Capirca is designed to utilize common definitions of networks, services and high-level policy files to facilitate the development and manipulation of network access control lists (ACLs) for various platforms. It was developed by Google for internal use, and is now open source.
Capirca consist of capirca Python package and the capirca tool.
The typical usage workflow consists of the following steps:
- Create object definitions containing "network" and "service" definitions
- Create a access control policy defining the desired state of access control and referencing the object definitions together with desired firewall platforms
- Generate ACL configurations by running
capircacommand referencing the access control policy and the object definitions. The command triggers a generator for each of the firewall platforms.
At a high-level capirca rationalizes objects (networks, services) against a security definition (.pol file) to generate a specific device configuration file via a platform specific ACL generator. Before getting started some objects must exist, the below table summarizes where objects are stored:
| path | description |
|---|---|
| /def/NETWORK.net | a list of network objects definitions |
| /def/SERVICES.svc | a list of service objects definitions |
Each network or service definition file has a very simple structure. A token is
defined, e.g. GUEST_NET, followed by an equal sign, then followed by a
definition, e.g. 10.10.10.0/24, and optional description field, e.g. # guest network range.
GUEST_NET = 10.10.10.0/24 # guest network range
The tool populates the access control policy from .pol files in a
particular directory, e.g. policies/. The tool searches
recursively for .pol files and add them to the policy, .e.g .pol files are
located in policies/pol.
Additionally, the .pol files MAY reference other policy definition files
located outside of the directory by using include directive. Please see
Includes section for documentation.
The files with .net extension contain the definitions of network objects, e.g.
IP networks and hosts. The following definition creates INTERNAL and RFC1918
network objects in the object definitions, whether INTERNAL references the IP
ranges of RFC 1918 defined in the RFC1918.
RFC1918 = 10.0.0.0/8 # non-public
172.16.0.0/12 # non-public
192.168.0.0/16 # non-public
INTERNAL = RFC1918
The files with .svc extension contain the definitions of service objects, e.g.
ports and protocols.
DNS = 53/tcp # transfers
53/udp # queries
The nesting of tokens is permitted only when both tokens are of the same type. The referencing of a "network" object by "service" object is not allowed, and vice versa.
The examples of nesting of the network and service object follow.
HTTP = 80/tcp # common web
HTTPS = 443/tcp # SSL web
HTTP_8080 = 8080/tcp # web on non-standard port
WEB_SERVICES = HTTP HTTP_8080 HTTPS # all our web services
DB_SERVICES = 3306/tcp # allow db access
HTTPS # and SSL access
NYC_NETWORK = 200.1.1.0/24 # New York office
ATL_NETWORK = 200.2.1.0/24 # Atlanta office
DEN_NETWORK = 200.5.1.0/24 # Denver office
REMOTE_OFFICES = NYC_NETWORK
ATL_NETWORK
DEN_NETWORK
The network objects may reference both IPv4 and IPv6 addresses at the same time.
LOOPBACK = 127.0.0.1/32 # loopback in IPv4
::1/128 # loopback in IPv6
LINKLOCAL = FE80::/10 # IPv6 link local address
NYC_NETWORK = 172.16.1.0/24 # NYC IPv4
2620:0:10A1::/48 # NYC IPv6
A policy file (/policies/pol/something.pol) has the security policy written using capirca specific meta-language and format. There are specific sections (e.g: header) that tell capirca how to generate the output configuration of the security policy.
The header section defines:
- target firewall platforms (which ACL generator to use)
- passes additional arguments to the generator responsible for that platform.
A single header may have many targets within a section. It will result in multiple outputs being generated for that policy.
The term sections defines the access control rules within an ACL, it contains keywords followed by an object (service or network) and policy decision ("action" keyword).
The term section specifies the network flow metadata for ACL matching.
- Addresses
- Ports
- Protocols
- Action (allow/deny)
Inside a term a mandatory keyword will be found followed by an object token
for rule evaluation.
Tokens are the names of services and networks loaded from the object definitions. Example:
| token_name | definition |
|---|---|
| "HTTPS" | 443 |
| "NYC_NETWORK" | 192.168.0.0/24 |
| keyword | description |
|---|---|
| required | must be supported by all output policy generators |
| optional | available in a subset of the generators and are intended to |
| : : provide additional flexibility when developing policies for that : | |
| : : specific target platform : |
- action
- accept
- deny
- reject
- next
- reject-with-tcp-rst
- comment
- a text comment enclosed in double-quotes. Comments may span multiple lines if desired.
- destination-address
- one or more destination address tokens.
- destination-exclude
- exclude one or more address tokens from the specified destination-address.
- destination-port
- one or more service definition tokens.
- icmp-type
- specific icmp-type code to match (IPv4/IPv6 types vary).
- IPv4:
- echo-reply
- unreachable
- source-quench
- redirect
- alternate-address
- echo-request
- router-advertisement
- router-solicitation
- time-exceeded
- parameter-problem
- timestamp-request
- timestamp-reply
- information-request
- information-reply
- mask-request
- mask-reply
- conversion-error
- mobile-redirect
- IPv6:
- destination-unreachable
- packet-too-big
- time-exceeded
- parameter-problem
- echo-request
- echo-reply
- multicast-listener-query
- multicast-listener-report
- multicast-listener-done
- router-solicit
- router-advertisement
- neighbor-solicit
- neighbor-advertisement
- redirect-message
- router-renumbering
- icmp-node-information-query
- icmp-node-information-response
- inverse-neighbor-discovery-solicitation
- inverse-neighbor-discovery-advertisement
- version-2-multicast-listener-report
- home-agent-address-discovery-request
- home-agent-address-discovery-reply
- mobile-prefix-solicitation
- mobile-prefix-advertisement
- certification-path-solicitation
- certification-path-advertisement
- multicast-router-advertisement
- multicast-router-solicitation
- multicast-router-termination
- specific icmp-type code to match (IPv4/IPv6 types vary).
- option
- connection options.
- established
- only permit established connections; implements tcp-established flag if protocol is tcp only, otherwise adds 1024-65535 to required destination-ports.
- tcp-established
- only permit established tcp connections, usually checked based on TCP flag settings. If protocol UDP is included in term, only adds 1024-65535 to required destination-ports.
- sample
- not supported by all generators. Samples traffic for netflow.
- intial
- currently only supported by juniper generator. Appends tcp-initial to the term.
- rst
- currently only supported by juniper generator. Appends "tcp-flags rst" to the term.
- first-fragment
- currently only supported by juniper generator. Appends 'first-fragment' to the term.
- established
- connection options.
- protocol
- network protocols this term will match, such as tcp, udp, icmp, or a numeric value.
- protocol-except
- network protocols that should be excluded from the protocol specification. This is rarely used.
- source-address
- one or more source address tokens.
- source-exclude
- exclude one or more address tokens from the specified source-address.
- source-port
- one or more service definition tokens.
- verbatim
- this specifies that the text enclosed within quotes should be rendered into the output without interpretation or modification. This is sometimes used as a temporary workaround while new required features are being added.
WARNING: These terms may or may not function properly on all generators. Always refer to the generator specific documentation and code base.
- address
- one or more network address tokens matches either source or destination.
- counter
- (Juniper only) enable filter-based generic routing encapsulation (GRE) tunneling using the specified tunnel template.
- destination-prefix
- (Juniper only) specify destination-prefix matching (e.g. source-prefix` configured-neighbors-only).
- ether-type
- (Juniper only) specify matching ether-type(e.g. ether-type` arp).
- fragement-offset
- (Juniper only) specify a fragment offset of a fragmented packet.
- logging
- (Juniper, speedway/iptables) specify that this packet should be logged via syslog.
- loss-priority
- (Juniper only) specify loss priority.
- packet-length
- (Juniper only) specify packet length.
- policer
- (Juniper only) specify which policer to apply to matching packets.
- precedence
- (Juniper only) specify precendence.
- qos
- (Juniper only) apply quality of service classification to matching packets (e.g. qos` af4).
- routing-instance
- (iptables, speedway only) specify specific interface a term should apply to (e.g. source-interface` eth3).
- source-prefix
- (Juniper only) specify source-prefix matching (e.g. source-prefix, configured-neighbors-only).
- traffic-type
- (Juniper only) specify traffic-type.
Policy files support the use of #include statements. An include may be used to
avoid duplication of commonly used text, such as a group of terms that permit or
block specific types of traffic.
An include directive will result in the contents of the included file being
injected into the current policy file in the exact location of the #include
directive. An example include:
#include 'includes/untrusted-networks-blocking.inc'
NOTE: Includes are only read from the subdirectories of your base_directory, all other directories will error out.
The .inc file extension and the includes/ folder is not required but it is
recommended to be used as a best practice and for easier readability.
WARNING: Not all generators have the same configuration options or feature set.
header {
target:: paloalto from-zone internal to-zone external
}
term ping-gdns{
source-address:: INTERNAL
destination-address:: GOOGLE_DNS
protocol:: icmp
action:: accept
}
The above example tells capirca to use paloalto.py to generate a platform specific configuration for Palo Alto.
The security policy is written within the term section using the meta-language:
- name/description: ping-gdns
- source: any INTERNAL network (check /def/NETWORK.net definition of 'INTERNAL')
- destination: service object named GOOGLE_DNS
- protocol: icmp
- action: accept
The above ACL controls traffic in one direction only (outbound towards the service) and there should be another header and term to control the traffic in the opposite direction. Unless the target generator features the ability to automatically create bi-directional configuration from a single ACL term. Always check the documentation of the generator or validate the output generated to validate final configuration and policy interpretation.
The following list contains links to the documentation of the individual policy generators:
arista: Aristaaruba: Arubabrocade: Brocadecisco: Ciscociscoasa: Cisco ASAcisconx: Cisco NXciscoxr: Cisco XRcloudarmor: cloudarmorgce: GCEgcp_hfipset: ipsetiptables: iptablesjuniper: Juniperjuniperevo: Juniper EVOjunipermsmpc: Juniperjunipersrx: Juniper SRXk8s: Kubernetes NetworkPolicynftables: nftablesnsxv: NSXnsxt: NSX-Tpacketfilter: PacketFilterpaloaltofw: Palo Alto PANOSpcap: PcapFiltersonic: SONiC ACLs in config_db.json formatspeedway: Speedwaysrxlo: Stateless Juniper ACLwindows_advfirewall: Windows Advanced Firewall
The following are examples of how to construct a term, and assumes that naming definition tokens used have been defined in the definitions files.
- Block incoming bogons and spoofed traffic
term block-bogons {
source-address:: BOGONS RFC1918
source-address:: COMPANY_INTERNAL
action:: deny
}
- Permit Public to Web Servers
term permit-to-web-servers {
destination-address:: WEB_SERVERS
destination-port:: HTTP
protocol:: tcp
action:: accept
}
- Permit Replies to DNS Servers From Primaries
term permit-dns-tcp-replies {
source-address:: DNS_PRIMARIES
destination-address:: DNS_SECONDARIES
source-address:: DNS
protocol:: tcp
option:: tcp-established
action:: accept
}
- Permit All Corporate Networks, Except New York, to FTP Server
This will "subtract" the CORP_NYC_NETBLOCK from the CORP_NETBLOCKS token.
For example, assume CORP_NETBLOCKS includes 200.0.0.0/20, and
CORP_NYC_NETBLOCK is defined as 200.2.0.0/24. The source-exclude will
remove the NYC netblock from the permitted source addresses. If the excluded
address is not contained with the source address, nothing is changed.
term allow-inbound-ftp-from-corp {
source-address:: CORP_NETBLOCKS
source-exclude:: CORP_NYC_NETBLOCK
destination-port:: FTP
protocol:: tcp
action:: accept
}
Below is an example policy file for a Juniper target platform. It contains two filters, each with a handful of terms. This examples assumes that the network and service naming definition tokens have been defined.
header {
comment:: "edge input filter for sample network."
target:: juniper edge-inbound
}
term discard-spoofs {
source-address:: RFC1918
action:: deny
}
term permit-ipsec-access {
source-address:: REMOTE_OFFICES
destination-address:: VPN_HUB
protocol:: 50
action:: accept
}
term permit-ike-access {
source-address:: REMOTE_OFFICES
destination-address:: VPN_HUB
protocol:: udp
destination-port:: IKE
action:: accept
}
term permit-public-web-access {
destination-address:: WEB_SERVERS
destination-port:: HTTP HTTPS HTTP_8080
protocol:: tcp
action:: accept
}
term permit-tcp-replies {
option:: tcp-established
action:: accept
}
term default-deny {
action:: deny
}
header {
comment:: "edge output filter for sample network."
target:: juniper edge-outbound
}
term drop-internal-sourced-outbound {
destination-address:: INTERNAL
destination-address:: RESERVED
action:: deny
}
term reject-internal {
source-address:: INTERNAL
action:: reject
}
term default-accept {
action:: accept
}
If setuptools Python package is not installed on your system, install it: For
example, the following commands installs the package with apt package manager.
sudo apt-get install python3-pip python3-setuptoolsNext, to install capirca from source, clone the capirca repository and run
its installer:
git clone https://github.com/google/capirca.git
cd capirca/
python3 setup.py install --userTypically, when provided --user argument, the installer creates the following
files, where 3.8 is Python version and 2.0.0 is the version of capirca:
~/.local/bin/capirca~/.local/lib/python3.8/site-packages/capirca-2.0.0-py3.8.egg
If necessary, remove build files:
rm -rf build capirca.egg-info distNext, test capirca by generating sample output filters for Cisco, Juniper,
iptables, and other firewall platforms.
~/.local/bin/capircaThe generation of sample output while in the capirca's source code directory
does not require command line parameters, because capirca inherits default
settings from the following configuration (see capirca/utils/config.py).
{
'base_directory': './policies',
'definitions_directory': './def',
'policy_file': None,
'output_directory': './filters',
'optimize': False,
'recursive': True,
'debug': False,
'verbose': False,
'ignore_directories': ['DEPRECATED', 'def'],
'max_renderers': 10,
'shade_check': False,
'exp_info': 2
}Although the policy_file is None, the tool processes all policies located in
base_directory, i.e. ./policies. The tool loads network and service
definitions from definitions_directory. The tool output generated ACLs to the
root of the source directory because output_directory is ./filters.
Currently, the PyPI is out of date. Nevertheless, a user can install an older
version of capirca with pip:
pip install capirca --userThere are a number of command-line arguments that can be used with capirca.
$ ~/.local/bin/capirca --helpfull
USAGE: capirca [flags]
flags:
absl.app:
-?,--[no]help: show this help
(default: 'false')
--[no]helpfull: show full help
(default: 'false')
--[no]helpshort: show this help
(default: 'false')
--[no]helpxml: like --helpfull, but generates XML output
(default: 'false')
--[no]only_check_args: Set to true to validate args and exit.
(default: 'false')
--[no]pdb: Alias for --pdb_post_mortem.
(default: 'false')
--[no]pdb_post_mortem: Set to true to handle uncaught exceptions with PDB post mortem.
(default: 'false')
--profile_file: Dump profile information to a file (for python -m pstats). Implies --run_with_profiling.
--[no]run_with_pdb: Set to true for PDB debug mode
(default: 'false')
--[no]run_with_profiling: Set to true for profiling the script. Execution will be slower, and the output format might change over time.
(default: 'false')
--[no]use_cprofile_for_profiling: Use cProfile instead of the profile module for profiling. This has no effect unless --run_with_profiling
is set.
(default: 'true')
absl.logging:
--[no]alsologtostderr: also log to stderr?
(default: 'false')
--log_dir: directory to write logfiles into
(default: '')
--logger_levels: Specify log level of loggers. The format is a CSV list of `name:level`. Where `name` is the logger name used with
`logging.getLogger()`, and `level` is a level name (INFO, DEBUG, etc). e.g. `myapp.foo:INFO,other.logger:DEBUG`
(default: '')
--[no]logtostderr: Should only log to stderr?
(default: 'false')
--[no]showprefixforinfo: If False, do not prepend prefix to info messages when it's logged to stderr, --verbosity is set to INFO level,
and python logging is used.
(default: 'true')
--stderrthreshold: log messages at this level, or more severe, to stderr in addition to the logfile. Possible values are 'debug', 'info',
'warning', 'error', and 'fatal'. Obsoletes --alsologtostderr. Using --alsologtostderr cancels the effect of this flag. Please also note
that this flag is subject to --verbosity and requires logfile not be stderr.
(default: 'fatal')
-v,--verbosity: Logging verbosity level. Messages logged at this level or lower will be included. Set to 1 for debug logging. If the flag
was not set or supplied, the value will be changed from the default of -1 (warning) to 0 (info) after flags are parsed.
(default: '-1')
(an integer)
capirca.capirca:
--base_directory: The base directory to look for acls; typically where you'd find ./corp and ./prod
(default: './policies')
--config_file: A yaml file with the configuration options for capirca;
repeat this option to specify a list of values
--[no]debug: Debug messages
(default: 'false')
--definitions_directory: Directory where the definitions can be found.
(default: './def')
--exp_info: Print a info message when a term is set to expire in that many weeks.
(default: '2')
(an integer)
--ignore_directories: Don't descend into directories that look like this string
(default: 'DEPRECATED,def')
(a comma separated list)
--max_renderers: Max number of rendering processes to use.
(default: '10')
(an integer)
-o,--[no]optimize: Turn on optimization.
(default: 'False')
--output_directory: Directory to output the rendered acls.
(default: './filters')
--policy_file: Individual policy file to generate.
--[no]recursive: Descend recursively from the base directory rendering acls
(default: 'true')
--[no]shade_check: Raise an error when a term is completely shaded by a prior term.
(default: 'false')
--[no]verbose: Verbose messages
(default: 'false')
absl.flags:
--flagfile: Insert flag definitions from the given file into the command line.
(default: '')
--undefok: comma-separated list of flag names that it is okay to specify on the command line even if the program does not define a flag
with that name. IMPORTANT: flags in this list that have arguments MUST use the --flag=value format.
(default: '')
Notably, the --config_file PATH argument allows passing one or more yaml
configuration files. These files will be prioritized from left to right, i.e.
any duplicate configurations will be overriden, not merged.
The command line arguments take precendence over any settings passed via the configuration files.
The default capirca configurations in a YAML format follows:
---
base_directory: ./policies
definitions_directory: ./def
output_directory: ./
optimize: false
recursive: true
debug: false
verbose: false
ignore_directories:
- DEPRECATED
- def
max_renderers: 10
shade_check: true
exp_info: 2The capirca tool uses capirca Python package.
Therefore, there is a way to access capirca programmatically.
policies/sample_paloalto.poldef/SERVICES.svcdef/NETWORK.net
Provided you have the following files in your directory, the following code
snippets create a naming definitions object, policy object, and render
generator filter output.
NOTE: Paste the code snippets one line at a time.
First, start Python interpreter:
$ python3
Python 3.8.7 (default, Dec 22 2020, 10:37:26)
[GCC 10.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>>
Next, import naming library and create naming object from definitions files
in ./def directory.
from pprint import pprint
from capirca.lib import naming
defs = naming.Naming('./def')
pprint(defs)
<capirca.lib.naming.Naming object at 0x7f8421b57280>The defs object follows:
<capirca.lib.naming.Naming object at 0x7f8421b57280>
The Naming object has various methods. The GetServiceNames method returns
the service name tokens.
>>> dir(defs)
['GetIpParents', 'GetNet', 'GetNetAddr', 'GetNetChildren', 'GetServiceNames',
<...intentionally omitted ..>
'unseen_networks', 'unseen_services']
>>>
>>> pprint(defs.GetServiceNames())
['WHOIS',
'SSH',
<...intentionally omitted ..>
'TRACEROUTE']
>>>
Then, import policy library, read in the policy configuration data from
./policies/sample_paloalto.pol, and create a policy object.
from capirca.lib import policy
conf = open('./policies/sample_paloalto.pol').read()
pol = policy.ParsePolicy(conf, defs, optimize=True)The policy object follows:
>>> pprint(pol)
Policy: {Target[paloalto], Comments [], Apply groups: [], except: []:[ name: ping-gdns
source_address: [IPv4('10.0.0.0/8'), IPv4('172.16.0.0/12'), IPv4('192.168.0.0/16')]
destination_address: [IPv4('8.8.4.4/32'), IPv4('8.8.8.8/32'), IPv6('2001:4860:4860::8844/128'), IPv6('2001:4860:4860::8888/128')]
protocol: ['icmp']
action: ['accept'], name: dns-gdns
source_address: [IPv4('10.0.0.0/8'), IPv4('172.16.0.0/12'), IPv4('192.168.0.0/16')]
destination_address: [IPv4('8.8.4.4/32'), IPv4('8.8.8.8/32'), IPv6('2001:4860:4860::8844/128'), IPv6('2001:4860:4860::8888/128')]
protocol: ['tcp']
destination_port: [(53, 53)]
action: ['accept'], name: allow-web-outbound
source_address: [IPv4('10.0.0.0/8'), IPv4('172.16.0.0/12'), IPv4('192.168.0.0/16')]
protocol: ['tcp']
destination_port: [(80, 80), (443, 443)]
action: ['accept']], Target[paloalto], Comments [], Apply groups: [], except: []:[ name: allow-icmp
protocol: ['icmp']
action: ['accept'], name: allow-only-pan-app
action: ['accept']
pan_application: ['http'], name: allow-web-inbound
destination_address: [IPv4('200.1.1.1/32'), IPv4('200.1.1.2/32')]
protocol: ['tcp']
destination_port: [(80, 80), (443, 443)]
action: ['accept']
pan_application: ['ssl', 'http']]}
>>>
Next, import a generator library (here paloaltofw for Palo Alto firewalls) and
output a policy in the desired format.
from capirca.lib import paloaltofw
for header in pol.headers:
if 'paloalto' in header.platforms:
jcl = True
if jcl:
output = paloaltofw.PaloAltoFW(pol, 1)
print(output)The following code initiates Palo Alto firewall ACL model with the default expiration of 1 week.
paloaltofw.PaloAltoFW(pol, 1)
If your use case is to just use the CLI and you don't want to go through the
process of installing capirca, you can use the dockerized version of the tool.
When using docker, mount your working directory to the /data directory of
the container and pass command-line arguments in the following way.
docker run -v "${PWD}:/data" docker.pkg.github.com/google/capirca/capirca:latest
docker run -v "${PWD}:/data" docker.pkg.github.com/google/capirca/capirca:latest --helpfull
docker run -v "${PWD}:/data" docker.pkg.github.com/google/capirca/capirca:latest --config_file /data/path/to/configThe Capirca threat model assumes some control and verification of policy definitions (in .pol files). This is either through human user verification, or that policies are generated by upstream systems that enforce correctness.
It is recommended that the ACL generated by Capirca is always tested for correctness before being applied to production. Not all generators support every feature, configuration option or term keywords. When something is unsupported, Capirca will error out. But due to the sensitive nature of network ACLs, it is always recommended to test any new generator being used, or new policies being generated.
External links, resources and references:
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent