LoginSign Up
← Back to all sparks
R

rdocdump

INFRA · APIS
Velocity0.0

Dump 'R' Package Source, Documentation, and Vignettes into One File

github.com/e-kotov/rdocdump ↗

rdocdump grew from a CRAN doc dumper into a resolver that pulls packages from anywhere

rdocumentationllm toolingpackage managementpak
◆Current state
rdocdump flattens an R package's documentation and source into plain text, a shape aimed squarely at feeding code and docs to language models. It reached CRAN in June 2025 handling CRAN packages only. Within a year it accepted GitHub, GitLab and Bitbucket references with branch, tag and commit pinning, and 0.3.0 in May 2026 moved resolution onto pak.
◆Where it's heading
The direction is to stop caring where a package lives. 0.2.0 opened remote repository references and direct URLs; 0.3.0 swapped remotes for pak, added auto-discovery of packages in subdirectories, broadened URL format support to Bioconductor and GitLab, and added git clone as an automatic fallback when resolution fails. Cross-platform handling moved to internal tar operations.
◆Prediction
With resolution generalized, remaining work is likely on output shaping — what gets included and how it is chunked — rather than on more sources. The entries do not say whether the text output format itself is stable.

◆Recent moves

  1. 2mo ago

    Resolution moves to pak, with git clone fallback and subdirectory discovery

    Migrates remote package resolution and downloads to pak, adds auto-discovery of R packages in repository subdirectories, and falls back to git clone automatically when resolution fails. URL format support broadens to Bioconductor and GitLab, and tar operations move in-package for cross-platform consistency. The infrastructure release that makes 0.2.0's promise reliable.

    View source ↗
  2. 3mo ago

    GitHub, GitLab and Bitbucket sources with branch/tag/commit pinning

    Extends rdd_to_txt() and rdd_extract_code() beyond CRAN to remote repository references across GitHub, GitLab and Bitbucket, including specific branches, tags and commits, subdirectory packages, and direct URLs. The change that turned a CRAN utility into a general-purpose source fetcher.

    View source ↗
  3. 11mo ago

    version argument pins which CRAN release to process

    Adds a version argument so a specific CRAN release can be downloaded and processed rather than the current one. The first step away from 'whatever is on CRAN right now'.

    View source ↗
  4. 1y ago

    Initial CRAN release with R documentation and source parsing

    The initial CRAN submission, with basic parsing of R documentation and source files. Everything since has been about widening where those files can come from.

    View source ↗
  5. 1y ago

    Proof-of-concept development version

    A pre-release proof of concept with no described functionality.

    View source ↗