Installing pandoc

Getting the pandoc source code

Source tarballs can be found at https://hackage.haskell.org/package/pandoc. For example, to fetch the source for version 1.17.0.3:

wget https://hackage.haskell.org/package/pandoc-1.17.0.3/pandoc-1.17.0.3.tar.gz
tar xvzf pandoc-1.17.0.3.tar.gz
cd pandoc-1.17.0.3

Or you can fetch the development code by cloning the repository:

git clone https://github.com/jgm/pandoc
cd pandoc

Note: there may be times when the development code is broken or depends on other libraries which must be installed separately. Unless you really know what you’re doing, install the last released version.

Quick stack method

The easiest way to build pandoc from source is to use stack:

1. Install stack. Note that Pandoc requires stack >= 1.7.0.

2. Change to the pandoc source directory and issue the following commands:

   stack setup
   stack install

   stack setup will automatically download the ghc compiler if you don’t have it. stack install will install the pandoc executable into ~/.local/bin, which you should add to your PATH. This process will take a while, and will consume a considerable amount of disk space.

Quick cabal method

1. Install the Haskell platform. This will give you GHC and the cabal-install build tool. Note that pandoc requires GHC >= 7.10 and cabal >= 2.0.

2. Update your package database:

   cabal update

3. Check your cabal version with

   cabal --version

   If you have a version less than 2.0, install the latest with:

   cabal install cabal-install

4. Use cabal to install pandoc and its dependencies:

   cabal install pandoc

   This procedure will install the released version of pandoc, which will be downloaded automatically from HackageDB.

   If you want to install a modified or development version of pandoc instead, switch to the source directory and do as above, but without the ‘pandoc’:

   cabal install

5. Make sure the $CABALDIR/bin directory is in your path. You should now be able to run pandoc:

   pandoc --help

   Not sure where $CABALDIR is?

6. By default pandoc uses the “i;unicode-casemap” method to sort bibliography entries (RFC 5051). If you would like to use the locale-sensitive unicode collation algorithm instead, specify the icu flag (which affects the dependency citeproc):

   cabal install pandoc -ficu

   Note that this requires the text-icu library, which in turn depends on the C library icu4c. Installation directions vary by platform. Here is how it might work on macOS with Homebrew:

   brew install icu4c
   stack install pandoc \
     --flag "citeproc:icu" \
     --extra-lib-dirs=/usr/local/opt/icu4c/lib \
     --extra-include-dirs=/usr/local/opt/icu4c/include

7. The pandoc.1 man page will be installed automatically. cabal shows you where it is installed: you may need to set your MANPATH accordingly. If MANUAL.txt has been modified, the man page can be rebuilt: make man/pandoc.1.

Custom cabal method

This is a step-by-step procedure that offers maximal control over the build and installation. Most users should use the quick install, but this information may be of use to packagers. For more details, see the Cabal User’s Guide. These instructions assume that the pandoc source directory is your working directory. You will need cabal version 2.0 or higher.

1. Install dependencies: in addition to the Haskell platform, you will need a number of additional libraries. You can install them all with

   cabal update
   cabal install --only-dependencies

2. Configure:

   cabal configure --prefix=DIR --bindir=DIR --libdir=DIR \
     --datadir=DIR --libsubdir=DIR --datasubdir=DIR --docdir=DIR \
     --htmldir=DIR --program-prefix=PREFIX --program-suffix=SUFFIX \
     --mandir=DIR --flags=FLAGSPEC --enable-tests

   All of the options have sensible defaults that can be overridden as needed.

   FLAGSPEC is a list of Cabal configuration flags, optionally preceded by a - (to force the flag to false), and separated by spaces. Pandoc’s flags include:

   • embed_data_files: embed all data files into the binary (default no). This is helpful if you want to create a relocatable binary.

   • lua53: embed support for Lua 5.3 instead of 5.4.

3. Build:

   cabal build
   cabal test

4. Build API documentation:

   cabal haddock --html-location=URL --hyperlink-source

5. Copy the files:

   cabal copy --destdir=PATH

   The default destdir is /.

6. Register pandoc as a GHC package:

   cabal register

   Package managers may want to use the --gen-script option to generate a script that can be run to register the package at install time.

Creating a relocatable binary

It is possible to compile pandoc such that the data files pandoc uses are embedded in the binary. The resulting binary can be run from any directory and is completely self-contained. With cabal, add -fembed_data_files to the cabal configure or cabal install commands.

With stack, use --flag pandoc:embed_data_files.

Running tests

Pandoc comes with an automated test suite. To run with cabal, cabal test; to run with stack, stack test.

To run particular tests (pattern-matching on their names), use the -p option:

cabal install pandoc --enable-tests
cabal test --test-options='-p markdown'

Or with stack:

stack test --test-arguments='-p markdown'

It is often helpful to add -j4 (run tests in parallel) and --hide-successes (don’t clutter output with successes) to the test arguments as well.

If you add a new feature to pandoc, please add tests as well, following the pattern of the existing tests. The test suite code is in test/test-pandoc.hs. If you are adding a new reader or writer, it is probably easiest to add some data files to the test directory, and modify test/Tests/Old.hs. Otherwise, it is better to modify the module under the test/Tests hierarchy corresponding to the pandoc module you are changing.

Running benchmarks

To build and run the benchmarks:

cabal configure --enable-benchmarks && cabal build
cabal bench

or with stack:

stack bench

To use a smaller sample size so the benchmarks run faster:

cabal bench --benchmark-options='-s 20'

To run just the markdown benchmarks:

cabal bench --benchmark-options='markdown'