From 3fe7ce1e1065d25baa6f44d4dcb35cd64c31c7b3 Mon Sep 17 00:00:00 2001
From: Sebastian Fischer <sebf.fischer@gmail.com>
Date: Mon, 22 Jan 2024 19:51:40 +0100
Subject: [PATCH] feat(calculate_hash): allow to customize hash input

---
 NAMESPACE             |  4 ++++
 R/calculate_hash.R    | 42 ++++++++++++++++++++++++++++++++----------
 man/calculate_hash.Rd |  5 ++++-
 man/hash_input.Rd     | 15 +++++++++++++++
 4 files changed, 55 insertions(+), 11 deletions(-)
 create mode 100644 man/hash_input.Rd

diff --git a/NAMESPACE b/NAMESPACE
index b7f35c90..4dcbe096 100644
--- a/NAMESPACE
+++ b/NAMESPACE
@@ -10,6 +10,9 @@ S3method(discard,default)
 S3method(distinct_values,default)
 S3method(distinct_values,factor)
 S3method(distinct_values,logical)
+S3method(hash_input,"function")
+S3method(hash_input,data.table)
+S3method(hash_input,default)
 S3method(insert_named,"NULL")
 S3method(insert_named,data.frame)
 S3method(insert_named,data.table)
@@ -82,6 +85,7 @@ export(formulate)
 export(get_private)
 export(get_seed)
 export(has_element)
+export(hash_input)
 export(ids)
 export(imap)
 export(imap_chr)
diff --git a/R/calculate_hash.R b/R/calculate_hash.R
index 1810a982..aae0c735 100644
--- a/R/calculate_hash.R
+++ b/R/calculate_hash.R
@@ -2,8 +2,11 @@
 #'
 #' @description
 #' Calls [digest::digest()] to calculate the hash for all objects provided.
+#' The hash is calculated using the [xxhash64] algorithm.
+#' By specifying methods for the [`hash_input`] generic, you can control which information of an object
+#' is used to calculate the hash.
 #'
-#' The following operations are performed to make hashing more robust:
+#' Methods exist for:
 #' * If an object is a [function()], the formals and the body are hashed separately.
 #'   This ensures that the bytecode or parent environment are not be included
 #'   in the hash.
@@ -21,13 +24,32 @@
 #' @examples
 #' calculate_hash(iris, 1, "a")
 calculate_hash = function(...) {
-  digest(lapply(list(...), function(x) {
-    if (is.function(x)) {
-      list(formals(x), as.character(body(x)))
-    } else if (is.data.table(x)) {
-      as.list(x)
-    } else {
-      x
-    }
-  }), algo = "xxhash64")
+  digest(lapply(list(...), hash_info), algo = "xxhash64")
 }
+
+#' Hash Input
+#'
+#' Returns the information of an object to be used to calculate its hash.
+#' @param x (any)\cr
+#'   Object for which to calculate the hash.
+#' @export
+hash_input = function(x) {
+  UseMethod("hash_input")
+}
+
+#' @export
+hash_input.function = function(x) {
+  list(formals(x), as.character(body(x)))
+}
+
+#' @export
+#' @method hash_input data.table
+hash_input.data.table = function(x) {
+  as.list(x)
+}
+
+#' @export
+hash_input.default = function(x) {
+  x
+}
+
diff --git a/man/calculate_hash.Rd b/man/calculate_hash.Rd
index 9a564e8d..b6a35087 100644
--- a/man/calculate_hash.Rd
+++ b/man/calculate_hash.Rd
@@ -15,8 +15,11 @@ Objects to hash.}
 }
 \description{
 Calls \code{\link[digest:digest]{digest::digest()}} to calculate the hash for all objects provided.
+The hash is calculated using the \link{xxhash64} algorithm.
+By specifying methods for the \code{\link{hash_input}} generic, you can control which information of an object
+is used to calculate the hash.
 
-The following operations are performed to make hashing more robust:
+Methods exist for:
 \itemize{
 \item If an object is a \code{\link[=function]{function()}}, the formals and the body are hashed separately.
 This ensures that the bytecode or parent environment are not be included
diff --git a/man/hash_input.Rd b/man/hash_input.Rd
new file mode 100644
index 00000000..96640411
--- /dev/null
+++ b/man/hash_input.Rd
@@ -0,0 +1,15 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/calculate_hash.R
+\name{hash_input}
+\alias{hash_input}
+\title{Hash Input}
+\usage{
+hash_input(x)
+}
+\arguments{
+\item{x}{(any)\cr
+Object for which to calculate the hash.}
+}
+\description{
+Returns the information of an object to be used to calculate its hash.
+}