Implementation of interpolated multiline strings


Version on this page:
LTS Haskell 22.26:
Stackage Nightly 2023-12-26:
Latest on Hackage:

See all snapshots qm-interpolated-string appears in

LicenseRef-PublicDomain licensed by Viacheslav Lotsmanov
Maintained by Viacheslav Lotsmanov
This version can be pinned in stack with:qm-interpolated-string-,2417

Module documentation for


Hackage Build Status

Implementation of interpolated multiline string QuasiQuoter that ignores indentation and trailing whitespaces.

Actually it’s modification of interpolatedstring-perl6 package. I’ve forked it to implemenent my own strings I really like.

This implementation based on qc from interpolatedstring-perl6 package but ignores any indentation, line breaks (except explicitly written using \n char) and trailing whitespaces.

  • ‘m’ in qm means ‘Multiline’.
  • ‘n’ in qn means ‘No interpolation’.
  • ‘b’ in qmb/qnb means ‘line Breaks’.
  • ‘s’ in qms/qns means ‘Spaces’.

Write a decoratively formatted string and your decorative indentation and line breaks wont go to result string, but when you really need it, you could just escape it using backslash.

Usage example

{-# LANGUAGE QuasiQuotes #-}

import Text.InterpolatedString.QM

main :: IO ()
main = do
  -- Hello, world! Pi is 3.14…
  putStrLn [qms| Hello,
                 Pi is {floor pi}.{floor $ (pi - 3) * 100}… |]

  -- Some examples with HTML below to demonstrate the difference
  -- between all of the quasi-quoters.

  let title = "Testing"
      text  = "Some testing text"

  -- <article><h1>Testing</h1><p>Some testing text</p></article>
  putStrLn [qm|

  -- <article><h1>{title}</h1><p>{text}</p></article>
  putStrLn [qn|

  -- <article> <h1>Testing</h1> <p>Some testing text</p> </article>
  putStrLn [qms|

  -- <article> <h1>{title}</h1> <p>{text}</p> </article>
  putStrLn [qns|

  -- <article>
  -- <h1>Testing</h1>
  -- <p>Some testing text</p>
  -- </article>
  putStrLn [qmb|

  -- <article>
  -- <h1>{title}</h1>
  -- <p>{text}</p>
  -- </article>
  putStrLn [qnb|


All QuasiQuoters

| QuasiQuoter | Interpolation | Indentation | Line breaks          | Trailing whitespaces |
| qm          | ✓             | Removed     | Removed              | Removed              |
| qn          | ✗             | Removed     | Removed              | Removed              |
| qmb         | ✓             | Removed     | Kept                 | Removed              |
| qnb         | ✗             | Removed     | Kept                 | Removed              |
| qms         | ✓             | Removed     | Replaced with spaces | Removed              |
| qns         | ✗             | Removed     | Replaced with spaces | Removed              |

About naming logic

| Contains in its name | What means                       | QuasiQuoters |
| m                    | Resolves interpolation blocks    | qm, qmb, qms |
| n                    | Without interpolation            | qn, qnb, qns |
| b                    | Keeps line breaks                | qmb, qnb     |
| s                    | Replaces line breaks with spaces | qms, qns     |

About interpolation blocks

Along with all specifics of any of the quoters (which supports interpolation blocks, which has m in their names) interpolation blocks work different. When curly bracket ({) opens everything inside until it closes (by }) is parsed as bare as possible to be given to haskell-src-meta without any modifications, to be parsed as bare haskell code.

But you might need use curly brackets inside an interpolation block. I don’t think it would be a good idea, because complicated logic there may cause code readability issues, but if you’re sure you need it then you get it. You just need to escape closing bracket to prevent interpolation block from closing, like this: \}. I know it could parsed and opening curly brackets inside could be used to prevent closing by next } symbol, but I chose do it this way to prevent any unobvious tricky behavior (e.g. consider } appear inside a string, [qm|foo {'x':'}':"y"} bar|], how that should be handled?). So I’ve decided to not make parser to be very smart, just to follow simple logic. You just need to explicitly escape every } symbol inside that isn’t closer of an interpolation block (you could find an example below).

About escaping

Symbols that can be escaped

Backslash is used for escaping these:

  1. \n - line break
  2. - space (space is supposed to be escaped when you’re going to keep some part of indentation)
  3. \↵ - backslash just before end of line cuts off line break (makes sense for qmb, qnb, qms and qns)
  4. \{ - opening bracket of interpolation block (only for qm, qmb and qms, to prevent interpolation and interpret this block as plain text)
  5. \t or \‣ (where is real tab symbol) - tab (escaping it to keep some part of indentation, or if you need tab symbol for some reason, escaping real tabs makes sense only for keeping some part of indentation)
  6. \\ - backslash itself (for situations when you don’t want to escape other symbols but just want backslash symbol, \\t, \\n, \\↵, \\{, etc., if backslash doesn’t come with any of symbols from this list it is interpreted just as backslash symbol, keep in mind that \\\ (without any of symbols from this list after) and \\\\ are producing same result - \\)
  7. \} - closing bracket inside an interpolation block (it works only inside opened interpolation block) to prevent interpolation block from closing (useful to escape records modification)

Escaping examples

[qm| foo\nbar  |] -- "foo\nbar"
[qm| foo\\nbar |] -- "foo\\nbar"
[qm| foo\tbar  |] -- "foo\tbar"
[qm| foo\\tbar |] -- "foo\\tbar"
[qm| foo\‣bar  |] -- "foo\tbar"   (`‣` is real tab symbol)
[qm| foo\\‣bar |] -- "foo\\\tbar" (`‣` is real tab symbol)
[qm| foo\ bar  |] -- "foo bar"
[qm| foo\\ bar |] -- "foo\\ bar"

[qm| foo\
     bar  |] -- "foobar"
[qm| foo\\
     bar  |] -- "foo\\bar"

[qmb| foo\
      bar  |] -- "foobar"
[qmb| foo\\
      bar  |] -- "foo\\\nbar"

[qm| foo\bar    |] -- "foo\\bar"
[qm| foo\\bar   |] -- "foo\\bar"
[qm| foo\\\bar  |] -- "foo\\\\bar"
[qm| foo\\\\bar |] -- "foo\\\\bar"

More examples

[qm|   you can escape spaces
     \ when you need them    |]
-- Result: "you can escape spaces when you need them"
        indentation and li
  ne bre
   aks are i
-- Result: "indentation and line breaks are ignored"
[qm|  \  You can escape indentation or\n
         line breaks when you need them! \  |]
-- Result: "  You can escape indentation or\nline breaks when you need them!  "
[qm| Interpolation blocks can be escaped too: {1+2} \{3+4} |]
-- Result: "Interpolation blocks can be escaped too: 3 {3+4}"

If you don’t need interpolation - just replace m to n in quasi-quoter name:

[qm| foo {1+2} |] -- Result: "foo 3"
[qn| foo {1+2} |] -- Result: "foo {1+2}"

[qms| foo {1+2} |] -- Result: "foo 3"
[qns| foo {1+2} |] -- Result: "foo {1+2}"

[qmb| foo {1+2} |] -- Result: "foo 3"
[qnb| foo {1+2} |] -- Result: "foo {1+2}"

That’s how you update some record inside interpolation block (you need to escape closing bracket):

{-# LANGUAGE QuasiQuotes #-}
import Text.InterpolatedString.QM (qm)
data Foo = Foo {bar :: Int, baz :: Int} deriving Show
main = let foo = Foo 10 20 in putStrLn [qm| Foo is: {foo {baz = 30\}} |]
-- Foo is: Foo {bar = 10, baz = 30}

Wanna make a contribution or maintain your own fork?

You can find some info for developers on wiki pages.


Viacheslav Lotsmanov


The Unlicense


  • Testing on GHC 8.2.2
  • More tests to cover more usage scenarios
  • Ability to use } inside an interpolation block by escaping it

WARNING! Breaking changes

  • \r characters are no longer pre-removed.

    Up to v0.2.1.0 all \r characters were pre-removed. When you compile your code with GHC you can use either LF or CRLF for line-breaks but not CR alone. When I changed handling of interpolation blocks (see below) I needed to get contents of interpolations blocks without any modifications, so I replaced pre-removing all CRs with explicit handling of CRLF in patterns. If your code ever was depending on \r symbols appearing alone inside quoters (that I can’t even imagine) it could break your code. But it will probably never happen, I’m just noticing it here.

  • Fix for interpolation blocks parsing.

    Once I noticed that [qm|{"\n"}|] compiles to "n", I considered this as a bug, I also realized that interpolation blocks aren’t interpreted as a bare haskell code as I was expecting. My bad, I’ve missed that, haven’t written enought tests to cover such scenarios, it migrated from original interpolatedstring-perl6 package. So I had to fix this mistake, notwithstanding it can break your code when you update the library. Now everything inside interpolation blocks is taken as bare haskell code as possible.

  • Support GHC 7.4.1
  • Internal modules moved to other-modules section

  • Added tab (\t) symbol escaping (breaks backward compatibility with v0.1.1.0)
  • Support new LTS Haskell 9.0 (ghc-8.0.2) (updated upper version of haskell-src-meta from 0.7.x to 0.8.x)
  • Added qmb QuasiQuoter, it’s qm + b (line-Breaks), it works just as qm but keeps line breaks (still ignores indendation)
  • Added qnb QuasiQuoter (qmb without interpolation), it’s qn + b (line-Breaks), it works just as qn but keeps line breaks (still ignores indendation)
  • Added qms QuasiQuoter, it’s qm + s (Spaces), it works just as qmb but kept line breaks replaced with spaces
  • Added qns QuasiQuoter (qms without interpolation), it’s qn + s (Spaces), it works just as qnb but kept line breaks replaced with spaces
  • More docs and tests

  • Added qn QuasiQuoter as alternative to qm but without interpolation
  • Some code refactoring
  • More docs and tests