I am currently trying to set an efficient algorithm to replace 0 values with the surrounding numbers if similar in R. Here is a replication of my data:

ID <- c("FR01", "FR02", "FR03", "FR04")
String <- c("0000001000100100100100220002000200020011", "0222000000001000010101110020020002002022", "0000000000001000010101110020020002002022", "2002220002200202010002222222222222222222")
df <- data.frame(ID, String)
#Results:
result<-df %>% mutate(String=c("1111111111111111111100222222222222220011","2222000000001111111111110022222222222222","1111111111111111111111110022222222222222","2222222222222222010002222222222222222222"))

Condition to replace , for both ways:

• if adjacent number is 0 check next number
• if both adjacent numbers are the same replace by this number
• if adjacent number are different keep 0 except for the start and the end of file where only 1 adjacent number is needed

Results needed

Anyone knows how to efficiently build algorithm to change those string numbers?

Thanks you for your help

CodePudding user response：

Here is something quick:

foo = \(x) {
  y  = unlist(strsplit(x, ""))
  ny = length(y)
  z  = gregexpr("0 ", x)[[1L]]
  if (z[1L] == -1L) return(x)
  for (i in seq_along(z)) {
    ml = attr(z, "match.length")[i]
    if      (i == 1L && ml < ny)       y[1L:ml]          = y[ml 1L]
    else if (z[i] ml > ny)             y[(ny-ml 1L):ny]  = y[ny-ml]
    else if (y[z[i]-1L] == y[z[i] ml]) y[z[i]:(z[i] ml)] = y[z[i] ml]
  }
  paste(y, collapse = "")
}

Example

df = data.frame(
  ID     = c("FR01", "FR02", "FR03"),
  String = c(
    "0000001000100100100100220002000200020010", 
    "0222000000001000010101110020020002002022", 
    "0000000000001000010101110020020002002022"
  )
)

df$result = sapply(df$String, foo)

#     ID                                   String                                   result
# 1 FR01 0000001000100100100100220002000200020010 1111111111111111111100222222222222220011
# 2 FR02 0222000000001000010101110020020002002022 2222000000001111111111110022222222222222
# 3 FR03 0000000000001000010101110020020002002022 1111111111111111111111110022222222222222