CodeGraph currently only indexes Python files. I'm working on a large epidemiological research project with 30+ R scripts (.R files) and many Markdown documentation files, but CodeGraph can only see the single Python file in the project.

[raf-159357]

Add tree-sitter R language support so CodeGraph can parse .R files and index:

• Function definitions (function_name <- function(...))
• Variable assignments (x <- value)
• Function calls
• Library imports (library(x), require(x))
• Source references (source("file.R"))

A simple R file like:

x <- 1
y <- function(a) a + 1
z <- y(x)

[psandeep02]

This is a well-defined and practical enhancement request. Adding Tree-sitter support for the R language would significantly improve CodeGraph's ability to analyze and index R codebases. The proposed scope covers the most important R constructs, including function definitions, variable assignments, function calls, library imports, and source references, which are essential for accurate code navigation and dependency tracking.
The example provided clearly demonstrates the expected parsing behavior and makes the implementation requirements easy to understand. This feature would be particularly valuable for data science, statistical computing, and research projects that rely heavily on R.
Recommendation: Approve and prioritize implementation, as it expands language coverage while providing immediate value to R developers and analysts.

[colbymchenry]

Shipped in #839, on main for the next release — CodeGraph now indexes R (.R / .r).

Everything you listed is covered, plus the class systems research code tends to accumulate:

• Function definitions in every assignment form — name <- function(...), name = function(...), lambdas (\(x)), and nested definitions (attributed to their enclosing function)
• Variable assignments at the top level (ALL_CAPS names index as constants)
• Function calls, including inside tidyverse pipe chains
• Library imports — library(x), require(x), requireNamespace("x")
• Source references — source("file.R") links to the sourced file
• Classes: S4 setClass, Reference classes, R6, and ggplot2-style ggproto — with their methods and inheritance (extends) edges — plus setGeneric/setMethod

Validated on three real R codebases: twitter/AnomalyDetection (100% of named definitions extracted), tidyverse/dplyr (1,000+ functions), and tidyverse/ggplot2 (150 ggproto classes, ~600 methods, 128 inheritance edges). In headless agent benchmarks on all three, an agent with CodeGraph answered cross-file architecture questions from the graph with zero or near-zero file reads, versus 6–28 reads/greps without — on ggplot2 the with-CodeGraph run was ~2.4× faster.

After upgrading, run codegraph init -i in your workspace. Two notes for your setup specifically:

• A folder holding several R projects as separate git repos indexes as one workspace since feat(index): multi-repo workspaces index as a whole (#514) #837 (also in the next release) — run codegraph init at the top-level directory.
• Plain Markdown docs aren't parsed as symbols (and .Rmd isn't yet — the R code chunks there are a natural follow-up; open a separate issue if that's a real need for your project and we'll look at it).