json - 避免数据框到嵌套 JSON 的嵌套循环

Question

给定一个表示某种层次结构的 data_frame，我想将此数据转换为具有特定结构的嵌套 JSON。

给定这个data_frame

df <- data_frame(
    "parent" = c("A", "A", "A", "A", "A", "B", "B", "B", "B", "C", "C", "C", "C", "C", "C"),
    "child1" = c("a", "a", "b", "b", "c", "d", "d", "e", "e", "f", "f", "f", "f", "g", "g"),
    "child2" = c("aa", "ab", "ba", "bb", "ca", "da", "db", "ea", "eb", "fa", "fb", "fc", "fd", "ga", "gb"),
    "value" = sample(seq(1,100,1), 15)
)

我想生成以下 JSON 结构:

[
  {
    "name": "A",
    "children": [
      {
        "name": "a",
        "children": [
          {"name": "aa", "value": 89},
          {"name": "ab", "value": 20}
        ]
      },
      {
        "name": "b",
        "children": [
          {"name": "ba", "value": 25},
          {"name": "bb", "value": 15}
        ]
      },
      {
        "name": "c",
        "children": [
          {"name": "ca","value": 95}
        ]
      }
    ]
  },
  {
    "name": "B",
    "children": [
      {
        "name": "d",
        "children": [
          {"name": "da", "value": 54},
          {"name": "db", "value": 62}
        ]
      },
      {
        "name": "e",
        "children": [
          {"name": "ea", "value": 100},
          {"name": "eb", "value": 56}
        ]
      }
    ]
  },
  {
    "name": "C",
    "children": [
      {
        "name": "f",
        "children": [
          {"name": "fa", "value": 69},
          {"name": "fb", "value": 98},
          {"name": "fc", "value": 83},
          {"name": "fd", "value": 63}
        ]
      },
      {
        "name": "g",
        "children": [
          {"name": "ga", "value": 91},
          {"name": "gb", "value": 77}
        ]
      }
    ]
  }
] 

目前我使用嵌套循环构造嵌套列表如下:

lll <- list()
i <- 1

for (a in unique(df$parent)) {

  lll[[i]] <- list(
    "name" = a,
    "children" = list()
  )

  ii <- 1

  for (b in unique(df$child1[df$parent == a])) {
    lll[[i]]$children[[ii]] <- list(
      "name" = b,
      "children" = list()
    )

    iii <- 1

    for(c in unique(df$child2[df$parent == a & df$child1 == b])) {

      lll[[i]]$children[[ii]]$children[[iii]] <- list(
        "name" = c,
        "value" = df$value[df$parent == a & df$child1 == b & df$child2 == c ]
      )

      iii <- iii + 1

    }

    ii <- ii + 1
  }

  i <- i + 1

}

使用 jsonlite::toJSON(lll, pretty = TRUE, auto_unbox = TRUE) 可以将此列表转换为嵌套的 JSON。

不知道有没有更优雅的方式。我尝试使用 purrr 解决这个问题，但没有成功。

Answer 1

您可以通过结合使用 dplyr::group_by() 和 tidyr::nest() 来实现:

library(dplyr)
library(tidyr)

df %>% 
  rename(name = child2) %>% 
  group_by(parent, child1) %>% 
  nest(.key = "children") %>% 
  rename(name = child1) %>% 
  group_by(parent) %>% 
  nest(.key = "children") %>% 
  rename(name = parent) %>% 
  jsonlite::toJSON(pretty = TRUE, auto_unbox = TRUE)
#> [
#>   {
#>     "name": "A",
#>     "children": [
#>       {
#>         "name": "a",
#>         "children": [
#>           {
#>             "name": "aa",
#>             "value": 64
#>           },
#>           {
#>             "name": "ab",
#>             "value": 29
#>           }
#>         ]
#>       },
#>       {
#>         "name": "b",
#>         "children": [
#>           {
#>             "name": "ba",
#>             "value": 73
#>           },
#>           {
#>             "name": "bb",
#>             "value": 45
#>           }
#>         ]
#>       },
#>       {
#>         "name": "c",
#>         "children": [
#>           {
#>             "name": "ca",
#>             "value": 95
#>           }
#>         ]
#>       }
#>     ]
#>   },
#>   {
#>     "name": "B",
#>     "children": [
#>       {
#>         "name": "d",
#>         "children": [
#>           {
#>             "name": "da",
#>             "value": 26
#>           },
#> ...

为了重现您的列名，代码通过调用 dplyr::rename 变得更加笨拙。没有它们，操作的结构将变得更加明显:

df %>% 
  group_by(parent, child1) %>% 
  nest() %>% 
  group_by(parent) %>% 
  nest() %>% 
  jsonlite::toJSON(pretty = TRUE, auto_unbox = TRUE)