R语言实现：除指定列外批量转数值型并嵌套数据

问题背景 #

需要批量处理结构略有差异的XML文件，每个文件生成一个嵌套DataFrame。原代码固定指定m1/m2/m3列转换数值型，遇到包含m4列的文件会报错，同时需要自动将除identite类变量外的所有数据嵌套。

解决方案代码 #

if(!require(tidyverse)) install.packages("tidyverse")
if(!require(xml2)) install.packages("xml2")
if(!require(lubridate)) install.packages("lubridate")

# 定义单文件处理函数
process_xml_file <- function(file_path) {
  # 读取并解析XML
  doc <- xml2::read_xml(file_path) %>%
    xml2::as_list() %>%
    purrr::pluck("bilans","bilan")
  
  # 提取identite核心信息
  identite <- tibble::tibble(
    siren = doc %>% purrr::pluck("identite", "siren", 1), 
    code_greffe = doc %>% purrr::pluck("identite", "code_greffe", 1),
    date_cloture_ex = doc %>% purrr::pluck("identite", "date_cloture_exercice", 1) %>% lubridate::as_date(),
    date_cloture_ex_n.1 = doc %>% purrr::pluck("identite", "date_cloture_exercice_n-1", 1) %>% lubridate::as_date(),
    duree_exercice_n = doc %>% purrr::pluck("identite", "duree_exercice_n", 1),
    duree_exercice_n.1 = doc %>% purrr::pluck("identite", "duree_exercice_n-1", 1),
    statut_saisie = doc %>% purrr::pluck("identite", "code_motif", 1),
    type_bilan = doc %>% purrr::pluck("identite", "code_type_bilan", 1),
    confidentialite = doc %>% purrr::pluck("identite", "code_confidentialite", 1),
    nom = doc %>% purrr::pluck("identite", "denomination", 1)
  )
  
  # 提取detail模块的属性数据
  data <- doc %>% purrr::pluck("detail") %>% 
    purrr::map_df(~purrr::map_df(.,~attributes(.)))
  
  # 动态转换：除code列外所有列转为数值型
  data_num <- data %>%
    dplyr::mutate(dplyr::across(-code, as.numeric))
  
  # 合并数据并自动嵌套非identite列
  full <- dplyr::bind_cols(identite, data_num)
  full_nested <- full %>%
    tidyr::nest(data = !names(identite))
  
  return(full_nested)
}

# 批量处理示例：替换为你的XML文件目录路径
xml_files <- list.files(path = "./xml_files", pattern = "*.xml", full.names = TRUE)
all_results <- purrr::map(xml_files, process_xml_file)

关键改进点 #

• 动态数值转换：使用dplyr::across(-code, as.numeric)自动识别所有非code的列（包括m1/m2/m3/m4等）并转换为数值型，避免固定列名导致的报错。
• 自动嵌套：通过nest(data = !names(identite))自动将所有不属于identite的列嵌套到data列中，无需手动指定列名，适配不同XML的结构差异。
• 批量处理：封装为函数后，结合purrr::map可以一次性处理目录下所有XML文件。

XML示例 #

<?xml version="1.0" encoding="UTF-8"?>
<bilans xmlns="fr:inpi:odrncs:bilansSaisisXML" version="1.0">
   <bilan>
      <identite>
         <siren>7541834956</siren>
         <date_cloture_exercice>20210101</date_cloture_exercice>
         <code_greffe>7803</code_greffe>
         <num_depot>0945383</num_depot>
         <num_gestion>1945852</num_gestion>
         <code_activite>4332B</code_activite>
         <date_cloture_exercice_n-1>0439543851</date_cloture_exercice_n-1>
         <duree_exercice_n>12</duree_exercice_n>
         <duree_exercice_n-1>12</duree_exercice_n-1>
         <date_depot>20210210</date_depot>
         <code_motif>00</code_motif>
         <code_type_bilan>S</code_type_bilan>
         <code_devise>EUR</code_devise>
         <code_origine_devise>O</code_origine_devise>
         <code_confidentialite>0</code_confidentialite>
         <info_traitement />
         <denomination><![CDATA[hidden]]></denomination>
         <adresse><![CDATA[nowhere]]></adresse>
      </identite>
      <detail>
         <page numero="01">
            <liasse code="014" m1="000000000001671" m2="000000000001671" />
            <liasse code="028" m1="000000000021557" m2="000000000019525" m3="000000000002032" />
            <liasse code="040" m1="000000000001049" m3="000000000001049" />
            <liasse code="044" m1="000000000024277" m2="000000000021196" m3="000000000003081" />
            <liasse code="050" m1="000000000060031" m3="000000000060031" />
            <liasse code="068" m1="000000000040570" m2="000000000002193" m3="000000000038376" />
            <liasse code="072" m1="000000000009806" m3="000000000009806" />
            <liasse code="084" m1="000000000032908" m3="000000000032908" />
            <liasse code="092" m1="000000000004742" m3="000000000004742" />
            <liasse code="096" m1="000000000148057" m2="000000000002193" m3="000000000145863" />
            <liasse code="110" m1="000000000172334" m2="000000000023390" m3="000000000148944" />
            <liasse code="120" m3="000000000007600" />
            <liasse code="126" m3="000000000000760" />
            <liasse code="134" m3="-000000000044214" />
            <liasse code="136" m3="-000000000012606" />
            <liasse code="142" m3="-000000000048461" />
            <liasse code="156" m3="000000000049178" />
            <liasse code="166" m3="000000000071965" />
            <liasse code="169" m2="000000000029740" />
            <liasse code="172" m3="000000000076261" />
            <liasse code="176" m3="000000000197405" />
            <liasse code="180" m3="000000000148944" />
            <liasse code="182" m3="000000000001415" />
         </page>
         <page numero="02">
            <liasse code="214" m1="000000000031915" />
            <liasse code="218" m1="000000000266718" />
            <liasse code="222" m1="000000000000835" />
            <liasse code="230" m1="000000000000178" />
            <liasse code="232" m1="000000000299646" />
            <liasse code="238" m1="000000000083038" />
            <liasse code="240" m1="000000000001500" />
            <liasse code="242" m1="000000000075261" />
            <liasse code="243" m1="-000000000000411" />
            <liasse code="244" m1="000000000003279" />
            <liasse code="250" m1="000000000100069" />
            <liasse code="252" m1="000000000045326" />
            <liasse code="254" m1="000000000001384" />
            <liasse code="256" m1="000000000002000" />
            <liasse code="262" m1="000000000000007" />
            <liasse code="264" m1="000000000311863" />
            <liasse code="270" m1="-000000000012217" />
            <liasse code="290" m1="000000000000564" />
            <liasse code="294" m1="000000000000024" />
            <liasse code="300" m1="000000000000930" />
            <liasse code="310" m1="-000000000012606" />
         </page>
      </detail>
   </bilan>
</bilans>

内容的提问来源于stack exchange，提问作者Greg Mansio