Use timer as a standalone class so you have full control on when to call a lap (most useful in while loops)…
… or use track2 to directly track a for loop
N = 50
info = None
for i in (tracker := track2(range(N), total=N)):
time.sleep(0.01)
info = f"My Info: {i*3.122:.2f}"
if i == N // 2:
print()
if i >= N // 2:
tracker.send(info)25/50 (0.31s - 0.31s remaining - 81.55 iters/s)
My Info: 152.98 50/50 (0.61s - 0.00s remaining - 82.52 iters/s) Warning! NEVER RUN tracker.send(None) as this will skip variables silently
@timeit decorates and prints time taken to execute a function as an info log
[01/02/25 20:59:04] INFO 0.00 seconds to execute `foo` 3546464892.py:inner:16
[01/02/25 20:59:06] INFO 2.01 seconds to execute `foo` 3546464892.py:inner:16
30@io will decorate to show inputs and outputs (along with time taken to execute) as a debug log
@io
def foo(a, b=None):
if b is None:
return a + 1
else:
time.sleep(2)
return a + b
with debug_mode():
foo(10)
foo(10, b=20)[01/02/25 20:59:10] DEBUG 3679299354.py:<module>:11 0.00 seconds to execute `foo` args() 0 - 10 (🏷️ int) kwargs outputs - 11 (🏷️ int)
[01/02/25 20:59:12] DEBUG 3679299354.py:<module>:12 2.01 seconds to execute `foo` args() 0 - 10 (🏷️ int) kwargs b - 20 (🏷️ int) outputs - 30 (🏷️ int)
@io can be forced to print as log/trace if needed
@io(level="trace")
def foo(a, b=None):
if b is None:
return a + 1
else:
time.sleep(2)
return a + b
with trace_mode():
foo(10)
import time
time.sleep(1)
foo(10, b=20)[06/15/25 18:58:59] TRACE <ipython-input-1-57a17d8aa123>:<module>:11
0.00 seconds to execute `foo`
args()
0 - 10 (🏷️ int)
kwargs
outputs - 11 (🏷️ int)
[06/15/25 18:59:02] TRACE <ipython-input-1-57a17d8aa123>:<module>:15
2.01 seconds to execute `foo`
args()
0 - 10 (🏷️ int)
kwargs
b - 20 (🏷️ int)
outputs - 30 (🏷️ int)
Source path:... <ipython-input-1-57a17d8aa123> Starting var:.. a = 10 Starting var:.. b = None 18:58:59.237764 call 1 SOURCE IS UNAVAILABLE 18:58:59.237840 line 3 SOURCE IS UNAVAILABLE 18:58:59.237855 line 4 SOURCE IS UNAVAILABLE 18:58:59.237865 return 4 SOURCE IS UNAVAILABLE Return value:.. 11 Elapsed time: 00:00:00.000309 Source path:... <ipython-input-1-57a17d8aa123> Starting var:.. a = 10 Starting var:.. b = 20 18:59:00.636968 call 1 SOURCE IS UNAVAILABLE 18:59:00.637033 line 3 SOURCE IS UNAVAILABLE 18:59:00.637046 line 6 SOURCE IS UNAVAILABLE 18:59:02.642140 line 7 SOURCE IS UNAVAILABLE 18:59:02.642571 return 7 SOURCE IS UNAVAILABLE Return value:.. 30 Elapsed time: 00:00:02.006072
@tryy
def do(a, b, c):
return 1 / 0
x = do(1, 2, c=10)
assert x is None # tryy returns None by default[01/02/25 20:59:38] WARNING Error for `do`: ZeroDivisionError: division by zero 579939537.py:wrapper:27
Use your own default on failure
WARNING Error for `do`: ZeroDivisionError: division by zero 579939537.py:wrapper:27
'😔'Optionally print the full stacktrace if needed
@tryy(print_traceback=True, output_to_return_on_fail="😔")
def do(a, b, c):
return 1 / 0
do(1, 2, c=10)WARNING Error for `do`: 579939537.py:wrapper:32 Traceback (most recent call last): File "/var/folders/1_/71dqv9vx2750gmyz77q_f45w0000gn/T/ipykernel_70802/579939537.py", line 22, in wrapper return f(*args, **kwargs) ^^^^^^^^^^^^^^^^^^ File "/var/folders/1_/71dqv9vx2750gmyz77q_f45w0000gn/T/ipykernel_70802/580638143.py", line 3, in do return 1 / 0 ~~^~~ ZeroDivisionError: division by zero
'😔'You can also silence the errors completely
@tryy(silence_errors=True, output_to_return_on_fail="😔")
def do(a, b, c):
return 1 / 0
do(1, 2, c=10)'😔'You can collect all your errors in a list
import random
errors = []
@tryy(silence_errors=True, store_errors=errors)
def do(a, b, c):
if random.randint(0, 100) < 50:
return 1 / 0
else:
raise NotImplementedError("🤔")
for _ in range(4):
do(1, random.randint(0, 10), c=random.randint(0, 100))
print(errors)[{'func': 'do', 'args': (1, 4), 'kwargs': {'c': 53}, 'tb': None, 'err_type': 'ZeroDivisionError'}, {'func': 'do', 'args': (1, 3), 'kwargs': {'c': 68}, 'tb': None, 'err_type': 'ZeroDivisionError'}, {'func': 'do', 'args': (1, 1), 'kwargs': {'c': 2}, 'tb': None, 'err_type': 'NotImplementedError'}, {'func': 'do', 'args': (1, 10), 'kwargs': {'c': 96}, 'tb': None, 'err_type': 'NotImplementedError'}]There’s onlly one usecase where you would want to send in a list by yourself - when you want to append your errors to an existing list. The sensible default is to always store the errors, especially because this is a debugging tool.
Just access all the errors in a dataframe like so
import random
random.seed(10)
@tryy(silence_errors=True)
def do(a, b, c):
if c < 50:
return 1 / 0
else:
raise NotImplementedError("🤔")
for _ in range(4):
do(1, random.randint(0, 10), c=random.randint(0, 100))
do.error_summary()| func | args | kwargs | tb | err_type | |
|---|---|---|---|---|---|
| 0 | do | (1, 9) | {'c': 4} | None | ZeroDivisionError |
| 1 | do | (1, 6) | {'c': 61} | None | NotImplementedError |
| 2 | do | (1, 9) | {'c': 1} | None | ZeroDivisionError |
| 3 | do | (1, 3) | {'c': 59} | None | NotImplementedError |
and the actual list of errors like so
[{'func': 'do',
'args': (1, 9),
'kwargs': {'c': 4},
'tb': None,
'err_type': 'ZeroDivisionError'},
{'func': 'do',
'args': (1, 6),
'kwargs': {'c': 61},
'tb': None,
'err_type': 'NotImplementedError'},
{'func': 'do',
'args': (1, 9),
'kwargs': {'c': 1},
'tb': None,
'err_type': 'ZeroDivisionError'},
{'func': 'do',
'args': (1, 3),
'kwargs': {'c': 59},
'tb': None,
'err_type': 'NotImplementedError'}]Finally, you want to run the function (without try) to reproduce the error and actually start debugging. Just use the .F attribute to access the original function that you created
ix = 2
data = do.error_store[ix]
try:
do.F(*data.args, **data.kwargs)
except Exception as e:
print(f"ERROR: ", e)ERROR: 'dict' object has no attribute 'args'cast_inputs (func)