R data.table split a row into multiple rows based on string values

Question

I have a data.table like this:

dt <- data.table("title"=c("First Title", "Second Title", "Third Title", "Fourth Title"), 
                 "sha"=c("12345", "2345; 66543; 33423", "22222; 12345678;", "666662345; 444"))

How can I get this:

      title                sha
1:  First Title           12345
2: Second Title           2345 
3: Second Title           66543
4: Second Title           33423
5:  Third Title           22222
6:  Third Title           12345678
7: Fourth Title           666662345
8: Fourth Title           444

Thanks in advance!

score 8 · Answer 1 · answered May 08 '20 at 20:52

Here is another solution using data.table:

dt[, .(sha = unlist(tstrsplit(sha, ";", type.convert = TRUE))), by = "title"]

#           title       sha
# 1:  First Title     12345
# 2: Second Title      2345
# 3: Second Title     66543
# 4: Second Title     33423
# 5:  Third Title     22222
# 6:  Third Title  12345678
# 7: Fourth Title 666662345
# 8: Fourth Title       444

Ian Campbell · Accepted Answer · 2020-05-08T17:47:37.150

Here's one approach with tstrsplit that should work for you:

library(data.table)
dt[, lapply(.SD, function(x) unlist(tstrsplit(x, "; ?"))),
   .SDcols = "sha",by = c("title","date")]
          title     date       sha
1:  First Title 1/1/2020     12345
2: Second Title 1/2/2020      2345
3: Second Title 1/2/2020     66543
4: Second Title 1/2/2020     33423
5:  Third Title 1/3/2020     22222
6:  Third Title 1/3/2020  12345678
7: Fourth Title 1/4/2020 666662345
8: Fourth Title 1/4/2020       444

Data

dt <- data.table("title"=c("First Title", "Second Title", "Third Title", "Fourth Title"), 
                  "sha"=c("12345", "2345; 66543; 33423", "22222; 12345678;", "666662345; 444"),
                  "date" = c("1/1/2020","1/2/2020","1/3/2020","1/4/2020"))

is it possible to keep other columns not just title and sha after this operation I have for example also a date column — Blind0ne, May 08 '20 at 17:43

score 3 · Answer 3 · answered May 08 '20 at 17:43

3

Here's my dplyr solution:

dt %>% 
  group_by(title) %>% 
  separate_rows(sha, sep = ";") %>% 
  mutate(sha = as.numeric(sha)) %>% 
  filter(!is.na(sha))

Should give you this:

# A tibble: 8 x 2
# Groups:   title [4]
  title              sha
  <chr>            <dbl>
1 First Title      12345
2 Second Title      2345
3 Second Title     66543
4 Second Title     33423
5 Third Title      22222
6 Third Title   12345678
7 Fourth Title 666662345
8 Fourth Title       444

answered May 08 '20 at 17:43

BellmanEqn

791
3
11

1

@Blind0ne a more simplistic approach: `dt %>% separate_rows(sha, sep = ";") %>% na_if("") %>% drop_na()` – nikn8 May 08 '20 at 17:51
@Neel Thank you! – BellmanEqn May 08 '20 at 17:55
You don't actually need `drop_na()`, but you do need to filter on empty vector. `separate_rows(dt,sha) %>% filter(sha != "")`. – Ian Campbell May 08 '20 at 18:01

R data.table split a row into multiple rows based on string values

3 Answers3