Lookup Class

Lookup class is the main feature of abstract package. It implements Abstraction Concepts in a user-friendly fashion, and allows users to perform dynamic lookups just as if they were accessing object attributes.

                                .------> TokenX.Y implementation
                               /
UserScript -> Lookup Target --+--------> Token X implementation
              (func/cls/var)   \
                                `------> Default (no token) implementation

Usages

When instanciated with a list of Abstraction Tokens, Lookup class allows the user to reference any Abstraction-Enabled Package available in the current namespace scope. This behavior can be generally summarized into the following:

• at miminum, a list of Abstraction Tokens is required in order to instanciate a new Lookup object.

• by default, all Abstraction-Enabled Packages imported and available at the scope where Lookup() is called, gets discovered and registered internally.

• if an package is a part of a parent package, it needs to be imported directly into the current namespace.

  # instead of
  import parent_package.my_abstracted_package
  
  # you must import it directly
  from parent_package import my_abstracted_package
  

• users can provide a dictionary of name: package to Lookup() and override the default discovery behavior. name is the alias to refer to the given package.

  import parent.my_package
  
  lookup = Lookup(*tokens, packages = {'pkg': parent.my_package})
  

• perform library lookups as if you were referencing attributes of an object.

  import my_abstracted_library
  
  lookup = Lookup(*tokens)
  
  # always start with the name of the library you want to search from
  lookup.my_abstracted_library.some_module.some_other_module.Target()
  

• the default Token Builder supports specifying mandatory tokens. This generator can be overwritten with builder argument to Lookup() (very advanced functionality).

  from genie import abstract
  from my_library import my_builder
  
  # use your default builder
  lookup = Lookup(*tokens, builder = my_builder)
  

• in addition, this global default builder setting can be modified by setting abstract.magic.DEFAULT_BUILDER to a builder of your liking. This will affect all newly created Lookup() object from this point onwards.

  from genie import abstract
  from my_library import my_default_builder
  
  # overwrite the default builder
  abstract.magic.DEFAULT_BUILDER = my_default_builder
  
  # any lookup object created hereonward will take on your builder
  lookup = Lookup(*tokens)
  

# Example
# -------
#
#   Lookup() class examples & features

# import the class from abstract
from genie.abstract import Lookup

# import any abstraction-enabled packages you need
import my_abstracted_library
from xbu_shared import genie, parser

# create the lookup object and provide it with tokens
# this auto discovers and registers the above imported packages:
#     my_abstracted_library, genie, parser
lookup = Lookup('iosxr')

# now use the lookup object and reference the above imported
# libraries using attribute queries. Eg:

result = lookup.my_abstracted_library.my_abstracted_function()
# runtime absolute path translation:
#   from my_abstracted_library.iosxr import my_abstracted_function
#   result = my_abstracted_function()

ospf = lookup.genie.conf.ospf.Ospf()
# runtime absolute path translation:
#   from xbu_shared.genie.conf.ospf.iosxr import Ospf
#   ospf = Ospf()

output = lookup.parser.ShowVersion(device = device)
# runtime absolute path translation:
#   from xbu_shared.parser.iosxr import ShowVersion
#   output = ShowVersion()

# --------------------------------------------------------------------------

# create new Lookup() instances if tokens requirements change
# you can also change the set of packages available for it,
# as well as its base reference name.
lookup = Lookup('token_a', 'token_b', '...', 'etc',
                packages = {'lib_1': my_abstracted_library,
                            'lib_2': genie',
                            'lib_3': parser})

# as new names are tokens are provided, we can now do:
result = lookup.lib_1.my_abstracted_function()
ospf = lookup.lib_2.conf.ospf.Ospf()
output = lookup.lib_3.ShowVersion(device = device)

Tip

always use meaningful package names.

Lookup Class Argument List

Argument	Description
*token	list of tokens to be used as input requirements for to this lookup
packages	dictionary of name/abstraction package to lookup from (optional)
builder	token permutation builder (optional)
**builder_kwargs	any keyword arguments/values to be passed to the builder (optional)

Integration with Topology

Lookup() class also features a classmethod constructor that enables it to understand pyATS topology module’s Device() object, and subsequently, create lookup objects based on the tokens specified under Device.custom.abstraction field.

# Example
# -------
#
#   example pyATS topology device yaml

device:
    my-example-device:
        type: router
        os: iosxe
        series: asr1k
        custom:
            abstraction:
                order: [os, series, context]
                context: yang

# Example
# -------
#
#   using the above testbed definition with abstraction

from pyats import topology
testbed = topology.loader.load('/path/to/above/testbed.yaml')
device = testbed.devices['my-example-device']

# create abstraction
from genie.abstract import Lookup

lookup = Lookup.from_device(device)
# eg, the above is equivalent to:
# os = device.custom.abstraction.get('os', device.os)
# series = device.custom.abstraction.get('series', device.series)
# context = device.custom.abstraction.get('context')
# lookup = Lookup(os, series, context)

In the above testbed YAML file, we defined a custom abstraction definition, specifying the expected token list [os, series, context], and the expected context = 'yang'.

When Lookup.from_device() method is called, the tokens associated with that device is automatically extracted following these rules:

• device.custom.abstraction is a dictionary

• device.custom.abstraction['tokens'] specifies the list of attributes to read from this device object, and converted into token values.

• the code prefers to read the attributes from device.custom.abstraction[attrbute], and falls back to device.<attribute> if needed.

All other arguments to Lookup(), such as builder, packages, builder_kwargs also applies to this classmethod.

If however you would like to not specify the device.custom.abstraction block in your testbed YAML file all the time, you can provide default_tokens as a list to Lookup.from_device(). Any tokens specified there would be looked-up from the provided device attribute.

# Example
# -------
#
#   Lookup.from_device using defaults

lookup = Lookup.from_device(device, default_tokens = ['os', 'series'])
# eg, the above is equivalent to:
# os = device.os
# series = device.serie
# lookup = Lookup(os, series)

Note

note that when using default_tokens, the lookup from device attribute is non-strict, eg: if tokens a, b, c are specified, and only a, c exists, it will not error and just use these values instead.

Tips & Tricks

Typically, abstraction should be used when the end library needs to handle differences (such as OS/Release/Mgmt Interface) etc. This leads to a per-device lookup model, where the set of abstraction-tokens per device differs. The best, pythonic method to tackle this is to follow the natural patterns of Python/pyATS programming:

• import all your packages at the top of your script/code, including all Abstraction-Enabled Packages.

• inside AEtest CommonSetup section, as soon as you have connected to your testbed devices and learnt about what they are, create your Lookup() objects and assign them as an attribute to each Device instance.

# Example
# -------
#
#   an example AEtest script with abstraction enabled

# import everything at the top
import logging
from genie import abstract
from pyats import aetest

# eg, these are my abstraction libraries
import my_abstracted_library
from xbu_shared import genie, parser

logger = logging.getLogger(__name__)

class CommonSetup(aetest.CommonSetup):

    @aetest.subsection
    def connect_to_testbed(self, testbed):
        for name, device in testbed.devices.items():
            device.connect()
            logger.info('connected to device %s' % device.name)

    @aetest.subsection
    def create_abstraction_lookup_objects(self, testbed, context):
        '''create_abstraction_lookup_objects

        Subsection to create abstraction Lookup object and assigns it to
        each corresponding device object as 'device.lib' attribute.

        In this example, we are using device object's attribute 'os', 'type'
        (from testbed YAML file) and script input parameter 'context' as
        tokens.
        '''
        for device in testbed.devices.values():
            device.lib = Lookup(device.os, device.type, context)

    # ... other subsections

# from here onwards, you can refer to libraries dynamically.

class Configure_Ospf(aetest.Testcase):

    @aetest.setup
    def setup(self, testbed):
        # iterate through all devices and configure device...
        for device in testbed.devices.values():
            device.lib.my_abstracted_library.configure_ospf(arg_1 = '...',
                                                            arg_2 = '...',
                                                            etc = '...')

    @aetest.test
    def test(self, testbed):
        for device in testbed.devices.values():
            output = device.lib.parser.ShowOspf(device = device)

            # validate values... etc
            # ...