---
title: "Using reticulate in an R Package"
output:
  rmarkdown::html_vignette
editor_options:
  markdown:
    wrap: 80
vignette: >
  %\VignetteIndexEntry{Using reticulate in an R Package}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

## Declaring Python Requirements

R package authors can use reticulate to make Python packages accessible to users
from R. This vignette documents best practices for how package authors can
declare and import their package's Python dependencies.

While `reticulate::import()` can be used to load a Python module, it does not
provide any mechanism for installing a Python package and actually making sure
the module is available. `reticulate::py_require()` helps fill that gap, by
giving R package authors a way to declare their Python package dependencies in a
way that can be collated and respected across multiple packages using
reticulate, each with their own unique requirements.

Beginning with Reticulate version 1.41, R packages can declare Python
requirements with `py_require()`. Python package dependencies requested via
`py_require()` will automatically be provisioned and made available for the user
when the Python session is later initialized, via an ephemeral Python virtual
environment. These requested packages can then be imported and used within your
R package as required.

### Typical Usage

`py_require()` is typically called from `.onLoad()`, as shown below:

``` r
.onLoad <- function(libname, pkgname) {
  reticulate::py_require("scipy")
}
```

`py_require()` can also be called from other package functions to modify
dependencies after the package has loaded. This is useful for packages that
support multiple configurations.

For example, the `keras3` R package supports multiple backends. In `.onLoad()`,
`keras3` configures a default backend, but users can choose a different one
using the `use_backend()` function. This function calls `py_require()` with
different values based on the selected backend:

``` r
.onLoad <- function(...) {
  py_require("keras")
  use_backend("tensorflow") # Default to TensorFlow
}

#' @export
use_backend <- function(backend, gpu = TRUE) {
  py_require("tensorflow", action = "remove") # Remove default backend

  switch(paste0(backend, "_", get_os()),
    jax_Linux = if (gpu) py_require("jax[cuda12]") else py_require("jax[cpu]"),
    jax_macOS = py_require(c("jax", if (gpu) "jax-metal")),
    jax_Windows = py_require("jax"),
    tensorflow_Linux = { ... },
    tensorflow_macOS = { ... },
    tensorflow_Windows = { ... },
    torch_Linux = { ... },
    torch_macOS = { ... },
    torch_Windows = { ... }
  )
}
```

`keras3` users can then specify a backend like this:

``` r
library(keras3)
use_backend("jax")
```

### Best Practices

Calling `py_require()` from a package is generally safe and recommended. It
ensures dependencies are declared while having no effect on users who manage
their own Python environments. `py_require()` replaces older approaches, such as
listing dependencies in the `DESCRIPTION` file or calling
`use_virtualenv(required = FALSE)` in `.onLoad()`.

Be mindful that other R packages and users may also declare Python requirements.
Avoid restrictive version constraints. If a version constraint is necessary,
prefer `>=` and `!=` over `<=`, as the latter can quickly become outdated. Also,
be mindful that an R package's requirements will be combined with a potentially
wide variety of user requirements, like `exclude_newer`.

An example user script header:

``` r
library(pysparklyr) # declares requirements for PySpark
library(keras3)     # declares requirements for default 'tensorflow' backend
use_backend("jax")  # removes 'tensorflow' requirements, adds 'jax' requirements

library(reticulate)
py_require(c("scipy", "polars"))         # user-declared requirements
py_require(python_version = ">=3.12")
py_require(exclude_newer = "2025-02-20")

np <- import("numpy")  # <-- Python initialized
...
```

### Declaring Optional Dependencies

It's recommended that all `py_require()` calls be made before reticulate
initializes the Python session. However, for rarely used optional dependencies,
the requirement can be declared right before use:

``` r
model_to_dot <- function(x, ...) {
  reticulate::py_require("pydot")
  keras$utils$model_to_dot(x, ...)
}
```

Calling `py_require()` after Python has initialized causes reticulate to
activate a new ephemeral virtual environment containing the additional
requirements. Only adding packages is permitted after Python has initialized;
calling `py_require()` with `action="set"` or `action="remove"` is not possible.

## Delay Loading Python Modules

If your R package wraps Python modules, it's common to import them within
`.onLoad()`. Use the `delay_load` flag in `import()` to allow:

1.  Successful R package loading even when Python packages are not installed
    (important for CRAN testing).
2.  Users to specify their Python installation before using your package.

Example:

``` r
scipy <- NULL

.onLoad <- function(libname, pkgname) {
  reticulate::py_require("scipy")
  scipy <<- reticulate::import("scipy", delay_load = TRUE)
}
```

Without `delay_load`, Python would load immediately, preventing users from
configuring their environment.

## Installing Python Dependencies

`py_require()` is the recommended approach for managing Python dependencies.
However, for users who prefer to manually manage a Python installation, you can
document what Python packages are required.

The `py_install()` function provides a high-level interface for installing
Python packages. The packages will by default be installed within the currently
active Python installation.

``` r
library(reticulate)
py_install("scipy")
```

Alternatively, create a wrapper function for `py_install()` (or
`virtualenv_create()`) that installs dependencies in a dedicated environment:

``` r
install_scipy <- function(envname = "r-scipy", method = "auto", ...) {
  reticulate::py_install("scipy", envname = envname, method = method, ...)
}
```

Note that calling `py_install()` on an ephemeral environment generated from
`py_require()` declared requirements will generate a warning.

## Checking and Testing on CRAN

To ensure your package is well behaved on CRAN:

1.  Use `delay_load` to defer module loading:

    ``` r
    scipy <- NULL

    .onLoad <- function(libname, pkgname) {
      # delay load foo module (will only be loaded when accessed via $)
      scipy <<- reticulate::import("scipy", delay_load = TRUE)
    }
    ```

2.  Skip tests when required modules are unavailable:

    ``` r
    skip_if_no_scipy <- function() {
      if (!reticulate::py_module_available("scipy"))
        skip("scipy not available for testing")
    }

    test_that("Things work as expected", {
      skip_if_no_scipy()
      # test code here...
    })
    ```

## Implementing S3 Methods

Python objects exposed by **reticulate** retain their Python classes in R,
allowing you to define S3 methods for them. This can be useful for customizing
how objects are printed or structured in R. However, Python objects do not
persist across R sessions, meaning an R object that previously pointed to a
Python object will become a `NULL` external pointer when reloaded.

To safely handle these cases, use `py_is_null_xptr()`, as shown in this example:

``` r
print.my_python_object <- function(x, ...) {
  if (py_is_null_xptr(x)) {
    cat("<Python object is no longer available>\n")
  } else {
    cat(py_to_r(x))
  }
}
```

This prevents errors when interacting with a Python object from a previous
session.

This prevents errors when attempting to interact with a Python object from a
previous session.

### Supporting Versions with Different S3 Classes

The Python S3 method for an object is generated from the Python modules and
submodules where the object is defined. In sophisticated Python packages, this
path might change between package versions. For instance, you can access the
`Model` object from `keras.Model` in Python. However, depending on the Keras
Python package version, the actual class definition for `Model` may be located
in a submodule like `keras._internals.src` or `keras._internals.models`, and
since the class module path is considered an internal implementation detail of
the Python package, it can vary across Python package versions. As a result, the
S3 class for the Python object will also change, depending on the Python package
version.

To support changing S3 classes, instead of registering methods in NAMESPACE with
roxygen, manually register them in `.onLoad()`:

``` r
# Python class `DocumentConverterResult` changes with different MarkItDown versions.
py_to_r.markitdown.DocumentConverterResult <- function(x) {
  paste0("# ", x$title, "\n\n", x$text_content)
}

.onLoad <- function(libname, pkgname) {
  reticulate::py_require("markitdown")

  reticulate::py_register_load_hook("markitdown", function() {
    markitdown <- reticulate::import("markitdown")
    registerS3method(
      "py_to_r",
      nameOfClass(markitdown$DocumentConverterResult),
      py_to_r.markitdown.DocumentConverterResult,
      environment(reticulate::py_to_r)
    )
  })
}
```

### Converting between R and Python

**reticulate** provides the generics `r_to_py()` for converting R objects into
Python objects, and `py_to_r()` for converting Python objects back into R
objects. Package authors can provide methods for these generics to convert
Python and R objects otherwise not handled by **reticulate**.

**reticulate** provides conversion operators for some of the most commonly used
Python objects, including:

-   Built-in Python objects (lists, dictionaries, numbers, strings, tuples)
-   NumPy arrays,
-   Pandas objects (`Index`, `Series`, `DataFrame`),
-   Python `datetime` objects.

If you see that **reticulate** is missing support for conversion of one or more
objects from these packages, please [let us
know](https://github.com/rstudio/reticulate/issues) and we'll try to implement
the missing converter. For Python packages not in this set, you can provide
conversion operators in your own extension package.

### Writing your own `r_to_py()` methods

`r_to_py()` accepts a `convert` argument, which controls how objects generated
from the created Python object are converted. To illustrate, consider the
difference between these two cases:

``` r
library(reticulate)

# [convert = TRUE] => convert Python objects to R when appropriate
sys <- import("sys", convert = TRUE)
class(sys$path)
# [1] "character"

# [convert = FALSE] => always return Python objects
sys <- import("sys", convert = FALSE)
class(sys$path)
# [1] "python.builtin.list" "python.builtin.object"
```

This is accomplished through the use of a `convert` flag, which is set on the
Python object wrappers used by `reticulate`. Therefore, if you're writing a
method `r_to_py.foo()` for an object of class `foo`, you should take care to
preserve the `convert` flag on the generated object. This is typically done by:

1.  Passing `convert` along to the appropriate lower-level `r_to_py()` method;

2.  Explicitly setting the `convert` attribute on the returned Python object.

As an example of the second:

``` r
# suppose 'make_python_object()' creates a Python object
# from R objects of class 'my_r_object'.
r_to_py.my_r_object <- function(x, convert) {
  object <- make_python_object(x)
  assign("convert", convert, envir = object)
  object
}
```

## Using GitHub Actions

For testing R packages with GitHub Actions, dependencies declared via
`py_require()` will resolve automatically with no additional steps. If there are
extra Python test dependencies, declare them using `py_require()` in
`tests/testthat/helper.R`. The standard R-CMD-check workflow should work:

``` yaml
- uses: r-lib/actions/setup-r@v2

- uses: r-lib/actions/setup-r-dependencies@v2
  with:
    extra-packages: rcmdcheck

- uses: r-lib/actions/check-r-package@v2
```

Optionally, you can pre-download Python dependencies in a separate step for
cleaner CI logs:

``` yaml
- uses: r-lib/actions/setup-r@v2
  with:
    r-version: release

- uses: r-lib/actions/setup-r-dependencies@v2
  with:
    extra-packages: rcmdcheck local::.

- run: |
    library(mypackage)      # <-- declare requirements in .onLoad()
    reticulate::py_config() # <-- resolves the ephemeral python environment

- uses: r-lib/actions/check-r-package@v2
  # The ephemeral python environment from previous step is reused from cache.
```

If you prefer to use a manually managed Python environment, you can do this:

``` yaml
- uses: actions/setup-python@v4
  with:
    python-version: "3.x"

- name: setup r-reticulate venv
  shell: Rscript {0}
  run: |
    path_to_python <- reticulate::virtualenv_create(
      envname = "r-reticulate",
      python = Sys.which("python"),
      packages = c("numpy", "other-packages")
    )
    writeLines(sprintf("RETICULATE_PYTHON=%s", path_to_python),
               Sys.getenv("GITHUB_ENV"))

- uses: r-lib/actions/check-r-package@v2
```