Python 实用宝典

Question 1

In pre-historic times (Python 1.4) we did:

fp = open('filename.txt')
while 1:
    line = fp.readline()
    if not line:
        break
    print line

after Python 2.1, we did:

for line in open('filename.txt').xreadlines():
    print line

before we got the convenient iterator protocol in Python 2.3, and could do:

for line in open('filename.txt'):
    print line

I’ve seen some examples using the more verbose:

with open('filename.txt') as fp:
    for line in fp:
        print line

is this the preferred method going forwards?

[edit] I get that the with statement ensures closing of the file… but why isn’t that included in the iterator protocol for file objects?

Question 2

There is exactly one reason why the following is preferred:

with open('filename.txt') as fp:
    for line in fp:
        print line

We are all spoiled by CPython’s relatively deterministic reference-counting scheme for garbage collection. Other, hypothetical implementations of Python will not necessarily close the file “quickly enough” without the with block if they use some other scheme to reclaim memory.

In such an implementation, you might get a “too many files open” error from the OS if your code opens files faster than the garbage collector calls finalizers on orphaned file handles. The usual workaround is to trigger the GC immediately, but this is a nasty hack and it has to be done by every function that could encounter the error, including those in libraries. What a nightmare.

Or you could just use the with block.

Bonus Question

(Stop reading now if are only interested in the objective aspects of the question.)

Why isn’t that included in the iterator protocol for file objects?

This is a subjective question about API design, so I have a subjective answer in two parts.

On a gut level, this feels wrong, because it makes iterator protocol do two separate things—iterate over lines and close the file handle—and it’s often a bad idea to make a simple-looking function do two actions. In this case, it feels especially bad because iterators relate in a quasi-functional, value-based way to the contents of a file, but managing file handles is a completely separate task. Squashing both, invisibly, into one action, is surprising to humans who read the code and makes it more difficult to reason about program behavior.

Other languages have essentially come to the same conclusion. Haskell briefly flirted with so-called “lazy IO” which allows you to iterate over a file and have it automatically closed when you get to the end of the stream, but it’s almost universally discouraged to use lazy IO in Haskell these days, and Haskell users have mostly moved to more explicit resource management like Conduit which behaves more like the with block in Python.

On a technical level, there are some things you may want to do with a file handle in Python which would not work as well if iteration closed the file handle. For example, suppose I need to iterate over the file twice:

with open('filename.txt') as fp:
    for line in fp:
        ...
    fp.seek(0)
    for line in fp:
        ...

While this is a less common use case, consider the fact that I might have just added the three lines of code at the bottom to an existing code base which originally had the top three lines. If iteration closed the file, I wouldn’t be able to do that. So keeping iteration and resource management separate makes it easier to compose chunks of code into a larger, working Python program.

Composability is one of the most important usability features of a language or API.

Question 3

Yes,

with open('filename.txt') as fp:
    for line in fp:
        print line

is the way to go.

It is not more verbose. It is more safe.

Question 4

if you’re turned off by the extra line, you can use a wrapper function like so:

def with_iter(iterable):
    with iterable as iter:
        for item in iter:
            yield item

for line in with_iter(open('...')):
    ...

in Python 3.3, the yield from statement would make this even shorter:

def with_iter(iterable):
    with iterable as iter:
        yield from iter

Question 5

f = open('test.txt','r')
for line in f.xreadlines():
    print line
f.close()

Question 6

python setup.py install will automatically install packages listed in requires=[] using easy_install. How do I get it to use pip instead?

Question 7

Yes you can. You can install a package from a tarball or a folder, on the web or your computer. For example:

Install from tarball on web

pip install https://pypi.python.org/packages/source/r/requests/requests-2.3.0.tar.gz

Install from local tarball

wget https://pypi.python.org/packages/source/r/requests/requests-2.3.0.tar.gz
pip install requests-2.3.0.tar.gz

Install from local folder

tar -zxvf requests-2.3.0.tar.gz
cd requests-2.3.0
pip install .

You can delete the requests-2.3.0 folder.

Install from local folder (editable mode)

pip install -e .

This installs the package in editable mode. Any changes you make to the code will immediately apply across the system. This is useful if you are the package developer and want to test changes. It also means you can’t delete the folder without breaking the install.

Question 8

You can pip install a file perhaps by python setup.py sdist first. You can also pip install -e . which is like python setup.py develop.

Question 9

If you are really set on using python setup.py install you could try something like this:

from setuptools import setup, find_packages
from setuptools.command.install import install as InstallCommand


class Install(InstallCommand):
    """ Customized setuptools install command which uses pip. """

    def run(self, *args, **kwargs):
        import pip
        pip.main(['install', '.'])
        InstallCommand.run(self, *args, **kwargs)


setup(
    name='your_project',
    version='0.0.1a',
    cmdclass={
        'install': Install,
    },
    packages=find_packages(),
    install_requires=['simplejson']
)

Question 10

I want to perform an element wise multiplication, to multiply two lists together by value in Python, like we can do it in Matlab.

This is how I would do it in Matlab.

a = [1,2,3,4]
b = [2,3,4,5]
a .* b = [2, 6, 12, 20]

A list comprehension would give 16 list entries, for every combination x * y of x from a and y from b. Unsure of how to map this.

If anyone is interested why, I have a dataset, and want to multiply it by Numpy.linspace(1.0, 0.5, num=len(dataset)) =).

Question 11

Use a list comprehension mixed with zip():.

[a*b for a,b in zip(lista,listb)]

Question 12

Since you’re already using numpy, it makes sense to store your data in a numpy array rather than a list. Once you do this, you get things like element-wise products for free:

In [1]: import numpy as np

In [2]: a = np.array([1,2,3,4])

In [3]: b = np.array([2,3,4,5])

In [4]: a * b
Out[4]: array([ 2,  6, 12, 20])

Question 13

Use np.multiply(a,b):

import numpy as np
a = [1,2,3,4]
b = [2,3,4,5]
np.multiply(a,b)

Question 14

You can try multiplying each element in a loop. The short hand for doing that is

ab = [a[i]*b[i] for i in range(len(a))]

Question 15

Yet another answer:

-1 … requires import
+1 … is very readable

import operator
a = [1,2,3,4]
b = [10,11,12,13]

list(map(operator.mul, a, b))

outputs [10, 22, 36, 52]

Question 16

Fairly intuitive way of doing this:

a = [1,2,3,4]
b = [2,3,4,5]
ab = []                        #Create empty list
for i in range(0, len(a)):
     ab.append(a[i]*b[i])      #Adds each element to the list

Question 17

you can multiplication using lambda

foo=[1,2,3,4]
bar=[1,2,5,55]
l=map(lambda x,y:x*y,foo,bar)

Question 18

For large lists, we can do it the iter-way:

product_iter_object = itertools.imap(operator.mul, [1,2,3,4], [2,3,4,5])

product_iter_object.next() gives each of the element in the output list.

The output would be the length of the shorter of the two input lists.

Question 19

create an array of ones; multiply each list times the array; convert array to a list

import numpy as np

a = [1,2,3,4]
b = [2,3,4,5]

c = (np.ones(len(a))*a*b).tolist()

[2.0, 6.0, 12.0, 20.0]

Question 20

gahooa’s answer is correct for the question as phrased in the heading, but if the lists are already numpy format or larger than ten it will be MUCH faster (3 orders of magnitude) as well as more readable, to do simple numpy multiplication as suggested by NPE. I get these timings:

0.0049ms -> N = 4, a = [i for i in range(N)], c = [a*b for a,b in zip(a, b)]
0.0075ms -> N = 4, a = [i for i in range(N)], c = a * b
0.0167ms -> N = 4, a = np.arange(N), c = [a*b for a,b in zip(a, b)]
0.0013ms -> N = 4, a = np.arange(N), c = a * b
0.0171ms -> N = 40, a = [i for i in range(N)], c = [a*b for a,b in zip(a, b)]
0.0095ms -> N = 40, a = [i for i in range(N)], c = a * b
0.1077ms -> N = 40, a = np.arange(N), c = [a*b for a,b in zip(a, b)]
0.0013ms -> N = 40, a = np.arange(N), c = a * b
0.1485ms -> N = 400, a = [i for i in range(N)], c = [a*b for a,b in zip(a, b)]
0.0397ms -> N = 400, a = [i for i in range(N)], c = a * b
1.0348ms -> N = 400, a = np.arange(N), c = [a*b for a,b in zip(a, b)]
0.0020ms -> N = 400, a = np.arange(N), c = a * b

i.e. from the following test program.

import timeit

init = ['''
import numpy as np
N = {}
a = {}
b = np.linspace(0.0, 0.5, len(a))
'''.format(i, j) for i in [4, 40, 400] 
                  for j in ['[i for i in range(N)]', 'np.arange(N)']]

func = ['''c = [a*b for a,b in zip(a, b)]''',
'''c = a * b''']

for i in init:
  for f in func:
    lines = i.split('\n')
    print('{:6.4f}ms -> {}, {}, {}'.format(
           timeit.timeit(f, setup=i, number=1000), lines[2], lines[3], f))

Question 21

Can use enumerate.

a = [1, 2, 3, 4]
b = [2, 3, 4, 5]

ab = [val * b[i] for i, val in enumerate(a)]

Question 22

The map function can be very useful here. Using map we can apply any function to each element of an iterable.

Python 3.x

>>> def my_mul(x,y):
...     return x*y
...
>>> a = [1,2,3,4]
>>> b = [2,3,4,5]
>>>
>>> list(map(my_mul,a,b))
[2, 6, 12, 20]
>>>

Of course:

map(f, iterable)

is equivalent to

[f(x) for x in iterable]

So we can get our solution via:

>>> [my_mul(x,y) for x, y in zip(a,b)]
[2, 6, 12, 20]
>>>

In Python 2.x map() means: apply a function to each element of an iterable and construct a new list. In Python 3.x, map construct iterators instead of lists.

Instead of my_mul we could use mul operator

Python 2.7

>>>from operator import mul # import mul operator
>>>a = [1,2,3,4]
>>>b = [2,3,4,5]
>>>map(mul,a,b)
[2, 6, 12, 20]
>>>

Python 3.5+

>>> from operator import mul
>>> a = [1,2,3,4]
>>> b = [2,3,4,5]
>>> [*map(mul,a,b)]
[2, 6, 12, 20]
>>>

Please note that since map() constructs an iterator we use * iterable unpacking operator to get a list. The unpacking approach is a bit faster then the list constructor:

>>> list(map(mul,a,b))
[2, 6, 12, 20]
>>>

Question 23

To maintain the list type, and do it in one line (after importing numpy as np, of course):

list(np.array([1,2,3,4]) * np.array([2,3,4,5]))

or

list(np.array(a) * np.array(b))

Question 24

you can use this for lists of the same length

def lstsum(a, b):
    c=0
    pos = 0
for element in a:
   c+= element*b[pos]
   pos+=1
return c

Question 25

I’m working on making a URL shortener for my site, and my current plan (I’m open to suggestions) is to use a node ID to generate the shortened URL. So, in theory, node 26 might be short.com/z, node 1 might be short.com/a, node 52 might be short.com/Z, and node 104 might be short.com/ZZ. When a user goes to that URL, I need to reverse the process (obviously).

I can think of some kludgy ways to go about this, but I’m guessing there are better ones. Any suggestions?

Question 26

ASCII to int:

ord('a')

gives 97

And back to a string:

in Python2: str(unichr(97))
in Python3: chr(97)

gives 'a'

Question 27

>>> ord("a")
97
>>> chr(97)
'a'

Question 28

If multiple characters are bound inside a single integer/long, as was my issue:

s = '0123456789'
nchars = len(s)
# string to int or long. Type depends on nchars
x = sum(ord(s[byte])<<8*(nchars-byte-1) for byte in range(nchars))
# int or long to string
''.join(chr((x>>8*(nchars-byte-1))&0xFF) for byte in range(nchars))

Yields '0123456789' and x = 227581098929683594426425L

Question 29

What about BASE58 encoding the URL? Like for example flickr does.

# note the missing lowercase L and the zero etc.
BASE58 = '123456789abcdefghijkmnopqrstuvwxyzABCDEFGHJKLMNPQRSTUVWXYZ' 
url = ''
while node_id >= 58:
    div, mod = divmod(node_id, 58)
    url = BASE58[mod] + url
    node_id = int(div)

return 'http://short.com/%s' % BASE58[node_id] + url

Turning that back into a number isn’t a big deal either.

Question 30

Use hex(id)[2:] and int(urlpart, 16). There are other options. base32 encoding your id could work as well, but I don’t know that there’s any library that does base32 encoding built into Python.

Apparently a base32 encoder was introduced in Python 2.4 with the base64 module. You might try using b32encode and b32decode. You should give True for both the casefold and map01 options to b32decode in case people write down your shortened URLs.

Actually, I take that back. I still think base32 encoding is a good idea, but that module is not useful for the case of URL shortening. You could look at the implementation in the module and make your own for this specific case. :-)

Question 31

A lot of our modules start with:

try:
    import json
except ImportError:
    from django.utils import simplejson as json  # Python 2.4 fallback.

…and it’s the only Pyflakes warning in the entire file:

foo/bar.py:14: redefinition of unused 'json' from line 12

How can I get Pyflakes to ignore this?

(Normally I’d go read the docs but the link is broken. If nobody has an answer, I’ll just read the source.)

Question 32

If you can use flake8 instead – which wraps pyflakes as well as the pep8 checker – a line ending with

# NOQA

(in which the space is significant – 2 spaces between the end of the code and the #, one between it and the NOQA text) will tell the checker to ignore any errors on that line.

Question 33

I know this was questioned some time ago and is already answered.

But I wanted to add what I usually use:

try:
    import json
    assert json  # silence pyflakes
except ImportError:
    from django.utils import simplejson as json  # Python 2.4 fallback.

Question 34

Yep, unfortunately dimod.org is down together with all goodies.

Looking at the pyflakes code, it seems to me that pyflakes is designed so that it will be easy to use it as an “embedded fast checker”.

For implementing ignore functionality you will need to write your own that calls the pyflakes checker.

Here you can find an idea: http://djangosnippets.org/snippets/1762/

Note that the above snippet only for for comments places on the same line. For ignoring a whole block you might want to add ‘pyflakes:ignore’ in the block docstring and filter based on node.doc.

Good luck!

I am using pocket-lint for all kind of static code analysis. Here are the changes made in pocket-lint for ignoring pyflakes: https://code.launchpad.net/~adiroiban/pocket-lint/907742/+merge/102882

Question 35

To quote from the github issue ticket:

While the fix is still coming, this is how it can be worked around, if you’re wondering:
try:
    from unittest.runner import _WritelnDecorator
    _WritelnDecorator; # workaround for pyflakes issue #13
except ImportError:
    from unittest import _WritelnDecorator
Substitude _unittest and _WritelnDecorator with the entities (modules, functions, classes) you need

— deemoowoor

Question 36

Here is a monkey patch for pyflakes that adds a # bypass_pyflakes comment option.

bypass_pyflakes.py

#!/usr/bin/env python

from pyflakes.scripts import pyflakes
from pyflakes.checker import Checker


def report_with_bypass(self, messageClass, *args, **kwargs):
    text_lineno = args[0] - 1
    with open(self.filename, 'r') as code:
        if code.readlines()[text_lineno].find('bypass_pyflakes') >= 0:
            return
    self.messages.append(messageClass(self.filename, *args, **kwargs))

# monkey patch checker to support bypass
Checker.report = report_with_bypass

pyflakes.main()

If you save this as bypass_pyflakes.py, then you can invoke it as python bypass_pyflakes.py myfile.py.

http://chase-seibert.github.com/blog/2013/01/11/bypass_pyflakes.html

Question 37

You can also import with __import__. It’s not pythonic, but pyflakes does not warn you anymore. See documentation for __import__ .

try:
    import json
except ImportError:
    __import__('django.utils', globals(), locals(), ['json'], -1)

Question 38

I created a little shell script with some awk magic to help me. With this all lines with import typing, from typing import or #$ (latter is a special comment I am using here) are excluded ($1 is the file name of the Python script):

result=$(pyflakes -- "$1" 2>&1)

# check whether there is any output
if [ "$result" ]; then

    # lines to exclude
    excl=$(awk 'BEGIN { ORS="" } /(#\$)|(import +typing)|(from +typing +import )/ { print sep NR; sep="|" }' "$1")

    # exclude lines if there are any (otherwise we get invalid regex)
    [ "$excl" ] &&
        result=$(awk "! /^[^:]+:(${excl}):/" <<< "$result")

fi

# now echo "$result" or such ...

Basically it notes the line numbers and dynamically creates a regex out it.

Question 39

In a project using SciPy and NumPy, should I use scipy.pi, numpy.pi, or math.pi?

Question 40

>>> import math
>>> import numpy as np
>>> import scipy
>>> math.pi == np.pi == scipy.pi
True

So it doesn’t matter, they are all the same value.

The only reason all three modules provide a pi value is so if you are using just one of the three modules, you can conveniently have access to pi without having to import another module. They’re not providing different values for pi.

Question 41

One thing to note is that not all libraries will use the same meaning for pi, of course, so it never hurts to know what you’re using. For example, the symbolic math library Sympy’s representation of pi is not the same as math and numpy:

import math
import numpy
import scipy
import sympy

print(math.pi == numpy.pi)
> True
print(math.pi == scipy.pi)
> True
print(math.pi == sympy.pi)
> False

Question 42

How can I remove all characters except numbers from string?

Question 43

In Python 2.*, by far the fastest approach is the .translate method:

>>> x='aaa12333bb445bb54b5b52'
>>> import string
>>> all=string.maketrans('','')
>>> nodigs=all.translate(all, string.digits)
>>> x.translate(all, nodigs)
'1233344554552'
>>>

string.maketrans makes a translation table (a string of length 256) which in this case is the same as ''.join(chr(x) for x in range(256)) (just faster to make;-). .translate applies the translation table (which here is irrelevant since all essentially means identity) AND deletes characters present in the second argument — the key part.

.translate works very differently on Unicode strings (and strings in Python 3 — I do wish questions specified which major-release of Python is of interest!) — not quite this simple, not quite this fast, though still quite usable.

Back to 2.*, the performance difference is impressive…:

$ python -mtimeit -s'import string; all=string.maketrans("", ""); nodig=all.translate(all, string.digits); x="aaa12333bb445bb54b5b52"' 'x.translate(all, nodig)'
1000000 loops, best of 3: 1.04 usec per loop
$ python -mtimeit -s'import re;  x="aaa12333bb445bb54b5b52"' 're.sub(r"\D", "", x)'
100000 loops, best of 3: 7.9 usec per loop

Speeding things up by 7-8 times is hardly peanuts, so the translate method is well worth knowing and using. The other popular non-RE approach…:

$ python -mtimeit -s'x="aaa12333bb445bb54b5b52"' '"".join(i for i in x if i.isdigit())'
100000 loops, best of 3: 11.5 usec per loop

is 50% slower than RE, so the .translate approach beats it by over an order of magnitude.

In Python 3, or for Unicode, you need to pass .translate a mapping (with ordinals, not characters directly, as keys) that returns None for what you want to delete. Here’s a convenient way to express this for deletion of “everything but” a few characters:

import string

class Del:
  def __init__(self, keep=string.digits):
    self.comp = dict((ord(c),c) for c in keep)
  def __getitem__(self, k):
    return self.comp.get(k)

DD = Del()

x='aaa12333bb445bb54b5b52'
x.translate(DD)

also emits '1233344554552'. However, putting this in xx.py we have…:

$ python3.1 -mtimeit -s'import re;  x="aaa12333bb445bb54b5b52"' 're.sub(r"\D", "", x)'
100000 loops, best of 3: 8.43 usec per loop
$ python3.1 -mtimeit -s'import xx; x="aaa12333bb445bb54b5b52"' 'x.translate(xx.DD)'
10000 loops, best of 3: 24.3 usec per loop

…which shows the performance advantage disappears, for this kind of “deletion” tasks, and becomes a performance decrease.

Question 44

Use re.sub, like so:

>>> import re
>>> re.sub('\D', '', 'aas30dsa20')
'3020'

\D matches any non-digit character so, the code above, is essentially replacing every non-digit character for the empty string.

Or you can use filter, like so (in Python 2):

>>> filter(str.isdigit, 'aas30dsa20')
'3020'

Since in Python 3, filter returns an iterator instead of a list, you can use the following instead:

>>> ''.join(filter(str.isdigit, 'aas30dsa20'))
'3020'

Question 45

s=''.join(i for i in s if i.isdigit())

Another generator variant.

Question 46

You can use filter:

filter(lambda x: x.isdigit(), "dasdasd2313dsa")

On python3.0 you have to join this (kinda ugly :( )

''.join(filter(lambda x: x.isdigit(), "dasdasd2313dsa"))

Question 47

along the lines of bayer’s answer:

''.join(i for i in s if i.isdigit())

Question 48

You can easily do it using Regex

>>> import re
>>> re.sub("\D","","£70,000")
70000

Question 49

x.translate(None, string.digits)

will delete all digits from string. To delete letters and keep the digits, do this:

x.translate(None, string.letters)

Question 50

The op mentions in the comments that he wants to keep the decimal place. This can be done with the re.sub method (as per the second and IMHO best answer) by explicitly listing the characters to keep e.g.

>>> re.sub("[^0123456789\.]","","poo123.4and5fish")
'123.45'

Question 51

A fast version for Python 3:

# xx3.py
from collections import defaultdict
import string
_NoneType = type(None)

def keeper(keep):
    table = defaultdict(_NoneType)
    table.update({ord(c): c for c in keep})
    return table

digit_keeper = keeper(string.digits)

Here’s a performance comparison vs. regex:

$ python3.3 -mtimeit -s'import xx3; x="aaa12333bb445bb54b5b52"' 'x.translate(xx3.digit_keeper)'
1000000 loops, best of 3: 1.02 usec per loop
$ python3.3 -mtimeit -s'import re; r = re.compile(r"\D"); x="aaa12333bb445bb54b5b52"' 'r.sub("", x)'
100000 loops, best of 3: 3.43 usec per loop

So it’s a little bit more than 3 times faster than regex, for me. It’s also faster than class Del above, because defaultdict does all its lookups in C, rather than (slow) Python. Here’s that version on my same system, for comparison.

$ python3.3 -mtimeit -s'import xx; x="aaa12333bb445bb54b5b52"' 'x.translate(xx.DD)'
100000 loops, best of 3: 13.6 usec per loop

Question 52

Use a generator expression:

>>> s = "foo200bar"
>>> new_s = "".join(i for i in s if i in "0123456789")

Question 53

Ugly but works:

>>> s
'aaa12333bb445bb54b5b52'
>>> a = ''.join(filter(lambda x : x.isdigit(), s))
>>> a
'1233344554552'
>>>

Question 54

$ python -mtimeit -s'import re;  x="aaa12333bb445bb54b5b52"' 're.sub(r"\D", "", x)'

100000 loops, best of 3: 2.48 usec per loop

$ python -mtimeit -s'import re; x="aaa12333bab445bb54b5b52"' '"".join(re.findall("[a-z]+",x))'

100000 loops, best of 3: 2.02 usec per loop

$ python -mtimeit -s'import re;  x="aaa12333bb445bb54b5b52"' 're.sub(r"\D", "", x)'

100000 loops, best of 3: 2.37 usec per loop

$ python -mtimeit -s'import re; x="aaa12333bab445bb54b5b52"' '"".join(re.findall("[a-z]+",x))'

100000 loops, best of 3: 1.97 usec per loop

I had observed that join is faster than sub.

Question 55

You can read each character. If it is digit, then include it in the answer. The str.isdigit() method is a way to know if a character is digit.

your_input = '12kjkh2nnk34l34'
your_output = ''.join(c for c in your_input if c.isdigit())
print(your_output) # '1223434'

Question 56

Not a one liner but very simple:

buffer = ""
some_str = "aas30dsa20"

for char in some_str:
    if not char.isdigit():
        buffer += char

print( buffer )

Question 57

I used this. 'letters' should contain all the letters that you want to get rid of:

Output = Input.translate({ord(i): None for i in 'letters'}))

Example:

Input = "I would like 20 dollars for that suit" Output = Input.translate({ord(i): None for i in 'abcdefghijklmnopqrstuvwxzy'})) print(Output)

Output: 20

Question 58

How can I write a lambda expression that’s equivalent to:

def x():
    raise Exception()

The following is not allowed:

y = lambda : raise Exception()

Question 59

There is more than one way to skin a Python:

y = lambda: (_ for _ in ()).throw(Exception('foobar'))

Lambdas accept statements. Since raise ex is a statement, you could write a general purpose raiser:

def raise_(ex):
    raise ex

y = lambda: raise_(Exception('foobar'))

But if your goal is to avoid a def, this obviously doesn’t cut it. It does, however allow you to conditionally raise exceptions, e.g.:

y = lambda x: 2*x if x < 10 else raise_(Exception('foobar'))

Alternatively you can raise an exception without defining a named function. All you need is a strong stomach (and 2.x for the given code):

type(lambda:0)(type((lambda:0).func_code)(
  1,1,1,67,'|\0\0\202\1\0',(),(),('x',),'','',1,''),{}
)(Exception())

And a python3 strong stomach solution:

type(lambda: 0)(type((lambda: 0).__code__)(
    1,0,1,1,67,b'|\0\202\1\0',(),(),('x',),'','',1,b''),{}
)(Exception())

Thanks @WarrenSpencer for pointing out a very simple answer if you don’t care which exception is raised: y = lambda: 1/0.

Question 60

How about:

lambda x: exec('raise(Exception(x))')

Question 61

Actually, there is a way, but it’s very contrived.

You can create a code object using the compile() built-in function. This allows you to use the raise statement (or any other statement, for that matter), but it raises another challenge: executing the code object. The usual way would be to use the exec statement, but that leads you back to the original problem, namely that you can’t execute statements in a lambda (or an eval(), for that matter).

The solution is a hack. Callables like the result of a lambda statement all have an attribute __code__, which can actually be replaced. So, if you create a callable and replace it’s __code__ value with the code object from above, you get something that can be evaluated without using statements. Achieving all this, though, results in very obscure code:

map(lambda x, y, z: x.__setattr__(y, z) or x, [lambda: 0], ["__code__"], [compile("raise Exception", "", "single"])[0]()

The above does the following:

the compile() call creates a code object that raises the exception;
the lambda: 0 returns a callable that does nothing but return the value 0 — this is used to execute the above code object later;
the lambda x, y, z creates a function that calls the __setattr__ method of the first argument with the remaining arguments, AND RETURNS THE FIRST ARGUMENT! This is necessary, because __setattr__ itself returns None;
the map() call takes the result of lambda: 0, and using the lambda x, y, z replaces it’s __code__ object with the result of the compile() call. The result of this map operation is a list with one entry, the one returned by lambda x, y, z, which is why we need this lambda: if we would use __setattr__ right away, we would lose the reference to the lambda: 0 object!
finally, the first (and only) element of the list returned by the map() call is executed, resulting in the code object being called, ultimately raising the desired exception.

It works (tested in Python 2.6), but it’s definitely not pretty.

One last note: if you have access to the types module (which would require to use the import statement before your eval), then you can shorten this code down a bit: using types.FunctionType() you can create a function that will execute the given code object, so you won’t need the hack of creating a dummy function with lambda: 0 and replacing the value of its __code__ attribute.

Question 62

Functions created with lambda forms cannot contain statements.

Question 63

If all you want is a lambda expression that raises an arbitrary exception, you can accomplish this with an illegal expression. For instance, lambda x: [][0] will attempt to access the first element in an empty list, which will raise an IndexError.

PLEASE NOTE: This is a hack, not a feature. Do not use this in any (non code-golf) code that another human being might see or use.

Question 64

I’d like to give an explanation of the UPDATE 3 of the answer provided by Marcelo Cantos:

type(lambda: 0)(type((lambda: 0).__code__)(
    1,0,1,1,67,b'|\0\202\1\0',(),(),('x',),'','',1,b''),{}
)(Exception())

Explanation

lambda: 0 is an instance of the builtins.function class.
type(lambda: 0) is the builtins.function class.
(lambda: 0).__code__ is a code object.
A code object is an object which holds the compiled bytecode among other things. It is defined here in CPython https://github.com/python/cpython/blob/master/Include/code.h. Its methods are implemented here https://github.com/python/cpython/blob/master/Objects/codeobject.c. We can run the help on the code object:

Help on code object:

class code(object)
 |  code(argcount, kwonlyargcount, nlocals, stacksize, flags, codestring,
 |        constants, names, varnames, filename, name, firstlineno,
 |        lnotab[, freevars[, cellvars]])
 |  
 |  Create a code object.  Not for the faint of heart.

type((lambda: 0).__code__) is the code class.
So when we say

type((lambda: 0).__code__)(
    1,0,1,1,67,b'|\0\202\1\0',(),(),('x',),'','',1,b'')

we are calling the constructor of the code object with the following arguments:

argcount=1
kwonlyargcount=0
nlocals=1
stacksize=1
flags=67
codestring=b’|\0\202\1\0′
constants=()
names=()
varnames=(‘x’,)
filename=”
name=”
firstlineno=1
lnotab=b”

You can read about what the arguments mean in the definition of the PyCodeObject https://github.com/python/cpython/blob/master/Include/code.h. The value of 67 for the flags argument is for example CO_OPTIMIZED | CO_NEWLOCALS | CO_NOFREE.

The most importand argument is the codestring which contains instruction opcodes. Let’s see what they mean.

>>> import dis
>>> dis.dis(b'|\0\202\1\0')
          0 LOAD_FAST                0 (0)
          2 RAISE_VARARGS            1
          4 <0>

The documentation of opcodes can by found here https://docs.python.org/3.8/library/dis.html#python-bytecode-instructions. The first byte is the opcode for LOAD_FAST, the second byte is its argument i.e. 0.

LOAD_FAST(var_num)
    Pushes a reference to the local co_varnames[var_num] onto the stack.

So we push the reference to x onto the stack. The varnames is a list of strings containing only ‘x’. We will push the only argument of the function we are defining to the stack.

The next byte is the opcode for RAISE_VARARGS and the next byte is its argument i.e. 1.

RAISE_VARARGS(argc)
    Raises an exception using one of the 3 forms of the raise statement, depending on the value of argc:
        0: raise (re-raise previous exception)
        1: raise TOS (raise exception instance or type at TOS)
        2: raise TOS1 from TOS (raise exception instance or type at TOS1 with __cause__ set to TOS)

The TOS is the top-of-stack. Since we pushed the first argument (x) of our function to the stack and argc is 1 we will raise the x if it is an exception instance or make an instance of x and raise it otherwise.

The last byte i.e. 0 is not used. It is not a valid opcode. It might as well not be there.

Going back to code snippet we are anylyzing:

type(lambda: 0)(type((lambda: 0).__code__)(
    1,0,1,1,67,b'|\0\202\1\0',(),(),('x',),'','',1,b''),{}
)(Exception())

We called the constructor of the code object:

type((lambda: 0).__code__)(
    1,0,1,1,67,b'|\0\202\1\0',(),(),('x',),'','',1,b'')

We pass the code object and an empty dictionary to the constructor of a function object:

type(lambda: 0)(type((lambda: 0).__code__)(
    1,0,1,1,67,b'|\0\202\1\0',(),(),('x',),'','',1,b''),{}
)

Let’s call help on a function object to see what the arguments mean.

Help on class function in module builtins:

class function(object)
 |  function(code, globals, name=None, argdefs=None, closure=None)
 |  
 |  Create a function object.
 |  
 |  code
 |    a code object
 |  globals
 |    the globals dictionary
 |  name
 |    a string that overrides the name from the code object
 |  argdefs
 |    a tuple that specifies the default argument values
 |  closure
 |    a tuple that supplies the bindings for free variables

We then call the constructed function passing an Exception instance as an argument. Consequently we called a lambda function which raises an exception. Let’s run the snippet and see that it indeed works as intended.

>>> type(lambda: 0)(type((lambda: 0).__code__)(
...     1,0,1,1,67,b'|\0\202\1\0',(),(),('x',),'','',1,b''),{}
... )(Exception())
Traceback (most recent call last):
  File "<stdin>", line 3, in <module>
  File "", line 1, in 
Exception

Improvements

We saw that the last byte of the bytecode is useless. Let’s not clutter this complicated expression needlesly. Let’s remove that byte. Also if we want to golf a little we could omit the instantiation of Exception and instead pass the Exception class as an argument. Those changes would result in the following code:

type(lambda: 0)(type((lambda: 0).__code__)(
    1,0,1,1,67,b'|\0\202\1',(),(),('x',),'','',1,b''),{}
)(Exception)

When we run it we will get the same result as before. It’s just shorter.

问题：如何在Python中逐行读取文件？

回答 0

奖金问题

Bonus Question

回答 1

回答 2

回答 3

问题：我可以使用`pip`代替`easy_install`来实现`python setup.py install`依赖关系解析吗？

回答 0

从网络上的tarball安装

从本地tarball安装

从本地文件夹安装

从本地文件夹安装（可编辑模式）

Install from tarball on web

Install from local tarball

Install from local folder

Install from local folder (editable mode)

回答 1

回答 2

问题：如何执行两个列表的按元素相乘？

回答 0

回答 1

回答 2

回答 3

回答 4

回答 5

回答 6

回答 7

回答 8

回答 9

回答 10

回答 11

回答 12

回答 13

问题：将int转换为ASCII并返回Python

回答 0

回答 1

回答 2

回答 3

回答 4

问题：我如何让Pyflakes忽略声明？

回答 0

回答 1

回答 2

回答 3

回答 4

passive_pyflakes.py

bypass_pyflakes.py

回答 5

回答 6

问题：我应该使用scipy.pi，numpy.pi还是math.pi？

回答 0

回答 1

问题：使用Python从字符串中删除数字以外的字符？

回答 0

回答 1

回答 2

回答 3

回答 4

回答 5

回答 6

回答 7

回答 8

回答 9

回答 10

回答 11

回答 12

回答 13

回答 14

问题：定义引发异常的lambda表达式

回答 0

回答 1

回答 2

回答 3

回答 4

回答 5

说明

改进措施

Explanation

Improvements