Generating a single string with an Oxford comma from a list

Question

What are some clever (brief and idiomatic) approaches to taking a list of strings and returning a single properly punctuated string built from the list, with each element quoted.

This came up for me while experimenting with Groovy, for which my too-literal, but illustrative solution is

def temp = things.collect({"\'${it}\'"})
switch (things.size()) {
    case 1:
        result = temp[0]
        break
    case 2:
        result = temp.join(" and ")
        break
    default:
        result = temp.take(temp.size()-1).join(", ") + ", and " + temp[-1]
        break
}

That is, ['1'] should yield '1', ['1','2'] should yield '1 and 2', [see what I did there?] and ['1','2','3'] should yield '1, 2, and 3'.

I have some good answers for Groovy, but I'd like to see what other languages can do.

What are some compact clever approaches in various languages that take advantage of the features and idioms of those languages?

Answer 1

CSS, 132 116 115 bytes

a:not(:last-child):nth-child(n+2):after,a:nth-last-child(n+3):after{content:","}a+:last-child:before{content:"and "

<p>
  <a>one</a>
</p>
<p>
  <a>one</a>
  <a>two</a>
</p>
<p>
  <a>one</a>
  <a>two</a>
  <a>three</a>
</p>
<p>
  <a>one</a>
  <a>two</a>
  <a>three</a>
  <a>four</a>
</p>

a:not(:last-child):nth-child(n+2):after,a:nth-last-child(n+3):after{content:","}a+:last-child:before{content:"and "

CSS is not seen too often in code golf because it can only format text, but it actually works for this challenge and I thought it would be fun to do. See it in action using the snippet above (click "Show code snippet").

List should be in a linked HTML file with each element surrounded by <a> tags and separated by line breaks. The list items should be the only elements in their parent element, e.g.

<a>one</a>
<a>two</a>
<a>three</a>

Explanation

a:not(:last-child):nth-child(n+2)::after,
a:nth-last-child(n+3)::after {
    content: ",";
}

a + :last-child::before {
    content: "and ";
}

Let's consider the ungolfed version above. If you're not familiar with how CSS works, everything outside the curly braces is a selector that determines the set of HTML elements to which the declarations inside the braces apply. Each selector-declaration pair is called a rule. (It's more complicated than that but will suffice for this explanation.) Before any styling is applied, the list appears separated by only spaces.

We want to add commas after every word except the last, except for two-word lists, which get no commas. The first selector, a:not(:last-child):nth-child(n+2):after, selects all elements except the first and the last. :nth-child(n+2) is a shorter way of saying :not(:first-child), and it basically works by selecting elements whose index (starting at 1) is greater than or equal to 2. (Yes, it still confuses me a little. The MDN docs might help.)

Now we just need to select the first element to get a comma if there are three or more elements total. a:nth-last-child(n+3):after works like :nth-child, but counting from the back, so it selects all elements except the last two. The comma takes the union of the two sets, and we use the :after pseudo-element to add content immediately after each selected element.

The second rule is easier. We need to add "and" before the last element in the list, unless it is a single element. In other words, we need to select the last element that is preceded by another element. + is the adjacent sibling selector in CSS.

Answer 2

Haskell: 81, 77 74 chars

f[x]=x
f[x,y]=x++" and "++y
f[x,y,z]=x++", "++f[y++",",z]
f(x:y)=x++", "++f y

Haskell features: Pattern matching

Answer 3

Ruby, 47 bytes

q=" and ";p$*[2]?(a=$*.pop;$**", "+?,+q+a):$**q

Explanation

• The input are the command line arguments ($*).
• When $* has a third element ($*[2] does not return nil), take all elements minus the last one and turn them into a comma-separated String using Array#*. Finally add an extra comma, the string " and " and the last command line argument.
• When $* has no third element, two, one, or zero arguments were given. The arguments can safely be joined with the string " and " and produce the correct result.

Answer 4

Python 2 (61)

s=input()
d=s.pop()
print", ".join(s)+", and "[8-7*len(s):]+d

The main trick is to cut off part of the final joiner ", and " for one and two elements. For one, all of it is cut out, and for two, the comma is removed. This is done by slicing out [8-7*len(s):] (noting that s is one shorter after the pop).

Unfortunately, d cannot just be replaced with its expression or the pop would happen too late.

Answer 5

CSS, 62 chars 112 chars

Inspired by the other entries, even shorter. Note that it requires that A elements are not separated by white spaces:

a+a:before{content:", "}a+a:last-child:before{content:", and "

http://jsfiddle.net/olvlvl/1Ls79ocb/

Fix for the "one and two", as pointed at by Dennis:

a+a:before{content:", "}a+a:last-child:before{content:", and "}a:first-child+a:last-child:before{content:" and "

http://jsfiddle.net/olvlvl/1Ls79ocb/3/

Answer 6

Perl (v 5.10+) - 37 35 34 28

@F>2&&s/ /, /g;s/.* \K/and /

to be run with perl -ape, with the list supplied space separated on STDIN.

Output for various inputs:

$ perl -ape '@F>2&&s/ /, /g;s/.* \K/and /'
oxford
oxford
oxford cambridge
oxford and cambridge
oxford cambridge comma
oxford, cambridge, and comma
oxford cambridge comma space
oxford, cambridge, comma, and space

Answer 7

GNU sed - 69 chars including 1 for -r flag

s/ /, /g
s/( [^ ]+)( [^ ]+)$/\1 and\2/
s/^([^,]+),([^,]+)$/\1 and\2/

Takes a space-separated list (fairly idiomatic for shell scripts).

Example

$ sed -r -f oxfordcomma.sed <<< "1"
1
$ sed -r -f oxfordcomma.sed <<< "1 2"
1 and 2
$ sed -r -f oxfordcomma.sed <<< "1 2 3"
1, 2, and 3
$

Answer 8

Javascript (63)

l=a.length;l>2&&(a[l-1]='and '+a[l-1]);a.join(l>2?', ':' and ')

Cases:

• a = [1] => 1
• a = [1, 2] => 1 and 2
• a = [1, 2, 3] => 1, 2, and 3

Caveat: this will modify the last element in an array with length > 2.

Answer 9

Google Sheets, 67 68 67 92 bytes

=SUBSTITUTE(JOIN(", ",FILTER(A:A,A:A<>"")),",",IF(COUNTA(A:A)>2,",","")&" and",COUNTA(A:A)-1

The input starts at cell A1 and continuing down for however many entries exist.
JOIN merges them all into a string with a comma-space between each.
FILTER removes any non-blanks so you don't end up with infinite commas at the end.
SUBSTITUTE replaces the last comma (found by COUNTA counting the non-blank inputs).

It's not very exciting but it's doable in a single cell.

Answer 10

Groovy, 47 43 57 characters , JavaScript ES6 56 characters

Groovy:

(a[1]?a[0..-2].join(", ")+(a[2]?",":"")+" and ":"")+a[-1]

Since the array is filled with characters, we can replace a.size>1 by a[1]

JavaScript, ES6:

a.join(', ').replace(/,([^,]+)$/,`${a[2]?',':''} and$1`)

Both cases assume that variable a has the array in question.

Answer 11

Python 2 - 71, 70 68

s=input()
l=len(s)-1
print', '.join(s[:l])+', and '[l<2:]*(l>0)+s[l]

Character count including both, input and print.

Answer 12

CJam, 35 30 28 27 bytes

q~)\_"and "L?@+a+_,2=", ">*

This is a program that reads from STDIN and prints to STDOUT. Try it online.

How it works

q~                                     " Q := eval(input())                               ";
  )                                    " P := Q.pop()                                     ";
   \_"and "L?@+                        " P := (Q ? 'and ' : '') + P                       ";
                a+                     " Q += [P]                                         ";
                  _,2=", ">            " J := ', '[(len(Q) == 2):]                        ";
                           *           " R := J.join(Q)                                   ";
                                       " print R (implicit)                               ";

Example run

$ cjam <(echo 'q~)\_"and "L?@+a+_,2=", ">*') <<< '["1"]'; echo
1
$ cjam <(echo 'q~)\_"and "L?@+a+_,2=", ">*') <<< '["1""2"]'; echo
1 and 2
$ cjam <(echo 'q~)\_"and "L?@+a+_,2=", ">*') <<< '["1""2""3"]'; echo
1, 2, and 3
$ cjam <(echo 'q~)\_"and "L?@+a+_,2=", ">*') <<< '["1""2""3""4"]'; echo
1, 2, 3, and 4

Answer 13

Ruby, 57 bytes

f=->l{s=l*', ';s.sub(/,(?!.*,)/,(l.size<3?'':?,)+' and')}

I'm joining the string with , and then I'm replacing the last comma in the string with an and (and optional comma depending on list length).

Answer 14

Cobra - 60

do(l as String[])=l.join(', ',if(l.length<3,'',',')+' and ')

Cobra's List<of T>.join function lets you specify a different separator for the final two elements of the list, which is what makes this so short.

Answer 15

Batch - 151 Bytes

@echo off&set f=for %%a in (%~1)do
%f% set/aa+=1
%f% set/ac+=1&if !c!==1 (set o=%%a)else if !c!==!a! (set o=!o!, and %%a)else set o=!o!, %%a
echo !o!

Note; you have to call the script from cmd with the /v switch set as on, this is so I don't have to include the lengthy setLocal enableDelayedExpansion in the script. Otherwise add 30 to the byte count, and call the script normally.

h:\uprof>cmd /von /c test.bat "1 2 3"
1, 2, and 3

h:\uprof>cmd /von /c test.bat "1 2 3 4 5"
1, 2, 3, 4, and 5

Answer 16

PHP, 192 167 146 136 characters:

$n=' and ';$s=count($a);echo ($s<3)?join($n,$a):join(', ',array_merge(array_slice($a,0,$s-2),Array(join(",$n",array_slice($a,-2,2)))));

Based on a function I wrote years ago at http://www.christopherbloom.com/2011/05/21/join-implode-an-array-of-string-values-with-formatting/

Answer 17

Perl - 59

sub f{$a=join', ',@_;$#_&&substr$a,(@_>2)-3,@_<3,' and';$a}

Joins the list with commas, then if the list has more than one element, either adds ' and' after the last comma (if length >= 3), or replaces the last comma with it (if length == 2).

Answer 18

PHP, 86 84 chars

$a = ['one', 'two', 'three', 'four', 'five'];

With the array initialized, we start counting:

$L=count($a)-1;$S=', ';$L<2?($S=' and '):($a[$L]='and '.$a[$L]);echo implode($S,$a);

Prettified:

$last_index = count($a) - 1;
$separator = ', ';
if ($last_index < 2) {
    $separator = ' and ';
} else {
    $a[$last_index] = 'and '.$a[$last_index];
}
echo implode($separator, $a);

The last item in the list is modified. This should be OK because in PHP, array assignment and function calls perform copies.

Answer 19

Excel VBA, 108 Bytes

Anonymous VBE immediate window function that takes input as space delimited array from range A1 and outputs to the VBE immediate window.

x=Split([A1]):y=UBound(x):For i=0To y-2:?x(i)", ";:Next:If y>0Then?x(i)IIf(y>1,",","")" and "x(i+1)Else?[A1]

Answer 20

JavaScript (ES6) 60

Assuming no empty strings in input array

f=(a,b=a.pop())=>a[0]?a.join(', ')+(a[1]?',':'')+' and '+b:b

Test In FireFox/Firebug console

console.log(f(['We invited the stripper','JFK','Stalin']))

Output

 We invited the stripper, JFK, and Stalin

Answer 21

JavaScript - 60 56 71 67 63

Not exactly an idiomatic approach, but I had fun writing it and I like regex.

Assuming the array is stored in var a:

a.join((a.length>2?',':'')+' ').replace(/([^,\s])$/,"and $1")

Shortened by simply checking for index [2]: (yay boolean coercion)

a.join((!!a[2]?',':'')+' ').replace(/([^,\s])$/,'and $1')

Apparently I suck at testing and skipped a single-entry test. Here's the fixed version:

a.join((!!a[2]?',':'')+' ').replace(/([^,\s])$/,(!!a[1]?'and ':'')+'$1')

Shaved off 2 chars by inverting my booleans and 3 more by moving the space from the concatenation into the then/else of the first ternary:

a.join((!a[2]?' ':', ')).replace(/([^,\s])$/,(!a[1]?'':'and ')+'$1')

Thanks to @tomsmeding for reminding me that I don't have to coerce my booleans because JS does that for me. I also realised I forgot to remove the parentheses separating the first ternary from the concatenation inside the join():

a.join(a[2]?', ':' ').replace(/([^,\s])$/,(a[1]?'and ':'')+'$1')

Did I do that right? Obligatory new golfer apology.

Answer 22

Golfscript - 31 39 34 30

Hello, World! I am a long time lurker, first time poster. Here is my 30 byte solution in Golfscript (given the array such as [1 2 3] is already on the stack).

.,3<" and ":A{~A(;\+]", "}if*

The results are proper for all test cases.

EDIT: Take that, CJam!

Answer 23

Xojo, 52 71 83 chars

dim i as int8=ubound(L)
L(i)=if(i=0,"","and ")+L(i)
Return L.Join(if(i=1," ",", ")

Note that UBound is one less than the array length.

Answer 24

Ruby, 34 bytes

Lambda takes an array as input, and joins every element but the last with the string ,. Then the last element of the array is added on with an "and" between.

->a{a[0..-2]*", "+" and #{a[-1]}"}

Answer 25

Haskell, 67 bytes

f[n]=n
f[m,n]=m++" and "++n
f n=(init n>>=(++", "))++"and "++last n

Abuses the list monad a bit

Explanation

Doing this later

Answer 26

Racket 87

(define(f x)(string-join(map ~a x)", "#:before-last(if(= 2(length x))" and "", and ")))

Answer 27

Python 62 chars

Assuming that i is the list of strings:

(" and", ", and")[len(i) < 2].join(", ".join(i).rsplit(",",1))

Answer 28

C# 102 Chars

var i=m.Count(),l=i;if(l>2)l--;var r=string.Join(", ",m.Take(l));if(l!=i)r+=" and "+m.Last();return r;

Answer 29

Julia (53)

print(join(ARGS,", ",(endof(ARGS)>2?",":"")" and "))

Takes args from STDIN and outputs to STDOUT

Solution using Base.join(items, delim[, last])

Edit:

Test cases

julia oxfordComma.jl 1

1

julia oxfordComma.jl 1 2

1 and 2

julia oxfordComma.jl 1 2 3

1, 2, and 3

Answer 30

Rant (108)

[$[l:@b]:[r:each][s:[cmp:[rc];2;[is:different;,]]\s][before:[last:[notfirst:and\s]]][sync:;ordered][arg:b]]

Ungolfed:

[$[l:@b]:
    [r:each]                            # Repeat for each item
    [s:[cmp:[rc];2;[is:different;,]]\s] # Separate by comma if n > 2
    [before:[last:[notfirst:and\s]]]    # Insert "and" before last item
    [sync:;ordered]                     # Force forward order
    [arg:b]                             # Read list
]

Usage:

[$l:{A|B|C|D|E}]

Try it online

Answer 31

Pyth, 28

j>", "qlQ2+_t_Q+?"and "tQkeQ

Examples:

$ pyth programs/oxford.pyth <<< "['1']"
1

$ pyth programs/oxford.pyth <<< "['1','2']"
1 and 2

$ pyth programs/oxford.pyth <<< "['1','2','3']"
1, 2, and 3

Answer 32

PHP - 72

Set up the input:

$i=[1,2,3,4,5];

And then the process:

$x=count($i)-1;if($x) {$i[$x]=" and ".$i[$x];}echo join($x>1?",":"",$i);

Answer 33

bash (79)

(($#<3))&&echo $1 ${2/#/and }||{ Y=${@/%/,};L=${@: -1};echo ${Y/%$L,/and $L}; }

Explaination

For 1 or 2 parameters

• echo $1 and, if$2 exists, prefix $2 with "and ".

Otherwise

• suffix all parameters with a comma and join together
• get last parameter
• replace "$L," with "and$L"

Answer 34

PHP, 51 bytes

<?=preg_replace("#.*\K,#"," and ",join(",",$_GET));

Try it online!

Answer 35

PHP, 67

if($c=count($a)-1)$a[$c]='and '.$a[$c];echo join($c-1?', ':' ',$a);

If there's more than one item, prepend the last with 'and '. (store the count minus one in $c).

Next, join on space or comma space, depending on whether the count is two.

if ($c = count($a) - 1) {
    $a[$c] = 'and '.$a[$c];
}
echo join($c - 1 ? ', ' : ' ', $a);

Answer 36

SmileBASIC, 80 bytes

DEF L A
L=LEN(A)WHILE LEN(A)>1?SHIFT(A);","*(L>2);" ";
WEND?"and "*(L>1);A[0]END

Answer 37

Ruby, 44 bytes

->*f,l{f[0]?f*', '+(f[1]??,:'')+' and '+l:l}

Try it online!

Answer 38

Python, 79 chars

Note: the char count does not include the list declaration.

l=['1','2','3']
print l[0] if len(l)<2 else (', '.join(l[:-1])+(',' if len(l)>2 else '')+' and '+l[-1])

Output for various inputs:

1
1 and 2
1, 2, and 3
...

Answer 39

Golfscript (61)

The fact that Javascript and others score better pretty much shows that this is probably not done very efficiently in terms of golfing.

.,.2={;' and '*}{(\{\(.{.)', '''if}{', and '}if@\+print}/;}if

Examples with link to try it out:

• ["1"] -> "1"
• ["1" "2"] -> "1 and 2"
• ["1" "2" "3"] -> "1, 2, and 3"

Answer 40

Perl 5 + Moose, 70 bytes

sub english_list{@_<3?join" and ",@_:do{$_="and ".pop;join", ",@_,$_}}

Not many people seem to know it, but Moose, the popular OO framework for Perl 5, has a function to do this built in. (Moose uses it internally in its error messages, but exposes it as a documented, public function.)

my @l = (1 .. 3);
use Moose;
print Moose::Util::english_list(@l);

Excluding the definition of the list @l, that can be easily golfed down to 41 characters. If you allow the use of the say function which was introduced in Perl 5.10, you can knock off two more characters. (I'll leave it as an exercise for the reader.)

Now, looking at the definition of the english_list function itself, the best I've been able to golf it down to is 70 characters:

sub english_list{@_<3?join" and ",@_:do{$_="and ".pop;join", ",@_,$_}}

... though obviously by choosing a shorter function name, that could be reduced further. (59 characters if you choose a one-letter function name.)

Answer 41

CSS, 89

Inspired by the other entry, but shorter :)

a:before{content:", "}a:last-child:before{content:" and "}a:first-child:before{content:""

Demo: http://jsfiddle.net/01gy4Lh1/

Answer 42

Groovy - 221 chars

This is not a compact implementation, but the usage is tight: it overrides toString() for ArrayList via meta-class functionality (and maps - no conditionals here!).

This is powerful stuff, but idiomatic when writing a framework. That is, the change is transparent to the client.

Golfed (assumes variable a is defined):

q={s->"'$s'"}
r={q a[it]}
f={"${it[0..-2].join ", "}, and ${it[-1]}"}
m=[:].withDefault{f}
m[0]={""}
m[1]={r 0}
m[2]={"${r 0} and ${r 1}"}
ArrayList.metaClass.toString={->m[delegate.size()].call(delegate.collect{q(it)})}

Sample assertions:

a=['oxford','cambridge','comma','space']
assert "'oxford', 'cambridge', 'comma', and 'space'" == a.toString()

a=['oxford','cambridge']
assert "'oxford' and 'cambridge'" == a.toString()

a=['oxford']
assert "'oxford'" == a.toString()

a=[]
assert "" == a.toString()

Ungolfed:

q = {s->"'$s'"}
r = {q a[it]}

f = {"${it[0..-2].join ", "}, and ${it[-1]}"}
m = [:].withDefault{f}
m[0] = {""}
m[1] = {r 0}
m[2] = {"${r 0} and ${r 1}"}

ArrayList.metaClass.toString = { ->
    m[ delegate.size() ].call( delegate.collect{q(it)} )
}