来自多个 XML 个文件的数据框

Question

我有一个包含多个 XML 文件的文件夹。所有文件都具有相同的基本结构。然而，每个文件实际上包含与分布在 16 个父节点中的单个实体相关的数据。有的节点有子节点，有的甚至有g运行dchildren，有的g运行dchildren.

我想从多个文件创建一个数据框，只选择 nodes/children/grandchildren。

作为第一步，我读取了一个 XML 文件作为列表。然后运行几行代码将需要的数据放到一个vector中。最终，将矢量转换为数据帧，就像我想要的那样。

此代码：

library(xml2)
library(tidyverse)

x = as_list(read_xml("ACTRN12605000003673.xml"))

tmp = c(ACTR_Number = as.character(x$ANZCTR_Trial$actrnumber),
         primary_sponsor_type = as.character(x$ANZCTR_Trial$sponsorship$primarysponsortype),
         primary_sponsor_name = as.character(x$ANZCTR_Trial$sponsorship$primarysponsorname),
         primary_sponsor_address = as.character(x$ANZCTR_Trial$sponsorship$primarysponsoraddress),
         primary_sponsor_country = as.character(x$ANZCTR_Trial$sponsorship$primarysponsorcountry),
         funding_source_type = as.character(x$ANZCTR_Trial$sponsorship$fundingsource$fundingtype),
         funding_source_name = as.character(x$ANZCTR_Trial$sponsorship$fundingsource$fundingname),
         funding_source_address = as.character(x$ANZCTR_Trial$sponsorship$fundingsource$fundingaddress),
         funding_source_country = as.character(x$ANZCTR_Trial$sponsorship$fundingsource$fundingcountry),
         secondary_sponsor_type = as.character(x$ANZCTR_Trial$sponsorship$secondarysponsor$sponsortype),
         secondary_sponsor_name = as.character(x$ANZCTR_Trial$sponsorship$secondarysponsor$sponsorname),
         secondary_sponsor_address = as.character(x$ANZCTR_Trial$sponsorship$secondarysponsor$sponsoraddress),
         secondary_sponsor_country = as.character(x$ANZCTR_Trial$sponsorship$secondarysponsor$sponsorcountry))
tmp = as.list(tmp)
tmp = as.data.frame(tmp)

对于下一步，我尝试同时处理 2 个 XML 文件。我尝试了以下代码同时读取两个文件。但是，除此之外，我不知道如何继续。

all_files = list.files(pattern=".xml", path = getwd(), full.names = TRUE)
x = lapply(all_files, read_xml)
class(x)

样本文件here

Answer 1

R 支持列表中的列表。当您定义一个列表时，它将包含您所有的 tmp 列表。然后将所有文件放入此列表后，您可以制作数据框或小标题。然后你的 tibble 包含带有列表（tmp）的列。然后您可以使用库中的地图函数解决它们（purrr）

Answer 2

我们可以将您的数据收集过程封装成一个函数。由于每个 XML 文件似乎代表所需输出数据框中的一行，因此我们可以使用 purrr::map_dfr 逐行构建数据。我稍微修改了你的代码。见下文：

get_row_data <- function(file_name) {
  xml <- xml2::read_xml(file_name)
  read_text <- \(x, xpath) rvest::html_elements(x, xpath = xpath) |> rvest::html_text(TRUE) 
  xpaths <- c(
    ACTR_Number = "/ANZCTR_Trial/actrnumber", 
    primary_sponsor_type = "/ANZCTR_Trial/sponsorship/primarysponsortype", 
    primary_sponsor_name = "/ANZCTR_Trial/sponsorship/primarysponsorname", 
    primary_sponsor_address = "/ANZCTR_Trial/sponsorship/primarysponsoraddress", 
    primary_sponsor_country = "/ANZCTR_Trial/sponsorship/primarysponsorcountry",
    funding_source_type = "/ANZCTR_Trial/sponsorship/fundingsource/fundingtype",
    funding_source_name = "/ANZCTR_Trial/sponsorship/fundingsource/fundingname",
    funding_source_address = "/ANZCTR_Trial/sponsorship/fundingsource/fundingaddress",
    funding_source_country = "/ANZCTR_Trial/sponsorship/fundingsource/fundingcountry",
    secondary_sponsor_type = "/ANZCTR_Trial/sponsorship/secondarysponsor/sponsortype",
    secondary_sponsor_name = "/ANZCTR_Trial/sponsorship/secondarysponsor/sponsorname",
    secondary_sponsor_address = "/ANZCTR_Trial/sponsorship/secondarysponsor/sponsoraddress",
    secondary_sponsor_country = "/ANZCTR_Trial/sponsorship/secondarysponsor/sponsorcountry"
  )
  x <- read_text(xml, paste0(xpaths, collapse = " | "))
  names(x) <- names(xpaths)
  as_tibble(as.list(x))
}

all_files <- list.files(pattern = ".xml", path = getwd(), full.names = TRUE)
purrr::map_dfr(all_files, get_row_data)

输出

# A tibble: 2 x 13
  ACTR_Number         primary_sponsor_~ primary_sponsor_n~ primary_sponsor_~ primary_sponsor~ funding_source_~ funding_source_~
  <chr>               <chr>             <chr>              <chr>             <chr>            <chr>            <chr>           
1 ACTRN12605000003673 Hospital          Barwon Health      "272-322 Ryrie S~ Australia        Commercial sect~ Astra Zeneca    
2 ACTRN12605000025639 Other Collaborat~ Australian Gastro~ "88 Mallett St\n~ Australia        Commercial sect~ Roche Products ~
# ... with 6 more variables: funding_source_address <chr>, funding_source_country <chr>, secondary_sponsor_type <chr>,
#   secondary_sponsor_name <chr>, secondary_sponsor_address <chr>, secondary_sponsor_country <chr>

来自多个 XML 个文件的数据框

Data frame from multiple XML files

r

xml

dataframe

xml2