Why does the digest function return the same value every time when used with dplyr's mutate?

Question

Here's a data frame containing a column of user ids:

> head(df)
       uid
1 14070210
2 14080815
3 14091420

For the sake of argument, I want to create a new column containing the square root of the user id, and another new column containing a hash of the user id. So I do this:

df_mutated <- df %>%
              mutate(sqrt_uid = sqrt(uid), hashed_uid = digest(uid))

... where digest() comes from the digest package.

While the square root appears to work, the digest function returns the same value for each user id.

> head(df_mutated)
       uid sqrt_uid                       hashed_uid
1 14070210 3751.028 f8c4b39403e57d85cd1698d2353954d0
2 14080815 3752.441 f8c4b39403e57d85cd1698d2353954d0
3 14091420 3753.854 f8c4b39403e57d85cd1698d2353954d0

This is weird to me. Without dplyr, the digest() function returns different values for different inputs. What am I not understanding about dplyr?

Thanks

MrFlick MrFlick · Accepted Answer · 2015-02-06T06:48:43

The digest() function isn't vectorized. So if you pass in a vector, you get one value for the whole vector rather than a digest for each element of the vector. Since it returns one value, that value is recycled for each row of your data.frame. You can create your own vectorized version

vdigest <- Vectorize(digest)
df %>% mutate(sqrt_uid = sqrt(uid), hashed_uid = vdigest(uid))
#        uid sqrt_uid                       hashed_uid
# 1 14070210 3751.028 cc90019421220a24f75b5ed5daec36ff
# 2 14080815 3752.441 9f7f643940b692dd9c7effad439547e8
# 3 14091420 3753.854 89e6666fdfdbfb532b2d7940def9d47d

which matches what you get when you pass in each vector element individually

digest(df$uid[1])
# [1] "cc90019421220a24f75b5ed5daec36ff"
digest(df$uid[3])
# [1] "89e6666fdfdbfb532b2d7940def9d47d"

Why does the digest function return the same value every time when used with dplyr's mutate?

1 Answers