xoople-sdk

Official Python SDK for the Xoople Products API.

from xoople.sdk import Config, XoopleClient, StaticToken, models

client = XoopleClient.from_credentials(
    "https://api.xoople.com", StaticToken("...")
)
# Or pass the complete transport configuration directly:
# client = XoopleClient(Config.from_env())

estimate = client.analyses.estimate(request)
analysis = client.analyses.create(request)

# `list` returns a paginator: iterate it for items across every page.
for run in client.analyses.runs.list(analysis.uid):
    print(run.state)

# Sub-resources take the parent analysis uid:
output = client.analyses.outputs.get(analysis.uid, output_uid)
access = client.analyses.outputs.generate_access_url(analysis.uid, output_uid, ttl_seconds=900)

# A table output resolves to a handle that reads itself:
table = client.table(analysis.uid)          # or client.table(uid, output=output_uid)
print(table.shared)                         # measures, grid, size, created time
df = table.to_pandas()                      # pip install "xoople-sdk[sharing]"
sdf = table.to_spark(spark)                 # your session, your connector

# Select an inclusive Delta Change Data Feed range. Start at 0 on the first sync.
changes = table.changes(starting_version=0)
change_df = changes.to_pandas()
change_sdf = changes.to_spark(spark)
latest = table.latest_changes()
latest_df = latest.to_pandas()
latest_sdf = latest.to_spark(spark)

# Custom verbs are plain methods:
client.analyses.cancel(analysis.uid)
client.analyses.schedules.pause(analysis.uid, schedule_uid)

# Cross-analysis reads and catalogs are top-level:
client.schedules.list(state=models.ScheduleStatus.ACTIVE)
client.usage.limits()
client.catalog.measures(product="tabular_time_series")

Config.from_env() reads XOOPLE_API_URL and XOOPLE_TOKEN.

API keys

An API key lets a script or a scheduled job call the API without a browser sign-in. Create one on the API keys section of your Xoople profile page (/profile). The key starts with sk_ and is shown once, so store it in a secret manager.

A key belongs to one organization membership. It acts as you, in the organization that was active when you created it, with that membership's roles. To work in a different organization, create a key there.

The SDK sends the key as the bearer token. Set XOOPLE_TOKEN to the key and use Config.from_env(), or pass it explicitly:

import os

from xoople.sdk import StaticToken, XoopleClient

client = XoopleClient.from_credentials(
    "https://api.xoople.com", StaticToken(os.environ["XOOPLE_TOKEN"])
)

The API accepts at most 300 API key requests a minute from one IP address. The SDK retries rate-limited GETs and replay-safe writes with backoff. Any rate-limited request that is not retried, or still fails after the retries, raises xoople.sdk.errors.RateLimitError. Revoke a key on the same profile page when you no longer need it.

Connection profiles

Config.from_profile(path) reads the URL and the credential from one file, for tools that take a file and not an environment:

{
  "version": 1,
  "api_url": "https://gateway.xoople.com",
  "auth": { "type": "bearer", "token": "<token>" },
  "expires_at": "2026-09-30T00:00:00Z"
}

To use the sign-in from xoople auth login instead of a token, set "auth": { "type": "xoople-cli" }. The profile then names only the API, and the session is read from $XDG_CONFIG_HOME/xoople/session.json (else ~/.config/xoople/session.json) and refreshed in place. xoople.sdk.session reads and writes that file. The profile must name the API the session signed in for, so sign in with xoople auth login --api-url <url>; another API raises CredentialError.

expires_at is optional and is what the issuer stamped. It carries a UTC offset. A profile past it raises ProfileExpiredError on load, not at the first request.

The file is a credential. Keep it out of projects and repositories. Restrict who can read it.

client.table() is the authenticated half of xoople.sdk.sharing, which reads a .share profile with no API credential at all. It downloads one profile for the client's lifetime and removes it on close(), so keep the client open through every read — Spark reads lazily. client.table(uid, profile_path=...) uses a profile you own instead; the SDK never writes or removes that one.

Change Data Feed reads accept either starting_version with an optional ending_version, or starting_timestamp with an optional ending_timestamp. Ranges are inclusive and return the standard _change_type, _commit_version, and _commit_timestamp columns. Persist the highest commit version you apply, then begin the next read at that value plus one. For inspection rather than cursor-based synchronization, latest_changes().to_pandas() and latest_changes().to_spark(spark) read only the newest available table commit.

Every failure raises a subclass of xoople.sdk.errors.XoopleError.