api-autopsy 0.1.0

Production-grade API performance analysis platform for Django and FastAPI

🔬 API Autopsy

Production-grade API performance analysis for Django and FastAPI.

Automatically detect performance issues — N+1 queries, blocking async calls, slow serializers — without external APM services.

---

⚡ Quickstart

Install

pip install api-autopsy

Django

# settings.py
INSTALLED_APPS += ["api_autopsy"]

MIDDLEWARE += ["api_autopsy.integrations.django.middleware.APIAutopsyMiddleware"]

FastAPI

from fastapi import FastAPI
from api_autopsy.integrations.fastapi import APIAutopsyMiddleware

app = FastAPI()
app.add_middleware(APIAutopsyMiddleware)

That's it. Visit /api-autopsy/ to see the dashboard.

---

🎯 What It Detects

Category	Detection
N+1 Queries	Repeated SQL patterns across a single request
Duplicate Queries	Identical SQL executed multiple times
Slow Queries	Individual queries exceeding thresholds
Missing Indexes	Slow WHERE clauses on non-primary columns
Slow Serializers	DRF serializer timing and nested cost
Async Violations	time.sleep(), requests.* in async handlers
Memory Bloat	Per-request memory allocation deltas
External Calls	Outbound HTTP call count and timing

Every detection comes with actionable recommendations.

---

📊 Performance Score

Each endpoint receives a 0–100 score based on:

• Latency (30%) — exponential decay from ideal threshold
• Query Efficiency (30%) — penalizes count, duplicates, N+1
• Memory Usage (20%) — penalizes large allocation deltas
• Async Correctness (20%) — binary penalties for blocking calls

---

🖥️ Dashboard

Auto-served at /api-autopsy/ with:

• Endpoint ranking table (sortable by score, latency, queries)
• Performance timeline (5-minute buckets over 24h)
• Slow endpoint heatmap (by hour)
• Analysis findings with severity and recommendations
• Dark/light mode toggle
• Auto-refresh every 30s

---

🔧 Configuration

Django

# settings.py
API_AUTOPSY = {
    "ENABLED": True,
    "STORAGE_BACKEND": "sqlite",      # sqlite | postgres | redis
    "STORAGE_DSN": "autopsy.db",
    "SAMPLE_RATE": 1.0,               # 0.0–1.0
    "RETENTION_HOURS": 72,
    "DASHBOARD_PREFIX": "/api-autopsy",
}

FastAPI

from api_autopsy.config import configure

configure(
    sample_rate=0.1,       # 10% sampling in production
    retention_hours=48,
    storage_dsn="/var/data/autopsy.db",
)

---

🚀 CI/CD Mode

# Generate JSON report
api-autopsy analyze --format json --output report.json

# Fail CI if score drops below threshold
api-autopsy analyze --threshold 70

# Detect regression against a baseline
api-autopsy analyze --baseline baseline.json --threshold 10

# Cleanup old data
api-autopsy cleanup --hours 24

Exit codes: 0 = pass, 1 = regression or critical findings.

---

🔌 Extensibility

Custom Analyzer

from api_autopsy.plugins import register_analyzer
from api_autopsy.analyzers.base import BaseAnalyzer
from api_autopsy.core.context import Finding

@register_analyzer
class MyCustomAnalyzer(BaseAnalyzer):
    def analyze(self, ctx):
        findings = []
        if ctx.duration_ms > 1000:
            findings.append(Finding(
                severity="warning",
                category="slow_response",
                message=f"Response took {ctx.duration_ms:.0f}ms",
                recommendation="Consider caching or pagination.",
            ))
        return findings

Via Entry Points

# pyproject.toml
[project.entry-points."api_autopsy.plugins"]
my_plugin = "my_package.autopsy:register"

---

🏗️ Architecture

api_autopsy/
├── config.py              # Central configuration + auto-detection
├── plugins.py             # Plugin discovery system
├── core/
│   ├── context.py         # contextvars-based RequestContext
│   └── registry.py        # Collector/Analyzer registry
├── collectors/
│   ├── base.py            # Abstract collector
│   ├── timing.py          # Request duration
│   ├── sql.py             # SQL queries + duplicates
│   ├── memory.py          # Memory delta (tracemalloc)
│   ├── http_calls.py      # Outbound HTTP tracking
│   └── cache.py           # Cache hit/miss tracking
├── analyzers/
│   ├── base.py            # Abstract analyzer
│   ├── orm.py             # N+1, duplicates, slow queries
│   ├── serializer.py      # DRF serializer profiling
│   └── async_safety.py    # Blocking I/O detection
├── scoring/
│   └── engine.py          # Weighted 0–100 scoring
├── storage/
│   ├── base.py            # Abstract storage
│   └── sqlite.py          # SQLite backend (default)
├── dashboard/
│   ├── views.py           # Django + ASGI handlers
│   └── templates/
│       └── index.html     # Dashboard UI
├── integrations/
│   ├── django/
│   │   ├── apps.py        # Django AppConfig
│   │   └── middleware.py   # WSGI middleware
│   └── fastapi/
│       └── middleware.py   # ASGI middleware
└── cli/
    └── main.py            # CLI entry point

---

🧪 Testing

pip install -e ".[dev]"
pytest tests/ -v

---

Production Safety

• Sampling: Set sample_rate to 0.01–0.1 for high-traffic
• Overhead: < 5% with sampling; collectors are lightweight
• Retention: Auto-cleanup via retention_hours or CLI
• Async-safe: Uses contextvars — no thread-local issues

---

License

MIT