diff MirbaseEntry.R @ 2:20d69a062da3 draft

planemo upload for repository https://github.com/workflow4metabolomics/lcmsmatching.git commit d4048accde6bdfd5b3e14f5394902d38991854f8

--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/MirbaseEntry.R	Thu Mar 02 08:55:00 2017 -0500
@@ -0,0 +1,47 @@
+#####################
+# CLASS DECLARATION #
+#####################
+
+MirbaseEntry <- methods::setRefClass("MirbaseEntry", contains = "BiodbEntry")
+
+###########
+# FACTORY #
+###########
+
+createMirbaseEntryFromHtml <- function(contents, drop = TRUE) {
+
+	entries <- list()
+
+	# Define fields regex
+	xpath.expr <- character()
+	xpath.expr[[BIODB.ACCESSION]]  <- "//td[text()='Accession number']/../td[2]"
+	xpath.expr[[BIODB.NAME]]       <- "//td[text()='ID']/../td[2]"
+	xpath.expr[[BIODB.SEQUENCE]]   <- "//td[text()='Sequence']/..//pre"
+
+	for (html in contents) {
+
+		# Create instance
+		entry <- MirbaseEntry$new()
+	
+		# Parse HTML
+		xml <-  XML::htmlTreeParse(html, asText = TRUE, useInternalNodes = TRUE)
+
+		# Test generic xpath expressions
+		for (field in names(xpath.expr)) {
+			v <- XML::xpathSApply(xml, xpath.expr[[field]], XML::xmlValue)
+			if (length(v) > 0)
+				entry$setField(field, v)
+		}
+
+		entries <- c(entries, entry)
+	}
+
+	# Replace elements with no accession id by NULL
+	entries <- lapply(entries, function(x) if (is.na(x$getField(BIODB.ACCESSION))) NULL else x)
+
+	# If the input was a single element, then output a single object
+	if (drop && length(contents) == 1)
+		entries <- entries[[1]]
+
+	return(entries)
+}