Programming in Python: selected questions and answers.

Updated frequently, refresh your browser.

From the course, http://staff.washington.edu/jon/python-course/

Here are answers to some of the questions asked in the weekly reports
or in personal email.  Some other questions will be answered in class
(perhaps as part of a lecture).  Not all questions will
be answered, due to limited time.  This does not reflect on the
worthiness of the unanswered questions.

Downey's solutions to some of his textbook exercises:

 http://greenteapress.com/thinkpython/code/

Other collections of Python questions and answers:

 http://docs.python.org/faq/programming

 http://mail.python.org/pipermail/tutor/

 http://stackoverflow.com/questions/tagged/python

Each question and its answer are numbered:  Q.2.1 and A.2.1 etc.
All Q.2 and A.2 are from week 2, etc., so you can search for a given week.


Dec 11 (week 9)

Q.9.4: I have a list of lists, say 'mylist', and I am interested in
sorting it, I use "mylist.sort()" and it works well but it sorts by
the first index.

For example, let
mylist =[[1,2],[3,1],[2,5],[8,1],[4,2]], then "mylist.sort()" sorts
mylist by the first index:
[[1, 2], [2, 5], [3, 1], [4, 2], [8, 1]].

How can I sort mylist by the second index?

A.9.4 You can use the optional 'key' keyword parameter to specify or
define a key function.  The list is sorted such that the sorted list
is ordered by the values obtained by applying the key function to each
list element (but the original list elements appear in the sorted list
also). Here the key function should return the second element of each
tuple, we define it with a lambda expression:

mylist.sort(key = lambda t: t[1])

Likewise, if you have a list of objects where each object has an
attribute named 'i' you sort by the value of this attribute this way:

objectlist.sort(key = lambda t: t.i)

The 'key' and other sort parameters are explained in more detail here:

Sorting HOW TO
http://docs.python.org/howto/sorting.html

See also sort.py at the class website.

Q.9.3:  2. In the vector.py example, you used

 super(vector,self).__repr__()

What is the meaning of "super(vector,self)"?

A.9.3: It refers to the superclass or base class, that is, list.  In this code:

'vector(%s)' % super(vector,self).__repr__()

this has the effect of calling list.__repr__(), then inserting the
result in place of the %s in 'vector(%s)'


Q.9.2: In the function def f(**kwargs) what is the underlying behavior of ** ?

A.9.2: It gathers multiple keyword arguments into a dictionary:

 def f(**kwargs): print kwargs 
 
Then f('a'=1, 'b'=2) prints { 'a':1, 'b':2 }

You can also use ** in a function invocation to scatter a dictionary
to positional parameters:

 def g(a,b,c): print a,b,c
 d = { 'c':3, 'b':2, 'a':1 }

Then g(d) prints 1,2,3

See also parameters.py on the course website.


Q.9.1: Since Python isn't a statically-typed language, is it a
recommended practice to put asserts or other forms of checking into
functions that implicitly expect certain types (or is try/except
preferred)?

A.9.1: I don't recall seeing many asserts or checks in function
bodies.  Sometimes there is a try/except around the function *call*
(not in the function body).

So typical Python style seems to be to omit any checks in the function
itself, leaving the caller to use try/except if they wish.

Note that in Python, the type of the argument is not what you really
need to know.  The important thing is whether the argument provides
the operations that the function body uses, and the type doesn't tell
you that (due to duck typing).  For example, your function body might
use +, so the argument has to support that, but the argument might not
even be a numeric type, it might be a list, or it might belong to some
class you never heard of where the programmer defined an __add__
method, that might make sense in your function, even if you didn't
anticipate it.

In my own code, I sometimes do put asserts at the beginning of
function bodies, to check types but also to check other conditions.  I
think of it as documentation --- it's no more difficult to write an
assert statement than write a comment.


Dec 5 (week 8)

Q.8.1: Is there some way to enforce that the objects (that belong to the same class) all have the same data members?

A.8.1:  Yes, use the __slots__ attribute.  See in

http://docs.python.org/reference/datamodel.html#special-method-names

and use a search engine to find examples.

Nov 15 (Week 5)

Q.5.3: (Install problems, versions and dependencies) It is not always
straightforward to figure out what versions of Python that third-party
packages such as NumPy requiire.  It turns out that NumPy works with
2.6.x.  Do you have any suggestions for dealing with these issues,
other than trial-and-error?

A.5.3: For some operating systems there are package managers that
automatically keep track of dependencies, for example Synaptic/apt for
Ubuntu and other Debian-like Linux distributions, or Macports for Mac
OS X.  IF your operating system has a package manager and IF the
package you want is known to the package manager, then when you
install a package it finds all the dependencies (including earlier
Python versions) and (optionally) installs them too. I'm not aware of
anything like this for Windows, unfortunately.


Nov 14 (Week 5)

Q.5.2: (Unicode) I copied and pasted the poem about Eric the half a bee
into a text file ... on output, the word D'you is rendered
'd\xe2\x80\x99you'. \xe2\x80\x99 is unicode for the forward-quote character.
How to deal with this?

A.5.2: I recommend avoiding the whole unicode issue by just converting
the file to ASCII by hand.  Just load the poem into your favorite
editor, change that word to D'you or D`you, using the single quote or
backquote key on your keyboard.  Then save that, now it should be
ASCII only.

This unicode/character encoding issue is complicated.  I think it's
only worth getting into if you need to work with a lot of foreign
language text.  If you do go that route, here are some pertinent
links:

 Unicode HOWTO
 http://docs.python.org/howto/unicode.html

 Unicode in Python, completely demystified
 http://farmdev.com/talks/unicode/

 How to Use UTF-8 with Python (with several more pertinent links at the end)
 http://evanjones.ca/python-utf8.html


Nov 13 (Week 5)

Q.5.1: What is the best way to copy a list? 

A.5.1: The unobvious but canonical solution is t = s[:].  You assign
the slice from the beginning to the end to the new list!  Let's try
it, confirm the new list t doesn't share contents with the original s:

>>> s = [ 1, 2, [3,4], 5]
>>> t = s[:]
>>> t
[1, 2, [3, 4], 5]
>>> s == t
True
>>> s is t
False
>>> s[1] = 99
>>> s == t
False
>>> s
[1, 99, [3, 4], 5]
>>> t
[1, 2, [3, 4], 5]

So it looks like we're not sharing.  BUT ---

>>> s[2][1] = 88
>>> s
[1, 99, [3, 88], 5]
>>> t
[1, 2, [3, 88], 5]

so we're still sharing the nested list.  t = s[:] only works for flat
(unnested) lists.  To copy nested lists we need a recursive function.
A general solution is provided by the copy module's deepcopy function,
which handles all nested structures, not just lists.


Nov 9 (Week 4)

Q.4.5: What is PYTHONPATH?  How do I set it up?

A.4.5: PYTHONPATH is an environment variable that the Python
interpreter uses to search for modules that you import.  When Python
executes 'import mymodule' it looks for a file named mymodule.py
... where?  It looks in the current working directory and in certain
special directories, which is often sufficient without you having to
do anything more.  But if you have modules in other directories that
you want Python to search, you can add to the search path by defining (or
redefining) PYTHONPATH at your system's command prompt (not the python
interpreter prompt).  The command is system dependent.  Here are some
examples (from one of my projects, the exact path contents would
be different for you):

For Windows (in a cmd window):

set PYTHONPATH=%PYTHONPATH%;C:\Users\jon\Documents\repy_v2\repy;C:\Users\jon\Doc
uments\repy_v2\portability

For Unix/Linux/Mac (using the bash shell):

export PYTHONPATH=$PYTHONPATH:/Users/jon/repy_v2/repy:/Users/jon/repy_v2/portabi
lity

To show the path that python actually uses in a session:

>>> import sys
>>> sys.path
...

If you assigned PYTHONPATH using a command like one of the above, you
should see the directories you named in the output here.


Q.4.4: How to "slice" a word by a character?  How to "slice" a file by
backspaces?

A.4.4: You probably want the string split (not slice) method.  It
takes a string and a separator string (which can be one character but
needn't be) and returns a list of strings.  For example, to split a
string by spaces:

>>> s = 'abc def'
>>> s.split(' ')
['abc', 'def']

The backspace character is indicated '\b'.  You can read an entire
file into a string using read(), so something like this should work:

>>> s = open('myfile.txt').read()
>>> s.split('\b')
...

Q.4.3: Is there any way to reload a module into a running Python session
(after editing it to fix an error)?

A.4.3: Not really.  You can try reload(<module name>), or repeat import
*then* reload, but these do not always have the intended effect.  The
most certain way is to just start a new Python session.  That's what
IDLE does.


Nov 6 (Week 4)

Q.4.2: Are the solutions to the textbook exercises available anywhere?

A.4.2: Some solutions are at http://greenteapress.com/thinkpython/code/


Q.4.1: What about Python 2.x vs. 3.x  ?

A.4.1: Most big libraries (NumPy, SciPy, ....) still use 2.x so I'm
sticking with that for now.  I don't have any experience with 3.x.
These pages may be helpful:

 What's New In Python 3.0
 http://docs.python.org/release/3.0.1/whatsnew/3.0.html
 
 On Teaching Programming With Python 3.0
 http://www.comp.leeds.ac.uk/nde/papers/teachpy3.html

 Ask HN: start with python 2.6 or 3?
 http://news.ycombinator.com/item?id=687087


Q.4.0: Why does the expression 'None or False' return False, but 
'False or None' does not return anything?

A.4.0. Actually, 'False or None' returns 'None', but the interpreter
prints nothing (try typing just None to the interpreter) .  'or'
always returns one of its arguments, see A.3.3 below.


Week 3

Q.3.0: What are some other sources of Python exercises and projects
(besides the Downey textbook)?  Also, what are some tutorials that use 
less math than Downey?

A.3.0: Here are a few I've heard of.  I haven't tried them, so these are not
recommendations, they are just for your information.  Google python
exercises or prthon projects turns up a lot more.

 Learning to Program by Alan Gauld (no exercises)
 http://www.freenetpages.co.uk/hp/alan.gauld/

 A Python Course (includes exercises, might use less math than Downey)
 http://www.upriss.org.uk/python/PythonCourse.html

 Basic Python Exercises (other exercises linked to list at lower left)
 http://code.google.com/edu/languages/google-python-class/exercises/basic.html

 The Python Challenge (all exercises, treasure-hunt format)
 http://www.pythonchallenge.com/ 

 Project Euler (math exercises, challenging)
 http://projecteuler.net/  


Q.3.1: How to represent enumerated types in Python?

A.3.1: If you just need symbols, you could use literal strings:  

 if color == 'green': ...

If enumerations must have numeric values a la C you could assign variables:

 red, yellow, green = range(3)

This is equivalent to the tuple assignment red, yellow, green = 0, 1,
2 which is equivalent to the sequence of assignments red = 0; yellow =
1; green = 2

If you need type checking a la C# you could define a class Color with
attributes Color.red etc.


Q.3.2: How to do type checking of function arguments?

A.3.2: You could use assert with isinstance:

    def f(i):
       assert isinstance(i, int)  # raise AssertionError if i is not int
       return i + 1


Q.3.3: Does 'and' evaluate both of its arguments?

A.3.3: Not always - it uses "short circuit evaluation" which returns
as soon as it has the answer.
From http://docs.python.org/reference/expressions.html#boolean-operations

  The expression x and y first evaluates x; if x is false, its value is
  returned; otherwise, y is evaluated and the resulting value is
  returned. ...

  (Note that neither and nor or restrict the value and type they return
  to False and True, but rather return the last evaluated argument. This
  is sometimes useful. ...

Recall that any value can be interpreted as true or false; for
example, nonzero numbers and nonempty collections are true, etc.


Week 2

Q.2.0: Is there a computerized means of converting source code from
Python 2 to Python 3?

A.2.0:  Yes, it's called 2to3: http://docs.python.org/py3k/library/2to3.html


Q.2.1:  What's this?
>>> 'Hello', 'World'
('Hello', 'World')

A.2.1: It's a tuple. The comma operator ',' creates a tuple.  You have
created a tuple of two strings.  Tuples are discussed in Ch. 12 in the
textbook, which we will cover in week 5.


Q.2.2: What is the syntactic significance of this result?
>>> type(17)
<type 'int'>

A.2.2 This is just how the Python interpreter prints out the the int
(integer) type.  Also:

>>> int
<type 'int'>


Week 1

Q.1.2: My question is the need to account for an additional space when using
the print statement with a comma. Is there a way to avoid extra space
when printing multiple variables separated by commas?

A.1.2: The usual way to get more control over printing is to use format
strings.  So, for example, to print the values of x and y with no
spaces between, this should work:

print '%s%s' % (x, y)

In the textbook, the index has an entry 'format strings' but the book
barely explains them.
The Python 2.6 Quick Reference linked to the course home page (in the
Readings section) has more, and the Python Tutorial linked in that
same section also explains more in section 7.1
(what I recommend above is Old String Formatting, section 7.1.1).


Q.1.3: I found in basics.py:

if x:
   print x, ' indicates True'
else:
   print x, ' indicates False'

I saved this code in a file and when I tried to run it, it gave me
this error message:

Traceback (most recent call last):
File "C:/Python26/week1.py", line 1, in <module>
   if x:
NameError: name 'x' is not defined

A.1.3:  In your module you need to define x before you use it, for example by
an assignment statement like this:

x = 'Hello'

Notice that this line does appear in my basics.py.  Put this line (or
something like it) before the if statement in your file.py and it
should work.