superclever

require(dplyr)

#Goal - collapse information from duplicate rows

name<-c("John","John","Frank","Frank","Sue","Sue","Bob")
rank<-c("","sergeant",""," ","lieutenant","major","")
force<-c("army","","navy","navy","army","army","navy")
testdf<-data.frame(name,rank,force)

testdf

#Test 1 dplyr with normal paste
output1 <- testdf %>%
  group_by(name) %>%
  summarize_all (paste,collapse=", ")

#Test 1 - it did what we want -- John's row is collapsed, but we'd rather not see duplicates and orphaned seperators
output1

#Define a conditional paste function

pastefun <- function(x){
  x<-unique(x)
  if(is.na(x[2])) {
    x[1]
  } else if(x[1]==x[2]) {
    x[1] 
  } else if(x[2]=="") {
    x[1]
  } else if(x[1]=="") {
    x[2]
  } else {
    paste(x,collapse=", ")
  } 
}

#test it on various inputs
test1<-"major"
test2<-c("major")
test3<-c("major","")
test4<-c("major","major")
test5<-c("major","lieutenant")
test6<-c("major","major","lieutenant")

#these all work
pastefun(test1)
pastefun(test2)
pastefun(test3)
pastefun(test4)
pastefun(test5)
pastefun(test6)

#this works too, I had expected it to fail
output2 <- testdf %>%
  group_by(name) %>%
  mutate_all(as.character) %>%
  summarize_all (pastefun)
output2

pastefun <- function(x){
  x<-unique(x)
  if(is.na(x[2])) {
    x[1]
  } else if(x[2]=="") {
    x[1]
  } else if(x[1]==x[2]) {
    x[1]
  } else {
    paste(x,collapse=", ")
  } 
}