Fixing code generation bug with underscores in names

cli/passengers_.answers

@@ -0,0 +1,17 @@
+'age' - what kind of feature is this? [0] integral [1] categorical: => 0
+'gender' - what kind of feature is this? [0] text [1] categorical: => 1
+'height' - what kind of feature is this? [0] integral [1] categorical: => 0
+'weight' - what kind of feature is this? [0] integral [1] categorical: => 0
+'description' - what kind of feature is this? [0] text [1] categorical: => 0
+'boarded' - what kind of feature is this? [0] integral [1] categorical: => 1
+'recordDate' - what kind of feature is this? [0] integral [1] categorical: => 0
+'Survived' - what kind of feature is this? [0] integral [1] categorical: => 1
+'P_class' - what kind of feature is this? [0] integral [1] categorical: => 0
+'Name' - what kind of feature is this? [0] text [1] categorical: => 0
+'Sex' - what kind of feature is this? [0] text [1] categorical: => 1
+'SibSp' - what kind of feature is this? [0] integral [1] categorical: => 0
+'Parch' - what kind of feature is this? [0] integral [1] categorical: => 0
+'Ticket' - what kind of feature is this? [0] text [1] categorical: => 1
+'Cabin' - what kind of feature is this? [0] text [1] categorical: => 0
+'Embarked' - what kind of feature is this? [0] text [1] categorical: => 0
+Cannot infer the kind of problem based on response field 'Survived'. What kind of problem is this? => binclass

cli/src/main/scala/com/salesforce/op/cli/CliExec.scala

@@ -50,16 +50,21 @@ class CliExec {
   }
 
   def main(args: Array[String]): Unit = try {
-    val outcome = for {
+    val ops = for {
       arguments <- CommandParser.parse(args, CliParameters())
       if arguments.command == "gen"
       settings <- arguments.values
-    } yield Ops(settings).run()
+    } yield Ops(settings)
 
-    outcome getOrElse {
+    ops getOrElse {
       CommandParser.showUsage()
       quit("wrong arguments", 1)
     }
+
+    val outcome = ops.map (_.run())
+
+    outcome getOrElse quit("Generation failed; see error messages", 1)
+
   } catch {
     case x: Exception =>
       if (DEBUG) x.printStackTrace()

cli/src/main/scala/com/salesforce/op/cli/CommandParser.scala

@@ -47,9 +47,12 @@ trait OpCli {
     else success
   }
 
+  private[cli] def tweak(file: File): File =
+    new File(file.getPath.replace("`pwd`/", ""))
+
   def fileExists(file: File): Either[String, Unit] = {
-    if (!file.exists()) failure(s"File '${file.getAbsolutePath}' not found")
-    else success
+    val f = tweak(file)
+    if (f.exists()) success else failure(s"File '${f.getAbsolutePath}' not found")
   }
 
   private val Identifier = "([a-zA-Z]\\w*)".r
@@ -80,7 +83,7 @@ object CommandParser extends scopt.OptionParser[CliParameters]("transmogrifai")
     .text("Input file for the TransmogrifAI project [required]")
     .validate(fileExists)
     .required
-    .action((inputFile, cfg) => cfg.copy(inputFile = Option(inputFile))),
+    .action((inputFile, cfg) => cfg.copy(inputFile = Option(tweak(inputFile)))),
 
   opt[String]("id")
     .text("Name for the ID field [required]")

cli/src/main/scala/com/salesforce/op/cli/gen/ProblemSchema.scala

@@ -33,9 +33,7 @@ package com.salesforce.op.cli.gen
 import java.io.File
 
 import com.salesforce.op.cli.SchemaSource
-import org.apache.avro.Schema
 
-import scala.collection.JavaConverters._
 import scala.io.Source
 import AvroField._
 
@@ -83,7 +81,7 @@ object ProblemSchema {
 
     val orf = MakeRawFeature(ops)
 
-    val (responseFeature :: features) = orderedFields.map { field =>
+    val responseFeature :: features = orderedFields.map { field =>
       orf.from(field, schemaSource.name, field == responseField)
     }
 
@@ -144,10 +142,15 @@ sealed trait OPRawFeature {
 
   /**
    * Gets the java method that Avro generates for this field. e.g. `getPassengerId` for field with name "passengerId"
+   * Note that variable names with underscores are converted to CamelCase
+   * by avro; so we should do the same
    *
    * @return The java getter as a string
    */
-  def avroGetter: String = s"get${avroField.name.capitalize}"
+  def avroGetter: String = {
+    val pieces = avroField.name.split("_").map(_.capitalize)
+    s"get${pieces.mkString("")}"
+  }
 
   /**
    * Gets a name corresponding to the scala `val` that will be generated for this feature builder, e.g.

cli/src/test/scala/com/salesforce/op/cli/CliCodeGenerationTest.scala

@@ -32,10 +32,13 @@ package com.salesforce.op.cli
 
 import language.postfixOps
 import java.io.File
+import java.nio.file.Paths
 
 import org.junit.runner.RunWith
 import org.scalatest.junit.JUnitRunner
 
+import scala.io.Source
+
 /**
  * Test for generator operations
  */
@@ -89,7 +92,7 @@ class CliCodeGenerationTest extends CliTestBase {
     assertResult(result, Succeeded)
   }
 
-  it should "read answers from a file" in {
+  it should "generage code, answers in a file" in {
     val sut = new Sut()
     val result = sut.run(
       "gen",
@@ -106,6 +109,33 @@ class CliCodeGenerationTest extends CliTestBase {
     }
   }
 
+  it should "not fail when fields have underscores in names" in {
+    val sut = new Sut()
+    val result = sut.run(
+      "gen",
+      "--input", TestCsvHeadless,
+      "--id", "passengerId",
+      "--response", "survived",
+      "--schema", TestAvscWithUnderscores,
+      "--answers", AnswersFileWithUnderscores,
+      ProjectName,
+      "--overwrite"
+    )
+    withClue(result.err) {
+      result.outcome shouldBe Succeeded
+    }
+    val scalaSourcesFolder = Paths.get(projectFolder, "src", "main", "scala", "com", "salesforce", "app")
+
+    val featuresFile = Source.fromFile(new File(scalaSourcesFolder.toFile, "Features.scala")).getLines
+    val testLines = featuresFile.dropWhile(!_.contains("val p_class = FB"))
+    testLines.hasNext shouldBe true
+    testLines.next
+    val integralPassenger = testLines.next
+    integralPassenger.trim shouldBe ".Integral[Passenger]"
+    val thisOneShouldNotHaveUnderscoreInGetter = testLines.next
+    thisOneShouldNotHaveUnderscoreInGetter.trim shouldBe ".extract(_.getPClass.toIntegral)"
+  }
+
   it should "work with autogeneration" in {
     val sut = new Sut
     val result = sut.run(

cli/src/test/scala/com/salesforce/op/cli/CliTestBase.scala

@@ -37,7 +37,7 @@ import com.salesforce.op.OpWorkflowRunType
 import com.salesforce.op.test.TestCommon
 import org.scalactic.source
 import org.scalatest.{Assertion, Assertions, BeforeAndAfter, FlatSpec}
-import org.slf4j.LoggerFactory
+import org.slf4j.{Logger, LoggerFactory}
 
 import scala.io.Source
 import scala.language.postfixOps
@@ -49,7 +49,8 @@ class CliTestBase extends FlatSpec with TestCommon with Assertions with BeforeAn
   CommandParser.AUTO_ENABLED = true
 
   protected val ProjectName = "CliGeneratedTestProject"
-  val log = LoggerFactory.getLogger("cli-test")
+  protected val projectFolder: String = ProjectName.toLowerCase
+  val log: Logger = LoggerFactory.getLogger("cli-test")
 
   trait Outcome
   case class Crashed(msg: String, code: Int) extends Throwable with Outcome {
@@ -107,9 +108,9 @@ class CliTestBase extends FlatSpec with TestCommon with Assertions with BeforeAn
 
   after { new Sut().delete(new File(ProjectName)) }
 
-  val expectedSourceFiles = "Features.scala" :: s"$ProjectName.scala"::Nil
+  val expectedSourceFiles: List[String] = "Features.scala" :: s"$ProjectName.scala"::Nil
 
-  val projectDir = ProjectName.toLowerCase
+  val projectDir: String = ProjectName.toLowerCase
 
   def checkAvroFile(source: File): Unit = {
     val avroFile = Paths.get(projectDir, "src", "main", "avro", source.getName).toFile
@@ -144,13 +145,15 @@ class CliTestBase extends FlatSpec with TestCommon with Assertions with BeforeAn
   }
 
   protected lazy val TestAvsc: String = findFile("test-data/PassengerDataAll.avsc")
+  protected lazy val TestAvscWithUnderscores: String = findFile("test-data/PassengerDataAll_.avsc")
   protected lazy val TestCsvHeadless: String = findFile("test-data/PassengerDataAll.csv")
   protected lazy val TestSmallCsvWithHeaders: String = findFile("test-data/PassengerDataWithHeader.csv")
   protected lazy val TestBigCsvWithHeaders: String = findFile("test-data/PassengerDataAllWithHeader.csv")
   protected lazy val AvcsSchema: String = findFile("templates/simple/src/main/avro/Passenger.avsc")
   protected lazy val AnswersFile: String = findFile("cli/passengers.answers")
+  protected lazy val AnswersFileWithUnderscores: String = findFile("cli/passengers_.answers")
 
-  protected def appRuntimeArgs(runType: OpWorkflowRunType) =
+  protected def appRuntimeArgs(runType: OpWorkflowRunType): String =
     s"--run-type=${runType.toString.toLowerCase} --model-location=/tmp/titanic-model " +
     s"--read-location Passenger=$TestCsvHeadless"
 }

gradlew

@@ -28,7 +28,8 @@ APP_NAME="Gradle"
 APP_BASE_NAME=`basename "$0"`
 
 # Add default JVM options here. You can also use JAVA_OPTS and GRADLE_OPTS to pass JVM options to this script.
-DEFAULT_JVM_OPTS=""
+# the one below is good for using with newer Java versions that don't have the options gradle tries to pass.
+DEFAULT_JVM_OPTS="-XX:+IgnoreUnrecognizedVMOptions"
 
 # Use the maximum available, or set MAX_FD != -1 to use that value.
 MAX_FD="maximum"

test-data/PassengerDataAll_.avsc

@@ -0,0 +1,43 @@
+{
+  "type" : "record",
+  "name" : "Passenger",
+  "namespace" : "com.salesforce.app.schema",
+  "fields" : [ {
+    "name" : "PassengerId",
+    "type" : [ "int", "null" ]
+  }, {
+    "name" : "Survived",
+    "type" : "int",
+    "default": 0
+  }, {
+    "name" : "P_class",
+    "type" : [ "int", "null" ]
+  }, {
+    "name" : "Name",
+    "type" : [ "string", "null" ]
+  }, {
+    "name" : "Sex",
+    "type" : [ "string", "null" ]
+  }, {
+    "name" : "Age",
+    "type" : [ "double", "null" ]
+  }, {
+    "name" : "SibSp",
+    "type" : [ "int", "null" ]
+  }, {
+    "name" : "Parch",
+    "type" : [ "int", "null" ]
+  }, {
+    "name" : "Ticket",
+    "type" : [ "string", "null" ]
+  }, {
+    "name" : "Fare",
+    "type" : [ "double", "null" ]
+  }, {
+    "name" : "Cabin",
+    "type" : [ "string", "null" ]
+  }, {
+    "name" : "Embarked",
+    "type" : [ "string", "null" ]
+  } ]
+}