Faster way to put large nested xml into R dataframe

I have a XML file that has nesting that is different from other examples I have seen. Here's the format it takes. I am interested in the data in the np nodes, but I also want the group and ID info in each row.

I have a working solution below, but the actual file is very large with thousands of <folder> nodes. The solution takes hours and hours to run on the large file, although it works on the sample code below.

My question is - is there a faster way to get the dataframe I want?

<File>
  <Time>
    <date>20220301</date>
    <Name>1</Name>
<folder>
      <group>800</group>
      <ID>ESK</ID>
      <Type>S</Type>
      <Customer>1</Customer>
      <currency>USD</currency>
      <Port>
        <ec>X</ec>
        <np>
          <A>FIRST</A>
          <B>ES</B>
          <C>GOR</C>
          <D>2021</D>
          <E>-1000</E>
        </np>
        <np>
          <A>TEST</A>
          <B>ES</B>
          <C>RUN</C>
          <D>202303</D>
          <E>202303</E>
          <F>C</F>
          <G>3200</G>
          <H>32</H>
        </np>
      </Port>
</folder>
<folder>
      <group>900</group>
      <ID>ABC</ID>
      <Type>D</Type>
      <Customer>1</Customer>
      <currency>USD</currency>
      <Port>
        <ec>X</ec>
        <np>
          <A>CAT</A>
          <B>ES</B>
          <C>GO</C>
          <D>202303</D>
          <E>-500</E>
        </np>
       </Port>
</folder>
</Time>
</File>

Here is my current working solution. This works on small xml files, but for large files with that xml format it is too slow and takes hours to run.

URL <- 'H:/testSO.xml'
doc <- read_xml(URL)

df <-
  xml_find_all(doc, ".//np") %>%
  map_df( function(x) {
    set_names( c(  
      xml_find_all( x, "./ancestor::folder/group") %>% xml_text(),
      xml_find_all( x, "./ancestor::folder/ID") %>% xml_text(),
      xml_find_all( x, ".//A") %>% xml_text(),
      xml_find_all( x, ".//B") %>% xml_text(),
      xml_find_all( x, ".//C") %>% xml_text(),
      xml_find_all( x, ".//D") %>% xml_text(),
      xml_find_all( x, ".//E") %>% xml_text()), 
      #set the column names
      c( "group","id", "A", "B", "C","D","E") ) %>% 
      as.list() %>% #make list
      flatten_df() 
  }) %>%
  type_convert() 

head(df)

Thanks!

You can take advantage of the vectorization of the xml2 library and avoid the map_df loop, this should provide a significant improvement in speed.

This routine finds all of the np nodes, extracts the requested information into a series of vectors, and then creates a data frame with the results.

library(xml2)
library(dplyr)
npnodes <- xml_find_all(doc, ".//np") 
   
group <- npnodes %>% xml_find_first( "./ancestor::folder/group") %>% xml_text()
id <- npnodes %>% xml_find_first("./ancestor::folder/ID") %>% xml_text()
A <- npnodes %>% xml_find_first(".//A") %>% xml_text()
B <- npnodes %>% xml_find_first(".//B") %>% xml_text()
C <- npnodes %>% xml_find_first(".//C") %>% xml_text()
D <- npnodes %>% xml_find_first(".//D") %>% xml_text()
E <- npnodes %>% xml_find_first(".//E") %>% xml_text() 

answer <- data.frame(group,id, A, B, C, D, E)
answer

  group  id     A  B   C      D      E
1   800 ESK FIRST ES GOR   2021  -1000
2   800 ESK  TEST ES RUN 202303 202303
3   900 ABC   CAT ES  GO 202303   -500