Writing an observation module to interface with observatories ============================================================= This guide will walk you through how to create a custom observation facility module using some mocked up endpoints to simulate a real observatory interface. It will also provide information on creating a custom manual observation facility for tracking observations that were not created through an API. You can use this example as the foundation to build an observing facility module to connect to a real observatory or track observations on non-API supported facilities. Be sure you’ve followed the :doc:`Getting Started ` guide before continuing onto this tutorial. What is a observing facility module? ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ A TOM Toolkit observing facility module is a python module which contains the code necessary to provide an interface to an observing facility in a TOM. Some examples of existing modules are the `Las Cumbres Observatory `__ and the `Gemini `__ modules. Both allow the submission of observation requests to their respective observatories through a TOM. Prerequisites ~~~~~~~~~~~~~ You should have a working TOM already. You can start where the :doc:`Getting Started ` guide leaves off. You should also be familiar with the observing facility’s API that you would like to work with. .. tip:: Read these first! The following Python/Django concepts are used in this tutorial. While this tutorial does not assume familiarity with the concepts, you will likely find the tutorial easier to understand and build upon if you read these in advance. - `Working with Django Forms `_ - `Requests Official API Docs `_ Creating a Custom Facility ~~~~~~~~~~~~~~~~~~~~~~~~~~ Defining the minimal implementation ----------------------------------- Within any existing module in your TOM you should create a new python module (file) named ``myfacility.py``. For example, if you have a fresh TOM installation you’ll have a directory structure that looks something like this: :: ├── data ├── db.sqlite3 ├── manage.py ├── mytom │ ├── __init__.py │ ├── settings.py │ ├── urls.py │ └── wsgi.py ├── static ├── templates └── tmp We’ll place our ``myfacility.py`` file inside the ``mytom`` directory, next to ``settings.py``. For now, copy the following lines into ``myfacility.py``: .. code:: python from tom_observations.facility import BaseRoboticObservationFacility, BaseRoboticObservationForm class MyObservationFacilityForm(BaseRoboticObservationForm): pass class MyObservationFacility(BaseRoboticObservationFacility): name = 'MyFacility' observation_types = [('OBSERVATION', 'Custom Observation')] We’ll go over what these lines mean soon. First, we’ll add a setting to our project’s ``settings.py`` to tell the TOM Toolkit to use our new class: .. code:: python TOM_FACILITY_CLASSES = [ 'tom_observations.facilities.lco.LCOFacility', 'tom_observations.facilities.gemini.GEMFacility', 'tom_observations.facilities.soar.SOARFacility', 'tom_observations.facilities.lt.LTFacility' 'mytom.myfacility.MyObservationFacility' ] Now go ahead and view a target in your TOM, you should see something like this: |image0| This means our new observation facility module has been successfully loaded. BaseRoboticObservationFacility and BaseRoboticObservationForm ------------------------------------------------------------- You will have noticed our module consists of two classes that inherit from two other classes. ``MyObservationFacility`` is the class that will contain the “business logic” for interacting with the remote observatory. This includes methods to submit observations, check observation status, etc. It inherits from ``BaseRoboticObservationFacility``, which contains some functionality that all observation facility classes will want. You can access the user within your facility implementation using ``self.user`` if you need it for any api requests. ``MyObservationFacilityForm`` is the class that will display a GUI form for our users to create an observation. We can submit observations programmatically, but it is also nice to have a GUI for our users to use. The ``BaseRoboticObservationForm`` class, just like the previous super class, contains logic and layout that all observation facility form classes should contain. Implementing observation submission ----------------------------------- Try to click on the button for ``MyFacility``. It should return an error that says everything it’s missing: :: Can't instantiate abstract class MyObservationFacility with abstract methods data_products, get_form, get_observation_status, get_observation_url, get_observing_sites, get_terminal_observing_states, submit_observation, validate_observation To start, let’s define new functions in ``MyObservationFacility`` for each missing function like so: .. code:: python class MyObservationFacility(BaseRoboticObservationFacility): name = 'MyFacility' observation_forms = { 'OBSERVATION': MyObservationFacilityForm } def data_products(self): return def get_form(self): return ... Reload the server, click the ``MyFacility`` button, and you should get . . . a different error! Progress! :: get_form() takes 1 positional argument but 2 were given To fix up ``get_form``, adjust it to: .. code:: python def get_form(self, observation_type): return MyObservationFacilityForm Reload the page and now it should look something like this: |image1| :Some notes: #. The form is empty, but we’ll fix that next. #. The ``name`` variable of ``MyObservationFacility`` determines what the top of the page says (``Submit an observation to MyFacility``). It also determines the name of the button under “Observe” on the target’s page. #. You should see a tab for ``Custom Observation`` as the only option on the page. This is read from the ``observation_forms`` variable in ``MyObservationFacility``. That variable is a dict. The value of each dict item is the observation form class. The key of each dict item is what should be used to distinguish different observation types in your code, which will be displayed in Pascal Case in the observation form tabs. To see a demonstration of this, check out the `Las Cumbres Observatory `__ facility’s ``observation_forms`` and ``get_form``. Now let’s populate the form. Let’s assume our observatory only requires us to send 2 parameters (besides the target data): ``exposure_time`` and ``exposure_count``. Let’s start by adding them to our form class: .. code:: python from crispy_forms.layout import Layout from django import forms from tom_observations.facility import BaseRoboticObservationFacility, BaseRoboticObservationForm class MyObservationFacilityForm(BaseRoboticObservationForm): exposure_time = forms.IntegerField() exposure_count = forms.IntegerField() def layout(self): return Layout( 'exposure_time', 'exposure_count' ) Notice that we’ve added the two field definitions on our form. We’ve also imported the django form module with ``from django import forms`` and a crispy_forms class with ``from crispy_forms.layout import Layout``. Finally, we've defined a function ``layout(self)`` that is used to display the fields that we've created. All fields must be named in the ``layout`` function in order to be displayed, and the ``layout`` function is also how we could make the layout more sophisticated. See the `django-crispy-forms documentation `__ and the `lco.py module `__ for examples. Now if we reload the page, we should see something like this: |image2| This is progress, but remember that most of the functions in ``MyObservationFacility`` have blank return statements. Next we’ll implement the methods that perform actions with our form when we submit the observation request: .. code:: python from crispy_forms.layout import Layout from django import forms from tom_observations.facility import BaseRoboticObservationFacility, BaseRoboticObservationForm class MyObservationFacilityForm(BaseRoboticObservationForm): exposure_time = forms.IntegerField() exposure_count = forms.IntegerField() class MyObservationFacility(BaseRoboticObservationFacility): name = 'MyFacility' observation_types = observation_forms = { 'OBSERVATION': MyObservationFacilityForm } def data_products(self, observation_id, product_id=None): return [] def get_form(self, observation_type): return MyObservationFacilityForm def get_observation_status(self, observation_id): return ['IN_PROGRESS'] def get_observation_url(self, observation_id): return '' def get_observing_sites(self): return {} def get_terminal_observing_states(self): return ['IN_PROGRESS', 'COMPLETED'] def submit_observation(self, observation_payload): print(observation_payload) return [1] def validate_observation(self, observation_payload): pass The important method here is ``submit_observation``. This method, when implemented fully, will send the observation payload to the remote observatory and then return a list of observation ids. Those ids will be stored in the database to be used later, in methods like ``get_observation_status(self, observation_id)``. In our dummy implementation, we simply print out the observation payload and return a single fake id with ``return [1]``. If you now “submit” an observation using the MyFacility module, you should see this in the server console: :: {'target_id': 1, 'params': '{"facility": "MyFacility", "target_id": 1, "observation_type": "(\'OBSERVATION\', \'Custom Observation\')", "exposure_time": 100, "exposure_count": 2}'} That was our print statement! Additionally, you should see ``1 upcoming observation`` on the target’s page, and if you navigate to its “Observations” tab you can see the parameters of the observation you just submitted in more detail. Filling in the rest of the functionality ---------------------------------------- You’ll notice we added many more methods other than ``submit_observation`` to our Facility class. For now they return dummy data, but when you adapt it to work with a real observatory you should fill them in with the correct logic so that the whole module works correctly with the TOM. You can view explanations of each method in the :doc:`Facility Modules <../api/tom_observations/facilities>` section of the documentation or `in the source code `__ Adding Sites to new Facilities ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ In our ``MyObservationFacility`` class, let’s define a new variable called ``SITES``. Modeling our ``SITES`` on the one defined for `Las Cumbres Observatory `__, we can easily put new sites to our facility that will then show up in the airmass plots: .. code:: python class MyObservationFacility(BaseRoboticObservationFacility): name = 'MyFacility' observation_types = observation_forms = { 'OBSERVATION': MyObservationFacilityForm } SITES = { 'Itagaki': { 'latitude': 38.188020, 'longitude': 140.335113, 'elevation': 350 } } ... def get_observing_sites(self): return self.SITES (Koichi Itagaki is an “amateur” astronomer in Japan who has discovered many extremely interesting supernovae.) Airmass plotting for new facilities ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Now that your facility has sites, they should appear on airmass plots. If you input two dates into the “Plan” form under the “Observe” tab on a target’s page, you’ll see the target’s visibility. By default, the plot shows you the airmass at LCO and Gemini sites. To remove these, you can delete these facilities from ``TOM_FACILITY_CLASSES`` in ``settings.py``. Even if the facilities you observe at are not API-accessible, you can still add them to your TOM’s airmass plots to judge what targets to observe when. Happy developing! .. |image0| image:: /_static/observation_module/myfacility.png .. |image1| image:: /_static/observation_module/empty_form.png .. |image2| image:: /_static/observation_module/fields.png