Pandoc   a universal document converter

Scripting with pandoc

This version of the tutorial is for pandoc 1.11.

A simple example

Suppose you wanted to replace all level 2+ headers in a markdown document with regular paragraphs, with text in italics. How would you go about doing this?

A first thought would be to use regular expressions. Something like this:

perl -pe 's/^##+ (.*)$/\*\1\*/' source.txt

This should work most of the time. But don't forget that ATX style headers can end with a sequence of #s that is not part of the header text:

## My header ##

And what if your document contains a line starting with ## in an HTML comment or delimited code block?

<!--
## This is just a comment
-->

~~~~
### A third level header in standard markdown
~~~~

We don't want to touch these lines. Moreover, what about setext style second-level headers?

A header
--------

We need to handle those too. Finally, can we be sure that adding asterisks to each side of our string will put it in italics? What if the string already contains asterisks around it? Then we'll end up with bold text, which is not what we want. And what if it contains a regular unescaped asterisk?

How would you modify your regular expression to handle these cases? It would be hairy, to say the least. What we need is a real parser.

Well, pandoc has a real markdown parser, the library function readMarkdown. This transforms markdown text to an abstract syntax tree (AST) that represents the document structure. Why not manipulate the AST directly in a short Haskell script, then convert the result back to markdown using writeMarkdown?

First, let's see what this AST looks like. We can use pandoc's native output format:

% cat test.txt
### my header

text with *italics*
% pandoc -t native test.txt
Pandoc (Meta {docTitle = [], docAuthors = [], docDate = []})
[ Header 3 ("my-header",[],[]) [Str "my",Space,Str "header"]
, Para [Str "text",Space,Str "with",Space,Emph [Str "italics"]] ]

A Pandoc document consists of a Meta block (with title, authors, and date) and a list of Block elements. In this case, we have two Blocks, a Header and a Para. Each has as its content a list of Inline elements. For more details on the pandoc AST, see the haddock documentation for Text.Pandoc.Definition.

Here's a short Haskell script that reads markdown, changes level 2+ headers to regular paragraphs, and writes the result as markdown. If you save it as behead.hs, you can run it using runhaskell behead.hs. It will act like a unix pipe, reading from stdin and writing to stdout. Or, if you want, you can compile it, using ghc --make behead, then run the resulting executable behead.

-- behead.hs
import Text.Pandoc

behead :: Block -> Block
behead (Header n _ xs) | n >= 2 = Para [Emph xs]
behead x = x

transformDoc :: Pandoc -> Pandoc
transformDoc = bottomUp behead

readDoc :: String -> Pandoc
readDoc = readMarkdown def

writeDoc :: Pandoc -> String
writeDoc = writeMarkdown def

main :: IO ()
main = interact (writeDoc . transformDoc . readDoc)

The magic here is the bottomUp function, which converts our behead function (a function from Block to Block) to a transformation on whole Pandoc documents.

(bottomUp is so named because it operates on the AST from the bottom up, starting at the leaves and working up the branches to the trunk. Suppose you have a BulletList than contains another BulletList in one of its items. Then bottomUp f will operate first on the nested list, and only later on the containing list. topDown f would operate first on the containing list, and then on the nested list.)

Digression: reader and writer options

The behead.hs script uses default options for the reader and writer. If you want more control, you can modify these defaults. (See the definitions for ReaderOptions and WriterOptions in the haddock documentation for Text.Pandoc.Options.)

For example, the following variants will disable pandoc markdown extensions ("strict mode") and write markdown using reference-style links instead of inline links:

readDoc = readMarkdown def{ readerExtensions = strictExtensions }

writeDoc = writeMarkdown def{ writerExtensions = strictExtensions,
                            , writerReferenceLinks = True -- use ref-style links
                            }

Queries: listing URLs

We can use this same technique to do much more complex transformations and queries. Here's how we could extract all the URLs linked to in a markdown document (again, not an easy task with regular expressions):

-- extracturls.hs
import Text.Pandoc

extractURL :: Inline -> [String]
extractURL (Link _ (u,_)) = [u]
extractURL (Image _ (u,_)) = [u]
extractURL _ = []

extractURLs :: Pandoc -> [String]
extractURLs = queryWith extractURL

readDoc :: String -> Pandoc
readDoc = readMarkdown def

main :: IO ()
main = interact (unlines . extractURLs . readDoc)

queryWith is the query counterpart of bottomUp: it lifts a function that operates on Inline elements to one that operates on the whole Pandoc AST.

LaTeX for WordPress

Another easy example. WordPress blogs require a special format for LaTeX math. Instead of $e=mc^2$, you need: $LaTeX e=mc^2$. How can we convert a markdown document accordingly?

Again, it's difficult to do the job reliably with regexes. A $ might be a regular currency indicator, or it might occur in a comment or code block or inline code span. We just want to find the $s that begin LaTeX math. If only we had a parser...

We do. Pandoc already extracts LaTeX math, so:

-- wordpressify.hs
import Text.Pandoc

main = interact (writeMarkdown def .
                 bottomUp wordpressify .
                 readMarkdown def)
  where wordpressify (Math x y) = Math x ("LaTeX " ++ y)
        wordpressify x = x

Mission accomplished. (I've omitted type signatures here, just to show it can be done.)

Include files

So none of our transforms have involved IO. How about a script that reads a markdown document, finds all the inline code blocks with attribute include, and replaces their contents with the contents of the file given?

-- includes.hs
import Text.Pandoc

doInclude :: Block -> IO Block
doInclude cb@(CodeBlock (id, classes, namevals) contents) =
  case lookup "include" namevals of
       Just f     -> return . (CodeBlock (id, classes, namevals)) =<< readFile f
       Nothing    -> return cb
doInclude x = return x

main :: IO ()
main = getContents >>= bottomUpM doInclude . readMarkdown def
                   >>= putStrLn . writeMarkdown def

Try this on the following:

Here's the pandoc README:

~~~~ {include="README"}
this will be replaced by contents of README
~~~~

The trick here is bottomUpM, which is just a monadic version of bottomUp, and lifts doInclude from Block -> IO Block to Pandoc -> IO Pandoc.

Documentation on bottomUp, queryWith, bottomUpM, and queryWithM can be found in the haddock documentation for Text.Pandoc.Definition.

What if we want to remove every link from a document, retaining the link's text?

-- delink.hs
import Text.Pandoc

delink :: [Inline] -> [Inline]
delink ((Link txt _) : xs) = txt ++ delink xs
delink (x : xs)            = x : delink xs
delink []                  = []

main = interact (writeMarkdown def . bottomUp delink . readMarkdown def)

Note that delink can't be a function of type Inline -> Inline, because the thing we want to replace the link with is not a single Inline element, but a list of them. So we make delink a function from lists of Inline elements to lists of Inline elements. bottomUp can still lift this function to a transformation of type Pandoc -> Pandoc.

JSON reader and writer

Starting with version 1.8, the pandoc command-line tool has a json reader and writer. Both are very fast (especially compared to the native reader and writer). So, instead of writing a script (as shown above) that converts from one format to another, you may want to write a general purpose script that accepts JSON and produces JSON. This can then be used in a pipe with pandoc.

Here is an example, using the convenience function toJsonFilter from Text.Pandoc. toJsonFilter converts a function into a program that reads pandoc's JSON output from standard input, transforms it by walking the AST and applying the specified function, and writes the result as JSON to standard output.

import Text.Pandoc

main = toJsonFilter removeLink

removeLink :: Inline -> Inline
removeLink (Link xs _) = Emph xs
removeLink x = x

To use it to remove the links from a rst document:

pandoc -f rst -t json | runghc myscript.hs | pandoc -f json -t rst

If you need more speed, you can compile the script:

ghc --make myscript.hs
pandoc -f rst -t json | ./myscript | pandoc -f json -t rst

Note that the argument of toJsonFilter can be any type (a -> a), (a -> IO a), (a -> [a]), or (a -> IO [a]), where a is an instance of Data. So, for example, a can be Pandoc, Inline, Block, [Inline], [Block], Meta, ListNumberStyle, Alignment, ListNumberDelim, QuoteType, etc. Here's a filter that left-aligns all table cells:

import Text.Pandoc

main = toJsonFilter leftAlign

leftAlign :: Alignment -> Alignment
leftAlign _ = AlignLeft

A filter for ruby text

Finally, here's a nice real-world example, developed on the pandoc-discuss list. Qubyte wrote:

I'm interested in using pandoc to turn my markdown notes on Japanese into nicely set HTML and (Xe)LaTeX. With HTML5, ruby (typically used to phonetically read chinese characters by placing text above or to the side) is standard, and support from browsers is emerging (Webkit based browsers appear to fully support it). For those browsers that don't support it yet (notably Firefox) the feature falls back in a nice way by placing the phonetic reading inside brackets to the side of each Chinese character, which is suitable for other output formats too. As for (Xe)LaTeX, ruby is not an issue.

At the moment, I use inline HTML to achieve the result when the conversion is to HTML, but it's ugly and uses a lot of keystrokes, for example

<ruby><rt></rt><rp>(</rp><rt>はん</rt><rp>)</rp></ruby>

sets ご飯 "gohan" with "han" spelt phonetically above the second character, or to the right of it in brackets if the browser does not support ruby. I'd like to have something more like

r[はん](飯)

or any keystroke saving convention would be welcome.

We came up with the following script, which uses the convention that a markdown link with a URL beginning with a hyphen is interpreted as ruby:

[はん](-飯)
import Text.Pandoc
import System.Environment (getArgs)

handleRuby :: String -> Inline -> Inline
handleRuby format (Link [Str ruby] ('-':kanji,_)) =
  case format of
       "html"   -> RawInline "html" $ "<ruby>" ++ kanji ++ "<rp>(</rp><rt>"
                     ++ ruby ++ "</rt><rp>)</rp></ruby>"
       "latex"  -> RawInline "latex" $ "\\ruby{" ++ kanji ++ "}{"
                     ++ ruby ++ "}"
       "kanji"  -> Str kanji
       "kana"   -> Str ruby
       _        -> Str ruby  -- default
handleRuby _ x = x

main :: IO ()
main = do
 args <- getArgs
 toJsonFilter $ handleRuby $ case args of
                              [f] -> f
                              _   -> "kana" -- default

We compile our script:

ghc --make handleRuby

Then run it:

% pandoc -t json | ./handleRuby html | pandoc -f json -t html
[はん](-飯)
^D
<p><ruby>飯<rp>(</rp><rt>はん</rt><rp>)</rp></ruby></p>
% pandoc -t json | ./handleRuby latex | pandoc -f json -t latex
[はん](-飯)
^D
\ruby{飯}{はん}

Exercises

  1. Put all the regular text in a markdown document in ALL CAPS (without touching text in URLs or link titles).

  2. Remove all horizontal rules from a document.

  3. Renumber all enumerated lists with roman numerals.

  4. Replace each delimited code block with class dot with an image generated by running dot -Tpng (from graphviz) on the contents of the code block.

  5. Find all code blocks with class python and run them using the python interpreter, printing the results to the console.