R根据行名传播数据帧

Question

I have a Data Frame with two columns. Row names are duplicated as the data domes from a list of reports with some common fields. Each report contains a different number of fields. I want to spread this data frame into multiple columns based on one of these duplicated row names. The end result would have each report in a row.

These reports come from an API that exists on a system at work. It returns a very nested JSON. I wanted to see if getting the data in to this format would provide me a way to clean up the data.

Minimal Example of Data

Column1       Column2
contentID      123
value1         California
value2         truck
value3         home
contentID      897
value1         Georgia
value2         car
value3         work
value4         boeing   
contentID      537
value2         truck
value4         private
value5         first class
value6         wheels

Desired outcome

ContentID   value1     value2 value3 value4   value5      value6
123         California truck  home   NA       NA          NA
897         Georgia    car    work   boeing   NA          NA  
537         NA         truck  NA     private  firstclass  wheels

Answer 1

One tidyverse possibility could be:

df %>%
 mutate(id = cumsum(grepl("content", Column1))) %>%
 group_by(id) %>%
 mutate(ContentID = first(Column2)) %>%
 filter(!grepl("content", Column1)) %>%
 ungroup() %>%
 select(-id) %>%
 spread(Column1, Column2)

  ContentID value1     value2 value3 value4  value5      value6
  <chr>     <chr>      <chr>  <chr>  <chr>   <chr>       <chr> 
1 123       California truck  home   <NA>    <NA>        <NA>  
2 537       <NA>       truck  <NA>   private first_class wheels
3 897       Georgia    car    work   boeing  <NA>        <NA>  

Here it, first, creates an ID variable based on the occurrence of content in "Column1" and groups by it. Second, it creates a "ContentID" variable with the values from the first row on "Column2" per group. Third, it filters out the rows that contains content in "Column1". Finally, it spreads the data.

Answer 2

You can simply do this-

library(data.table)
library(zoo)
setDT(dt)
dt[,id:=ifelse(Column1 %like% "contentID",paste(Column2),NA)]
dt[,id:=na.locf(id)]
dcast.data.table(dt,id~Column1,value.var="Column2",subset = .(Column1!="contentID"))


 id     value1 value2 value3  value4     value5 value6
1: 123 California  truck   home    <NA>       <NA>   <NA>
2: 537       <NA>  truck   <NA> private firstclass wheels
3: 897    Georgia    car   work  boeing       <NA>   <NA>

Note - It will be efficient if you have large dataset.