Skip to main content

Tools to generate documentation.

Project description

Documentation tools for enterprise-quality documentation from Markdown source. Dactyl has advanced features to enable single-sourcing and an extensible syntax for building well-organized, visually attractive docs. It generates output in HTML (natively), and can make PDFs if you have Prince installed.


Dactyl requires Python 3. Install with pip:

sudo pip3 install dactyl

Or a local install in a virtualenv:

# Create an activate a virtualenv so the package and dependencies are localized
virtualenv -p `which python3` venv_dactyl
source venv_dactyl/bin/activate

# Check out this repo
git clone

# Install
pip3 install dactyl/

# Where 'dactyl/' is the top level directory of the repo, containing
# And note the trailing '/' which tells pip to use a local directory to install it.


Simple (“Ad-Hoc”) usage:

$ dactyl_build --pages

By default, the resulting HTML pages are written to a folder called out/ in the current working directory. You can specify a different output path in the config file or by using the -o parameter.

Building PDF

Dactyl generates PDFs by making temporary HTML files and running Prince. Use the --pdf command to generate a PDF. Dactyl tries to come up with a sensible output filename by default, or you can provide one (which must end in .pdf):

$ dactyl_build --pages --pdf MyGuide.pdf

Advanced Usage

Dactyl is intended to be used with a config file containing a list of pages to parse. Pages are grouped into “targets” that represent a group of documents to be built together; a page can belong to multiple targets, and can even contain conditional syntax so that it builds slightly different depending on the target in question. Targets and pages can also use different templates from each other, and pages can inherit semi-arbitrary key/value pairs from the targets.

For more information on configuration, see the default-config.yml and the examples folder.

The input pages in the config file should be specified relative to the content_path, which is content/ by default. You can also specify a URL to pull in a markdown file from a remote source, but if you do, Dactyl won’t run any pre-processing on it.

For a full list of Dactyl options, use the -h parameter.

Specifying a Config File

By default, Dactyl looks for a config file named dactyl-config.yml in the current working directory. You can specify an alternate config file with the -c or --config parameter:

$ dactyl_build -c path/to/alt-config.yml

For more information on configuration, see the default-config.yml and the examples folder.

Specifying a Target

If your config file contains more than one target, Dactyl builds the first one by default. You can specify a different target by passing its name value with the -t parameter:

$ dactyl_build -t non-default-target

Static Files

Your templates may require certain static files (such as JavaScript, CSS, and images) to display properly. Your content may have its own static files (such as diagrams and figures). By default, Dactyl assumes that templates have static files in the assets/ folder. You can configure this path and also specify one or more paths to static files referenced by your content. When you build, Dactyl copies files from these folders to the output folder by default depending on which mode you’re building:

Build Mode

Files copied to output folder by default


Both template and content static files


Neither template nor content static files (cannot be overridden)


Content static files only

ElasticSearch JSON

Neither template nor content static files

You can use a commandline flag to explicitly specify what gets copied to the output folder, except in the case of PDF. (In PDF mode, Dactyl writes only the final PDF to the output folder.) The flags are as follows:

Flag (long version)

Short version




Copy all static files to the out dir.



Don’t copy any static files to the out dir.



Copy only templates’ static files to the out dir



Copy only the content’s static files to the out dir

The following config file parameters control what paths Dactyl checks for static content:






Static files belonging to the templates.



Static files belonging to content. This can be a single folder path, as a string, or an array of paths to files or folders. Dactyl copies all files and folders (regardless of whether the current target uses them).

Listing Available Targets

If you have a lot of targets, it can be hard to remember what the short names for each are. If you provide the -l flag, Dactyl will list available targets and then quit without doing anything:

$ dactyl_build -l
tests       Dactyl Test Suite
rc-install      Ripple Connect v2.6.3 Installation Guide
kc-rt-faq       Ripple Trade Migration FAQ

Building Markdown

This mode runs the pre-processor only, so you can generate Markdown files that are more likely to display properly in conventional Markdown parsers (like the one built into GitHub). Use the --md flag to output Markdown files, skipping the HTML/PDF templates entirely.

$ dactyl_build --md

Building Only One Page

If you only want to build a single page, you can use the --only flag, followed by the filename you want to build (either the input filename ending in .md or the output filename ending in .html):

dactyl_build --only index.html --pdf

This command can be combined with the --pdf or --md flags. You can also use it with the --target setting (in case you want the context from the target even though you’re only building one page.)

Watch Mode

You can use the -w flag to make Dactyl run continuously, watching for changes to its input templates or markdown files. Whenever it detects that a file has changed, Dactyl automatically rebuilds the output in whatever the current mode is, (HTML, PDF, or Markdown).

To be detected as a change, the file has to match one of the following patterns:


Beware: some configurations can lead to an infinite loop. (For example, if your output directory is a subdirectory of your content directory and you use Dactyl in --md mode.)

Limitations: Watch mode can be combined with --only, but re-builds the page even when it detects changes to unrelated pages. Watch mode doesn’t detect changes to the config file, static files, or filters.

To stop watching, interrupt the Dactyl process (Ctrl-C in most terminals).

ElasticSearch Compatibility

Dactyl has the ability to build JSON formatted for upload to ElasticSearch and even upload it directly.

To build JSON files for upload to ElasticSearch, use the --es mode:

dactyl_build --es

This writes files to the usual output directory using an ElasticSearch JSON template. Dactyl skips any files that do not have a md source parameter in this mode. The output filenames are the pages’ html filenames, except ending in .json instead of .html. You can specify a custom template for these JSON files using the top-level default_es_template field in the config file. This template must be a valid JSON file and has several special properties as described in ElasticSearch JSON Templates.

Dactyl can also upload these files directly to an ElasticSearch instance, even when building for another mode. For example, to build the HTML version of a target named filterdemos but also upload that target’s JSON-formatted data to an ElasticSearch instance:

dactyl_build -t filterdemos --html --es_upload

The parameter to --es_upload should be the base URL of your ElasticSearch index. You can omit the parameter to use the default base URL of http://localhost:9200.

ElasticSearch JSON Templates

Dactyl has a special format for JSON templates meant for creating ElasticSearch data. These templates must be valid JSON and are processed according to the following rules:

  • Any strings in the fields’ values are “preprocessed” in a similar context to the Jinja2-based Markdown preprocessor. For example, the string {{}} evaluates to the page’s name.

  • Any object containing the key __dactyl_eval__ is evaluated as a Python expression. The object is replaced with the results of the expression, with lists becoming JSON arrays and dictionaries becoming JSON objects.

  • The above rules apply recursively to values nested in arrays and objects. All other values are preserved literally.

The context provided to the preprocessing and to the __dactyl_eval__ expressions is the same and contains the following:


Python Type




The current page definition (usually derived from the config file)



The current target definition (usually derived from the config file)



A list of unique category values used by pages in the current target, in order of appearance.



A list of the names of Dactyl filters applied to the current page.



Always equal to es in this context



The current time, in the time_format specified in the config. (Defaults to YYYY-MM-DD)



If true, this build is running with the option to continue through errors where possible.

The currentpage dictionary has the following special fields in this mode:


Python Type




A plaintext-only version of the page’s markdown content, with all Markdown and HTML syntax removed.



A mapping of the page’s headers to the unique IDs of those headers in the generated HTML version.



An introductory blurb generated from the page’s first paragraph of text.

Style Checking

The style checker is experimental. It reads lists of discouraged words and phrases from the word_substitutions_file and phrase_substitutions_file paths (respectively) in the config. For each such word or phrase that appears in the output HTML (excluding code, pre, and tt elements), it counts and prints a violation, suggesting a replacement based on the word/phrase file.

The style checker re-generates HTML in-memory (never writing it out). It uses the first target in the config file unless you specify another target with -t.

Example usage:

$ dactyl_style_checker -t rippledevportal
Style Checker - checking all pages in target rippledevportal
Found 6 issues:
Page: Gateway Guide
   Discouraged phrase: in order to (1 instances); suggest 'to' instead.
   Discouraged phrase: and/or (1 instances); suggest '__ or __ or both' instead.
   Discouraged word: feasible (1 instances); suggest 'can be done, workable' instead.
   Discouraged phrase: in an effort to (1 instances); suggest 'to' instead.
   Discouraged phrase: comply with (1 instances); suggest 'follow' instead.
Page: Amendments
   Discouraged phrase: limited number (1 instances); suggest 'limits' instead.

You can add an exemption to a specific style rule with an HTML comment. The exemption applies to the whole output (HTML) file in which it appears.

Maybe the word "will" is a discouraged word, but you really want to use it here without flagging it as a violation? Adding a comment like this <!-- STYLE_OVERRIDE: will --> makes it so.


Many parts of Dactyl are configurable. An advanced setup would probably have the following folders in your directory structure:

./                      # Top-level dir; this is where you run dactyl_*
./dactyl-config.yml     # Default config file name
./content               # Dir containing your .md source files
---------/*/*.md        # You can sort .md files into subdirs if you like
---------/static/*      # Static images referencd in your .md files
./templates/template-*.html # Custom HTML Templates
./assets                # Directory for static files referenced by templates
./out                   # Directory where output gets generated. Can be deleted

(All of these paths can be configured.)


A target represents a group of pages, which can be built together or concatenated into a single PDF. You should have at least one target defined in the targets array of your Dactyl config file. A target definition should consist of a short name (used to specify the target in the commandline and elsewhere in the config file) and a human-readable display_name (used mostly by templates but also when listing targets on the commandline).

A simple target definition:

    -   name: kc-rt-faq
        display_name: Ripple Trade Migration FAQ

In addition to name and display_name, a target definition can contain arbitrary key-values to be inherited by all pages in this target. Dictionary values are inherited such that keys that aren’t set in the page are carried over from the target, recursively. The rest of the time, fields that appear in a page definition take precedence over fields that appear in a target definition.

Some things you may want to set at the target level include filters (an array of filters to apply to pages in this target), template (template to use when building HTML), and pdf_template (template to use when building PDF). You can also use the custom values in templates and preprocessing. Some filters define additional fields that affect the filter’s behavior.

The following field names cannot be inherited: name, display_name, and pages.


Each page represents one HTML file in your output. A page can belong to one or more targets. When building a target, all the pages belonging to that target are built in the order they appear in the pages array of your Dactyl config file.

Example of a pages definition with two files:

    -   name: RippleAPI
        category: References
        html: reference-rippleapi.html
            - remove_doctoc
            - add_version
            - local

    -   name: rippled
        category: References
        html: reference-rippled.html
            - local

Each individual page definition can have the following fields:






The short names of the targets that should include this page.



(Optional) The filename where this file should be written in the output directory. If omitted, Dactyl chooses a filename based on the md field (if provided), the name field (if provided), or the current time (as a last resort). By default, generated filenames flatten the folder structure of the md files. To instead replicate the folder structure of the source documents in auto-generated filenames, add flatten_default_html_paths: true to the top level of your Dactyl config file.



(Optional) Human-readable display name for this page. If omitted but md is provided, Dactyl tries to guess the right file name by looking at the first two lines of the md source file.



(Optional) The markdown filename to parse to generate this page, relative to the content_path in your config. If this is not provided, the source file is assumed to be empty. (You might do that if you use a nonstandard template for this page.)



(Optional) The name of a category to group this page into. This is used by Dactyl’s built-in templates to organize the table of contents.



(Optional) The filename of a custom Jinja HTML template to use when building this page for HTML, relative to the template_path in your config.



(Optional) The filename of a custom Jinja HTML template to use when building this page for PDF, relative to the template_path in your config.


Additional arbitrary key-value pairs as desired. These values can be used by templates or pre-processing.


Dactyl supports extended Markdown syntax with the Python-Markdown Extra module. This correctly parses most GitHub-Flavored Markdown syntax (such as tables and fenced code blocks) as well as a few other features.


Dactyl pre-processes Markdown files by treating them as Jinja Templates, so you can use Jinja’s templating syntax to do advanced stuff like include other files or pull in variables from the config or commandline. Dactyl passes the following fields to Markdown files when it pre-processes them:




The target definition of the current target.


The array of page definitions in the current target. Use this to generate navigation across pages. (The default templates don’t do this, but you should.)


The definition of the page currently being rendered.


A de-duplicated array of categories that are used by at least one page in this target, sorted in the order they first appear.


The global Dactyl config object.


The parsed HTML content of the page currently being rendered.


The current date as of rendering. The format is YYYY-MM-DD by default; you can also set the time_format field to a custom stftime format string.


The output format: either html (default), pdf, or md.

Adding Variables from the Commandline

You can pass in a JSON or YAML-formatted list of variables using --vars commandline switch. Any such variables get added as fields of target and inherited by currentpage in any case where currentpage does not already have the same variable name set. For example:

$ cat md/
Myvar is: '{{ target.myvar }}'

$ dactyl_build --vars '{"myvar":"foo"}'
rendering pages...
writing to file: out/index.html...
Preparing page
reading markdown from file:
... parsing markdown...
... modifying links for target: default
... re-rendering HTML from soup...
writing to file: out/test_vars.html...
done rendering
copying static pages...

$ cat out/test_vars.html | grep Myvar
<p>Myvar is: 'foo'</p></main>

If argument to --vars ends in .yaml or .json, Dactyl treats the argument as a filename and opens it as a YAML file. (YAML is a superset of JSON, so this works for JSON files.) Otherwise, Dactyl treats the argument as a YAML/JSON object directly. Be sure that the argument is quoted and escaped as necessary based on the commandline shell you use.

You cannot set the following reserved keys:

  • name

  • display_name (Instead, use the --title argument to set the display name of the target on the commandline.)

  • pages


Furthermore, Dactyl supports additional custom post-processing through the use of filters. Filters can operate on the markdown (after it’s been pre-processed), on the raw HTML (after it’s been parsed), or on a BeautifulSoup object representing the output HTML. Filters can also export functions and values that are available to the preprocessor.

Dactyl comes with several filters, which you can enable in your config file. You can also write your own filters. If you do, you must specify the paths to the folder(s) containing your filter files in the filter_paths array of the config file.

To enable a filter for a target or page, set the filters field of the config to be an array of filter names, where the filter names are derived from the Python source files in the format filter_<filtername>.py. Filter names must be valid Python variable names, so they can’t start with a numeral and must contain only alphanumeric and underscore characters.

Dactyl automatically runs the following functions from filter files (skipping any that aren’t defined):

  1. Before running the preprocessor on a page, Dactyl adds all items from each filter’s export global dictionary to the preprocessor environment.

  2. Dactyl runs the filter_markdown(md, **kwargs) function of each filter after the preprocessor. This function receives the preprocessed markdown as a string in the md argument and must return a string with the markdown as filtered.

  3. Dactyl runs the filter_html(html, **kwargs) function after the markdown processor. This function receives the parsed markdown content as an HTML string in the html argument and must return a string with the HTML as filtered.

  4. Dactyl runs the filter_soup(soup, **kwargs) function after the HTML filters. This function is expected to directly modify the soup argument, which contains a BeautifulSoup 4 object representing the HTML contents.

The keyword arguments (**kwargs) for the functions may change in future versions. As of Dactyl 0.5.0, the arguments are as follows:






The current page, as defined in the config file plus values inherited from the current target and any processing or calculations. (For example, Dactyl automatically adds a name field if one isn’t present.)



A de-duplicated, ordered list of category fields present among pages in this target.



A list of page objects for all pages in the current target, in the same order they appear in the config file.



The current target definition, as derived from the config file.



The time this build was started. The format is defined by your config’s global time_format field (in stftime format), defaulting to YYYY-MM-DD.



Either html, pdf, or md depending on what output Dactyl is building.



The global config object, based on the config file plus any commandline switches.



The logging object Dactyl uses, with the verbosity set to match user input.

See the examples for examples of how to do many of these things.


Dactyl provides the following information to templates, which you can access with Jinja’s templating syntax (e.g. {{ target.display_name }}):




The target definition of the current target.


The array of page definitions in the current target. Use this to generate navigation across pages. (The default templates don’t do this, but you should.)


The definition of the page currently being rendered.


A de-duplicated array of categories that are used by at least one page in this target, sorted in the order they first appear.


The global Dactyl config object.


The parsed HTML content of the page currently being rendered.


The current date as of rendering. The format is YYYY-MM-DD by default; you can also set the time_format field to a custom stftime format string.


The output format: either html (default), pdf, or md.


A table of contents generated from the current page’s headers. Wrap this in a <ul> element.


(Deprecated alias for page_toc.)

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

dactyl-0.8.2.tar.gz (58.5 kB view hashes)

Uploaded Source

Built Distribution

dactyl-0.8.2-py2.py3-none-any.whl (49.3 kB view hashes)

Uploaded Python 2 Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page