I'm trying to convert a dataframe to xml. It is about 600K records. I'm using the XML package:
library(XML)
con <- xmlOutputDOM("mydata")
for(i in seq(nrow(mydata))){
con$addTag("person", attrs = mydata[i,])
}
The code above is taking too long to run, Is there a way for my to rewrite this code or use a different package to improve the performance?
library('XML')
data
df1 <- data.frame(a = 1:7, b = letters[1:7], stringsAsFactors = FALSE)
code
# create a new xml doc
doc_xml <- newXMLDoc(isHTML = FALSE)
# create a table node
table_node <- newXMLNode("table", doc = doc_xml)
# row data
row_data <- apply(df1, 1, function(x) {
z1 <- newXMLNode('row') # create a new node for each row
addChildren(z1, lapply(names(x), function(y) newXMLNode(y, x[y])))
})
# add row data to table node
xmlParent(row_data) <- table_node
# save as xml file
saveXML(doc_xml, file = "df1.xml")
Output
doc_xml
# <?xml version="1.0"?>
# <table>
# <row>
# <a>1</a>
# <b>a</b>
# </row>
# <row>
# <a>2</a>
# <b>b</b>
# </row>
# <row>
# <a>3</a>
# <b>c</b>
# </row>
# <row>
# <a>4</a>
# <b>d</b>
# </row>
# <row>
# <a>5</a>
# <b>e</b>
# </row>
# <row>
# <a>6</a>
# <b>f</b>
# </row>
# <row>
# <a>7</a>
# <b>g</b>
# </row>
# </table>
verify nodes
getNodeSet(doc_xml, "//a")
getNodeSet(doc_xml, "//b")
Convert xml to dataframe
# using xpath expression of xml data inside R
xmlToDataFrame(nodes = getNodeSet(doc_xml, "//table/*"),
stringsAsFactors = FALSE,
colClasses = c('integer', 'character'))
# using name of xml data inside R
xmlToDataFrame(doc = doc_xml,
stringsAsFactors = FALSE,
colClasses = c('integer', 'character'))
# from xml file
xmlToDataFrame(doc = "df1.xml",
stringsAsFactors = FALSE,
colClasses = c('integer', 'character'))
# a b
# 1 1 a
# 2 2 b
# 3 3 c
# 4 4 d
# 5 5 e
# 6 6 f
# 7 7 g
Instead of using a loop try lapply. Maybe that will solve.
applyis around ~2x faster for me on a small dataset:apply(mydata, 1, function(x) {con$addTag("person", attrs = x); invisible()})