MediaMath / t1-python

Python SDK for MediaMath Platform APIs
https://developer.mediamath.com/
Apache License 2.0
18 stars 28 forks source link
ignore-list sdk

TerminalOne-Python

image

Codacy
Badge

Codacy Badge
Coverage

Python library for MediaMath\'s APIs. This library consists of classes for working with T1 APIs and managing entities. It is written for Python 2.7 and >=3.3. Compatibility with Python 3 is made possible by bundling the module six.

API Documentation is available at [\https://apidocs.mediamath.com/\ \https://apidocs.mediamath.com/\]{.title-ref}__.

Installation

Installation is simple with pip in a virtual environment:

$ pip install TerminalOne

Alternatively, download the latest tag of the repository as a tarball or zip file and run:

$ python setup.py install

Execution and Management API

Service Object

class terminalone.T1(username=None, password=None, api_key=None, client_secret=None, auth_method=None, session_id=None, environment="production", api_base=None, token=None, token_updater=None)

The starting point for this package. Authentication and session, entity retrieval, creation, etc. are handled here. Parameters:

T1-Python includes support for resource-owner code grant. Include a client ID and secret alongside your credentials:

>>> t1 = T1(auth_method='oauth2-resourceowner', client_id="my_client_id", client_secret="my_secret", username="my@user", password="mypass")

If you already have a valid access token (e.g. by following the authorization code flow - outside this library), you can also pass it in order to get authenticated:

>>> t1 = t1 = terminalone.T1(access_token=token, environment=environment, auth_method='oauth2-existingaccesstoken', json=True)

If you have a specific API base (for instance, if you are testing against a sandbox deployment) (Note: sandbox environments are not yet useable), you can use the api_base keyword with the domain. For production endpoints, neither environment nor api_base should be provided:

>>> t1 = terminalone.T1("myusername", "mypassword", "my_api_key", api_base="myqaserver.domain.com", auth_method="cookie")

If you are receiving a (cloned) session ID, for instance the norm for apps, you will not have user credentials to log in with. Instead, provide the session ID and API key:

>>> t1 = terminalone.T1(session_id="13ea5a26e77b64e7361c7ef84910c18a8d952cf0", api_key="my_api_key")

Cookie Auth (username/password) exists, but it is not recommended for use.

>>> import terminalone
>>> t1 = terminalone.T1("myusername", "mypassword")

Fetching Entities and Collections

Entity and collection retrieval. Parameters:

T1.get(collection, entity=None, child=None, limit=None, include=None, full=None, page_limit=100, page_offset=0, sort_by="id", get_all=False, parentNone, query=None, count=False)

| Raises: terminalone.errors.ClientError if page_limit > 100, terminalone.errors.APIError on >399 HTTP status code. | Returns: If single entity is specified, returns a single entity object. If multiple entities, generator yielding each entity.

Collections

>>> advertisers = t1.get("advertisers")
>>> for advertiser in advertisers:
...     print(advertiser)
...
Advertiser(id=1, name="My Brand Advertiser", _type="advertiser")

Returns generator over the first 100 advertisers (or fewer if the user only has access to fewer), ordered ascending by ID. Each entity is the limited object, containing just id, name, and _type (_type just signifies the type returned by the API, in this case, \"advertiser\").

>>> ag_advertisers = t1.get("advertisers",
...                         limit={"agency": 123456},
...                         include="agency",
...                         full="advertiser")
>>> for advertiser in ag_advertisers:
...     print(advertiser)
...
Advertiser(id=1, name="My Brand Advertiser", agency=Agency(id=123456, name="Operating Agency", _type="agency"), agency_id=123456, status=True, ...)

Generator over up to 100 advertisers within agency ID 123456. Each advertiser includes its parent agency object as an attribute. The advertiser objects are the full entities, so all fields are returned. Agency objects are limited and have the same fields as advertisers in the previous example.

>>> campaigns, count = t1.get("campaigns",
...                           get_all=True,
...                           full=True,
...                           sort_by="-updated_on")
>>> print(count)
539
>>> for campaign in campaigns:
...     print(campaign)
Campaign(id=123, name="Summer Acquisition", updated_on=datetime.datetime(2015, 4, 4, 0, 15, 0, 0), ...)
Campaign(id=456, name="Spring Acquisition", updated_on=datetime.datetime(2015, 4, 4, 0, 10, 0, 0), ...)
...

Generator over every campaign accessible by the user, sorted in descending order of last update. Second argument is integer number of campaigns retrieved, as returned by the API. get_all=True removes the need to worry about pagination --- it is handled by the SDK internally.

>>> _, count = t1.get("advertisers",
...                   page_limit=1,
...                   count=True)
>>> print(count)
23

Sole purpose is to get the count of advertisers accessible by the user. Use page_limit=1 to minimize unnecessary resources, and assign to _ to throw away the single entity retrieved.

Searching for entities

Limiting entities by relation ID is one way to limit entities, but we can also search with more intricate queries using find:

T1.find(collection, variable, operator, candidates, **kwargs)

module terminalone.filters

>>> greens = t1.find("atomic_creatives",
...                  "name",
...                  terminalone.filters.CASE_INS_STRING,
...                  "*Green*",
...                  include="concept",
...                  get_all=True)

Generator over all creatives with \"Green\" in the name. Include concept.

>>> my_campaigns = t1.find("campaigns",
...                       "id",
...                       terminalone.filers.IN,
...                       [123, 234, 345],
...                       full=True)

Generator over campaign IDs 123, 234, and 345. Note that when using terminalone.filers.IN, variable is automatically ID, so that argument is effectively ignored. Further, candidates must be a list of integer IDs.

>>> pixels = t1.find("pixel_bundles",
...                  "keywords",
...                  terminalone.filters.NOT_NULL,
...                  None)

Generator over first 100 pixels with non-null keywords field.

>>> strats = t1.find("strategies",
...                  "status",
...                  terminalone.filters.EQUALS,
...                  True,
...                  limit={"campaign": 123456})

Active strategies within campaign ID 123456.

Entities

A specific entity can be retrieved by using get with an entity ID as the second argument, or using the entity keyword. You can then access that entity\'s properties using instance attributes:

>>> my_advertiser = t1.get("advertisers", 111111)
>>> my_advertiser.id
111111

class terminalone.Entity

: Save the entity. If data is provided, send that. Typically used with no arguments.

(Note: you will typically interact with subclasses, not ``Entity`` itself)

If for some reason you need to access the object like a dictionary (for instance, if you need to iterate over fields or dump to a CSV), the method get_properties() is available. However, you shouldn\'t modify _properties directly, as it will cause incorrect behaviour.

Once you have your instance, you can modify its values, and then save it back. A return value of None indicates success. Otherwise, an error is raised.

>>> my_advertiser.name = "Updated name"
>>> my_advertiser.save()
>>>

Create new entities by calling T1.new on your instance.

T1.new(collection, report=None, properties=None)

>>> new_properties = {
...     "name": "Spring Green",
...     "status": True,
... }
>>> new_concept = t1.new("concept", properties=new_properties)
>>> new_concept.advertiser_id = 123456
>>> new_concept.save()
>>>

properties is an optional mapping object with properties to get passed in. You can use a string representation of the object (such as "concept" above); or, you can use the object itself from terminalone.models:

>>> new_concept = t1.new(terminalone.models.Concept, properties=new_properties)
>>> 

Child Entities

To retrieve child entities (for instance, /users/:id/permissions), include the child argument in a call to T1.get:

>>> permissions = t1.get("users", 1, child="permissions")

Reports

To use MediaMath\'s Reports API instantiate an instance with T1.new:

>>> rpts = t1.new("report")

class terminalone.Report

This is a metadata object, and can be used to retrieve information about which reports are available.

>>> pprint.pprint(rpts.metadata)
{'reports': {...
             'geo': {'Description': 'Standard Geo Report',
                     'Name': 'Geo Report',
                     'URI_Data': 'https://api.mediamath.com/reporting/v1/std/geo',
                     'URI_Meta': 'https://api.mediamath.com/reporting/v1/std/geo/meta'},
...}
>>> pprint.pprint(rpts.metadata, depth=2)
{'reports': {'audience_index': {...},
             'audience_index_pixel': {...},
             'day_part': {...},
             'device_technology': {...},
             'geo': {...},
             'performance': {...},
             'pulse': {...},
             'reach_frequency': {...},
             'site_transparency': {...},
             'technology': {...},
             'video': {...},
             'watermark': {...}}}

You can retrieve the URI stub of any report by calling Report.report_uri:

>>> print(rpts.report_uri("geo"))
'geo'

Which is just a short-cut to getting the final part of the path of Report.metadata[report]['URI_Data']. Getting the URI from the specification is preferred to assuming that the name is the same as the stub. This is more directly applicable by instantiating the object for it:

>>> report = t1.new("report", rpts.report_uri("performance"))

The Reporting Service has two version of the API: /reporting/v1/std and reporting-beta/v1/std/. To call the beta version of reporting API:

>>> rpts = t1.new("report", version="beta")
>>> report = t1.new("report", rpts.report_uri("performance"), version="beta")

A short way to do it if the url is known:

>>> report = t1.new( "report", "deals?v1", version="beta" )

You can access metadata about this report from the Report.metadata property as well. To get data, first set properties about the query with Report.set, and use the Report.get method, which returns a tuple (headers, data).:

>>> report.set({
...     'dimensions': ['campaign_id', 'strategy_name'],
...     'filter': {'campaign_id': 126173},
...     'metrics': ['impressions', 'total_spend'],
...     'time_rollup': 'by_day',
...     'start_date': '2013-01-01',
...     'end_date': '2013-12-31',
...     'order': ['date'],
... })
>>> headers, data = report.get()
>>> print(headers)
['start_date', 'end_date', 'campaign_id', 'strategy_name', 'impressions']
>>> for line in data:
...     # do work on line
...     print(line)
...
['2013-06-27', '2013-06-27', '126173', 'PS', '231']
...

headers is a list of headers, while data is a csv.reader object. Type casting is not present in the current version, but is tentatively planned for a future date.

More information about these parameters can be found here.

Appendix

Why don\'t we import the object classes directly? For instance, why doesn\'t this work?

>>> from terminalone import Campaign

The answer here is that we need to keep a common session so that we can share session information across requests. This allows you to work with many objects, only passing in authentication information once.

>>> t1 = T1("myusername", "mypassword", "my_api_key")
>>> t1.authenticate("cookie")
>>> c = t1.new("campaign")
>>> c.session is t1.session
True

Contact

For questions about either API workflow or this library, email [\developers\@mediamath.com\ \mailto:developers\@mediamath.com\]{.title-ref}__.

Copyright

Copyright MediaMath 2015-2017. All rights reserved.