Return ses() output in more computable form?

[hadley]

i.e.. I'm doing this:

ses <- function(x, y) {
  out <- diffobj::ses(x, y, max.diffs = 100)
  out <- rematch2::re_match(out, paste0(
    "(?:(?<x1>\\d+),)?(?<x2>\\d+)",
    "(?<t>[acd])",
    "(?:(?<y1>\\d+),)?(?<y2>\\d+)"
  ))[1:5]

  out$x1 <- ifelse(out$x1 == "", out$x2, out$x1)
  out$y1 <- ifelse(out$y1 == "", out$y2, out$y1)

  out$x1 <- as.integer(out$x1)
  out$x2 <- as.integer(out$x2)
  out$y1 <- as.integer(out$y1)
  out$y2 <- as.integer(out$y2)

  out
}

but I'm sure you could compute that more efficiently at a lower level.

[brodieG]

Internally diffobj::ses calls diffobj:::diff_myers and the returns the raw data (note ?diff_myers is out of date, I'm realizing):

> xx <- diffobj:::diff_myers(letters[1:5], letters[2:6])
> str(xx)
Formal class 'MyersMbaSes' [package "diffobj"] with 6 slots
  ..@ a     : chr [1:5] "a" "b" "c" "d" ...
  ..@ b     : chr [1:5] "b" "c" "d" "e" ...
  ..@ type  : Factor w/ 3 levels "Match","Insert",..: 3 1 2
  ..@ length: int [1:3] 1 4 1
  ..@ offset: int [1:3] 1 2 5
  ..@ diffs : int 2

The strings ses outputs are produced by the as.character method used by show.

So short term you can get the data that way. Longer term I can add a method that returns this.

[brodieG]

Also, if you can give me a sense of what time frame this feature would be useful to you (i.e. when is too late to be useful) let me know. I'm way behind on all sorts of things so absent specific timelines things are likely to sit for a while.

[hadley]

I’m already using the function above so there’s no rush. The only advantage would be a small performance boost, but I doubt it’s a bottleneck for me anyway.

[brodieG]

Just added ses_dat which I think gives you the data in a good format for your pruposes. This is the example. It's now in the development branch.

[hadley]

Thanks!