apimemo 0.1.0

Automatic HTTP request logging for Python. Records every outgoing API call (httpx, requests) to your database with batch buffering, Django Admin & FastAPI admin panel support. Zero-config, production-ready.

apimemo

Automatic HTTP request logging for Python. Records every outgoing API call to your database — zero config, batch-buffered, production-ready.

---

Why apimemo?

Most applications make dozens of outgoing HTTP requests — to payment gateways, notification services, third-party APIs. When something goes wrong, you need answers: What did we send? What came back? How long did it take?

apimemo wraps your HTTP client (httpx or requests) and silently logs every request to your database. No manual logging. No code changes to every call site. Just plug in and see everything in your admin panel.

Key Features

• Zero-effort logging — wrap your HTTP client once, every request is captured automatically
• Batch-buffered writes — logs are collected in memory and flushed in configurable batches to minimize database overhead
• Thread-safe — safe for multi-threaded applications with proper locking and daemon timers
• Framework-native — works with your existing ORM and migration tool, no separate database needed
• Admin panel included — Django Admin and starlette-admin (FastAPI) views out of the box
• Configurable filtering — ignore health checks, internal services, or specific paths with glob patterns
• Body truncation — automatically truncates large request/response bodies to keep your database lean
• Graceful shutdown — atexit hook ensures no logs are lost when your process exits

Installation

pip install apimemo[django]        # Django + httpx
pip install apimemo[sqlalchemy]    # FastAPI + SQLAlchemy (async)
pip install apimemo[tortoise]      # FastAPI + Tortoise ORM
pip install apimemo[httpx]         # httpx interceptor only (bring your own storage)
pip install apimemo[requests]      # requests interceptor only
pip install apimemo[fastapi-admin] # starlette-admin panel for FastAPI
pip install apimemo[all]           # everything

Quick Start

Django

Add "apimemo" to your installed apps and run migrations — that's it.

# settings.py
INSTALLED_APPS = [
    "apimemo",
    ...
]

python manage.py migrate

# anywhere in your code
from apimemo.integrations.django import DjangoIntegration

integration = DjangoIntegration()
client = integration.get_client()

resp = client.get("https://api.stripe.com/v1/charges")
# → automatically logged to api_logs table
# → visible at /admin/apimemo/apilog/

FastAPI + SQLAlchemy

from sqlalchemy.ext.asyncio import create_async_engine
from apimemo.integrations.sqlalchemy import SqlAlchemyIntegration, ApiLogMixin

# 1. Define your model with the mixin
class ApiLog(Base, ApiLogMixin):
    __tablename__ = "api_logs"

# 2. Generate & apply migration
#    alembic revision --autogenerate -m "add api_logs"
#    alembic upgrade head

# 3. Create integration
engine = create_async_engine("postgresql+asyncpg://...")
integration = SqlAlchemyIntegration(engine)

# 4. Use the logged client
client = integration.get_async_client()
resp = await client.get("https://api.example.com/users")

# 5. (Optional) Mount admin panel
#    pip install apimemo[fastapi-admin]
integration.mount_admin(app)

FastAPI + Tortoise ORM

# tortoise config
TORTOISE_ORM = {
    "apps": {
        "apimemo": {
            "models": ["apimemo.integrations.tortoise"],
            "default_connection": "default",
        }
    }
}
# aerich migrate && aerich upgrade

from apimemo.integrations.tortoise import TortoiseIntegration

integration = TortoiseIntegration()
client = integration.get_async_client()
resp = await client.get("https://api.example.com/users")

Using requests instead of httpx

Every integration also provides a get_session() method that returns a requests.Session:

session = integration.get_session()
resp = session.post("https://api.example.com/webhook", json={"event": "test"})

Configuration

from apimemo import configure

configure(
    enabled=True,              # kill switch — disable all logging
    max_body_size=10240,       # truncate bodies larger than 10KB
    batch_size=50,             # flush after collecting 50 entries
    flush_interval=5.0,        # or flush every 5 seconds, whichever comes first
    ignore_hosts=("localhost", "*.internal.io"),  # skip these hosts (glob patterns)
    ignore_paths=("/health", "/metrics*"),         # skip these paths (glob patterns)
    log_request_body=True,     # capture request body
    log_response_body=True,    # capture response body
    log_headers=False,         # disabled by default — headers may contain auth tokens
)

Configuration is incremental — calling configure(log_headers=True) preserves all other settings. Call configure() with no arguments to reset to defaults.

What Gets Logged

Every outgoing HTTP request produces a row in api_logs:

Column	Type	Description
id	UUID	Primary key
method	VARCHAR(10)	GET, POST, PUT, DELETE, etc.
url	TEXT	Full request URL
host	VARCHAR(255)	Target hostname (indexed)
path	VARCHAR(2048)	URL path
status_code	INTEGER	Response status code, 0 if request failed (indexed)
request_headers	TEXT	JSON-encoded request headers (if enabled)
request_body	TEXT	Request body (truncated to max_body_size)
response_headers	TEXT	JSON-encoded response headers (if enabled)
response_body	TEXT	Response body (truncated to max_body_size)
duration_ms	FLOAT	Request duration in milliseconds
error	TEXT	Exception message if request failed
created_at	DATETIME	Timestamp in UTC (indexed)

Architecture

┌─────────────┐     ┌──────────────┐     ┌───────────┐     ┌──────────┐
│ httpx/       │────▶│  LogBuffer   │────▶│  _flush() │────▶│ Database │
│ requests     │     │ (thread-safe │     │ (batch    │     │          │
│ interceptor  │     │  in-memory)  │     │  insert)  │     │          │
└─────────────┘     └──────────────┘     └───────────┘     └──────────┘
                     ▲ batch_size trigger
                     ▲ timer trigger (flush_interval)
                     ▲ atexit trigger

• Interceptors wrap httpx transports and requests sessions to capture request/response data
• LogBuffer collects entries in a thread-safe list and flushes on batch size, timer, or process exit
• Integrations provide the _flush() implementation specific to each ORM (SQLAlchemy async, Tortoise bulk_create, Django raw SQL)

Supported Frameworks

Framework	ORM	Migration Tool	Admin Panel
Django	Django ORM	manage.py migrate	Django Admin
FastAPI	SQLAlchemy 2.0 (async)	Alembic --autogenerate	starlette-admin
FastAPI	Tortoise ORM	Aerich	—
Any	—	—	— (bring your own storage)

License

MIT