Represent Packages with pkglite.txt

The output file of pkglite::pack() uses a standard file format to allow the storage, exchange, and parsing of the packaged assets. The specification for this file is detailed below.

Filename and extension

Unless specified, the output file is automatically named.

The file extension is .txt so that one can recognize, open, and inspect it directly using regular text editors.

File format

The overall goal here is to make the file format unambiguous, human-friendly, and machine-readable. For pkglite.txt, we follow the DCF (Debian Control File) format, used by Debian, R, and RStudio IDE. The minimalist, time-tested format allows us to save key-value data in plain text that humans can easily read and write. The format is also extendable to include more information about the assets being packed.

File structure

A pkglite.txt is structured following these rules:

Field names and values

Each block has at least four key-value pairs called fields. For example:

Package: pkglite
File: R/pkglite-package.R
Format: text
Content:
  #' @keywords internal
  "_PACKAGE"

The keys and values in these fields are separated by a colon and a whitespace except for Content, whose value part starts in a new line.

Package

R package name. Since one pkglite.txt might contain files from multiple packages, this field is used to declare the package the file is under explicitly.

This design implies that each pkglite.txt can only store multiple R packages with unique names.

File

The relative path (to the package root) of the file.

Format

File format category. It can be text or binary. This determines how the file content will be read and written.

Content

The file content. The text format files are stored as-is, while the binary files are stored in the hexadecimal format, following the RTF format convention. In both cases, two whitespaces are added before each line of the value part.