#!/usr/bin/python

#

# CDDL HEADER START

#

# The contents of this file are subject to the terms of the

# Common Development and Distribution License (the "License").

# You may not use this file except in compliance with the License.

#

# You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE

# or http://www.opensolaris.org/os/licensing.

# See the License for the specific language governing permissions

# and limitations under the License.

#

# When distributing Covered Code, include this CDDL HEADER in each

# file and include the License file at usr/src/OPENSOLARIS.LICENSE.

# If applicable, add the following below this CDDL HEADER, with the

# fields enclosed by brackets "[]" replaced with your own identifying

# information: Portions Copyright [yyyy] [name of copyright owner]

#

# CDDL HEADER END

#

#

# Copyright (c) 2010, 2016, Oracle and/or its affiliates. All rights reserved.

#

# Some userland consolidation specific lint checks

import pkg.lint.base as base

from pkg.lint.engine import lint_fmri_successor

import pkg.elf as elf

import pkg.fmri

import platform

import re

import os.path

import subprocess

import sys

class UserlandActionChecker(base.ActionChecker):

        """An opensolaris.org-specific class to check actions."""

        name = "userland.action"

        def __init__(self, config):

                self.description = _(

                    "checks Userland packages for common content errors")

		path = os.getenv('PROTO_PATH')

		if path != None:

			self.proto_path = path.split()

		else:

			self.proto_path = None

		solaris_ver = os.getenv('SOLARIS_VERSION', '')

		#

		# These lists are used to check if a 32/64-bit binary

		# is in a proper 32/64-bit directory.

		#

		self.pathlist32 = [

			"i86",

			"sparcv7",

			"32",

			"i86pc-solaris-64int",  # perl path

			"sun4-solaris-64int",   # perl path

			"i386-solaris" + solaris_ver,	# ruby path

			"sparc-solaris"	+ solaris_ver	# ruby path

		]

		self.pathlist64 = [

			"amd64",

			"sparcv9",

			"64",

			"i86pc-solaris-64",     # perl path

			"sun4-solaris-64",      # perl path

                        "i86pc-solaris-thread-multi-64", # perl path

                        "sun4-solaris-thread-multi-64", # perl path

			"amd64-solaris" + solaris_ver,	# ruby path

			"sparcv9-solaris" + solaris_ver,# ruby path

			"sparcv9-sun-solaris" + solaris_ver,# ruby path

                        "amd64-solaris-" + solaris_ver,  # ruby path

                        "sparcv9-solaris-" + solaris_ver,# ruby path

			"x86_64-pc-solaris" + solaris_ver  # GCC path

		]

		self.runpath_re = [

			re.compile('^/lib(/.*)?$'),

			re.compile('^/usr/'),

			re.compile('^\$ORIGIN/')

		]

		self.runpath_64_re = [

			re.compile('^.*/64(/.*)?$'),

			re.compile('^.*/amd64(/.*)?$'),

			re.compile('^.*/sparcv9(/.*)?$'),

			re.compile('^.*/i86pc-solaris-64(/.*)?$'), # perl path

			re.compile('^.*/sun4-solaris-64(/.*)?$'),  # perl path

                        re.compile('^.*/i86pc-solaris-thread-multi-64(/.*)?$'),

                                # perl path

                        re.compile('^.*/sun4-solaris-thread-multi-64(/.*)?$'),

                                # perl path

			re.compile('^.*/amd64-solaris2\.[0-9]+(/.*)?$'),

				# ruby path

			re.compile('^.*/sparcv9-solaris2\.[0-9]+(/.*)?$'),

				# ruby path

			re.compile('^.*/sparcv9-sun-solaris2\.[0-9]+(/.*)?$'),

				# GCC path

			re.compile('^.*/x86_64-sun-solaris2\.[0-9]+(/.*)?$')

				# GCC path

		]

		self.initscript_re = re.compile("^etc/(rc.|init)\.d")

                self.lint_paths = {}

                self.ref_paths = {}

                super(UserlandActionChecker, self).__init__(config)

        def startup(self, engine):

                """Initialize the checker with a dictionary of paths, so that we

                can do link resolution.

                This is copied from the core pkglint code, but should eventually

                be made common.

                """

                def seed_dict(mf, attr, dic, atype=None, verbose=False):

                        """Updates a dictionary of { attr: [(fmri, action), ..]}

                        where attr is the value of that attribute from

                        actions of a given type atype, in the given

                        manifest."""

                        pkg_vars = mf.get_all_variants()

                        if atype:

                                mfg = (a for a in mf.gen_actions_by_type(atype))

                        else:

                                mfg = (a for a in mf.gen_actions())

                        for action in mfg:

                                if atype and action.name != atype:

                                        continue

                                if attr not in action.attrs:

                                        continue

                                variants = action.get_variant_template()

                                variants.merge_unknown(pkg_vars)

                                # Action attributes must be lists or strings.

                                for k, v in variants.iteritems():

                                        if isinstance(v, set):

                                                action.attrs[k] = list(v)

                                        else:

                                                action.attrs[k] = v

                                p = action.attrs[attr]

                                dic.setdefault(p, []).append((mf.fmri, action))

                # construct a set of FMRIs being presented for linting, and

                # avoid seeding the reference dictionary with any for which

                # we're delivering new packages.

                lint_fmris = {}

                for m in engine.gen_manifests(engine.lint_api_inst,

                    release=engine.release, pattern=engine.pattern):

                        lint_fmris.setdefault(m.fmri.get_name(), []).append(m.fmri)

                for m in engine.lint_manifests:

                        lint_fmris.setdefault(m.fmri.get_name(), []).append(m.fmri)

                engine.logger.debug(

                    _("Seeding reference action path dictionaries."))

                for manifest in engine.gen_manifests(engine.ref_api_inst,

                    release=engine.release):

                        # Only put this manifest into the reference dictionary

                        # if it's not an older version of the same package.

                        if not any(

                            lint_fmri_successor(fmri, manifest.fmri)

                            for fmri

                            in lint_fmris.get(manifest.fmri.get_name(), [])

                        ):

                                seed_dict(manifest, "path", self.ref_paths)

                engine.logger.debug(

                    _("Seeding lint action path dictionaries."))

                # we provide a search pattern, to allow users to lint a

                # subset of the packages in the lint_repository

                for manifest in engine.gen_manifests(engine.lint_api_inst,

                    release=engine.release, pattern=engine.pattern):

                        seed_dict(manifest, "path", self.lint_paths)

                engine.logger.debug(

                    _("Seeding local action path dictionaries."))

                for manifest in engine.lint_manifests:

                        seed_dict(manifest, "path", self.lint_paths)

                self.__merge_dict(self.lint_paths, self.ref_paths,

                    ignore_pubs=engine.ignore_pubs)

        def __merge_dict(self, src, target, ignore_pubs=True):

                """Merges the given src dictionary into the target

                dictionary, giving us the target content as it would appear,

                were the packages in src to get published to the

                repositories that made up target.

                We need to only merge packages at the same or successive

                version from the src dictionary into the target dictionary.

                If the src dictionary contains a package with no version

                information, it is assumed to be more recent than the same

                package with no version in the target."""

                for p in src:

                        if p not in target:

                                target[p] = src[p]

                                continue

                        def build_dic(arr):

                                """Builds a dictionary of fmri:action entries"""

                                dic = {}

                                for (pfmri, action) in arr:

                                        if pfmri in dic:

                                                dic[pfmri].append(action)

                                        else:

                                                dic[pfmri] = [action]

                                return dic

                        src_dic = build_dic(src[p])

                        targ_dic = build_dic(target[p])

                        for src_pfmri in src_dic:

                                # we want to remove entries deemed older than

                                # src_pfmri from targ_dic.

                                for targ_pfmri in targ_dic.copy():

                                        sname = src_pfmri.get_name()

                                        tname = targ_pfmri.get_name()

                                        if lint_fmri_successor(src_pfmri,

                                            targ_pfmri,

                                            ignore_pubs=ignore_pubs):

                                                targ_dic.pop(targ_pfmri)

                        targ_dic.update(src_dic)

                        l = []

                        for pfmri in targ_dic:

                                for action in targ_dic[pfmri]:

                                        l.append((pfmri, action))

                        target[p] = l

        def __realpath(self, path, target):

		"""Combine path and target to get the real path."""

		result = os.path.dirname(path)

		for frag in target.split(os.sep):

			if frag == '..':

				result = os.path.dirname(result)

			elif frag == '.':

				pass

			else:

				result = os.path.join(result, frag)

		return result

	def __elf_aslr_check(self, path, engine):

		result = None

		ei = elf.get_info(path)

		type = ei.get("type");

		if type != "exe":

			return result

		# get the ASLR tag string for this binary

		aslr_tag_process = subprocess.Popen(

			"/usr/bin/elfedit -r -e 'dyn:sunw_aslr' "

			+ path, shell=True,

			stdout=subprocess.PIPE, stderr=subprocess.PIPE)

		# aslr_tag_string will get stdout; err will get stderr

		aslr_tag_string, err = aslr_tag_process.communicate()

		# No ASLR tag was found; everything must be tagged

		if aslr_tag_process.returncode != 0:

			engine.error(

				_("'%s' is not tagged for aslr") % (path),

				msgid="%s%s.5" % (self.name, "001"))

			return result

		# look for "ENABLE" anywhere in the string;

		# warn about binaries which are not ASLR enabled

		if re.search("ENABLE", aslr_tag_string) is not None:

			return result

		engine.warning(

			_("'%s' does not have aslr enabled") % (path),

			msgid="%s%s.6" % (self.name, "001"))

		return result

	def __elf_runpath_check(self, path, engine):

		result = None

		list = []

		ed = elf.get_dynamic(path)

		ei = elf.get_info(path)

		bits = ei.get("bits")

		for dir in ed.get("runpath", "").split(":"):

			if dir == None or dir == '':

				continue

			match = False

			for expr in self.runpath_re:

				if expr.match(dir):

					match = True

					break

			if match == False:

				list.append(dir)

			# Make sure RUNPATH matches against a packaged path.

			# Don't check runpaths starting with $ORIGIN, which

			# is specially handled by the linker.

			elif not dir.startswith('$ORIGIN/'):

			# Strip out leading and trailing '/' in the 

			# runpath, since the reference paths don't start 

			# with '/' and trailing '/' could cause mismatches.

			# Check first if there is an exact match, then check

			# if any reference path starts with this runpath

			# plus a trailing slash, since it may still be a link

			# to a directory that has no action because it uses

			# the default attributes.

				relative_dir = dir.strip('/')

				if not relative_dir in self.ref_paths and \

				    not any(key.startswith(relative_dir + '/')

				        for key in self.ref_paths):

			# If still no match, if the runpath contains

			# an embedded symlink, emit a warning; it may or may

			# not resolve to a legitimate path.

			# E.g., for usr/openwin/lib, usr/openwin->X11 and

			# usr/X11/lib are packaged, but usr/openwin/lib is not.

			# Otherwise, runpath is bad; add it to list.

					embedded_link = False

					pdir = os.path.dirname(relative_dir)

					while pdir != '':

						if (pdir in self.ref_paths and 

						    self.ref_paths[pdir][0][1].name == "link"):

							embedded_link = True

							engine.warning(

								_("runpath '%s' in '%s' not found in reference paths but contains symlink at '%s'") % (dir, path, pdir),

								msgid="%s%s.3" % (self.name, "001"))

							break

						pdir = os.path.dirname(pdir)

					if not embedded_link:

						list.append(dir)

			if bits == 32:

				for expr in self.runpath_64_re:

					if expr.search(dir):

						engine.warning(

							_("64-bit runpath in 32-bit binary, '%s' includes '%s'") % (path, dir),

							msgid="%s%s.3" % (self.name, "001"))

			else:

				match = False

				for expr in self.runpath_64_re:

					if expr.search(dir):

						match = True

						break

				if match == False:

					engine.warning(

						_("32-bit runpath in 64-bit binary, '%s' includes '%s'") % (path, dir),

						msgid="%s%s.3" % (self.name, "001"))

		if len(list) > 0:

			result = _("bad RUNPATH, '%%s' includes '%s'" %

				   ":".join(list))

		return result

	def __elf_wrong_location_check(self, path, inspath):

		result = None

		ei = elf.get_info(path)

		bits = ei.get("bits")

		type = ei.get("type");

                elems = os.path.dirname(inspath).split("/")

		path32 = False

                path64 = False

		# Walk through the path elements backward and at the first

		# 32/64 bit specific element, flag it and break.

		for p in elems[::-1]:

		    if (p in self.pathlist32):

		    	path32 = True

			break

		    if (p in self.pathlist64):

		    	path64 = True

			break

		# ignore 64-bit executables in normal (non-32-bit-specific)

		# locations, that's ok now.

		if (type == "exe" and bits == 64 and path32 == False and path64 == False):

			return result

		if bits == 32 and path64:

			result = _("32-bit object '%%s' in 64-bit path(%s)" % elems)

		elif bits == 64 and not path64:

			result = _("64-bit object '%s' in 32-bit path")

		return result

	def file_action(self, action, manifest, engine, pkglint_id="001"):

		"""Checks for existence in the proto area."""

		if action.name not in ["file"]:

			return

		inspath=action.attrs["path"]

		path = action.hash

		if path == None or path == 'NOHASH':

			path = inspath

		# check for writable files without a preserve attribute

		if "mode" in action.attrs:

			mode = action.attrs["mode"]

			if (int(mode, 8) & 0222) != 0 and "preserve" not in action.attrs:

				engine.error(

				_("%(path)s is writable (%(mode)s), but missing a preserve"

				  " attribute") %  {"path": path, "mode": mode},

				msgid="%s%s.0" % (self.name, pkglint_id))

		elif "preserve" in action.attrs:

			if "mode" in action.attrs:

				mode = action.attrs["mode"]

				if (int(mode, 8) & 0222) == 0:

					engine.error(

					_("%(path)s has a preserve action, but is not writable (%(mode)s)") %  {"path": path, "mode": mode},

				msgid="%s%s.4" % (self.name, pkglint_id))

			else:

				engine.error(

				_("%(path)s has a preserve action, but no mode") %  {"path": path, "mode": mode},

				msgid="%s%s.3" % (self.name, pkglint_id))

		# checks that require a physical file to look at

		if self.proto_path is not None:

			for directory in self.proto_path:

				fullpath = directory + "/" + path

				if os.path.exists(fullpath):

					break

			if not os.path.exists(fullpath):

				engine.info(

					_("%s missing from proto area, skipping"

					  " content checks") % path, 

					msgid="%s%s.1" % (self.name, pkglint_id))

			elif elf.is_elf_object(fullpath):

				# 32/64 bit in wrong place

				result = self.__elf_wrong_location_check(fullpath, inspath)

				if result != None:

					engine.error(result % inspath, 

						msgid="%s%s.2" % (self.name, pkglint_id))

				result = self.__elf_runpath_check(fullpath, engine)

				if result != None:

					engine.error(result % path, 

						msgid="%s%s.3" % (self.name, pkglint_id))

				result = self.__elf_aslr_check(fullpath, engine)

	file_action.pkglint_desc = _("Paths should exist in the proto area.")

	def link_resolves(self, action, manifest, engine, pkglint_id="002"):

		"""Checks for link resolution."""

		if action.name not in ["link", "hardlink"]:

			return

		path = action.attrs["path"]

		target = action.attrs["target"]

		realtarget = self.__realpath(path, target)

		# Check against the target image (ref_paths), since links might

		# resolve outside the packages delivering a particular

		# component.

		# links to files should directly match a patch in the reference

		# repo.

		if self.ref_paths.get(realtarget, None):

			return

		# If it didn't match a path in the reference repo, it may still

		# be a link to a directory that has no action because it uses

		# the default attributes.  Look for a path that starts with

		# this value plus a trailing slash to be sure this it will be

		# resolvable on a fully installed system.

		realtarget += '/'

		for key in self.ref_paths:

			if key.startswith(realtarget):

				return

		engine.error(_("%s %s has unresolvable target '%s'") %

				(action.name, path, target),

			msgid="%s%s.0" % (self.name, pkglint_id))