span.py

# Copyright 2017 Quark Security, Inc.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# * Redistributions of source code must retain the above copyright notice, this
#   list of conditions and the following disclaimer.
#
# * Redistributions in binary form must reproduce the above copyright notice,
#   this list of conditions and the following disclaimer in the documentation
#   and/or other materials provided with the distribution.
#
# * Neither the name of the copyright holder nor the names of its
#   contributors may be used to endorse or promote products derived from
#   this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

#
# Basic Utilities for Policy Analysis Within Jupyter Notebook
#

import configparser
import os
import os.path
import re

import pandas as pd
import setools as se
from IPython.display import display, Markdown
from setools.policyrep import Type as PolicyRepType
from setools.policyrep import TypeAttribute as PolicyRepTypeAttribute

from . import indexed_terulequery
from .util import compile_policy

pd.options.display.max_rows = 2000
pd.set_option('max_colwidth', 2000)


def dataframe_hide_none(val):
    if val is None or (isinstance(val, set) and len(val) == 0):
        return "color: white"

    return "color: red"


permmap = se.PermissionMap()
# Some common permission sets for use in rule queries
file_w_perms = [x.perm for x in permmap.perms("file") if x.direction == "w"]
file_r_perms = [x.perm for x in permmap.perms("file") if x.direction == "r"]

file_classes = ["file", "blk_file", "chr_file", "lnk_file"]
file_dir_classes = file_classes + ["dir"]

all_object_classes = ['bluetooth_socket', 'netlink_audit_socket', 'tcp_socket', 'msgq', 'rose_socket', 'x_property', 'binder', 'db_procedure', 'dir', 'peer', 'tipc_socket', 'blk_file', 'chr_file', 'db_table', 'db_tuple', 'dbus', 'ipc', 'ipx_socket', 'lnk_file', 'netlink_connector_socket', 'process', 'atmsvc_socket', 'capability2', 'fd', 'nfc_socket', 'packet', 'socket', 'bridge_socket', 'cap_userns', 'fifo_file', 'file', 'node', 'process2', 'x_cursor', 'x_server', 'bpf', 'decnet_socket', 'irda_socket', 'phonet_socket', 'db_view', 'netlink_nflog_socket', 'rds_socket', 'sctp_socket', 'xdp_socket', 'key', 'netlink_netfilter_socket', 'ib_socket', 'netlink_iscsi_socket', 'netlink_tcpdiag_socket', 'unix_stream_socket', 'x_synthetic_event', 'db_database', 'db_language', 'kernel_service', 'netlink_route_socket', 'pppox_socket', 'x_extension', 'db_sequence', 'ieee802154_socket', 'infiniband_endport', 'netlink_rdma_socket', 'netrom_socket', 'shm', 'x_resource', 'llc_socket', 'netlink_selinux_socket', 'capability', 'mpls_socket', 'netlink_ip6fw_socket', 'cap2_userns', 'dccp_socket', 'iucv_socket', 'netlink_firewall_socket', 'sock_file', 'unix_dgram_socket', 'kcm_socket', 'netlink_kobject_uevent_socket', 'vsock_socket', 'db_blob', 'filesystem', 'netlink_xfrm_socket', 'rxrpc_socket', 'x_device', 'can_socket', 'db_schema', 'netlink_dnrt_socket', 'netlink_generic_socket', 'x_client', 'x_gc', 'atmpvc_socket', 'context', 'nscd', 'passwd', 'x_event', 'x_font', 'ax25_socket', 'netlink_scsitransport_socket', 'service', 'x25_socket', 'isdn_socket', 'key_socket', 'netif', 'packet_socket', 'memprotect', 'msg', 'qipcrtr_socket', 'tun_socket', 'udp_socket', 'appletalk_socket', 'netlink_crypto_socket', 'proxy', 'x_colormap', 'x_screen', 'rawip_socket', 'x_application_data', 'association', 'caif_socket', 'x_selection', 'db_column', 'netlink_socket', 'x_drawable', 'infiniband_pkey', 'sem', 'system', 'x_keyboard', 'alg_socket', 'icmp_socket', 'netlink_fib_lookup_socket', 'security', 'smc_socket', 'x_pointer']


def load_policy(fname):
    return Policy(fname)

def load_policy_from_source(fname):
    binary_fname = compile_policy(fname)
    p = Policy(binary_fname)
    os.unlink(binary_fname)
    return p

def _load_policy_from_section(config_section):
    p = None
    ps = None

    if "binary" in config_section:
        p = load_policy(config_section["binary"])
    if "source" in config_section:
        ps = load_refpolicy_source(config_section["source"])

    return p, ps


def load_policies_from_config(config_fname):
    config = configparser.ConfigParser()
    config.read(config_fname)

    p = None
    ps = None
    bp = None
    bs = None

    if config.has_section("Policy"):
        p, ps = _load_policy_from_section(config["Policy"])

    if config.has_section("BasePolicy"):
        bp, bs = _load_policy_from_section(config["BasePolicy"])

    return p, ps, bp, bs


def cond_expr(rule):
    try:
        return rule.conditional
    except Exception as e:
        return None


def type_names(result):
    return [str(x) for x in result]


def filter_types(types, exclude_re):
    r = re.compile(exclude_re)
    return [x for x in types if not r.match(str(x))]


def collect_types(p, raw, expand_attrs=True):
    # This takes a list of type, conditional expression, perms tuples and merges all of the types (removing dups)
    # and expands attributes (if expand_attrs is true).
    #
    # It returns a dictionary of t: perms for the unconditional rules and a dictionary of (t, cond): perms for the
    # conditional rules.
    u = {}
    c = {}

    # We first process the unconditional rules because we are not going to show conditional access if
    # there is also unconditional access
    for t, cond, perms in raw:
        if cond:
            continue

        if expand_attrs:
            keys = list(t.expand())
        else:
            keys = [t]

        for k in keys:
            if k in u:
                u[k] = u[k] | perms
            else:
                u[k] = perms

    for t, cond, perms in raw:
        if not cond or t in u:
            continue

        if expand_attrs:
            keys = [(x, cond) for x in t.expand()]
        else:
            keys = [(t, cond)]

        for k in keys:
            if t in c:
                c[k] = c[k] | perms
            else:
                c[k] = perms

    data = []
    for k in sorted(u.keys()):
        v = u[k]
        data.append({"Type": Type(k), "Conditional": None, "Permissions": sorted(v)})

    for k in sorted(c.keys()):
        t, cond = k
        v = c[k]
        data.append({"Type": Type(t), "Conditional": cond, "Permissions": sorted(v)})

    df = pd.DataFrame(data)[["Type", "Conditional", "Permissions"]]
    df.style.applymap(dataframe_hide_none)

    return df

class Delegator:
    def __init__(self, child) -> None:
        if isinstance(child, Delegator):
            child = child.child
        self.child = child

    def __getattr__(self, name: str):
        return getattr(self.child, name)

    def __hash__(self):
        return hash(self.child)

    def __eq__(self, other):
        if not hasattr(other, "child"):
            return self.child == other
        return self.child == other.child

    def __lt__(self, other):
        if not hasattr(other, "child"):
            return self.child < other
        return self.child < other.child


class Type(Delegator):
    def __repr__(self) -> str:
        return self.name

    def attributes(self):
        attributes = self.child.attributes()
        return [TypeAttribute(x) for x in attributes]

class TypeAttribute(Delegator):
    def __repr__(self) -> str:
        return self.name

def wrap(instance):
    if isinstance(instance, PolicyRepType):
        return Type(instance)
    elif isinstance(instance, PolicyRepTypeAttribute):
        return TypeAttribute(instance)
    else:
        return instance

class Policy(se.SELinuxPolicy):
    # common trusted domain types we can filter out to remove noise
    example_trusted_domain_types = ["kernel_t", "rpm_script_t", "setfiles_t", "setfiles_mac_t", "restorecond_t",
                                    "smbd_t", "mount_t", "files_unconfined_type", "xauth_t", "hostname_t",
                                    "readahead_t", "rpm_t", "nmbd_t", "init_t", "initrc_t", "insmod_t", "mdadm_t",
                                    "devices_unconfined_type", "udev_t", "bootloader_t", "system_cronjob_t"]
    trusted_domain_types = []

    domain_attribute = "domain"

    def terules_query_orig(self, **kwargs):
        if "ruletype" not in kwargs:
            kwargs["ruletype"] = [se.TERuletype.allow]
        results = se.TERuleQuery(self, **kwargs).results()
        filtered_results = [x for x in results if x.source not in self.trusted_domain_types]
        return sorted(filtered_results)

    def terules_query_raw(self, **kwargs):
        if "ruletype" not in kwargs:
            kwargs["ruletype"] = [se.TERuletype.allow]
        results = indexed_terulequery.TERuleQueryIndexed(self, **kwargs).results()

        if "trusted_domain_types" in kwargs:
            td = kwargs["trusted_domain_types"]
        else:
            td = self.trusted_domain_types
        if td is not None:
            filtered_results = [x for x in results if x.source not in td]
        else:
            filtered_results = results

        if "trusted_target_types" in kwargs:
            tt = kwargs["trusted_target_types"]
            filtered_results = [x for x in filtered_results if x.target not in tt]

        return sorted(filtered_results)

    BASE_RULE_ATTRS = ["source", "target", "tclass"]
    AVRULE_ATTRS = ["perms"]
    AVRULE_XPERMS_RULE_ATTRS = ["perms", "xperm_type"]
    TYPE_RULE_ATTRS = ["default", "filename"]

    def terules_query(self, **kwargs):
        r = self.terules_query_raw(**kwargs)

        data = []
        extra_indexes = set()
        for rule in r:
            rt = se.TERuletype
            if rule.ruletype in (rt.allow, rt.dontaudit, rt.neverallow, rt.dontaudit):
                attrs = self.AVRULE_ATTRS
            elif rule.ruletype in (rt.allowxperm, rt.neverallowxperm, rt.auditallowxperm, rt.dontauditxperm):
                attrs = self.AVRULE_XPERMS_RULE_ATTRS
            else:
                attrs = self.TYPE_RULE_ATTRS
            extra_indexes.update(attrs)

            row = {}
            for attr in self.BASE_RULE_ATTRS + attrs:
                try:
                    row[attr] = wrap(getattr(rule, attr))
                except:
                    row[attr] = None
                    continue
            row["cond"] = cond_expr(rule)

            data.append(row)

        if not len(data):
            return None

        df = pd.DataFrame(data)[self.BASE_RULE_ATTRS + list(extra_indexes) + ["cond"]]
        df.style.applymap(dataframe_hide_none)

        return df

    def constraint_query(self, **kwargs):
        if "ruletype" not in kwargs:
            kwargs["ruletype"] = ["mlsconstrain"]
        results = se.ConstraintQuery(self, **kwargs).results()
        return sorted(results)

    def transrules_query(self, **kwargs):
        if "ruletype" not in kwargs:
            kwargs["ruletype"] = [se.TERuletype.type_transition]

        return self.terules_query(**kwargs)

    def roles_query(self, **kwargs):
        return sorted(se.RoleQuery(self, **kwargs).results())

    def types_in_role(self, role_name):
        return sorted([str(x) for x in self.roles_query(name=role_name)[0].types()])

    def roles_for_type(self, type_name):
        roles = sorted([str(x) for x in self.roles()])
        out = []
        for role in roles:
            rtypes = self.types_in_role(role)
            if type_name in rtypes:
                out.append(role)
        return out

    def roletrans_query(self, **kwargs):
        if "ruletype" not in kwargs:
            kwargs["ruletype"] = [se.RBACRuletype.role_transition, se.RBACRuletype.allow]
        return sorted(se.RBACRuleQuery(self, **kwargs).results())

    def dta_analysis(self, domain, *args, **kwargs):
        return list(se.DomainTransitionAnalysis(self, *args, **kwargs).transitions(domain))

    def types_re(self, s, **kwargs):
        q = se.TypeQuery(self, name_regex=True, **kwargs)
        q.name = s
        return [Type(x) for x in sorted(q.results())]

    def lookup_type(self, name):
        return Type(super().lookup_type(name))

    def lookup_type_or_attrs(self, type_names):
        """
        Convert a list of type/attribute names into a list of types.
        
        :param type_names: An iterable of type names 
        :return: An list of Type and TypeAttribute objects
        """

        return sorted([self.lookup_type_or_attr(x) for x in type_names])

    def expand_attributes(self, tlist):
        expanded = set()
        [expanded.update(x.expand()) for x in tlist]

        return expanded

    def attributes_re(self, s, **kwargs):
        q = se.TypeAttributeQuery(self, name_regex=True, **kwargs)
        q.name = s
        return sorted(q.results())

    def attributes_for_type(self, tname):
        attrs = sorted(self.lookup_type(tname).attributes())
        return attrs

    def types_in_attribute(self, attr):
        return sorted(self.attributes_re("^%s$" % attr)[0].expand())

    def new_types(self, base_policy):
        """
        Determine what types are new in this policy compared to another
        policy.
        :param base_policy: The policy to compare with. 
        :return: Tuple contains (list of new types, list of new domains)
        """
        # We have to do this as strings because the hasing
        # from different policies doesn't work correctly
        p_domains = as_strset(self.types())
        bp_domains = as_strset(base_policy.types())

        custom_types = self.lookup_type_or_attrs(p_domains - bp_domains)
        custom_domains = self.lookup_type_or_attrs(self.domain_types(custom_types))

        return custom_types, custom_domains

    def domain_types(self, types=None):
        out = set()
        if types is None:
            types = self.types()
        for t in types:
            if isinstance(t, str):
                t = self.lookup_type(t)
            if self.domain_attribute in list(t.attributes()):
                out.add(t)
        return out

    def domains_with(self, target_name, tclass=["file", "dir"], perms=file_w_perms, expand_attrs=False):
        raw = [(x.source, cond_expr(x), x.perms) for x in
               self.terules_query_raw(target=target_name, tclass=tclass, perms=perms, trusted_domain_types=[])]
        return collect_types(self, raw, expand_attrs=expand_attrs)

    def domains_with_file_w_perms(self, target_name, expand_attrs=False):
        return self.domains_with(self, target_name, tclass=["file", "dir"], perms=file_w_perms,
                                 expand_attrs=expand_attrs)

    def domains_with_file_r_perms(self, target_name, expand_attrs=False):
        return self.domains_with(self, target_name, tclass=["file", "dir"], perms=file_r_perms,
                                 expand_attrs=expand_attrs)

    def domains_that_can_relabel(self, from_type, to_type, expand_attrs=False):
        f = self.domains_with(from_type, tclass=None, perms=["relabelfrom"], expand_attrs=expand_attrs)
        t = self.domains_with(to_type, tclass=None, perms=["relabelto"], expand_attrs=expand_attrs)

        def build_sets(result):
            unconditional = set()
            conditional = {}
            for row in result.itertuples():
                if row[2] == "":
                    unconditional.add(row.type)
                else:
                    if row.Type not in conditional:
                        conditional[row.Type] = []
                    conditional[row.Type].append(row.Conditional)

            return unconditional, conditional

        f_u, f_c = build_sets(f)
        t_u, t_c = build_sets(t)

        data = []

        def append_row(d, from_conditional=None, to_conditional=None):
            data.append({"type": d, "from_conditional": from_conditional, "to_conditional": to_conditional})

        # For unconditional rules everything is easy - we just need to which types
        # are in both sets (meaning they both have relabel rules
        for d in sorted(t_u | f_u):
            append_row(d)

        # Conditional rules are harder. We have to look for any matching rules that are conditional
        # or unconditional. And we don't want to merge all of the different conditionals together.
        # So we will have duplicate rows in some cases (which makes it clear in the results which conditionals
        # must be active to allow the relabeling).
        #
        # Since we are looking for pairs and we show _all_ of the conditional relabels we only have to go through
        # the from (or the to) for the conditional rules. Because any domains that we don't hit by going through one
        # by definition don't have a matching pair in the other. BUT the to rules might have a match in the unconditional
        # rules. So we will go through t_c just to match unconditional.
        for d in sorted(f_c.keys()):
            # if we have a match
            if d in t_u:
                append_row(d, from_conditional=f_c[d])
            if d in t_c:
                append_row(d, from_conditional=f_c[d], to_conditional=t_c[d])

        for d in sorted(t_c.keys()):
            if d in f_u:
                append_row(d, to_conditional=t_c[d])

        df = pd.DataFrame(data)[["type", "from_conditional", "to_conditional"]]
        df.style.applymap(dataframe_hide_none)

        return df

    DIR_WRITE = "w"
    DIR_READ = "r"

    def info_flow_perms(self, tclass=["file"], direction=DIR_WRITE, min_weight=7):
        perms = set()
        for c in tclass:
            perms.update([x.perm for x in permmap.perms(c) if
                          (x.direction == direction or x.direction == "b") and x.weight >= min_weight])

        return list(perms)

    def domain_info_flow(self, domain, tclass=file_classes, direction=DIR_WRITE, min_weight=10, expand_attrs=False):
        """
        Return the set of object types that information can flow into or out of for a given domain.
    
        domain - name of the domain type
        tclass - list of target class names
        direction - DIR_WRITE or DIR_READ
        """
        perms = self.info_flow_perms(tclass, direction, min_weight)

        raw = [(x.target, cond_expr(x), as_strset(x.perms)) for x in
               self.terules_query_raw(source=domain, tclass=tclass, perms=perms, trusted_domain_types=[])]

        return collect_types(self, raw, expand_attrs)

    def object_info_flow(self, object_type, tclass=file_classes, direction=DIR_WRITE, min_weight=10,
                         expand_attrs=False):
        """
        Return the set of domain types that information can flow into or out of for a given object.
    
        object_type - name of the object type
        tclass - list of target class names
        direction - DIR_WRITE or DIR_READ
        """
        perms = self.info_flow_perms(tclass, direction, min_weight)

        raw = [(x.source, cond_expr(x), as_strset(x.perms)) for x in
               self.terules_query_raw(target=object_type, tclass=tclass, perms=perms, trusted_domain_types=[])]

        return collect_types(self, raw, expand_attrs)

    def policy_caps(self, **kwargs):
        return list(se.PolCapQuery(self, **kwargs).results())

    def types_summary(self, types):
        data = []
        for t in sorted(types):
            t = Type(t)
            data.append({"name": str(t), "attributes": sorted([x.name for x in t.attributes()])})

        df = pd.DataFrame(data)[["name", "attributes"]]
        df.style.applymap(dataframe_hide_none)

        return df

    def domain_summary(self, domain):
        pp_markdown(domain_summary_raw(self, domain))

    def file_summary(self, file_type):
        obj_classes_except_process = as_str(self.classes())
        obj_classes_except_process.remove("process")
        data = {
            "file_type": file_type,
            "attributes": markdown_list(self.attributes_for_type(file_type)),
            "other_trans": markdown_code_from_results(
                self.transrules_query(target=file_type, target_indirect=False, tclass=obj_classes_except_process)),
            "fread": markdown_code_from_results(
                self.terules_query_raw(target=file_type, tclass=file_dir_classes, perms=["read"], target_indirect=False)),
            "fwrite": markdown_code_from_results(
                self.terules_query_raw(target=file_type, tclass=file_dir_classes, perms=["write", "append"],
                                       target_indirect=False)),
        }

        pp_markdown(file_summary_template.format(**data))

    def packet_summary(self, packet_type):
        obj_classes_except_process = as_str(self.classes())
        obj_classes_except_process.remove("process")
        data = {
            "packet_type": packet_type,
            "attributes": markdown_list(self.attributes_for_type(packet_type)),
            "packet": markdown_code_from_results(
                self.terules_query_raw(target=packet_type, tclass=["packet"], target_indirect=False)),
        }

        pp_markdown(packet_summary_template.format(**data))

    def attribute_summary(self, attribute):
        # Attributes are tricky because we do not know if an attribute is really meant for files or for domains
        #  We are going to take some educated guesses with this that may or may not work out well
        #  If we find the word file in the attribute name we treat it as a file
        #  Anything else we're going to great as a domain

        r_file = re.compile(".*file.*")
        obj_classes_except_process = as_str(self.classes())
        obj_classes_except_process.remove("process")

        if r_file.match(attribute):
            data = {
                "file_type": attribute,
                "types": markdown_list(self.types_in_attribute(attribute)),
                "other_trans": markdown_code_from_results(
                    self.transrules_query(target=attribute, target_indirect=False, tclass=obj_classes_except_process)),
                "fread": markdown_code_from_results(
                    self.terules_query_raw(target=attribute, tclass=file_dir_classes, perms=["read"],
                                           target_indirect=False)),
                "fwrite": markdown_code_from_results(
                    self.terules_query_raw(target=attribute, tclass=file_dir_classes, perms=["write", "append"],
                                           target_indirect=False)),
            }

            pp_markdown(file_attribute_summary_template.format(**data))

        else:
            data = {
                "attribute": attribute,
                "types": markdown_list(self.types_in_attribute(attribute)),
                "capabilities": markdown_code_from_results(
                    self.terules_query_raw(source=attribute, tclass=["capability"], source_indirect=False)),
                "fread": markdown_code_from_results(
                    self.terules_query_raw(source=attribute, tclass=file_dir_classes, perms=["read"],
                                           source_indirect=False)),
                "fwrite": markdown_code_from_results(
                    self.terules_query_raw(source=attribute, tclass=file_dir_classes, perms=["write", "append"],
                                           source_indirect=False)),
                "packet": markdown_code_from_results(
                    self.terules_query_raw(source=attribute, tclass=["packet"], source_indirect=False)),
                "process": markdown_code_from_results(
                    self.terules_query_raw(source=attribute, tclass=["process"], source_indirect=False)),
                "socket": markdown_code_from_results(
                    self.terules_query_raw(source=attribute, tclass="socket", tclass_regex=True, source_indirect=False))
            }

            pp_markdown(domain_attribute_summary_template.format(**data))


def pp_markdown(markdown):
    display(Markdown(markdown))


def as_str(l):
    return [str(x) for x in l]


def as_strset(l):
    return set(as_str(l))


def markdown_code_from_results(results):
    out = """
```
%s
```
        """ % "\n".join(as_str(results))
    return out


def markdown_list(l):
    out = ""
    for item in as_str(l):
        out = out + "  * `%s`\n" % item

    return out


domain_summary_template = """
### Domain Summary: {domain}

####  Type Attributes

{attributes}

#### Domain Transitions In

{dta_in}

#### Domain Transitions Out

{dta_out}

#### Entrypoints

{entrypoints}

#### Non-Process Transitions

{other_trans}

#### Roles

{roles}

#### Capability Permissions

{capabilities}

#### File Read Permissions

{fread}

#### File Write Permissions

{fwrite}

#### Packet Permissions

{packet}

#### Process Permissions

{process}

#### Socket Permissions

{socket}


"""


def domain_summary_raw(p, domain):
    obj_classes_except_process = as_str(p.classes())
    obj_classes_except_process.remove("process")
    data = {
        "domain": domain,
        "attributes": markdown_list(p.attributes_for_type(domain)),
        "dta_in": markdown_list(
            [x.source for x in p.terules_query_raw(target=domain, tclass=["process"], perms=["transition"])]),
        "dta_out": markdown_list(
            [x.target for x in p.terules_query_raw(source=domain, tclass=["process"], perms=["transition"])]),
        "entrypoints": markdown_list(
            [x.target for x in p.terules_query_raw(source=domain, tclass=["file"], perms=["entrypoint"])]),
        "other_trans": markdown_code_from_results(
            p.transrules_query(source=domain, source_indirect=False, tclass=obj_classes_except_process)),
        "roles": markdown_list(p.roles_for_type(domain)),
        "capabilities": markdown_code_from_results(
            p.terules_query_raw(source=domain, tclass=["capability"], source_indirect=False)),
        "fread": markdown_code_from_results(
            p.terules_query_raw(source=domain, tclass=file_dir_classes, perms=["read"], source_indirect=False)),
        "fwrite": markdown_code_from_results(
            p.terules_query_raw(source=domain, tclass=file_dir_classes, perms=["write", "append"],
                                source_indirect=False)),
        "packet": markdown_code_from_results(
            p.terules_query_raw(source=domain, tclass=["packet"], source_indirect=False)),
        "process": markdown_code_from_results(
            p.terules_query_raw(source=domain, tclass=["process"], source_indirect=False)),
        "socket": markdown_code_from_results(
            p.terules_query_raw(source=domain, tclass="socket", tclass_regex=True, source_indirect=False))
    }

    return domain_summary_template.format(**data)


domain_attribute_summary_template = """
### Domain Attribute Summary: {attribute}

####  Types in Attribute

{types}

#### Capability Permissions

{capabilities}

#### File Read Permissions

{fread}

#### File Write Permissions

{fwrite}


#### Packet Permissions

{packet}

#### Process Permissions

{process}

#### Socket Permissions

{socket}

"""

file_attribute_summary_template = """
### File Attribute Summary: {file_type}

####  Types in Attribute

{types}

#### Transitions

{other_trans}

#### File Read Permissions

{fread}

#### File Write Permissions

{fwrite}

"""



file_summary_template = """
### File Type Summary: {file_type}

####  Type Attributes

{attributes}

#### Transitions

{other_trans}

#### File Read Permissions

{fread}

#### File Write Permissions

{fwrite}

"""





packet_summary_template = """
### Packet Type Summary: {packet_type}

####  Type Attributes

{attributes}

#### Packet Permissions

{packet}

"""


# Monkey patch to get some better __repr__ - these should be expanded over time
# to cover more types.
#
# With this you don't have to print the last statement - e.g.,
#
#     p.terules_query(source="smbd_t")
#
# You can also just use the built-in pretty printer (aliased a pp)
#
#    pp(p.terules_query(source="smbd_t"))

def str_repr(self):
    return str(self)

# Now that setools moved many of these over to cython, the monkey patching won't work
# for these types. Just leaving this here as a reminder to do somethign about this at
# some point in the future.
#se.policyrep.BaseType.__repr__ = str_repr
#se.policyrep.BaseTERule.__repr__ = str_repr
#se.policyrep.BaseRole.__repr__ = str_repr

import subprocess
from pygments import highlight
from pygments.lexers import DiffLexer
from pygments.formatters import HtmlFormatter


def diff_to_html(diff_text):
    html = HtmlFormatter()
    return highlight(diff_text, DiffLexer(), html) + "<style>%s</style" % html.get_style_defs()


# PolicySource makes it easier to grep through the sources inside the notebook to
# find the policy statements granting the access we care about.
class RefPolicySource(object):
    def __init__(self, path):
        self.path = path
        self.abspath = os.path.abspath(self.path)
        self.modules_path = self.path + "/policy/modules"

    def grep(self, search_str, file_type, path=None, extended=False):
        if path is None:
            path = self.path
        return subprocess.run(["grep", "-r", "-n" ] + [ "--extended-regexp" ] * extended +
                              [ "--include", file_type, search_str, "."], cwd=path, stdout=subprocess.PIPE,
                              stderr=subprocess.STDOUT).stdout.decode('utf-8')

    def diff(self, patha, pathb):
        return subprocess.run(["diff", "-u", pathb, patha], cwd=self.path,
                              stdout=subprocess.PIPE).stdout.decode('utf-8')

    def diff_relative(self, fname, other_source):
        return diff_to_html(self.diff(self.abspath + fname, other_source.abspath + fname))

    def find_def(self, name):
        return self.grep(name, "*.te", self.modules_path)

    def type_def(self, name):
        return self.find_def("type " + name)

    def attr_def(self, name):
        return self.find_def("attribute " + name)

    def file_contexts(self, type_name, extended=False):
        return self.grep(type_name, "*.fc", self.modules_path, extended)

    def genfscon(self, fs_type):
        return self.find_def("genfscon " + fs_type)

    def rules_search(self, search_str):
        return self.grep(search_str, "*", self.modules_path)

    def get_module(self, file_name):
        fd = open(self.modules_path + "/" + file_name)
        return fd.read()

    def diff_mls_constraints(self, other_source):
        return self.diff_relative("/policy/mls", other_source)

    def diff_mcs_constraints(self, other_source):
        return self.diff_relative("/policy/mcs", other_source)

    def diff_constraints(self, other_source):
        return self.diff_relative("/policy/constraints", other_source)


def load_refpolicy_source(path):
    return RefPolicySource(path)


if __name__ == '__main__':
    p = load_policy("policy.30")
    p.build_index_if_needed()

    # t = timeit.timeit(lambda: p.terules_query(source="smb", source_regex=True, source_indirect=True), number=4)
    # print(t)
    # t = timeit.timeit(lambda: p.terules_query_orig(source="smb", source_regex=True, source_indirect=True), number=4)
    # print(t)

    # print(domain_info_flow(p, "passwd_t", tclass=["file"]))

    p.domain_types()
    print(p.types_summary())