Stabby McDamage posted:

I'm surprised nobody's posted the underlying problem.

It's common to all floating-point arithmetic. Basically, as with integers, the number is stored using a binary representation.

Avenging Dentist posted:

I know why it's that way but it's beyond retarded that they don't check for it in the assignment, especially since it's trivially easy (A[3::-1].base is A).

Modern Pragmatist posted:

Do you mind elaborating why this is the case? Is it because you don't want to unnecessarily copy the entire array in memory? What benefits does the current way have?

UberJumper posted:

a. comtypes people get pissy, and can they basically tell me to take it down?

Scaevolus posted:

It doesn't really matter. It's distributed under the MIT License, so they can't prevent you from modifying it to work for your particular case.

Lurchington posted:

In the past I've used template strings to do stuff like this.

you could then have the option of making your wrapper something like:

code:

def my_match(regex, s):
   your_subexpression_dict = {"THING1": <your_regex1>, "THING2": <your_regex2>}
   regex_template = Template(regex)
   return re.match(regex_template.safe_substitute(your_subexpression_dict), s)

which would allow you to write your expressions as:

code:

my_match('$THING1,\s*$THING2', s)

UberJumper posted:

thanks!

Thermopyle posted:

Concise method of creating a new list from every nth item of a list?

I seem to write multiple lines of code for this and it seems kinda ridiculous.

Example:

Every 3rd item from list a makes list b.

a = [1, 2, 3, 4, 5, 6, 7, 8, 9]
b = [3, 6, 9]

UberJumper posted:

b = [x for x in a[::3]]

Thermopyle posted:

I was not aware of that syntax with slicing. Thanks guys.

Another question:

alist = ['a', 'b', 'c', 'd', 'e', 'f']

Given an index, let's say 2, what's a concise, yet readable, way to create a list that begins with alist[3] and wraps around to alist[2], so I end up with:

blist = ['d', 'e', 'f', 'a', 'b', 'c']

Thermopyle posted:

alist = ['a', 'b', 'c', 'd', 'e', 'f']

Given an index, let's say 2, what's a concise, yet readable, way to create a list that begins with alist[3] and wraps around to alist[2], so I end up with:

blist = ['d', 'e', 'f', 'a', 'b', 'c']

king_kilr posted:

blist = alist[idx:] + alist[:idx]

Thermopyle posted:

God, I'm an idiot. I think I've stayed up too late.

Captain Capacitor posted:

Don't feel too bad. List comprehensions are powerful and mysterious creatures. They're a feature I miss quite a bit when switching to other languages.

dv6speed posted:

I now decided to try interfacing a 2x16 Character LCD module to my parallel port. This is my first program dealing with ASCII to binary conversion, in addition to accepting command line arguments.

Any newbish mistakes I should know about?

Stabby McDamage posted:

Avenging Dentist posted:

code:

def first_N(n):
    assert n >= 0, 'n must be a positive integer'
    return [i**2 for i in xrange(1, 2*n, 2)]

A perfect square is odd iff its square root is odd. (I'd give hints but it's a one-liner so it's hard to hint at, really.)

EDIT: here's how you do it without list comprehensions:

code:

def first_N(n):
    assert n >= 0, 'n must be a positive integer'
    perf_squares = []
    for i in xrange(1, 2*n, 2):
        perf_squares.append(i**2)
    return perf_squares

dv6speed posted:

Awesome stuff, thanks!

Avenging Dentist posted:

code:

def first_N(n):
    assert n >= 0, 'n must be a positive integer'
    return [i**2 for i in xrange(1, 2*n, 2)]

A perfect square is odd iff its square root is odd.

RobotEmpire posted:

devilmouse posted:

Google is failing me (or rather, it's just cluttered up with old old no longer maintained code). Anyone know a serviceable package for editing video files in Python? There was a promising looking thing called DVEdit, but it's nowhere to be found. PyMedia hasn't been touched since '06.

I'm looking to open up some AVI files, copy some audio/video from certain timeslices into a new file and mix in some audio over it. Just some really primitive video-editing.

Pyglet is a possibility, but the A/V support seems pretty limited from a brief look.

Am I missing some sweet secret media package?

Scaevolus posted:

This uses a generator expression:

code:

for x in (n ** 2 for n in xrange(1000)):
    pass

The above will create two iterators-- objects that yield values every time next() is called on them-- that don't store every value like the lists do, just the values that are necessary for the moment.

xrange() returns an iterator. range returns a list.

Scaevolus posted:

This uses a generator expression:

code:

for x in (n ** 2 for n in xrange(1000)):
    pass

RobotEmpire posted:

So I assume a generator expression creates a tuple vice a list? Or are the outer parens for grouping?

mystes posted:

The python bindings for gstreamer may be your best bet. It's too bad there isn't any good up-to-date ffmpeg module since that was my first thought too.

RobotEmpire posted:

So I assume a generator expression creates a tuple vice a list? Or are the outer parens for grouping?

nbv4 posted:

All though, I think that syntax is 2.6/3.x only.

Avenging Dentist posted:

This is really important: an expression inside parens but without commas (commas inside other parens don't count) aren't tuples. You have to type (foo,) for a one-element tuple. Also function calls aren't tuples either.

LuckySevens posted:

Stupid python question, I'm sorting a 2x20 list with a string and a % integer, and I want to sort by the absolute value and put the highest 5 entries into a list. I suck at this but this is what I came up with:

code:

def bigmovers():
    list = __request('ST')
    change = []
    for m in list:
        change.append([m[2], m[20]])
    sorted(change, key=lambda stocks: stocks[1])
    return change

Now this sorts by value, negative and positive, but not absolute value. It won't let me just add abs(stocks[1]) to the lambda like I thought it might in my head, obviously for a good reason. Any ideas?