Super thin Object Document Mapper for MongoDB written in Python
Python Shell
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


PyPI version Build Status

Tavi (as in Rikki Tikki Tavi) is an extremely thin Mongo object mapper for Python. It is a thin abstraction over pymongo that allows you to easily model your applications and persist your data in MongoDB.

Getting Started

Install using pip:

pip install Tavi

or clone the project and run:

python install

Define your models:

import tavi

class Product(tavi.documents.Document):
    name        = tavi.fields.StringField("name", required=True)
    sku         = tavi.fields.StringField("sku", required=True)
    description = tavi.fields.StringField("description", required=True)
    price       = tavi.fields.FloatField("price", min_value=0)

Use your models to create and find documents:

>>> product = Product(name="Spam", sku="123", price=2.99)

>>> product.valid

>>> product.errors.full_messages
['Description is required']

>>> product.description = "A tasty canned precooked meat product."

>>> product.valid


>>> for p in Product.find():
...     print

For more details see Using Tavi.


  • pymongo >= 2.5.2
  • inflection >= 0.2.0

Using Tavi

Connecting to MongoDB

MongoDB connections are handled by the tavi.Connection class (which delegates to pymongo). Set up a connection like this:

import tavi

tavi.Connection.setup("my_test_database", host="localhost", port=27017)

Alternatively, you can also use the MongoDB URI format:

import tavi

tavi.Connection.setup("my_test_database", host="mongodb://localhost:27017/")

Defining Documents

Documents are the building blocks for defining your models. An instantiated tavi.documents.Document class represents a single document in a MongoDB collection. It also provides a number of class methods used for querying the collection itself. You can embed documents inside other documents (rather than in their own collections) using the tavi.documents.EmbeddedDocument class.


Document objects inherit from tavi.documents.Document. You can persist them to collections and they can contain embedded documents. They also come with support for validations and querying.

import tavi

class Order(tavi.documents.Document):
    name     = tavi.fields.StringField("name", required=True)
    address  = tavi.fields.EmbeddedField("address", Address)
    email    = tavi.fields.StringField("email", required=True)
    pay_type = tavi.fields.StringField("pay_type", required=True, default="Mastercard")

Note: The collection name that is stored in Mongo is derived from the class name. In the example above, the collection in Mongo would be named orders. For the initial version of Tavi the collection name is not customizable. I have plans to support it in a future version, however.

Document objects are initialized like any other object. Additionally, they may be initialized with keyword arguments that set the field values.

>>> order = Order(name="My Order", email="")

"My Order"


>>> order.address

>>> order.pay_type

Any fields that are omitted from the list of keyword arguments are set to either None or their default value, if provided. Any keyword argument that is not a valid field is simply ignored.

Document objects have several attributes for retrieving information about them:

  • #bson_id returns the ID of the document; this value can also be retrieved using #_id
  • #collection_name returns the name of the collection
  • #data returns a dictionary that contains all the fields and their values
  • #errors returns a tavi.errors.Errors object (see validations for more info)
  • #fields returns the list of fields defined for the document


Document objects can be (de-)serialized from/to JSON. Under the hood it delegates to pymongo's bson.json_util. The #to_json and #from_json methods convert to JSON and from JSON, respectively. In addition, the #to_json instance method can be given an optional array of fields to convert to JSON. By default, all fields are serialized.

import tavi

class Order(tavi.documents.Document):
    name     = tavi.fields.StringField("name", required=True)
    address  = tavi.fields.EmbeddedField("address", Address)
    email    = tavi.fields.StringField("email", required=True)
    pay_type = tavi.fields.StringField("pay_type", required=True, default="Mastercard")
>>> order = Order(name="My Order", email="", pay_type="Visa")

>>> order.to_json(["name", "email"])
... '{"name": "My Order", "email": ""}'

Embedded Documents

Embedded documents are almost identical to Documents with one exception: they are saved inside of another document instead of in their own collection. They inherit from tavi.documents.EmbeddedDocument and have support for validations.

import tavi

class Address(tavi.documents.EmbeddedDocument):
    street      = tavi.fields.StringField("street")
    city        = tavi.fields.StringField("city")
    state       = tavi.fields.StringField("state")
    postal_code = tavi.fields.StringField("postal_code")


Fields are how Tavi maps the attributes in your objects to attributes in the document for your collections in MongoDB. All fields inherit from tavi.base.fields.BaseField which provides some common validations.

Fields are initialized with a name, which will be used when persisting the document to Mongo. This name can be different from the class's field name, for example:

import tavi

class User(tavi.documents.Document):
    email  = tavi.fields.StringField("email")
    status = tavi.fields.StringField("my_status")

In the example above, you would refer to the class attribute status, but when the document is persisted my_status will be used as the name. This is useful if you have to support an existing Mongo database that has a different field name than the one you would like to use in your Python class.

Basic Fields

There are several field types supported:

tavi.fields.DateTimeField: Represents a naive datetime for a Mongo Document.

tavi.fields.FloatField: Represents a floating point number for a Mongo Document. Supports the following additional validations:

  • min_value: validates the minimum value the field value
  • max_value: validates the maximum value the field value

tavi.fields.IntegerField: Represents a integer number for a Mongo Document. Supports the following additional validations:

  • min_value: validates the minimum value the field value
  • max_value: validates the maximum value the field value

tavi.fields.StringField: Represents a String field for a Mongo Document. Supports the following additional validations:

  • length : validates the field value has an exact length; default is None
  • min_length: ensures field has a minimum number of characters; default is None
  • max_length: ensures field is not more than a maximum number of characters; default is None
  • pattern : validates the field matches the given regular expression pattern; default is None

Note that leading and trailing whitespace is automatically stripped from StringField values.

Custom Fields

If you need to add your own field types you may inherit from either tavi.base.fields.BaseField or one of the other field types. Any classes that inherit from tavi.base.fields.BaseField must implement the #validate method and call #super in order for validations to work. For example:

    class MyCustomField(tavi.base.field.BaseField):
        def validate(self, instance, value):
            super(MyCustomField, self).validate(instance, value)
            # Your validation logic goes here...

Array Fields

tavi.fields.ArrayFields are used for fields where the value must be a list of primitive items (numbers, strings, etc., as opposed to embedded documents).

import tavi

def validate(field, document, item):
    if not isinstance(item, basestring):
        document.errors.add(, "values must be strings")

class MontyPython(Document):
    names = tavi.fields.ArrayField("names", validate_item=validate)

p = MontyPython()
p.names = ["Graham Chapman", "Eric Idle", "John Cleese", "Michael Palin",
           "Terry Jones"]
p.names.append("Terry Gilliam")

Embedded Fields

tavi.fields.EmbeddedField's are how embedded documents are placed in documents. For example, let's say we have defined an embedded document for an address.

import tavi

class Address(tavi.documents.EmbeddedDocument):
    street      = tavi.fields.StringField("street")
    city        = tavi.fields.StringField("city")
    state       = tavi.fields.StringField("state")
    postal_code = tavi.fields.StringField("postal_code")

This embedded document can be placed into a user document using a tavi.fields.EmbeddedField.

class User(tavi.documents.Document):
    name    = tavi.fields.StringField("name")
    address = tavi.fields.EmbeddedField("address", Address)

The address field can now be accessed through the user object...

user = User()
user.address.street = "123 Elm Street" = "Anywhere"
user.address.state = "NY"
user.address.postal_code = "00000"

...and when the user is saved, the address is persisted along with it.

Embedded List Fields

tavi.fields.ListFields are used for embedding a list of Embedded fields. For example:

import tavi

class OrderLine(tavi.documents.EmbeddedDocument):
    quantity    = tavi.fields.IntegerField("quantity")
    total_price = tavi.fields.FloatField("total_price")

class Order(Document):
    name        = tavi.fields.StringField("name")
    address     = tavi.fields.EmbeddedField("address", Address)
    email       = tavi.fields.StringField("email")
    pay_type    = tavi.fields.StringField("pay_type")
    order_lines = tavi.fields.ListField("order_lines")

In the above example, OrderLine is an EmbeddedDocument and Order is it's container Document. An OrderLine object can be appended to an Order like this:

>>> order = Order(
...     name     = "John Doe",
...     email    = "",
...     pay_type = "Mastercard"
... )

>>> line_a = OrderLine(quantity=1, total_price=19.99)
>>> line_b = OrderLine(quantity=3, total_price=39.99)

>>> order.order_lines.append(line_a)
>>> order.order_lines.append(line_b)

When order is saved, it's order_lines are persisted as an array in the document.

>>> order.order_lines[0].price


Document objects support field validations through two attributes:

#valid: returns True or False indicating if all field validations are met #errors: returns a tavi.errors.Errors object that holds information about all field errors

tavi.errors.Errors is a dictionary-like object with the following interface:

import tavi

>>> errors = tavi.errors.Errors()

>>> errors.add("email", "is required")

>>> errors.get("email")
"is required"

>>> errors.full_messages_for("email")
"Email is required"

>>> errors.full_messages
["Email is required"]

>>> errors.count

>>> errors.clear

>>> errors.count

In practice, fields will handle adding and clearing errors themselves.

import tavi

class User(tavi.documents.Document)
    email = StringField("email", required=True)
>>> user = User()

>>> user.valid

>>> user.errors.get("email")
"is required"

>>> user.errors.full_messages_for("email")
"Email is required"

>>> user.errors.full_messages
["Email is required"]

>>> = ""

>>> user.valid

>>> user.errors.count

Knowing how tavi.errors.Errors works is useful when you need to define your own custom fields.

Specifying Validations

All fields that inherit from tavi.base.fields.BaseField support the following validations:

required: indicates if the field is required; default is False

default: default value for the field; None if not given

choices: validates field value is a member of specified list

persist: boolean indicating if field should be persisted to Mongo; default is True

Here are some examples:

import tavi

class User(tavi.documents.Document):
    email    = tavi.fields.StringField("email", required=True)
    age      = tavi.fields.IntegerField("age", default=0)
    pay_type = tavi.fields.StringField("pay_type", choices=["Mastercard", "Visa"])
    password = tavi.fields.StringField("password", persist=False)

Refer to Basic Fields for a list of field types and their validations.


Saving Documents

Document objects are persisted using the #save method. This method inserts the document into the collection if it does not exist or updates it if it does. The method returns True if the save was successful. Before performing the save, it ensures that the Document is valid and returns False if it was not.

If the document object has a field named created_at, this field's value will be set to the current time when the document is inserted. Also, if a field named last_modified_at is defined, this value will be set when the document is either inserted or updated.

Finding Documents

Document objects can be retrieved using finder classmethods. There are two main finder methods: #find and #find_one. These are wrappers around the pymongo #find and #find_one methods and support all the same arguments. The difference is these methods wrap the return result into a Document object.

It is important to note that when using these methods, if you restrict the fields that are returned, the resulting document object(s) will have these fields set to None. If you later try to persist one of these objects, you will overwrite the value of the field. Therefore I recommend that you use the collection directly and have it return a dictionary result set.

Document objects also support two convenience finder methods: #find_by_id and #find_all which delegate to #find_one and #find, respectively.

You may also want to define your own custom finder methods. I recommend you delegate to the main finder methods like this:

import tavi

class User(tavi.documents.Document):
    email     = tavi.fields.StringField("email")
    last_name = tavi.fields.StringField("last_name")

    def find_by_email(cls, email)
        """Example custom finder."""
        return cls.find_one({"email": email})

    def find_by_last_name(cls, last_name)
        """Example custom finder."""
        return cls.find({"last_name": last_name})

This way you will not have to wrap the results as document objects, since it will be done for you.

Document objects also support a #count method that will return the total number of documents in the collection.

Deleting Documents

Document objects may be removed from the collection using the #delete method. There is no support for undoing this operation.


Tavi defines several custom exceptions:

TaviError: Base error for all Tavi exceptions. If you need to define your own exceptions you may inherit from this class.

TaviTypeError: Raised when an operation or function is applied to an object of inappropriate type. For example, this error will be raised if you try to add an object that does not derive from tavi.documents.EmbeddedDocument to a tavi.fields.EmbeddedField.

TaviConnectionError: Raised when Tavi cannot connect to Mongo.

Using pymongo

Tavi is just a thin wrapper for pymongo. When you need to work with pymongo directly, Tavi has a couple of convenience features to help you out.

Once a connection has been established, you can grab a handle to the database directly, like this:


The database object that is returned is a pymongo database and supports all of its methods.

You can access a collection directly from a document model. Given a document model:

import tavi

class User(tavi.documents.Document):
    email =      tavi.fields.StringField("email")
    first_name = tavi.fields.StringField("first_name")
    last_name =  tavi.fields.StringField("last_name")

If you call


You will get back a handle to the users collection. This collection is a pymongo collection and supports all of it's methods.


Environment Setup

Clone the repo, cd into folder then:

virtualenv --no-site-packages env
. env/bin/activate
pip install -r requirements.txt

Run the tests using:

python nosetests

Flake8 is used for tracking PEP8 compliance, cyclomatic complexity, etc. Run it using:

flake8 tavi

Releasing a New Version

  1. Increment version in
  2. Update CHANGES.txt
  3. Commit and tag version (ie. git tag -a 1.4 -m 'Version 1.4')
  4. Push tags (ie. git push origin 1.4)
  5. Run python sdist upload

Issues / Roadmap

Use GitHub issues for reporting bugs and feature requests. This library is meant to be lightweight so I probably won't be adding many features but feel free to submit pull requests for any critical features you think may be missing.

Patches / Pull Requests

  • Fork the project.
  • Make your feature addition or bug fix.
  • Add tests for it. This is important so I don't break it in a future version unintentionally.
  • Commit, do not mess with version or history (if you want to have your own version that is fine, but please bump version in a commit by itself I can ignore when I pull).
  • Send me a pull request. Bonus points for topic branches.