来自字符向量的数据帧，其中变量名称及其数据被联合存储

Question

我有这种情况：

foo <- data.frame("vars" = c("animal: mouse | wks: 12 | site: cage | PI: 78",
                            "animal: dog | wks: 32 | GI: 0.2",
                            "animal: cat | wks: 8 | site: wild | PI: 13"))

其中变量名和相对数据存储在字符串中，如上例所示。特别是，每个variable_name/its_data单元由|分隔。在之后：是相对数据。

我想要一个这样的最终数据帧：

  animal  wks  site  PI   GI
  mouse   12   cage  78   NA
    dog   32   <NA>  NA  0.2
    cat    8   wild  13   NA

Answer 1

下面是一个dplyr解决方案：

library(dplyr)
library(tidyr)

tibble(foo) %>%
  mutate(row = row_number()) %>% 
  separate_rows(vars, sep = '\\|') %>% 
  separate(vars, c("a", "b"), sep = '\\:') %>% 
  mutate(across(everything(), str_trim)) %>% 
  group_by(a) %>% 
  pivot_wider(names_from = a, values_from = b) %>% 
  type.convert(as.is = TRUE) %>% 
  select(-row)

  animal   wks site     PI    GI
  <chr>  <int> <chr> <int> <dbl>
1 mouse     12 cage     78  NA  
2 dog       32 NA       NA   0.2
3 cat        8 wild     13  NA 

Answer 2

另一个使用Reducemerge

type.convert(
  Reduce(
    function(x, y) merge(x, y, all = TRUE),
    lapply(
      strsplit(foo$vars, ":|\\|"),
      function(x) {
        m <- matrix(trimws(x), 2)
        setNames(data.frame(m[2, , drop = FALSE]), m[1, ])
      }
    )
  ),
  as.is = TRUE
)

给予

  animal wks site PI  GI
1    cat   8 wild 13  NA
2    dog  32 <NA> NA 0.2
3  mouse  12 cage 78  NA

Answer 3

下面是一种使用scan（）行方式，然后使用merge。

lapply(1:nrow(foo), \(x) 
       scan(text=foo[x, ], what=character(), sep='|', strip.white=T, qui=T) |>
  (\(.) do.call(rbind, strsplit(., ': ')))() |>
  (\(.) setNames(data.frame(t(.[, 2])), .[, 1]))()) |>
  (\(.) Reduce(\(...) merge(..., all=TRUE), .))()
#   animal wks site   PI   GI
# 1    cat   8 wild   13 <NA>
# 2    dog  32 <NA> <NA>  0.2
# 3  mouse  12 cage   78 <NA>

注：R