#!/usr/bin/env python3
# rpmconf-matt [--sync-only]

# Known blockers to submission to Fedora: Basically removing assumptions specific to my setup.
# - Decide how to name it vs. existing "rpmconf" in Fedora
# - Make it configurable what part of the filesystem to scan.  Is there a
#   default that's sufficiently safe for people to use on systems with untrusted
#   users?
# - Should save its own files in another tree rather than blacklisting
#   directories known to cause problems if it adds its files there.
# - Make diff program configurable / smarter default

# TODO: Move the "sync" part into a dnf plugin that runs after every transaction
# to make it more robust.  Perhaps there can be an option to run the merge
# interactively.  With tighter integration, we might be able to skip scanning the
# whole filesystem, but that would add one more kind of state that can get wrong.

# Python 3 conversion, 2017-11-12
# One could argue we should treat filenames and RPM names as bytes, but it's too
# much of a pain to fix all the resulting fallout now, and we don't expect
# adversarial names anyway. ~ Matt 2017-11-12
#
# Fedora 30 -> 32 upgrade: remove some .decode() calls (guess rpm-python changed
# some things from bytes to str) ~ Matt 2020-06-01

import collections
import hashlib
import os
import pwd, grp
import re
import shutil
import stat
import subprocess
import sys
import tempfile
import time

import rpm

def getFileDigest(path, algoHint):
	# TODO: Get the digest algorithm from RPM??
	# File a bug to have rpmfiDigestAlgo exposed to Python?
	if len(algoHint) == 32:  # adobe packages
		algo = hashlib.md5
	elif len(algoHint) == 64:
		algo = hashlib.sha256
	else:
		raise ValueError('Failed to guess digest algorithm')
	with open(path, 'rb') as f:
		return algo(f.read()).hexdigest()

def setAttributes(fe, path):
  # C.f. https://github.com/rpm-software-management/rpm/blob/rpm-4.13.0-rc1/lib/fsm.c#L713
	# If user or group is not found, we'll get a KeyError.
	os.chown(path, pwd.getpwnam(fe.user).pw_uid, grp.getgrnam(fe.group).gr_gid)
	os.chmod(path, fe.mode & 0o7777)
	os.utime(path, (fe.mtime, fe.mtime))

def makeConfFindCommand(expr):
	# Make a find command intended to catch any files ever managed by RPM,
	# without getting confused by trees such as /proc
	# or even malicious user-created files.  Unfortunately, we can't just ask RPM
	# which files it previously managed.  The following heuristic should work for
	# ml*: root filesystem (and avoid an error on /mnt/root, which -xdev doesn't
	# catch), and exclude world-writable dirs such as /var/tmp.
	# TODO: Make the search criteria more general.
	return r"find / -xdev \( -path /mnt -or -path /var/www/html -or -perm /002 \) -prune -or " + expr

def doMerge(a, b, c, output):
	# TODO: Make diff program customizable.
	# FIXME: Stop leaving ".orig" files around.  At least name them distinctively.
	# kdiff3 wrapper to work around issue with Qt apps running as root under Qubes ~ 2015-08-20
	args = ['rpmconf-matt-merge', output, a if os.path.lexists(a) else '', b, c]
	subprocess.check_call(args)

class NeededPackage(object):
	def __init__(self, header):
		self.nvra = header.nvra
		self.paths = dict()  # live path -> (digest, path to download to)

def rpmconf(syncOnly=False):
	if os.geteuid() != 0:
		print('This tool needs to run as root.', file=sys.stderr)
		sys.exit(1)

	# First, rename any rpmsave files and corresponding rpmbase files before we
	# would overwrite the rpmbase files.  We'll do this to any file type, even
	# though there will only ever be base files for regular files.
	print('Scanning for rpmsave files that need to be stamped.')
	saveStamp = str(int(time.time()))
	for savePath in subprocess.check_output(makeConfFindCommand("-name '*.rpmsave' -print"), shell=True).decode().splitlines():
		livePath = savePath[:-len('.rpmsave')]
		liveBasePath = livePath + '.rpmbase'
		stampedSavePath = livePath + '.rpmsave-' + saveStamp
		stampedSaveBasePath = stampedSavePath + '-base'
		print('-- Timestamping rpmsave file for %s.' % livePath)
		# XXX: Make sure we are not clobbering existing files?
		os.rename(savePath, stampedSavePath)
		if os.path.lexists(liveBasePath):
			os.rename(liveBasePath, stampedSaveBasePath)

	filesToMerge = collections.defaultdict(lambda: [[], False])  # live path -> (list of rpmsave stamps, bool if rpmnew)

	print('Scanning for config files that need base files created.')
	filesDone = {}  # live path -> (nevra, digest)
	needPackages = {}  # nevra -> NeededPackage; no more defaultdict because NeededPackage needs header
	ts = rpm.ts()
	mi = ts.dbMatch()
	for header in mi:
		nevra = header.nevra
		for fe in rpm.files(header):
			# Only installed config files.
			if fe.state != rpm.RPMFILE_STATE_NORMAL: continue
			if not (fe.fflags & rpm.RPMFILE_CONFIG): continue
			# For a ghost, we have no base content to write.  Probably best to let this be a two-way merge if the file becomes non-ghost later.
			if fe.fflags & rpm.RPMFILE_GHOST: continue
			# For now, we only handle regular files.  Conflicts on config symlinks seem to be rare.
			if not stat.S_ISREG(fe.mode): continue
			if fe.caps != '': raise NotImplementedError('File capabilities are not implemented: %s' % fe.name)

			# Extension point directories whose readers can't handle additional *.rpm* files.
			# /etc/skel/ is not actually causing a problem but leads to ugly persistent state.
			# TODO: Find a better workaround.
			if re.search('^(/etc/skel/|/etc/rpm/macros|/etc/logrotate.d/|/etc/grub.d/)', fe.name):
				continue

			# We need this check to avoid thrashing a conflicted base file.
			if fe.name in filesDone:
				(oldNevra, oldDigest) = filesDone[fe.name]
				if fe.digest != oldDigest:
					print('Conflict at %s: have %s from %s, ignoring %s from %s' %
						(fe.name, oldDigest, oldNevra, fe.digest, nevra),
						file=sys.stderr)
				continue
			filesDone[fe.name] = (nevra, fe.digest)

			path_new = fe.name + '.rpmnew'
			download_path = None
			if os.path.lexists(path_new):
				filesToMerge[fe.name][1] = True
				# The live config file is not based on the current DB entry.
				# Hopefully we already have a base for the live config file; if not, there's nothing we can do about it now.
				# We do want to make sure the rpmnew file is correct.
				if getFileDigest(path_new, fe.digest) != fe.digest:
					download_path = path_new
			else:
				path_base = fe.name + '.rpmbase'
				if not (os.path.lexists(path_base) and getFileDigest(path_base, fe.digest) == fe.digest):
					if (os.path.lexists(fe.name) and getFileDigest(fe.name, fe.digest) == fe.digest):
						# The live file has the original content.
						# Copy the content and set the original attributes manually.
						path_tmp = fe.name + '.rpmbase-tmp'
						shutil.copyfile(fe.name, path_tmp)
						setAttributes(fe, path_tmp)
						os.rename(path_tmp, path_base)
						print('- %s: Copied %s from %s.' % (nevra, path_base, fe.name))
					else:
						download_path = path_base
			if download_path:
				if nevra not in needPackages:
					needPackages[nevra] = NeededPackage(header)
				needPackages[nevra].paths[fe.name] = (fe.digest, download_path)
				print('- %s: Need to download %s.' % (nevra, download_path))

	if needPackages:
		print('Downloading %d packages.' % len(needPackages))
		packages_tmpdir = tempfile.mkdtemp(prefix='rpmconf-packages')
		# Make sure the cpio archive is covered by a valid signature
		# before we use it.  Since dnf-download-signed currently doesn't
		# check that the package is the one we asked for, this only
		# ensures that the cpio archive is safe to extract.  Then we
		# check the digest on each needed file before using it.  We're
		# still correct if an attacker substitutes a different signed
		# package in which the files we need have the same content.
		# ~ Matt 2019-05-18
		#
		# Ideally, we'd only require a signature if the package came
		# from a repository with gpgcheck=1.  Right now, I use no
		# unsigned packages.  If I build my own packages again, I can
		# either sign them or just fix them manually if they reach this
		# code.
		# ~ Matt 2017-11-11
		subprocess.check_call(['dnf-download-signed'] + list(needPackages), cwd=packages_tmpdir)
		for nevra, neededPkg in needPackages.items():
			packagePath = '%s/%s.rpm' % (packages_tmpdir, neededPkg.nvra)
			extract_tmpdir = tempfile.mkdtemp(prefix='rpmconf-extract-%s' % nevra)
			cpioNeedPaths = ['.' + p for p in neededPkg.paths]  # go figure
			subprocess.check_call(['/bin/bash', '-c', 'p="$1"; shift; rpm2cpio "$p" | cpio --extract --quiet --preserve-modification-time --make-directories "$@"', '--', packagePath] + cpioNeedPaths, cwd=extract_tmpdir)
			print('- Extracted %s.' % nevra)
			for livePath, (needDigest, downloadPath) in neededPkg.paths.items():
				tmpPath = extract_tmpdir + livePath
				tmpDigest = getFileDigest(tmpPath, needDigest)
				if tmpDigest != needDigest:
					print('%s: got digest %s, wanted %s' % (livePath, tmpDigest, needDigest), file=sys.stderr)
					continue
				# This is easiest in case it is cross-filesystem, etc.  mv should preserve all attributes.
				subprocess.check_call(['mv', '-f', tmpPath, downloadPath])
				print('-- Installed %s.' % downloadPath)
			shutil.rmtree(extract_tmpdir)
		shutil.rmtree(packages_tmpdir)

	print('Scanning for obsolete rpmnew files.')
	for newPath in subprocess.check_output(makeConfFindCommand("-type f -name '*.rpmnew' -print"), shell=True).decode().splitlines():
		livePath = newPath[:-len('.rpmnew')]
		if livePath not in filesToMerge:  # only rpmnew files will be recorded in filesToMerge yet
			print('-- Deleting %s. UNTESTED' % newPath)
			#os.unlink(newPath)

	# Remove rpmbase files for config files that are no longer managed, to not leave cruft.
	# This intentionally does not remove rpmsave-base files. ~ 2014-07-03
	# Note: If the config file had been modified, RPM would move it to
	# rpmsave on package removal (whether or not it was noreplace), so we'd
	# stamp the base file before we get here. ~ 2017-11-12
	print('Scanning for obsolete rpmbase files.')
	for basePath in subprocess.check_output(makeConfFindCommand("-type f -name '*.rpmbase' -print"), shell=True).decode().splitlines():
		livePath = basePath[:-len('.rpmbase')]
		if livePath not in filesDone:
			print('-- Deleting %s.' % basePath)
			os.unlink(basePath)

	# "sync vs. merge" terminology is inspired by Perforce.  We'll want a
	# better term for "sync" before releasing this to the public.
	if syncOnly:
		print('rpmconf sync complete.')
		return
	else:
		print('rpmconf sync complete.  You can interrupt if you don\'t wish to merge now.')

	print('Scanning for rpmsave files.')
	for savePath in subprocess.check_output(makeConfFindCommand("-type f -name '*.rpmsave-*' -print"), shell=True).decode().splitlines():
		m = re.search('^(.*)\.rpmsave-(\d+)$', savePath)
		if not m: continue
		(livePath, stamp) = (m.group(1), int(m.group(2)))
		filesToMerge[livePath][0].append(stamp)

	# Nested function for the ability to return from a nested loop...
	def mergeFile(livePath):
		print('- Merging %s.' % livePath)
		(saveStamps, haveRpmnew) = filesToMerge[livePath]
		saveStamps.sort()  # mutates the original, that's OK
		# TODO: If a package was uninstalled, we could have an rpmsave and rpmsave-base with no live.
		# We want to alert the user that the configuration change is no longer having an effect.  How?
		# Currently kdiff3 comes up with a bunch of error dialogs and I have to manually intervene.
		for i in range(len(saveStamps)):
			c_output = '%s.rpmsave-%d' % (livePath, saveStamps[i+1]) if i+1 < len(saveStamps) else livePath
			b = '%s.rpmsave-%d' % (livePath, saveStamps[i])
			a = b + '-base'
			try:
				doMerge(a, b, c_output, c_output)
			except subprocess.CalledProcessError:
				print('- Leaving %s merge unfinished.' % livePath)
				return 1
			os.unlink(b)
			if os.path.lexists(a): os.unlink(a)
		if haveRpmnew:
			try:
				doMerge(livePath + '.rpmbase', livePath, livePath + '.rpmnew', livePath)
			except subprocess.CalledProcessError:
				print('- Leaving %s merge unfinished.' % livePath)
				return 1
			os.rename(livePath + '.rpmnew', livePath + '.rpmbase')
		print('- Merged %s.' % livePath)
		return 0

	unfinishedFiles = 0
	for livePath in filesToMerge:
		unfinishedFiles += mergeFile(livePath)

	if unfinishedFiles == 0:
		print('rpmconf merge complete!')
	else:
		print('No more files to consider.  %d files left unfinished.' % unfinishedFiles)

if __name__ == '__main__':
	# TODO: Adopt a real option-parsing library.
	args = sys.argv[1:]
	if args == []:
		rpmconf()
	elif args == ['--sync-only']:
		rpmconf(syncOnly=True)
	else:
		print('Unrecognized arguments.', file=sys.stderr)
		sys.exit(1)