Home - News ( United States | United Kingdom | Italy | Germany ) - Football scores

Showing content from https://gto76.github.io/python-cheatsheet/ below:

Comprehensive Python Cheatsheet

ToC = {
    '1. Collections': [List, Dictionary, Set, Tuple, Range, Enumerate, Iterator, Generator],
    '2. Types':       [Type, String, Regular_Exp, Format, Numbers, Combinatorics, Datetime],
    '3. Syntax':      [Function, Inline, Import, Decorator, Class, Duck_Type, Enum, Except],
    '4. System':      [Exit, Print, Input, Command_Line_Arguments, Open, Path, OS_Commands],
    '5. Data':        [JSON, Pickle, CSV, SQLite, Bytes, Struct, Array, Memory_View, Deque],
    '6. Advanced':    [Operator, Match_Stmt, Logging, Introspection, Threading, Coroutines],
    '7. Libraries':   [Progress_Bar, Plot, Table, Console_App, GUI, Scraping, Web, Profile],
    '8. Multimedia':  [NumPy, Image, Animation, Audio, Synthesizer, Pygame, Pandas, Plotly]
}

if __name__ == '__main__':      
    main()                      

<list> = [<el_1>, <el_2>, ...]  

<el>   = <list>[index]          
<list> = <list>[<slice>]        

<list>.append(<el>)             
<list>.extend(<collection>)     

<list>.sort()                   
<list>.reverse()                
<list> = sorted(<collection>)   
<iter> = reversed(<list>)       

<el>  = max(<collection>)       
<num> = sum(<collection>)       

elementwise_sum  = [sum(pair) for pair in zip(list_a, list_b)]
sorted_by_second = sorted(<collection>, key=lambda el: el[1])
sorted_by_both   = sorted(<collection>, key=lambda el: (el[1], el[0]))
flatter_list     = list(itertools.chain.from_iterable(<list>))

• For details about sort(), sorted(), min() and max() see Sortable.
• Module operator has function itemgetter() that can replace listed lambdas.
• This text uses the term collection instead of iterable. For rationale see Collection.

<int> = len(<list>)             
<int> = <list>.count(<el>)      
<int> = <list>.index(<el>)      
<el>  = <list>.pop()            
<list>.insert(<int>, <el>)      
<list>.remove(<el>)             
<list>.clear()                  

<dict> = {key_1: val_1, key_2: val_2, ...}      

<view> = <dict>.keys()                          
<view> = <dict>.values()                        
<view> = <dict>.items()                         

value  = <dict>.get(key, default=None)          
value  = <dict>.setdefault(key, default=None)   
<dict> = collections.defaultdict(<type>)        
<dict> = collections.defaultdict(lambda: 1)     

<dict> = dict(<collection>)                     
<dict> = dict(zip(keys, values))                
<dict> = dict.fromkeys(keys [, value])          

<dict>.update(<dict>)                           
value = <dict>.pop(key)                         
{k for k, v in <dict>.items() if v == value}    
{k: v for k, v in <dict>.items() if k in keys}  

>>> from collections import Counter
>>> counter = Counter(['blue', 'blue', 'blue', 'red', 'red'])
>>> counter['yellow'] += 1
>>> print(counter.most_common())
[('blue', 3), ('red', 2), ('yellow', 1)]

<set> = {<el_1>, <el_2>, ...}                   

<set>.add(<el>)                                 
<set>.update(<collection> [, ...])              

<set>  = <set>.union(<coll.>)                   
<set>  = <set>.intersection(<coll.>)            
<set>  = <set>.difference(<coll.>)              
<set>  = <set>.symmetric_difference(<coll.>)    
<bool> = <set>.issubset(<coll.>)                
<bool> = <set>.issuperset(<coll.>)              

<el> = <set>.pop()                              
<set>.remove(<el>)                              
<set>.discard(<el>)                             

• Is immutable and hashable.
• That means it can be used as a key in a dictionary or as an element in a set.

<frozenset> = frozenset(<collection>)

Tuple is an immutable and hashable list.

<tuple> = ()                               
<tuple> = (<el>,)                          
<tuple> = (<el_1>, <el_2> [, ...])         

Tuple's subclass with named elements.

>>> from collections import namedtuple
>>> Point = namedtuple('Point', 'x y')
>>> p = Point(1, y=2)
>>> print(p)
Point(x=1, y=2)
>>> p.x, p[1]
(1, 2)

Immutable and hashable sequence of integers.

<range> = range(stop)                      
<range> = range(start, stop)               
<range> = range(start, stop, ±step)        

>>> [i for i in range(3)]
[0, 1, 2]

for i, el in enumerate(<coll>, start=0):   
    ...

Potentially endless stream of elements.

<iter> = iter(<collection>)                
<iter> = iter(<function>, to_exclusive)    
<el>   = next(<iter> [, default])          
<list> = list(<iter>)                      

import itertools as it

<iter> = it.count(start=0, step=1)         
<iter> = it.repeat(<el> [, times])         
<iter> = it.cycle(<collection>)            

<iter> = it.chain(<coll>, <coll> [, ...])  
<iter> = it.chain.from_iterable(<coll>)    

<iter> = it.islice(<coll>, to_exclusive)   
<iter> = it.islice(<coll>, from_inc, …)    

• Any function that contains a yield statement returns a generator.
• Generators and iterators are interchangeable.

def count(start, step):
    while True:
        yield start
        start += step

>>> counter = count(10, 2)
>>> next(counter), next(counter), next(counter)
(10, 12, 14)

• Everything is an object.
• Every object has a type.
• Type and class are synonymous.

<type> = type(<el>)                          
<bool> = isinstance(<el>, <type>)            

>>> type('a'), 'a'.__class__, str
(<class 'str'>, <class 'str'>, <class 'str'>)

from types import FunctionType, MethodType, LambdaType, GeneratorType, ModuleType

Each abstract base class specifies a set of virtual subclasses. These classes are then recognized by isinstance() and issubclass() as subclasses of the ABC, although they are really not. ABC can also manually decide whether or not a specific class is its virtual subclass, usually based on which methods the class has implemented. For instance, Iterable ABC looks for method iter(), while Collection ABC looks for iter(), contains() and len().

>>> from collections.abc import Iterable, Collection, Sequence
>>> isinstance([1, 2, 3], Iterable)
True

┏━━━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━┯━━━━━━━━━━━━┯━━━━━━━━━━━━┓
┃                  │  Iterable  │ Collection │  Sequence  ┃
┠──────────────────┼────────────┼────────────┼────────────┨
┃ list, range, str │     ✓      │     ✓      │     ✓      ┃
┃ dict, set        │     ✓      │     ✓      │            ┃
┃ iter             │     ✓      │            │            ┃
┗━━━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━┷━━━━━━━━━━━━┷━━━━━━━━━━━━┛

>>> from numbers import Number, Complex, Real, Rational, Integral
>>> isinstance(123, Number)
True

┏━━━━━━━━━━━━━━━━━━━━┯━━━━━━━━━━┯━━━━━━━━━━┯━━━━━━━━━━┯━━━━━━━━━━┯━━━━━━━━━━┓
┃                    │  Number  │  Complex │   Real   │ Rational │ Integral ┃
┠────────────────────┼──────────┼──────────┼──────────┼──────────┼──────────┨
┃ int                │    ✓     │    ✓     │    ✓     │    ✓     │    ✓     ┃
┃ fractions.Fraction │    ✓     │    ✓     │    ✓     │    ✓     │          ┃
┃ float              │    ✓     │    ✓     │    ✓     │          │          ┃
┃ complex            │    ✓     │    ✓     │          │          │          ┃
┃ decimal.Decimal    │    ✓     │          │          │          │          ┃
┗━━━━━━━━━━━━━━━━━━━━┷━━━━━━━━━━┷━━━━━━━━━━┷━━━━━━━━━━┷━━━━━━━━━━┷━━━━━━━━━━┛

Immutable sequence of characters.

<str>  = <str>.strip()                       
<str>  = <str>.strip('<chars>')              

<list> = <str>.split()                       
<list> = <str>.split(sep=None, maxsplit=-1)  
<list> = <str>.splitlines(keepends=False)    
<str>  = <str>.join(<coll_of_strings>)       

<bool> = <sub_str> in <str>                  
<bool> = <str>.startswith(<sub_str>)         
<int>  = <str>.find(<sub_str>)               

<str>  = <str>.lower()                       
<str>  = <str>.casefold()                    
<str>  = <str>.replace(old, new [, count])   
<str>  = <str>.translate(<table>)            

<str>  = chr(<int>)                          
<int>  = ord(<str>)                          

• Use 'unicodedata.normalize("NFC", <str>)' on strings like 'Motörhead' before comparing them to other strings, because 'ö' can be stored as one or two characters.
• 'NFC' converts such characters to a single character, while 'NFD' converts them to two.

<bool> = <str>.isdecimal()                   
<bool> = <str>.isdigit()                     
<bool> = <str>.isnumeric()                   
<bool> = <str>.isalnum()                     
<bool> = <str>.isprintable()                 
<bool> = <str>.isspace()                     

Functions for regular expression matching.

import re
<str>   = re.sub(r'<regex>', new, text, count=0)  
<list>  = re.findall(r'<regex>', text)            
<list>  = re.split(r'<regex>', text, maxsplit=0)  
<Match> = re.search(r'<regex>', text)             
<Match> = re.match(r'<regex>', text)              
<iter>  = re.finditer(r'<regex>', text)           

• Raw string literals do not interpret escape sequences, thus enabling us to use regex-specific escape sequences that cause SyntaxWarning in normal string literals (since 3.12).
• Argument 'new' of re.sub() can be a function that accepts Match object and returns a str.
• Argument 'flags=re.IGNORECASE' can be used with all functions that are listed above.
• Argument 'flags=re.MULTILINE' makes '^' and '$' match the start/end of each line.
• Argument 'flags=re.DOTALL' makes '.' also accept the '\n' (besides all other chars).
• 're.compile(<regex>)' returns a Pattern object with methods sub(), findall(), etc.

<str>   = <Match>.group()                         
<str>   = <Match>.group(1)                        
<tuple> = <Match>.groups()                        
<int>   = <Match>.start()                         
<int>   = <Match>.end()                           

'\d' == '[0-9]'                                   
'\w' == '[a-zA-Z0-9_]'                            
'\s' == '[ \t\n\r\f\v]'                           

• By default, decimal characters and alphanumerics from all alphabets are matched unless 'flags=re.ASCII' is used. It restricts special sequence matches to the first 128 Unicode characters and also prevents '\s' from accepting '\x1c', '\x1d', '\x1e' and '\x1f' (non-printable characters that divide text into files, tables, rows and fields, respectively).
• Use a capital letter for negation (all non-ASCII characters will be matched when used in combination with ASCII flag).

<str> = f'{<el_1>}, {<el_2>}'            
<str> = '{}, {}'.format(<el_1>, <el_2>)  
<str> = '%s, %s' % (<el_1>, <el_2>)      

>>> Person = collections.namedtuple('Person', 'name height')
>>> person = Person('Jean-Luc', 187)
>>> f'{person.name} is {person.height / 100} meters tall.'
'Jean-Luc is 1.87 meters tall.'

{<el>:<10}                               
{<el>:^10}                               
{<el>:>10}                               
{<el>:.<10}                              
{<el>:0}                                 

• Objects are rendered by calling the 'format(<el>, "<options>")' function.
• Options inside curly braces can be generated dynamically: f'{<el>:{<str/int>}[…]}'.
• Adding '=' to the expression prepends it to the output: f'{1+1=}' returns '1+1=2'.
• Adding '!r' to the expression converts object to string by calling its repr() method.

{'abcde':10}                             
{'abcde':10.3}                           
{'abcde':.3}                             
{'abcde'!r:10}                           

{123456:10}                              
{123456:10,}                             
{123456:10_}                             
{123456:+10}                             
{123456:=+10}                            
{123456: }                               
{-123456: }                              

{1.23456:10.3}                           
{1.23456:10.3f}                          
{1.23456:10.3e}                          
{1.23456:10.3%}                          

┏━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━━┓
┃              │    {<float>}   │   {<float>:f}  │   {<float>:e}  │   {<float>:%}  ┃
┠──────────────┼────────────────┼────────────────┼────────────────┼────────────────┨
┃  0.000056789 │   '5.6789e-05' │    '0.000057'  │ '5.678900e-05' │    '0.005679%' ┃
┃  0.00056789  │   '0.00056789' │    '0.000568'  │ '5.678900e-04' │    '0.056789%' ┃
┃  0.0056789   │   '0.0056789'  │    '0.005679'  │ '5.678900e-03' │    '0.567890%' ┃
┃  0.056789    │   '0.056789'   │    '0.056789'  │ '5.678900e-02' │    '5.678900%' ┃
┃  0.56789     │   '0.56789'    │    '0.567890'  │ '5.678900e-01' │   '56.789000%' ┃
┃  5.6789      │   '5.6789'     │    '5.678900'  │ '5.678900e+00' │  '567.890000%' ┃
┃ 56.789       │  '56.789'      │   '56.789000'  │ '5.678900e+01' │ '5678.900000%' ┃
┗━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━━┛

┏━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━━┓
┃              │  {<float>:.2}  │  {<float>:.2f} │  {<float>:.2e} │  {<float>:.2%} ┃
┠──────────────┼────────────────┼────────────────┼────────────────┼────────────────┨
┃  0.000056789 │    '5.7e-05'   │      '0.00'    │   '5.68e-05'   │      '0.01%'   ┃
┃  0.00056789  │    '0.00057'   │      '0.00'    │   '5.68e-04'   │      '0.06%'   ┃
┃  0.0056789   │    '0.0057'    │      '0.01'    │   '5.68e-03'   │      '0.57%'   ┃
┃  0.056789    │    '0.057'     │      '0.06'    │   '5.68e-02'   │      '5.68%'   ┃
┃  0.56789     │    '0.57'      │      '0.57'    │   '5.68e-01'   │     '56.79%'   ┃
┃  5.6789      │    '5.7'       │      '5.68'    │   '5.68e+00'   │    '567.89%'   ┃
┃ 56.789       │    '5.7e+01'   │     '56.79'    │   '5.68e+01'   │   '5678.90%'   ┃
┗━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━━┛

• '{<float>:g}' is '{<float>:.6}' with stripped zeros, exponent starting at '1e+06'.
• When both rounding up and rounding down are possible, the one that returns result with even last digit is chosen. That makes '{6.5:.0f}' a '6' and '{7.5:.0f}' an '8'.
• This rule only effects numbers that can be represented exactly by a float (.5, .25, …).

{90:c}                                   
{90:b}                                   
{90:X}                                   

<int>      = int(<float/str/bool>)             
<float>    = float(<int/str/bool>)             
<complex>  = complex(real=0, imag=0)           
<Fraction> = fractions.Fraction(<int>, <int>)  
<Decimal>  = decimal.Decimal(<str/int/tuple>)  

• 'int(<str>)' and 'float(<str>)' raise ValueError if passed string is malformed.
• Decimal objects store numbers exactly, unlike most floats where '1.1 + 2.2 != 3.3'.
• Floats can be compared with: 'math.isclose(<float>, <float>, rel_tol=1e-09)'.
• Precision of decimal operations is set with: 'decimal.getcontext().prec = <int>'.
• Bools can be used anywhere ints can, because bool is a subclass of int: 'True + 1 == 2'.

<num> = pow(<num>, <num>)                      
<num> = abs(<num>)                             
<num> = round(<num> [, ±ndigits])              
<num> = min(<collection>)                      
<num> = sum(<collection>)                      

from math import floor, ceil, trunc            
from math import pi, inf, nan, isnan           
from math import sqrt, factorial               
from math import sin, cos, tan                 
from math import log, log10, log2              

from statistics import mean, median, mode      
from statistics import variance, stdev         

from random import random, randint, uniform    

<float> = random()                             
<num>   = randint/uniform(a, b)                
<float> = gauss(mean, stdev)                   
<el>    = choice(<sequence>)                   
shuffle(<list>)                                

<int> = 0x<hex>                                
<int> = int('±<hex>', 16)                      
<str> = hex(<int>)                             

<int> = <int> & <int>                          
<int> = <int> | <int>                          
<int> = <int> ^ <int>                          
<int> = <int> << n_bits                        
<int> = ~<int>                                 

import itertools as it

>>> list(it.product('abc', repeat=2))        
[('a', 'a'), ('a', 'b'), ('a', 'c'),         
 ('b', 'a'), ('b', 'b'), ('b', 'c'),         
 ('c', 'a'), ('c', 'b'), ('c', 'c')]         

>>> list(it.permutations('abc', 2))          
[('a', 'b'), ('a', 'c'),                     
 ('b', 'a'), ('b', 'c'),                     
 ('c', 'a'), ('c', 'b')]                     

>>> list(it.combinations('abc', 2))          
[('a', 'b'), ('a', 'c'),                     
 ('b', 'c')                                  
]                                            

Provides 'date', 'time', 'datetime' and 'timedelta' classes. All are immutable and hashable.

from datetime import date, time, datetime, timedelta, timezone
import zoneinfo, dateutil.tz

<D>  = date(year, month, day)               
<T>  = time(hour=0, minute=0, second=0)     
<DT> = datetime(year, month, day, hour=0)   
<TD> = timedelta(weeks=0, days=0, hours=0)  

• Times and datetimes that have defined timezone are called aware and ones that don't, naive. If time or datetime object is naive, it is presumed to be in the system's timezone!
• 'fold=1' means the second pass in case of time jumping back (usually for one hour).
• Timedelta normalizes arguments to ±days, seconds (< 86 400) and microseconds (< 1M). Its str() method returns '[±D, ]H:MM:SS[.…]' and total_seconds() a float of all seconds.
• Use '<D/DT>.weekday()' to get the day of the week as an integer, with Monday being 0.

<D/DTn> = D/DT.today()                      
<DTa>   = DT.now(<tzinfo>)                  

• To extract time use '<DTn>.time()', '<DTa>.time()' or '<DTa>.timetz()'.

<tzinfo> = timezone.utc                     
<tzinfo> = timezone(<timedelta>)            
<tzinfo> = dateutil.tz.tzlocal()            
<tzinfo> = zoneinfo.ZoneInfo('<iana_key>')  
<DTa>    = <DT>.astimezone([<tzinfo>])      
<Ta/DTa> = <T/DT>.replace(tzinfo=<tzinfo>)  

• Timezones returned by tzlocal(), ZoneInfo() and implicit local timezone of naive objects have offsets that vary through time due to DST and historical changes of the base offset.
• To get ZoneInfo() to work on Windows run '> pip3 install tzdata'.

<D/T/DT> = D/T/DT.fromisoformat(<str>)      
<DT>     = DT.strptime(<str>, '<format>')   
<D/DTn>  = D/DT.fromordinal(<int>)          
<DTn>    = DT.fromtimestamp(<float>)        
<DTa>    = DT.fromtimestamp(<float>, <tz>)  

• ISO strings come in following forms: 'YYYY-MM-DD', 'HH:MM:SS.mmmuuu[±HH:MM]', or both separated by an arbitrary character. All parts following the hours are optional.
• Python uses the Unix Epoch: '1970-01-01 00:00 UTC', '1970-01-01 01:00 CET', …

<str>    = <D/T/DT>.isoformat(sep='T')      
<str>    = <D/T/DT>.strftime('<format>')    
<int>    = <D/DT>.toordinal()               
<float>  = <DTn>.timestamp()                
<float>  = <DTa>.timestamp()                

>>> dt = datetime.strptime('2025-08-14 23:39:00.00 +0200', '%Y-%m-%d %H:%M:%S.%f %z')
>>> dt.strftime("%dth of %B '%y (%a), %I:%M %p %Z")
"14th of August '25 (Thu), 11:39 PM UTC+02:00"

• '%z' accepts '±HH[:]MM' and returns '±HHMM' or empty string if object is naive.
• '%Z' accepts 'UTC/GMT' and local timezone's code and returns timezone's name, 'UTC[±HH:MM]' if timezone is nameless, or an empty string if object is naive.

<bool>   = <D/T/DTn> > <D/T/DTn>            
<bool>   = <DTa>     > <DTa>                
<TD>     = <D/DTn>   - <D/DTn>              
<TD>     = <DTa>     - <DTa>                
<D/DT>   = <D/DT>    ± <TD>                 
<TD>     = <TD>      * <float>              
<float>  = <TD>      / <TD>                 

Independent block of code that returns a value when called.

def <func_name>(<nondefault_args>): ...                  
def <func_name>(<default_args>): ...                     
def <func_name>(<nondefault_args>, <default_args>): ...  

• Function returns None if it doesn't encounter 'return <obj/exp>' statement.
• Run 'global <var_name>' inside the function before assigning to global variable.
• Default values are evaluated when function is first encountered in the scope. Any mutation of a mutable default value will persist between invocations!

<obj> = <function>(<positional_args>)                    
<obj> = <function>(<keyword_args>)                       
<obj> = <function>(<positional_args>, <keyword_args>)    

Splat expands a collection into positional arguments, while splatty-splat expands a dictionary into keyword arguments.

args, kwargs = (1, 2), {'z': 3}
func(*args, **kwargs)

func(1, 2, z=3)

Splat combines zero or more positional arguments into a tuple, while splatty-splat combines zero or more keyword arguments into a dictionary.

def add(*a):
    return sum(a)

>>> add(1, 2, 3)
6

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━━┓
┃                           │  func(1, 2)  │ func(1, y=2) │ func(x=1, y=2) ┃
┠───────────────────────────┼──────────────┼──────────────┼────────────────┨
┃ func(x, *args, **kwargs): │      ✓       │      ✓       │       ✓        ┃
┃ func(*args, y, **kwargs): │              │      ✓       │       ✓        ┃
┃ func(*, x, **kwargs):     │              │              │       ✓        ┃
┗━━━━━━━━━━━━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━━┛

<list>  = [*<collection> [, ...]]  
<tuple> = (*<collection>, [...])   
<set>   = {*<collection> [, ...]}  
<dict>  = {**<dict> [, ...]}       

head, *body, tail = <collection>   

<func> = lambda: <return_value>                     
<func> = lambda <arg_1>, <arg_2>: <return_value>    

<list> = [i+1 for i in range(10)]                   
<iter> = (i for i in range(10) if i > 5)            
<set>  = {i+5 for i in range(10)}                   
<dict> = {i: i*2 for i in range(10)}                

>>> [l+r for l in 'abc' for r in 'abc']             
['aa', 'ab', 'ac', ..., 'cc']

from functools import reduce

<iter> = map(lambda x: x + 1, range(10))            
<iter> = filter(lambda x: x > 5, range(10))         
<obj>  = reduce(lambda out, x: out + x, range(10))  

<bool> = any(<collection>)                          
<bool> = all(<collection>)                          

<obj> = <exp> if <condition> else <exp>             

>>> [i if i else 'zero' for i in (0, 1, 2, 3)]      
['zero', 1, 2, 3]

<obj> = <exp> and <exp> [and ...]                   
<obj> = <exp> or <exp> [or ...]                     

>>> [i for ch in '0123' if (i := int(ch)) > 0]      
[1, 2, 3]

from collections import namedtuple
Point = namedtuple('Point', 'x y')                  
point = Point(0, 0)                                 

from enum import Enum
Direction = Enum('Direction', 'N E S W')            
direction = Direction.N                             

from dataclasses import make_dataclass
Player = make_dataclass('Player', ['loc', 'dir'])   
player = Player(point, direction)                   

Mechanism that makes code in one file available to another file.

import <module>            
import <package>           
import <package>.<module>  

• Package is a collection of modules, but it can also define its own functions, classes, etc.
• On a filesystem this corresponds to a directory of Python files with an optional init script.
• Running 'import <package>' does not automatically provide access to the package's modules unless they are explicitly imported in the '<package>/__init__.py' script.
• Directory of the file that is passed to python command serves as a root of local imports.
• For relative imports use 'from .[…][<pkg/module>[.…]] import <obj>'.

We have/get a closure in Python when a nested function references a value of its enclosing function and then the enclosing function returns its nested function.

def get_multiplier(a):
    def out(b):
        return a * b
    return out

>>> multiply_by_3 = get_multiplier(3)
>>> multiply_by_3(10)
30

• Any value that is referenced from within multiple nested functions gets shared.

from functools import partial
<function> = partial(<function> [, <arg_1> [, ...]])

>>> def multiply(a, b):
...     return a * b
>>> multiply_by_3 = partial(multiply, 3)
>>> multiply_by_3(10)
30

• Partial is also useful in cases when a function needs to be passed as an argument because it enables us to set its arguments beforehand ('collections.defaultdict(<func>)', 'iter(<func>, to_exc)' and 'dataclasses.field(default_factory=<func>)').

If variable is being assigned to anywhere in the scope, it is regarded as a local variable, unless it is declared as a 'global' or a 'nonlocal'.

def get_counter():
    i = 0
    def out():
        nonlocal i
        i += 1
        return i
    return out

>>> counter = get_counter()
>>> counter(), counter(), counter()
(1, 2, 3)

A decorator takes a function, adds some functionality and returns it. It can be any callable, but is usually implemented as a function that returns a closure.

@decorator_name
def function_that_gets_passed_to_decorator():
    ...

Decorator that prints function's name every time the function is called.

from functools import wraps

def debug(func):
    @wraps(func)
    def out(*args, **kwargs):
        print(func.__name__)
        return func(*args, **kwargs)
    return out

@debug
def add(x, y):
    return x + y

• Wraps is a helper decorator that copies the metadata of the passed function (func) to the function it is wrapping (out). Without it, 'add.__name__' would return 'out'.

Decorator that caches function's return values. All function's arguments must be hashable.

from functools import cache

@cache
def fib(n):
    return n if n < 2 else fib(n-2) + fib(n-1)

• Potential problem with cache is that it can grow indefinitely. To clear stored values run '<func>.cache_clear()', or use '@lru_cache(maxsize=<int>)' decorator instead.
• CPython interpreter limits recursion depth to 3000 by default. To increase it run 'sys.setrecursionlimit(<int>)'.

A decorator that accepts arguments and returns a normal decorator that accepts a function.

from functools import wraps

def debug(print_result=False):
    def decorator(func):
        @wraps(func)
        def out(*args, **kwargs):
            result = func(*args, **kwargs)
            print(func.__name__, result if print_result else '')
            return result
        return out
    return decorator

@debug(print_result=True)
def add(x, y):
    return x + y

• Using only '@debug' to decorate the add() function would not work here, because debug would then receive the add() function as a 'print_result' argument. Decorators can however manually check if the argument they received is a function and act accordingly.

A template for creating user-defined objects.

class MyClass:
    def __init__(self, a):
        self.a = a
    def __str__(self):
        return str(self.a)
    def __repr__(self):
        class_name = self.__class__.__name__
        return f'{class_name}({self.a!r})'

    @classmethod
    def get_class_name(cls):
        return cls.__name__

>>> obj = MyClass(1)
>>> obj.a, str(obj), repr(obj)
(1, '1', 'MyClass(1)')

• Methods whose names start and end with two underscores are called special methods. They are executed when object is passed to a built-in function or used as an operand, for example, 'print(a)' calls 'a.__str__()' and 'a + b' calls 'a.__add__(b)'.
• Methods decorated with '@staticmethod' receive neither 'self' nor 'cls' argument.
• Return value of str() special method should be readable and of repr() unambiguous. If only repr() is defined, it will also be used for str().

print(<obj>)
f'{<obj>}'
logging.warning(<obj>)
csv.writer(<file>).writerow([<obj>])

print/str/repr([<obj>])
print/str/repr({<obj>: <obj>})
f'{<obj>!r}'
Z = make_dataclass('Z', ['a']); print/str/repr(Z(<obj>))

• Inheritance is a mechanism that enables a class to extend some other class (that is, subclass to extend its parent), and by doing so inherit all of its methods and attributes.
• Subclass can then add its own methods and attributes or override inherited ones by reusing their names.

class Person:
    def __init__(self, name):
        self.name = name
    def __repr__(self):
        return f'Person({self.name!r})'
    def __lt__(self, other):
        return self.name < other.name

class Employee(Person):
    def __init__(self, name, staff_num):
        super().__init__(name)
        self.staff_num = staff_num
    def __repr__(self):
        return f'Employee({self.name!r}, {self.staff_num})'

>>> people = {Person('Ann'), Employee('Bob', 0)}
>>> sorted(people)
[Person('Ann'), Employee('Bob', 0)]

• They add type hints to variables, arguments and functions ('def f() -> <type>:').
• Hints are used by type checkers like mypy, data validation libraries such as Pydantic and lately also by Cython compiler. However, they are not enforced by CPython interpreter.

from collections import abc

<name>: <type> [| ...] [= <obj>]
<name>: list/set/abc.Iterable/abc.Sequence[<type>] [= <obj>]
<name>: tuple/dict[<type>, ...] [= <obj>]

Decorator that uses class variables to generate init(), repr() and eq() special methods.

from dataclasses import dataclass, field, make_dataclass

@dataclass(order=False, frozen=False)
class <class_name>:
    <attr_name>: <type>
    <attr_name>: <type> = <default_value>
    <attr_name>: list/dict/set = field(default_factory=list/dict/set)

• Objects can be made sortable with 'order=True' and immutable with 'frozen=True'.
• For object to be hashable, all attributes must be hashable and 'frozen' must be True.
• Function field() is needed because '<attr_name>: list = []' would make a list that is shared among all instances. Its 'default_factory' argument accepts any callable object.
• For attributes of arbitrary type use 'typing.Any'.

P = make_dataclass('P', ['x', 'y'])
P = make_dataclass('P', [('x', float), ('y', float)])
P = make_dataclass('P', [('x', float, 0), ('y', float, 0)])

Pythonic way of implementing getters and setters.

class Person:
    @property
    def name(self):
        return ' '.join(self._name)

    @name.setter
    def name(self, value):
        self._name = value.split()

>>> person = Person()
>>> person.name = '\t Guido  van Rossum \n'
>>> person.name
'Guido van Rossum'

Mechanism that restricts objects to attributes listed in 'slots'.

class MyClassWithSlots:
    __slots__ = ['a']
    def __init__(self):
        self.a = 1

from copy import copy, deepcopy
<object> = copy/deepcopy(<object>)

A duck type is an implicit type that prescribes a set of special methods. Any object that has those methods defined is considered a member of that duck type.

• If eq() method is not overridden, it returns 'id(self) == id(other)', which is the same as 'self is other'. That means all user-defined objects compare not equal by default (because id() returns object's memory address that is guaranteed to be unique).
• Only the left side object has eq() method called, unless it returns NotImplemented, in which case the right object is consulted. Result is False if both return NotImplemented.
• Ne() automatically works on any object that has eq() defined.

class MyComparable:
    def __init__(self, a):
        self.a = a
    def __eq__(self, other):
        if isinstance(other, type(self)):
            return self.a == other.a
        return NotImplemented

• Hashable object needs both hash() and eq() methods and its hash value must not change.
• Hashable objects that compare equal must have the same hash value, meaning default hash() that returns 'id(self)' will not do. That is why Python automatically makes classes unhashable if you only implement eq().

class MyHashable:
    def __init__(self, a):
        self._a = a
    @property
    def a(self):
        return self._a
    def __eq__(self, other):
        if isinstance(other, type(self)):
            return self.a == other.a
        return NotImplemented
    def __hash__(self):
        return hash(self.a)

• With 'total_ordering' decorator, you only need to provide eq() and one of lt(), gt(), le() or ge() special methods (used by <, >, <=, >=) and the rest will be automatically generated.
• Functions sorted() and min() only require lt() method, while max() only requires gt(). However, it is best to define them all so that confusion doesn't arise in other contexts.
• When two lists, strings or dataclasses are compared, their values get compared in order until a pair of unequal values is found. The comparison of this two values is then returned. The shorter sequence is considered smaller in case of all values being equal.
• To sort collection of strings in proper alphabetical order pass 'key=locale.strxfrm' to sorted() after running 'locale.setlocale(locale.LC_COLLATE, "en_US.UTF-8")'.

from functools import total_ordering

@total_ordering
class MySortable:
    def __init__(self, a):
        self.a = a
    def __eq__(self, other):
        if isinstance(other, type(self)):
            return self.a == other.a
        return NotImplemented
    def __lt__(self, other):
        if isinstance(other, type(self)):
            return self.a < other.a
        return NotImplemented

• Any object that has methods next() and iter() is an iterator.
• Next() should return next item or raise StopIteration exception.
• Iter() should return an iterator of remaining items, i.e. 'self'.
• Any object that has iter() method can be used in a for loop.

class Counter:
    def __init__(self):
        self.i = 0
    def __next__(self):
        self.i += 1
        return self.i
    def __iter__(self):
        return self

>>> counter = Counter()
>>> next(counter), next(counter), next(counter)
(1, 2, 3)

• Sequence iterators returned by the iter() function, such as list_iterator, etc.
• Objects returned by the itertools module, such as count, repeat and cycle.
• Generators returned by the generator functions and generator expressions.
• File objects returned by the open() function, etc.

• All functions and classes have a call() method that is executed when they are called.
• Use 'callable(<obj>)' or 'isinstance(<obj>, collections.abc.Callable)' to check if object is callable. You can also call the object and check if it raised TypeError.
• When this cheatsheet uses '<function>' as an argument, it means '<callable>'.

class Counter:
    def __init__(self):
        self.i = 0
    def __call__(self):
        self.i += 1
        return self.i

>>> counter = Counter()
>>> counter(), counter(), counter()
(1, 2, 3)

• With statements only work on objects that have enter() and exit() special methods.
• Enter() should lock the resources and optionally return an object (file, lock, etc.).
• Exit() should release the resources (for example close a file, release a lock, etc.).
• Any exception that happens inside the with block is passed to the exit() method.
• The exit() method can suppress the exception by returning a true value.

class MyOpen:
    def __init__(self, filename):
        self.filename = filename
    def __enter__(self):
        self.file = open(self.filename)
        return self.file
    def __exit__(self, exc_type, exception, traceback):
        self.file.close()

>>> with open('test.txt', 'w') as file:
...     file.write('Hello World!')
>>> with MyOpen('test.txt') as file:
...     print(file.read())
Hello World!

• Only required method is iter(). It should return an iterator of object's items.
• Contains() automatically works on any object that has iter() defined.

class MyIterable:
    def __init__(self, a):
        self.a = a
    def __iter__(self):
        return iter(self.a)
    def __contains__(self, el):
        return el in self.a

>>> obj = MyIterable([1, 2, 3])
>>> [el for el in obj]
[1, 2, 3]
>>> 1 in obj
True

• Only required methods are iter() and len(). Len() should return the number of items.
• This cheatsheet actually means '<iterable>' when it uses the '<collection>'.
• I chose not to use the name 'iterable' because it sounds scarier and more vague than 'collection'. The main drawback of this decision is that the reader could think a certain function doesn't accept iterators when it does, since iterators are the only built-in objects that are iterable but are not collections.

class MyCollection:
    def __init__(self, a):
        self.a = a
    def __iter__(self):
        return iter(self.a)
    def __contains__(self, el):
        return el in self.a
    def __len__(self):
        return len(self.a)

• Only required methods are getitem() and len(). Getitem() should return the item at the passed index or raise IndexError (it may also support negative indices and/or slices).
• Iter() and contains() automatically work on any object that has getitem() defined.
• Reversed() automatically works on any object that has getitem() and len() defined. It returns reversed iterator of object's items.

class MySequence:
    def __init__(self, a):
        self.a = a
    def __iter__(self):
        return iter(self.a)
    def __contains__(self, el):
        return el in self.a
    def __len__(self):
        return len(self.a)
    def __getitem__(self, i):
        return self.a[i]
    def __reversed__(self):
        return reversed(self.a)

• Python's glossary defines iterable as any object with special methods iter() and/or getitem() and sequence as any object with getitem() and len(). It doesn't define collection.
• Passing ABC Iterable to isinstance() or issubclass() only checks whether object/class has special method iter(), while ABC Collection checks for iter(), contains() and len().

• It's a richer interface than the basic sequence that also requires just getitem() and len().
• Extending it generates iter(), contains(), reversed(), index() and count() special methods.
• Unlike 'abc.Iterable' and 'abc.Collection', it is not a duck type. That is why 'issubclass(MySequence, abc.Sequence)' would return False even if MySequence had all the methods defined. It however recognizes list, tuple, range, str, bytes, bytearray, array, memoryview and deque, since they are registered as Sequence's virtual subclasses.

from collections import abc

class MyAbcSequence(abc.Sequence):
    def __init__(self, a):
        self.a = a
    def __len__(self):
        return len(self.a)
    def __getitem__(self, i):
        return self.a[i]

┏━━━━━━━━━━━━┯━━━━━━━━━━━━┯━━━━━━━━━━━━┯━━━━━━━━━━━━┯━━━━━━━━━━━━━━┓
┃            │  Iterable  │ Collection │  Sequence  │ abc.Sequence ┃
┠────────────┼────────────┼────────────┼────────────┼──────────────┨
┃ iter()     │     !      │     !      │     ✓      │      ✓       ┃
┃ contains() │     ✓      │     ✓      │     ✓      │      ✓       ┃
┃ len()      │            │     !      │     !      │      !       ┃
┃ getitem()  │            │            │     !      │      !       ┃
┃ reversed() │            │            │     ✓      │      ✓       ┃
┃ index()    │            │            │            │      ✓       ┃
┃ count()    │            │            │            │      ✓       ┃
┗━━━━━━━━━━━━┷━━━━━━━━━━━━┷━━━━━━━━━━━━┷━━━━━━━━━━━━┷━━━━━━━━━━━━━━┛

• Method iter() is required for 'isinstance(<obj>, abc.Iterable)' to return True, however any object with getitem() will work with any code expecting an iterable.
• MutableSequence, Set, MutableSet, Mapping and MutableMapping ABCs are also extendable. Use '<abc>.__abstractmethods__' to get names of required methods.

Class of named constants called members.

from enum import Enum, auto

class <enum_name>(Enum):
    <member_name> = auto()              
    <member_name> = <value>             
    <member_name> = <el_1>, <el_2>      

• Methods receive the member they were called on as the 'self' argument.
• Accessing a member named after a reserved keyword causes SyntaxError.

<member> = <enum>.<member_name>         
<member> = <enum>['<member_name>']      
<member> = <enum>(<value>)              
<str>    = <member>.name                
<obj>    = <member>.value               

<list>   = list(<enum>)                 
<list>   = <enum>._member_names_        
<list>   = [m.value for m in <enum>]    

<enum>   = type(<member>)               
<iter>   = itertools.cycle(<enum>)      
<member> = random.choice(list(<enum>))  

Cutlery = Enum('Cutlery', 'FORK KNIFE SPOON')
Cutlery = Enum('Cutlery', ['FORK', 'KNIFE', 'SPOON'])
Cutlery = Enum('Cutlery', {'FORK': 1, 'KNIFE': 2, 'SPOON': 3})

from functools import partial
LogicOp = Enum('LogicOp', {'AND': partial(lambda l, r: l and r),
                           'OR':  partial(lambda l, r: l or r)})

try:
    <code>
except <exception>:
    <code>

try:
    <code_1>
except <exception_a>:
    <code_2_a>
except <exception_b>:
    <code_2_b>
else:
    <code_2_c>
finally:
    <code_3>

• Code inside the 'else' block will only be executed if 'try' block had no exceptions.
• Code inside the 'finally' block will always be executed (unless a signal is received).
• All variables that are initialized in executed blocks are also visible in all subsequent blocks, as well as outside the try statement (only the function block delimits scope).
• To catch signals use 'signal.signal(signal_number, handler_function)'.

except <exception>: ...
except <exception> as <name>: ...
except (<exception>, [...]): ...
except (<exception>, [...]) as <name>: ...

• Also catches subclasses, e.g. 'IndexError' is caught by 'except LookupError:'.
• Use 'traceback.print_exc()' to print the full error message to standard error stream.
• Use 'print(<name>)' to print just the cause of the exception (that is, its arguments).
• Use 'logging.exception(<str>)' to log the passed message, followed by the full error message of the caught exception. For details about setting up the logger see Logging.
• Use 'sys.exc_info()' to get exception type, object, and traceback of caught exception.

raise <exception>
raise <exception>()
raise <exception>(<obj> [, ...])

except <exception> [as <name>]:
    ...
    raise

arguments = <name>.args
exc_type  = <name>.__class__
filename  = <name>.__traceback__.tb_frame.f_code.co_filename
func_name = <name>.__traceback__.tb_frame.f_code.co_name
line_str  = linecache.getline(filename, <name>.__traceback__.tb_lineno)
trace_str = ''.join(traceback.format_tb(<name>.__traceback__))
error_msg = ''.join(traceback.format_exception(type(<name>), <name>, <name>.__traceback__))

BaseException
 ├── SystemExit                   
 ├── KeyboardInterrupt            
 └── Exception                    
      ├── ArithmeticError         
      ├── AssertionError          
      ├── AttributeError          
      ├── EOFError                
      ├── LookupError             
      │    ├── IndexError         
      │    └── KeyError           
      ├── MemoryError             
      ├── NameError               
      │    └── UnboundLocalError  
      ├── OSError                 
      │    └── ConnectionError    
      ├── RuntimeError            
      │    ├── NotImplementedEr…  
      │    └── RecursionError     
      ├── StopIteration           
      ├── TypeError               
      └── ValueError              

┏━━━━━━━━━━━┯━━━━━━━━━━━━┯━━━━━━━━━━━━┯━━━━━━━━━━━━┓
┃           │    List    │    Set     │    Dict    ┃
┠───────────┼────────────┼────────────┼────────────┨
┃ getitem() │ IndexError │            │  KeyError  ┃
┃ pop()     │ IndexError │  KeyError  │  KeyError  ┃
┃ remove()  │ ValueError │  KeyError  │            ┃
┃ index()   │ ValueError │            │            ┃
┗━━━━━━━━━━━┷━━━━━━━━━━━━┷━━━━━━━━━━━━┷━━━━━━━━━━━━┛

raise TypeError('Argument is of the wrong type!')
raise ValueError('Argument has the right type but an inappropriate value!')
raise RuntimeError('I am too lazy to define my own exception!')

class MyError(Exception): pass
class MyInputError(MyError): pass

Exits the interpreter by raising SystemExit exception.

import sys
sys.exit()                        
sys.exit(<int>)                   
sys.exit(<obj>)                   

print(<el_1>, ..., sep=' ', end='\n', file=sys.stdout, flush=False)

• Use 'file=sys.stderr' for messages about errors so they can be processed separately.
• Stdout and stderr streams hold output in a buffer until they receive a string containing '\n' or '\r', buffer reaches 4096 characters, 'flush=True' is used, or program exits.

from pprint import pprint
pprint(<collection>, width=80, depth=None, compact=False, sort_dicts=True)

• Each item is printed on its own line if collection exceeds 'width' characters.
• Nested collections that are 'depth' levels deep get printed as '…'.

• Reads a line from the user input or pipe if present (trailing newline gets stripped).
• If argument is passed, it gets printed to the standard output before input is read.
• EOFError is raised if user hits EOF (ctrl-d/ctrl-z⏎) or if stream is already exhausted.

import sys
scripts_path = sys.argv[0]
arguments    = sys.argv[1:]

from argparse import ArgumentParser, FileType
p = ArgumentParser(description=<str>)                             
p.add_argument('-<short_name>', '--<name>', action='store_true')  
p.add_argument('-<short_name>', '--<name>', type=<type>)          
p.add_argument('<name>', type=<type>, nargs=1)                    
p.add_argument('<name>', type=<type>, nargs='+')                  
p.add_argument('<name>', type=<type>, nargs='?/*')                
args  = p.parse_args()                                            
<obj> = args.<name>                                               

• Use 'help=<str>' to set argument description that will be displayed in help message.
• Use 'default=<obj>' to override None as option's or optional argument's default value.
• Use 'type=FileType(<mode>)' for files. It accepts 'encoding', but 'newline' is None.

Opens a file and returns the corresponding file object.

<file> = open(<path>, mode='r', encoding=None, newline=None)

• 'encoding=None' means that the default encoding is used, which is platform dependent. Best practice is to use 'encoding="utf-8"' until it becomes the default (Python 3.15).
• 'newline=None' means all different end of line combinations are converted to '\n' on read, while on write all '\n' characters are converted to system's default line separator.
• 'newline=""' means no conversions take place, but input is still broken into chunks by readline() and readlines() on every '\n', '\r' and '\r\n'.

• 'r' - Read. Used by default.
• 'w' - Write. Deletes existing contents.
• 'x' - Write or fail if the file already exists.
• 'a' - Append. Creates new file if it doesn't exist.
• 'w+' - Read and write. Deletes existing contents.
• 'r+' - Read and write from the start.
• 'a+' - Read and write from the end.
• 'b' - Binary mode ('rb', 'wb', 'xb', …).

• 'FileNotFoundError' can be raised when reading with 'r' or 'r+'.
• 'FileExistsError' can be raised when writing with 'x'.
• 'IsADirectoryError' and 'PermissionError' can be raised by any.
• 'OSError' is the parent class of all listed exceptions.

<file>.seek(0)                      
<file>.seek(offset)                 
<file>.seek(0, 2)                   
<bin_file>.seek(±offset, origin)    

<str/bytes> = <file>.read(size=-1)  
<str/bytes> = <file>.readline()     
<list>      = <file>.readlines()    
<str/bytes> = next(<file>)          

<file>.write(<str/bytes>)           
<file>.writelines(<collection>)     
<file>.flush()                      
<file>.close()                      

• Methods do not add or strip trailing newlines, not even writelines().

def read_file(filename):
    with open(filename, encoding='utf-8') as file:
        return file.readlines()

def write_to_file(filename, text):
    with open(filename, 'w', encoding='utf-8') as file:
        file.write(text)

import os, glob
from pathlib import Path

<str>  = os.getcwd()                
<str>  = os.path.join(<path>, ...)  
<str>  = os.path.realpath(<path>)   

<str>  = os.path.basename(<path>)   
<str>  = os.path.dirname(<path>)    
<tup.> = os.path.splitext(<path>)   

<list> = os.listdir(path='.')       
<list> = glob.glob('<pattern>')     

<bool> = os.path.exists(<path>)     
<bool> = os.path.isfile(<path>)     
<bool> = os.path.isdir(<path>)      

<stat> = os.stat(<path>)            
<num>  = <stat>.st_mtime/st_size/…  

Unlike listdir(), scandir() returns DirEntry objects that cache isfile, isdir, and on Windows also stat information, thus significantly increasing the performance of code that requires it.

<iter> = os.scandir(path='.')       
<str>  = <DirEntry>.path            
<str>  = <DirEntry>.name            
<file> = open(<DirEntry>)           

<Path> = Path(<path> [, ...])       
<Path> = <path> / <path> [/ ...]    
<Path> = <Path>.resolve()           

<Path> = Path()                     
<Path> = Path.cwd()                 
<Path> = Path.home()                
<Path> = Path(__file__).resolve()   

<Path> = <Path>.parent              
<str>  = <Path>.name                
<str>  = <Path>.suffix              
<str>  = <Path>.stem                
<tup.> = <Path>.parts               

<iter> = <Path>.iterdir()           
<iter> = <Path>.glob('<pattern>')   

<str>  = str(<Path>)                
<file> = open(<Path>)               

import os, shutil, subprocess

os.chdir(<path>)                    
os.mkdir(<path>, mode=0o777)        
os.makedirs(<path>, mode=0o777)     

shutil.copy(from, to)               
shutil.copy2(from, to)              
shutil.copytree(from, to)           

os.rename(from, to)                 
os.replace(from, to)                
shutil.move(from, to)               

os.remove(<path>)                   
os.rmdir(<path>)                    
shutil.rmtree(<path>)               

• Paths can be either strings, Path objects, or DirEntry objects.
• Functions report OS related errors by raising OSError or one of its subclasses.

<pipe> = os.popen('<commands>')     
<str>  = <pipe>.read(size=-1)       
<int>  = <pipe>.close()             

>>> subprocess.run('bc', input='1 + 1\n', capture_output=True, text=True)
CompletedProcess(args='bc', returncode=0, stdout='2\n', stderr='')

>>> from shlex import split
>>> os.popen('echo 1 + 1 > test.in')
>>> subprocess.run(split('bc -s'), stdin=open('test.in'), stdout=open('test.out', 'w'))
CompletedProcess(args=['bc', '-s'], returncode=0)
>>> open('test.out').read()
'2\n'

Text file format for storing collections of strings and numbers.

import json
<str>  = json.dumps(<list/dict>)    
<coll> = json.loads(<str>)          

def read_json_file(filename):
    with open(filename, encoding='utf-8') as file:
        return json.load(file)

def write_to_json_file(filename, collection):
    with open(filename, 'w', encoding='utf-8') as file:
        json.dump(collection, file, ensure_ascii=False, indent=2)

Binary file format for storing Python objects.

import pickle
<bytes>  = pickle.dumps(<object>)   
<object> = pickle.loads(<bytes>)    

def read_pickle_file(filename):
    with open(filename, 'rb') as file:
        return pickle.load(file)

def write_to_pickle_file(filename, an_object):
    with open(filename, 'wb') as file:
        pickle.dump(an_object, file)

Text file format for storing spreadsheets.

import csv

<file>   = open(<path>, newline='')       
<reader> = csv.reader(<file>)             
<list>   = next(<reader>)                 
<list>   = list(<reader>)                 

• Without the 'newline=""' argument, every '\r\n' sequence that is embedded inside a quoted field will get converted to '\n'! For details about newline argument see Open.
• To print the spreadsheet to the console use Tabulate or PrettyTable library.
• For XML and binary Excel files (xlsx, xlsm and xlsb) use Pandas library.
• Reader accepts any iterator (or collection) of strings, not just files.

<file>   = open(<path>, 'w', newline='')  
<writer> = csv.writer(<file>)             
<writer>.writerow(<collection>)           
<writer>.writerows(<coll_of_coll>)        

• If file is opened without the 'newline=""' argument, '\r' will be added in front of every '\n' on platforms that use '\r\n' line endings (i.e., newlines may get doubled on Windows)!
• Open existing file with 'mode="a"' to append to it or 'mode="w"' to overwrite it.

• 'dialect' - Master parameter that sets the default values. String or a 'csv.Dialect' object.
• 'delimiter' - A one-character string that separates fields (comma, tab, semicolon, etc.).
• 'lineterminator' - How writer terminates rows. Reader looks for '\n', '\r' and '\r\n'.
• 'quotechar' - Character for quoting fields containing delimiters, quotechars, '\n' or '\r'.
• 'escapechar' - Character for escaping quotechars (not needed if doublequote is True).
• 'doublequote' - Whether quotechars inside fields are/get doubled instead of escaped.
• 'quoting' - 0: As necessary, 1: All, 2: All but numbers which are read as floats, 3: None.
• 'skipinitialspace' - Is space character at the start of the field stripped by the reader.

┏━━━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━┓
┃                  │     excel    │   excel-tab  │     unix     ┃
┠──────────────────┼──────────────┼──────────────┼──────────────┨
┃ delimiter        │       ','    │      '\t'    │       ','    ┃
┃ lineterminator   │    '\r\n'    │    '\r\n'    │      '\n'    ┃
┃ quotechar        │       '"'    │       '"'    │       '"'    ┃
┃ escapechar       │      None    │      None    │      None    ┃
┃ doublequote      │      True    │      True    │      True    ┃
┃ quoting          │         0    │         0    │         1    ┃
┃ skipinitialspace │     False    │     False    │     False    ┃
┗━━━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━┛

def read_csv_file(filename, **csv_params):
    with open(filename, encoding='utf-8', newline='') as file:
        return list(csv.reader(file, **csv_params))

def write_to_csv_file(filename, rows, mode='w', **csv_params):
    with open(filename, mode, encoding='utf-8', newline='') as file:
        writer = csv.writer(file, **csv_params)
        writer.writerows(rows)

A server-less database engine that stores each database into its own file.

import sqlite3
<conn> = sqlite3.connect(<path>)               
<conn>.close()                                 

<cursor> = <conn>.execute('<query>')           
<tuple>  = <cursor>.fetchone()                 
<list>   = <cursor>.fetchall()                 

<conn>.execute('<query>')                      
<conn>.commit()                                
<conn>.rollback()                              

with <conn>:                                   
    <conn>.execute('<query>')                  

<conn>.execute('<query>', <list/tuple>)        
<conn>.execute('<query>', <dict/namedtuple>)   
<conn>.executemany('<query>', <coll_of_coll>)  

• Passed values can be of type str, int, float, bytes, None, or bool (stored as 1 or 0).
• SQLite does not restrict columns to any type unless table is declared as strict.

Values are not actually saved in this example because 'conn.commit()' is omitted!

>>> conn = sqlite3.connect('test.db')
>>> conn.execute('CREATE TABLE person (name TEXT, height INTEGER) STRICT')
>>> conn.execute('INSERT INTO person VALUES (?, ?)', ('Jean-Luc', 187))
>>> conn.execute('SELECT rowid, * FROM person').fetchall()
[(1, 'Jean-Luc', 187)]

Library for interacting with various DB systems via SQL, method chaining, or ORM.

from sqlalchemy import create_engine, text
<engine> = create_engine('<url>')              
<conn>   = <engine>.connect()                  
<cursor> = <conn>.execute(text('<query>'), …)  
with <conn>.begin(): ...                       

┏━━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃ Dialect         │ pip3 install │           Dependencies           ┃
┠─────────────────┼──────────────┼──────────────────────────────────┨
┃ mysql           │ mysqlclient  │ www.pypi.org/project/mysqlclient ┃
┃ postgresql      │ psycopg2     │ www.pypi.org/project/psycopg2    ┃
┃ mssql           │ pyodbc       │ www.pypi.org/project/pyodbc      ┃
┃ oracle+oracledb │ oracledb     │ www.pypi.org/project/oracledb    ┃
┗━━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┛

A bytes object is an immutable sequence of single bytes. Mutable version is called bytearray.

<bytes> = b'<str>'                       
<int>   = <bytes>[index]                 
<bytes> = <bytes>[<slice>]               
<bytes> = <bytes>.join(<coll_of_bytes>)  

<bytes> = bytes(<coll_of_ints>)          
<bytes> = bytes(<str>, 'utf-8')          
<bytes> = bytes.fromhex('<hex>')         
<bytes> = <int>.to_bytes(n_bytes, …)     

<list>  = list(<bytes>)                  
<str>   = str(<bytes>, 'utf-8')          
<str>   = <bytes>.hex()                  
<int>   = int.from_bytes(<bytes>, …)     

def read_bytes(filename):
    with open(filename, 'rb') as file:
        return file.read()

def write_bytes(filename, bytes_obj):
    with open(filename, 'wb') as file:
        file.write(bytes_obj)

• Module that performs conversions between a sequence of numbers and a bytes object.
• System’s type sizes, byte order, and alignment rules are used by default.

from struct import pack, unpack

<bytes> = pack('<format>', <el_1> [, ...])  
<tuple> = unpack('<format>', <bytes>)       

>>> pack('>hhl', 1, 2, 3)
b'\x00\x01\x00\x02\x00\x00\x00\x03'
>>> unpack('>hhl', b'\x00\x01\x00\x02\x00\x00\x00\x03')
(1, 2, 3)

• '=' - System's byte order (usually little-endian).
• '<' - Little-endian (i.e. least significant byte first).
• '>' - Big-endian (also '!').

• 'c' - A bytes object with a single element. For pad byte use 'x'.
• '<n>s' - A bytes object with n elements (not effected by byte order).

• 'b' - char (1/1)
• 'h' - short (2/2)
• 'i' - int (2/4)
• 'l' - long (4/4)
• 'q' - long long (8/8)

• 'f' - float (4/4)
• 'd' - double (8/8)

List that can only hold numbers of a predefined type. Available types and their minimum sizes in bytes are listed above. Type sizes and byte order are always determined by the system, however bytes of each element can be reversed (by calling the byteswap() method).

from array import array

<array> = array('<typecode>', <coll_of_nums>)  
<array> = array('<typecode>', <bytes>)         
<array> = array('<typecode>', <array>)         
<array>.fromfile(<file>, n_items)              

<bytes> = bytes(<array>)                       
<file>.write(<array>)                          

A sequence object that points to the memory of another bytes-like object. Each element can reference a single or multiple consecutive bytes, depending on format. Order and number of elements can be changed with slicing.

<mview> = memoryview(<bytes/bytearray/array>)  
<obj>   = <mview>[index]                       
<mview> = <mview>[<slice>]                     
<mview> = <mview>.cast('<typecode>')           
<mview>.release()                              

<bytes> = bytes(<mview>)                       
<bytes> = <bytes>.join(<coll_of_mviews>)       
<array> = array('<typecode>', <mview>)         
<file>.write(<mview>)                          

<list>  = list(<mview>)                        
<str>   = str(<mview>, 'utf-8')                
<str>   = <mview>.hex()                        

List with efficient appends and pops from either side.

from collections import deque

<deque> = deque(<collection>)                  
<deque>.appendleft(<el>)                       
<deque>.extendleft(<collection>)               
<deque>.rotate(n=1)                            
<el> = <deque>.popleft()                       

Module of functions that provide the functionality of operators. Functions are grouped by operator precedence, from least to most binding. Functions and operators in first, third and fifth line are also ordered by precedence within a group.

import operator as op

<bool> = op.not_(<obj>)                                        
<bool> = op.eq/ne/lt/ge/is_/is_not/contains(<obj>, <obj>)      
<obj>  = op.or_/xor/and_(<int/set>, <int/set>)                 
<int>  = op.lshift/rshift(<int>, <int>)                        
<obj>  = op.add/sub/mul/truediv/floordiv/mod(<obj>, <obj>)     
<num>  = op.neg/invert(<num>)                                  
<num>  = op.pow(<num>, <num>)                                  
<func> = op.itemgetter/attrgetter/methodcaller(<obj> [, ...])  

elementwise_sum  = map(op.add, list_a, list_b)
sorted_by_second = sorted(<coll>, key=op.itemgetter(1))
sorted_by_both   = sorted(<coll>, key=op.itemgetter(1, 0))
first_element    = op.methodcaller('pop', 0)(<list>)

• Most operators call the object's special method that is named after them (second object is passed as an argument), while logical operators call their own code that relies on bool().
• Comparisons can be chained: 'x < y < z' gets converted to '(x < y) and (y < z)'.

Executes the first block with matching pattern.

match <object/expression>:
    case <pattern> [if <condition>]:
        <code>
    ...

<value_pattern> = 1/'abc'/True/None/math.pi        
<class_pattern> = <type>()                         
<wildcard_patt> = _                                
<capture_patt>  = <name>                           
<as_pattern>    = <pattern> as <name>              
<or_pattern>    = <pattern> | <pattern> [| ...]    
<sequence_patt> = [<pattern>, ...]                 
<mapping_patt>  = {<value_pattern>: <patt>, ...}   
<class_pattern> = <type>(<attr_name>=<patt>, ...)  

• Sequence pattern can also be written as a tuple, either with or without the brackets.
• Use '*<name>' and '**<name>' in sequence/mapping patterns to bind remaining items.
• Sequence pattern must match all items of the collection, while mapping pattern does not.
• Patterns can be surrounded with brackets to override their precedence: '|' > 'as' > ','. For example, '[1, 2]' is matched by the 'case 1|2, 2|3 as x if x == 2:' block.
• All names that are bound in the matching case, as well as variables initialized in its body, are visible after the match statement (only function block delimits scope).

>>> from pathlib import Path
>>> match Path('/home/gto/python-cheatsheet/README.md'):
...     case Path(
...         parts=['/', 'home', user, *_]
...     ) as p if p.name.lower().startswith('readme') and p.is_file():
...         print(f'{p.name} is a readme file that belongs to user {user}.')
README.md is a readme file that belongs to user gto.

import logging as log

log.basicConfig(filename=<path>, level='DEBUG')   
log.debug/info/warning/error/critical(<str>)      
<Logger> = log.getLogger(__name__)                
<Logger>.<level>(<str>)                           
<Logger>.exception(<str>)                         

log.basicConfig(
    filename=None,                                
    format='%(levelname)s:%(name)s:%(message)s',  
    level=log.WARNING,                            
    handlers=[log.StreamHandler(sys.stderr)]      
)

<Formatter> = log.Formatter('<format>')           
<Handler> = log.FileHandler(<path>, mode='a')     
<Handler>.setFormatter(<Formatter>)               
<Handler>.setLevel(<int/str>)                     
<Logger>.addHandler(<Handler>)                    
<Logger>.setLevel(<int/str>)                      
<Logger>.propagate = <bool>                       

• Parent logger can be specified by naming the child logger '<parent>.<name>'.
• If logger doesn't have a set level, it inherits it from the first ancestor that does.
• Formatter also accepts: pathname, filename, funcName, lineno, thread and process.
• RotatingFileHandler creates and deletes files based on 'maxBytes', 'backupCount' args.
• An object with 'filter(<LogRecord>)' method (or the method itself) can be added to loggers and handlers via addFilter(). Message is dropped if filter() returns a false value.

>>> logger = log.getLogger('my_module')
>>> handler = log.FileHandler('test.log', encoding='utf-8')
>>> handler.setFormatter(log.Formatter('%(asctime)s %(levelname)s:%(name)s:%(message)s'))
>>> logger.addHandler(handler)
>>> logger.setLevel('DEBUG')
>>> log.basicConfig()
>>> log.root.handlers[0].setLevel('WARNING')
>>> logger.critical('Running out of disk space.')
CRITICAL:my_module:Running out of disk space.
>>> print(open('test.log').read())
2023-02-07 23:21:01,430 CRITICAL:my_module:Running out of disk space.

<list> = dir()                      
<dict> = vars()                     
<dict> = globals()                  

<list> = dir(<obj>)                 
<dict> = vars(<obj>)                
<bool> = hasattr(<obj>, '<name>')   
value  = getattr(<obj>, '<name>')   
setattr(<obj>, '<name>', value)     
delattr(<obj>, '<name>')            

<Sig>  = inspect.signature(<func>)  
<dict> = <Sig>.parameters           
<memb> = <Param>.kind               
<type> = <Param>.annotation         

CPython interpreter can only run a single thread at a time. Using multiple threads won't result in a faster execution, unless at least one of the threads contains an I/O operation.

from threading import Thread, Lock, RLock, Semaphore, Event, Barrier
from concurrent.futures import ThreadPoolExecutor, as_completed

<Thread> = Thread(target=<function>)           
<Thread>.start()                               
<Thread>.join()                                

• Use 'kwargs=<dict>' to pass keyword arguments to the function.
• Use 'daemon=True', or the program won't be able to exit while the thread is alive.

<lock> = Lock/RLock()                          
<lock>.acquire()                               
<lock>.release()                               

<Semaphore> = Semaphore(value=1)               
<Event>     = Event()                          
<Barrier>   = Barrier(n_times)                 

<Queue> = queue.Queue(maxsize=0)               
<Queue>.put(<el>)                              
<Queue>.put_nowait(<el>)                       
<el> = <Queue>.get()                           
<el> = <Queue>.get_nowait()                    

<Exec> = ThreadPoolExecutor(max_workers=None)  
<iter> = <Exec>.map(<func>, <args_1>, ...)     
<Futr> = <Exec>.submit(<func>, <arg_1>, ...)   
<Exec>.shutdown()                              

<bool> = <Future>.done()                       
<obj>  = <Future>.result(timeout=None)         
<bool> = <Future>.cancel()                     
<iter> = as_completed(<coll_of_Futures>)       

• Map() and as_completed() also accept 'timeout'. It causes futures.TimeoutError when next() is called/blocking. Map() times from original call and as_completed() from first call to next(). As_completed() fails if next() is called too late, even if all threads are done.
• Exceptions that happen inside threads are raised when map iterator's next() or Future's result() are called. Future's exception() method returns an exception object or None.
• ProcessPoolExecutor provides true parallelism but: everything sent to/from workers must be pickable, queues must be sent using executor's 'initargs' and 'initializer' parameters, and executor should only be reachable via 'if __name__ == "__main__": ...'.

• Coroutines have a lot in common with threads, but unlike threads, they only give up control when they call another coroutine and they don’t use as much memory.
• Coroutine definition starts with 'async' and its call with 'await'.
• Use 'asyncio.run(<coroutine>)' to start the first/main coroutine.

import asyncio as aio

<coro> = <async_function>(<args>)          
<obj>  = await <coroutine>                 
<task> = aio.create_task(<coroutine>)      
<obj>  = await <task>                      

<coro> = aio.gather(<coro/task>, ...)      
<coro> = aio.wait(<tasks>, return_when=…)  
<iter> = aio.as_completed(<coros/tasks>)   

import asyncio, collections, curses, curses.textpad, enum, random

P = collections.namedtuple('P', 'x y')     
D = enum.Enum('D', 'n e s w')              
W, H = 15, 7                               

def main(screen):
    curses.curs_set(0)                     
    screen.nodelay(True)                   
    asyncio.run(main_coroutine(screen))    

async def main_coroutine(screen):
    moves = asyncio.Queue()
    state = {'*': P(0, 0)} | {id_: P(W//2, H//2) for id_ in range(10)}
    ai    = [random_controller(id_, moves) for id_ in range(10)]
    mvc   = [human_controller(screen, moves), model(moves, state), view(state, screen)]
    tasks = [asyncio.create_task(coro) for coro in ai + mvc]
    await asyncio.wait(tasks, return_when=asyncio.FIRST_COMPLETED)

async def random_controller(id_, moves):
    while True:
        d = random.choice(list(D))
        moves.put_nowait((id_, d))
        await asyncio.sleep(random.triangular(0.01, 0.65))

async def human_controller(screen, moves):
    while True:
        key_mappings = {258: D.s, 259: D.n, 260: D.w, 261: D.e}
        if d := key_mappings.get(screen.getch()):
            moves.put_nowait(('*', d))
        await asyncio.sleep(0.005)

async def model(moves, state):
    while state['*'] not in (state[id_] for id_ in range(10)):
        id_, d = await moves.get()
        deltas = {D.n: P(0, -1), D.e: P(1, 0), D.s: P(0, 1), D.w: P(-1, 0)}
        state[id_] = P((state[id_].x + deltas[d].x) % W, (state[id_].y + deltas[d].y) % H)

async def view(state, screen):
    offset = P(curses.COLS//2 - W//2, curses.LINES//2 - H//2)
    while True:
        screen.erase()
        curses.textpad.rectangle(screen, offset.y-1, offset.x-1, offset.y+H, offset.x+W)
        for id_, p in state.items():
            screen.addstr(offset.y + (p.y - state['*'].y + H//2) % H,
                          offset.x + (p.x - state['*'].x + W//2) % W, str(id_))
        screen.refresh()
        await asyncio.sleep(0.005)

if __name__ == '__main__':
    curses.wrapper(main)

>>> import tqdm, time
>>> for el in tqdm.tqdm([1, 2, 3], desc='Processing'):
...     time.sleep(1)
Processing: 100%|████████████████████| 3/3 [00:03<00:00,  1.00s/it]

import matplotlib.pyplot as plt

plt.plot/bar/scatter(x_data, y_data [, label=<str>])  
plt.legend()                                          
plt.title/xlabel/ylabel(<str>)                        
plt.show()                                            
plt.clf()                                             

import csv, tabulate
with open('test.csv', encoding='utf-8', newline='') as file:
    rows = list(csv.reader(file))
print(tabulate.tabulate(rows, headers='firstrow'))

import curses, os
from curses import A_REVERSE, KEY_UP, KEY_DOWN, KEY_LEFT, KEY_RIGHT

def main(screen):
    ch, first, selected, paths = 0, 0, 0, os.listdir()
    while ch != ord('q'):
        height, width = screen.getmaxyx()
        screen.erase()
        for y, filename in enumerate(paths[first : first+height]):
            color = A_REVERSE if filename == paths[selected] else 0
            screen.addnstr(y, 0, filename, width-1, color)
        ch = screen.getch()
        selected -= (ch == KEY_UP) and (selected > 0)
        selected += (ch == KEY_DOWN) and (selected < len(paths)-1)
        first -= (first > selected)
        first += (first < selected-(height-1))
        if ch in [KEY_LEFT, KEY_RIGHT, ord('\n')]:
            new_dir = '..' if ch == KEY_LEFT else paths[selected]
            if os.path.isdir(new_dir):
                os.chdir(new_dir)
                first, selected, paths = 0, 0, os.listdir()

if __name__ == '__main__':
    curses.wrapper(main)

import PySimpleGUI as sg

text_box = sg.Input(default_text='100', enable_events=True, key='QUANTITY')
dropdown = sg.InputCombo(['g', 'kg', 't'], 'kg', readonly=True, enable_events=True, k='UNIT')
label    = sg.Text('100 kg is 220.462 lbs.', key='OUTPUT')
button   = sg.Button('Close')
window   = sg.Window('Weight Converter', [[text_box, dropdown], [label], [button]])

while True:
    event, values = window.read()
    if event in [sg.WIN_CLOSED, 'Close']:
        break
    try:
        quantity = float(values['QUANTITY'])
    except ValueError:
        continue
    unit = values['UNIT']
    factors = {'g': 0.001, 'kg': 1, 't': 1000}
    lbs = quantity * factors[unit] / 0.45359237
    window['OUTPUT'].update(value=f'{quantity} {unit} is {lbs:g} lbs.')
window.close()

import requests, bs4, os

response   = requests.get('https://en.wikipedia.org/wiki/Python_(programming_language)')
document   = bs4.BeautifulSoup(response.text, 'html.parser')
table      = document.find('table', class_='infobox vevent')
python_url = table.find('th', text='Website').next_sibling.a['href']
logo_url   = table.find('img')['src']
filename   = os.path.basename(logo_url)
with open(filename, 'wb') as file:
    file.write(requests.get(f'https:{logo_url}').content)
print(f'{python_url}, file://{os.path.abspath(filename)}')

Library for scraping websites with dynamic content.

from selenium import webdriver

<WebDrv> = webdriver.Chrome/Firefox/Safari/Edge()     
<WebDrv>.get('<url>')                                 
<str>  = <WebDrv>.page_source                         
<El>   = <WebDrv/El>.find_element('css selector', …)  
<list> = <WebDrv/El>.find_elements('xpath', …)        
<str>  = <El>.get_attribute(<str>)                    
<El>.click/clear()                                    

<xpath>     = //<element>[/ or // <element>]          
<xpath>     = //<element>/following::<element>        
<element>   = <tag><conditions><index>                
<condition> = [<sub_cond> [and/or <sub_cond>]]        
<sub_cond>  = @<attr>[="<val>"]                       
<sub_cond>  = contains(@<attr>, "<val>")              
<sub_cond>  = [//]<element>                           

Flask is a micro web framework/server. If you just want to open a html file in a web browser use 'webbrowser.open(<path>)' instead.

import flask as fl

app = fl.Flask(__name__)                   
app.run(host=None, port=None, debug=None)  

• Starts the app at 'http://localhost:5000'. Use 'host="0.0.0.0"' to run externally.
• Install a WSGI server like Waitress and a HTTP server such as Nginx for better security.
• Debug mode restarts the app whenever script changes and displays errors in the browser.

@app.route('/img/<path:filename>')
def serve_file(filename):
    return fl.send_from_directory('DIRNAME', filename)

@app.route('/<sport>')
def serve_html(sport):
    return fl.render_template_string('<h1>{{title}}</h1>', title=sport)

• 'fl.render_template(filename, <kwargs>)' renders a file located in 'templates' dir.
• 'fl.abort(<int>)' returns error code and 'return fl.redirect(<url>)' redirects.
• 'fl.request.args[<str>]' returns parameter from query string (URL part right of '?').
• 'fl.session[<str>] = <obj>' stores session data. It requires secret key to be set at the startup with 'app.secret_key = <str>'.

@app.post('/<sport>/odds')
def serve_json(sport):
    team = fl.request.form['team']
    return {'team': team, 'odds': [2.09, 3.74, 3.68]}

>>> import threading, requests
>>> threading.Thread(target=app.run, daemon=True).start()
>>> url = 'http://localhost:5000/football/odds'
>>> response = requests.post(url, data={'team': 'arsenal f.c.'})
>>> response.json()
{'team': 'arsenal f.c.', 'odds': [2.09, 3.74, 3.68]}

from time import perf_counter
start_time = perf_counter()
...
duration_in_seconds = perf_counter() - start_time

>>> from timeit import timeit
>>> timeit('list(range(10000))', number=1000, globals=globals(), setup='pass')
0.19373

$ pip3 install line_profiler
$ echo '@profile
def main():
    a = list(range(10000))
    b = set(range(10000))
main()' > test.py
$ kernprof -lv test.py
Line #      Hits         Time  Per Hit   % Time  Line Contents
==============================================================
     1                                           @profile
     2                                           def main():
     3         1        253.4    253.4     32.2      a = list(range(10000))
     4         1        534.1    534.1     67.8      b = set(range(10000))

$ apt/brew install graphviz && pip3 install gprof2dot snakeviz  
$ tail --lines=+2 test.py > test.py                             
$ python3 -m cProfile -o test.prof test.py                      
$ gprof2dot --format=pstats test.prof | dot -T png -o test.png  
$ xdg-open/open test.png                                        
$ snakeviz test.prof                                            

┏━━━━━━━━━━━━━━┯━━━━━━━━━━━━┯━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┯━━━━━━━┯━━━━━━┓
┃ pip3 install │   Target   │          How to run           │ Lines │ Live ┃
┠──────────────┼────────────┼───────────────────────────────┼───────┼──────┨
┃ pyinstrument │    CPU     │ pyinstrument test.py          │   ×   │  ×   ┃
┃ py-spy       │    CPU     │ py-spy top -- python3 test.py │   ×   │  ✓   ┃
┃ scalene      │ CPU+Memory │ scalene test.py               │   ✓   │  ×   ┃
┃ memray       │   Memory   │ memray run --live test.py     │   ✓   │  ✓   ┃
┗━━━━━━━━━━━━━━┷━━━━━━━━━━━━┷━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┷━━━━━━━┷━━━━━━┛

Array manipulation mini-language. It can run up to one hundred times faster than the equivalent Python code. An even faster alternative that runs on a GPU is called CuPy.

import numpy as np

<array> = np.array(<list/list_of_lists/…>)              
<array> = np.zeros/ones/empty(<shape>)                  
<array> = np.arange(from_inc, to_exc, ±step)            
<array> = np.random.randint(from_inc, to_exc, <shape>)  

<view>  = <array>.reshape(<shape>)                      
<array> = <array>.flatten()                             
<view>  = <array>.transpose()                           

<array> = np.copy/abs/sqrt/log/int64(<array>)           
<array> = <array>.sum/max/mean/argmax/all(axis)         
<array> = np.apply_along_axis(<func>, axis, <array>)    

<array> = np.concatenate(<list_of_arrays>, axis=0)      
<array> = np.vstack/column_stack(<list_of_arrays>)      
<array> = np.tile/repeat(<array>, <int/list> [, axis])  

• Shape is a tuple of dimension sizes. A 100x50 RGB image has shape (50, 100, 3).
• Axis is an index of a dimension. Leftmost dimension has index 0. Summing the RGB image along axis 2 will return a greyscale image with shape (50, 100).

<el>       = <2d>[row_index, col_index]                 
<1d_view>  = <2d>[row_index]                            
<1d_view>  = <2d>[:, col_index]                         
<2d_view>  = <2d>[from:to_row_i, from:to_col_i]         

<1d_array> = <2d>[row_indices, col_indices]             
<2d_array> = <2d>[row_indices]                          
<2d_array> = <2d>[:, col_indices]                       
<2d_array> = <2d>[np.ix_(row_indices, col_indices)]     

<2d_bools> = <2d> > <el/1d/2d>                          
<1/2d_arr> = <2d>[<2d/1d_bools>]                        

• ':' returns a slice of all dimension's indices. Omitted dimensions default to ':'.
• Python converts 'obj[i, j]' to 'obj[(i, j)]'. This makes '<2d>[row_i, col_i]' and '<2d>[row_indices]' indistinguishable to NumPy if tuple of two indices is passed!
• 'ix_([1, 2], [3, 4])' returns '[[1], [2]]' and '[[3, 4]]'. Due to broadcasting rules, this is the same as using '[[1, 1], [2, 2]]' and '[[3, 4], [3, 4]]'.
• Any value that is broadcastable to the indexed shape can be assigned to the selection.

A set of rules by which NumPy functions operate on arrays of different shapes.

left  = np.array([0.1,  0.6,  0.8])                     
right = np.array([[0.1], [0.6], [0.8]])                 

left  = np.array([[0.1,  0.6,  0.8]])                   
right = np.array([[0.1], [0.6], [0.8]])                 

left  = np.array([[0.1,  0.6,  0.8],                    
                  [0.1,  0.6,  0.8],
                  [0.1,  0.6,  0.8]])

right = np.array([[0.1,  0.1,  0.1],                    
                  [0.6,  0.6,  0.6],
                  [0.8,  0.8,  0.8]])

>>> print(points := np.array([0.1, 0.6, 0.8]))
[0.1  0.6  0.8]
>>> print(wrapped_points := points.reshape(3, 1))
[[0.1]
 [0.6]
 [0.8]]
>>> print(deltas := points - wrapped_points)
[[ 0.   0.5  0.7]
 [-0.5  0.   0.2]
 [-0.7 -0.2  0. ]]
>>> deltas[range(3), range(3)] = np.inf
>>> print(distances := np.abs(deltas))
[[inf  0.5  0.7]
 [0.5  inf  0.2]
 [0.7  0.2  inf]]
>>> print(distances.argmin(axis=1))
[1 2 1]

from PIL import Image

<Image> = Image.new('<mode>', (width, height))  
<Image> = Image.open(<path>)                    
<Image> = <Image>.convert('<mode>')             
<Image>.save(<path>)                            
<Image>.show()                                  

<int/tup> = <Image>.getpixel((x, y))            
<ImgCore> = <Image>.getdata()                   
<Image>.putpixel((x, y), <int/tuple>)           
<Image>.putdata(<list/ImgCore>)                 
<Image>.paste(<Image>, (x, y))                  

<Image> = <Image>.filter(<Filter>)              
<Image> = <Enhance>.enhance(<float>)            

<array> = np.array(<Image>)                     
<Image> = Image.fromarray(np.uint8(<array>))    

• 'L' - Lightness (greyscale image). Each pixel is an integer between 0 and 255.
• 'RGB' - Red, green, blue (true color image). Each pixel is a tuple of three integers.
• 'RGBA' - RGB with alpha. Low alpha (i.e. fourth int) makes pixel more transparent.
• 'HSV' - Hue, saturation, value. Three ints representing color in HSV color space.

WIDTH, HEIGHT = 100, 100
n_pixels = WIDTH * HEIGHT
hues = (255 * i/n_pixels for i in range(n_pixels))
img = Image.new('HSV', (WIDTH, HEIGHT))
img.putdata([(int(h), 255, 255) for h in hues])
img.convert('RGB').save('test.png')

from random import randint
add_noise = lambda value: max(0, min(255, value + randint(-20, 20)))
img = Image.open('test.png').convert('HSV')
img.putdata([(add_noise(h), s, v) for h, s, v in img.getdata()])
img.show()

from PIL import ImageDraw
<Draw> = ImageDraw.Draw(<Image>)                
<Draw>.point((x, y))                            
<Draw>.line((x1, y1, x2, y2 [, ...]))           
<Draw>.arc((x1, y1, x2, y2), deg1, deg2)        
<Draw>.rectangle((x1, y1, x2, y2))              
<Draw>.polygon((x1, y1, x2, y2, ...))           
<Draw>.ellipse((x1, y1, x2, y2))                
<Draw>.text((x, y), <str>, font=<Font>)         

• Use 'fill=<color>' to set the primary color.
• Use 'width=<int>' to set the width of lines or contours.
• Use 'outline=<color>' to set the color of the contours.
• Color can be an int, tuple, '#rrggbb[aa]' string or a color name.

from PIL import Image, ImageDraw
import imageio

WIDTH, HEIGHT, R = 126, 126, 10
frames = []
for velocity in range(1, 16):
    y = sum(range(velocity))
    frame = Image.new('L', (WIDTH, HEIGHT))
    draw = ImageDraw.Draw(frame)
    draw.ellipse((WIDTH/2-R, y, WIDTH/2+R, y+R*2), fill='white')
    frames.append(frame)
frames += reversed(frames[1:-1])
imageio.mimsave('test.gif', frames, duration=0.03)

<Wave>  = wave.open('<path>')         
<int>   = <Wave>.getframerate()       
<int>   = <Wave>.getnchannels()       
<int>   = <Wave>.getsampwidth()       
<tuple> = <Wave>.getparams()          
<bytes> = <Wave>.readframes(nframes)  

<Wave> = wave.open('<path>', 'wb')    
<Wave>.setframerate(<int>)            
<Wave>.setnchannels(<int>)            
<Wave>.setsampwidth(<int>)            
<Wave>.setparams(<tuple>)             
<Wave>.writeframes(<bytes>)           

• Bytes object contains a sequence of frames, each consisting of one or more samples.
• In a stereo signal, the first sample of a frame belongs to the left channel.
• Each sample consists of one or more bytes that, when converted to an integer, indicate the displacement of a speaker membrane at a given moment.
• If sample width is one byte, then the integer should be encoded unsigned. For all other sizes, the integer should be encoded signed with little-endian byte order.

┏━━━━━━━━━━━┯━━━━━━━━━━━┯━━━━━━┯━━━━━━━━━━━┓
┃ sampwidth │    min    │ zero │    max    ┃
┠───────────┼───────────┼──────┼───────────┨
┃     1     │         0 │  128 │       255 ┃
┃     2     │    -32768 │    0 │     32767 ┃
┃     3     │  -8388608 │    0 │   8388607 ┃
┗━━━━━━━━━━━┷━━━━━━━━━━━┷━━━━━━┷━━━━━━━━━━━┛

def read_wav_file(filename):
    def get_int(bytes_obj):
        an_int = int.from_bytes(bytes_obj, 'little', signed=(p.sampwidth != 1))
        return an_int - 128 * (p.sampwidth == 1)
    with wave.open(filename) as file:
        p = file.getparams()
        frames = file.readframes(-1)
    bytes_samples = (frames[i : i + p.sampwidth] for i in range(0, len(frames), p.sampwidth))
    return [get_int(b) / pow(2, (p.sampwidth * 8) - 1) for b in bytes_samples], p

def write_to_wav_file(filename, samples_f, p=None, nchannels=1, sampwidth=2, framerate=44100):
    def get_bytes(a_float):
        a_float = max(-1, min(1 - 2e-16, a_float))
        a_float += (p.sampwidth == 1)
        a_float *= pow(2, (p.sampwidth * 8) - 1)
        return int(a_float).to_bytes(p.sampwidth, 'little', signed=(p.sampwidth != 1))
    if p is None:
        p = wave._wave_params(nchannels, sampwidth, framerate, 0, 'NONE', 'not compressed')
    with wave.open(filename, 'wb') as file:
        file.setparams(p)
        file.writeframes(b''.join(get_bytes(f) for f in samples_f))

from math import pi, sin
samples_f = (sin(i * 2 * pi * 440 / 44100) for i in range(100_000))
write_to_wav_file('test.wav', samples_f)

from random import uniform
samples_f, params = read_wav_file('test.wav')
samples_f = (f + uniform(-0.05, 0.05) for f in samples_f)
write_to_wav_file('test.wav', samples_f, p=params)

from simpleaudio import play_buffer
with wave.open('test.wav') as file:
    frames, p = file.readframes(-1), file.getparams()
    play_buffer(frames, p.nchannels, p.sampwidth, p.framerate).wait_done()

import pyttsx3
engine = pyttsx3.init()
engine.say('Sally sells seashells by the seashore.')
engine.runAndWait()

import itertools as it, math, array, simpleaudio

def play_notes(notes, bpm=132, f=44100):
    get_pause   = lambda n_beats: it.repeat(0, int(n_beats * 60/bpm * f))
    sin_f       = lambda i, hz: math.sin(i * 2 * math.pi * hz / f)
    get_wave    = lambda hz, n_beats: (sin_f(i, hz) for i in range(int(n_beats * 60/bpm * f)))
    get_hz      = lambda note: 440 * 2 ** ((int(note[:2]) - 69) / 12)
    get_nbeats  = lambda note: 1/2 if '♩' in note else 1/4 if '♪' in note else 1
    get_samples = lambda n: get_wave(get_hz(n), get_nbeats(n)) if n else get_pause(1/4)
    samples_f   = it.chain.from_iterable(get_samples(n) for n in notes.split(','))
    samples_i   = array.array('h', (int(fl * 5000) for fl in samples_f))
    simpleaudio.play_buffer(samples_i, 1, 2, f).wait_done()

play_notes('83♩,81♪,,83♪,,78♪,,74♪,,78♪,,71♪,,,,83♪,,81♪,,83♪,,78♪,,74♪,,78♪,,71♪,,,,'
           '83♩,85♪,,86♪,,85♪,,86♪,,83♪,,85♩,83♪,,85♪,,81♪,,83♪,,81♪,,83♪,,79♪,,83♪,,,,')

import pygame as pg

pg.init()
screen = pg.display.set_mode((500, 500))
rect = pg.Rect(240, 240, 20, 20)
while not pg.event.get(pg.QUIT):
    for event in pg.event.get(pg.KEYDOWN):
        dx = (event.key == pg.K_RIGHT) - (event.key == pg.K_LEFT)
        dy = (event.key == pg.K_DOWN) - (event.key == pg.K_UP)
        rect = rect.move((dx * 20, dy * 20))
    screen.fill(pg.Color('black'))
    pg.draw.rect(screen, pg.Color('white'), rect)
    pg.display.flip()
pg.quit()

Object for storing rectangular coordinates.

<Rect> = pg.Rect(x, y, width, height)           
<int>  = <Rect>.x/y/centerx/centery/…           
<tup.> = <Rect>.topleft/center/…                
<Rect> = <Rect>.move((delta_x, delta_y))        

<bool> = <Rect>.collidepoint((x, y))            
<bool> = <Rect>.colliderect(<Rect>)             
<int>  = <Rect>.collidelist(<list_of_Rect>)     
<list> = <Rect>.collidelistall(<list_of_Rect>)  

Object for representing images.

<Surf> = pg.display.set_mode((width, height))   
<Surf> = pg.Surface((width, height))            
<Surf> = pg.image.load(<path/file>)             
<Surf> = pg.surfarray.make_surface(<np_array>)  
<Surf> = <Surf>.subsurface(<Rect>)              

<Surf>.fill(color)                              
<Surf>.set_at((x, y), color)                    
<Surf>.blit(<Surf>, (x, y))                     

from pygame.transform import scale, rotate      
<Surf> = scale(<Surf>, (width, height))         
<Surf> = rotate(<Surf>, angle)                  
<Surf> = flip(<Surf>, flip_x=False)             

from pygame.draw import line, arc, rect         
line(<Surf>, color, (x1, y1), (x2, y2))         
arc(<Surf>, color, <Rect>, from_rad, to_rad)    
rect(<Surf>, color, <Rect>, width=0)            

<Font> = pg.font.Font(<path/file>, size)        
<Surf> = <Font>.render(text, antialias, color)  

<Sound> = pg.mixer.Sound(<path/file/bytes>)     
<Sound>.play/stop()                             

import collections, dataclasses, enum, io, itertools as it, pygame as pg, urllib.request
from random import randint

P = collections.namedtuple('P', 'x y')          
D = enum.Enum('D', 'n e s w')                   
W, H, MAX_S = 50, 50, P(5, 10)                  

def main():
    def get_screen():
        pg.init()
        return pg.display.set_mode((W*16, H*16))
    def get_images():
        url = 'https://gto76.github.io/python-cheatsheet/web/mario_bros.png'
        img = pg.image.load(io.BytesIO(urllib.request.urlopen(url).read()))
        return [img.subsurface(get_rect(x, 0)) for x in range(img.get_width() // 16)]
    def get_mario():
        Mario = dataclasses.make_dataclass('Mario', 'rect spd facing_left frame_cycle'.split())
        return Mario(get_rect(1, 1), P(0, 0), False, it.cycle(range(3)))
    def get_tiles():
        border = [(x, y) for x in range(W) for y in range(H) if x in [0, W-1] or y in [0, H-1]]
        platforms = [(randint(1, W-2), randint(2, H-2)) for _ in range(W*H // 10)]
        return [get_rect(x, y) for x, y in border + platforms]
    def get_rect(x, y):
        return pg.Rect(x*16, y*16, 16, 16)
    run(get_screen(), get_images(), get_mario(), get_tiles())

def run(screen, images, mario, tiles):
    clock = pg.time.Clock()
    pressed = set()
    while not pg.event.get(pg.QUIT):
        clock.tick(28)
        pressed |= {e.key for e in pg.event.get(pg.KEYDOWN)}
        pressed -= {e.key for e in pg.event.get(pg.KEYUP)}
        update_speed(mario, tiles, pressed)
        update_position(mario, tiles)
        draw(screen, images, mario, tiles)

def update_speed(mario, tiles, pressed):
    x, y = mario.spd
    x += 2 * ((pg.K_RIGHT in pressed) - (pg.K_LEFT in pressed))
    x += (x < 0) - (x > 0)
    y += 1 if D.s not in get_boundaries(mario.rect, tiles) else (pg.K_UP in pressed) * -10
    mario.spd = P(x=max(-MAX_S.x, min(MAX_S.x, x)), y=max(-MAX_S.y, min(MAX_S.y, y)))

def update_position(mario, tiles):
    x, y = mario.rect.topleft
    n_steps = max(abs(s) for s in mario.spd)
    for _ in range(n_steps):
        mario.spd = stop_on_collision(mario.spd, get_boundaries(mario.rect, tiles))
        x, y = x + (mario.spd.x / n_steps), y + (mario.spd.y / n_steps)
        mario.rect.topleft = x, y

def get_boundaries(rect, tiles):
    deltas = {D.n: P(0, -1), D.e: P(1, 0), D.s: P(0, 1), D.w: P(-1, 0)}
    return {d for d, delta in deltas.items() if rect.move(delta).collidelist(tiles) != -1}

def stop_on_collision(spd, bounds):
    return P(x=0 if (D.w in bounds and spd.x < 0) or (D.e in bounds and spd.x > 0) else spd.x,
             y=0 if (D.n in bounds and spd.y < 0) or (D.s in bounds and spd.y > 0) else spd.y)

def draw(screen, images, mario, tiles):
    screen.fill((85, 168, 255))
    mario.facing_left = mario.spd.x < 0 if mario.spd.x else mario.facing_left
    is_airborne = D.s not in get_boundaries(mario.rect, tiles)
    image_index = 4 if is_airborne else next(mario.frame_cycle) if mario.spd.x else 6
    screen.blit(images[image_index + (mario.facing_left * 9)], mario.rect)
    for t in tiles:
        is_border = t.x in [0, (W-1)*16] or t.y in [0, (H-1)*16]
        screen.blit(images[18 if is_border else 19], t)
    pg.display.flip()

if __name__ == '__main__':
    main()

Data analysis library. For examples see Plotly.

import pandas as pd, matplotlib.pyplot as plt

Ordered dictionary with a name.

>>> s = pd.Series([1, 2], index=['x', 'y'], name='a'); s
x    1
y    2
Name: a, dtype: int64

<S>  = pd.Series(<list>)                       
<S>  = pd.Series(<dict>)                       

<el> = <S>.loc[key]                            
<S>  = <S>.loc[coll_of_keys]                   
<S>  = <S>.loc[from_key : to_key_inc]          

<el> = <S>[key/i]                              
<S>  = <S>[coll_of_keys/coll_of_i]             
<S>  = <S>[<S_of_bools>]                       

<S>  = <S> > <el/S>                            
<S>  = <S> + <el/S>                            

<S>  = <S>.head/describe/sort_values()         
<S>  = <S>.str.strip/lower/contains/replace()  
<S>  = <S>.dt.year/month/day/hour              
<S>  = <S>.dt.to_period('y/m/d/h')             

<S>.plot.line/area/bar/pie/hist()              
plt.show()                                     

• Use 'print(<S>.to_string())' to print a Series that has more than sixty items.
• Use '<S>.index' to get collection of keys and '<S>.index = <coll>' to update them.
• Only pass a list or Series to loc/iloc because 'obj[x, y]' is converted to 'obj[(x, y)]' and '<S>.loc[key_1, key_2]' is how you retrieve a value from a multi-indexed Series.
• Pandas uses NumPy types like 'np.int64'. Series is converted to 'float64' if np.nan is assigned to any item. Use '<S>.astype(<str/type>)' to get converted Series.

<el> = <S>.sum/max/mean/std/idxmax/count()     
<S>  = <S>.rank/diff/cumsum/ffill/interpol…()  
<S>  = <S>.isna/fillna/isin([<el/coll>])       

┏━━━━━━━━━━━━━━┯━━━━━━━━━━━━━┯━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━┓
┃              │    'sum'    │   ['sum']   │ {'s': 'sum'}  ┃
┠──────────────┼─────────────┼─────────────┼───────────────┨
┃ s.apply(…)   │      3      │    sum  3   │     s  3      ┃
┃ s.agg(…)     │             │             │               ┃
┗━━━━━━━━━━━━━━┷━━━━━━━━━━━━━┷━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━┛

┏━━━━━━━━━━━━━━┯━━━━━━━━━━━━━┯━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━┓
┃              │    'rank'   │   ['rank']  │ {'r': 'rank'} ┃
┠──────────────┼─────────────┼─────────────┼───────────────┨
┃ s.apply(…)   │             │      rank   │               ┃
┃ s.agg(…)     │    x  1.0   │   x   1.0   │   r  x  1.0   ┃
┃              │    y  2.0   │   y   2.0   │      y  2.0   ┃
┗━━━━━━━━━━━━━━┷━━━━━━━━━━━━━┷━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━┛

Table with labeled rows and columns.

>>> df = pd.DataFrame([[1, 2], [3, 4]], index=['a', 'b'], columns=['x', 'y']); df
   x  y
a  1  2
b  3  4

<DF>   = pd.DataFrame(<list_of_rows>)          
<DF>   = pd.DataFrame(<dict_of_columns>)       

<el>   = <DF>.loc[row_key, col_key]            
<S/DF> = <DF>.loc[row_key/s]                   
<S/DF> = <DF>.loc[:, col_key/s]                
<DF>   = <DF>.loc[row_bools, col_bools]        

<S/DF> = <DF>[col_key/s]                       
<DF>   = <DF>[<S_of_bools>]                    
<DF>   = <DF>[<DF_of_bools>]                   

<DF>   = <DF> > <el/S/DF>                      
<DF>   = <DF> + <el/S/DF>                      

<DF>   = <DF>.set_index(col_key)               
<DF>   = <DF>.reset_index(drop=False)          
<DF>   = <DF>.sort_index(ascending=True)       
<DF>   = <DF>.sort_values(col_key/s)           

<DF>   = <DF>.head/tail/sample(<int>)          
<DF>   = <DF>.describe()                       
<DF>   = <DF>.query('<query>')                 

<DF>.plot.line/area/bar/scatter(x=col_key, …)  
plt.show()                                     

>>> df_2 = pd.DataFrame([[4, 5], [6, 7]], index=['b', 'c'], columns=['y', 'z']); df_2
   y  z
b  4  5
c  6  7

┏━━━━━━━━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━┯━━━━━━━━━━━━┯━━━━━━━━━━━━┯━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃                       │    'outer'    │   'inner'  │   'left'   │       Description         ┃
┠───────────────────────┼───────────────┼────────────┼────────────┼───────────────────────────┨
┃ df.merge(df_2,        │    x   y   z  │ x   y   z  │ x   y   z  │ Merges on column if 'on'  ┃
┃          on='y',      │ 0  1   2   .  │ 3   4   5  │ 1   2   .  │ or 'left_on/right_on' are ┃
┃          how=…)       │ 1  3   4   5  │            │ 3   4   5  │ set, else on shared cols. ┃
┃                       │ 2  .   6   7  │            │            │ Uses 'inner' by default.  ┃
┠───────────────────────┼───────────────┼────────────┼────────────┼───────────────────────────┨
┃ df.join(df_2,         │    x yl yr  z │            │ x yl yr  z │ Merges on row keys.       ┃
┃         lsuffix='l',  │ a  1  2  .  . │ x yl yr  z │ 1  2  .  . │ Uses 'left' by default.   ┃
┃         rsuffix='r',  │ b  3  4  4  5 │ 3  4  4  5 │ 3  4  4  5 │ If Series is passed, it   ┃
┃         how=…)        │ c  .  .  6  7 │            │            │ is treated as a column.   ┃
┠───────────────────────┼───────────────┼────────────┼────────────┼───────────────────────────┨
┃ pd.concat([df, df_2], │    x   y   z  │     y      │            │ Adds rows at the bottom.  ┃
┃           axis=0,     │ a  1   2   .  │     2      │            │ Uses 'outer' by default.  ┃
┃           join=…)     │ b  3   4   .  │     4      │            │ A Series is treated as a  ┃
┃                       │ b  .   4   5  │     4      │            │ column. To add a row use  ┃
┃                       │ c  .   6   7  │     6      │            │ pd.concat([df, DF([s])]). ┃
┠───────────────────────┼───────────────┼────────────┼────────────┼───────────────────────────┨
┃ pd.concat([df, df_2], │    x  y  y  z │            │            │ Adds columns at the       ┃
┃           axis=1,     │ a  1  2  .  . │ x  y  y  z │            │ right end. Uses 'outer'   ┃
┃           join=…)     │ b  3  4  4  5 │ 3  4  4  5 │            │ by default. A Series is   ┃
┃                       │ c  .  .  6  7 │            │            │ treated as a column.      ┃
┗━━━━━━━━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━┷━━━━━━━━━━━━┷━━━━━━━━━━━━┷━━━━━━━━━━━━━━━━━━━━━━━━━━━┛

<S>  = <DF>.sum/max/mean/std/idxmax/count()    
<DF> = <DF>.rank/diff/cumsum/ffill/interpo…()  
<DF> = <DF>.isna/fillna/isin([<el/coll>])      

┏━━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━┓
┃                 │     'sum'     │    ['sum']    │ {'x': 'sum'}  ┃
┠─────────────────┼───────────────┼───────────────┼───────────────┨
┃ df.apply(…)     │      x  4     │        x  y   │     x  4      ┃
┃ df.agg(…)       │      y  6     │   sum  4  6   │               ┃
┗━━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━┛

┏━━━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━┯━━━━━━━━━━━━━━━┓
┃                 │     'rank'    │    ['rank']   │ {'x': 'rank'} ┃
┠─────────────────┼───────────────┼───────────────┼───────────────┨
┃ df.apply(…)     │               │       x    y  │               ┃
┃ df.agg(…)       │       x    y  │    rank rank  │         x     ┃
┃ df.transform(…) │  a  1.0  1.0  │  a  1.0  1.0  │    a  1.0     ┃
┃                 │  b  2.0  2.0  │  b  2.0  2.0  │    b  2.0     ┃
┗━━━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━┷━━━━━━━━━━━━━━━┛

• Listed methods process the columns unless they receive 'axis=1'. Exceptions to this rule are '<DF>.dropna()', '<DF>.drop(row_key/s)' and '<DF>.rename(<dict/func>)'.
• Fifth result's columns are indexed with a multi-index. This means we need a tuple of column keys to specify a column: '<DF>.loc[row_key, (col_key_1, col_key_2)]'.

<DF> = <DF>.loc[row_key_1]                     
<DF> = <DF>.loc[:, (slice(None), col_key_2)]   
<DF> = <DF>.set_index(col_keys)                
<DF> = <DF>.pivot_table(index=col_key/s)       
<S>  = <DF>.stack/unstack(level=-1)            

<S/DF> = pd.read_json/pickle(<path/url/file>)  
<DF>   = pd.read_csv/excel(<path/url/file>)    
<list> = pd.read_html(<path/url/file>)         
<S/DF> = pd.read_parquet/feather/hdf(<path…>)  
<DF>   = pd.read_sql('<table/query>', <conn>)  

<DF>.to_json/csv/html/latex/parquet(<path>)    
<DF>.to_pickle/excel/feather/hdf(<path>)       
<DF>.to_sql('<table_name>', <connection>)      

• '$ pip3 install "pandas[excel]" odfpy lxml pyarrow' installs dependencies.
• Csv functions use the same dialect as standard library's csv module (e.g. 'sep=","').
• Read_csv() only parses dates of columns that are listed in 'parse_dates'. It automatically tries to detect the format, but it can be helped with 'date_format' or 'dayfirst' arguments.
• We get a dataframe with DatetimeIndex if 'parse_dates' argument includes 'index_col'. Its 'resample("y/m/d/h")' method returns Resampler object that is similar to GroupBy.

Object that groups together rows of a dataframe based on the value of the passed column.

<GB> = <DF>.groupby(col_key/s)                 
<DF> = <GB>.apply/filter(<func>)               
<DF> = <GB>.get_group(<el>)                    
<S>  = <GB>.size()                             
<GB> = <GB>[col_key]                           

<DF> = <GB>.sum/max/mean/std/idxmax/count()    
<DF> = <GB>.rank/diff/cumsum/ffill()           
<DF> = <GB>.fillna(<el>)                       

>>> df = pd.DataFrame([[1, 2, 3], [4, 5, 6], [7, 8, 6]], list('abc'), list('xyz'))
>>> gb = df.groupby('z'); gb.apply(print)
   x  y  z
a  1  2  3
   x  y  z
b  4  5  6
c  7  8  6
>>> gb.sum()
    x   y
z
3   1   2
6  11  13

Object for rolling window calculations.

<RS/RDF/RGB> = <S/DF/GB>.rolling(win_size)     
<RS/RDF/RGB> = <RDF/RGB>[col_key/s]            
<S/DF>       = <R>.mean/sum/max()              

import plotly.express as px, pandas as pd

<Fig> = px.line(<DF> [, y=col_key/s [, x=col_key]])   
<Fig>.update_layout(paper_bgcolor='#rrggbb')          
<Fig>.write_html/json/image('<path>')                 

<Fig> = px.area/bar/box(<DF>, x=col_key, y=col_keys)  
<Fig> = px.scatter(<DF>, x=col_key, y=col_keys)       
<Fig> = px.scatter_3d(<DF>, x=col_key, y=col_key, …)  
<Fig> = px.histogram(<DF>, x=col_keys, y=col_key)     

covid = pd.read_csv('https://raw.githubusercontent.com/owid/covid-19-data/8dde8ca49b'
                    '6e648c17dd420b2726ca0779402651/public/data/owid-covid-data.csv',
                    usecols=['iso_code', 'date', 'population', 'total_deaths'])
continents = pd.read_csv('https://gto76.github.io/python-cheatsheet/web/continents.csv',
                         usecols=['Three_Letter_Country_Code', 'Continent_Name'])
df = pd.merge(covid, continents, left_on='iso_code', right_on='Three_Letter_Country_Code')
df = df.groupby(['Continent_Name', 'date']).sum().reset_index()
df['Total Deaths per Million'] = df.total_deaths * 1e6 / df.population
df = df[df.date > '2020-03-14']
df = df.rename({'date': 'Date', 'Continent_Name': 'Continent'}, axis='columns')
px.line(df, x='Date', y='Total Deaths per Million', color='Continent')

import pandas as pd, selenium.webdriver, io, plotly.graph_objects as go

def main():
    covid, (bitcoin, gold, dow) = get_covid_cases(), get_tickers()
    df = wrangle_data(covid, bitcoin, gold, dow)
    display_data(df)

def get_covid_cases():
    url = 'https://catalog.ourworldindata.org/garden/covid/latest/compact/compact.csv'
    df = pd.read_csv(url, parse_dates=['date'])
    df = df[df.country == 'World']
    s = df.set_index('date').total_cases
    return s.rename('Total Cases')

def get_tickers():
    with selenium.webdriver.Chrome() as driver:
        driver.implicitly_wait(10)
        symbols = {'Bitcoin': 'BTC-USD', 'Gold': 'GC=F', 'Dow Jones': '%5EDJI'}
        return [get_ticker(driver, name, symbol) for name, symbol in symbols.items()]

def get_ticker(driver, name, symbol):
    url = f'https://finance.yahoo.com/quote/{symbol}/history/'
    driver.get(url + '?period1=1579651200&period2=9999999999')
    if buttons := driver.find_elements('xpath', '//button[@name="reject"]'):
        buttons[0].click()
    html = io.StringIO(driver.page_source)
    dataframes = pd.read_html(html, parse_dates=['Date'])
    s = dataframes[0].set_index('Date').Open
    return s.rename(name)

def wrangle_data(covid, bitcoin, gold, dow):
    df = pd.concat([bitcoin, gold, dow], axis=1)  
    df = df.sort_index().interpolate()            
    df = df.loc['2020-02-23':'2021-12-20']        
    df = (df / df.iloc[0]) * 100                  
    df = df.join(covid)                           
    return df.sort_values(df.index[-1], axis=1)   

def display_data(df):
    figure = go.Figure()
    for col_name in reversed(df.columns):
        yaxis = 'y1' if col_name == 'Total Cases' else 'y2'
        trace = go.Scatter(x=df.index, y=df[col_name], yaxis=yaxis, name=col_name)
        figure.add_trace(trace)
    figure.update_layout(
        width=944,
        height=423,
        yaxis1=dict(title='Total Cases', rangemode='tozero'),
        yaxis2=dict(title='%', rangemode='tozero', overlaying='y', side='right'),
        colorway=['#EF553B', '#636EFA', '#00CC96', '#FFA152'],
        legend=dict(x=1.08)
    )
    figure.show()

if __name__ == '__main__':
    main()

Library that compiles Python-like code into C.

import pyximport; pyximport.install()                
import <cython_script>                               

cdef <type> <var_name> [= <obj/var>]                 
cdef <ctype> *<pointer_name> [= &<var>]              
cdef <ctype>[size] <array_name> [= <coll/array>]     
cdef <ctype> *<array_name> [= <coll/array/pointer>]  

cdef <type> <func_name>(<type> [*]<arg_name>): ...   

cdef class <class_name>:                             
    cdef public <type> [*]<attr_name>                
    def __init__(self, <type> <arg_name>):           
        self.<attr_name> = <arg_name>                

System for installing libraries directly into project's directory.

$ python3 -m venv NAME      
$ source NAME/bin/activate  
$ pip3 install LIBRARY      
$ python3 FILE              
$ deactivate                

Run the script with '$ python3 FILE' or '$ chmod u+x FILE; ./FILE'. To automatically start the debugger when uncaught exception occurs run '$ python3 -m pdb -cc FILE'.

from sys import argv, exit
from collections import defaultdict, namedtuple
from dataclasses import make_dataclass
from enum import Enum
import functools as ft, itertools as it, operator as op, re


def main():
    pass






def read_file(filename):
    with open(filename, encoding='utf-8') as file:
        return file.readlines()


if __name__ == '__main__':
    main()

• Ctrl+F / ⌘F is usually sufficient.
• Searching '#<title>' will limit the search to the titles.
• Click on the title's '#' to get a link to its section.

RetroSearch is an open source project built by @garambo | Open a GitHub Issue

Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo

HTML: 3.2 | Encoding: UTF-8 | Version: 0.7.4