apiauto — Python client for the Encar, Che168, Copart and IAAI API

apiauto is a typed Python client for the apiauto.space REST API: used-car listings from Encar (South Korea), Che168 (China) and the Copart and IAAI auctions (USA) in one JSON record format, without writing a scraper. Search the catalogs by make, model, year, price and mileage, page through whole markets, follow a feed of new cars, sold cars and price changes, and read quota headers — from sync or async code. It is an unofficial client: the project is not affiliated with Encar, Che168/Autohome, Copart or IAAI; the data comes through the apiauto.space API.

• Sync Client and AsyncClient on top of httpx, the only dependency
• Iterators that follow the cursor for you, with max_items that never over-spends the quota
• Automatic retries on 429 and 503 (honouring Retry-After) and on 5xx and network errors
• One exception class per API error: QuotaExceededError, RateLimitError, NotFoundError ...
• TypedDict types for every response, py.typed, checked with mypy --strict
• Python 3.10+

Install

pip install apiauto-space

The distribution is called apiauto-space on PyPI; the module you import is apiauto.

Get a key

Open the Telegram bot @apiautospace_bot and take a free demo key: one day, 100 data requests, up to 20 cars per request, all three markets. Paid plans are on the pricing page.

Save the key in the APIAUTO_KEY environment variable — the client reads it from there:

export APIAUTO_KEY="your-key"

Keep the key on your server. The API does not accept browser requests (CORS is off), and a key in page code is visible to every visitor.

Quick start

import apiauto

client = apiauto.Client()  # reads APIAUTO_KEY

page = client.cars("kr", brand="Kia", model="Sorento", year_min=2021, limit=20)
for car in page["items"]:
    print(car["id"], car["title"], car["price"]["amount"], car["price"]["currency"])

Markets are kr (Encar, Korea), cn (Che168, China) and us (Copart and IAAI, USA). Every record has the same keys in every market; a value the source does not provide is None.

Examples

Search Encar by make, model, year and price

page = client.cars(
    "kr",
    brand="Hyundai",
    model="Grandeur",
    year_min=2020,
    year_max=2023,
    price_min=10_000_000,  # in the market's currency: KRW for Korea; skips monthly lease payments
    price_max=30_000_000,
    mileage_max=80_000,  # km
    sort="-first_seen",  # newest listings first (the default)
    limit=20,
)
print(page["count"], "cars match")  # None when more than 10,000 match

Prices in dollars at the ECB rate work in every market: price_usd_min=15000, price_usd_max=25000. The full list of filters: brand, model, generation (Korea), trim, year_min, year_max, price_min, price_max, price_usd_min, price_usd_max, mileage_max, fuel, transmission, body_type, has_photos, is_active, q (words of the title), sort, limit, cursor.

Search Che168 (China)

for car in client.cars("cn", brand="BYD", fuel="electric", has_photos=True, limit=10)["items"]:
    print(car["title"], car["price"]["amount"], "CNY", car["location"]["city"])

US auction lots: Copart and IAAI

for lot in client.cars("us", brand="Toyota", model="Camry", year_min=2020, limit=10)["items"]:
    raw = lot["raw"]  # auction-only fields as the source sends them
    print(lot["id"], lot["vin"], lot["mileage_km"], raw.get("damage_main"), raw.get("sale_date"))

Cars that are not in the copy yet can be searched live at the auction data provider (Pro and Max plans, a separate monthly counter): client.us_live(make="Toyota", model="Camry", year_min=2020).

Iterate over all pages

for car in client.iter_cars("kr", brand="Genesis", limit=100, max_items=1000):
    save(car)

iter_cars follows next_cursor until the last page. With max_items the last request asks only for the cars still needed, so the quota is not spent on records that would be thrown away. The demo allows limit up to 20; paid plans up to 100.

Get a car by id

car = client.car("encar:43020893")
print(car["is_active"], car["listing_url"])

Ids look like encar:…, che168:…, copart:… and iaai:… and never change. A sold or removed car stays available for 14 days with is_active=False.

Keep a catalog in sync with the change feed

import time
from datetime import datetime, timezone

since = datetime.now(timezone.utc)
while True:
    time.sleep(3600)  # the market copy is refreshed hourly
    polled_at = datetime.now(timezone.utc)
    for event in client.iter_changes("kr", since, limit=100):
        if event["type"] == "added":
            save(client.car(event["car_id"]))
        elif event["type"] == "disappeared":
            mark_sold(event["car_id"])
        else:  # price_drop or price_up
            update_price(event["car_id"], event["new_price"])
    since = polled_at

since takes a datetime (a naive one is treated as UTC) or an ISO 8601 string. Filter by type="price_drop" (or a list of types) and min_pct=5. The feed comes with the Pro, Max and demo plans; each event counts as one record. A complete script is in examples/sync_changes.py.

References: makes, models, canonical values, freshness

These calls are free.

client.brands("cn")  # [{"name": "Mercedes-Benz", "count": 20185}, ...]
client.models("kr", "Hyundai")  # the make's models with car counts
client.dictionaries()["fuel"]  # {"gasoline": ["gasoline", "petrol", ...], "diesel": [...], ...}
for market in client.markets():
    print(market["market"], market["active_cars"], market["data_age_seconds"], market["stale"])

Usage and quota headers

The quota counts returned cars (records), not requests; the demo counts data requests.

usage = client.usage()  # free
print(usage["plan"], usage["quota_used"], "/", usage["quota_limit"], usage["quota_unit"])

client.cars("us", limit=5)
info = client.last_response
print(info.quota_remaining, info.quota_reset, info.rate_remaining, info.data_age_seconds)

client.last_response is a frozen ResponseInfo read from the X-Quota-*, X-RateLimit-* and X-Data-Age-Seconds headers of the last response; a header that was not sent is None.

Async

import asyncio
import apiauto


async def main() -> None:
    async with apiauto.AsyncClient() as client:
        page = await client.cars("cn", brand="Toyota", limit=10)
        async for car in client.iter_cars("kr", brand="Kia", max_items=50):
            print(car["title"])


asyncio.run(main())

AsyncClient has the same methods as Client; iter_cars and iter_changes return async iterators.

Errors and retries

try:
    car = client.car("encar:1")
except apiauto.NotFoundError:
    car = None
except apiauto.QuotaExceededError as error:
    print("quota renews at", error.reset_at)
except apiauto.ApiautoError as error:
    print(error.status_code, error.code, error.detail)

All of them derive from ApiautoError. Retries follow the API's error guide: max_retries=2 by default, a wait of at most 60 seconds; Client(max_retries=0) turns them off. Other options: Client(api_key=..., base_url=..., timeout=30.0, http_client=httpx.Client(...)).

Photos

car["photos"] are signed links that work for about a day. Store the car's id, not the links, and fetch a fresh record with client.car(id) when you need the photos again. Photo requests are free and may redirect to the original host, so follow redirects when downloading them.

Prices are decimal strings

Prices, dollar prices and percentages come as strings such as "23900000.00" so no precision is lost. Convert them with Decimal:

from decimal import Decimal

total = sum(Decimal(car["price_usd"]) for car in page["items"] if car["price_usd"])

A record

A real Encar car from the documentation, shortened: two photo links and a few raw fields.

{
  "id": "encar:43020893",
  "market": "kr",
  "source": "encar",
  "source_id": "43020893",
  "title": "2020 Kia Mohave The Master 3.0 4WD 7-seater Masters",
  "brand": "Kia",
  "model": "Mohave",
  "generation": "Mohave The Master",
  "trim": "3.0 4WD 7-seater Masters",
  "year": 2020,
  "mileage_km": 93724,
  "fuel": "diesel",
  "transmission": null,
  "engine_cc": null,
  "body_type": null,
  "color": null,
  "vin": null,
  "price": {"amount": "23900000.00", "currency": "KRW"},
  "price_usd": "18033.96",
  "location": {"country": "KR", "region": "Gyeonggi", "city": null},
  "photos": [
    "https://api.apiauto.space/img/encar:43020893/1.jpg?exp=…&sig=…",
    "https://api.apiauto.space/img/encar:43020893/2.jpg?exp=…&sig=…"
  ],
  "photo_count": 4,
  "listing_url": "https://fem.encar.com/cars/detail/43020893",
  "is_active": true,
  "first_seen_at": "2026-10-26T04:37:32.553458Z",
  "last_seen_at": "2026-10-26T04:37:32.553458Z",
  "disappeared_at": null,
  "updated_at": "2026-10-26T04:37:32.553458Z",
  "source_mode": "replica",
  "raw": {"badge": "디젤 3.0 4WD 7인승", "form_year": "202004", "sell_type": "normal", "has_accident": false}
}

The fields each source adds are described on the Encar API, Che168 API, Copart API and IAAI API pages.

Methods

The API reference: apiauto.space/en/docs and the OpenAPI schema.

Other languages

По-русски

apiauto — Python-клиент для API Encar, Che168, Copart и IAAI, без парсинга: объявления о подержанных автомобилях из Южной Кореи, Китая и с аукционов США приходят в одном формате JSON через apiauto.space. Клиент умеет искать по марке, модели, году, цене и пробегу, сам проходит по страницам курсором, повторяет запрос при 429 и 503 и работает в синхронном и асинхронном коде. Бесплатный демо-ключ на сутки выдаёт Telegram-бот @apiautospace_bot; сохраните его в переменной APIAUTO_KEY и держите на сервере, не в браузере. Документация на русском: apiauto.space/docs.

pip install apiauto-space

import apiauto

client = apiauto.Client()  # ключ из APIAUTO_KEY
for car in client.cars("kr", brand="Kia", year_min=2021, limit=20)["items"]:
    print(car["title"], car["price"]["amount"], car["price"]["currency"])

License

MIT © 2026 apiauto.space

Disclaimer

This is an unofficial client of the apiauto.space API. It is not affiliated with, endorsed by or sponsored by Encar, Che168, Autohome, Copart or IAAI; their names are used only to describe where the listings come from, and all trademarks belong to their owners. Use of the data is subject to the apiauto.space terms.