amasto 0.2.2

No project description provided

amasto

Fully async, type-safe Python client for the Mastodon API.

[!WARNING] This project contains code generated by LLMs (Large Language Models). A significant portion of the codebase has not been fully reviewed or tested. Use at your own risk.

Features

• Async-first — All I/O uses async/await via httpx
• Type-safe — Typed endpoint descriptors and Pydantic response models; ships with py.typed
• Version-aware — Automatic server version detection via NodeInfo; models mark field availability with since() / Unsupported
• Pagination — PaginatedHttpMethod.paginate() async iterator transparently follows Link: rel="next" headers across pages
• Streaming — Real-time WebSocket streaming with typed events, automatic reconnection, and exponential back-off
• Minimal surface area — Small, deliberate public API

Requirements

• Python ≥ 3.14

Installation

pip install amasto

Or with uv:

uv add amasto

Quick start

import asyncio
from amasto import Amasto

async def main() -> None:
    client = Amasto("https://mastodon.social", "YOUR_ACCESS_TOKEN")

    # Post a status
    status = await client.api.v1.statuses.post(body={"status": "Hello from amasto!"})

    # Read a single status by ID
    status = await client.api.v1.statuses["123456"].get()

    # List accounts the user is following
    accounts = await client.api.v1.accounts["123456"].following.get()

asyncio.run(main())

Client

Amasto is the main entry point. It wraps an httpx.AsyncClient and automatically discovers the server's Mastodon version via the NodeInfo protocol on first use.

from semver import Version

client = Amasto(
    "https://mastodon.social",   # base URL
    "YOUR_ACCESS_TOKEN",         # API key (Bearer token)
    mastodon_version=Version(4, 3, 0),  # optional: skip auto-detection
)

Resources

API access uses a resource-based pattern where every endpoint is reachable as a chain of attribute accesses on the client:

client.api.v1.<resource>.<method>(params=..., body=...)
client.api.v1.<resource>["id"].<sub_resource>.<method>(...)

Each leaf node is an HttpMethod instance that is:

• Async-callable — await method(params=..., body=...) executes the HTTP request and returns a validated response.
• Introspectable — .method, .path, and .requires expose the HTTP verb, URL path, and minimum server version.
• Test-friendly — .parse(data) validates data against the response type without making HTTP calls.

List endpoints (bookmarks, timelines, followers, …) use PaginatedHttpMethod which adds a .paginate() async iterator:

# Single page (default behaviour, backwards compatible)
statuses = await client.api.v1.timelines.home.get()

# Iterate across all pages automatically
async for status in client.api.v1.timelines.home.get.paginate(params={"limit": 40}):
    print(f"{status.account.username}: {status.content}")

# Stop after a maximum number of items
async for status in client.api.v1.timelines.home.get.paginate(max_items=200):
    print(status.content)

# Validate data without HTTP (useful in tests)
status = client.api.v1.statuses["123"].get.parse({
    "id": "123",
    "content": "<p>Hello</p>",
    "account": {"id": "1", "username": "alice", ...},
    ...
})

API v1 (client.api.v1)

Resource	Access pattern
accounts	.get, .post, .verify_credentials.get, ["id"].get, ["id"].follow.post, …
announcements	.get, ["id"].dismiss.post, ["id"].reactions["name"].put
apps	.post, .verify_credentials.get
blocks	.get
bookmarks	.get
conversations	.get, ["id"].delete, ["id"].read.post
custom_emojis	.get
directory	.get
domain_blocks	.get, .post, .delete
emails	.confirmations.post
endorsements	.get
favourites	.get
featured_tags	.get, .post, .suggestions.get, ["id"].delete
follow_requests	.get, ["id"].authorize.post, ["id"].reject.post
followed_tags	.get
instance	.get, .peers.get, .activity.get, .rules.get, .domain_blocks.get, …
lists	.get, .post, ["id"].get, ["id"].put, ["id"].delete, ["id"].accounts.get
markers	.get, .post
media	["id"].get, ["id"].put, ["id"].delete
mutes	.get
notifications	.get, .clear.post, .unread_count.get, .requests.*, ["id"].get, ["id"].dismiss.post
polls	["id"].get, ["id"].votes.post
preferences	.get
profile	.avatar.delete, .header.delete
push	.subscription.get, .subscription.post, .subscription.put, .subscription.delete
reports	.post
scheduled_statuses	.get, ["id"].get, ["id"].put, ["id"].delete
search	.get
statuses	.get, .post, ["id"].get, ["id"].put, ["id"].delete, ["id"].context.get, ["id"].favourite.post, …
streaming	.user(), .user_notification(), .public(), .public_local(), .public_remote(), .hashtag(tag), .list(list_id), .direct()
suggestions	.get, ["id"].delete
tags	["key"].get, ["key"].follow.post, ["key"].unfollow.post
timelines	.public.get, .home.get, .link.get, .direct.get, .tag["hashtag"].get, .list["id"].get
trends	.tags.get, .statuses.get, .links.get

API v2 (client.api.v2)

Resource	Access pattern
filters	.get, .post, ["id"].get/put/delete, ["id"].keywords.get/post, .keywords["id"].get/put/delete
instance	.get
media	.post
notifications	.get, .unread_count.get, .policy.get/patch, ["group_key"].dismiss.post, ["group_key"].accounts.get
search	.get
suggestions	.get

OEmbed (client.api.oembed)

Access	Description
.get	Fetch oEmbed data for a status URL

OAuth (client.oauth)

Resource	Access
authorize	.get — Authorization URL
token	.post — Obtain a token
revoke	.post — Revoke a token
userinfo	.get — Authenticated user info

Health (client.health)

Access	Description
.get	Server health check

Examples

Reading timelines

# Home timeline
statuses = await client.api.v1.timelines.home.get(params={"limit": 20})
for s in statuses:
    print(f"{s.account.username}: {s.content}")

# Local public timeline
statuses = await client.api.v1.timelines.public.get(params={"local": True})

# Hashtag timeline
statuses = await client.api.v1.timelines.tag["python"].get()

Posting and interacting with statuses

# Post a new status
status = await client.api.v1.statuses.post(body={
    "status": "Hello, Fediverse!",
    "visibility": "public",
    "language": "en",
})

# Post a status with a poll
status = await client.api.v1.statuses.post(body={
    "status": "What's your favourite language?",
    "poll": {
        "options": ["Python", "Rust", "TypeScript"],
        "expires_in": 86400,
        "multiple": False,
    },
})

# Favourite / boost / bookmark
await client.api.v1.statuses["123456"].favourite.post()
await client.api.v1.statuses["123456"].reblog.post()
await client.api.v1.statuses["123456"].bookmark.post()

# Get thread context
context = await client.api.v1.statuses["123456"].context.get()
print(context.ancestors, context.descendants)

Accounts

# Verify your own credentials
me = await client.api.v1.accounts.verify_credentials.get()
print(me.username, me.followers_count)

# Look up another account
account = await client.api.v1.accounts["12345"].get()

# List followers with pagination
followers = await client.api.v1.accounts["12345"].followers.get(
    params={"limit": 40},
)

Search

results = await client.api.v1.search.get(params={
    "q": "python",
    "resolve": True,
    "limit": 10,
})
for account in results.accounts:
    print(account.acct)
for status in results.statuses:
    print(status.content)

Notifications

# Fetch only mentions
mentions = await client.api.v1.notifications.get(params={
    "types": ["mention"],
    "limit": 30,
})
for n in mentions:
    print(f"{n.account.username} mentioned you: {n.status.content}")

Pagination

List endpoints accept max_id, since_id, min_id, and limit for cursor-based pagination:

# First page
statuses = await client.api.v1.timelines.home.get(params={"limit": 40})

# Next page — use the last item's ID
if statuses:
    older = await client.api.v1.timelines.home.get(params={
        "max_id": statuses[-1].id,
        "limit": 40,
    })

Streaming

The client supports real-time streaming via WebSocket. Each streaming method returns an async iterator that yields typed event models:

async for event in client.api.v1.streaming.user():
    match event:
        case UpdateEvent() as e:
            print(f"New status: {e.status.content}")
        case DeleteEvent() as e:
            print(f"Deleted: {e.status_id}")
        case NotificationEvent() as e:
            print(f"Notification: {e.notification.type}")

Available streams

Method	Stream	Description
streaming.user()	user	Statuses, notifications, and other events for the authenticated user
streaming.user_notification()	user:notification	Notifications only
streaming.public()	public	All public statuses (pass only_media=True for media-only)
streaming.public_local()	public:local	Local public statuses
streaming.public_remote()	public:remote	Remote public statuses
streaming.hashtag(tag)	hashtag	Statuses with the given hashtag (pass local=True for local only)
streaming.list(list_id)	list	Statuses from a specific list
streaming.direct()	direct	Direct messages

Reconnection

All streaming methods automatically reconnect with exponential back-off when the connection drops. Customize the policy with ReconnectPolicy:

from amasto import ReconnectPolicy

policy = ReconnectPolicy(max_retries=10, initial_delay=0.5, max_delay=60.0)

async for event in client.api.v1.streaming.user(reconnect=policy):
    ...

Event types

Event class	Trigger	Key fields
UpdateEvent	New status	status
DeleteEvent	Status deleted	status_id
NotificationEvent	New notification	notification
FiltersChangedEvent	Filters updated	—
ConversationEvent	Direct conversation updated	conversation
AnnouncementEvent	New announcement	announcement
AnnouncementReactionEvent	Reaction on announcement	name, count, announcement_id
AnnouncementDeleteEvent	Announcement deleted	announcement_id
StatusUpdateEvent	Status edited	status
EncryptedMessageEvent	Encrypted message	encrypted_message
NotificationsMergedEvent	Notification requests merged	—

Models

Response models live under amasto.models and are re-exported from amasto.models.v1 and amasto.models.v2. All models are frozen Pydantic BaseModel subclasses.

V1 models

Account, AccountRole, AccountSource, AccountWarning, Announcement, AnnouncementAccount, AnnouncementDeleteEvent, AnnouncementEvent, AnnouncementReactionEvent, AnnouncementStatus, Appeal, Application, AsyncRefresh, Context, Conversation, ConversationEvent, CredentialAccount, CredentialApplication, CustomEmoji, DeleteEvent, DomainBlock, EncryptedMessage, EncryptedMessageEvent, Error, ExtendedDescription, FamiliarFollowers, FeaturedTag, Field, FiltersChangedEvent, IdentityProof, InstanceStats, InstanceUrls, List, Marker, MediaAttachment, MutedAccount, Notification, NotificationEvent, NotificationRequest, NotificationsMergedEvent, Poll, PollOption, Preferences, PreviewCard, PreviewCardAuthor, PrivacyPolicy, Quote, QuoteApproval, Reaction, Relationship, RelationshipSeveranceEvent, Report, Role, Rule, ScheduledStatus, ScheduledStatusParams, ScheduledStatusParamsPoll, Search, ShallowQuote, Status, StatusEdit, StatusEditPoll, StatusEditPollOption, StatusMention, StatusSource, StatusTag, StatusUpdateEvent, StreamEvent, Suggestion, Tag, TagHistory, TermsOfService, Token, Translation, TranslationAttachment, TranslationPoll, TranslationPollOption, TrendsLink, UpdateEvent, WebPushAlerts, WebPushSubscription

V2 models

Filter, FilterKeyword, FilterResult, FilterStatus, Instance, InstanceConfiguration, InstanceConfigurationAccounts, InstanceConfigurationMediaAttachments, InstanceConfigurationPolls, InstanceConfigurationStatuses, InstanceConfigurationTimelinesAccess, InstanceConfigurationTimelinesFeedAccess, InstanceConfigurationTranslation, InstanceConfigurationUrls, InstanceConfigurationVapid, InstanceContact, InstanceIcon, InstanceRegistrations, InstanceThumbnail, InstanceThumbnailVersions, InstanceUsage, InstanceUsageUsers, NotificationPolicy, NotificationPolicySummary

Version awareness

Model fields annotated with since("x.y.z") resolve to Unsupported when the connected server is older than the specified version, so your code can safely handle missing data.

A field can be in one of three states:

State	Meaning
str, int, …	Supported and has a value
None	Supported but has no value
Unsupported	Server too old to support this field

from amasto import Unsupported

status = await client.api.v1.statuses["123456"].get()

# Pattern matching (Python 3.14+)
match status.text:
    case str() as t:
        print(t)          # supported, has a value
    case None:
        pass              # supported, no value
    case Unsupported():
        pass              # server too old for this field

# Or use isinstance
if not isinstance(status.poll, Unsupported):
    print(status.poll)

Endpoints can also declare requires="x.y.z" to indicate the minimum server version.

Error handling

amasto raises httpx.HTTPStatusError for non-2xx responses:

import httpx

try:
    status = await client.api.v1.statuses["nonexistent"].get()
except httpx.HTTPStatusError as e:
    print(e.response.status_code)  # 404
    print(e.response.json())       # {"error": "Record not found"}

Dependencies

Package	Purpose
httpx ≥ 0.28.1	Async HTTP client
pydantic ≥ 2.12.5	Response validation and models
semver ≥ 3.0.4	Server version parsing
websockets ≥ 16.0	WebSocket streaming

License

See LICENSE for details.