# This program will convert Profusion XML files into a single, combined CSV file that lists staging and scored events
# Prerequisites: XML package
# Upon running for first time, uncomment the "install.packages("XML")" line
#
# How to use: Put XML files into 'xml_path' folder below and run program.

#install.packages("XML")
library(XML)

# Set project path
project_path <- "C:/NSRR/chat"

# Set path for XML files
xml_path <- paste(project_path, "xml", sep="/")

# Set path for CSV exports of XML files
csv_path <- paste(project_path, "csv", sep="/")

# Set path for merged output CSV file
output_path <- paste(project_path, "output", sep="/")

# Grab list of XML files
xmlFiles <- list.files(path = xml_path, pattern = "*.xml")

n <- length(xmlFiles)

for(i in 1:n)  {
  pxml <- xmlParse(paste(xml_path, xmlFiles[i], sep="/"))
  xmldata <- xmlToDataFrame(getNodeSet(pxml,"//SleepStage"))
  write.csv(xmldata, file = file.path(project_path, "sleepstages.csv"), row.names = TRUE)
  xml_in <- read.csv(file.path(project_path, "sleepstages.csv"), header = TRUE, sep = ",", )
  xml_in$epoch <- xml_in$X
  xml_in$SleepStage <- xml_in$text
  xml_in$X <- xml_in$text <- NULL
  xml_in$EpochStartTime <- ((xml_in$epoch * 30) - 30)
  xml_event <- xmlToDataFrame(getNodeSet(pxml, "//ScoredEvent"))
  xml_event$Start <- as.numeric(xml_event$Start)
  xml_event$Duration <- as.numeric(xml_event$Duration)
  xml_event$epoch <- ceiling(xml_event$Start/30)
  xml_merged <- merge(xml_in, xml_event, by = "epoch", all.x = TRUE)
  xml_merged$pptid <- paste(xmlFiles[i])
  xml_merged <- xml_merged[c("pptid", "epoch", "EpochStartTime", "SleepStage", "Name", "Start", "Duration", "Input", "LowestSpO2", "Desaturation")]
  filename <- as.character(paste(substr(xmlFiles[[i]], 1, 20), ".csv", sep=""))
  write.csv(xml_merged, file = file.path(csv_path, filename), row.names = FALSE)
}

# Remove temporary 'sleepstages.csv' file
file.remove(file.path(project_path, "sleepstages.csv"))

# Merge all the CSVs of interest
csvs_to_merge <- list.files(path = csv_path, pattern = "*.csv")
all_events <- do.call("rbind", lapply(paste(csv_path,"/",csvs_to_merge,sep=""), read.csv, header = TRUE))
write.csv(all_events, file = file.path(output_path, "all_events.csv"), row.names = FALSE)