How to extract outstanding values from an object returned by waldo::compare()?

Question

I'm trying to use a new R package called waldo (see at the tidyverse blog too) that is designed to compare data objects to find differences. The waldo::compare() function returns an object that is, according to the documentation:

a character vector with class "waldo_compare"

The main purpose of this function is to be used within the console, leveraging coloring features to highlight outstanding values that are not equal between data objects. However, while just examining in console is useful, I do want to take those values and act on them (filter them out from the data, etc.). Therefore, I want to programmatically extract the outstanding values. I don't know how.

Example

1. Generate a vector of length 10:

set.seed(2020)

vec_a <- sample(0:20, size = 10)

## [1]  3 15 13  0 16 11 10 12  6 18

2. Create a duplicate vector, and add additional value (4) into an 11th vector element.

vec_b <- vec_a
vec_b[11] <- 4
vec_b <- as.integer(vec_b) 

## [1]  3 15 13  0 16 11 10 12  6 18  4

3. Use waldo::compare() to test the differences between the two vectors

waldo::compare(vec_a, vec_b)

## `old[8:10]`: 12 6 18  
## `new[8:11]`: 12 6 18 4

The beauty is that it's highlighted in the console:

---

But now, how do I extract the different value?

I can try to assign waldo::compare() to an object:

waldo_diff <- waldo::compare(vec_a, vec_b)

and then what? when I try to do waldo_diff[[1]] I get:

[1] "`old[8:10]`: \033[90m12\033[39m \033[90m6\033[39m \033[90m18\033[39m  \n`new[8:11]`: \033[90m12\033[39m \033[90m6\033[39m \033[90m18\033[39m \033[34m4\033[39m"

and for waldo_diff[[2]] it's even worse:

Error in waldo_diff[3] : subscript out of bounds

Any idea how I could programmatically extract the outstanding values that appear in the "new" vector but not in the "old"?

Answer 1

As a disclaimer, I didn't know anything about this package until you posted so this is far from an authoritative answer, but you can't easily extract the different values using the compare() function as it returns an ANSI formatted string ready for pretty printing. Instead the workhorses for vectors seem to be the internal functions ses() and ses_context() which return the indices of the differences between the two objects. The difference seems to be that ses_context() splits the result into a list of non-contiguous differences.

waldo:::ses(vec_a, vec_b)

# A tibble: 1 x 5
     x1    x2 t        y1    y2
  <int> <int> <chr> <int> <int>
1    10    10 a        11    11

The results show that there is an addition in the new vector beginning and ending at position 11.

The following simple function is very limited in scope and assumes that only additions in the new vector are of interest:

new_diff_additions <- function(x, y) {
  res <- waldo:::ses(x, y)
  res <- res[res$t == "a",]  # keep only additions
  if (nrow(res) == 0) {
    return(NULL)
  }  else {
    Map(function(start, end) {
      d <- y[start:end]
      `attributes<-`(d, list(start = start, end = end))
    },
    res[["y1"]], res[["y2"]])
  }
}
    
new_diff_additions(vec_a, vec_b)

[[1]]
[1] 4
attr(,"start")
[1] 11
attr(,"end")
[1] 11

Answer 2

At least for the simple case of comparing two vectors, you’ll be better off using diffobj::ses_dat() (which is from the package that waldo uses under the hood) directly:

waldo::compare(1:3, 2:4)
#> `old`: 1 2 3  
#> `new`:   2 3 4

diffobj::ses_dat(1:3, 2:4)
#>       op val id.a id.b
#> 1 Delete   1    1   NA
#> 2  Match   2    2   NA
#> 3  Match   3    3   NA
#> 4 Insert   4   NA    3

For completeness, to extract additions you could do e.g.:

extract_additions <- function(x, y) {
  ses <- diffobj::ses_dat(x, y)
  y[ses$id.b[ses$op == "Insert"]]
}

old <- 1:3
new <- 2:4

extract_additions(old, new)
#> [1] 4