#!/usr/bin/python
# Copyright 1999-2005 Gentoo Foundation
# Distributed under the terms of the GNU General Public License v2
# $Header: $

# TODO:
# * add a --root=/path/to/some/dir option:
#   - check /$ROOT/var/db/pkg exists
#   - set the env var before importing portage module
#   - get the right vardbapi
#   - takes only file paths with ${ROOT} prefix into account

__author__      = "Thomas de Grenier de Latour (TGL)"
__email__       = "degrenier@easyconnect.fr"
__version__     = "0.1"
__date__        = "2005/04/27"
__productname__ = "fix_contents"
__description__ = "\
This script can update some files mtime and/or md5 in the CONTENTS list of\n\
their owning package. Handle with care, this is about messing with your\n\
portage installed packages database."

import sys
import os, os.path
import getopt
sys.path.insert(0, "/usr/lib/portage/pym")
from portage import settings, db, dblink
import portage_checksum


###############################################################################
## Globals:

vdbapi = db["/"]["vartree"].dbapi
options = { \
	'stdin'   : False, \
	'pretend' : False, \
	'quiet'   : False, \
	'verbose' : False }
files = {}
filenames = {}
owners = {}


###############################################################################
## CLI functions:

def print_version():
	print "This is \"%s\" %s (%s)" % (__productname__, __version__, __date__)
	print "Author: %s <%s>" % (__author__, __email__)

def doing(message, stream=sys.stdout):
	if not options['quiet']:
		print >>stream,">>>",message,"... "
def warn(message, stream=sys.stderr):
	print >>stream,"!!!",message
def info(message, stream=sys.stdout):
	if options['verbose']:
		print >>stream,"---",message

def print_help():
	print_version()
	print
	print "Description:\n%s" % __description__
	print
	print \
"\
Usage: \n\
    %(name)s [-q|-v] [-p] </path/to/file> ...\n\
    %(name)s [-q|-v] [-p] -s\n\
    %(name)s -h|-V\n\
\n\
Otions:\n\
    -s, --stdin      : read files list from standard input instead of\n\
                       command line arguments. It's one file per line.\n\
    -p, --pretend    : fake mode. Don't write the CONTENTS files.\n\
    -q, --quiet      : minimal noise level\n\
    -v, --verbose    : maximal noise level\n\
    -V, --version    : get version informations\n\
    -h, --help       : display this help screen\n\
" % { 'name': __productname__ }


###############################################################################
## Parse command line options and collect the files list:

def parse_cmdline():
	global options, files, filenames
	try: opts, args = getopt.getopt(sys.argv[1:], "hpqsvV", \
			["help","pretend","quiet","stdin","verbose","version"])
	except:
		warn("Wrong command line options. Exiting")
		sys.exit(2)
	for o, a in opts:
		if o in ("-s", "--stdin"):
			options['stdin'] = True
		elif o in ("-p", "--pretend"):
			options['pretend'] = True
		elif o in ("-q", "--quiet"):
			options['quiet'] = True
		elif o in ("-v", "--verbose"):
			options['verbose'] = True
		elif o in ("-h", "--help"):
			print_help()
			sys.exit(0)
		elif o in ("-V", "--version"):
			print_version()
			sys.exit(0)
	if options['verbose'] and options['quiet']:
		warn("Error: Using both --quiet and --verbose doesn't make sense.")
		sys.exit(2)
	if options['stdin']:
		if args:
			warn("Error: Arguments found on both the command line and stdin.")
			sys.exit(2)
		args = sys.stdin.readlines()
	for my_file in args:
		my_file = os.path.normpath(my_file.strip())
		my_real_file = os.path.realpath(my_file)
		if not os.path.isfile(my_real_file):
			warn("%s: not a regular file. Ignoring." % my_file)
		else: 
			files[my_real_file] = { 'owners' : [] }
			filenames[os.path.basename(my_real_file)] = None


###############################################################################
## Find the packages owning the files we want to update:

def get_files_owners():
	global files, owners
	for my_cpv in vdbapi.cpv_all():
		my_split = my_cpv.split('/')
		my_contents = \
			dblink(my_split[0], my_split[1], '/', settings).getcontents()
		for my_file in my_contents:
			if my_contents[my_file][0] != 'obj': continue
			my_filename = os.path.basename(my_file)
			if my_filename not in filenames: continue
			my_real_file = os.path.realpath(my_file)
			if my_real_file not in files: continue
			files[my_real_file]['owners'].append(my_cpv)
			owners[my_cpv] = None
			info("%s: %s" % (my_real_file, my_cpv))
	for my_file in files:
		if not files[my_file]['owners']:
			warn("No owner package was found for %s" % my_file)
		elif len(files[my_file]['owners']) >= 2:
			warn("Several owner packages were found for %s: %s" \
			     % (my_file, ', '.join(files[my_file]['owners'])))


###############################################################################
## Get the files new md5sums and mtimes:

def get_files_info():
	global files
	for my_file in files:
		if not files[my_file]['owners']:
			continue
		files[my_file]['md5'] = \
			portage_checksum.perform_md5(my_file, calc_prelink=1)
		files[my_file]['mtime'] = os.path.getmtime(my_file)
		info("%s: md5 = %s -- mtime = %s" \
		     % (my_file, files[my_file]['md5'], files[my_file]['mtime']))


###############################################################################
## Update CONTENTS files of the owner packages with the new files md5/mtimes:

def write_files_info():
	for my_cpv in owners:
		my_split = my_cpv.split('/')
		my_dblink = dblink(my_split[0], my_split[1], '/', settings)
		my_contents = open(os.path.join(my_dblink.dbdir,'CONTENTS'),'r')
		my_contents_lines = my_contents.readlines()
		my_contents.close()
		my_new_contents_lines = []
		for my_line in my_contents_lines:
			my_new_contents_lines.append(my_line)
			my_line_split = my_line.split()
			if my_line_split[0] != 'obj': continue
			my_file = my_line_split[1]
			my_filename = os.path.basename(my_file)
			if my_filename not in filenames: continue
			my_real_file = os.path.realpath(my_file)
			if my_real_file not in files: continue
			my_new_line = "obj %s %s %s\n" \
				% (my_file, 
				   files[my_real_file]['md5'], 
				   files[my_real_file]['mtime'])
			info("%s: %s" % (my_cpv, my_new_line.strip()))
			my_new_contents_lines[-1] = my_new_line
		my_new_contents_data = ''.join(my_new_contents_lines)
		if not options['pretend']:
			my_dblink.setfile('CONTENTS', my_new_contents_data)


###############################################################################
## Do stuffs:

parse_cmdline()
if os.geteuid() and not options['pretend']:
	warn("Error: Only root can run this script for real.")
	sys.exit(1)
if not files:
	info("Nothing to do.")
	sys.exit(0)
doing("Looking for packages owning this files")
get_files_owners()
doing("Reading files new mtime and md5")
get_files_info()
doing("Fixing packages CONTENTS files")
write_files_info()
sys.exit(0)