R/getUpdates.R

In jasenfinch/pubchemPIP: Putative Ionisation Product querying of the pubchem database

getUpdates <- function(out_path,url="ftp://ftp.ncbi.nlm.nih.gov/pubchem/Compound/Monthly/"){
	library(RCurl)
	# get month directories
	dirnames <- getURL(url,ftp.use.epsv = FALSE,dirlistonly = TRUE)
	dirnames <- strsplit(dirnames,"\r\n")[[1]]
	for (x in 1:length(dirnames)){
		# Create month folder for output
		dir.create(paste(out_path,dirnames[x],sep="/"))
		# Get availble months
		dir_contents <- getURL(paste(url,dirnames[x],"/",sep=""),ftp.use.epsv = FALSE,dirlistonly = TRUE)
		dir_contents <- strsplit(dir_contents,"\r\n")[[1]]
		# Download updated cids, killed cids
		download.file(paste(url,dirnames[x],"updated-CIDs",sep="/"),paste(out_path,dirnames[x],"updated-CIDs",sep="/"))
		download.file(paste(url,dirnames[x],"killed-CIDs",sep="/"),paste(out_path,dirnames[x],"killed-CIDs",sep="/"))
		# Download SDF files
		dir.create(paste(out_path,dirnames[x],"SDF",sep="/"))
		sdf_contents <- getURL(paste(url,dirnames[x],"SDF/",sep="/"),ftp.use.epsv = FALSE,dirlistonly = TRUE)
		sdf_contents <- strsplit(sdf_contents,"\r\n")[[1]]
		for (i in 1:length(sdf_contents)){
			download.file(paste(url,dirnames[x],"SDF",sdf_contents[i],sep="/"),paste(out_path,dirnames[x],"SDF",sdf_contents[i],sep="/"))
		}
	}
}