Skip to content

upgrading_to_v4.md

Latest commit

 

History

History
200 lines (137 loc) · 9.43 KB

upgrading_to_v4.md

File metadata and controls

200 lines (137 loc) · 9.43 KB
Raw
id upgrading-to-v4
title Upgrading to v4
description Breaking changes and migration guide from Apify SDK v3.x to v4.0.

This guide lists the breaking changes between Apify Python SDK v3.x and v4.0.

Python 3.11+ required

Support for Python 3.10 has been dropped. The Apify Python SDK v4.x now requires Python 3.11 or later — make sure your environment is on a compatible version before upgrading.

Keyword-only arguments

Secondary parameters in these signatures can no longer be passed positionally:

  • Actor — get_value, push_data, charge, use_state.
  • ChargingManager — charge.
# Before (v3)
value = await Actor.get_value('my-key', default_value)
await Actor.push_data(data, 'my-event')
await Actor.charge('my-event', 5)

# After (v4)
value = await Actor.get_value('my-key', default_value=default_value)
await Actor.push_data(data, charged_event_name='my-event')
await Actor.charge('my-event', count=5)

Removal of deprecated APIs

Methods and arguments that had been deprecated in v3 are removed in v4.

api_public_base_url argument of storage clients

The deprecated api_public_base_url argument has been removed from ApifyDatasetClient and ApifyKeyValueStoreClient. It had no effect already in v3, passing it emitted only a DeprecationWarning. Drop it from your call sites. The public base URL is taken from Configuration.api_public_base_url, which is unchanged.

# Before (v3)
client = ApifyDatasetClient(
    api_client=api_client,
    api_public_base_url='https://api.apify.com',
    lock=lock,
)

# After (v4)
client = ApifyDatasetClient(
    api_client=api_client,
    lock=lock,
)

Actor.start and Actor.call: RemainingTime

The deprecated RemainingTime value of the timeout argument has been removed from Actor.start() and Actor.call(). Use inherit instead, the signature and behavior are identical.

# Before (v3)
run = await Actor.call('user/actor', timeout='RemainingTime')

# After (v4)
run = await Actor.call('user/actor', timeout='inherit')

Deprecated Configuration fields

The deprecated latest_sdk_version, log_format, and standby_port fields have been removed from Configuration:

  • In place of standby_port, use web_server_port.
  • latest_sdk_version and log_format don't have replacement. SDK version checking isn't supported for the Python SDK and the log format should be adjusted in code instead.

Built on apify-client v3

The SDK is now built on apify-client v3 and no longer depends on apify-shared. The sections below cover the user-visible consequences; see the client's Upgrading to v3 guide for the full list of changes in the client itself.

Environment variable enums moved

If you imported the platform environment-variable enums from apify_shared.consts (ApifyEnvVars, ActorEnvVars), import them from apify instead — they are now vendored in the SDK and re-exported from the top-level package.

# Before (v3)
from apify_shared.consts import ApifyEnvVars

# After (v4)
from apify import ApifyEnvVars

Typed responses

Actor.start, Actor.abort, Actor.call, and Actor.call_task now return apify_client._models.Run instead of the SDK-side ActorRun. Both are Pydantic models with the same snake_case fields, so field access is unchanged — only the type and import path differ. The SDK no longer ships its own response models (apify._models has been removed); response shapes come from apify-client.

Literal string aliases instead of StrEnum classes

Generated enum-like types are now Literal string aliases instead of StrEnum classes. Pass plain strings instead of enum members.

  • apify.WebhookEventType is now a Literal[...] instead of a StrEnum. Use plain string values ('ACTOR.RUN.FAILED') instead of enum members.
  • apify_shared.consts.ActorEventTypes (a StrEnum) is replaced by apify.ActorEventTypes, now a Literal['systemInfo', 'persistState', 'migrating', 'aborting']. For runtime values, use apify.Event (re-exported from Crawlee) instead of enum members.
# Before (v3)
from apify import Actor
from apify_shared.consts import ActorEventTypes

Actor.on(ActorEventTypes.SYSTEM_INFO, callback)

# After (v4)
from apify import Actor, Event

Actor.on(Event.SYSTEM_INFO, callback)

Actor pricing info models

The Actor pricing-info models exposed through Actor.configuration.actor_pricing_info — FreeActorPricingInfo, FlatPricePerMonthActorPricingInfo, PricePerDatasetItemActorPricingInfo, PayPerEventActorPricingInfo, and the nested ActorChargeEvent / PricingPerEvent — are now thin subclasses of the corresponding apify-client models instead of standalone SDK copies. The discriminated-union shape is unchanged, so existing access (pricing_model, per-event titles and prices) keeps working; the models now expose the full apify-client field set, and a charge event's event_price_usd is optional (it is unset for tier-priced events). ChargingManager.get_pricing_info() is unchanged.

Webhook API simplified

The Webhook model has been slimmed down to only the fields a user sets when defining a webhook. Server-populated response fields (id, created_at, modified_at, user_id, is_ad_hoc, condition, last_dispatch, stats) and the unused WebhookCondition helper class have been removed. The description and should_interpolate_strings fields have also been removed — they are not part of the ad-hoc webhook representation (event_types, request_url, payload_template, headers_template) that Actor.start / Actor.call / Actor.call_task and Actor.add_webhook now send. Webhook is now a plain @dataclass instead of a Pydantic BaseModel — construct it with snake_case kwargs; .model_dump() / .model_validate() are gone.

The retry and idempotency kwargs that used to live on Actor.add_webhook have moved onto the Webhook instance itself.

# Before (v3)
from apify import Actor, Webhook

await Actor.add_webhook(
    Webhook(event_types=['ACTOR.RUN.FAILED'], request_url='https://example.com'),
    ignore_ssl_errors=False,
    do_not_retry=False,
    idempotency_key='my-key',
)

# After (v4)
from apify import Actor, Webhook

await Actor.add_webhook(
    Webhook(
        event_types=['ACTOR.RUN.FAILED'],
        request_url='https://example.com',
        ignore_ssl_errors=False,
        do_not_retry=False,
        idempotency_key='my-key',
    )
)

The idempotency_key kwarg form on Actor.add_webhook still works for one more release but emits a DeprecationWarning and will be removed in v5.0. The ignore_ssl_errors and do_not_retry kwargs have been removed outright — set them on the Webhook instance.

apify.WebhookCondition is no longer exported; the SDK now binds the webhook to the current Actor run internally.

The webhooks argument on Actor.start, Actor.call, and Actor.call_task still accepts list[Webhook] and the fields used at the call site (event_types, request_url, payload_template, headers_template) are unchanged.

Actor.new_client — timeout scales all tiers

apify-client v3 split its single timeout into four tiers (short / medium / long / max). Actor.new_client(timeout=...) still takes a single timedelta; the SDK uses it as the medium-tier baseline and scales the other tiers proportionally (short = timeout / 6, long = timeout * 12, max = timeout * 12). The public signature is unchanged — no migration needed.

Using the client from Actor.new_client

Actor.new_client() (and the Actor.apify_client property) now returns an apify-client v3 ApifyClientAsync. When you use that client directly, the client's v3 breaking changes apply — the most impactful ones are below. See the client's Upgrading to v3 guide for the complete reference.

404 raises NotFoundError on ambiguous endpoints

Direct .get(id) and .delete(id) calls still swallow 404 into None. But where a 404 could mean either the parent or the sub-resource is missing, the client now raises NotFoundError instead of returning None.

# Before (v3)
client = Actor.new_client()

# Returned None on 404.
dataset = await client.run('some-run-id').dataset().get()

# After (v4)
from apify_client.errors import NotFoundError

client = Actor.new_client()

# Raises NotFoundError; handle it explicitly.
try:
    dataset = await client.run('some-run-id').dataset().get()
except NotFoundError:
    dataset = None

Keyword-only arguments

Secondary parameters on several client methods can no longer be passed positionally.

# Before (v3)
await client.key_value_store('my-store').set_record('my-key', {'data': 1}, 'application/json')
await client.run('my-run').charge('my-event', 5)

# After (v4)
await client.key_value_store('my-store').set_record('my-key', {'data': 1}, content_type='application/json')
await client.run('my-run').charge('my-event', count=5)

Async iterate_* are no longer coroutine functions

DatasetClientAsync.iterate_items() and KeyValueStoreClientAsync.iterate_keys() are now plain def functions returning AsyncIterator[T]. Consumer code (async for ...) is unchanged; if you annotate the call's return value, change AsyncGenerator[T, None] to AsyncIterator[T].