beda42 / prospyr

A Python client library for ProsperWorks

Prospyr

A Python client library for ProsperWorks.

Prospyr runs on Python 2.7 or Python 3.4+.

Installation

pip install prospyr

Quickstart

If you've used Django, Prospyr might feel strangely familiar.

from prospyr import connect, Person, Company

# see https://www.prosperworks.com/developer_api/token_generation to obtain
# a token.
cn = connect(email='[email protected]', token='1aefcc3...')

# collections can be ordered and sliced.
newest_person = Person.objects.order_by('-date_modified')[0]

# new records can be created.
art = Person(
    name='Art Vandelay',
    emails=[{'email': '[email protected]', 'category': 'work'}]
)
art.create()  # Art is local-only until .create() is called

# related objects can be read and assigned
art.company = Company.objects.all()[0]
art.update()

# and deleting works too.
art.delete()

Resources

The following ProsperWorks resources are supported by Prospyr:

• Account (read-only)
• Activity (read–only)
• ActivityType
• Company
• CustomerSource (read–only)
• Identifier
• Lead
• LossReason (read–only)
• Opportunity
• Person
• Pipeline (read–only)
• PipelineStage (read–only)
• Task (read–only)
• User (read–only)
• Webhook (read-only)

The following resources are not supported, but will still appear when referenced by the supported resources above. In this case, they come only with an id attribute.

• Project

Note you will receive errors trying to deal with the Lead resource if the Leads feature is not enabled in your ProsperWorks account. You can change this at Settings / Customize ProsperWorks / Lead Management.

Usage

Connecting

To connect, you'll need an email and token per token generation.

from prospyr import connect

cn = connect(email='...', token='...')

All reads are cached per–connection for five minutes. You can pass a custom cache instance when connecting to ProsperWorks to change this behaviour.

from prospyr import connect
from prospyr.cache import NoOpCache, InMemoryCache

# only cache the last request
cn = connect(email='...', token='...', cache=InMemoryCache(size=1))

# no caching
cn = connect(email='...', token='...', cache=NoOpCache())

You can also substitute your own custom cache here to use e.g. Redis or memcached.

Prospyr also supports multiple named connections. Provide a name='...' argument when calling connect() and refer to the connection when interacting with the API later, e.g. Person.objects.get(id=1, using='...').

Create

You can create new records in ProsperWorks.

from prospyr import Person

steve = Person(
    name='Steve Cognito',
    emails=[{'category': 'work', 'email': '[email protected]'}]
)

# steve only exists locally at this stage
steve.id
>>> None

# now he exists remotely too
steve.create()
>>> True
steve.id
>>> 1

Read

There are two ways to read a single record from ProsperWorks. A new instance can be fetched using the resource's objects.get() method, or you can call read() on an existing instance to have its attributes refreshed.

from prospyr import Person

# a new instance
steve = Person.objects.get(id=1)
steve.name
>>> 'Steve Cognito'

# update an existing instance
steve = Person(id=1)
steve.read()
>>> True
steve.name
>>> 'Steve Cognito'

# as a special case, People can be read by email as well as ID:
steve = Person.objects.get(email='[email protected]')

Update

Note that “update” means to push an update to ProsperWorks using your local data, rather than to refresh local data using ProsperWorks. In this example, Steve is fetched from ProsperWorks and given a new title. Hey, congrats on the promotion Steve.

from prospyr import Person

steve = Person.objects.get(id=1)
steve.title = 'Chairman'
steve.update()
>>> True

Delete

When Steve has reached the end of his useful lifespan, he can be deleted too.

from prospyr import Person

steve = Person.objects.get(id=1)
steve.delete()
>>> True

Ordering

Resource collections can be ordered. Check the ProsperWorks API documentation to learn which fields can be ordered. However, Prospyr does check that the fields you argue are correct.

from prospyr import Person

# oldest first
rs = Person.objects.order_by('date_modified')

# newest first (note the hyphen)
rs = Person.objects.order_by('-date_modified')

# At this stage, no requests have been made. Results are lazily evaluated
# and paging is handled transparently.

# The results can be indexed and sliced like a Python list. Doing so forces
# evaluation. The below causes the first page of results to be fetched.
rs[0]
>>> <Person: Steve Cognito>

# No request is required here, as the Bones was on the first page requested
# above. The default page size is 200.
rs[1]
>>> <Person: Bones Johannson>

# This result is on the second page, so another request is fired.
rs[200]
>>> <Person: Alfons Tundra>

Once ResultSet instances have been evaluated they are cached for their lifetime. However, the filter() and order_by() methods return new ResultSet instances which require fresh evaluation. While you are dealing with a single ResultSet, it is safe to iterate and slice it as many times as necessary.

Filtering

Resource collections can be filtered. Check the ProsperWorks API documentation to learn which filters can be used. Prospyr does not currently validate your filter arguments, and note that ProsperWorks does not either; if you make an invalid filter argument, results will be returned as though you had not filtered at all.

Multiple filters are logically ANDed together. A single call to filter() with many parameters is equivalent to many calls with single parameters.

from prospyr import Company

active = Company.objects.filter(minimum_interaction_count=10)
active_in_china = active.filter(country='CN')

# this is equivalent
active_in_china = Company.objects.filter(
    minimum_interaction_count=10,
    country='CN'
)

As with ordering, filtered results are evaluated lazily and then cached indefinitely. Re-ordering or re-filtering results in a new ResultSet which requires fresh evaluation.

ProsperWorks' “Secondary Resources”, such as Pipeline Stages, cannot be filtered or ordered. These resources use ListSet rather than ResultSet instances; these only support the all() method:

from prospyr import PipelineStage

PipelineStage.objects.all()
>>> <ListSet: Qualifying, Quoted, ...>

Related resources

The ProsperWorks API allows querying related objects for some records, such as all people for a company, etc. To use it in prospyr, you would use corresponding attributes on the relevant instance, such as company.people. These are implemented as managers (similar to .objects), so you need to run .all() on them.

Because the API does not return full records for related objects, but only their IDs, it might be expensive to fetch all the objects - it costs n+1 requests to get n objects. To make it possible to skip this overhead when the actual content of the related objects is not necessary, it is possible to pass lazy_instances keyword argument to .all() and the returned object will only contain the corresponding ID. It is possible to fill the object with data later on using its .read() method.

There is also a convenience .count() method which returns the number of related records as integer.

Example:

for company in Company.objects.all():
    print(company.name)
    # the following does 1 request for opportunities and 1 request for people
    print("Opportunities: {0}, persons: {1}".format(company.opportunities.all().count(),
                                                    company.people.all().count()))
    for person in company.people.all(lazy_instances=True):  # one request for the list
        print(person.id)  # the person does not have anything else than .id
        person.read()     # fetch person data - 1 request
        print(person)

If you just iterate over the result of .all() without the lazy_instances attr being set to True, one request will be fired for each iteration. However, it is done lazily, so if you prematurely stop the loop, it will save you some requests.

Example:

for person in company.people.all():  # 1 request for the list, 1 per person
    print(person)

The related resources are as follows:

• Company - people, opportunities, tasks
• Person - companies, opportunities, tasks
• Opportunity - companies, people, tasks
• Task - companies, people, opportunities, leads

Account

The Account resource represents the ProsperWorks account which you are currently working with. The name of the account can be read like so:

from prospyr import Account

account = Account.objects.get()
account.name
>>> 'So-and-so Company'

Collection Error Handling

Prospyr validates data delivered from ProsperWorks when building representative Python objects for local use. Because there are no documented details on the validation that ProsperWorks itself uses, Prospyr's validation rules are sometimes incorrect or more strict than necessary. The author suspects that sometimes ProsperWorks also delivers data that is simply invalid.

This can cause exceptions to be raised when iterating over result sets (e.g. for person in Person.objects.all()...) which prevent the remainder of the collection from being accessed.

To make your life easier while such a mismatch is corrected in Prospyr, you can choose to have these validation errors collected instead of being raised:

from prospyr import Person

errs = []
for person in Person.objects.store_invalid(errs).all():
    # ...

if errs:
    # handle errors

The argument to store_invalid must, like a list, have a working append method. It will be filled with ValidationError instances which each have errors, raw_data and resource_cls attributes.

If your use–case allows you to correct the problem in raw_data, you can recover like so:

for err in errs:
    good_data = make_corrections(err.raw_data)
    instance = err.resource_cls.from_api_data(good_data)

Tests

pip install -r dev-requirements

# test using the current python interpreter
make test

# test with all supported interpreters
tox