Siren
Minimal Python debug helper with automatic cleanup.
A tiny debugging utility for Python that prints variables with file/line context, traces function calls, measures execution time, and safely removes debug calls from your code.
Install
pip install siren-debug
The package also installs two commands: siren-clean (remove debug calls) and siren-autoload (use siren without importing it).
Quick Start
from siren import siren
x = 10
user = {"name": "Alex", "items": [1, 2, 3]}
siren(x)
siren(user)
[🧜 SIREN core.py:10] x = 10
[🧜 SIREN core.py:11] user = {'name': 'Alex', 'items': [1, 2, 3]}
Siren automatically uses pprint for complex objects, and picks up the file/line it was called from.
Features
- Works with Python 2.7 and 3.6+
- Zero external dependencies
- Prints values with file and line number
- Uses
pprintautomatically for complex data - Function tracing with
@siren.trace, object diffing withsiren.diff, an interactivesiren.breakpoint(), memory snapshots withsiren.memory(), and colored traceback capture withsiren.catch - Quiet mode, conditional logging, and file logging
- Removes
siren(...)calls automatically withsiren-clean - Use
sirenanywhere without importing it viasiren-autoload - Project/file scaffolding with
siren-scaffold,.envdrift checks withsiren-env - Terminal snippet manager (
siren-snippet) and a dependency-free HTTP client (siren-http) - Local code-quality checks with
siren-quality(dead code, lint, cyclomatic complexity) - Works in scripts, CLI tools, Django, Flask, FastAPI, and more
- Colored output with emoji for easy visual scanning
Usage
Call siren(...) with one or more values. It returns them unchanged, so it can be inlined:
from siren import siren
siren(x, data, user)
result = siren(compute()) # still returns compute()'s value
Label — tag a call for easier scanning:
siren(value, label="BEFORE SAVE")
Timer — measure execution time for a call:
siren(x, timeit=True)
# [🧜 SIREN core.py:10] x = 10
# [🧜 SIREN TIME] 0.000123s
Quiet mode — suppress output without removing the call:
siren(x, quiet=True) # this call only, still returns x
siren.set_quiet(True) # every call, until set_quiet(False)
Conditional logging — only print when a condition holds:
siren(x, if_equals=5) # only if x == 5
siren(items, if_len_gt=100) # only if len(items) > 100
siren(items, if_len_lt=5) # only if len(items) < 5
siren(result, if_true=True) # only if result is truthy
siren(error, if_false=True) # only if error is falsy
Logging to file — mirror output to a file:
siren.set_logfile("debug.log")
siren(x) # prints to stdout AND writes to debug.log
Inspect configuration:
config = siren.get_config()
print(config) # {"quiet": False, "logfile": None, "enabled": True}
Function tracing
@siren.trace logs a function's calls, arguments, return value, execution time, and exceptions automatically:
from siren import trace
@siren.trace
def add(a, b):
return a + b
add(2, 3)
[🧜 SIREN core.py:10] Calling add(a=2, b=3)
[🧜 SIREN core.py:11] Returned from add -> 5 [int] (0.000123s)
Configuration options (all default to True):
| Option | Effect |
|---|---|
timeit |
Show execution time |
show_args |
Show function arguments |
show_return |
Show return value |
show_type |
Show return type in brackets |
@siren.trace(timeit=True, show_args=False, show_type=False)
def multiply(a, b):
return a * b
Exceptions are logged before being re-raised, so @siren.trace never swallows an error:
@siren.trace
def divide(a, b):
return a / b
divide(5, 0) # Logs exception before raising
Diff, breakpoint, memory, and catch
siren.diff compares two dicts, lists, tuples, or any comparable objects:
before = {"name": "Alice", "age": 30}
after = {"name": "Alice", "age": 31, "city": "NYC"}
siren.diff(before, after)
[🧜 SIREN test.py:10] DIFF
[🧜 SIREN test.py:11] [~] age: 30 → 31 (changed)
[🧜 SIREN test.py:12] [+] city: NYC (new)
siren.breakpoint() pauses execution and prints local variables:
x = 42
data = {"items": [1, 2, 3]}
siren.breakpoint() # Pauses and displays all locals
# Press Ctrl+C to continue, or type 'd' to drop into pdb
siren.memory() prints current/peak traced memory usage (requires Python 3.4+; prints a clear message instead of failing on Python 2):
siren.memory() # [🧜 SIREN MEMORY ...] current=1.2MB peak=1.5MB
siren.memory(top=5) # also print the top 5 allocation sites
siren.catch is a context manager that prints a colored traceback on exception and re-raises it — it never swallows errors:
with siren.catch():
risky_call()
Cleaning debug calls
Run siren-clean in a project folder to remove all siren(...) calls and their import lines — comments and string literals are left untouched:
siren-clean
Before:
from siren import siren
siren(x)
print("hello")
siren(data)
After:
print("hello")
Autoload (no per-file imports)
By default you still need from siren import siren in every file that uses it. If you'd rather call siren(x) anywhere in a project without importing it each time, enable autoload once per environment (virtualenv, Docker image, CI job, etc.):
siren-autoload on
siren-autoload status # check whether it's enabled
siren-autoload off # disable again
This writes a .pth file into the current environment's site-packages, injecting siren into Python's builtins as soon as any interpreter starts in that environment — no import needed anywhere, including in Django apps, Flask views, scripts, or the shell. It's opt-in per environment, so it won't silently affect environments where you didn't run on.
Beyond debugging
Siren also ships a handful of small, dependency-free CLI tools for everyday project work.
Scaffolding — siren-scaffold
Generate a small file or project skeleton:
siren-scaffold script my_tool # a single script with a main() guard
siren-scaffold package my_package # a package dir with __init__.py, core.py, and tests/
siren-scaffold class Widget # a plain class
siren-scaffold dataclass Point # a plain-Python value object (no dataclasses module needed)
siren-scaffold test Widget # a unittest.TestCase stub
It refuses to overwrite existing files.
.env drift check — siren-env
siren-env diff # compares .env.example against .env
siren-env diff --example .env.sample --env .env.local
Reports keys present in one file but missing from the other, and exits non-zero on drift — usable as a CI check.
Snippets — siren-snippet
echo "print('hello')" | siren-snippet save greet --tag python
siren-snippet save query --file query.sql --tag sql # from a file instead of stdin
siren-snippet show greet
siren-snippet copy greet # sends it straight to the clipboard
siren-snippet edit greet # opens it in $EDITOR
siren-snippet rename greet hello
siren-snippet list [--tag sql]
siren-snippet tags # every tag in use, with counts
siren-snippet search select # matches by name, tag, or content
siren-snippet remove greet
save refuses to overwrite an existing snippet unless you pass --force — this also applies to rename.
Snippets can hold {{placeholder}} markers, filled in on the way out instead of when saved:
echo 'SELECT * FROM {{table}};' | siren-snippet save query --tag sql
siren-snippet copy query --var table=users # copies "SELECT * FROM users;"
siren-snippet show query --var table=users # same, printed instead of copied
Back up or move your snippets between machines with export/import (content, tags, and timestamps all round-trip; import skips names that already exist unless you pass --force):
siren-snippet export backup.json
siren-snippet import backup.json
Snippets are stored as plain text files under ~/.siren/snippets/, with tags/timestamps tracked separately in ~/.siren/snippets/_index.json (so any snippet saved before this existed keeps working unchanged, just without tags).
HTTP client — siren-http
A tiny httpie-like client built on urllib only:
siren-http GET https://api.example.com/items
siren-http POST https://api.example.com/items --json '{"name": "x"}' -H "Authorization: Bearer TOKEN"
siren-http GET https://api.example.com/items --save my-request # save it as a local collection
siren-http replay my-request # resend a saved request
siren-http list # list saved requests
You can also log every HTTP call your own code makes through requests or httpx, without touching that code — requests/httpx are not siren dependencies, they're only imported when you call these:
siren.patch_requests() # every requests.Session call now logs method/url/status/duration
siren.patch_httpx() # same, for httpx.Client (sync only)
siren.unpatch_requests()
siren.unpatch_httpx()
Code quality — siren-quality
Local checks built on the stdlib ast module (no pyflakes/radon/etc dependency):
siren-quality deadcode . # unused imports and module-level defs never referenced in the same file
siren-quality lint . # bare `except:`, leftover pdb.set_trace()/breakpoint(), TODO/FIXME comments
siren-quality complexity . # cyclomatic complexity per function, flags anything above --threshold (default 10)
deadcode is a same-file heuristic — it can't see usage from other files, so treat its findings as candidates to double-check, not certainties.
Pro tier
Everything above is free and runs entirely offline. The siren-debug package also ships a couple of pro-tier commands that talk to a small backend (separate, closed-source repo) for a paid feature: exception capture with a searchable history, instead of only a local siren.catch().
siren-login signup you@example.com # creates an account + API key, stored in ~/.siren/credentials.json
siren-login status # check your plan/license
siren-login logout
try:
risky()
except Exception:
siren.report() # sends the exception (with traceback) to your workspace
siren-events list # recent exceptions reported from any of your machines
siren-events show <id> # full traceback for one of them
siren.report() never raises on its own — if you're not logged in, or the backend can't be reached, it prints a message and returns None instead of breaking your error handling. Point the CLI at a different backend with SIREN_API_URL (defaults to the hosted one). The hosted backend runs on Render's free tier, so it sleeps after inactivity — the first request after a while can take 30-60s to wake it up.
Subscribing:
siren-login upgrade --currency brl # or usd / eur — prints a Stripe Checkout link to open in a browser
Team workspaces — invite a teammate (creates their account if they don't have one yet, and hands you their API key to pass along since there's no email delivery yet):
siren-login invite teammate@example.com
Notifications — post to a Slack/Discord incoming webhook whenever an exception is captured for your workspace:
siren-login set-webhook https://hooks.slack.com/services/...
siren-login set-webhook # no URL clears it
Framework examples
Django
from django.http import JsonResponse
from siren import siren
def my_view(request):
user_data = request.GET.dict()
siren(user_data, label="REQUEST_PARAMS")
result = process_data(user_data)
siren(result)
return JsonResponse(result)
Flask
from flask import Flask, request
from siren import siren, trace
app = Flask(__name__)
@app.route("/api/users")
def get_users():
query = request.args.get("q")
siren(query, label="SEARCH_QUERY")
users = search_users(query)
return {"users": users}
@siren.trace
def search_users(query):
# Function entry/exit will be logged automatically
return [{"id": 1, "name": "Alice"}]
FastAPI
from fastapi import FastAPI
from siren import siren, trace
app = FastAPI()
@app.get("/items/{item_id}")
async def get_item(item_id: int, q: str = None):
siren({"item_id": item_id, "q": q}, label="QUERY_PARAMS")
item = await fetch_item(item_id)
return item
@siren.trace(timeit=True)
async def fetch_item(item_id: int):
# Execution time and arguments will be logged
return {"id": item_id, "name": "Item"}
Why use Siren?
Debug prints are easy to add, but hard to remove later. Siren gives you a fast debug workflow and a safe cleanup step so your temporary debug code does not stay in production.
Project
- Package name:
siren-debug - Python versions:
2.7,3.6+ - License: MIT
- PyPI: https://pypi.org/project/siren-debug/
License
MIT
Release files for siren-debug 0.7.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| siren_debug-0.7.0.tar.gz | 43.4 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| siren_debug-0.7.0-py2.py3-none-any.whl | Python 2, Python 3 | none | any | Details |
Total release size: 77.6 kB
Release files / siren_debug-0.7.0.tar.gz
| Download URL | siren_debug-0.7.0.tar.gz |
|---|---|
| Size | 43.4 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
e044bd1eb1d7907a59e3fd06b3bed170bae3eedc8a25698fecb75d6f428fd19f
|
|
BLAKE2b-256 checksum How to use checksums |
6b942a940b27f24fe894590ce041cd67f0b8b9b5d0a005619490bef42f05db13
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 20, 2026.
Transparency logRelease files / siren_debug-0.7.0-py2.py3-none-any.whl
| Download URL | siren_debug-0.7.0-py2.py3-none-any.whl |
|---|---|
| Size | 34.3 kB |
| Tags | Python 2 Python 3 |
|
SHA-256 checksum How to use checksums |
56face9d34ec45afccd881f471ee07bd686bc4ed71bb5bc8759b6e6cad6f3ac0
|
|
BLAKE2b-256 checksum How to use checksums |
5018f28e7557b39ab72c150b843b8da4ebdeafccfddf640ce38b5f8ebd756fad
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 20, 2026.
Transparency log