Re: [Tutor] Stack unwind using exceptions.
Kent Johnson wrote: [..] Why not just return the value from the function and pass it up the call chain? If a call fails return None. Something like this: That's what I ended up doing but the first thing occurred to me and I was just wondering if there's any production code that relies on the technique. -- ~noufal http://nibrahim.net.in/ ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor
Re: [Tutor] reading variables in a data set?
On Sat, Jul 4, 2009 at 12:09 PM, Steven Buckbucks...@gmail.com wrote: I've used a module (StataTools) from (http://presbrey.mit.edu/PyDTA ) to get a Stata .dta file into Python. In Stata the data set is an NXK matrix where N is the number of observations (households) and K is the number of variables. I gather it's now a list where each element of the list is an observation (a vector) for one household. The name of my list is data; I gather Python recognizes the first observation by: data[1] . Example, data = [X_1, X_2, X_3, . . . . , X_N] where each X_i for all i, is vector of household characteristics, eg X_1 = (age_1, wage_1, . . . , residence_1). I also have a list for variable names called varname; although I'm not sure the module I used to extract the .dta into Python also created a correspondence between the varname list and the data list--the python interpreter won't print anything when I type one of the variable names, I was hoping it would print out a vector of ages or the like. varname is probably just a list of strings without any direct connection to the data. In anycase, I'd like to make a scatter plot in pylab, but don't know how to identify a variable in data (i.e. I'd like a vector listing the ages and another vector listing the wages of households). Perhaps, I need to run subroutine to collect each relevant data point to create a new list which I define as my variable of interest? From the above example, I'd like to create a list such as: age = [age_1, age_2, . . . , age_N] and likewise for wages. You can use a list comprehension to collect columns from the data. If age is the first element of each observation (index 0), and wages the second (index 1), then ages = [ observation[0] for observation in data ] wages = [ observation[1] for observation in data ] Any help you could offer would be very much appreciated. Also, this is my first time using the python tutor, so let me know if I've used it appropriately or if I should change/narrow the structure of my question. It's very helpful if you show us the code you have so far. Kent ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor
Re: [Tutor] Stack unwind using exceptions.
Noufal Ibrahim wrote: Kent Johnson wrote: [..] Why not just return the value from the function and pass it up the call chain? If a call fails return None. Something like this: That's what I ended up doing but the first thing occurred to me and I was just wondering if there's any production code that relies on the technique. I use the exception technique in my Python Pipelines parser. The recursive routine has 6 raise statements. The exceptions are caught by the program that calls the recursive routine. There would be more code and complexity to work back up the call chain. -- Bob Gailer Chapel Hill NC 919-636-4239 ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor
[Tutor] Poor style to use list as array?
The problem this time is, given an amount of US currency in cents (and less than a dollar), to print out a description of the coins needed to represent it, using the smallest number of coins. E.g.: How many cents this time? 59 2 quarters, 1 nickel and 4 pennies. How many cents this time? 55 2 quarters and 1 nickel. How many cents this time? 75 3 quarters. My program relaxes the restriction to amounts less than a dollar, and initialises the necessary data structures as follows: denom = ['dollar', 'quarter', 'dime', 'nickel', 'penny'] value = [100, 25, 10, 5, 1] LEN = 5 plural = [None] * LEN plural[-1] = 'pennies' count = [None] * LEN # Kludge I use the list 'count' as a kind of array, so that e.g. count[-1] is the number of pennies needed, count[2] is the number of dimes needed, and so on. Any initial values I might store as elements of the list are never used, hence the use of 'None'. (The use of 'None' in the list 'plural' is much more defensible, I think, as it is used to denote the formation of a regular plural by adding 's' to the name.) Instead, the actual values are assigned in a 'for' loop, as follows: for i in range(LEN - 1): (count[i], amnt) = divmod(amnt, value[i]) This feels like a bit of a cheat, as if I am trying to program in Python as if it were some other more familiar language. Should I have coded this differently? The full source code is here, in case anyone wants to look at it (but I'm not soliciting any more attention, as I've already been given quite a lot of it, and fear being offered the comfy chair!): http://python.pastebin.com/d5e321ae0 (retention: 1 day) Could I have used dictionaries instead, with the denomination names as keys? Is it possible to guarantee a sequence in which the keys of a dictionary are iterated through? (If not, I suppose I could keep the list 'denom' as it is here, and iterate through it with for key in denom:, although this seems a bit redundant.) -- Angus Rodgers ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor
Re: [Tutor] Poor style to use list as array?
Angus Rodgers wrote: [..] This feels like a bit of a cheat, as if I am trying to program in Python as if it were some other more familiar language. Should I have coded this differently? I can't offer anything concrete but here are some things that occur to me at a glance. 0. You can try to use the collections module in 2.6 (especially the namedtuple object which gives you access to things through attributes as well as through order. 1. You have 3 lists (denom, plural and count) which you will iterate through in lockstep as far as I can tell. It's an option to use a list of 3-tuples or a list of dictionaries with three keys (denomination, plural and count). Thanks. -- ~noufal http://nibrahim.net.in/ ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor
Re: [Tutor] Is my style OK in this elementary student exercise?
[Angus Rogers, suffering eval-angst] ... On the other hand, so long as I AM only executing the function myself, I am no more at risk than I already am every single time I type a command into a Python interpreter, of any description. (A somewhat Existentialist thought, perhaps! Virtual suicide is always a possibility.) - Does that seem reasonable? You've made me clearly aware of a risk that I was only vaguely aware of before (I ruminated only briefly as to whether any harm might come from entering general Python expressions, but my main feeling about this facility was that it would probably be useful - in some future exercise), but isn't there a role for functions that one can use oneself, but never ever distribute to the general public? Certainly! I use eval() and input() all the time in my private code (but never in released code), and you're never going to screw yourself by accident doing so. Especially if you have an interest in writing mathematical code, it's a tremendous convenience for prompted input to support arbitrary computation. If so, are the precautions I have suggested above sufficient? In private code, any precautions are probably just a waste of time. You really don't, for example, need to remind /yourself/ not to enter a convoluted expression that emails your credit card info to a hacker site in Nigeria. Or if you do need to remind yourself not to do things like that, you probably shouldn't be using a computer to begin with ;-) ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor
Re: [Tutor] Stack unwind using exceptions.
On Sunday 05 July 2009, Noufal Ibrahim wrote: Kent Johnson wrote: [..] Why not just return the value from the function and pass it up the call chain? If a call fails return None. Something like this: That's what I ended up doing but the first thing occurred to me and I was just wondering if there's any production code that relies on the technique. The Pyparsing library uses exceptions a lot internally. If I understood it right, exceptions are used to tell that a pattern does not match. If the pattern matches the results are transported with a regular 'return'. As it happens quite often that a pattern does not match, exceptions can be considered a regular mechanism for information transport in Pyparsing. There are two types of exceptions in Pyparsing. An exception that means: try the next pattern; and there are exceptions that mean: there was a fatal error, stop parsing. Pyparsing: http://pyparsing.wikispaces.com/ Kind regards, Eike. ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor
Re: [Tutor] Poor style to use list as array?
On Sun, Jul 5, 2009 at 2:48 PM, Angus Rodgersang...@bigfoot.com wrote: for i in range(LEN - 1): (count[i], amnt) = divmod(amnt, value[i]) How about this: counts = [] for val in value: count, amnt = divmod(amnt, val) counts.append(count) This feels like a bit of a cheat, as if I am trying to program in Python as if it were some other more familiar language. Should I have coded this differently? Generally it's more straighforward to iterate over a list directly rather than using an index. The full source code is here, in case anyone wants to look at it (but I'm not soliciting any more attention, as I've already been given quite a lot of it, and fear being offered the comfy chair!): Don't worry, we have retired the comfy chair. Could I have used dictionaries instead, with the denomination names as keys? Is it possible to guarantee a sequence in which the keys of a dictionary are iterated through? In general no. Python 2.7 and 3.1 do have an ordered dictionary: http://www.python.org/dev/peps/pep-0372/ Kent ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor
Re: [Tutor] Poor style to use list as array?
Angus Rodgers ang...@bigfoot.com wrote as keys? Is it possible to guarantee a sequence in which the keys of a dictionary are iterated through? Indirectly yes: for key in sorted( dct ): print key I would definitely tend to go with a dictionary for the denominations/values in this case (Actually I'd probably go for a class but that's another ball game entirely!) -- Alan Gauld Author of the Learn to Program web site http://www.alan-g.me.uk/ ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor
[Tutor] append question
Hi Python Tutors: I have a data structure that looks like: test=[[1,2,3],[4,5,6],[7,8,9]] I want to define a new variable that captures the second element of each sublist from above: testvar2 = [] Next I try to capture the aforementioned elements: for i in len(test): testvar2.append(test[i][2]) I want testvar2 = [2,5,8] but instead I get the following error message: Traceback (most recent call last): File pyshell#34, line 1, in module for i in len(test): TypeError: 'int' object is not iterable Any insight would be appreciated. Thanks Steve -- Steven Buck Ph.D. Student Department of Agricultural and Resource Economics University of California, Berkeley ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor
Re: [Tutor] Poor style to use list as array?
On Sun, 5 Jul 2009 18:49:32 -0400, Kent Johnson wrote: On Sun, Jul 5, 2009 at 2:48 PM, Angus Rodgersang...@bigfoot.com wrote: for i in range(LEN - 1): (count[i], amnt) = divmod(amnt, value[i]) Incidentally, I forgot to quote the next line: count[-1] = m Of course, this is much better incorporated into the loop, thus: for i in range(LEN): (count[i], amnt) = divmod(amnt, value[i]) and this lends itself to being rewritten in terms of some other kind of iteration (as below). How about this: counts = [] for val in value: count, amnt = divmod(amnt, val) counts.append(count) I like that very much, because it is in the nature of the problem that the numbers in the list 'value' are all distinct, and so can be used as keys. However, as this remark suggests, I think I am going to need 'count' to be a dictionary, rather than a list, and the same goes for 'denom', and 'plural' (although this should be keyed by the strings in 'denom', and we don't need the 'None's). So it looks like we should have something like: plural = {'penny':'pennies'} counts = {} for val in value: (count, amnt) = divmod(amnt, val) counts[val] = count and, later in the program, something like this (comments stripped for brevity, and one new comment added): for val in value: amnt = counts[val] name = denom[val] if amnt: to_print -= 1 if printed: if to_print: buff += , else: buff += and printed += 1 if amnt 1: # This should become a function if name in plural: name = plural[name] else: name += 's' buff += %d %s % (amnt, name) I haven't run this, but I'll try rewriting the program tomorrow. It looks like there's nothing to it. I'm much happier now, thanks! -- Angus Rodgers ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor
Re: [Tutor] append question
Read your error message... It highlighted the first line of your for loop ansd said ints aren't iterable. len(list) returns an integer. You want a list of items... for i in range(len(list)): On 7/5/09, Steven Buck bucks...@gmail.com wrote: Hi Python Tutors: I have a data structure that looks like: test=[[1,2,3],[4,5,6],[7,8,9]] I want to define a new variable that captures the second element of each sublist from above: testvar2 = [] Next I try to capture the aforementioned elements: for i in len(test): testvar2.append(test[i][2]) I want testvar2 = [2,5,8] but instead I get the following error message: Traceback (most recent call last): File pyshell#34, line 1, in module for i in len(test): TypeError: 'int' object is not iterable Any insight would be appreciated. Thanks Steve -- Steven Buck Ph.D. Student Department of Agricultural and Resource Economics University of California, Berkeley -- Sent from my mobile device ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor
Re: [Tutor] Poor style to use list as array?
On Mon, 06 Jul 2009 01:02:10 +0100, I hastily wrote: Incidentally, I forgot to quote the next line: count[-1] = m That was copied-and-pasted from an older version of the program, with less descriptive identifiers. 'm' should be 'amnt'. Of course, this is much better incorporated into the loop, thus: for i in range(LEN): (count[i], amnt) = divmod(amnt, value[i]) [...] So it looks like we should have something like: plural = {'penny':'pennies'} counts = {} for val in value: (count, amnt) = divmod(amnt, val) counts[val] = count Better, of course (I still haven't run it, but it should work): counts = {} for val in value: (counts[val], amnt) = divmod(amnt, val) I'm much happier now, thanks! ... but also rather annoyed with myself. -- Angus Rodgers ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor
Re: [Tutor] append question
2009/7/5 Steven Buck bucks...@gmail.com: for i in len(test): testvar2.append(test[i][2]) I want testvar2 = [2,5,8] but instead I get the following error message: Traceback (most recent call last): File pyshell#34, line 1, in module for i in len(test): TypeError: 'int' object is not iterable Any insight would be appreciated. Thanks Steve -- Steven Buck Ph.D. Student Department of Agricultural and Resource Economics University of California, Berkeley This sounds like a homework assignment, and we're not supposed to give out answers to homework. The error message and the docs explain what you're doing wrong if you take a moment to look. from http://www.python.org/doc/2.6/reference/compound_stmts.html#for for_stmt ::= for target_list in expression_list : suite [else : suite] The expression list is evaluated once; it should yield an iterable object. An iterator is created for the result of the expression_list. The suite is then executed once for each item provided by the iterator, in the order of ascending indices. Each item in turn is assigned to the target list using the standard rules for assignments, and then the suite is executed. As Luke said, len returns an int, which as your error tells you, is not iterable. From the same page: The for statement is used to iterate over the elements of a sequence (such as a string, tuple or list) or other iterable object: Therefore you have an iterable, there is no need to try and construct a new one. Does that help? It is extremly unpythonic to iterate over range(len(...)), as it adds in the overhead of two function calls, and ruins the readability of code. The latter is probably the most important of the two. An even more pythonic way to do this would be a list comprehension, http://www.python.org/doc/2.6/tutorial/datastructures.html#list-comprehensions If it's not homework, let us know, and we'll be more than willing to give you code if you still need it. -- Richard Roadie Rich Lovely, part of the JNP|UK Famile www.theJNP.com ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor
Re: [Tutor] append question
In [1]: test=[[1,2,3],[4,5,6],[7,8,9]] In [3]: testvar2 = [] In [16]: for i in range(len(test)): : testvar2.append(test[i][1]) : : In [17]: testvar2 Out[17]: [2, 5, 8] Robert On Sun, 2009-07-05 at 15:57 -0700, Steven Buck wrote: Hi Python Tutors: I have a data structure that looks like: test=[[1,2,3],[4,5,6],[7,8,9]] I want to define a new variable that captures the second element of each sublist from above: testvar2 = [] Next I try to capture the aforementioned elements: for i in len(test): testvar2.append(test[i][2]) I want testvar2 = [2,5,8] but instead I get the following error message: Traceback (most recent call last): File pyshell#34, line 1, in module for i in len(test): TypeError: 'int' object is not iterable Any insight would be appreciated. Thanks Steve -- Steven Buck Ph.D. Student Department of Agricultural and Resource Economics University of California, Berkeley ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor
Re: [Tutor] Poor style to use list as array?
if name in plural: name = plural[name] else: name += 's' This could be written more cleanly (although arguably not as readably) as name = plural.get(name, name + s) d.get(key, default) returns the value from d mapped to key if it exists, or default otherwise. You might also want to split your calculation and display code into two separate loops. This might seem wasteful, but it will make your code easier to read and maintain, and the waste is only marginal with the loops you're running - there is a maximum of only 17 passes (once for each value of coin and note) -- Richard Roadie Rich Lovely, part of the JNP|UK Famile www.theJNP.com ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor
Re: [Tutor] Poor style to use list as array?
Angus Rodgers wrote: The problem this time is, given an amount of US currency in cents (and less than a dollar), to print out a description of the coins needed to represent it, using the smallest number of coins. E.g.: How many cents this time? 59 2 quarters, 1 nickel and 4 pennies. How many cents this time? 55 2 quarters and 1 nickel. How many cents this time? 75 3 quarters. My program relaxes the restriction to amounts less than a dollar, and initialises the necessary data structures as follows: denom = ['dollar', 'quarter', 'dime', 'nickel', 'penny'] value = [100, 25, 10, 5, 1] LEN = 5 plural = [None] * LEN plural[-1] = 'pennies' count = [None] * LEN # Kludge I'm inclined to use a class. I omitted the input and loop for simplicity. class Coin: def __init__(self, name, value, plural=None): self.name = name self.value = value if plural: self.plural = plural else: self.plural = self.name + 's' self.count = 0 def display(self): if self.count == 0: return None if self.count == 1: return %d %s % (self.count, name) else: return %d %s % (self.count, self.plural) coins = Coin('dollar', 100), Coin('quarter', 25), Coin('dime', 10), Coin('nickel', 5), Coin('penny', 1, 'pennies') amnt = 99 buff = [] for coin in coins: (coin.count, amnt) = divmod(amnt, coin.value) d = coin.display() if d: buff.append(d) if len(buff) 2: print buff else: print ', '.join(buff[:-1]) + and + buff[-1] I use the list 'count' as a kind of array, so that e.g. count[-1] is the number of pennies needed, count[2] is the number of dimes needed, and so on. Any initial values I might store as elements of the list are never used, hence the use of 'None'. (The use of 'None' in the list 'plural' is much more defensible, I think, as it is used to denote the formation of a regular plural by adding 's' to the name.) Instead, the actual values are assigned in a 'for' loop, as follows: for i in range(LEN - 1): (count[i], amnt) = divmod(amnt, value[i]) This feels like a bit of a cheat, as if I am trying to program in Python as if it were some other more familiar language. Should I have coded this differently? The full source code is here, in case anyone wants to look at it (but I'm not soliciting any more attention, as I've already been given quite a lot of it, and fear being offered the comfy chair!): http://python.pastebin.com/d5e321ae0 (retention: 1 day) Could I have used dictionaries instead, with the denomination names as keys? Is it possible to guarantee a sequence in which the keys of a dictionary are iterated through? (If not, I suppose I could keep the list 'denom' as it is here, and iterate through it with for key in denom:, although this seems a bit redundant.) -- Bob Gailer Chapel Hill NC 919-636-4239 ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor
Re: [Tutor] append question
2009/7/6 Steven Buck bucks...@gmail.com: Thanks for the previous responses. This isn't homework--I'm beyond coursework, although I am a newbie to Python (and I've never had to do much real programming since I've just used Stata for econometric analysis). I'm testing Python as a more powerful alternative to Stata. I've learned from the responses I received, although now see my problem differently. The data structure I have uses a dictionary and know now that the append command doesn't work. Having said that, perhaps my variables of interest have already been created--perhaps I just don't know how to identify them. I've been using some borrowed code to get me started; my modified version is below: import sys # The modules below help me get a .dta file into Python. # Although I'm not sure what form they take; I suppose a list of lists??? from StataTools import Reader from StataTypes import MissingValue # I call my data set the psid (Panel Study of Income Dynamics) # In Stata this would look like and NXK matrix (N observations and K variables) psid=Reader(file('data3.dta')) # I gather this next just creates a list of the variable names. varnames=[x.name for x in psid.variables()] # It's not clear what these next two lines gain me. labels=psid.file_headers()['vlblist'] Labels=dict(zip(varnames,labels)) From here, I'd like Python to identify the Nx1 vectors (or n-tuples) that correspond to the varnames list defined above. I can't seem grab the vectors representing age, wage, etc.. I've tried things like age, psid['age'], psid.age. My last email was an attempt to create the vectors myself, although the Reader module puts the data in a dictionary structure so the append command I was trying to use doesn't work. Hopefully once I learn to create and call on my own vectors and matrices I'll be better off--I'm comfortable working with these in MATLAB and Stata. Bottom line: Given the above data I've imported/extracted from Stata .dta file, how do I create an Nx1 vector which I call 'age'? Thanks for your patience with this newbie. Steve Sorry about suggesting this was homework... I noticed the word University, but not the line above it saying PhD student... If you're new to python, the standard path we recommend is to take an online tutorial. Dive into Python (http://www.diveintopython.org/) has been recommended for people with some programming experience, and Think Python for those without, although I'm sure that if you ask 10 members of this list, you'll get 20 different suggestions. Real veterans get pointed at the standard libraries, which are extremly well commented in the most part. I personally learned from the tutorial in the python documentation, but it does leave a fair bit of the thought processes involved in programming out. See what I mean? One person, three recomendations. -- Richard Roadie Rich Lovely, part of the JNP|UK Famile www.theJNP.com ___ Tutor maillist - Tutor@python.org http://mail.python.org/mailman/listinfo/tutor