how to calculate the cumulative sum of sequences of values

I have a matrix

x <- cbind(label = c(1,1,1,-1,-1,0,0,-1,1,1),
            val = c(1,3,7,5,4,2,6,8,5,1))
 x
      label val
 [1,]     1   1
 [2,]     1   3
 [3,]     1   7
 [4,]    -1   5
 [5,]    -1   4
 [6,]     0   2
 [7,]     0   6
 [8,]    -1   8
 [9,]     1   5
[10,]     1   1

I want to calculate the cumulative sum of sequences of values in the val, but the sequence is determined by the label column.
That is:

1. label 1,1,1 cumsum val 1, 1+3, 1+4+7
2. label -1,-1 cumsum val -5, -5 + -4
3. if the label is zero then we ignore it
   and so on..

label can only have three values -1, 0, 1

Also I want to have a column with the last value from each cumulative sum
In the final result I would like to get a matrix like this

res_x <- cbind(x,
                cum_val = c(1,4,11,-5,-9,0,0,-8,5,6),
                last_cum_val = c(NA,NA,11,NA,-9,NA,NA,8,NA,6))
 
 res_x
      label val cum_val last_cum_val
 [1,]     1   1       1           NA
 [2,]     1   3       4           NA
 [3,]     1   7      11           11
 [4,]    -1   5      -5           NA
 [5,]    -1   4      -9           -9
 [6,]     0   2       0           NA
 [7,]     0   6       0           NA
 [8,]    -1   8      -8           -8
 [9,]     1   5       5           NA
[10,]     1   1       6            6

>Solution :

x |> 
  mutate(group = c(0, cumsum(diff(label) != 0))) |> 
  mutate(cum_val = sign(label) *  cumsum(val), .by = group) |> 
  mutate(last_cum_val = ifelse(row_number() == n(), last(cum_val), NA), .by = group)

      label val group cum_val last_cum_val
[1,]      1   1     0       1           NA
[2,]      1   3     0       4           NA
[3,]      1   7     0      11           11
[4,]     -1   5     1      -5           NA
[5,]     -1   4     1      -9           -9
[6,]      0   2     2       0           NA
[7,]      0   6     2       0            0
[8,]     -1   8     3      -8           -8
[9,]      1   5     4       5           NA
[10,]     1   1     4       6            6