Raw File
Tip revision: 2a99f4af4a352d2e082988d364e01e2e2809a87b authored by Daniel Himmelstein on 04 June 2019, 15:58 UTC
process command: make parent directories
Tip revision: 2a99f4a
# The manuscript bot for automated scholarly publishing

[![Travis Linux Build Status](](
[![AppVeyor Windows Build Status](](

The Manubot Python package prepares scholarly manuscripts for Pandoc consumption.
It automates and scripts several aspects of manuscript creation, including fetching bibliographic metadata for citations.

This program is designed to be used with clones of [Manubot Rootstock](, which perform Pandoc conversion and continuous deployment.
See the Manubot Rootstock [usage guide]( for more information.

To cite the Manubot project or for more information on its design and history, see:

> **Open collaborative writing with Manubot**<br>
Daniel S. Himmelstein, Vincent Rubinetti, David R. Slochower, Dongbo Hu, Venkat S. Malladi, Casey S. Greene, Anthony Gitter<br>
_Manubot Preprint_ (2019) <>

## Usage

Installing the python package creates the `manubot` command line program.
Here is the usage information as per `manubot --help`:

<!-- test codeblock contains output of `manubot --help` -->
usage: manubot [-h] [--version] {process,cite} ...

Manubot: the manuscript bot for scholarly writing

optional arguments:
  -h, --help      show this help message and exit
  --version       show program's version number and exit

  All operations are done through subcommands:

    process       process manuscript content
    cite          citation to CSL command line utility

Note that all operations are done through the following sub-commands.

### Process

The `manubot process` program is the primary interface to using Manubot.
There are two required arguments: `--content-directory` and `--output-directory`, which specify the respective paths to the content and output directories.
The content directory stores the manuscript source files.
Files generated by Manubot are saved to the output directory.

One common setup is to create a directory for a manuscript that contains both the `content` and `output` directory.
Under this setup, you can run the Manubot using:

manubot process \
  --content-directory=content \

See `manubot process --help` for documentation of all command line arguments:

<!-- test codeblock contains output of `manubot process --help` -->
usage: manubot process [-h] --content-directory CONTENT_DIRECTORY
                       --output-directory OUTPUT_DIRECTORY
                       [--template-variables-path TEMPLATE_VARIABLES_PATH]
                       [--cache-directory CACHE_DIRECTORY]
                       [--log-level {DEBUG,INFO,WARNING,ERROR,CRITICAL}]

Process manuscript content to create outputs for Pandoc consumption. Performs
bibliographic processing and templating.

optional arguments:
  -h, --help            show this help message and exit
  --content-directory CONTENT_DIRECTORY
                        Directory where manuscript content files are located.
  --output-directory OUTPUT_DIRECTORY
                        Directory to output files generated by this script.
  --template-variables-path TEMPLATE_VARIABLES_PATH
                        Path or URL of a JSON file containing template
                        variables for jinja2. Specify this argument multiple
                        times to read multiple files. Variables can be applied
                        to a namespace (i.e. stored under a dictionary key)
                        like `--template-variables-
                        path=namespace=path_or_url`. Namespaces must match the
                        regex `[a-zA-Z_][a-zA-Z0-9_]*`.
  --cache-directory CACHE_DIRECTORY
                        Custom cache directory. If not specified, caches to
                        Set the logging level for stderr logging

#### Manual references

Manubot has the ability to rely on user-provided reference metadata rather than generating it.
`manubot process` searches the content directory for files containing manually-provided reference metadata that match the glob `manual-references*.*`.
If a manual reference filename ends with `.json` or `.yaml`, it's assumed to contain CSL Data (i.e. Citation Style Language JSON).
Otherwise, the format is inferred from the extension and converted to CSL JSON using the `pandoc-citeproc --bib2json` [utility](
The standard citation for manual references is inferred from the CSL JSON `id` or `note` field.
When no prefix is provided, such as `doi:`, `url:`, or `raw:`, a `raw:` prefix is automatically added.
If multiple manual reference files load metadata for the same standard citation `id`, precedence is assigned according to descending filename order.

### Cite

`manubot cite` is a command line utility to create [CSL JSON items]( for one or more citations.
Citations should be in the format `source:identifier`.
For example, the following example generates CSL JSON for four references:

manubot cite doi:10.1098/rsif.2017.0387 pmid:29424689 pmcid:PMC5640425 arxiv:1806.05726

The following [terminal recording]( demonstrates the main features of `manubot cite`:

![manubot cite demonstration](media/terminal-recordings/manubot-cite-cast.gif)

Additional usage information is available from `manubot cite --help`:

<!-- test codeblock contains output of `manubot cite --help` -->
usage: manubot cite [-h] [--render] [--csl CSL]
                    [--format {plain,markdown,docx,html,jats}]
                    [--output OUTPUT] [--allow-invalid-csl-data]
                    [--log-level {DEBUG,INFO,WARNING,ERROR,CRITICAL}]
                    citations [citations ...]

Retrieve bibliographic metadata for one or more citation identifiers.

positional arguments:
  citations             One or more (space separated) citations to produce CSL

optional arguments:
  -h, --help            show this help message and exit
  --render              Whether to render CSL Data into a formatted reference
                        list using Pandoc. Pandoc version 2.0 or higher is
                        required for complete support of available output
  --csl CSL             When --render, specify an XML CSL definition to style
                        references (i.e. Pandoc's --csl option). Defaults to
                        Manubot's style.
  --format {plain,markdown,docx,html,jats}
                        When --render, format to use for output file. If not
                        specified, attempt to infer this from filename
                        extension. Otherwise, default to plain.
  --output OUTPUT       Specify a file to write output, otherwise default to
                        Allow CSL Items that do not conform to the JSON
                        Schema. Skips CSL pruning.
                        Set the logging level for stderr logging

## Installation

Install the version specified by a git commit hash using:

pip install git+$COMMIT

Use the `--upgrade` argument to reinstall `manubot` with a different commit hash.

## Development

Create a development environment using:

conda create --name manubot-dev --channel conda-forge \
  python=3.6 jinja2 pandas pytest pandoc
conda activate manubot-dev  # assumes conda >= 4.4
pip install --editable .

Inside this environment, use `pytest` to run the test suite.
You can also use the `manubot` CLI to build manuscripts.
For example:

manubot process \
  --content-directory=tests/manuscripts/example/content \
  --output-directory=tests/manuscripts/example/output \

## Release instructions


This section is only relevant for project maintainers.
Travis CI deployments are used to upload releases to [PyPI](
To create a new release, bump the `__version__` in [`manubot/`](manubot/
Then run the following commands:

TAG=v`python --version`
# Commit updated __version__ info
git add manubot/ release-notes/$
git commit --message="Prepare $TAG release"
git push
# Create & push tag (assuming upstream is greenelab remote)
git tag --annotate $TAG --file=release-notes/$
git push upstream $TAG

## Goals & Acknowledgments

Our goal is to create scholarly infrastructure that encourages open science and assists reproducibility.
Accordingly, we hope for the Manubot software and philosophy to be adopted widely, by both academic and commercial entities.
As such, Manubot is free/libre and open source software (see [``](

We would like to thank the contributors and funders whose support makes this project possible.
Specifically, Manubot development has been financially supported by:

- the **Alfred P. Sloan Foundation** in [Grant G-2018-11163]( to [**@dhimmel**](
- the **Gordon & Betty Moore Foundation** ([**@DDD-Moore**]( in [Grant GBMF4552]( to [**@cgreene**](
back to top