Tagging the gentoolkit-0.2.4 releasegentoolkit-0.2.4

svn path=/tags/gentoolkit-0.2.4/; revision=564

9 files changed, 1067 insertions, 0 deletions

diff --git a/src/eclean/AUTHORS b/src/eclean/AUTHORS
new file mode 100644
index 0000000..9263cbb
--- /dev/null
+++ b/src/eclean/AUTHORS
@@ -0,0 +1 @@
+Thomas de Grenier de Latour (tgl) <degrenier@easyconnect.fr>
diff --git a/src/eclean/ChangeLog b/src/eclean/ChangeLog
new file mode 100644
index 0000000..36d9a28
--- /dev/null
+++ b/src/eclean/ChangeLog
@@ -0,0 +1,27 @@
+2005-12-19 Paul Varner <fuzzyray@gentoo.org>
+	* Add support for reqular expression matching for file names in the
+	  exclude files.
+
+2005-08-28 Thomas de Grenier de Latour (tgl) <degrenier@easyconnect.fr>
+	* Version 0.4.1
+	* added support for some "eclean-dist" and "eclean-pkg" symlinks on eclean
+	  (and thus refactored command-line parsing and help screen code)
+	* accept file names in exclude files for specific distfiles protection
+	  (useful to protect the OOo i18n files for instance, which are not in
+	  $SRC_URI but put there manually)
+	* minor rewrite of some findDistfiles() code
+	* added /usr/lib/portage/pym python path, just to be sure it comes first
+	  (after all, "ouput" is a pretty generic name for a python module...)
+	* updated manpage
+
+2005-08-27 Thomas de Grenier de Latour (tgl) <degrenier@easyconnect.fr>
+	* Version 0.4
+	* added exclusion files support
+	* added time limit option
+	* added size limit option (for distfiles only)
+	* added fetch-restricted distfile optionnal protection
+	* added --package-names option for protection of all versions of installed
+	  packages.
+	* removed support of multiple actions on command-line. That would have been
+	  hell with action-specific options.
+	* updated manpage
diff --git a/src/eclean/Makefile b/src/eclean/Makefile
new file mode 100644
index 0000000..79c5895
--- /dev/null
+++ b/src/eclean/Makefile
@@ -0,0 +1,24 @@
+# Copyright 2004 Karl Trygve Kalleberg <karltk@gentoo.org>
+# Copyright 2004 Gentoo Technologies, Inc.
+# Distributed under the terms of the GNU General Public License v2
+#
+# $Header$
+
+include ../../makedefs.mak
+
+all:
+
+dist:
+	mkdir -p ../../$(distdir)/src/eclean
+	cp eclean eclean.1 Makefile *.exclude ../../$(distdir)/src/eclean
+	cp AUTHORS THANKS TODO ChangeLog ../../$(distdir)/src/eclean
+
+install:
+	install -m 0755 eclean $(bindir)/
+	ln -sf eclean $(bindir)/eclean-pkg
+	ln -sf eclean $(bindir)/eclean-dist
+	install -d $(sysconfdir)/eclean
+	install -m 0644 distfiles.exclude packages.exclude $(sysconfdir)/eclean/
+	install -d $(docdir)/eclean
+	install -m 0644 AUTHORS THANKS TODO ChangeLog $(docdir)/eclean/
+	install -m 0644 eclean.1 $(mandir)/
diff --git a/src/eclean/THANKS b/src/eclean/THANKS
new file mode 100644
index 0000000..6b8dc2e
--- /dev/null
+++ b/src/eclean/THANKS
@@ -0,0 +1,7 @@
+The starting point ideas were found here:
+http://forums.gentoo.org/viewtopic.php?t=3011
+
+Thanks to eswanson and far for their contributions, and to wolf31o2 for his
+support.  Thanks also to karltk, some of this code was at some point inspired
+by his "equery" tool.  And thanks to people who had a look on bug #33877: 
+Benjamin Braatz, fuzzyray, genone, etc.
diff --git a/src/eclean/TODO b/src/eclean/TODO
new file mode 100644
index 0000000..04e64ca
--- /dev/null
+++ b/src/eclean/TODO
@@ -0,0 +1,16 @@
+- exclusion file syntax could be improved (maybe it should support real
+   dep-atoms, or wildcards, etc.)
+
+- some policy to keep the X latest versions of a package (in each of its 
+   SLOT maybe) would be really cool...
+
+- add an option to protect system binary packages
+   => yup, but later... (needs some portage modifications to be done right)
+
+- add actions for PORT_LOGDIR and/or /var/tmp/portage cleaning?
+   => bah, don't know... imho tmpreaper or find+rm onliners are enough here
+
+- cleanup of DISTDIR/cvs-src when action=distfiles
+   => i never use cvs ebuilds, i should check what it does exactly
+
+- rewrite for a decent Portage API if there ever is one
diff --git a/src/eclean/distfiles.exclude b/src/eclean/distfiles.exclude
new file mode 100644
index 0000000..a31be55
--- /dev/null
+++ b/src/eclean/distfiles.exclude
@@ -0,0 +1,5 @@
+# /etc/eclean/distfiles.exclude
+# In this file you can list some categories or cat/pkg-name for which you want
+# to protect distfiles from "ecleaning". You can also name some specific files.
+# See `man eclean` for syntax details.
+metadata.dtd
diff --git a/src/eclean/eclean b/src/eclean/eclean
new file mode 100644
index 0000000..4fdec10
--- /dev/null
+++ b/src/eclean/eclean
@@ -0,0 +1,807 @@
+#!/usr/bin/env python
+# Copyright 2003-2005 Gentoo Foundation
+# Distributed under the terms of the GNU General Public License v2
+# $Header: $
+
+
+###############################################################################
+# Meta:
+__author__ = "Thomas de Grenier de Latour (tgl)"
+__email__ = "degrenier@easyconnect.fr"
+__version__ = "0.4.1"
+__productname__ = "eclean"
+__description__ = "A cleaning tool for Gentoo distfiles and binaries."
+
+
+###############################################################################
+# Python imports:
+import sys
+import os, stat
+import re
+import time
+import getopt
+import fpformat
+import signal
+try:
+	import portage
+except ImportError:
+	sys.path.insert(0, "/usr/lib/portage/pym")
+	import portage
+try:
+	from portage.output import *
+except ImportError:
+	from output import *
+
+listdir = portage.listdir
+
+###############################################################################
+# Misc. shortcuts to some portage stuff:
+port_settings = portage.settings
+distdir = port_settings["DISTDIR"]
+pkgdir = port_settings["PKGDIR"]
+
+
+###############################################################################
+# printVersion:
+def printVersion():
+	print "%s (version %s) - %s" \
+			% (__productname__, __version__, __description__)
+	print "Author: %s <%s>" % (__author__,__email__)
+	print "Copyright 2003-2005 Gentoo Foundation"
+	print "Distributed under the terms of the GNU General Public License v2"
+
+
+###############################################################################
+# printUsage: print help message. May also print partial help to stderr if an
+# error from {'options','actions'} is specified.
+def printUsage(error=None,help=None):
+	out = sys.stdout
+	if error: out = sys.stderr
+	if not error in ('actions', 'global-options', \
+			'packages-options', 'distfiles-options', \
+			'merged-packages-options', 'merged-distfiles-options', \
+			'time', 'size'):
+		error = None
+	if not error and not help: help = 'all'
+	if error == 'time':
+		eerror("Wrong time specification")
+		print >>out, "Time specification should be an integer followed by a"+ \
+				" single letter unit."
+		print >>out, "Available units are: y (years), m (months), w (weeks), "+ \
+				"d (days) and h (hours)."
+		print >>out, "For instance: \"1y\" is \"one year\", \"2w\" is \"two"+ \
+				" weeks\", etc. "
+		return
+	if error == 'size':
+		eerror("Wrong size specification")
+		print >>out, "Size specification should be an integer followed by a"+ \
+				" single letter unit."
+		print >>out, "Available units are: G, M, K and B."
+		print >>out, "For instance: \"10M\" is \"ten megabytes\", \"200K\" "+ \
+				"is \"two hundreds kilobytes\", etc."
+		return
+	if error in ('global-options', 'packages-options', 'distfiles-options', \
+			'merged-packages-options', 'merged-distfiles-options',):
+		eerror("Wrong option on command line.")
+		print >>out
+	elif error == 'actions':
+		eerror("Wrong or missing action name on command line.")
+		print >>out
+	print >>out, white("Usage:")
+	if error in ('actions','global-options', 'packages-options', \
+	'distfiles-options') or help == 'all':
+		print >>out, " "+turquoise(__productname__), \
+			yellow("[global-option] ..."), \
+			green("<action>"), \
+			yellow("[action-option] ...")
+	if error == 'merged-distfiles-options' or help in ('all','distfiles'):
+		print >>out, " "+turquoise(__productname__+'-dist'), \
+			yellow("[global-option, distfiles-option] ...")
+	if error == 'merged-packages-options' or help in ('all','packages'):
+		print >>out, " "+turquoise(__productname__+'-pkg'), \
+			yellow("[global-option, packages-option] ...")
+	if error in ('global-options', 'actions'):
+		print >>out, " "+turquoise(__productname__), \
+			yellow("[--help, --version]")
+	if help == 'all':
+		print >>out, " "+turquoise(__productname__+"(-dist,-pkg)"), \
+			yellow("[--help, --version]")
+	if error == 'merged-packages-options' or help == 'packages':
+		print >>out, " "+turquoise(__productname__+'-pkg'), \
+			yellow("[--help, --version]")
+	if error == 'merged-distfiles-options' or help == 'distfiles':
+		print >>out, " "+turquoise(__productname__+'-dist'), \
+			yellow("[--help, --version]")
+	print >>out
+	if error in ('global-options', 'merged-packages-options', \
+	'merged-distfiles-options') or help:
+		print >>out, "Available global", yellow("options")+":"
+		print >>out, yellow(" -C, --nocolor")+ \
+			"             - turn off colors on output"
+		print >>out, yellow(" -d, --destructive")+ \
+			"         - only keep the minimum for a reinstallation"
+		print >>out, yellow(" -e, --exclude-file=<path>")+ \
+			" - path to the exclusion file"
+		print >>out, yellow(" -i, --interactive")+ \
+			"         - ask confirmation before deletions"
+		print >>out, yellow(" -n, --package-names")+ \
+			"       - protect all versions (when --destructive)"
+		print >>out, yellow(" -p, --pretend")+ \
+			"             - only display what would be cleaned"
+		print >>out, yellow(" -q, --quiet")+ \
+			"               - be as quiet as possible"
+		print >>out, yellow(" -t, --time-limit=<time>")+ \
+			"   - don't delete files modified since "+yellow("<time>")
+		print >>out, "   "+yellow("<time>"), "is a duration: \"1y\" is"+ \
+				" \"one year\", \"2w\" is \"two weeks\", etc. "
+		print >>out, "   "+"Units are: y (years), m (months), w (weeks), "+ \
+				"d (days) and h (hours)."
+		print >>out, yellow(" -h, --help")+ \
+			"                - display the help screen"
+		print >>out, yellow(" -V, --version")+ \
+			"             - display version info"
+		print >>out
+	if error == 'actions' or help == 'all':
+		print >>out, "Available", green("actions")+":"
+		print >>out, green(" packages")+ \
+			"      - clean outdated binary packages from:"
+		print >>out, "                  ",teal(pkgdir)
+		print >>out, green(" distfiles")+ \
+			"     - clean outdated packages sources files from:"
+		print >>out, "                  ",teal(distdir)
+		print >>out
+	if error in ('packages-options','merged-packages-options') \
+	or help in ('all','packages'):
+		print >>out, "Available", yellow("options"),"for the", \
+				green("packages"),"action:"
+		print >>out, yellow(" NONE  :)")
+		print >>out
+	if error in ('distfiles-options', 'merged-distfiles-options') \
+	or help in ('all','distfiles'):
+		print >>out, "Available", yellow("options"),"for the", \
+				green("distfiles"),"action:"
+		print >>out, yellow(" -f, --fetch-restricted")+ \
+			"   - protect fetch-restricted files (when --destructive)"
+		print >>out, yellow(" -s, --size-limit=<size>")+ \
+			"  - don't delete distfiles bigger than "+yellow("<size>")
+		print >>out, "   "+yellow("<size>"), "is a size specification: "+ \
+				"\"10M\" is \"ten megabytes\", \"200K\" is"
+		print >>out, "   "+"\"two hundreds kilobytes\", etc.  Units are: "+ \
+				"G, M, K and B."
+		print >>out
+	print >>out, "More detailed instruction can be found in", \
+			turquoise("`man %s`" % __productname__)
+
+
+###############################################################################
+# einfo: display an info message depending on a color mode
+def einfo(message="", nocolor=False):
+	if not nocolor: prefix = " "+green('*')
+	else: prefix = ">>>"
+	print prefix,message
+
+
+###############################################################################
+# eerror: display an error depending on a color mode
+def eerror(message="", nocolor=False):
+	if not nocolor: prefix = " "+red('*')
+	else: prefix = "!!!"
+	print >>sys.stderr,prefix,message
+
+
+###############################################################################
+# eprompt: display a user question depending on a color mode.
+def eprompt(message, nocolor=False):
+	if not nocolor: prefix = " "+red('>')+" "
+	else: prefix = "??? "
+	sys.stdout.write(prefix+message)
+	sys.stdout.flush()
+
+
+###############################################################################
+# prettySize: integer -> byte/kilo/mega/giga converter. Optionnally justify the
+# result. Output is a string.
+def prettySize(size,justify=False):
+	units = [" G"," M"," K"," B"]
+	approx = 0
+	while len(units) and size >= 1000:
+		approx = 1
+		size = size / 1024.
+		units.pop()
+	sizestr = fpformat.fix(size,approx)+units[-1]
+	if justify: 
+		sizestr = " " + blue("[ ") + " "*(7-len(sizestr)) \
+		          + green(sizestr) + blue(" ]")
+	return sizestr
+
+
+###############################################################################
+# yesNoAllPrompt: print a prompt until user answer in yes/no/all. Return a 
+# boolean for answer, and also may affect the 'accept_all' option.
+# Note: i gave up with getch-like functions, to much bugs in case of escape
+# sequences. Back to raw_input.
+def yesNoAllPrompt(myoptions,message="Do you want to proceed?"):
+	user_string="xxx"
+	while not user_string.lower() in ["","y","n","a","yes","no","all"]:
+		eprompt(message+" [Y/n/a]: ", myoptions['nocolor'])
+		user_string = raw_input()
+	if user_string.lower() in ["a","all"]:
+		myoptions['accept_all'] = True
+	myanswer = user_string.lower() in ["","y","a","yes","all"]
+	return myanswer
+
+
+###############################################################################
+# ParseArgsException: for parseArgs() -> main() communication
+class ParseArgsException(Exception):
+	def __init__(self, value):
+		self.value = value
+	def __str__(self):
+		return repr(self.value)
+
+
+###############################################################################
+# parseSize: convert a file size "Xu" ("X" is an integer, and "u" in [G,M,K,B])
+# into an integer (file size in Bytes). Raises ParseArgsException('size') in
+# case of failure.
+def parseSize(size):
+	myunits = { \
+		'G': (1024**3), \
+		'M': (1024**2), \
+		'K': 1024, \
+		'B': 1 \
+	}
+	try:
+		mymatch = re.match(r"^(?P<value>\d+)(?P<unit>[GMKBgmkb])?$",size)
+		mysize = int(mymatch.group('value'))
+		if mymatch.group('unit'):
+			mysize *= myunits[mymatch.group('unit').capitalize()]
+	except:
+		raise ParseArgsException('size')
+	return mysize
+
+
+###############################################################################
+# parseTime: convert a duration "Xu" ("X" is an int, and "u" a time unit in 
+# [Y,M,W,D,H]) into an integer which is a past EPOCH date. 
+# Raises ParseArgsException('time') in case of failure.
+# (yep, big approximations inside... who cares?)
+def parseTime(timespec):
+	myunits = {'H' : (60 * 60)}
+	myunits['D'] = myunits['H'] * 24
+	myunits['W'] = myunits['D'] * 7
+	myunits['M'] = myunits['D'] * 30
+	myunits['Y'] = myunits['D'] * 365
+	try:
+		# parse the time specification
+		mymatch = re.match(r"^(?P<value>\d+)(?P<unit>[YMWDHymwdh])?$",timespec)
+		myvalue = int(mymatch.group('value'))
+		if not mymatch.group('unit'): myunit = 'D'
+		else: myunit = mymatch.group('unit').capitalize()
+	except: raise ParseArgsException('time')
+	# calculate the limit EPOCH date
+	mytime = time.time() - (myvalue * myunits[myunit])
+	return mytime
+
+
+###############################################################################
+# parseCmdLine: parse the command line arguments. Raise exceptions on errors or
+# non-action modes (help/version). Returns an action, and affect the options
+# dict.
+def parseArgs(myoptions={}):
+
+	# local function for interpreting command line options 
+	# and setting myoptions accordingly
+	def optionSwitch(myoption,opts,action=None):
+		return_code = True
+		for o, a in opts:
+			if o in ("-h", "--help"):
+				if action: raise ParseArgsException('help-'+action)
+				else: raise ParseArgsException('help')
+			elif o in ("-V", "--version"):
+				raise ParseArgsException('version')
+			elif o in ("-C", "--nocolor"):
+				myoptions['nocolor'] = True
+				nocolor()
+			elif o in ("-d", "--destructive"):
+				myoptions['destructive'] = True
+			elif o in ("-i", "--interactive") and not myoptions['pretend']:
+				myoptions['interactive'] = True
+			elif o in ("-p", "--pretend"):
+				myoptions['pretend'] = True
+				myoptions['interactive'] = False
+			elif o in ("-q", "--quiet"):
+				myoptions['quiet'] = True
+			elif o in ("-t", "--time-limit"):
+				myoptions['time-limit'] = parseTime(a)
+			elif o in ("-e", "--exclude-file"):
+				myoptions['exclude-file'] = a
+			elif o in ("-n", "--package-names"):
+				myoptions['package-names'] = True
+			elif o in ("-f", "--fetch-restricted"):
+				myoptions['fetch-restricted'] = True
+			elif o in ("-s", "--size-limit"):
+				myoptions['size-limit'] = parseSize(a)
+			else: return_code = False
+		# sanity check of --destructive only options:
+		for myopt in ('fetch-restricted', 'package-names'):
+			if (not myoptions['destructive']) and myoptions[myopt]:
+				if not myoptions['quiet']:	
+					eerror("--%s only makes sense in --destructive mode." \
+							% myopt, myoptions['nocolor'])
+				myoptions[myopt] = False
+		return return_code
+
+	# here are the different allowed command line options (getopt args)
+	getopt_options = {'short':{}, 'long':{}}
+	getopt_options['short']['global'] = "Cdipqe:t:nhV"
+	getopt_options['long']['global'] = ["nocolor", "destructive", \
+			"interactive", "pretend", "quiet", "exclude-file=", "time-limit=", \
+			"package-names", "help", "version"]
+	getopt_options['short']['distfiles'] = "fs:"
+	getopt_options['long']['distfiles'] = ["fetch-restricted", "size-limit="]
+	getopt_options['short']['packages'] = ""
+	getopt_options['long']['packages'] = [""]
+	# set default options, except 'nocolor', which is set in main()
+	myoptions['interactive'] = False
+	myoptions['pretend'] = False
+	myoptions['quiet'] = False
+	myoptions['accept_all'] = False
+	myoptions['destructive'] = False
+	myoptions['time-limit'] = 0
+	myoptions['package-names'] = False
+	myoptions['fetch-restricted'] = False
+	myoptions['size-limit'] = 0
+	# if called by a well-named symlink, set the acction accordingly:
+	myaction = None
+	if os.path.basename(sys.argv[0]) in \
+			(__productname__+'-pkg', __productname__+'-packages'):
+		myaction = 'packages'
+	elif os.path.basename(sys.argv[0]) in \
+			(__productname__+'-dist', __productname__+'-distfiles'):
+		myaction = 'distfiles'
+	# prepare for the first getopt
+	if myaction:
+		short_opts = getopt_options['short']['global'] \
+				+ getopt_options['short'][myaction]
+		long_opts = getopt_options['long']['global'] \
+				+ getopt_options['long'][myaction]
+		opts_mode = 'merged-'+myaction
+	else:
+		short_opts = getopt_options['short']['global']
+		long_opts = getopt_options['long']['global']
+		opts_mode = 'global'
+	# apply getopts to command line, show partial help on failure
+	try: opts, args = getopt.getopt(sys.argv[1:], short_opts, long_opts)
+	except: raise ParseArgsException(opts_mode+'-options')
+	# set myoptions accordingly
+	optionSwitch(myoptions,opts,action=myaction)
+	# if action was already set, there should be no more args
+	if myaction and len(args): raise ParseArgsException(opts_mode+'-options')
+	# if action was set, there is nothing left to do
+	if myaction: return myaction
+	# So, we are in "eclean --foo action --bar" mode. Parse remaining args...
+	# Only two actions are allowed: 'packages' and 'distfiles'.
+	if not len(args) or not args[0] in ('packages','distfiles'):
+		raise ParseArgsException('actions')
+	myaction = args.pop(0)
+	# parse the action specific options
+	try: opts, args = getopt.getopt(args, \
+			getopt_options['short'][myaction], \
+			getopt_options['long'][myaction])
+	except: raise ParseArgsException(myaction+'-options')
+	# set myoptions again, for action-specific options
+	optionSwitch(myoptions,opts,action=myaction)
+	# any remaning args? Then die!
+	if len(args): raise ParseArgsException(myaction+'-options')
+	# returns the action. Options dictionary is modified by side-effect.
+	return myaction
+
+###############################################################################
+# isValidCP: check wether a string is a valid cat/pkg-name
+# This is for 2.0.51 vs. CVS HEAD compatibility, i've not found any function
+# for that which would exists in both. Weird...
+def isValidCP(cp):
+	if not '/' in cp: return False
+	try: portage.cpv_getkey(cp+"-0")
+	except: return False
+	else: return True
+
+
+###############################################################################
+# ParseExcludeFileException: for parseExcludeFile() -> main() communication
+class ParseExcludeFileException(Exception):
+	def __init__(self, value):
+		self.value = value
+	def __str__(self):
+		return repr(self.value)
+
+
+###############################################################################
+# parseExcludeFile: parses an exclusion file, returns an exclusion dictionnary
+# Raises ParseExcludeFileException in case of fatal error.
+def parseExcludeFile(filepath):
+	excl_dict = { \
+			'categories':{}, \
+			'packages':{}, \
+			'anti-packages':{}, \
+			'garbage':{} }
+	try: file = open(filepath,"r")
+	except IOError:
+		raise ParseExcludeFileException("Could not open exclusion file.")
+	filecontents = file.readlines()
+	file.close()
+	cat_re = re.compile('^(?P<cat>[a-zA-Z0-9]+-[a-zA-Z0-9]+)(/\*)?$')
+	cp_re = re.compile('^(?P<cp>[-a-zA-Z0-9_]+/[-a-zA-Z0-9_]+)$')	
+	for line in filecontents:
+		line = line.strip()
+		if not len(line): continue
+		if line[0] == '#': continue
+		try: mycat = cat_re.match(line).group('cat')
+		except: pass
+		else: 
+			if not mycat in portage.settings.categories:
+				raise ParseExcludeFileException("Invalid category: "+mycat)
+			excl_dict['categories'][mycat] = None
+			continue
+		dict_key = 'packages'
+		if line[0] == '!':
+			dict_key = 'anti-packages'
+			line = line[1:]
+		try:
+			mycp = cp_re.match(line).group('cp')
+			if isValidCP(mycp):
+				excl_dict[dict_key][mycp] = None
+				continue
+			else: raise ParseExcludeFileException("Invalid cat/pkg: "+mycp)
+		except: pass
+		#raise ParseExcludeFileException("Invalid line: "+line)
+		try:
+			excl_dict['garbage'][line] = re.compile(line)
+		except:
+			try:
+				excl_dict['garbage'][line] = re.compile(re.escape(line))
+			except:
+				raise ParseExcludeFileException("Invalid file name/regular expression: "+line)
+	return excl_dict
+
+
+###############################################################################
+# exclDictExpand: returns a dictionary of all CP from porttree which match
+# the exclusion dictionary
+def exclDictExpand(excl_dict):
+	mydict = {}
+	if 'categories' in excl_dict:
+		# XXX: i smell an access to something which is really out of API...
+		for mytree in portage.portdb.porttrees:
+			for mycat in excl_dict['categories']:
+				for mypkg in listdir(os.path.join(mytree,mycat),ignorecvs=1):
+					mydict[mycat+'/'+mypkg] = None
+	if 'packages' in excl_dict:
+		for mycp in excl_dict['packages']:
+			mydict[mycp] = None
+	if 'anti-packages' in excl_dict:
+		for mycp in excl_dict['anti-packages']:
+			if mycp in mydict:
+				del mydict[mycp]
+	return mydict
+
+
+###############################################################################
+# exclDictMatch: checks whether a CP matches the exclusion rules
+def exclDictMatch(excl_dict,pkg):
+	if 'anti-packages' in excl_dict \
+	   and pkg in excl_dict['anti-packages']:
+		return False
+	if 'packages' in excl_dict \
+	   and pkg in excl_dict['packages']:
+		return True
+	mycat = pkg.split('/')[0]
+	if 'categories' in excl_dict \
+	   and mycat in excl_dict['categories']:
+		return True
+	return False
+
+
+###############################################################################
+# findDistfiles: find all obsolete distfiles.
+# XXX: what about cvs ebuilds? i should install some to see where it goes...
+def findDistfiles( \
+		exclude_dict={}, \
+		destructive=False,\
+		fetch_restricted=False, \
+		package_names=False, \
+		time_limit=0, \
+		size_limit=0):
+	# this regexp extracts files names from SRC_URI. It is not very precise,
+	# but we don't care (may return empty strings, etc.), since it is fast.
+	file_regexp = re.compile('([a-zA-Z0-9_,\.\-\+\~]*)[\s\)]')
+	clean_dict = {}
+	keep = []
+	pkg_dict = {}
+
+	# create a big CPV->SRC_URI dict of packages whose distfiles should be kept
+	if (not destructive) or fetch_restricted:
+		# list all CPV from portree (yeah, that takes time...)
+		for package in portage.portdb.cp_all():
+			for my_cpv in portage.portdb.cp_list(package):
+				# get SRC_URI and RESTRICT from aux_get
+				try: (src_uri,restrict) = \
+					portage.portdb.aux_get(my_cpv,["SRC_URI","RESTRICT"])
+				except KeyError: continue
+				# keep either all or fetch-restricted only
+				if (not destructive) or ('fetch' in restrict):
+					pkg_dict[my_cpv] = src_uri
+	if destructive:
+		if not package_names:
+			# list all CPV from vartree
+			pkg_list = portage.db[portage.root]["vartree"].dbapi.cpv_all()
+		else:
+			# list all CPV from portree for CP in vartree
+			pkg_list = []
+			for package in portage.db[portage.root]["vartree"].dbapi.cp_all():
+				pkg_list += portage.portdb.cp_list(package)
+		for my_cp in exclDictExpand(exclude_dict):
+			# add packages from the exclude file
+			pkg_list += portage.portdb.cp_list(my_cp)
+		for my_cpv in pkg_list:
+			# skip non-existing CPV (avoids ugly aux_get messages)
+			if not portage.portdb.cpv_exists(my_cpv): continue
+			# get SRC_URI from aux_get
+			try: pkg_dict[my_cpv] = \
+					portage.portdb.aux_get(my_cpv,["SRC_URI"])[0]
+			except KeyError: continue
+		del pkg_list
+
+	# create a dictionary of files which should be deleted 
+	for file in os.listdir(distdir):
+		filepath = os.path.join(distdir, file)
+		try: file_stat = os.stat(filepath)
+		except: continue
+		if not stat.S_ISREG(file_stat[stat.ST_MODE]): continue
+		if size_limit and (file_stat[stat.ST_SIZE] >= size_limit):
+			continue
+		if time_limit and (file_stat[stat.ST_MTIME] >= time_limit):
+			continue
+		if 'garbage' in exclude_dict:
+			# Try to match file name directly
+			if file in exclude_dict['garbage']:
+				file_match = True
+			# See if file matches via regular expression matching
+			else:
+				file_match = False
+				for file_entry in exclude_dict['garbage']:
+					if exclude_dict['garbage'][file_entry].match(file):
+						file_match = True
+						break
+
+			if file_match:
+				continue
+		# this is a candidate for cleaning
+		clean_dict[file]=[filepath]
+	# remove files owned by some protected packages
+	for my_cpv in pkg_dict:
+		for file in file_regexp.findall(pkg_dict[my_cpv]+"\n"):
+			if file in clean_dict:
+				del clean_dict[file]
+		# no need to waste IO time if there is nothing left to clean
+		if not len(clean_dict): return clean_dict
+	return clean_dict
+
+
+###############################################################################
+# findPackages: find all obsolete binary packages.
+# XXX: packages are found only by symlinks. Maybe i should also return .tbz2
+#      files from All/ that have no corresponding symlinks.
+def findPackages( \
+		exclude_dict={}, \
+		destructive=False, \
+		time_limit=0, \
+		package_names=False):
+	clean_dict = {}
+	# create a full package dictionnary
+	for root, dirs, files in os.walk(pkgdir):
+		if root[-3:] == 'All': continue
+		for file in files:
+			if not file[-5:] == ".tbz2":
+				# ignore non-tbz2 files
+				continue
+			path = os.path.join(root, file)
+			category = os.path.split(root)[-1]
+			cpv = category+"/"+file[:-5]
+			mystat = os.lstat(path)
+			if time_limit and (mystat[stat.ST_MTIME] >= time_limit):
+				# time-limit exclusion
+				continue
+			# dict is cpv->[files] (2 files in general, because of symlink)
+			clean_dict[cpv] = [path]
+			#if os.path.islink(path):
+			if stat.S_ISLNK(mystat[stat.ST_MODE]):
+				clean_dict[cpv].append(os.path.realpath(path))
+	# keep only obsolete ones
+	if destructive:
+		mydbapi = portage.db[portage.root]["vartree"].dbapi
+		if package_names: cp_all = dict.fromkeys(mydbapi.cp_all())
+		else: cp_all = {}
+	else:
+		mydbapi = portage.db[portage.root]["porttree"].dbapi
+		cp_all = {}
+	for mycpv in clean_dict.keys():
+		if exclDictMatch(exclude_dict,portage.cpv_getkey(mycpv)):
+			# exclusion because of the exclude file
+			del clean_dict[mycpv]
+			continue
+		if mydbapi.cpv_exists(mycpv):
+			# exclusion because pkg still exists (in porttree or vartree)
+			del clean_dict[mycpv]
+			continue
+		if portage.cpv_getkey(mycpv) in cp_all:
+			# exlusion because of --package-names 
+			del clean_dict[mycpv]
+
+	return clean_dict
+
+
+###############################################################################
+# doCleanup: takes a dictionnary {'display name':[list of files]}. Calculate
+# size of each entry for display, prompt user if needed, delete files if needed
+# and return the total size of files that [have been / would be] deleted.
+def doCleanup(clean_dict,action,myoptions):
+	# define vocabulary of this action
+	if action == 'distfiles': file_type = 'file'
+	else: file_type = 'binary package'
+	# sorting helps reading
+	clean_keys = clean_dict.keys()
+	clean_keys.sort()
+	clean_size = 0
+	# clean all entries one by one
+	for mykey in clean_keys:
+		key_size = 0
+		for file in clean_dict[mykey]:
+			# get total size for an entry (may be several files, and
+			# symlinks count zero)
+			if os.path.islink(file): continue
+			try: key_size += os.path.getsize(file)
+			except: eerror("Could not read size of "+file, \
+			               myoptions['nocolor'])
+		if not myoptions['quiet']:
+			# pretty print mode
+			print prettySize(key_size,True),teal(mykey)
+		elif myoptions['pretend'] or myoptions['interactive']:
+			# file list mode
+			for file in clean_dict[mykey]: print file
+		#else: actually delete stuff, but don't print anything
+		if myoptions['pretend']: clean_size += key_size
+		elif not myoptions['interactive'] \
+		     or myoptions['accept_all'] \
+		     or yesNoAllPrompt(myoptions, \
+		                       "Do you want to delete this " \
+				       + file_type+"?"):
+			# non-interactive mode or positive answer. 
+			# For each file,...
+			for file in clean_dict[mykey]:
+				# ...get its size...
+				filesize = 0
+				if not os.path.exists(file): continue
+				if not os.path.islink(file):
+					try: filesize = os.path.getsize(file)
+					except: eerror("Could not read size of "\
+					               +file, myoptions['nocolor'])
+				# ...and try to delete it.
+				try: os.unlink(file)
+				except: eerror("Could not delete "+file, \
+				               myoptions['nocolor'])
+				# only count size if successfully deleted
+				else: clean_size += filesize
+	# return total size of deleted or to delete files
+	return clean_size
+
+
+###############################################################################
+# doAction: execute one action, ie display a few message, call the right find*
+# function, and then call doCleanup with its result.
+def doAction(action,myoptions,exclude_dict={}):
+	# define vocabulary for the output
+	if action == 'packages': files_type = "binary packages"
+	else: files_type = "distfiles"
+	# find files to delete, depending on the action
+	if not myoptions['quiet']:
+		einfo("Building file list for "+action+" cleaning...", \
+		      myoptions['nocolor'])
+	if action == 'packages':
+		clean_dict = findPackages( \
+			exclude_dict=exclude_dict, \
+			destructive=myoptions['destructive'], \
+			package_names=myoptions['package-names'], \
+			time_limit=myoptions['time-limit'])
+	else: 
+		clean_dict = findDistfiles( \
+			exclude_dict=exclude_dict, \
+			destructive=myoptions['destructive'], \
+			fetch_restricted=myoptions['fetch-restricted'], \
+			package_names=myoptions['package-names'], \
+			time_limit=myoptions['time-limit'], \
+			size_limit=myoptions['size-limit'])
+	# actually clean files if something was found
+	if len(clean_dict.keys()):
+		# verbose pretend message
+		if myoptions['pretend'] and not myoptions['quiet']:
+			einfo("Here are "+files_type+" that would be deleted:", \
+			      myoptions['nocolor'])
+		# verbose non-pretend message
+		elif not myoptions['quiet']:
+			einfo("Cleaning "+files_type+"...",myoptions['nocolor'])
+		# do the cleanup, and get size of deleted files
+		clean_size = doCleanup(clean_dict,action,myoptions)
+		# vocabulary for final message
+		if myoptions['pretend']: verb = "would be"
+		else: verb = "has been"
+		# display freed space
+		if not myoptions['quiet']:
+			einfo("Total space that "+verb+" freed in " \
+			      + action + " directory: " \
+			      + red(prettySize(clean_size)), \
+			      myoptions['nocolor'])
+	# nothing was found, return
+	elif not myoptions['quiet']:
+		einfo("Your "+action+" directory was already clean.", \
+		      myoptions['nocolor'])
+
+
+###############################################################################
+# main: parse command line and execute all actions
+def main():
+	# set default options
+	myoptions = {}
+	myoptions['nocolor'] = port_settings["NOCOLOR"] in ('yes','true') \
+	                       and sys.stdout.isatty()
+	if myoptions['nocolor']: nocolor()
+	# parse command line options and actions
+	try: myaction = parseArgs(myoptions)
+	# filter exception to know what message to display
+	except ParseArgsException, e:
+		if e.value == 'help':
+			printUsage(help='all')
+			sys.exit(0)
+		elif e.value[:5] == 'help-':
+			printUsage(help=e.value[5:])
+			sys.exit(0)
+		elif e.value == 'version':
+			printVersion()
+			sys.exit(0)
+		else: 
+			printUsage(e.value)
+			sys.exit(2)
+	# parse the exclusion file
+	if not 'exclude-file' in myoptions:
+		my_exclude_file = "/etc/%s/%s.exclude" % (__productname__ , myaction)
+		if os.path.isfile(my_exclude_file):
+			myoptions['exclude-file'] = my_exclude_file
+	if 'exclude-file' in myoptions:
+		try: exclude_dict = parseExcludeFile(myoptions['exclude-file'])
+		except ParseExcludeFileException, e:
+			eerror(e, myoptions['nocolor'])
+			eerror("Invalid exclusion file: %s" % myoptions['exclude-file'], \
+					myoptions['nocolor'])
+			eerror("See format of this file in `man %s`" % __productname__, \
+					myoptions['nocolor'])	
+			sys.exit(1)
+	else: exclude_dict={}
+	# security check for non-pretend mode
+	if not myoptions['pretend'] and portage.secpass == 0:
+		eerror("Permission denied: you must be root or belong to the portage group.", \
+		       myoptions['nocolor'])
+		sys.exit(1)
+	# execute action
+	doAction(myaction, myoptions, exclude_dict=exclude_dict)
+
+
+###############################################################################
+# actually call main() if launched as a script
+if __name__ == "__main__":
+	try: main()
+	except KeyboardInterrupt:
+		print "Aborted."
+		sys.exit(130)
+	sys.exit(0)
+
diff --git a/src/eclean/eclean.1 b/src/eclean/eclean.1
new file mode 100644
index 0000000..7d785af
--- /dev/null
+++ b/src/eclean/eclean.1
@@ -0,0 +1,176 @@
+.TH "eclean" "1" "0.4.1" "gentoolkit"
+.SH "NAME"
+eclean \- A cleaning tool for Gentoo distfiles and binary packages.
+.SH "SYNOPSIS"
+.LP
+.B eclean \fR[\fIglobal\-options\fR] ... <\fIactions\fR> \fR[\fIaction\-options\fR] ...
+.LP
+.B eclean\-dist \fR[\fIglobal\-options, distfiles\-options\fR] ...
+.LP
+.B eclean\-pkg \fR[\fIglobal\-options, packages\-options\fR] ...
+.LP
+.B eclean(-dist,-pkg) \fR[\fI\-\-help, \-\-version\fR]
+.SH "DESCRIPTION"
+\fBeclean\fP is small tool to remove obsolete portage sources files and binary packages.  
+Used on a regular basis, it prevents your DISTDIR and PKGDIR directories to 
+infinitely grow, while not deleting files which may still be useful.
+.PP
+By default, eclean will protect all distfiles or binary packages corresponding to some
+ebuilds available in the Portage tree.  This is the safest mode, since it will protect
+whatever may still be useful, for instance to downgrade a package without downloading
+its sources for the second time, or to reinstall a package you unmerge by mistake
+without recompiling it.  Sure, it's also a mode in which your DISTDIR and PKGDIR will
+stay rather big (although still not growing infinitely).  For the 'distfiles', this
+mode is also quit slow mode because it requiries some access to the whole Portage tree.
+.PP
+If you use the \-\-destructive option, eclean will only protect files corresponding to
+some currently installed package (taking their exact version into account).  It will
+save much more space, while still preserving sources files around for minor revision
+bumps, and binaries for reinstallation of corrupted packages.  But it won't keep files
+for less usual operations like downgrading or reinstalling an unmerged package.  This
+is also the fastest execution mode (big difference for distfiles), and the one used by
+most other cleaning scripts around like yacleaner (at least in its version 0.3).
+.PP
+Somewhere in the middle, adding the \-\-package\-names option when using \-\-destructive
+will protect files corresponding to all existing versions of installed packages.  It will
+allow easy downgrading without recompilation or redownloading in case of trouble, but
+won't protect you against package uninstallation.
+.PP
+In addition to this main modes, some options allow to declare a few special cases file 
+protection rules:
+.IP o
+\-\-time-limit is useful to protect files which are more recent than a given amount of time. 
+.IP o
+\-\-size-limit (for distfiles only) is useful if you want to protect files bigger than  a given size.
+.IP o
+\-\-fetch-restricted (for distfiles only) is useful to protect manually downloaded files.
+But it's also very slow (again, it's a reading of the whole Portage tree data)...
+.IP o
+Finally, you can list some categories or package names to protect in exclusion files (see
+\fBEXCLUSION FILES\fP below).
+.SH "PARAMETERS"
+.SS "Global options"
+.TP
+\fB\-C, \-\-nocolor\fP				turn off colors on output
+.TP
+\fB\-d, \-\-destructive\fP			only keep the minimum for a reinstallation
+.TP
+\fB\-e, \-\-exclude\-file=<path>\fP	path to the exclusion file
+\fB<path>\fP is the absolute path to the exclusion file you want to use.
+When this option is not used, default paths are /etc/eclean/{packages,distfiles}.exclude
+(if they exist).  Use /dev/null if you have such a file at it standard location and
+you want to temporary ignore it.
+.TP
+\fB\-i, \-\-interactive\fP          ask confirmation before deleting
+.TP
+\fB\-n, \-\-package\-names\fP       protect all versions (\-\-destructive only)
+.TP
+\fB\-p, \-\-pretend\fP              only display what would be cleaned
+.TP
+\fB\-q, \-\-quiet\fP                be as quiet as possible, only display errors
+.TP
+\fB\-t, \-\-time-limit=<time>\fP    don't delete files modified since <time>
+\fB<time>\fP is an amount of time: "1y" is "one year", "2w" is "two weeks", etc.
+.br
+Units are: y (years), m (months), w (weeks), d (days) and h (hours).
+.TP
+\fB\-h, \-\-help\fP                 display the help screen
+.TP
+\fB\-V, \-\-version\fP              display version informations
+.SS "Actions"
+.TP
+\fBdistfiles\fR
+Clean files from /usr/portage/distfiles (or whatever else is your DISTDIR in /etc/make.conf).
+This action should be useful to almost any Gentoo user, we all have to big DISTDIRs sometime...
+.br
+\fBeclean\-dist\fP is a shortcut to call eclean with the "distfiles" action, for simplified
+command\-line.
+.TP
+\fBpackages\fR
+Clean files from /usr/portage/packages (or whatever else is your PKGDIR in /etc/make.conf).
+This action is in particular useful for people who use the "buildpkg" or "buildsyspkg"
+FEATURES flags.
+.br
+\fBeclean\-pkg\fP is a shortcut to call eclean with the "packages" action, for simplified
+command\-line.
+.SS "Options for the 'distfiles' action"
+.TP
+\fB\-f, \-\-fetch-restricted\fP		protect fetch-restricted files (\-\-destructive only)
+.TP
+\fB\-s, \-\-size-limit=<size>\fP	don't delete distfiles bigger than <size>
+<size> is a size specification: "10M" is "ten megabytes", "200K" is "two hundreds kilobytes",
+etc.
+.br
+Units are: G, M, K and B.
+.SS "Options for the 'packages' action"
+.TP
+There is no specific option for this action.
+.SH "EXCLUSION FILES"
+Exclusions files are lists of packages names or categories you want to protect
+in particular.  This may be useful to protect more binary packages for some system
+related packages for instance.  Syntax is the following:
+.IP o
+blank lines and lines starting with a "#" (comments) are ignored.
+.IP o
+only one entry per line is allowed.
+.IP o
+if a line contains a category name, like "sys\-apps", then all packages from this 
+category will be protected.  "sys\-apps/*" is also allowed for aesthetic reasons, but
+that does NOT mean that wildcard are supported in any way for any other usage.
+.IP o
+if a line contains a package name ("app\-shells/bash"), then this package will be
+protected.  Versioned atoms like ">=app\-shells/bash\-3" are NOT supported.  Also, the
+full package name (with category) is mandatory.
+.IP o
+if a line contains a package name with an exclamation mark in front ("!sys\-apps/portage"),
+then this package will be excluded from protection.  This is only useful if the category 
+itself was protected.
+.IP o
+for distfiles protection, a line can also a filename to protect. This is useful if you have
+there some files which are not registered by the ebuilds, like OpenOffice.org i18n files
+("helpcontent_33_unix.tgz" for instance).
+.LP
+By default, if it exists, /etc/eclean/packages.exclude (resp. distfiles.exclude) will be use
+when action is "packages" (resp. "distfiles").  This can be overide with the \-\-exclude\-file
+option.
+.SH "EXAMPLES"
+.LP
+Clean distfiles only, with per file confirmation prompt:
+.br
+.B # eclean \-i distfiles
+.LP
+Check which binary packages could be removed, with a no-color display:
+.br
+.B # eclean \-Cp packages
+.LP
+Clean binary packages of uninstalled packages, but keep all versions of installed ones:
+.br
+.B # eclean-pkg \-d \-n
+.LP
+Clean all distfiles except for installed packages (exact version), those which
+are less than one month old, bigger than 50MB, or fetch-restricted:
+.br
+.B # eclean-dist \-d \-t1m -s50M -f
+.LP
+From a crontab, silently clean packages in the safest mode, and then distfiles in destructive
+mode but protecting files less than a week old, every sunday at 1am:
+.br
+.B 0 1 * * sun \ \ eclean \-C \-q packages ; eclean \-C \-q \-d \-t1w distfiles
+.".SH "BUGS"
+.".TP
+."The policy used to decide wether a distfile can be removed or not relies on the SRC_URI variables ."of ebuilds.  It means that if an ebuild uses files that are not part of its SRC_URI, eclean will ."probably remove them.  This are ebuilds bugs, please report them as such on ."http://bugs.gentoo.org. 
+.".TP
+."In safest mode (default, without the \-\-destructive option), this script can be very slow.  There
+."is not much to do about it without hacking outside of the portage API.
+.SH "SEE ALSO"
+.TP
+The Gentoo forum thread that gave birth to eclean:
+.B http://forums.gentoo.org/viewtopic.php?t=3011
+.TP
+The bug report requesting eclean inclusion in gentoolkit:
+.B http://bugs.gentoo.org/show_bug.cgi?id=33877
+.TP
+Yacleaner, one of the other similar tools:
+.B http://blog.tacvbo.net/data/files/yacleaner/
+.SH "AUTHORS"
+Thomas de Grenier de Latour (tgl) <degrenier@easyconnect.fr>
diff --git a/src/eclean/packages.exclude b/src/eclean/packages.exclude
new file mode 100644
index 0000000..8277155
--- /dev/null
+++ b/src/eclean/packages.exclude
@@ -0,0 +1,4 @@
+# /etc/eclean/packages.exclude
+# In this file you can list some categories or cat/pkg-name for which you want
+# to protect binary packages from "ecleaning".
+# See `man eclean` for syntax details.