如何让自定义R6类NumericInterval作为data.frame列类型正常工作？

Let's break down why you're seeing that error with base data.frame, and walk through two solutions that meet your requirements: displaying the interval string in the table while keeping the R6 objects intact for modification.

Why Base data.frame Throws an Error #

Base R's data.frame expects columns to be atomic vectors (like numeric, character) or structures that can be coerced to a data frame. R6 objects are reference-type objects, so as.data.frame.default doesn't know how to handle them—hence the error you ran into.

Solution 1: Make Base data.frame Work with Your R6 Class #

We'll create a custom class for lists of NumericInterval objects, then define formatting methods so the data frame displays interval strings while storing the actual R6 objects.

Step 1: Extend Your Existing Code #

Keep your original R6 class and S3 methods, then add code to handle list columns:

require(R6)

# Your original NumericInterval class
NumericInterval <- R6Class(
  "NumericInterval",
  public = list(
    lower_bound = NA,
    upper_bound = NA,
    initialize = function(low, up) {
      self$lower_bound <- low
      self$upper_bound <- up
    },
    as_character = function() {
      paste0("[", self$lower_bound, ", ", self$upper_bound, "]")
    }
  )
)

# Your existing S3 methods
as.character.NumericInterval <- function(x, ...) {
  x$as_character()
}
print.NumericInterval <- function(x, ...) {
  cat(x$as_character(), "\n")
}

# New: Custom class for lists of NumericInterval objects
as.NumericIntervalList <- function(x) {
  stopifnot(all(sapply(x, inherits, "NumericInterval")))
  structure(x, class = "NumericIntervalList")
}

# New: Format method to display interval strings
format.NumericIntervalList <- function(x, ...) {
  sapply(unclass(x), as.character)
}

Step 2: Create the Data Frame #

Use I() to prevent base data.frame from trying to coerce the list column, and wrap your R6 objects with our custom list class:

df <- data.frame(
  X = c("I1", "I2", "I3"),
  Y = I(as.NumericIntervalList(list(
    NumericInterval$new(0,1),
    NumericInterval$new(1,2),
    NumericInterval$new(2,3)
  ))),
  stringsAsFactors = FALSE
)

Step 3: Verify Behavior #

Now the data frame displays correctly, and you can modify the R6 objects directly:

print(df)
#   X       Y
# 1 I1 [0, 1]
# 2 I2 [1, 2]
# 3 I3 [2, 3]

# Modify an object
df$Y[[2]]$lower_bound <- 0

# Print again to see the update
print(df)
#   X       Y
# 1 I1 [0, 1]
# 2 I2 [0, 2]
# 3 I3 [2, 3]

# Access object properties
df$Y[[2]]$upper_bound
# [1] 2

Solution 2: Optimize Tibble Display for Your R6 Class #

Tibbles already support list columns of R6 objects, but we can fix the display using the pillar package to show interval strings instead of <NmrcIntr>.

Step 1: Add a Pillar Formatting Method #

Keep your original R6 class and S3 methods, then add a pillar_shaft method for NumericInterval:

require(R6)
require(tibble)
require(pillar)

# Your original NumericInterval class and S3 methods (same as before)
NumericInterval <- R6Class(
  "NumericInterval",
  public = list(
    lower_bound = NA,
    upper_bound = NA,
    initialize = function(low, up) {
      self$lower_bound <- low
      self$upper_bound <- up
    },
    as_character = function() {
      paste0("[", self$lower_bound, ", ", self$upper_bound, "]")
    }
  )
)

as.character.NumericInterval <- function(x, ...) {
  x$as_character()
}
print.NumericInterval <- function(x, ...) {
  cat(x$as_character(), "\n")
}

# New: Pillar formatting method for tibble display
pillar_shaft.NumericInterval <- function(x, ...) {
  new_pillar_shaft_simple(as.character(x), align = "left")
}

Step 2: Create the Tibble #

Just pass your R6 objects as a list column:

df <- tibble(
  X = c("I1", "I2", "I3"),
  Y = list(
    NumericInterval$new(0,1),
    NumericInterval$new(1,2),
    NumericInterval$new(2,3)
  )
)

Step 3: Verify Behavior #

Now the tibble displays interval strings, and you can modify the R6 objects seamlessly:

print(df)
# # A tibble: 3 × 2
#   X     Y          
#   <chr> <chr>      
# 1 I1    [0, 1]
# 2 I2    [1, 2]
# 3 I3    [2, 3]

# Modify an object
df$Y[[2]]$lower_bound <- 0

# See the update
print(df)
# # A tibble: 3 × 2
#   X     Y          
#   <chr> <chr>      
# 1 I1    [0, 1]
# 2 I2    [0, 2]
# 3 I3    [2, 3]

Which Solution Should You Choose? #

• Use the base data.frame solution if you prefer working with traditional R data frames and want behavior consistent with base R.
• Use the tibble solution if you're part of the tidyverse ecosystem and want a more streamlined setup for list columns.

内容的提问来源于stack exchange，提问作者pietrodito