Functional Programming with Python

[MUSIC] 0:00 How many of you. 0:04 So, a lot of you do Python. How many of you have done 0:05 functional programming in Python, JavaScript, any language that supports it, 0:07 which is most of them? 0:11 Some of you? 0:13 All right. So, 0:14 functional programming is a fun little area. 0:15 So the basic idea is that you do all of your computing, 0:21 all your programming just by evaluating functions. 0:24 So you always have expressions like this thing is equal to that thing. 0:28 Thank you. 0:34 And it really helps when firs, functions are first class citizens of the language, 0:37 which means you can pass functions around. 0:41 Ooh. 0:44 So you can send functions to other functions. 0:46 You can make a variable just equal to a function, and 0:49 then call that thing over and over again. 0:52 You can do all that fun stuff. 0:54 All of this applies to Python, which is great. 0:56 Another rule for functional programming. 1:00 And this is one we're gonna bring up a lot, 1:03 is that functions shouldn't have side effects. 1:04 So if I run a function with the same data, 1:07 I should get the exact same output every single time. 1:10 That seems kinda obvious, but let's talk about it. 1:15 So in Python, if I do this. 1:19 What is the list that I passed then on the outside? 1:23 Right, I had a list and I passed it in, yeah? 1:26 Let's just do it. [BLANK_AUDIO] 1:29 All right. 1:32 [SOUND] So we're in Python. 1:33 [BLANK_AUDIO] 1:38 42, 28, 1, 105 and 2. 1:42 All right, messy list, yeah? 1:49 [SOUND] Now cheating. 1:51 All right, so there's my function. 1:58 If I call reverse list, and I pass in a. 2:00 What'd I get back? 2:04 >> [INAUDIBLE] >> Right. 2:05 So now what is A? 2:10 Is it my old list or is it the reversed list? 2:12 >> [INAUDIBLE] >> Nope. 2:15 This function has a side effect, because this function calls list.reverse, 2:19 which reverses list in place, because lists are mutable. 2:23 You don't want that. 2:27 You don't want side effects. 2:28 'Kay. 2:30 So. 2:32 Eww. 2:34 >> [LAUGH]. 2:35 >> Because we don't want side effects. 2:35 So we're gonna talk about addressing side effects in a moment. 2:38 But we'll go through some more bullet points here. 2:41 Functions should also be limited in their scope and their functionality. 2:44 You wanna write small little functions. 2:46 Lots of them. 2:48 But they're really small. 2:49 Right? They just do a little bit. 2:50 For instance, this is very ew. 2:51 Let's look at this and we'll see part of why it's bad too. 2:58 So we taken a list, we reversed the list, we swapped the first and 3:01 last items of the list, we stick a thing into the list. 3:05 If it's a string, then we reverse the first thing that's in the list. 3:08 Or sorry, if the first thing in the list is a string, 3:13 we reverse the first thing in the list. 3:15 Otherwise if it's a number, then we int divide it by 2, 3:17 and then we eventually return the list. 3:21 This is horrible, it does a lot of stuff. 3:23 It doesn't have any side effects though. 3:25 Which is nice and weird but it's still ugly. 3:29 So we can break it apart into nice, small functions, right? 3:33 Cuz we want them to be limited in what they do. 3:36 Small scopes, small amounts of functionality. 3:38 This isn't necessarily better cuz it's still doing really ugly stuff down 3:41 there on the bottom but it's better. 3:44 Cuz it's small. 3:48 Right? 3:49 Okay, last thing and we'll get to some code. 3:51 Well, almost last thing. 3:55 It's great if functions can chain. 3:58 So if I can multiple functions that always give me back more data, 4:00 I can put those functions inside of each other, 4:04 and I can do a whole lot of stuff in one line that's all very explicit, but 4:05 it's not scattered out doing crazy weird stuff in each function. 4:10 Okay, so that means we're gonna use return a lot, way more than you may be expecting. 4:15 And, like we said before with the list.reverse, we don't want mutations, 4:22 we don't want to return like, things that have been mutated. 4:27 Okay, we get some chaining-friendly things from Python, yay. 4:32 So let's start with map. 4:37 Anybody heard of map? 4:39 Few people. Anybody want to tell me what map does? 4:41 Without reading this! 4:45 I know it's already up there, but without reading it what does map do? 4:46 So what map does [LAUGH] map takes a function. 4:49 Map is a function, it takes a function. 4:53 And then it takes an iterable. 4:54 And map runs that function on everything that's in the iterable. 4:56 And gives us back a new iterable, right? 5:00 So, if I have a function that turns numbers, turns things into strings. 5:05 Right? 5:12 It's really annoying that I can't Cmd+Tab from there. 5:14 All right. 5:17 You've all seen str, right? 5:18 The string class? 5:20 And if I do str and a number, I get back a string of the number. 5:22 Yeah? 5:26 Nothing crazy there? 5:27 'Kay. 5:28 So if I send in the string function to map. 5:30 You know, hold on. 5:33 There we go. 5:35 If I send in the string function to map and I give it a list of numbers. 5:37 Thanks Python 3 for making me look silly. 5:45 I get back this map object which is a generator. 5:48 All right, there we go. 5:50 I get back a list of strings of numbers. 5:52 So really handy if I want to make sure everything in here's a string. 5:55 Make sure this is all strings, right? 5:57 So all I've done is I applied the string function to everything that's in the list. 6:01 All right. 6:06 We could do the opposite of that. 6:08 [BLANK_AUDIO] 6:09 Now we see how well I can, [SOUND] wow. 6:14 >> [INAUDIBLE] [SOUND]. 6:18 >> There we go. And 6:19 now we got ints instead of strings, 'kay? 6:20 We're gonna do some crazier stuff with map in a moment. 6:24 But, that's the idea of map. 6:26 Run a function over everything that's in a list, or a dict, or a tuple, or whatever. 6:28 Okay. 6:37 Filter is our next one. 6:37 How many of you have messed with filter? 6:38 Nah, a couple people. 6:41 All right. 6:41 So filter is a handy way of saying, make sure all this stuff is true. 6:43 Or is like I want it to be. 6:47 So, ignore the or none for right now. 6:49 If we pass a function into filter, that function must return true or false. 6:52 For everything that's in the iterable, we run that function. 6:58 If that function gives back true, that thing that was in the iterable gets kept. 7:02 If it returns false that thing is discarded. 7:06 'Kay? 7:10 So if I'm like, I've got a list of people and 7:10 I wanna make sure all these people are 21 or over. 7:12 They'll have an age attribute. 7:15 I can write a function that checks the age. 7:16 Make sure the age is 21 or up. 7:19 And if it is, returns true, otherwise returns false. 7:21 All right? 7:23 So a quick easy way to filter through a bunch of people. 7:24 The none is a fun little shortcut. 7:28 You just pass none instead of a function, 7:30 it only returns things in the iterable that are truthy in Python. 7:32 So no zeroes, no empty strings, no empty lists, nothing like that. 7:36 But you'll get back all of your lists that, or 7:40 all of your strings that actually have content. 7:41 All of your numbers that are not zero. 7:43 All that kind of stuff. 7:47 We'll do a lot of stuff with filter too. 7:48 And our last one we're gonna talk about is reduce. 7:51 Reduce is a little weird, compared to filter and map. 7:54 Reduce always gives you back just one thing. 7:58 And the function always takes two arguments. 8:02 So what it does, is for 8:06 the first two items in the iterable, it runs the function with them. 8:06 Takes whatever the answer to that function is, runs it against the third item in 8:09 the iterable and so on, until there's just one answer that comes out. 8:13 So if my function just adds two numbers together, I pass it a big list, 8:18 I get the sum of the list. 8:22 If it multiplies things together, I get the product of the list. 8:24 All right? 8:28 We'll do stuff with reduce as well. 8:29 That's enough talk. 8:32 so. 8:35 Let's go do stuff. 8:38 All right. 8:40 Wow, I got a lot of things open. 8:41 [SOUND] So, there was a GitHub repo that had these three files over here, 8:42 test_filter, test_map, and test_reduce in it. 8:48 And we can test all of those at once by running python unittest. 8:53 All of our tests pass. 9:00 Because we ran 15 tests and we had 15 failures. 9:01 So lets fix that. 9:04 So which one do you want to do first, filter, map, or reduce? 9:04 >> I don't care, we can do any of them. 9:09 >> Filter, map and reduce? 9:13 All right, lets do filter. 9:13 I'm going to close this one. 9:18 And all right, there we go. 9:21 So we just have Emacs. 9:22 Yes, Emacs. 9:26 I think I was on vim the last time I presented up here. 9:26 All right, so we got a bunch of stuff up here. 9:30 We've got some nums, we got the states, we got some some values, 9:33 we got a bunch of strings, and then we've got this string of numbers. 9:35 Which this is a generator pattern. 9:40 I talked about how filter gives us back generators. 9:42 We're not gonna worry about what generators are. 9:44 Just, yeah. 9:47 That's why it's a stream instead of like number_list. 9:48 And then we have this word_list function that opens up 9:52 my dictionary that's built into my computer. 9:55 And gives me back all of the words that are in that dictionary. 9:58 Which is a lot of words. 10:02 >> [LAUGH] >> Safe enough for this. 10:03 All right, so then we've got some tests. 10:07 So let's look here at our first test. 10:09 Our first test is positive numbers, so I wanna make sure that all 10:11 of the numbers that are in this positive numbers variable, are greater than or 10:16 equal to zero. 10:20 So no negative numbers. 10:21 'Kay, but if I look at nums I got a negative two in there. 10:25 And I want to get rid of that. 10:29 And I'm going to be, somewhat obtuse and not just use remove. 10:31 So let's write this. 10:37 So positive numbers has to equal something. 10:40 So filter, there's gonna be a function and there's gonna be nums right? 10:45 So, I want interaction on this by the way, you can all yell out things at me. 10:52 And I will ignore the things I don't like. 10:57 So, we need to write a function that only gives us back the numbers that 10:58 are greater than zero. 11:02 So what are we gonna call it? 11:03 Anybody? 11:05 How about positives? 11:06 Yeah, does that work? 11:10 Okay, and it needs to take a single item, so it's gonna take a num. 11:11 Or an nu. 11:15 Okay, inside of this function how do we check the numbers greater than zero? 11:17 Or greater than negative one, rather. 11:22 Return num greater than or equal to zero. 11:25 Does that work? 11:28 Good for everybody? 11:31 'Kay? 11:34 That's it. That's my entire function. 11:36 All I'm doing is making sure that number's bigger than that. 11:38 And in here I have to tell it what function to run. 11:42 All right, I'll scroll that up a little. 11:46 As much as I can since I'm at the bottom of my file. 11:49 So all we're doing is we're using filter. 11:54 We're using our positive function we just wrote. 11:56 And we're using our nums list that's up at the top. 11:58 'Kay? 12:03 Shall we see if it works? 12:05 Yay. 12:16 It does. Cuz we only have six errors out of 12:16 seven tests. 12:18 So we got that one to work. 12:19 Somewhere up here will be a little sign that says it worked. 12:22 Yeah, there's our dot. 12:24 The rest are all e's. 12:25 All right that's our way we're gonna do this, this whole TDD thing so. 12:28 Let's go to another one, yeah? 12:35 [BLANK_AUDIO] 12:36 Okay, our next test is no non-zero numbers. 12:38 So all the numbers that are not zero. 12:43 So short cut on this one, anybody tell me the short cut real quick? 12:45 We talked about what filter does. 12:48 >> [INAUDIBLE]. 12:50 >> Right. 12:53 So, non_zero_numbers is what we have to call this. 12:54 So, filter None and nums, right? 13:00 Cuz zero is falsey thanks to the glory of Python. 13:04 Sorry for things like, flashing up there, my, my Emacs likes to do a lot of things. 13:11 All right. 13:17 I also think it's kinda cool that I wrote way more lines of code for 13:18 tests than I have to write for solving this. 13:21 And we're gonna make it even shorter by the end if we get to lambdas. 13:24 So, yay. 13:26 Or list comprehensions. 13:28 All right, let's see if that passes. 13:30 It does. Five errors, not seven. 13:33 All right. 13:36 What's our third one? 13:38 No vowels in one word. 13:42 So I got a word. 13:46 My word is California. 13:47 I don't want there to be any vowels in it. 13:50 So vowel is my variable here. 13:55 Sorry let me find my word back up here again. 14:05 State, that's my word. 14:07 Okay so what function would I write? 14:11 [BLANK_AUDIO] 14:14 Actually that's not gonna be a word that comes in. 14:22 Gonna be a letter. 14:24 Because when we look at a word as an interval we just get the letters. 14:26 Right? 14:30 So how do we get rid of the vowels? 14:32 If letter in aeiou, actually 14:35 we'll do it easier, not in [NOISE]. 14:41 Or return true, either way. 14:48 Returning the letter is truthy, returning true is very truthy, because it's true. 14:53 All right, and then no vowels. 14:58 All right. 15:06 See if it works. 15:07 Oh, it didn't work. 15:09 Why didn't that work? 15:10 No_vowel_word. 15:13 That's what I'm supposed to call it? 15:17 No_vowel_word. 15:19 Wow, I picked weird variable names. 15:25 Still didn't work. 15:29 [BLANK_AUDIO] 15:30 Okay, I'm gonna cheat for a second. 15:39 [BLANK_AUDIO] 15:40 no_vowel_word. 15:49 Oh, that's right. 15:52 I gotta pass in an actual word. 15:53 All right. 15:54 Okay. 15:56 [BLANK_AUDIO] 15:56 I apologize. 16:08 I wrote these like late one afternoon. 16:09 So, some of them are gonna make a lot more sense than others. 16:12 There we go, four errors. 16:16 Okay, I'll go back to how I did that one. 16:17 So what we did was we just joined all the letters together. 16:19 Cuz we get back this list-like object from the filter. 16:22 So we just joined them all back together into one word. 16:28 Just to make sure it was a word. 16:31 I probably shoulda left it as a list but whatever. 16:34 It works. 16:38 Okay so we call a filter with no vowels on our state name lower cased. 16:44 Right, and then that gives us back this list-like object and 16:50 so we join it back together with no spaces. 16:55 So that it's a string again instead of this list-like object. 16:58 That's only cuz of how the test was written that we have to do that. 17:05 If the test wasn't written to where 17:08 it's calling this .lower, then it wouldn't have had to do that. 17:14 But it's okay. [COUGH] 17:19 [BLANK_AUDIO] 17:21 Cool, okay any questions at this point? 17:27 We've done three things for filter. 17:30 Everybody feeling okay with filter? 17:34 All right, you want me to leave the rest for you to do later, and 17:36 we can go onto map. 17:39 Let's do one more filter. 17:43 All right, that was awfully quiet. 17:45 [LAUGH] Let's skip the no false-y cuz that's pretty easy. 17:47 What I got here? 17:53 Only modulo three? 17:55 Only words over ten characters. 17:56 That'll be a fun one. 17:58 All right, so, I got my word list here, right? 17:59 All these words. 18:06 Tons and tons of words. 18:08 In fact if we do ipython from test_filter import as let's do import test_filter. 18:11 We get back this generator. 18:26 [SOUND] Wait I need two of those. 18:27 [BLANK_AUDIO] 18:30 Len list. 18:33 So we get back 235,000 words. 18:36 I only want the ones that are over ten characters. 18:37 All right? 18:41 That's easy, right? 18:42 How hard is it to, to only have words over ten characters? 18:44 'Kay. So let's do that. 18:48 >> Can you show what you had [INAUDIBLE] last one? 18:50 >> Yup. 18:53 That right there. [BLANK_AUDIO] 18:55 I will make all of the answers available at the end as well. 19:01 I'll set up a, a branch on the repo that's answers, and 19:05 that'll have the completed ones. 19:07 [BLANK_AUDIO] 19:09 Just in case anything gets missed. 19:12 All right so let's start ten letter words. 19:17 And I actually had to do this several times to make sure that like ten letter 19:20 words was still something feasible. 19:23 Cuz I started at like eight and it was still like 10,000 words. 19:25 And that was just stupid. 19:28 okay. 19:31 So ten_letter_words has to be words that are ten letters or more. 19:31 So let's do like word_length for our function. 19:37 And then we're gonna pass in word_list. 19:41 All right. 19:46 So help me write word_length. 19:47 It's gonna get a word, what do I wanna send back? 19:53 [BLANK_AUDIO] 19:57 So return, len word greater than or equal to 10. 20:04 >> So you're changing [UNKNOWN] cuz if it's only words over 10 characters. 20:11 >> [INAUDIBLE] Characters. 20:16 >> yeah. Good catch. 20:21 Bonus points. 20:24 All right so, our word length just says, 20:27 hey is this word more than ten letters long or equal to ten letters. 20:31 We could have also just done greater than 9. 20:38 Since words always have whole numbers, for 20:42 their length, you can't have a half a letter. 20:44 Give that a try. 20:48 Only three problems, yay. 20:53 All right. 20:55 I'll go back to this in case anybody missed word length, or 10 letter words. 20:57 This filter make sense? 21:03 You get a pretty good idea of how filter works. 21:04 It's just for truthy things, right? 21:06 [COUGH]. 21:08 >> Inside of, now we say, well, 21:09 my memory works, maybe Craig can remind me itertools, I think. 21:12 There's filter false. 21:15 Is that itertools or functools? 21:17 One of those two, either inside of itertools or 21:19 functools, there's filter false, which is the opposite of filter. 21:22 So, filter only gives you back truthy things, 21:25 filter false only gives you back falsey things. 21:28 So depending on what your scenario is. 21:32 It's really not that hard to swap a Boolean, 21:34 when you're writing filter stuff and just, make it work the other way. 21:35 But, if you just can't do that for some reason, [UNKNOWN] does exist. 21:39 And it's pretty cool. 21:43 Works exact same way. 21:45 There's tons of cool stuff in functools and itertools yeah. 21:47 We'll eventually teach them, cuz I like them. 21:52 Okay, so let's move on, let's do map. 21:55 All right, I think maps actually are little bit simpler. 21:58 But, that's cool, alright, so we've got a bunch of numbers again. 21:59 We've got numbers from one to 20. 22:05 We've got some time zone strings, how many of you watched the dates and time scores. 22:07 Couple, all right, it's not the most exciting course. 22:13 And we've got a format string that we're gonna put stuff into and 22:17 we have some dictionaries that we're going to put into that format string. 22:21 Alright, so the first thing I want is I want to 22:27 get back the double of all of my numbers. 22:30 So I've got all these numbers, I want to double them all. 22:34 So instead of one, I want two, instead of two, I want four, that kinda thing, right? 22:36 And we're gonna put those in a variable called double_nums cuz, creative. 22:41 Double_nums = map, something against nums. 22:47 All right. This should be an easy function to write. 22:54 How do we double numbers? 22:55 [BLANK_AUDIO] 22:57 So we'll return num times 2. 23:02 [BLANK_AUDIO] 23:04 All right. 23:09 So double num, so we're just gonna run map. 23:10 And I'm gonna pass it doubler, 23:12 which is our function that returns number times 2 for each of the numbers. 23:14 All right? 23:17 [SOUND] Let's do another one before we before we test this. 23:19 Let's do the half numbers, cuz that's, that's the same idea, right? 23:23 [SOUND] So we've got half nums, and it's gonna be, we'll say, halver. 23:26 Halfer? 23:33 >> [LAUGH]. 23:35 >> Whatever it would be. 23:36 [LAUGH] naming things is one of the hard problems in computer science. 23:37 [BLANK_AUDIO] 23:41 All right, and this would return num divided by 2. 23:45 Right? 23:48 So double nums we're multiplying by 2, half nums we're dividing by 2. 23:51 Really straightforward, yeah? 23:55 Okay? 23:59 By the way, I like map. 24:01 Map is just like, oh yeah. 24:02 [NOISE] Obviously that's the one I'm gonna do. 24:03 So five tests, only three of them failed. 24:05 So the time zones, the num strings, and the format strings are all bad still. 24:07 Let's fix those. 24:13 All right. 24:14 Num strings looks pretty easy. 24:16 Num strings actually looks insanely easy, because all we're wanting to do is 24:17 make each of our numbers a string. 24:24 I did this one like 15 minute ago. 24:29 Remember what it was? 24:31 All right, we're just gonna 24:32 pass our string function, or our string class that acts like a function or other. 24:38 And we give it to the, the nums variable, all right? 24:43 And the the nice thing about this is since there's no side effects, 24:48 our nums never changes, right? 24:51 We can do all sorts of stuff to nums, 24:53 we can do anything we want, and nums is never gonna change. 24:55 That's really good because if I'm getting data from a source like say, Twitter. 24:59 And I wanna parse this data repeatedly. 25:03 If I go and change it, I'm never really sure what state its in. 25:05 But if I'm always just processing it from its raw state, 25:09 then I know what state its in. 25:12 Its in the state that Twitter gave it to me. 25:14 So I don't have to deal with, okay let's make sure that this thing is still there. 25:16 Or, make sure that I didn't rename this part or whatever. 25:20 All right [SOUND] yay. 25:23 All right. 25:27 So, we just have two more. 25:28 You wanna solve these? 25:29 Let's solve these. 25:31 Which one do you wanna do first? 25:33 Do you wanna do time zones? 25:34 Whoo, time zones. 25:35 Or do you [LAUGH] wanna do the string formatting? 25:37 >> [SOUND] Time zones. 25:41 >> Time zones, all right, time zones. 25:43 Time zones. 25:47 Map. 25:49 [SOUND] And, what did I call that thing up there? 25:50 Time zone strings. 25:58 All right. 26:00 So what our test is wanting. 26:05 Where is it? There's what are test is wanting. 26:09 Are test wants all of the, we want a datetime.tzinfo instance, for 26:11 each of the timezones that are there. 26:17 So, actually forgot how I did that one. 26:19 I've been doing way too much time on stuff lately. 26:25 I used pytz, of course, which is why pytz is imported. 26:29 All right. So, let's make a method, 26:33 [SOUND] that makes timezones for us. 26:41 All right, if you haven't watched the course, 26:50 you may not know how to do this one, or if you haven't played with pytz yourself. 26:52 So, all we want to do is we want to 26:56 return pytz.timezone [SOUND] with the name that was passed in 26:59 [BLANK_AUDIO] 27:04 That makes sense? 27:14 We got the names, we're just generating all these timezones. 27:15 All right? 27:17 All right. 27:19 Let's see if that one passes, and then we're gonna hopefully make it simpler. 27:20 Air pass all right. 27:26 So, we know that map calls a function on everything. 27:27 What do we pass in, right? 27:33 So, I should be able to do, 27:35 [SOUND] pytz.timezone because that's a function, right? 27:37 It generates a time zone or 27:42 a tz info thing, we really don't want to go into what timezones are. 27:44 So I should be able to just call that method straight from there, 27:51 without having to use my time zoner. 27:53 Yep, I can. 27:56 So, I don't even have to write this code, that's useless code. 27:57 I wasted two whole lines and a blank liner, too. 28:01 Don't waste lines of code, they're precious. 28:06 Okay. 28:08 So, our last one is this format strings, 28:09 we've got the string up top, let's go look at it. 28:13 This format string, that it has a couple of named placeholders. 28:17 One of them's called name, the other is called thing. 28:21 It's a very creative sentence. 28:24 And then we have these dictionaries, that have keys and values in them. 28:27 We're gonna put those into the string. 28:31 Yeah? 28:35 All right. 28:36 So, we actually have to write code for this one. 28:38 What's this supposed to be? 28:42 Statements, map let's do formatter. 28:44 And we want to pass in here, is the dictionaries. 28:52 Cuz that's what we're gonna loop through, all these dictionaries. 28:56 [NOISE]. 28:59 [BLANK_AUDIO] 29:06 See how hard it is naming things, I can't even come up with a good variable name. 29:10 All right trying to get this so you can see more. 29:15 So what we wanna do is we wanna return format-string.format and 29:19 we're gonna pass in the values as keyword arguments. 29:27 Is the star star surprising and new to anybody? 29:30 All right. So, star star is really cool in Python. 29:34 It turns a dictionary, into keyword arguments. 29:36 So, normally it have like, key colon value in your dictionary. 29:39 In your function call, you now have key equals value. 29:43 So, you don't have to find those values yourself, 29:47 you don't have to deal with reformatting stuff in whatever weird way. 29:49 You don't have to do, like, you know, 29:53 my_dict(key)=my_dict(value) when you're filling something 29:55 out, because that would be really annoying and take forever. 30:04 So the star star turns that into keyword arguments, which is really handy. 30:10 All right. 30:15 So, we call formatter on our bunch of dictionaries. 30:16 That in turn has whatever value comes in, or 30:21 sets the value since the dictionary with multiple values, right? 30:25 We send into our format strings format function, method. 30:28 Yay, all the tests pass. 30:32 All right, are there any questions about map? 30:38 Is map weird or confusing to anybody? 30:40 It's a little strange so if you have questions, feel free to ask. 30:42 No? Everybody's good? 30:47 All right. 30:48 You had your chance. 30:50 Now we're going on to reduce. 30:51 All right. 30:54 So, reduce. 30:55 The refresher takes a function, applies that function to the first two items in 30:57 an interval, takes that result and applies it to the third item, fourth, fifth, 31:01 sixth, so on, so we end up with only one item at the end. 31:06 'Kay? 31:11 And it looks like I left the functions in there. 31:17 So, that's handy. 31:20 All right. 31:21 [LAUGH] So, we wanna add some numbers. 31:22 so, add nums is reduced. 31:25 Which I imported, yay. 31:31 And our function is going to be add_two_nums. 31:34 And we're going to pass it nums. 31:37 So what does add_two_nums do? 31:38 It adds two numbers. 31:40 This is probably the worst example I have tonight, 31:42 because you should really just do the sum function with this, instead of reduce and 31:45 a function that adds numbers together. 31:51 But regardless, it's there, you get the idea of how it works. 31:54 All right. 31:59 Let's do multiply numbers as well. 32:01 So, this should be multiply nums. 32:04 We're gonna write a call reduce, 32:05 multiply two nums, and our nums. 32:15 So multiply two nums, as you can probably guess, multiplies two numbers together. 32:18 So again, instead of the sum of the product, we get the, or sum of the list, 32:24 we get the product of the list. 32:28 This isn't one that's built in as far as I know. 32:30 So, you'd have to write this one. 32:33 >> [INAUDIBLE]. 32:35 >> There, not for 32:40 multiplying a list as it is there is a built in way to solve this problem without 32:42 having to write a function though, and we can talk about that in a minute. 32:47 It's a really cool library called Operator, 32:51 that's part of the standard library. 32:53 That has tons and tons of stuff in it, but one of the cool things it 32:55 has is a function called mul that multiplies two numbers together. 32:58 So you just use operator mul, instead of writing multiply two nums. 33:02 All right and then let's solve longest. 33:08 [SOUND] And what did I call it? 33:16 Did I call it states? 33:20 I did. 33:21 [BLANK_AUDIO] 33:22 Okay. 33:26 So we'll look at this one. 33:27 It takes two strings. 33:28 Yeah? Two strings. 33:31 And it returns if one is bigger than two, 33:33 then it returns one, otherwise it returns two. 33:36 There's another way of doing this too, 33:39 but we'll talk about that later if we have time. 33:41 All right, so 33:45 we used three already written functions, but this should solve our tests. 33:46 We're good. 33:56 There's our tests. 33:57 All right, any questions about reduce? 34:00 Does that one make sense? 34:04 Function that rates over sets of things. 34:09 Gives you back one final answer. 34:12 What was that? 34:13 Oh, work they were just doing? 34:15 Yeah, so, we have two strings that come in, 34:18 which we're creatively calling string one and string two. 34:21 We're checking to see if the length of string one is greater than the length of 34:26 string two. 34:30 And if it is then we send back string one. 34:31 If it's not, which means that string two must be longer or 34:34 they're equal, which I've made sure they weren't then it returns string two. 34:37 Yeah, so the first time through it's gonna have Ohio and Alaska. 34:46 So which one's longer? 34:51 Alaska, so the next time though, 34:53 Alaska becomes state one, and California comes in. 34:55 So, which one's longer? 34:59 California. 35:00 So, state one becomes California, state two becomes Oklahoma. 35:02 California is still longer so, there's nothing else to take as another argument, 35:06 so it says okay, California's your answer. 35:11 And sends that back. 35:13 So this works with a lot of things. 35:16 If you're like hey, I've got, again, maybe all these people, I need to know 35:17 which one's been employed the longest, and they all have, you know, start date. 35:20 Make a date time from when they got started, 35:24 subtract it from now, you find out hey, this dude's been here for 35:26 10 years and this women's been here for 12, so she's got seniority. 35:30 Reduce is actually a lot more handy than filter or 35:36 map, but it's one that you won't see used as often, which is weird. 35:38 So, lets see what's next in my key note. 35:43 [NOISE] More blank slides, functions to know and love. 35:46 I don't want to cover every single of these, but 35:51 if you look through the tests you'll see some of these used. 35:53 These are functions that come up pretty often and 35:57 functional programming in python. 35:59 So the first two are related, max and min. 36:01 Max gives back the biggest of the things that are sent in, 36:05 min gives back the smallest of the things that are sent in. 36:07 So, your state names you should be able to do max and min on those. 36:10 Although that may do alphabetical sorting, I don't remember. 36:14 We can test it later. 36:17 Any and all are related. 36:18 Any takes an iterable, and it gives you back a true, 36:21 if any of the things in the interable are true. 36:24 So, it can have 1,000 false things and one true thing, and it'll come back as true. 36:27 Cuz at least one of those things is, is true, right. 36:32 All is kind of the exact opposite of that, 36:35 it will only tell you true if every single thing in there is true. 36:37 You'll see me use all a lot in the tests. 36:42 And if you could see your code challenge tests, you'll see all used a lot, 36:44 because I love all. 36:47 All's especially great if you put conditions inside of there. 36:50 So, your list items are like, x is 5 or x is less than 2. 36:53 Those are gonna evaluated as the true or false. 36:58 And then reversed takes an irritable and 37:01 you remember how we did list dot reverse and we got, the list reversed in place. 37:03 So reversed will actually send you back a new list, 37:08 that has been reversed from the old list. 37:11 So, it doesn't overwrite your old one. 37:14 Cuz we don't wanna mutate things. 37:16 And then sorted is the same idea, instead of calling list.sort you 37:18 can use sorted on a list and you get back a sorted version of the list. 37:23 Sorted has some other stuff you can add to it, to change how it sorts. 37:27 But that is beyond tonight. 37:31 We will end up covering all of this stuff at some point so, yeah. 37:34 [INAUDIBLE] It'll do a natural sort. 37:39 okay. First, let me repeat that. 37:46 [LAUGH] since [UNKNOWN] pointed that out. 37:48 so, if you don't provide a way to sort, to sort it, what does it do? 37:51 So it does a natural sort. 37:56 And what a natural sort is, is kind of how you think of things as being sorted. 37:58 So, like, numbers will come first, and then characters. 38:02 Right? I may have that backwards. 38:05 I'm pretty sure that's right. 38:07 So you'd get, 38:09 you know, numbers one through ten, or whatever your order is, yeah? 38:10 And then you'd get your words and they would be sorted alphabetically. 38:14 >> So [INAUDIBLE] 38:17 >> I think caps may come first, but I don't remember. 38:21 So all these things you can play with. 38:23 Sorted is pretty handy on its own. 38:27 Once you start writing the sorting functions that it expects, 38:30 it opens up like whole new worlds. 38:33 Cuz you can just sort by crazy, 38:35 like I wanna sort by these five nested attributes, and it'll handle it for you. 38:37 What else is in here? 38:45 Questions, we'll come back to that one, cuz it hasn't been that long. 38:46 Okay. 38:51 Let's talk about lambdas. 38:52 Anybody use lambdas? 38:54 You have, probably. 38:56 Who's written anonymous functions in Javascript? 38:57 Where you just pass function into something, those are lambdas. 39:00 They just don't call them a cool name. 39:02 okay. 39:05 So let's go write a lambda real quick. 39:06 So we'll have nums, I'll again make it a list of one through 20. 39:11 All right. 39:17 So, nums is just some numbers, right? 39:19 And we had earlier, we had this map and 39:24 we had a doubler function when we passed the nums, right? 39:27 You remember doing that? 39:30 All right, I don't wanna write doubler, because I'm lazy. 39:32 All right, so nums I still have, 39:35 I don't wanna write doubler, so I'm gonna write map. 39:39 I'm going to write the word lambda, which I have a really hard time typing so 39:44 tell me if I misspell it. 39:47 So lambda is an ominous function. 39:49 Functions take arguments, right? 39:50 So after we say lambda which says give me an ominous function we can tell it 39:53 what arguments that function is going to take. 39:57 So. 39:59 >> Doesn't matter, I'm gonna call it num. 40:00 Okay. 40:04 And we do a colon, cuz it's a function, it starts with a colon. 40:05 And the weird thing with lambdas and 40:08 Python is you can't press return, so they all have to fit on one line. 40:11 There's a cheat around that, but don't worry about that for right now. 40:15 You're new to lambdas, just write them like they are. 40:18 And then they have an implicit return, 40:21 they automatically return whatever they calculate. 40:23 So we're just gonna do num times 2, 'kay. 40:26 That's my function. 40:30 This lambda thing, all it does is return num times 2. 40:32 Then I pass in nums. 40:37 And I'm gonna make this a list so that you'll be able to 40:40 see the numbers instead of seeing a map object, which will make no sense. 40:44 Yay! 40:50 So I don't have to write functions if I don't want to. 40:51 As long as it's something short. 40:53 I mean, if you were doing all sorts of complex stuff, you obviously probably 40:55 don't wanna use lambda's, unless you can break it up into a lot of little steps. 41:00 But if you're just doing something short and quick, lambdas are amazing. 41:04 So what that means [SOUND] is we 41:10 can do all of these without functions. 41:15 So let's solve solve this modulo three one, 41:22 modulo_three_nums is gonna be a filter, and we're doing it with nums. 41:27 All right, so I have a lambda, right? 41:36 And it takes a number, and 41:41 I'm gonna return if that number modulo 3 is equal to 0, 'kay. 41:43 Cuz modulo is weird as Dave can tell you. 41:51 And we have to make sure that it's equal to zero, all right. 41:53 Now, we'll, we'll see if anybody gets this. 41:58 You should. 42:04 Why do I have to make sure that it's equal to 0, instead of just doing modulo three? 42:05 Because if I just do modulo three, I get back either a number or I get back zero. 42:10 Right? 42:15 So on ones where it actually is modulo three, I'd get back zero, which is falsy. 42:18 Which would make filter not included. 42:23 So that's what we have to do the, is equal to zero, so we get back true. 42:25 It's a weird one but it works. 42:30 And it makes sense sorta. 42:33 Lambdas in their very simple, simple sense. 42:36 Anybody completely lost on lambdas? 42:40 There's all sorts of crazy ways you can break them, 42:41 but we're not gonna worry about those right now. 42:44 And then you can break Python using lambdas too which is really fun. 42:48 I'll show you how to do that in a minute if you want. 42:53 All right, so that was in filter. 42:55 Let's run that. 42:56 All right. 42:58 So we just have two wrong ones now instead of three. 43:00 We had three wrong ones before, just two now. 43:03 All right. 43:05 That is lambda. 43:09 Let's do lambda inside of one of our maps as well, oh we already did a doubler. 43:12 [NOISE] so we can do it here too. 43:18 So we've got lambda, and we've got values. 43:24 I'm actually just basically copying what happened up there in formatter, 43:28 format_string.format (**values). 43:31 All right. 43:38 It's literally the exact same code I just write, 43:40 I just wrote lambda instead of def formatter(values) return. 43:43 So I saved what, 30 characters? 43:48 But if I'm only ever using this once or maybe I'm assigning my map 43:51 to something else somehow so that it can be called multiple times with multiple 43:57 values then you know, I save myself a decent amount of typing. 43:59 Let's make sure that still passes. 44:05 [SOUND] It does, yay! 44:06 And then, I guess the last thing, before we do question. 44:10 Actually, we'll go ahead and do questions. 44:16 I'll leave list comprehensions for another time. 44:18 But before we do questions, I've got one other thing that's important. 44:20 So if you want help, if you're like, 44:23 Python's really cool, but some of the stuff is confusing. 44:25 There are two awesome groups here in Portland. 44:28 One of them is the PyLadies. 44:30 Yeah, PyLadies. 44:33 They meet like 8,000 times a week. 44:36 It, seriously their, their meetings are crazy. 44:39 But they're an awesome group of women beginner all the way up to expert. 44:43 Great group if you need help and are of the womanly persuasion. 44:48 And then every Monday night, 44:53 there's a thing that they call the Python Flying Circus. 44:54 This happens over at the Code Guild headquarters, which I can never 44:56 remember the street they're on, I think it's Water, or Paper, or some other noun. 45:01 But if you join the PDX Python group, 45:06 which is our general Python thing, you'll get tons of meetups. 45:08 But they meet up every single Monday night and it's pretty awesome. 45:11 You show up, if you've got questions they've got mentors. 45:15 If you want to mentor people, they've got questions. 45:17 awesome, awesome group to hang out with. 45:21 All right, so now I'll go back to my questions slide, just so it looks cool. 45:24 Anybody have questions about functional programming in Python? 45:29 Yes. >> [INAUDIBLE] How do I make my lambdas on 45:31 two lines? 45:35 >> How do you make lambdas on two lines or more? 45:36 But that's cheating. 45:38 >> [INAUDIBLE] >> All right. 45:39 So I'll show you a couple of fun things with lambdas. 45:44 Cuz I like lambdas. 45:47 I think lambdas are really cool. 45:47 But I don't like a couple of the rules about lambdas. 45:49 One of which is they're temporary, they're anonymous. 45:51 The other which is they can't be on multiple lines. 45:55 And the third of which is they don't give me any help. 45:56 So let's solve those in not that order. 45:58 So multiple lines. 46:01 map, sorry, list map. 46:04 [NOISE] There we go 46:09 I'll eventually get 46:14 enough parenthesis. 46:20 You make it a container. 46:27 It's a hack but it works. 46:30 If you find that you're constantly writing lambdas that are three, four, 46:32 five lines long, just write a function. 46:36 Don't, don't mess with lambdas like that. 46:39 It's just, it's really not worth it. 46:41 Cuz at some point you're gonna forget that open parenthesis and 46:42 then, you're just lost. 46:45 So let's talk about breaking lambdas, 46:48 cuz lambdas, that lambda doesn't really exist, right? 46:49 I mean it's, it's there but it only exists so long as map is running. 46:53 But if I've written a really cool lambda and 46:56 I wanna have it around for a long time, I can just name it. 46:57 [NOISE] So now I can do. 47:03 [NOISE] 47:08 Yeah. 47:13 So now doubler exists. 47:20 It's not a real function and 47:22 it's not really a lambda, it's just a thing that's there. 47:23 Cuz if I do like type doubler. 47:27 I mean it's a function, it is, 47:30 because lambda's are functions, but yeah, it's not really a function anymore. 47:34 Yes? >> So then like, 47:39 what's the advantage then, you don't have to type return, is that? 47:40 >> So the advantage to lambdas is, yeah, you don't have to type return. 47:45 And if it's something you're just doing once, it's, 47:48 seems kinda silly to go write a two line function if you don't have to. 47:53 If I'm only ever gonna do with once, 47:57 why am I wasting memory holding on to a function? 47:59 Why am I wasting lines of code by writing the function? 48:01 >> So in the case of like [INAUDIBLE] would that make sense 48:04 [INAUDIBLE] inside that function to like, scope it or whatever? 48:09 >> The naming like this, right at the doubler equals? 48:17 >> Like if you want to use the same lambda more than once [INAUDIBLE]. 48:20 >> So the question is, does the naming only make sense if 48:23 you need to use it more than once in a function? 48:26 Yeah, pretty much. 48:29 It's, I, I, I know I'm gonna need to do this a couple of times I don't 48:30 wanna write it over again. 48:34 And for some reason I'm opposed to nested functions, so I named my lambda. 48:35 So then, the other thing with breaking lambdas. 48:41 If I do help on the doubler. 48:44 I, of course don't get anything because it's just a lambda but meta programming. 48:48 So doubler.doc equals doubles numbers. 48:53 And now I can do, help(doubler) and I get hey, it doubles numbers. 48:59 But it's still a lambda. 49:03 It still doesn't really know what's going on, but it's a lambda and it's there. 49:05 You're not supposed to do these things to lambdas, don't, 49:09 don't do these things to lambdas. 49:10 But lambdas like all functions in Python they can do star args, they can do 49:13 star star kwargs they can they are functions for all intents and purposes. 49:16 They just can't have returns in them or new lines in them. 49:20 Like one of these. 49:30 That's just me not counting parentheses correctly. 49:31 And then Python going, hey you missed a parenthesis. 49:33 Yeah, it's, I do that a lot. 49:36 I love having syntax highlighting in browsers [LAUGH] or editors. 49:39 Cuz, yeah. 49:43 All right. 49:45 Anybody else? 49:46 Any other questions or problems with functional programming? 49:47 Yeah, Portia. 49:51 >> This is more of a high level question. 49:52 >> Okay. 49:54 >> What problem is functional programming solving like, 49:55 what convenience is it bringing to the program? 49:58 >> Okay. So 50:01 what convenience, what problem does functional programming address? 50:02 So functional programming isn't like a panacea, right, 50:06 just like object oriented programming isn't. 50:11 Object oriented programming will not be applicable in 50:12 every single scenario that you find yourself in. 50:16 And functional programming won't either, but you can often do 50:18 things a lot faster and cleaner using a functional approach than you can in 50:22 writing a class or, well a class, yeah writing, you know, creating objects. 50:26 Or writing a function that maybe does all this work in one function. 50:32 And then you have to remember that is does all that stuff and, 50:36 what if something changes? 50:38 What if I no longer wanna do that step, I have to go edit this thing. 50:40 Functional programming, that mentality where you're writing things very small and 50:44 reusable, you get to the point where, if you're always writing like that, 50:48 editing your approaches to code becomes a lot simpler. 50:53 I just take out one function call. 50:55 Or I go change this one function, and I don't have to worry about 50:57 what happens elsewhere, because, you know, I'm paying attention to this one thing. 51:02 So it's, 51:06 the benefit it gives is that things are in smaller more digestible packages. 51:09 And, yeah, that's pretty much it, smaller packages. 51:12 >> So I guess it prevents the programmer from making a function and 51:18 having the function do like five or six things. 51:21 >> Can prevent somebody from [LAUGH] yeah, if if you have the the functional 51:25 programing mindset you're, you're used to always creating these small little, 51:30 I mean they almost seem pointless these little functions, cuz they're so tiny. 51:35 The ones that we wrote tonight are, 51:41 are very short, you'll ending up writing ones that are five, ten lines long. 51:42 But they do one thing, one major thing, right? 51:45 So it gets to where you stop trying to put everything into classes that 51:49 you create for, just, I need a thing that does a thing, right, so you build a class. 51:54 That works, but that's often way more overhead mentally and in your program. 51:59 And it stops you from writing functions that do 50 lines or a 100 lines of stuff. 52:04 Because you always break things up nice and small. 52:10 [NOISE] I mean, I'm gonna pick on Chris. 52:13 He's always done object-oriented stuff and when I first started learning 52:17 functional programming I'd be like, hey try and do this thing! 52:20 And he'd start writing a class. 52:23 Which worked [LAUGH] and it does it. 52:24 But it's like that's not the approach. 52:28 >> What. >> [LAUGH] Yes, that's why you're here. 52:31 All right, anybody else? 52:33 We're good? 52:38 All right, I will put the answers up in a branch, I'll try to 52:39 make sure they all have lambda options and stuff like that as well. 52:44 But, they, some of them may not. 52:47 And then, that's it. 52:49 That's our code. 52:50 [APPLAUSE] 52:51