* parsing/test_addtokens.py Refactored tests. Made up better names.

2001-09-25  Phil Dawes  <[e-mail]>

	Laptop broke. Continuing with an older version of the code (hence
	the gap in changelog)

	* parsing/tokenutils.py (TokenConverter.getTokensToColon): Added
	function which just gobbles tokens to the next colon. Refactored
	getTokensStartBlock() to use this instead of walking the nodelist,
	since I think all 

	* parsing/addtokens.py (AddTokens.visitFunction): Now that
	function and class no longer need to pass in partial nodelists,
	the code is common. Refactored common code into
	visitCompoundNode().

	(hence the gap in the changelog)
	
	* parsing/tokenhandler.py: removed the preceeding tokens code -
	you ain't gonna need it!

	* parsing/test_addtokens.py: changed visitclass tests to use
	mockobjects

2001-09-10  Phil Dawes  <[e-mail]>

	* parsing/tokenhandler.py: 

	* parsing/*: Integrated code with bike parsing package

	Split tokenhandler base classes into seperate module
	Merged brmast stuff into extended_ast package
	
	
2001-09-08  Phil Dawes  <[e-mail]>

	* test_tokenutils.py: Added more unit tests 

2001-09-06  Phil Dawes  <[e-mail]>

	* test_tokenutils.py: Added more unit tests 


2001-09-04  Phil Dawes  <[e-mail]>

	* test_addtokens.py: Added some mock objects and tests
	
2001-08-31  Phil Dawes  <[e-mail]>

	* test_tokenutils.py: beefed up the unit tests a bit

2001-08-29  Phil Dawes  <[e-mail]>

	* brmtransformer.py: This module now just contains code to add the
	nodelist member to each ast node as the compiler generates
	it. This is so the code in addtokens has access to the original
	python parser nodes to generate tokens from.

	* addtokens.py: Moved the code to add the tokens to the ast nodes
	to this module. It now relies on there being a 'nodelist' member
	on each ast object, containing the original python parser
	nodelist.

	* brmast.py: Added hack to retrofit TokenHandler base class to
	existing compiler.ast classes. This means that we no longer have
	to include compiler classes with the distribution. 
	Not sure if this is completely legal - doesn't work with jython.

	Removed stock Python compiler classes from src tree.
	

2001-08-27  Phil Dawes  <[e-mail]>
	Set about reimplementing the token stuff - will now add tokens to
	ast objects *after* the nodes have been processed by the compiler
	module.
	
2001-08-22  Phil Dawes  <[e-mail]>

	* tokenutils.py: Added support for tokenising strings as well as
	files.

	* test_tokenutils.py: removed testdata directory, and moved all
	the testdata into this module, now that the tokenutils stuff can
	take strings as well as files

	* brmtransformer.py: Renamed TransformerWithTokens to Transformer,
	now that it's in a seperate module.

	* brmast.py: Refactored the added token code out of ast module
	into this one. Created a seperate mixin called TokenHandler rather
	than sticking this code in the Node base class. Each class now
	re-inherits this baseclass.
	Unfortunately this means that I had to modify the transformer.py
	file to import brmast rather than ast. Need to think of a better
	way of doing this.

2001-08-21  Phil Dawes  <[e-mail]>

	* brmtransformer.py: Created new brmtransformer module and
	refactored token code from transformer.Transformer into seperate
	class TransformerWithTokens.  Refactored common code from