Skip to content

Repository files navigation

OilPriceAPI Python SDK

Official Python client for source-timestamped OilPriceAPI energy data. It provides typed synchronous and asynchronous clients, bounded retries, explicit errors, optional pandas helpers, and executable example manifests.

PyPI version Python Tests License: MIT

Mutable offer, catalog, freshness, entitlement, and data-rights wording is governed by the reviewed product-facts.json contract. Latest available values include source timestamps; cadence, history depth, and access vary by source, market hours, dataset, and account entitlement.

Install

python -m pip install oilpriceapi

Optional extras are installed only when the application uses them:

python -m pip install "oilpriceapi[pandas]"
python -m pip install "oilpriceapi[stream]"

An installed helper does not imply that every dataset or workflow is enabled for an account. Confirm access in the current API response and documentation.

Authenticate

Create an API key in the OilPriceAPI dashboard and provide it through the environment. Do not put a key in source code, a notebook cell, a URL, logs, screenshots, or issue text.

export OILPRICEAPI_KEY="your-key-from-the-dashboard"

The API authentication header is Authorization: Token YOUR_API_KEY.

First Request With Source Context

The canonical first request is GET /v1/prices/latest?by_code=BRENT_CRUDE_USD. This example fails closed if the response omits the context needed to interpret the value:

import math
import os

from oilpriceapi import OilPriceAPI

with OilPriceAPI(api_key=os.environ["OILPRICEAPI_KEY"], max_retries=1) as client:
    payload = client.request(
        "GET",
        "/v1/prices/latest",
        params={"by_code": "BRENT_CRUDE_USD"},
        timeout=30,
    )

record = payload.get("data")
if not isinstance(record, dict):
    raise RuntimeError("EMPTY_RESPONSE: no price record returned")

price = record.get("price")
if isinstance(price, bool) or not isinstance(price, (int, float)) or not math.isfinite(price):
    raise RuntimeError("MALFORMED_RESPONSE: price is not a finite number")

source = record.get("source")
metadata = record.get("metadata")
if not source and isinstance(metadata, dict):
    source = metadata.get("source")

timestamp_field = next(
    (
        field
        for field in ("as_of", "source_timestamp", "created_at", "updated_at")
        if isinstance(record.get(field), str) and record[field].strip()
    ),
    None,
)

required_text = {
    "code": record.get("code"),
    "currency": record.get("currency"),
    "unit": record.get("unit"),
    "source": source,
}
if timestamp_field is None or any(
    not isinstance(value, str) or not value.strip()
    for value in required_text.values()
):
    raise RuntimeError("MALFORMED_RESPONSE: source context is incomplete")

print(
    {
        **required_text,
        "price": float(price),
        "api_timestamp_field": timestamp_field,
        "api_timestamp": record[timestamp_field],
        "freshness": record.get("data_status") or record.get("freshness"),
    }
)

The reviewed standalone form is examples/snippets/latest_price.py. CI executes it against production-shaped fixtures and publishes its code and checksum in the release snippet manifest.

Typed Client

For applications that only need the normalized core fields:

import os

from oilpriceapi import OilPriceAPI

with OilPriceAPI(api_key=os.environ["OILPRICEAPI_KEY"]) as client:
    price = client.prices.get("BRENT_CRUDE_USD")

print(
    price.commodity,
    price.value,
    price.currency,
    price.unit,
    price.timestamp.isoformat(),
)

Use the raw first-request pattern when downstream logic requires the exact source and timestamp-field semantics from the API response.

Complete pandas DataFrames

Install the optional pandas support, then request a historical DataFrame:

df = client.historical.to_dataframe(
    commodity="BRENT_CRUDE_USD",
    start="2026-01-01",
    end="2026-06-30",
    per_page=500,
)

to_dataframe() fetches every page automatically. per_page controls the request page size, not the total result size, and must be an integer from 1 to 1000. The DataFrame preserves each API row's currency and unit; a missing currency remains missing rather than being labeled USD. Exact records repeated by an overlapping page boundary are returned once, while distinct records are retained. Empty results have a stable schema with a date index.

The same per_page behavior applies to date-range queries through client.prices.to_dataframe(...). See the DataFrames and pagination guide for the complete contract.

Dates and Commodity Codes

Date strings must be real calendar dates in exact YYYY-MM-DD form. Malformed values are rejected before an API request, while well-formed ranges are still validated authoritatively by the server.

Search the current API catalog instead of maintaining a local code list:

matches = client.commodities.search("brent crude", limit=5)
print([commodity["code"] for commodity in matches])

Invalid-code API responses expose sanitized recovery values through error.suggestions and error.invalid_codes. See the dates and commodity-code guide for sync/async examples and failure behavior.

Recovery

The package exposes typed errors for the customer-recoverable boundaries:

from oilpriceapi import (
    AuthenticationError,
    OilPriceAPIError,
    RateLimitError,
    TimeoutError,
)

try:
    price = client.prices.get("BRENT_CRUDE_USD")
except AuthenticationError:
    print("Replace the missing, expired, or revoked API key.")
except RateLimitError as error:
    print("Wait for the API-provided reset window.", error.seconds_until_reset)
except TimeoutError:
    print("Retry once, then check https://status.oilpriceapi.com.")
except OilPriceAPIError as error:
    # Safe support context: do not log your API key or request headers.
    if error.request_id:
        print("Support request ID:", error.request_id)
    if error.status_code in (402, 403):
        print(
            "Review dataset access for this account.",
            error.required_plan,
            error.required_feature,
            error.remediation_url,
        )
    else:
        raise

All non-2xx responses share the same OilPriceAPIError attributes, including status_code, code, request_id, plan/feature recovery fields, retry metadata, commodity suggestions and invalid_codes, sanitized response headers, raw_body, and raw_text. Canonical nested, fail/data, and legacy flat API error envelopes are normalized into that contract. Transport failures use NetworkError; timeouts remain the more specific TimeoutError.

Executable recovery examples cover 401, 403, 429, and timeout responses under examples/snippets/. Empty or malformed successful responses should stop analysis rather than inventing a price, unit, currency, source, or timestamp.

Capabilities

The client includes resources for latest and historical values plus additional dataset and workflow families. Availability is determined by the live API and account entitlement, not by the presence of a helper method in the package.

Standard plans provide API access, normalization, monitoring, and delivery; they do not transfer ownership of underlying source data or unrestricted raw data redistribution rights.

Reproducible Examples

Website and documentation snippets are maintained in examples/snippets/. Every release attaches a versioned manifest containing the package version, minimum runtime, source commit, expected response shape, exact code, and SHA-256 for each example.

python scripts/generate_snippet_manifest.py \
  --source-commit "$(git rev-parse HEAD)" \
  --output artifacts/snippets/oilpriceapi-python-snippets-v1.json

Development

The performance guide documents timeout, connection-pooling, batching, retry, and troubleshooting behavior without making a universal latency promise. The release process documents the actual PyPI gate and immutable-version recovery procedure.

python -m pip install -e '.[dev]'
python scripts/validate_storefront_claims.py
pytest tests/ --ignore=tests/integration --ignore=tests/contract -m 'not slow'
python -m build

Live tests require an explicitly supplied non-customer test credential. Unit and snippet tests use local fixtures and do not print or persist credentials.

Support

Licensed under the MIT License.

Releases

Packages

Used by

Contributors

Languages