SDK for the Seam API written in Python.
Seam makes it easy to integrate IoT devices with your applications. This is an official SDK for the Seam API. Please refer to the official Seam Docs to get started.
Parts of this SDK are generated from always up-to-date type information provided by @seamapi/types node package. This ensures all API methods, request shapes, and response shapes are accurate and fully typed.
- Installation
- Usage
- Development and Testing
- GitHub Actions
- Contributing
- License
- Warranty
This package is registered on the Python Package Index (PyPI) as seam.
Install it with:
$ pip install seam
Note: These examples assume `SEAM_API_KEY` is set in your environment.
from seam import Seam
seam = Seam()
devices = seam.devices.list()from seam import Seam
seam = Seam()
lock = seam.locks.get(name="Front Door")
seam.locks.unlock_door(device_id=lock.device_id)The SDK supports API key and personal access token authentication mechanisms.
Authentication may be configured by passing the corresponding options directly to the Seam constructor, or with the more ergonomic static factory methods.
An API key is scoped to a single workspace and should only be used on the server. Obtain one from the Seam Console.
# Set the `SEAM_API_KEY` environment variable
seam = Seam()
# Pass as the first argument to the constructor
seam = Seam("your-api-key")
# Pass as a keyword argument to the constructor
seam = Seam(api_key="your-api-key")
# Use the factory method
seam = Seam.from_api_key("your-api-key")A Personal Access Token is scoped to a Seam Console user. Obtain one from the Seam Console. A workspace ID must be provided when using this method and all requests will be scoped to that workspace.
# Set the `SEAM_PERSONAL_ACCESS_TOKEN` and `SEAM_WORKSPACE_ID` environment variables
seam = Seam()
# Pass as an option to the constructor
seam = Seam(
personal_access_token="your-personal-access-token",
workspace_id="your-workspace-id",
)
# Use the factory method
seam = Seam.from_personal_access_token(
"your-personal-access-token",
"your-workspace-id",
)Some asynchronous operations, e.g., unlocking a door, return an action attempt. Seam tracks the progress of the requested operation and updates the action attempt when it succeeds or fails.
To make working with action attempts more convenient for applications,
this library provides the wait_for_action_attempt option and enables it by default.
When the wait_for_action_attempt option is enabled, the SDK:
- Polls the action attempt up to the
timeoutat thepolling_interval(both in seconds). - Resolves with a fresh copy of the successful action attempt.
- Raises a
SeamActionAttemptFailedErrorif the action attempt is unsuccessful. - Raises a
SeamActionAttemptTimeoutErrorif the action attempt is still pending when thetimeoutis reached. - Both errors expose an
action_attemptproperty.
If you already have an action attempt ID and want to wait for it to resolve, simply use
seam.action_attempts.get(action_attempt_id=action_attempt_id)Or, to get the current state of an action attempt by ID without waiting,
seam.action_attempts.get(
action_attempt_id=action_attempt_id,
wait_for_action_attempt=False,
)To disable this behavior, set the default option for the client:
seam = Seam(
api_key="your-api-key",
wait_for_action_attempt=False,
)
seam.locks.unlock_door(device_id=device_id)or the behavior may be configured per-request:
seam.locks.unlock_door(
device_id=device_id,
wait_for_action_attempt=False,
)The polling_interval and timeout may be configured for the client or per-request.
For example:
from seam import Seam, SeamActionAttemptFailedError, SeamActionAttemptTimeoutError
seam = Seam("your-api-key")
lock = seam.locks.list()
if len(locks) == 0:
raise Exception("No locks in this workspace")
lock = locks[0]
try:
seam.locks.unlock_door(
device_id=lock.device_id,
wait_for_action_attempt={
"timeout": 5.0,
"polling_interval": 1.0,
},
)
print("Door unlocked")
except SeamActionAttemptFailedError as e:
print("Could not unlock the door")
except SeamActionAttemptTimeoutError as e:
print("Door took too long to unlock")The Seam API tells an omitted param apart from one explicitly set to null. In an update request, an omitted param leaves the current value unchanged, while a null param unsets it.
Python has a single nil value None which represents an undefined parameter. This SDK provides an explicit null value to send in requests.
A param set to None is omitted, and a param set to NULL is sent as null:
from seam import NULL, Seam
seam = Seam()
# Leaves the name unchanged.
seam.devices.update(device_id="your-device-id", name=None)
# Unsets the name.
seam.devices.update(device_id="your-device-id", name=NULL)Because unsetting a value cannot be undone, None means the safe option of
omitting the param, and sending null is always explicit.
This is why a param is never sent as null by default,
even though None is the natural way to spell null in Python.
NULL behaves the same way in a request body and in a URL search param.
Its type is exported as Null for annotating your own code:
from typing import Optional, Union
from seam import NULL, Null
name: Optional[Union[str, Null]] = NULLOnly params the Seam API documents as nullable accept NULL.
The generated method signatures say which ones those are,
so a type checker rejects NULL anywhere else:
# name is nullable, so it may be unset.
seam.devices.update(device_id="your-device-id", name=NULL)
# is_managed is not, so this fails the type check.
seam.devices.update(device_id="your-device-id", is_managed=NULL)Some Seam API endpoints that return lists of resources support pagination.
Use the SeamPaginator class to fetch and process resources across multiple pages.
from seam import Seam
seam = Seam()
paginator = seam.create_paginator(seam.devices.list, {"limit": 20})
devices, pagination = paginator.first_page()
if pagination.has_next_page:
more_devices, _ = paginator.next_page(pagination.next_page_cursor)Get the first page on initial load and store the state (e.g., in memory or a file):
import json
from seam import Seam
seam = Seam()
params = {"limit": 20}
paginator = seam.create_paginator(seam.devices.list, params)
devices, pagination = paginator.first_page()
# Example: Store state for later use (e.g., in a file or database)
pagination_state = {
"params": params,
"next_page_cursor": pagination.next_page_cursor,
"has_next_page": pagination.has_next_page,
}
with open("/tmp/seam_devices_list.json", "w") as f:
json.dump(pagination_state, f)Get the next page at a later time using the stored state:
import json
from seam import Seam
seam = Seam()
# Example: Load state from where it was stored
with open("/tmp/seam_devices_list.json", "r") as f:
pagination_state = json.load(f)
if pagination_state.get("has_next_page"):
paginator = seam.create_paginator(
seam.devices.list, pagination_state["params"]
)
more_devices, _ = paginator.next_page(
pagination_state["next_page_cursor"]
)from seam import Seam
seam = Seam()
paginator = seam.create_paginator(seam.devices.list, {"limit": 20})
for account in paginator.flatten():
print(account.account_type_display_name)from seam import Seam
seam = Seam()
paginator = seam.create_paginator(seam.devices.list, {"limit": 20})
all_devices = paginator.flatten_to_list()Some Seam API endpoints do not require a workspace in scope.
The SeamWithoutWorkspace client is not bound to a specific workspace
and may use those endpoints with an appropriate authentication method.
A Personal Access Token is scoped to a Seam Console user. Obtain one from the Seam Console.
from seam import SeamWithoutWorkspace
# Set the `SEAM_PERSONAL_ACCESS_TOKEN` environment variable
seam = SeamWithoutWorkspace()
# Pass as an option to the constructor
seam = SeamWithoutWorkspace(personal_access_token="your-personal-access-token")
# Use the factory method
seam = SeamWithoutWorkspace.from_personal_access_token("your-personal-access-token")
# List workspaces authorized for this Personal Access Token
workspaces = seam.workspaces.list()The Seam API implements webhooks using Svix.
This SDK exports a thin wrapper SeamWebhook around the svix package.
Use it to parse and validate Seam webhook events.
Refer to the Svix docs on Consuming Webhooks for an in-depth guide on best-practices for handling webhooks in your application.
This example is for Flask, see the Svix docs for more examples in specific frameworks.
import os
from flask import Flask, request
from seam import SeamWebhook
app = Flask(__name__)
webhook = SeamWebhook(os.getenv('SEAM_WEBHOOK_SECRET'))
@app.route('/webhook', methods=['POST'])
def handle_webhook():
try:
data = webhook.verify(request.get_data(), request.headers)
except Exception:
return 'Bad Request', 400
try:
store_event(data)
except Exception:
return 'Internal Server Error', 500
return '', 204
def store_event(data):
print(data)
if __name__ == '__main__':
app.run(port=8080)Some contexts may need to override the API endpoint, e.g., testing or proxy setups.
Either pass the endpoint option to the constructor, or set the SEAM_ENDPOINT environment variable.
Requests time out after 30 seconds by default.
Pass the timeout option, in seconds, to override this:
from seam import Seam
seam = Seam(api_key="your-api-key", timeout=60)Setting it to None disables the timeout entirely.
A request that exceeds the timeout raises httpx.TimeoutException.
By default, the SDK makes up to three attempts: the initial request and two
retries. Retries are limited to GET, HEAD, OPTIONS, PUT, and
DELETE requests that fail because of a transport error, timeout, HTTP 429
response, or HTTP 5xx response. POST and PATCH requests are not retried.
Retries use exponential backoff with jitter: approximately 200–240 ms before
the first retry and 400–480 ms before the second. A Retry-After header is
honored instead of the calculated backoff. The request timeout is reset for
each attempt.
Pass the retries option to configure retry behavior.
Retries are handled by httpx-retries,
and its Retry class is re-exported from seam for convenience:
from seam import Seam, Retry
seam = Seam(
api_key="your-api-key",
retries=Retry(total=3, backoff_factor=0.5, status_forcelist=[503]),
)For control the options above do not cover, pass httpx_options.
These are handed to the underlying httpx Client and take
precedence over the defaults the SDK sets:
from httpx import Limits
seam = Seam(
api_key="your-api-key",
httpx_options={
"limits": Limits(max_connections=25, max_keepalive_connections=20),
},
)The Seam API parses URL search params as complex types.
If you call it with your own HTTP client,
serialize_url_search_params is exported for that purpose:
import httpx
from seam import serialize_url_search_params
httpx.get(
"https://connect.getseam.com/devices/list",
params=serialize_url_search_params({"device_ids": ["device1", "device2"]}),
headers={"Authorization": "Bearer your-api-key"},
)The serialization defines the name and value of each search param,
where every value is a string.
UrlSearchParams holds those pairs and renders the query string,
as URLSearchParams does for the reference implementation:
from seam import UrlSearchParams, update_url_search_params
search_params = UrlSearchParams()
update_url_search_params(search_params, {"device_ids": ["device1", "device2"]})
list(search_params)
# => [('device_ids', 'device1'), ('device_ids', 'device2')]
str(search_params)
# => 'device_ids=device1&device_ids=device2'Pass either the query string or the pairs to your HTTP client.
A client may percent-encode a few characters differently than
URLSearchParams does, e.g. httpx escapes * and unescapes ~,
which the Seam API reads as the same params either way.
A param set to None is omitted, while a param set to NULL
is serialized to an empty value, which the Seam API reads as null,
as described in Setting a Param to Null.
A param that cannot be represented raises a seam.UnserializableParamError.
The Seam API parses these params with the corresponding parser.
$ git clone https://github.com/seamapi/python.git $ cd python $ uv sync
Run each command below in a separate terminal window:
$ just watch
Primary development tasks are defined in the justfile.
The source code is hosted on GitHub. Clone the project with
$ git clone https://github.com/seamapi/python.git
You will need Python 3 and uv and Node.js with npm and just.
Install the development dependencies with
$ uv sync $ npm install
Lint code with
$ just lint
Run tests with
$ just test
Run tests on changes with
$ just watch
New versions are created with uv version.
New versions are released automatically with semantic-release as long as commits follow the Angular Commit Message Conventions.
Publish a new version by triggering a version workflow_dispatch on GitHub Actions.
The version input will be passed as the first argument to uv version.
This may be done on the web or using the GitHub CLI with
$ gh workflow run version.yml --raw-field version=<version>
GitHub Actions should already be configured: this section is for reference only.
The following repository secrets must be set on GitHub Actions.
PYPI_API_TOKEN: API token for publishing on PyPI.
These must be set manually.
The version, format, generate, and semantic-release GitHub actions require a user with write access to the repository including access to read and write packages. Set these additional secrets to enable the action:
GH_TOKEN: A personal access token for the user.GIT_USER_NAME: The name to set for Git commits.GIT_USER_EMAIL: The email to set for Git commits.GPG_PRIVATE_KEY: The GPG private key.GPG_PASSPHRASE: The GPG key passphrase.
Please submit and comment on bug reports and feature requests.
To submit a patch:
- Fork it (https://github.com/seamapi/python/fork).
- Create your feature branch (git checkout -b my-new-feature).
- Make changes.
- Commit your changes (git commit -am 'Add some feature').
- Push to the branch (git push origin my-new-feature).
- Create a new Pull Request.
This Python package is licensed under the MIT license.
This software is provided by the copyright holders and contributors "as is" and any express or implied warranties, including, but not limited to, the implied warranties of merchantability and fitness for a particular purpose are disclaimed. In no event shall the copyright holder or contributors be liable for any direct, indirect, incidental, special, exemplary, or consequential damages (including, but not limited to, procurement of substitute goods or services; loss of use, data, or profits; or business interruption) however caused and on any theory of liability, whether in contract, strict liability, or tort (including negligence or otherwise) arising in any way out of the use of this software, even if advised of the possibility of such damage.
