Merge pull request #31 from INCATools/issue-28

Add handling of CSV.

.travis.yml

@@ -1,5 +1,5 @@
 language: scala
 scala:
-  - "2.11.8"
+  - "2.12.3"
 jdk:
   - oraclejdk8

build.sbt

@@ -6,9 +6,9 @@ organization  := "org.monarchinitiative"
 
 name          := "dosdp-tools"
 
-version       := "0.5"
+version       := "0.6-SNAPSHOT"
 
-scalaVersion  := "2.11.8"
+scalaVersion  := "2.12.3"
 
 scalacOptions := Seq("-unchecked", "-deprecation", "-encoding", "utf8")
 

src/main/scala/org/monarchinitiative/dosdp/cli/Common.scala

@@ -17,6 +17,9 @@ import io.circe.syntax._
 import io.circe.yaml.parser
 import org.monarchinitiative.dosdp._
 import com.typesafe.scalalogging.LazyLogging
+import com.github.tototoshi.csv.CSVFormat
+import com.github.tototoshi.csv.DefaultCSVFormat
+import com.github.tototoshi.csv.TSVFormat
 
 trait Common extends Command with LazyLogging {
 
@@ -27,6 +30,7 @@ trait Common extends Command with LazyLogging {
   var prefixesFileOpt = opt[Option[File]](name = "prefixes", default = None, description = "CURIE prefixes (YAML)")
   var oboPrefixes = opt[Boolean](name = "obo-prefixes", default = false, description = "Assume prefixes are OBO ontologies; predefine rdf, rdfs, and owl")
   var outfile = opt[File](name = "outfile", default = new File("dosdp.out"), description = "Output file (OWL or TSV)")
+  var tableFormat = opt[String](name = "table-format", default = "tsv", description = "Tabular format: TSV (default) or CSV")
 
   def ontologyOpt: Option[OWLOntology] = ontOpt.map { ontPath =>
     val ontIRI = if (ontPath.startsWith("http")) IRI.create(ontPath) else IRI.create(new File(ontPath))
@@ -50,4 +54,10 @@ trait Common extends Command with LazyLogging {
     if (oboPrefixes) specifiedPrefixes.orElse(OBOPrefixes) else specifiedPrefixes
   }
 
+  def tabularFormat: CSVFormat = tableFormat.toLowerCase match {
+    case "csv" => new DefaultCSVFormat {}
+    case "tsv" => new TSVFormat {}
+    case other => throw new RuntimeException(s"Invalid tabular format requested: $other")
+  }
+
 }

src/main/scala/org/monarchinitiative/dosdp/cli/Generate.scala

@@ -17,7 +17,6 @@ import org.semanticweb.owlapi.model.OWLOntology
 import org.semanticweb.owlapi.model.parameters.Imports
 
 import com.github.tototoshi.csv.CSVReader
-import com.github.tototoshi.csv.TSVFormat
 
 import cats.implicits._
 
@@ -28,11 +27,12 @@ object Generate extends Command(description = "generate ontology axioms for TSV
   val LocalLabelProperty = IRI.create("http://example.org/TSVProvidedLabel")
 
   def run: Unit = {
+    val sepFormat = tabularFormat
     val dosdp = inputDOSDP
     val eDOSDP = ExpandedDOSDP(dosdp, prefixes)
     val readableIDIndex = ontologyOpt.map(ont => createReadableIdentifierIndex(eDOSDP, ont)).getOrElse(Map.empty)
     val axioms = (for {
-      row <- CSVReader.open(infile, "utf-8")(new TSVFormat {}).iteratorWithHeaders
+      row <- CSVReader.open(infile, "utf-8")(sepFormat).iteratorWithHeaders
     } yield {
       val (varBindingsItems, localLabelItems) = (for {
         vars <- dosdp.vars.toSeq

src/main/scala/org/monarchinitiative/dosdp/cli/Main.scala

@@ -29,6 +29,12 @@ import org.apache.jena.system.JenaSystem
 object Main extends App {
 
   JenaSystem.init()
-  Cli.parse(args).withProgramName("dosdp-tools").withCommands(Generate, Query).foreach(_.run)
+  try {
+    Cli.parse(args).withProgramName("dosdp-tools").withCommands(Generate, Query).foreach(_.run)
+  } catch {
+    case e: Exception =>
+      println(e.getMessage)
+      throw e
+  }
 
 }

src/main/scala/org/monarchinitiative/dosdp/cli/Query.scala

@@ -1,10 +1,5 @@
 package org.monarchinitiative.dosdp.cli
 
-import java.io.FileOutputStream
-import java.io.OutputStreamWriter
-import java.io.PrintWriter
-import java.nio.charset.StandardCharsets
-
 import scala.collection.JavaConverters._
 
 import org.apache.jena.query.QueryExecutionFactory
@@ -18,6 +13,8 @@ import org.semanticweb.elk.owlapi.ElkReasonerFactory
 import org.semanticweb.owlapi.apibinding.OWLManager
 import org.semanticweb.owlapi.model.OWLOntology
 
+import com.github.tototoshi.csv.CSVWriter
+
 import uk.ac.manchester.cs.jfact.JFactFactory
 
 object Query extends Command(description = "query an ontology for terms matching a Dead Simple OWL Design Pattern") with Common {
@@ -26,6 +23,7 @@ object Query extends Command(description = "query an ontology for terms matching
   var printQuery = opt[Boolean](name = "print-query", default = false, description = "Print generated query without running against ontology")
 
   def run: Unit = {
+    val sepFormat = tabularFormat
     val sparqlQuery = SPARQL.queryFor(ExpandedDOSDP(inputDOSDP, prefixes))
     val reasonerFactoryOpt = reasonerNameOpt.map(_.toLowerCase).map { name =>
       name match {
@@ -59,13 +57,13 @@ object Query extends Command(description = "query an ontology for terms matching
       val query = QueryFactory.create(processedQuery)
       val results = QueryExecutionFactory.create(query, model).execSelect()
       val columns = results.getResultVars.asScala.toList
-      val pw = new PrintWriter(new OutputStreamWriter(new FileOutputStream(outfile), StandardCharsets.UTF_8))
-      pw.println(columns.mkString("\t"))
+      val writer = CSVWriter.open(outfile, "utf-8")(sepFormat)
+      writer.writeRow(columns)
       while (results.hasNext()) {
         val qs = results.next()
-        pw.println(columns.map(variable => Option(qs.get(variable)).map(_.toString).getOrElse("")).mkString("\t"))
+        writer.writeRow(columns.map(variable => Option(qs.get(variable)).map(_.toString).getOrElse("")))
       }
-      pw.close()
+      writer.close()
     }
   }