[Tutor] map vs. list comprehension
I read somewhere that the function 'map' might one day be deprecated in favor of list comprehensions. But I can't see a way to do this in a list comprehension: >>> map (pow, [2, 2, 2, 2], [1, 2, 3, 4]) [2, 4, 8, 16] Is there a way? Cheers, Mike ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor
Re: [Tutor] map vs. list comprehension
On Feb 14, 2006, at 3:46 PM, Andre Roberge wrote: > [2**i for i in [1, 2, 3, 4]] Ah yes, I'm sorry, I was thinking of the most general case, where the arguments are two arbitrary lists. My example was too specific. Is there a way to do something like the following in a list comprehension? map(pow, [1, 7, 6, 2], [3, 8, 2, 5]) Mike ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor
[Tutor] Unexpected behavior of +=
I just discovered the following behavior, but can't find any documentation about it: >>> list = [] >>> list = list + 'abc' Traceback (most recent call last): File "", line 1, in ? TypeError: can only concatenate list (not "str") to list but: >>> list = [] >>> list += 'abc' >>> list ['a', 'b', 'c'] Is this a special characteristic that has been added to the augmented assignment operator +=; or is it an automatic consequence of += assignment being performed'in place'? (Tho I can't see how it could be...) It just seems very un-Pythonesque to be able to successfully concatenate objects of different types like this. And it seems very inconsistent with standard assignment. Indeed, the Python Reference Manual, section 6.3.1 states: "With the exception of assigning to tuples and multiple targets in a single statement, the assignment done by augmented assignment statements is handled the same way as normal assignments. Similarly, with the exception of the possible in-place behavior, the binary operation performed by augmented assignment is the same as the normal binary operations." ...which is patently not the case here. I was scandalized lol! ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor
[Tutor] 'in-place' methods
I think I understand this sorting-a-list 'in place' stuff, and things of that kind (reversing for example); but I am finding it very difficult to get used to, since sorting a list doesn't return the sorted list as a value, but simply does the work as a side effect. The place where it really trips me up is when I want to do something like pass a sorted list as a value in a program, say. Here is a trivial example. Say I want to define a function to always print lists in sorted form. I can't do it as follows, tho it's what I want to do intuitively: def sort_print(L): print L.sort() This is on a strict analogy to writing a function that prints strings in upper case only: def upper_print(s): print s.upper() (The fact that the second works and the first doesn't really does bug me as a newbie.) Anyway, first question: is the fact that the first doesn't work purely a function of the fact that lists are mutable? (At least I could kind of understand that, that methods work differently for objects of different types). Or is it possible to have methods like .upper() that return the result of performing the operation even for lists? Second question. Do I really have to write the sort_print() function like this: def sort_print(L): L.sort() print L i.e. first perform the operation in-place, then pass the variable? Is this the idiomatic way of doing it? Third question: is this in-place behavior of methods in effect restricted to lists, or are there other places I should be on the lookout for it? Cheers, Mike ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor
[Tutor] Unicode and regexes
Does Python support the Unicode-flavored class-specifications in
regular expressions, e.g. \p{L} ? It doesn't work in the following
code, any ideas?
-
#! /usr/local/bin/python
""" usage: ./uni_read.py file
"""
import codecs
import re
text = codecs.open(sys.argv[1], mode='r', encoding='utf-8').read()
unicode_property_pattern = re.compile(r"\p{L}")
dot_pattern = re.compile(".")
letters = unicode_property_pattern.findall(text)
characters = dot_pattern.findall(text)
print 'var letters =', letters
print 'var characters = ', characters
-
The input file, encoded in utf-8 is
abc
The output is:
var letters = []
var characters = [u'a', u'b', u'c', u' ', u'\u03b1', u'\u03b2',
u'\u03b3']
___
Tutor maillist - Tutor@python.org
http://mail.python.org/mailman/listinfo/tutor
Re: [Tutor] Unicode and regexes
Thanks Kent, for breaking the bad news. I'm not angry, just terribly,
terribly disappointed. :)
"From http://www.unicode.org/unicode/reports/tr18/ I see that \p{L} is
intended to select Unicode letters, and it is part of a large number of
selectors based on Unicode character properties."
Yeah, that's the main cite, and yeah, a large, large number. The only
sane way to use regexes with Unicode. Also see Friedl's 'Mastering
Regular Expressions' Chapter 3: or actually, if you are a Python only
person, don't: it will make you weep.
"Python doesn't support this syntax. It has limited support for
Unicode character properties [...]".
Umm Earth to Python-guys, you *have heard* of Unicode, right? Call me
crazy, but in this day and age, I assume a scripting language with
regex support will implement standard Unicode conventions, unless
there is a compelling reason not to. Very odd.
Back to Perl. Right now. Just kidding. Not. Sheesh. This is a big
flaw in Python, IMHO. I never saw it coming.
___
Tutor maillist - Tutor@python.org
http://mail.python.org/mailman/listinfo/tutor
[Tutor] Bigrams and nested dictionaries
I'm playing with the whole idea of creating bigram (digram?)
frequencies for text analysis and cryptographic and entropy analysis
etc (this is as much an exercise in learning Python and programming
as anything else, I realise everything has already been done
somewhere somehow :) Though I *am* aiming to run this over unicoded
phonetic representations of natural languages, which doesn't seem to
be that common.
I implemented a single character count using a dictionary, and it
worked really well:
Character_Count = {}
for character in text:
Character_Count[character] = Character_Count.get(character, 0) + 1
And so for bigrams I was thinking of creating a data-structure that
was a nested dictionary, where the key was the character in position
1, and the value was a sub-dictionary that gave a count of each
character in position 2.
So the dictionary might begin something like:
{'a': {'a':1, 'b':8, 'c':10,...}, 'b' : {'a':23, 'b':0, 'c':
1,...},..., 'z' : {'a':11, 'b':0, 'c':0,...}}
The count of the character in position one could be retrieved by
summing over the characters in position 2, so the bigram and single-
character counts can be done in one pass.
I don't want anyone to tell me how to do this :) I'd just like to get
a feel for:
(i) is this idea of a nested dictionary a good/efficient/tractable
data-structure?
(ii) is there a straightforward path to the goal of constructing this
thing in a single pass over the input string (yes or no will suffice
for the moment!), or is there a can of worms lurking in my future.
I'd rather have less information than more, but there is something
about grabbing characters two-at-a-time, but moving forward one-at-a-
time, and sticking the count down a level inside the dictionary
that's just a little baffling at the moment. On the other hand it's
like moving a two-character window across the text and recording
information as you go, which seems like it should be a good thing to
do computationally. I like being baffled, I'd just kinda like to know
if this is a good problem to work on to gain enlightenment, or a bad
one and I should think about a totally different path with a less
jazzy data structure.
___
Tutor maillist - Tutor@python.org
http://mail.python.org/mailman/listinfo/tutor
Re: [Tutor] Bigrams and nested dictionaries
Well I ran into an interesting glitch already. For a dictionary D, I
can pull out a nested value using this syntax:
>>> D['b']['a']
23
and I can assign to this dictionary using
>>> D['d'] = {'a':7, 'b':0'}
but I can't assign like this:
>>> D['d']['c'] = 1
TypeError: object does not support item assignment.
hmmm.
___
Tutor maillist - Tutor@python.org
http://mail.python.org/mailman/listinfo/tutor
Re: [Tutor] Bigrams and nested dictionaries
Aha! John wrote:
"Are you sure you haven't mistakenly assigned something other than a
dict to D or D['d'] ?"
Thanks for the tip! Yup that was it (and apologies for not reporting
the problem more precisely). I hadn't initialized the nested
dictionary before trying to assign to it. (I think Perl doesn't
require initialization of dictionaries prior to assignment, which in
this case, would be a nice thing...)
>>> D['a'] = {'a':1, 'b':2}#oops
Traceback (most recent call last):
File "", line 1, in ?
NameError: name 'D' is not defined
>>> D = {}
>>> D['a'] = {'a':1, 'b':2}
>>> D
{'a': {'a': 1, 'b': 2}}
>>> D['c']['a'] = 1 #ooops
Traceback (most recent call last):
File "", line 1, in ?
KeyError: 'c'
>>> D['c'] = {}
>>> D['c']['a'] = 1
>>> D
{'a': {'a': 1, 'b': 2}, 'c': {'a': 1}}
And Kent wrote:
"Encapsulating this in a generator is probably a good plan."
Yay, I get to play with generators... thanks for the suggestion, I
would never have looked in that direction.
--
Python: [x for S in L for x in S]
Mathematica: Flatten[L] (but where's the fun in that?)
___
Tutor maillist - Tutor@python.org
http://mail.python.org/mailman/listinfo/tutor
Re: [Tutor] Bigrams and nested dictionaries
Well coming up with this has made me really love Python. I worked on
this with my online pythonpenpal Kyle, and here is what we came up
with. Thanks to all for input so far.
My first idea was to use a C-type indexing for-loop, to grab a two-
element sequence [i, i+1]:
dict = {}
for i in range(len(t) - 1):
if not dict.has_key(t[i]):
dict[t[i]] = {}
if not dict[t[i]].has_key(t[i+1]):
dict[t[i]][t[i+1]] = 1
else:
dict[t[i]][t[i+1]] += 1
Which works, but. Kyle had an alternative take, with no indexing, and
after we worked on this strategy it seemed very Pythonesque, and ran
almost twice as fast.
#!/usr/local/bin/python
import sys
file = open(sys.argv[1], 'rb').read()
# We imagine a 2-byte 'window' moving over the text from left to right
#
# +---+
# L o n | d o | n . M i c h a e l m a s t e
r m ...
# +---+
#
# At any given point, we call the leftmost byte visible in the window
'L', and the
# rightmost byte 'R'.
#
# +---+
# L o n | L=d R=o | n . M i c h a e l m a s t
e r m ...
# +---+
#
# When the program begins, the first byte is preloaded into L, and we
position R
# at the second byte of the file.
#
dict = {}
L = file[0]
for R in file[1:]: # move right edge of window across the file
if not L in dict:
dict[L] = {}
if not R in dict[L]:
dict[L][R] = 1
else:
dict[L][R] += 1
L = R # move character in R over to L
# that's it. here's a printout strategy:
for entry in dict:
print entry, ':', sum(dict[entry].values())
print dict[entry]
print
___
Tutor maillist - Tutor@python.org
http://mail.python.org/mailman/listinfo/tutor
