When you are given a boring text, you just want to get it over with as quick as possible. Let's observe the following text:

Ens colligi timenda etc priorem judicem. De quascunque ii at contingere 
repugnaret explicetur intellectu. Adjuvetis hoc fortassis suspicari opportune 
obversari vix eam? Dei praemia prudens hominum iii constet requiri haberem. Ima 
sane nemo modi fuit lus pro dem haud. Vestro age negare tactum hoc cui lor. Ne et 
ut quod id soli soni deus. At constare innumera is occurret ea. Nia calebat seu 
acquiro fraudem effingo dicimus.

Note: This text has newlines for readability, while the test cases you need to handle don't have newline characters.

One way to skim text is to find questions that the writer has put into the text. For example, there is 1 question in the text above. Namely:

Adjuvetis hoc fortassis suspicari opportune obversari vix eam?

This can be done by splitting the text into sentences. A sentence will always end with one of the following punctuation symbols: .?! (others don't have to be handled). Other symbols than these are part of the sentence.

The Task

Given a line of text consisting of only

  • letters (A-Za-z)
  • numbers (0-9)
  • spaces
  • punctuation (,.;:?!) (a sentence cannot start with one of these).

Output the questions that are in the text in any reasonable format. You may assume that the text will always have at least 1 question. Outputting trailing and leading spaces before and after a question is allowed.

Important: Next to a punctuation symbol, there will never be another punctuation symbol (e.g. ?? is invalid and will not be required to handle).

Test cases

In the format:


The test cases:

Huh? I haven't heard what you just said. Could you repeat that please?
Could you repeat that please?

plz can i haz cheesburgr? i am cat pls.
plz can i haz cheesburgr?

This is a badly formatted question.Can u please help me,or my friends,with formatting this question    ?thankyou.
Can u please help me,or my friends,with formatting this question    ?


Does this question have a question mark? yes
Does this question have a question mark?

Why example.com resolves to in 99.9 percent of cases?
9 percent of cases?

A? b? c? d!

This is , so the submission with the lowest number of bytes wins!

  • 2
    \$\begingroup\$ .?! Others must not be handled, as your rules specify, but you have said others don't have to be handled. \$\endgroup\$ – Erik the Outgolfer Jul 20 '16 at 7:27
  • \$\begingroup\$ No testcase with more than one question? \$\endgroup\$ – edc65 Jul 20 '16 at 13:47
  • \$\begingroup\$ @edc65 The first test case has 2 questions, but I'll add another test case. \$\endgroup\$ – Adnan Jul 20 '16 at 13:48

11 Answers 11


Retina, 13 11 bytes


Try it online!

!`       print all matches
[^.?!]*  any number of non-ending-punctuation symbols
\?       followed by a question mark

Thanks to @MartinEnder for 2 bytes!


Python, 46 Bytes

import re
f=lambda s:re.findall("[^!?.]*\?",s)

Call with:

f("your string here")

output on tests:

['Can u please help me,or my friends,with formatting this question    ?', 'Can u please help me,or my friends,with formatting this question    ?', ' Huh?', ' Could you repeat that please?', ' plz can i haz cheesburgr?', 'd?', 'Does this question have a question mark?', '9 percent of cases?', 'A?', ' b?', ' c?']

another idea, 77 bytes (in python3 you'd need a list around filter):

import re
f=lambda s:filter(lambda x:x[-1]=="?",re.split("(?<=[\.\?!]).",s)))

I'm new to this so this, so this could probably be much shorter.

-17 (!) bytes thanks to Martin

-2 bytes by matching anything that is not "!","?" or "." (Getting close to the shell solutions, but I doubt I could save much more)

  • 1
    \$\begingroup\$ Welcome to Programming Puzzles and Code Golf! Very nice first answer :). \$\endgroup\$ – Adnan Jul 20 '16 at 10:50
  • \$\begingroup\$ I don't think you need that lookbehind at all and neither do you need to make the [\w,:; ]* ungreedy (because that group can't go past a punctuation character anyway), and then you also don't need to prepend . to your input. You calso shorten the remaining character class to [^.!?]. \$\endgroup\$ – Martin Ender Jul 20 '16 at 11:42
  • \$\begingroup\$ Thank you, Martin! I have tried around with this a bit, but didn't see the obvious. \$\endgroup\$ – KarlKastor Jul 20 '16 at 19:06

JavaScript, 35 24 bytes


Returns all substrings that start after a ., ?, or ! (or the beginning of the text) and end in a ?.

  • \$\begingroup\$ Urgh. And I thought I did good with 40 bytes. Good Job OP and @MartinEnder \$\endgroup\$ – MayorMonty Jul 29 '16 at 1:56

V, 12 bytes


Try it online!

A very straightforward answer.

Í             "Remove every occurrence, on every line
 [^.!?]       "Of any character that isn't '.', '!', or '?'
       *      "Repeated any number of times
        [.!]  "Followed by a a '.' or a '!'

Thankfully, handling newlines, or verifying all test-cases does not add any bytes.


Jelly, 16 bytes


Try it online! or verify all test cases

  • \$\begingroup\$ 28 bytes, isn't it? (16 UTF-8 chars) \$\endgroup\$ – Fabio Iotti Jul 20 '16 at 6:09
  • 6
    \$\begingroup\$ @bruce965 Jelly uses a custom code page that encodes each of the 256 characters it understands as the single byte each. The bytes link in the header points to it. \$\endgroup\$ – Dennis Jul 20 '16 at 6:10
  • \$\begingroup\$ Oh, cool! I'm not a codegolfer yet, so I'm not aware of this tricks, sorry for the question. \$\endgroup\$ – Fabio Iotti Jul 20 '16 at 6:33
  • 4
    \$\begingroup\$ @bruce965 For the record, it's not really a trick: the language could just as well use ISO 8859-1 (or some other existing single-byte encoding) and be just as powerful, but using a custom code page allows you to use more easily typable characters and better mnemonics than if you had to code with control characters for example. At the end of the day, it's just a stream of bytes, where every byte has been assigned some meaning. \$\endgroup\$ – Martin Ender Jul 20 '16 at 7:18
  • 2
    \$\begingroup\$ OK, "trick" might have sounded with a bad connotation, I should have said "stratagem" or something. I couldn't find any better word than "trick". \$\endgroup\$ – Fabio Iotti Jul 20 '16 at 7:55

Shell utilities, 43 38 bytes

Thanks to rexkogitans for trimming 5 bytes!

tr ? "\n"|sed "s/.*[\.!]//;s/.\+/&?/"

Pipe input in, like this:

echo Huh? I haven't heard what you just said. Could you repeat that please?|tr ? "\n"|sed "s/.*[\.!]//;s/.\+/&?/"

If if matters, I'm using:

  • GNU tr 5.3.0
  • GNU sed 4.2.1
  • Microsoft's cmd.exe, shipped with Windows 5.1.2600
  • \$\endgroup\$
    • 1
      \$\begingroup\$ I've never submitted an answer using multiple utilities before, so if I'm doing something wrong, let me know. \$\endgroup\$ – Joe Jul 19 '16 at 23:52
    • \$\begingroup\$ Do you use windows versions of the GNU utililties, or a shell for windows? \$\endgroup\$ – Erik the Outgolfer Jul 20 '16 at 8:02
    • \$\begingroup\$ tr ? "\n"|sed "s/.*[\.!]//;s/.\+/&?/" saves 5 Bytes (two added in tr and 7 saved in sed - this was tested in bash). /g is not necessary, as it is processed line by line. \$\endgroup\$ – rexkogitans Jul 20 '16 at 8:53
    • \$\begingroup\$ @EʀɪᴋᴛʜᴇGᴏʟғᴇʀ, I'm using the Windows ports of the GNU utilities. \$\endgroup\$ – Joe Jul 20 '16 at 21:15
    • \$\begingroup\$ @rexkogitans, thanks! I forgot about []; my first attempt at doing that looked something like \(\.|!\). \$\endgroup\$ – Joe Jul 20 '16 at 21:16

    Perl 5.10, 21 18 bytes (with -n flag)

    say m/[^?.!]+\?/g

    Straightforward implementation of the question.

    Try it here!

    • \$\begingroup\$ You can get rid of the leading m of your regex, and then you'll be able to remove the space between say and / \$\endgroup\$ – Dada Jul 22 '16 at 11:01

    Ruby 1.9, 17 bytes


    A 5 bytes program that must be invoked with the following command line options:

    • \$\begingroup\$ I didn't know Ruby flags could be wrestled with in such a manner, +1! Feels kind of odd, though, since consecutive questions will be together as one string within that array while other questions are separate, right? Unless there's a Ruby 1.9 quirk I'm not aware of. \$\endgroup\$ – Value Ink Jul 29 '16 at 7:14
    • \$\begingroup\$ @ValueInk The contents of the array will be concatenated so the program outputs a single string, not an array literal. You can try it out at golf.shinh.org/check.rb which has ruby 1.9. The flags can be set in the shebang. \$\endgroup\$ – xsot Jul 29 '16 at 9:26
    • \$\begingroup\$ Aha, that explains why you need 1.9 since 2.0 and up output it to look like an actual array. \$\endgroup\$ – Value Ink Jul 29 '16 at 21:11

    Batch, 174 bytes

    @echo off
    set t=
    set c=%s:~0,1%
    set t=%t%%c%
    if "%c%"=="?" echo %t%&set t=
    if "%c%"=="!" set t=
    if "%c%"=="." set t=
    set s=%s~1%
    if not "%s%"=="" goto l

    Reading a line from STDIN is a byte shorter than using set s=%*.


    PowerShell v4+, 43 bytes


    Really straightforward. Takes input $args and feeds that in as the first parameter to a .NET [regex]::Matches(...) static function. The regex we're matching is [^?!.]*\? -- that is, any number of non-sentence-ending characters that are followed by a question mark. The static function returns an array of objects detailing what capture group, index, etc., but we only want the .Values, so the return is encapsulated in parens and we call that property. This is where the v4+ requirement comes into play, as in prior versions you'd need to instead do something like a loop |%{$_.Value} or |Select Value to get the appropriate properties.

    Example without the parens and .Value

    PS C:\Tools\Scripts\golfing> .\what-exactly-was-your-question.ps1 "Huh? I haven't heard what you just said! Could you repeat that please?"
    Groups   : {Huh?}
    Success  : True
    Captures : {Huh?}
    Index    : 0
    Length   : 4
    Value    : Huh?
    Groups   : { Could you repeat that please?}
    Success  : True
    Captures : { Could you repeat that please?}
    Index    : 40
    Length   : 30
    Value    :  Could you repeat that please?

    Example with the parens and .Value

    PS C:\Tools\Scripts\golfing> .\what-exactly-was-your-question.ps1 "Huh? I haven't heard what you just said! Could you repeat that please?"
     Could you repeat that please?

    Python 3, 91 bytes

    def f(x,a=0):
     for n in range(len(x)):
      if x[n]in".!":a=n+1
      if x[n]is"?":print(x[a:n+1])

    Saves 1 byte in Python 2:

    def f(x,a=0):
     for n in range(len(x)):
      if x[n]in".!":a=n+1
      if x[n]is"?":print x[a:n+1]

    Your Answer

    By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

    Not the answer you're looking for? Browse other questions tagged or ask your own question.