Compare two data.frames and generate the code to patch the differences

make_patch.R

# function to identify what changed where
which_changed <- function(x, y = x) {
  
  # argument validation
  stopifnot(inherits(x, "data.frame"))
  stopifnot(inherits(y, "data.frame"))
  stopifnot(identical(dim(x), dim(y)))
  y <- y[, names(x)]
  
  # compare objects
  w <- which(as.vector(!(x == y)))
  if (length(w)) {
      rowpos <- w %% nrow(x)
      colpos <- names(x)[(w %/% nrow(x)) + 1]
      
      values <- list()
      for (i in seq_along(rowpos)) {
          values[[i]] <- y[rowpos[i], colpos[i], drop = FALSE]
      }
      
      out <- mapply(function(row, col, val) {
          list(x = row, y = col, value = val)
      }, rowpos, colpos, values, SIMPLIFY = FALSE)
  } else {
    out <- list()  
  }
  structure(out, class = "data_diff")
}

# print function
print.data_diff <- function(x, ...) {
    message(paste0("Data differs in ", length(x), ngettext(length(x), " place", " places")))
    invisible(x)
}

# function to make code to change `x` into `y`
make_data_patch <- function(x, y = x) {
  diff <- which_changed(x, y)
  if (length(diff)) {
      xname <- as.character(substitute(x))
      changes <- lapply(diff, function(change) {
          paste0(xname, "[", change[["x"]], ", '", change[["y"]], "'] <- ", change[["value"]])
      })
      out <- paste0(unlist(changes), collapse = "\n")
  } else {
      out <- paste0("# no value changes to make")
  }
}

# Examples

# example with no changes
(chg1 <- which_changed(mtcars, mtcars))
## Data differs in 0 places
p1 <- make_patch_code(mtcars, mtcars)
cat(p1, "\n")
## # no value changes to make


# example with changes
mtcars1 <- mtcars
mtcars2 <- mtcars
mtcars2[3,4] <- 99
mtcars2[2,1] <- 0
mtcars2[6,8] <- -99

(chg2 <- which_changed(mtcars1, mtcars2))
## Data differs in 3 places
p2 <- make_patch_code(mtcars1, mtcars2)
cat(p2, "\n")
## mtcars1[2, 'mpg'] <- 0
## mtcars1[3, 'hp'] <- 99
## mtcars1[6, 'vs'] <- -99