fixed negative selection; inc version to 1.2.3

Author: holgerbrandl

build.gradle

@@ -51,7 +51,7 @@ task javadocJar(type: Jar, dependsOn: javadoc) {
 //group = 'de.mpicbg.scicomp'
 group = 'com.github.holgerbrandl'
 //version = '1.1.9'
-version = '1.2.3-SNAPSHOT'
+version = '1.2.3'
 
 
 artifacts {

src/main/kotlin/kscript/KscriptUtil.kt

@@ -7,7 +7,7 @@ import kotlin.system.exitProcess
  */
 
 /**
- * Just used interally to prevent [stopIfNot] to quit the process when running in unit-test mode.
+ * Just used internally to prevent [stopIfNot] to quit the process when running in unit-test mode.
  * It throw an IllegalArgumentException instead.
  */
 internal var isTestMode = false

src/main/kotlin/kscript/text/Tables.kt

@@ -83,18 +83,33 @@ fun List<Row>.print(separator: String = "\t") = asSequence().print(separator)
 /** Internal representations for column selection indices. Usually not use directly but rather via [with] and [without].
  */
 abstract class ColSelect(val indices: Array<Int> = emptyArray()) {
+
+    // irrespective of selection mode (positive or negative) indices must be positive in here
+    init {
+        stopIfNot(indices.all { it > 0 }) {
+            "kscript.text.* is using 1-based arrays to ease awk transition, so indices must be strictly positive"
+        }
+    }
+
     abstract fun and(column: Int): ColSelect
     abstract fun and(range: IntRange): ColSelect
+    abstract fun process(lines: Sequence<Row>): Sequence<Row>
 }
 
-class PosSelect(arrayOf: Array<Int>) : ColSelect(arrayOf) {
+class PosSelect(columnIndices: Array<Int>) : ColSelect(columnIndices) {
     override fun and(column: Int) = PosSelect(arrayOf(*indices, column))
     override fun and(range: IntRange) = PosSelect(arrayOf(*indices, *range.toList().toTypedArray()))
+
+    override fun process(lines: Sequence<Row>): Sequence<Row> = lines.map { row -> Row(indices.map { row[it] }) }
 }
 
-class NegSelect(arrayOf: Array<Int>) : ColSelect(arrayOf) {
+class NegSelect(columnIndices: Array<Int>) : ColSelect(columnIndices) {
     override fun and(column: Int) = NegSelect(arrayOf(*indices, column))
     override fun and(range: IntRange) = NegSelect(arrayOf(*indices, *range.toList().toTypedArray()))
+
+    override fun process(lines: Sequence<Row>): Sequence<Row> = lines.map {
+        Row(it.filterIndexed { index, _ -> !indices.contains(index+1) })
+    }
 }
 
 /** Starts building a column selection index. Both positive and negative indices are supported. */
@@ -116,24 +131,20 @@ fun Sequence<Row>.select(vararg colIndices: Int): Sequence<Row> {
         "Can not mix positive and negative selections"
     }
 
-    val selector = if (isPositive) PosSelect(arrayOf(*colIndices.toTypedArray())) else NegSelect(arrayOf(*colIndices.toTypedArray()))
+    val selector = if (isPositive) {
+        PosSelect(arrayOf(*colIndices.toTypedArray()))
+    } else {
+        NegSelect(arrayOf(*colIndices.map { -it }.toTypedArray()))
+    }
 
     return select(selector)
 }
 
-fun Sequence<Row>.select(columns: ColSelect): Sequence<Row> {
+fun Sequence<Row>.select(columnSelector: ColSelect): Sequence<Row> {
     // more efficient but does not allow to change the order
-    //    return map { it.filterIndexed { index, _ -> retainColumn(columns, index + 1) } }
-
-    stopIfNot(columns.indices.all { it != 0 }) { "kscript.text.* is using 1-based arrays to ease awk transition" }
+    //    return map { it.filterIndexed { index, _ -> retainColumn(columnSelector, index + 1) } }
 
-    return if (columns is PosSelect) {
-        // positive selection
-        map { row -> Row(columns.indices.map { row[it] }) }
-    } else {
-        // negative selection
-        map { Row(it.filterIndexed { index, _ -> !columns.indices.contains(index) }) }
-    }
+    return columnSelector.process(this)
 }
 
 

src/test/kotlin/kscript/test/SupportApiTest.kt

@@ -1,9 +1,11 @@
 package kscript.test
 
 import io.kotlintest.matchers.*
-import io.kotlintest.specs.StringSpec
-import kscript.isTestMode
-import kscript.text.*
+import kscript.text.resolveArgFile
+import kscript.text.select
+import kscript.text.split
+import kscript.text.with
+import org.junit.Test
 
 /**
  * @author Holger Brandl
@@ -16,12 +18,15 @@ fun flightsZipped() = resolveArgFile(arrayOf("src/test/resources/flights.tsv.gz"
 fun flights() = resolveArgFile(arrayOf("src/test/resources/flights.txt"))
 
 
-class SupportApiTest : StringSpec() { init {
+class SupportApiTest {
 
-    isTestMode = true
+    init {
+        kscript.isTestMode = true
+    }
 
 
-    "extract field with column filter" {
+    @Test
+    fun `extract field with column filter`() {
         someFlights().split().
                 filter { it[12] == "N14228" }.
                 map { it[13] }.
@@ -33,23 +38,34 @@ class SupportApiTest : StringSpec() { init {
     }
 
 
-    "allow to select columsn" {
+    @Test
+    fun `allow to select column`() {
         someFlights().split()
                 .select(with(3).and(11..13).and(1))
                 .first().data shouldBe listOf("day", "flight", "tailnum", "origin", "year")
     }
 
 
-    "rejeced mixed select" {
+    @Test
+    fun `is should perform a negative selection`() {
+        someFlights().split()
+                .select(1, 2, 3)
+                .select(-2)
+                .first().data shouldBe listOf("year", "day")
+    }
+
+
+    @Test
+    fun `rejeced mixed select`() {
         shouldThrow<IllegalArgumentException> {
             someFlights().split().select(1, -2)
         }.message shouldBe "[ERROR] Can not mix positive and negative selections"
     }
 
 
-    "compressed lines should be unzipped on the fly"{
+    @Test
+    fun `compressed lines should be unzipped on the fly`() {
         resolveArgFile(arrayOf("src/test/resources/flights.tsv.gz")).
                 drop(1).first() should startWith("2013")
     }
-}
 }