# Copyright (C) 2012 Sebastian Pipping <sebastian@pipping.org>
# Licender under GPL v2 or later

from __future__ import print_function

VERSION_STR = '0.5.3'

import sys
import os
import portage
import re

try:
	import argparse
except ImportError:
	print("ERROR: You need Python 2.7+ unless you have module argparse "
		"(package dev-python/argparse on Gentoo) installed independently.", file=sys.stderr)
	sys.exit(1)


_revision_matcher = re.compile('-r([0-9]+)$')
_ignore_matcher = re.compile('(?:^# (?:\\$Header|Copyright)|KEYWORDS=)')


def get_portdir():
	return portage.settings["PORTDIR"]


def find_atoms(repository_path, category_package):
	versions = list()
	category, package = category_package.split('/')
	for root, dirs, files in os.walk(os.path.join(repository_path, category_package)):
		for f in files:
			if not f.endswith('.ebuild'):
				continue
			versions.append(f[len(package) + 1:-len('.ebuild')])
	return versions


def highest_revision_only(versions):
	def make_version(version, revison):
		if revision == 0:
			return version
		return '%s-r%s' % (version, revison)

	d = dict()
	for v in versions:
		match = _revision_matcher.search(v)
		if match is None:
			revision = 0
			version = v
		else:
			revision = int(match.group(1))
			version = v[:-len(match.group(0))]
		d[version] = max(d.get(version, 0), revision)
	return [make_version(version, revison) for version, revison in d.items()]


def find_missed_bumps(gentoo_versions, overlay_versions):
	missed_version_bumps = list()
	missed_revision_bumps = list()
	gentoo_zero_revisions = set(_revision_matcher.sub('', gv) for gv in gentoo_versions)
	for ov in overlay_versions:
		if '999' in ov:
			continue
		newer_than_gentoo = True
		for gv in gentoo_versions:
			if '999' in gv:
				continue
			if portage.vercmp(ov, gv) <= 0:
				newer_than_gentoo = False
				break
		if newer_than_gentoo:
			ov_without_revision = _revision_matcher.sub('', ov)
			if ov_without_revision in gentoo_zero_revisions:
				missed_revision_bumps.append(ov)
			else:
				missed_version_bumps.append(ov)
	return (missed_revision_bumps, missed_version_bumps)


def find_lagging_behind_gentoo(gentoo_versions, overlay_versions):
	lagging_behind_gentoo = list()
	for ov in overlay_versions:
		if '999' in ov:
			continue
		for gv in gentoo_versions:
			if '999' in gv:
				continue
			if ov == gv:
				continue
			if portage.vercmp(ov, gv) < 0 and ov not in gentoo_versions:
				lagging_behind_gentoo.append(ov)
				break
	return lagging_behind_gentoo


def _iter_important_lines(ebuild_f):
	return (line_br for line_br in ebuild_f if _ignore_matcher.search(line_br) is None)


def ebuilds_are_equivalent(filename_a, filename_b):
	fa = open(filename_a, 'r')
	fb = open(filename_b, 'r')
	for (la, lb) in zip(_iter_important_lines(fa), _iter_important_lines(fb)):
		if la != lb:
			return False
	return True


def find_ebuild_changes(category_package, overlay_path, gentoo_versions, overlay_versions):
	ebuild_changes = list()
	intersection = set(gentoo_versions) & set(overlay_versions)
	if not intersection:
		return list()
	category, package = category_package.split('/')
	for version in intersection:
		ebuild_name = '%s-%s.ebuild' % (package, version)
		filename_a = os.path.join(get_portdir(), category_package, ebuild_name)
		filename_b = os.path.join(overlay_path, category_package, ebuild_name)
		if ebuilds_are_equivalent(filename_a, filename_b):
			continue
		# print("meld '%s' '%s'" % (filename_a, filename_b))
		ebuild_changes.append(version)
	return ebuild_changes


def sorted_versions(versions):
	return sorted(versions, key=portage.versions.cmp_sort_key(portage.vercmp))


def dump_tree(tree, title):
	print('===============================================================')
	print(title)
	print('===============================================================')
	for category, package_versions in sorted(tree.items()):
		print('%s/' % category)
		for package, versions in sorted(package_versions.items()):
			print('  %s :: %s' % (package, ', '.join(sorted_versions(versions))))
	print()


def parse_command_line(args):
	parser = argparse.ArgumentParser(description='Simple tool for static analysis of Gentoo overlays')
	parser.add_argument(
		'--version',
		action='version', version='%(prog)s ' + VERSION_STR)
	parser.add_argument('overlay_path',
		metavar='PATH', action='store',
		help='Path to Gentoo overlay')
	conf = parser.parse_args(args=args[1:])

	return conf


def main(args):
	conf = parse_command_line(args)

	def sanitize_overlay_path(path):
		path = path.rstrip('/')
		if path == '':
			return '/'
		return path

	conf.overlay_path = sanitize_overlay_path(conf.overlay_path)

	if not os.path.isdir(conf.overlay_path):
		print("ERROR: Path \"%s\" is not a diretory" % conf.overlay_path, file=sys.stderr)
		return 1

	missed_revision_bumps_tree = dict()
	missed_version_bumps_tree = dict()
	ebuild_changes_tree = dict()
	lagging_behind_gentoo__tree = dict()

	for root, dirs, files in os.walk(conf.overlay_path):
		if '.svn' in dirs:
			dirs[:] = [d for d in dirs if d != '.svn']


		for d in dirs:
			full_path_overlay = os.path.join(root, d)
			category_package = full_path_overlay[len(conf.overlay_path + '/'):]
			if len(category_package.split('/')) != 2:
				continue
			full_path_gentoo = os.path.join(get_portdir(), category_package)
			found = os.path.exists(full_path_gentoo)
			if not found:
				continue

			overlay_versions = find_atoms(conf.overlay_path, category_package)
			gentoo_versions = find_atoms(get_portdir(), category_package)
			(missed_revision_bumps, missed_version_bumps) = find_missed_bumps(gentoo_versions, overlay_versions)
			lagging_behind_gentoo = find_lagging_behind_gentoo(gentoo_versions, overlay_versions)
			ebuild_changes = find_ebuild_changes(category_package, conf.overlay_path, gentoo_versions, overlay_versions)
			category, package = category_package.split('/')
			if missed_revision_bumps:
				missed_revision_bumps_tree.setdefault(category, dict())[package] = highest_revision_only(missed_revision_bumps)
			if missed_version_bumps:
				missed_version_bumps_tree.setdefault(category, dict())[package] = highest_revision_only(missed_version_bumps)
			if ebuild_changes:
				ebuild_changes_tree.setdefault(category, dict())[package] = ebuild_changes
			if lagging_behind_gentoo:
				lagging_behind_gentoo__tree.setdefault(category, dict())[package] = lagging_behind_gentoo


	dump_tree(missed_version_bumps_tree, 'Version bumps missing from Gentoo main tree')
	dump_tree(missed_revision_bumps_tree, 'Revision bumps missing from Gentoo main tree')
	dump_tree(lagging_behind_gentoo__tree, 'Ebuilds lagging behind the Gentoo main tree')
	dump_tree(ebuild_changes_tree, 'Ebuilds that differ at same revision')
	return 0


def go():
	ret = main(sys.argv)
	sys.exit(ret)