将嵌入列表的json解析为半长数据帧

Question

我有一个带有多层嵌套的json文件，但我正努力将其放入可行的数据框中。我创建了一个基于真实结构的模拟数据玩具示例：here是要点。

这是我想要的输出。输出可能是“更长”的，也可能是原始json中有其他变量，但我正在显示核心问题。

这是json的一部分，它显示了我想要进入半长格式的嵌套的最深层次，如上面的白色所示（完全宽的格式就可以了）。

我已经用这个对象尝试了很多事情：

myList <- jsonlite::fromJSON("example.json", flatten=TRUE)$results

从尝试对[][[]]和cbind()进行子集化，到尝试对嵌套列表进行嵌套的其他尝试。没什么好说的。从最佳方法的建议中我将受益匪浅。

Answer 1

这会让您更进一步吗？ （这是一个粗糙的结构）：

library(tidyverse)

x <- (jsonlite::fromJSON("/Users/hrbrmstr/r7/gh/labs-research/2018-11-portland-ciso-event/example.json"))

jsonlite::stream_out(x$results, con = gzfile("ex-res.json.gz"))

y <- ndjson::stream_in("ex-res.json.gz", "tbl")

gather(y, path, path_val, starts_with("path")) %>%
  gather(flow, flow_val, starts_with("flow")) %>%
  gather(name, name_val, starts_with("values.pdep")) %>%
  gather(intervention, interv_val, starts_with("values.inter")) %>%
  glimpse()
## Observations: 87,696
## Variables: 18
## $ contact.name <chr> "Person 1", "Person 2", "Person 1", "Person 2", "Person 1", "Person 2", "Person 1", "Person 2"...
## $ contact.uuid <chr> "k0dcjs", "rd3jfui", "k0dcjs", "rd3jfui", "k0dcjs", "rd3jfui", "k0dcjs", "rd3jfui", "k0dcjs", ...
## $ created_on   <chr> "2016-02-08T07:00:15.093813Z", "2016-02-08T07:00:15.093813Z", "2016-02-08T07:00:15.093813Z", "...
## $ id           <dbl> 1234, 1235, 1234, 1235, 1234, 1235, 1234, 1235, 1234, 1235, 1234, 1235, 1234, 1235, 1234, 1235...
## $ modified_on  <chr> "2016-02-09T04:42:54.812323Z", "2016-02-08T08:09:51.545160Z", "2016-02-09T04:42:54.812323Z", "...
## $ responded    <lgl> TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE...
## $ start.uuid   <chr> "dnxh4g", "kfj4dsi", "dnxh4g", "kfj4dsi", "dnxh4g", "kfj4dsi", "dnxh4g", "kfj4dsi", "dnxh4g", ...
## $ uuid         <chr> "esn4dk", "qask9dj", "esn4dk", "qask9dj", "esn4dk", "qask9dj", "esn4dk", "qask9dj", "esn4dk", ...
## $ exit_type    <chr> NA, "completed", NA, "completed", NA, "completed", NA, "completed", NA, "completed", NA, "comp...
## $ exited_on    <chr> NA, "2016-02-08T08:09:51.544998Z", NA, "2016-02-08T08:09:51.544998Z", NA, "2016-02-08T08:09:51...
## $ path         <chr> "path.0.node", "path.0.node", "path.0.time", "path.0.time", "path.1.node", "path.1.node", "pat...
## $ path_val     <chr> "ecb4cb11-6cca-4791-a950-c448e9300846", "ecb4cb11-6cca-4791-a950-c448e9300846", "2016-02-08T07...
## $ flow         <chr> "flow.name", "flow.name", "flow.name", "flow.name", "flow.name", "flow.name", "flow.name", "fl...
## $ flow_val     <chr> "weeklyratings", "weeklyratings", "weeklyratings", "weeklyratings", "weeklyratings", "weeklyra...
## $ name         <chr> "values.pdeps1.category", "values.pdeps1.category", "values.pdeps1.category", "values.pdeps1.c...
## $ name_val     <chr> "0 - 7", "0 - 7", "0 - 7", "0 - 7", "0 - 7", "0 - 7", "0 - 7", "0 - 7", "0 - 7", "0 - 7", "0 -...
## $ intervention <chr> "values.intervention", "values.intervention", "values.intervention", "values.intervention", "v...
## $ interv_val   <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA...

完全方法：

gather(y, path, path_val, starts_with("path")) %>%
  gather(flow, flow_val, starts_with("flow")) %>%
  gather(name, name_val, starts_with("values.pdep")) %>%
  gather(intervention, interv_val, starts_with("values.inter")) %>%
  filter(grepl(".value", name)) %>% 
  filter(grepl("node", path)) %>%
  mutate(variable = gsub("values.", "", name)) %>% 
  mutate(variable = gsub(".value", "", variable)) %>% 
  distinct(contact.name, uuid, name, .keep_all = TRUE) %>% 
  select(id, uuid, contact.uuid, variable, name_val, created_on, modified_on) %>% 
  arrange(id, created_on) # optional wide %>% spread(variable, name_val)