An image of Heer de Beer

Heer de Beer.org

Exploring the Computational Medium

Pandocomatic

Automate the use of pandoc

Huub de Beer

April 6th, 2019

Part I. Introduction

Pandocomatic automates the use of pandoc. With pandocomatic you can express common patterns of using pandoc for generating your documents. Applied to a directory, pandocomatic acts as a static site generator. For example, this manual is generated with pandocomatic!

Pandocomatic is free software; pandocomatic is released under the GPLv3. You will find the source code of pandocomatic in its repository on Github.

Note. Pandocomatic is build on top of paru, which is a wrapper around pandoc.

Chapter 1. Acknowledgements

Thanks to Ian for his contribution of patches, bug reports, fixes, and suggestions. With your help pandocomatic is growing beyond a simple tool for personal use into a useful addition to the pandoc ecosystem.

Chapter 2. Installation

Pandocomatic is a Ruby program and can be installed through RubyGems as follows:

gem install pandocomatic

This will install pandocomatic and paru, a Ruby wrapper around pandoc. To use pandocomatic, you also need a working pandoc installation. See pandoc’s installation guide for more information about installing pandoc.

You can also download the latest gem, pandocomatic-1.1.2, from Github and install it manually as follows:

cd /directory/you/downloaded/the/gem/to
gem install pandocomatic-1.1.2.gem

Chapter 3. Why pandocomatic?

I use pandoc a lot. I use it to write all my papers, notes, websites, reports, outlines, summaries, and books. Time and again I was invoking pandoc like:

pandoc --from markdown \
  --to html \
  --standalone \
  --csl apa.csl \
  --bibliography my-bib.bib \
  --mathjax \
  --output result.html \
  source.md

Sure, when I write about history, the CSL file and bibliography change. And I do not need the --mathjax option like I do when I am writing about mathematics education. Still, all these invocations are quite similar.

I already wrote the program do-pandoc.rb as part of a Ruby wrapper around pandoc, paru. Using do-pandoc.rb I can specify the options to pandoc in a metadata block in the source file itself. With do-pandoc.rb the invocation above is simplified to:

do-pandoc.rb source.md

It saves me from typing out the whole pandoc invocation each time I run pandoc on a source file. However, I have still to setup the same options to use in each document that I am writing, even though these options do not differ that much from document to document.

Pandocomatic is a tool to re-use these common configurations by specifying a so-called pandocomatic template in a YAML configuration file. For example, by placing the following file, pandocomatic.yaml, in pandoc’s data directory:

templates:
  education-research:
    preprocessors: []
    pandoc:
      from: markdown
      to: html
      standalone: true
      csl: 'apa.csl'
      toc: true
      bibliography: /path/to/bibliography.bib
      mathjax: true
    postprocessors: []

In this configuration file a single pandocomatic template is being defined: education-research. This template specifies that the source files it is applied to are not being preprocessed. Furthermore, the source files are converted with pandoc by invoking pandoc --from markdown --to html --standalone --csl apa.csl --toc --bibliography /path/to/bibliography.bib --mathjax. Finally, the template specifies that pandoc’s output is not being postprocessed.

I now can create a new document that uses this template by including the following metadata block in my source file, on_teaching_maths.md:

 ---
 title: On teaching mathematics
 author: Huub de Beer
 pandocomatic_:
   use-template: education-research
   pandoc:
     output: on_teaching_mathematics.html
 ...
 
 Here goes the contents of my new paper ...

To convert this file to on_teaching_mathematics.html I run pandocomatic:

pandocomatic -i on_teaching_maths.md

With just two extra lines in a metadata block I can tell pandocomatic what template to use when converting a file. You can also use multiple templates in a document, for example to convert a markdown file to both HTML and PDF. Adding file-specific pandoc options to the conversion process is as easy as adding a pandoc property with those options to the pandocomatic_ metadata property in the source file like I did with the output property in the example above.

Once I had written a number of related documents this way, it was a small step to enable pandocomatic to convert directories as well. Just like that, pandocomatic can be used as a static site generator!


Part II. Using pandocomatic: Quick start and overview

Chapter 4. Converting a single document

Pandocomatic allows you to put pandoc command-line options in the document to be converted itself. Instead of a complex pandoc command-line invocation, pandocomatic allows you to convert your markdown document hello_world.md with just:

pandocomatic hello_world.md

Pandocomatic starts by extracting the YAML metadata blocks in hello_world.md, looking for a pandocomatic_ property. If such a property exists, it is treated as an internal pandocomatic template and the file is converted according to that pandocomatic template. For more information about pandocomatic templates, see the chapter about templates later in this manual.

For example, if hello_world.md contains the following pandoc markdown text:

 ---
 title: My first pandocomatic-converted document
 pandocomatic_:
     pandoc:
         to: html
 ...
 
 Hello *World*!

pandocomatic is instructed by the pandoc property to convert the document to the HTML file hello_world.html. If you would like to instruct pandocomatic to convert hello_world.md to goodday_world.html instead, use command-line option --output goodday_world.html. For more information about pandocomatic’s command-line options, see the chapter about command-line options.

You can tell pandocomatic to apply any pandoc command-line option in a template’s pandoc property. For example, to use a custom pandoc template and add a custom CSS file to the generated HTML, extend the pandoc property above as follows:

pandoc:
    to: html
    css:
    -   style.css
    template: hello-template.html

Besides the pandoc property to configure the pandoc conversion, pandocomatic templates can also contain a list of preprocessors and a list of postprocessors. Preprocessors are run before the document is converted with pandoc and postprocessors are run afterwards (see the Figure below):

Figure 4.1 How pandocomatic works: a simple conversion
How pandocomatic works: a simple conversion

For example, you can use the following script to clean up the HTML generated by pandoc:

 #!/bin/bash
 tidy -quiet -indent -wrap 78 -utf8

This script tidy.sh is a simple wrapper script around the html-tidy program. To tell pandocomatic to use it as a postprocessor, you have to change the pandocomatic_ property to:

pandocomatic_:
    pandoc:
        to: html
        css:
        -   style.css
        template: hello-template.html
    postprocessors:
    - ./tidy.sh

The “./” in the path ./tidy.sh tells pandocomatic to look for the tidy.sh script in the same directory as the file to convert. You can also specify an absolute path (starting with a slash “/”) or a path relative to the pandocomatic data directory like we do in the path in the template property in the example above. See the Section about specifying paths in pandocomatic for more information. If you use a path relative to the pandocomatic data directory, you have to use the --data-dir option to tell pandocomatic where to find its data directory. If you do not, pandocomatic will default to pandoc’s data directory.

To convert the example with a data directory, use:

pandocomatic --data-dir my_data_dir hello_world.md

Like pandoc, pandocomatic does support multiple input files. These input files are concatenated by pandocomatic and then treated as a single input file. For example, instead of writing a book in one big markdown file, you could separate the chapters into separate markdown files. To generate the final book, invoke pandocomatic like:

pandocomatic -i frontmatter.md -i c01.md -i c02.md -i c03.md -i c04.md -o book.html

Note. If multiple files do have a pandocomatic_ property in their metadata blocks, only the first pandocomatic_ property is used; all other occurrences are discarded. If this happens, pandocomatic will show a warning.

Chapter 5. Converting a series of documents

5.1 Using external templates

Adding an internal pandocomatic template to a markdown file helps a lot by simplifying converting that file with pandoc. Once you start using pandocomatic more and more, you will discover that most of these internal pandocomatic templates are a lot alike. You can re-use these internal pandocomatic templates by moving the common parts to an external pandocomatic template.

External pandocomatic templates are defined in a pandocomatic configuration file. A pandocomatic configuration file is a YAML file. Templates are specified in the templates property as named sub properties. For example, the internal pandocomatic template specified in the hello_world.md file (see previous chapter) can be specified as the external pandocomatic template hello in the pandocomatic configuration file my-config.yaml as follows:

 templates:
     hello:
         pandoc:
             to: html
             css:
                 - ./style.css
             template: hello-template.html
         postprocessors:
             - ./tidy.sh

You use it in a pandoc markdown file by specifying the use-template sub property in the pandocomatic_ property. The hello_world.md example then becomes:

  ---
  title: My second pandocomatic-converted document
  pandocomatic_:
      use-template: hello
  ...
  
  Hello *World*!

To convert external_hello_world.md you need to tell pandocomatic where to find the external pandocomatic template via the --config command-line option. For example, to convert external_hello_world.md to out.html, use:

pandocomatic -d my_data_dir --config my-config.yaml -i external_hello_world.md -o out.html

5.2 Customizing external templates with an internal template

Because you can use an external pandocomatic templates in many files, these external templates tend to setup more general aspects of a conversion process. You can customize a such a general conversion process in a specific document by extending the internal pandocomatic template. For example, if you want to apply a different CSS style sheet and add a table of contents, customize the hello template with the following internal pandocomatic template:

pandocomatic_:
    use-template: hello
    pandoc:
        toc: true
        css:
            remove:
            - './style.css'
            add:
            - './goodbye-style.css'

hello’s pandoc property is extended with the --toc option, the style.css is removed, and goodbye-style.css is added. If you want to add the goodbye-style.css rather than having it replace style.css, you specify:

css:
    -   './goodbye-style.css'

Lists and properties in internal pandocomatic templates are merged with external pandocomatic templates; simple values, such as strings, numbers, or Booleans, are replaced. Besides a template’s pandoc property you can also customize any other property of the template.

5.3 Extending templates

In a similar way that an internal pandocomatic template extends an external pandocomatic template you can also extend an external pandocomatic template directly in the pandocomaitc configuration file. For example, instead of customizing the hello template, you could also have extended hello as follows:

 templates:
     hello:
         pandoc:
             to: html
             css:
                 - ./style.css
             template: hello-template.html
         postprocessors:
             - ./tidy.sh
     goodbye:
         extends: ['hello']
         pandoc:
             toc: true
             css:
                 - ./goodbye-style.css

The ‘goodbye’ template extends the hello template. A template can extend multiple templates. For example, you could write a template author in which you configure the author metadata variable:

templates:
    author:
        metadata:
            author: Huub de Beer
    ...

This author template specifies the metadata property of a template. This metadata will be mixed into each document that uses this template. If you want the goodbye template to also set the author automatically, you can change its extends property to:

templates:
    ...
    goodbye:
        extends: ['author', 'hello']
        ...

Setting up templates by extending other smaller templates makes for a modular setup. If you share your templates with someone else, they only have to change the author template in one place to use their own names on all their documents while using your templates.

See the Section on extending pandocomatic templates for more information about this extension mechanism.

Chapter 6. Converting a directory tree of documents

Once you have created a number of documents that can be converted by pandocomatic and you change something significant in one of the external pandocomatic templates, you have to run pandocomatic on all of the documents again to propagate the changes. That is fine for a document or two, but more than that and it becomes a chore.

For example, if you change the HTML template hello-template.html in the pandocomatic data directory, or switch to another template, you need to regenerate all documents you have already converted with the old HTML template. If you run pandocomatic on an input directory rather than on an input file, it will convert all files in that directory, recursively.

Thus, to convert the example files used in this chapter, you can run

pandocomatic -d my_data_dir -c my-extended-config.yaml -i manual -o output_dir

It will convert all files in the directory manual and place the generated documents in the output directory output_dir.

From here it is but a small step to use pandocomatic as a static-site generator. For that purpose some configuration options are available:

With these features, you can (re)generate a website with the pandocomatic invocation:

pandocomatic -d data_dir -c intitial_config.yaml -i src -o www --modified-only

For more detailed information about pandocomatic, please see the Reference section of this manual.


Part III. Reference: All about pandocomatic

Chapter 7. Pandocomatic command-line interface

Pandocomatic takes a number of arguments which should at least include the input file or directory. The general form of a pandocomatic invocation is:

pandocomatic options [INPUT]

7.1 General arguments: help and version

-v, --version

Show the version. If this option is used, all other options are ignored.

-h, --help

Show a short help message. If this option is used, all other options except --version or -v are ignored.

7.2 Input/output arguments

-i PATH, --input PATH

Convert PATH. If this option is not given, INPUT is converted. INPUT and --input or -i cannot be used together. You can use this option multiple times, denoting to concatenate each input file in the order they are specified on the command-line. Pandocomatic treats the concatenated files as a single input file.

-o PATH, --output PATH

Create converted files and directories in PATH.

You can specify the output file in the metadata of a pandoc markdown input file. In that case, you can omit the output argument. Furthermore, if no output file is specified whatsoever, pandocomatic defaults to output to HTML by replacing the extension of the input file with html.

The input and output should both be files or both be directories. Pandocomatic will complain if the input and output types do not match.

-s, --stdout

Print result of converstion to standard output.

You cannot combine this option with --output or with a directory as input.

7.3 Arguments to configure pandocomatic

-d DIR, --data-dir DIR

Configure pandocomatic to use DIR as its data directory. The default data directory is pandoc’s data directory. (Run pandoc --version to find pandoc’s data directory on your system.)

-c FILE, --config FILE

Configure pandocomatic to use FILE as its configuration file during the conversion process. Default is DATA_DIR/pandocomatic.yaml.

7.4 Arguments to change how pandocomatic operates

-m, --modified-only

Only convert files that have been modified since the last time pandocomatic has been run. Or, more precisely, only those source files that have been updated at a later time than the corresponding destination files will be converted, copied, or linked. Default is false.

-q, --quiet

By default pandocomatic is verbose when you convert a directory. It tells you about the number of commands to execute. When executing these commands, pandocomatic tells you what it is doing, and how many commands still have to be executed. Finally, when pandocomatic is finished, it tells you how long it took to perform the conversion.

If you do not like this verbosity, use the --quiet or -q argument to run pandocomatic quietly. Default is false.

-y, --dry-run

Inspect the files and directories to convert, but do not actually run the conversion. Default is false.

-b, --debug

Run pandocomatic in debug mode. At the moment this means that all pandoc invocations are printed as well.

-r PATH, --root-path PATH

Set the root path for use with the root path relative path specification in templates (see Specifying paths). It is used mostly with the –css pandoc option. It defaults to the directory of the specified output.

Note. This option is experimental.

7.5 Status codes

When pandocomatic runs into a problem, it will return with status codes 1266 or 1267. The former is returned if something goes wrong before any conversion is started and the latter when something goes wrong during the conversion process.

Chapter 8. Pandocomatic configuration

Pandocomatic can be configured by means of pandocomatic configuration files, which are YAML files. For example, the following YAML code is a valid pandocomatic configuration file:

 settings:
     data-dir: ~/my_data_dir
     recursive: true
     follow-links: false
     skip: ['.*']
 templates:
     webpage:
         glob: ['*.md']
         pandoc:
             to: html
             toc: true
             css:
                 - assets/style.css
         postprocessors:
             - postprocessors/tidy.sh

By default, pandocomatic looks for the configuration files in both the pandoc data directory and the pandocomatic data directory; by convention, pandocomatic expects these files to be named pandocomatic.yaml.

On top of that, you can tell pandocomatic to use a specific configuration file via the command-line option --config. For example, if you want to use a configuration file my-config.yaml, invoke pandocomatic as follows:

pandocomatic --config my-config.yaml some-file-to-convert.md

A pandocomatic configuration file contains two properties:

  1. global settings
  2. external templates

I discuss these two properties next. Before I do, however, I describe pandocomatic’s configuration hierarchy. For more in-depth information about pandocomatic templates, please see the Chapter on pandocomatic templates.

8.1 Pandocomatic’s configuration hierarchy

If you’re using pandocomatic a lot, you might have collected many useful templates that you often use. Instead of having to copy them to a new configuration file each time you start a new project, you can put them in a configuration file in one of pandocomatic’s data directories. Pandocomatic will construct a configuration hierarchy based on these more global configuration files as follows:

  1. This configuration hierarchy starts with the default configuration that’s part of pandocomatic. This is the base configuration that’s always there.

  2. Then pandocomatic mixes in the configuration files named pandocomatic.yaml found in the data directories in the following order:

    1. The configuration file in the pandocomatic data directory specified on the command line with option “–data-dir” or “-d”.
    2. The configuration file in the pandoc data directory, if it exists. Otherwise, as a fall-back, the current working directory is used. Note, a missing pandoc data directory is likely a sign of a broken pandoc installation.
  3. Finally, the configuration file given on the command line with the “–config” or “-c” option is mixed in.

Pandodomatic always constructs the configuration hierarchy in this exact order, skipping any configuration that’s missing. Thus, the default configuration is extended first by the configuration from pandocomatic’s data directory, then by the configuration from pandoc’s data directory, and finally by the configuration specified with the --config option.

If you run pandocomatic with command-line option --verbose, it should print this configuration hierarchy.

8.2 Settings

You can configure five optional global settings:

  1. data-dir
  2. match-files
  3. skip
  4. recursive
  5. follow-links

The latter three are used only when converting a whole directory tree with pandocomatic. These are discussed in the next sub section.

The first setting, data-dir (String), tells pandocomatic where its data directory is. You can also specify the pandocomatic data directory via the command-line option --data-dir. For example, if you want to use ~/my-data-dir as the pandocomatic data directory, invoke pandocomatic as follows:

pandocomatic --data-dir ~/my-data-dir some-file-to-convert.md

If no pandocomatic data directory is specified whatsoever, pandocomatic defaults to pandoc’s data directory.

Any directory can be used as a pandocomatic data directory, there are no conventions or requirements for this directory other than being a directory. However, it is recommended to create a meaningful sub directory structure. For example, a sub directory for processors, filters, CSL files, and pandoc templates makes it easier to manage and point to these assets.

The setting match-files controls how pandocomatic selects the template to use to convert a file. Possible values for match-files are first and all. Pandocomatic matches a file to a template as follows:

  1. If the file has one or more use-template statements in the pandocomatic metadata, it will use these specified templates.

  2. However, if no such templates are specified in the file, pandocomatic tries to find global templates as follows:

    1. If the setting match-files has value all, all templates with a glob pattern that matches the input filename are used to convert that input file. For example, you can specify a template www to convert *.md files to HTML and a template pdf to convert *.md to PDF. In this case, a markdown file will be converted to both HTML and PDF. For example, you could use this to generate a website with a print PDF page for each web page.

    2. If the setting match-files has value first, the first template with a glob pattern that matches the input file is used to convert the file.

      This is the default.

Configuring converting a directory tree

You can convert a directory tree by invoking pandocomatic with a single directory as the input rather than one or more files. Of course, once you start converting directories, more fine-grained control over what files to convert than “convert all files” is useful. There are four settings you can use to control which files to convert. Three of them are global settings, the other one is the glob property of an external pandocomatic template. The glob property is discussed later.

The three global settings to control which files to convert are:

  1. recursive (Boolean), which tells pandocomatic to convert sub directories or not. This setting defaults to true.
  2. follow-links (Boolean), which tells pandocomatic to treat symbolic links as files and directories to convert or not. This setting defaults to false.
  3. skip (Array of glob patterns), which tells pandocomatic which files not to convert at all. This setting defaults to ['.*', 'pandocomatic.yaml']: ignore all hidden files (starting with a period) and also ignore default pandocomatic configuration files.

Default configuration

Pandocomatic’s default configuration file is defined in the file lib/pandocomatic/default_configuration.yaml. This default configuration is used when

When converting a directory tree, each time pandocomatic enters a (sub) directory, it also looks for a default configuration file to update the current settings. In other words, you can have pandocomatic behave differently in a sub directory than the current directory by putting a pandocomatic.yaml file in that sub directory that changes the global settings or external pandocomatic templates.

8.3 Templates

Besides the global settings property, a pandocomatic configuration file can also contain a templates property. In the templates property you define the external pandocomatic templates you want to use when converting files with pandocomatic. Pandocomatic templates are discussed in detail in the Chapter on pandocomatic templates. The glob property of external pandocomatic templates is related to configuring pandocomatic when converting a directory. It tells pandocomatic which files in a directory are to be converted with a template.

If you look at the example pandocomatic configuration file at the start of this chapter, you see that the webpage template is configured with property glob: ['*.md']. This tells pandocomatic to apply the template webpage to all markdown files with extension .md. In other words, given a directory with the following files:

directory/
+ sub directory/
| + index.md
- index.md
- image.png 

Running pandocomatic with the example pandocomatic configuration file will result in the following result”

directory/
+ sub directory/
| + index.html
- index.html
- image.png 

That is, all .md files are converted to HTML and all other files are copied, recursively.

Chapter 9. Pandocomatic templates

Pandocomatic automates the use of pandoc by extracting common patterns of using pandoc into so called pandocomatic templates. You can then apply these templates to your documents. As described in Part II, there are internal and external pandocomatic templates. The difference between these two types of templates is their scope: internal pandocomatic templates only affect the document they are defined in, whereas external pandocomatic templates, which are defined in a pandocomatic configuration file, affect all documents that use that template.

Although you can create an one-off internal pandocomatic template for a document—sometimes you just have an odd writing project that differs too much from your regular writings—, most often you use an external pandocomatic template and customize it in the internal pandocomatic template.

In this Chapter the definition, extension, customization, and use of templates are discussed in detail.

9.1 Defining a template

An external pandocomatic template is defined in the templates property of a pandocomatic configuration file. For example, in the following YAML code, the template webpage is defined:

 settings:
     data-dir: ~/my_data_dir
     recursive: true
     follow-links: false
     skip: ['.*']
 templates:
     webpage:
         glob: ['*.md']
         pandoc:
             to: html
             toc: true
             css:
                 - assets/style.css
         postprocessors:
             - postprocessors/tidy.sh

Each template is a sub property in the templates property. The property name is the template name. The property value is the template’s definition. A template definition can contain the following sub properties:

Before discussing these properties in detail, the way pandocomatic resolves paths used in these sections is described first because paths can be used in most of these properties.

Specifying paths

Because templates can be used in any document, specifying paths pointing to assets to use in the conversion process is not straightforward. Using global paths works, but has the disadvantage that the templates are no longer easily shareable with others. Using local paths works if the assets and the document to convert are located in the same directory, but that does not hold for more general external pandocomatic templates. As a third alternative, pandocomatic also supports paths that are relative to the pandocomatic data directory. Finally, to handle paths that are meaningful in the output documents, pandocomatic has a way to specify paths relative to a root path.

You can specify these types of paths as follows:

  1. All local paths start with ./. These paths are local to the document being converted. When converting a directory tree, the current directory is being prepended to the path minus the ./.

    On the Windows operating system, a local path starts with .\. Note that backslashes might need escaping, like .\\.

  2. Global paths start with a /. These paths are resolved as is. On the Windows operating system, a global path starts with a letter followed by a colon and a backslash, for example C:\. Note that backslashes might need escaping, like C:\\.

  3. Paths relative to the pandocomatic data directory do not start with a ./ nor a /. These paths are resolved by appending the path to the pandocomatic data directory. These come in handy for defining general usable external pandocomatic templates.

    Note. For filters, processors, and start-up or clean-up scripts, the path is first checked against the PATH. If pandocomatic finds an executable matching the path, it will resolve that executable instead.

  4. Paths relative to a root path start with $ROOT$. These paths are resolved by adding ../ to first go to the root path, then from there to go to the path specified.

    The root path is set by means of the command-line option --root-path. If no such option is used, the directory of the output is used instead.

    Note. This root relative path alternative is experimental.

Template properties

extends

A template can extend zero or more templates by supplying a list of template names to extend. The extension builds from left to right.

For more detailed information about extending templates, see the Section about extending templates below.

Examples

glob

When a template is used for converting files in a directory tree, you can specify which files in the directory should be converted by a template. The glob section expects a list of glob patterns. All files that match any of these glob patterns are converted using this template.

When there are more templates that have matching glob patterns, the first one is used.

If there is also a skip configured (see the Section on global settings, the skip setting has precedence over the glob setting. Thus, if skip is ['*.md'] and glob is ['*.md'], the template will not be applied.

Examples

setup

For more involved conversion patterns, some setup of the environment might be needed. Think of setting Bash environment variables, creating temporary directories, or even installing third party tools needed in the conversion. Startup scripts can be any executable script or program.

Setup scripts are run before the conversion process starts.

Examples

preprocessors

After setup, pandocomatic executes all preprocessors in order of specification in the preprocessor property, which is a list. A preprocessor is any executable script or program that takes as input the document to convert and outputs that document after “preparing” it somehow. You can use a preprocessor to add metadata, include other files, replace strings, and so on.

Examples

metadata

Metadata is used in pandoc’s templates as well as a means of communicating with a filter. Some metadata is common to many documents, such as language, author, keywords, and so on. In the metadata property of a template you can specify this global metadata. The metadata property is a key-value list.

Examples

pandoc

To actually control the pandoc conversion process itself, you can specify any pandoc command-line option in the pandoc property, which is a key-value list.

Examples

For convenience, the virtual output format pdf is added by pandocomatic. It allows you to specify PDF output without needing to use the output option. This allows for general pandoc configurations for generating PDF files. You specify the PDF output format by to: pdf. Pandocomatic will determine the actual output format based on the value of pdf-engine. If that option is not set, pandocomatic defaults to latex.

To give the use more control over what filename extension will be used, the virtual pandoc option use-extension has been added. If set, and the output option is not being used, the value of the use-extension option is used as the extension of the output file. For example, to generate a PDF presentation using the beamer output format, you can specify the following pandoc options:

pandoc:
    from: markdown
    to: beamer
    use-extension: pdf

Finally, the virtual pandoc option rename has been added to allow you to rename the output file via a script. This script will receive the destination path on STDIN and is supposed to write the renamed output path to STDOUT. It allows you to perform quite complex behavior with regards to the output directory and name of output files.

I use this virtual pandoc option when I am generating my static sites with both HTML and PDF output and my input file is named index.md. For the HTML format I want index.html as the output file name, but for the PDF output I do not want index.pdf as output filename. Instead, I prefer to use the name of the input directory with extenstion .pdf. To that end I setup pandocomatic as follows:

pandoc:
    from: markdown
    to: pdf
    rename: use-dirname.rb

and use-dirname.rb:

#!/usr/bin/env ruby

current_dst = $stdin.read
current_dst_dir = File.dirname current_dst
current_dst_filename = File.basename current_dst
current_dst_extname = File.extname current_dst

dirname = File.split(current_dst_dir).last

if current_dst_filename.start_with? "index" and not dirname.nil?
    puts File.join(current_dst_dir, "#{dirname}.#{current_dst_extname}")
else 
    puts current_dst
end

Note Pandoc’s command-line option --verbose might interfere with the inner working of pandocomatic. As a result, pandocomatic will ignore pandoc’s --verbose option and warn you about this.

postprocessors

Similar to the preprocessors property, the postprocessors property is a list of scripts or programs to run after the pandoc conversion has finished. Each postprocessor takes as input the converted document and outputs that document with the changes made by the postprocessor. Postprocessors come in handy for cleaning up output, checking for dead links, do string replacing, and so on.

Examples

cleanup

The counterpart of the setup property. The cleanup property is a list of scripts or programs to run after the conversion of the document. It can be used to clean up temporary files, resetting the environment, uploading the resulting document, and so on.

Examples

9.2 Extending pandocomatic templates

Using the extends property of a template, you can mix and extend multiple templates. For example, building on the webpage template, I can create a my-webpage template like so:

 settings:
     data-dir: ~/my_data_dir
     recursive: true
     follow-links: false
     skip: ['.*']
 templates:
     author:
         metadata:
             author: Huub de Beer
     today:
         preprocessors:
             -   preprocessors/today.rb
     webpage:
         glob: ['*.md']
         pandoc:
             to: html
             toc: true
             css:
                 - assets/style.css
         postprocessors:
             - postprocessors/tidy.sh
     my-webpage:
         extends: ['author', 'today', 'webpage']
         pandoc:
             to: html5
             bibliography: 'assets/my-bibliography.bib'

This my-webpage templates extends the original by:

Extension rules

Although extension of templates is relatively straightforward, there are some nuances to the extension rules to keep in mind. Basically there are three cases:

  1. If the parent template has a property, but the child does not, the resulting template has the parent’s property. Examples:

    parent = 4 ∧ child = ⊘ ⇒ 4
    parent = [4, 5] ∧ child = ⊘ ⇒ [4, 5]
  2. If the parent template does not have a property, but the child does, the resulting template has the child’s property.

    parent = ⊘ ∧ child = 4 ⇒ 4
    parent = ⊘ ∧ child = {a: 1} ⇒ {a: 1}
  3. If both parent and child templates do have a property, the resulting template will have that property and its value is determined as follows:

    1. If the child’s value is of a simple type, such as a string, number, or Boolean, the resulting property will have the value of the child. Examples:

      parent = 4 ∧ child = true ⇒ true
      parent = [4, 5] ∧ child = "yes" ⇒ "yes"
      parent = {key: true} ∧ child = 12 ⇒ 12
    2. If parent and child values both are key-value lists, the resulting value will be the child’s key-value list merged with the parent’s key-value list. Examples:

      parent = {a: 1, b: 2} ∧ child = {a: 2, c: 3} ⇒ {a: 2, b: 2, c: 3}
      parent = {a: 1, b: 2} ∧ child = {a: , c: 3} ⇒ {b: 2, c: 3}
    3. If the parent value is a list, two different extension mechanisms can take effect depending on the type of the child’s value:

      1. If the child is a list as well, the resulting value will be the child’s list merged with the parent’s list. Duplicate values will be removed. Lists in pandocomatic templates are treated as sets. Examples:

        parent = [1] ∧ child = [2] ⇒ [1, 2]
        parent = [1] ∧ child = [1, 2] ⇒ [1, 2]
      2. If the child is a key-value list, it is assumed to have keys remove and add. The resulting value will be the parent’s value with the items from the remove list removed and items from the add list added. Examples:

        parent = [1] ∧ child = {'remove': [1], 'add': [3]} ⇒ [3]
        parent = [1, 2] ∧ child = {'remove': [1]} ⇒ [2]

To remove a property in a child template, that child’s value should be nil. You can create a nil value in YAML by having a key without a value.

9.3 Customizing an external template in an internal template

To use an external pandocomatic template you have to use it in a document by creating an internal pandocomatic template which has the use-template property set to the name of the external pandocomatic template. After that, you can customize the template to suit the document it is used in, for example adding extra pandoc command-line options or adding another preprocessor.

You create an internal pandocomatic template by adding a pandocomatic_ property to the document’s YAML metadata. The pandocomatic_ property can have the same properties as an external pandocomatic template except for the glob and extends properties. (Actually, you can add these two properties as well, but they are ignored.)

For example, if you use the my-webpage template, but you would like to use a different bibliography and check all links in the converted document, your document would look like:

 ---
 title: About using templates
 pandocomatic_:
     use-template: my-webpage
     pandoc:
         bibliography: ./a_different_bibliography.bib
     postprocessors:
     -   postprocessors/check_links.sh
 ...
 
 # Introduction
 
 To use a template, ...

Multiple conversions

The use-template property can also be a list of external pandocomatic template names. In that case, the document is converted once for each of these templates. For example, this allows you to generate both a HTML and a PDF version of a document at the same time:

 ---
 title: About using templates
 pandocomatic_:
     use-template: 
     -   my-webpage
     -   my-pdf
     pandoc:
         bibliography: ./a_different_bibliography.bib
     postprocessors:
     -   postprocessors/check_links.sh
 ...
 
 # Introduction
 
 To use a template, ...

Do note, however, that an internal pandocomatic template will apply to all used external pandocomatic templates. It is not possible to customize one used template differently than another. This means that you have to move the customization to the used external pandocomatic templates or you have customize the internal pandocomatic template such that it is applicable to all used external pandocomatic templates (as in the example above).

9.4 Making templates more flexible with environment variable substitution

You can use environment variables in your templates to make them more flexible. For example, if you want to vary the output format depending on the value of environment variable OUTPUT_FORMAT, add the following format property to your template:

pandocomatic_:
  # ...
  pandoc:
    # ...
    format: $(OUTPUT_FORMAT)$

When pandocomatic reads this template, it replaces $(OUTPUT_FORMAT)$ with the value of the environment variable OUTPUT_FORMAT. If no such variable exists, pandocomatic stops after printing an error message. Pandocomatic tells you in which template it encountered this non-existing environment variable so you can easily investigate and resolve the issue.

In other words, all occurrences of $(X)$ in an internal or external template are replaced by the the value of environment variable X. You can use environment variables anywhere in your templates, as property names, keys, or values. You can even use it to add YAML snippets to your templates.

Combined with other tools, like bash scripts, make files, and the like, you can create a powerful and flexible template system for your situation.


Part IV. Appendix

Chapter 10. Frequently asked questions (FAQ)

10.1 How do I use pandoc2 with pandocomatic?

Pandocomatic uses paru to run pandoc. Paru itself uses the pandoc executable in your PATH. If that already is pandoc2, you do not need to do anything.

If you have pandoc1 installed by default, however, and you want to run a nightly version of pandoc2, you have to set the PARU_PANDOC_PATH to point to the pandoc2 executable. For example:

export PARU_PANDOC_PATH=~/Downloads/pandoc-amd64-7c20fab3/pandoc
pandocomatic some-file-to-convert.md

10.2 Pandocomatic has too much output! How do I make pandocomatic more quiet?

You can run pandoc in quiet mode by using the --quiet or -q command-line option. For example:

pandocomatic --quiet some-file-to-export.md

Chapter 11. Glossary

pandocomatic template

A pandocomatic template specified the conversion process executed by pandocomatic. It can contain the following properties:

internal pandocomatic template

A pandocomatic template specified in a pandoc markdown file itself via the YAML metadata property pandocomatic_.

external pandocomatic template

A pandocomatic template specified in a pandocomatic configuration file.

preprocessors

A preprocessor applied to an input document before running pandoc.

postprocessors

A postprocessor applied to an input document after running pandoc.

pandocomatic data directory

The directory used by pandocomatic to resolve relative paths. Use this directory to store preprocessors, pandoc templates, pandoc filters, postprocessors, setup scripts, and cleanup scripts. It defaults to pandoc’s data directory.

pandocomatic configuration file

The configuration file specifying external pandocomatic templates as well as settings for converting a directory tree. Defaults to pandocomatic.yaml.

extending pandocomatic templates

External pandocomatic templates can extend other external pandocomatic templates. By using multiple smaller external pandocomatic templates it is possible to setup your templates in a modular way. Pandocomatic supports extending from multiple external pandocomatic templates.

static-site generator

Pandocomatic can be used as a static-site generator by running pandocomatic recursivel on a directory. Pandocomatic has some specific congiguration options to be used as a static-site generator.