Port to Scala 3

broadinstitute · Jan 10, 2024 · 63ae339 · 63ae339
1 parent 4723212
commit 63ae339
Show file tree

Hide file tree

Showing 78 changed files with 256 additions and 376 deletions.
diff --git a/.scalafix.conf b/.scalafix.conf
@@ -7,5 +7,4 @@ OrganizeImports {
   expandRelative = true
   groups = ["re:javax?\\.", "scala.", "*"]
   groupedImports = Merge
-  removeUnused = true
 }
diff --git a/.scalafmt.conf b/.scalafmt.conf
@@ -14,4 +14,4 @@ rewrite.redundantBraces.generalExpressions = false
 rewrite.redundantBraces.includeUnitMethods = false
 rewrite.redundantBraces.maxBreaks = 16
 rewrite.redundantBraces.stringInterpolation = true
-runner.dialect = scala213source3
+runner.dialect = scala3
diff --git a/build.sbt b/build.sbt
@@ -2,17 +2,15 @@ val artifactId = "poolq"
 
 inThisBuild(
   List(
-    scalaVersion := "2.13.11",
+    scalaVersion := "3.3.1",
     semanticdbEnabled := true,
     semanticdbVersion := scalafixSemanticdb.revision,
     versionScheme := Some("early-semver")
   )
 )
 
 lazy val versions = new {
-  val acyclic = "0.2.1"
   val betterFiles = "3.9.2"
-  val betterMonadicFor = "0.3.1"
   val catsEffect3 = "3.5.2"
   val cats = "2.10.0"
   val commonsIo = "2.15.1"
@@ -28,15 +26,13 @@ lazy val versions = new {
   val scalaCheck = "1.17.0"
   val scalaCsv = "1.3.10"
   val scalaTest = "3.2.17"
-  val scalaTestPlusScalaCheck = "3.2.2.0"
+  val scalaTestPlusScalaCheck = "3.2.17.0"
   val scopt = "4.1.0"
   val slf4j = "1.7.36"
 }
 
 lazy val libraries = new {
-  val acyclic = "com.lihaoyi" %% "acyclic" % versions.acyclic
   val betterFiles = "com.github.pathikrit" %% "better-files" % versions.betterFiles
-  val betterMonadicFor = "com.olegpy" %% "better-monadic-for" % versions.betterMonadicFor
   val cats = "org.typelevel" %% "cats-core" % versions.cats
   val catsEffect3 = "org.typelevel" %% "cats-effect" % versions.catsEffect3
   val commonsIo = "commons-io" % "commons-io" % versions.commonsIo
@@ -59,12 +55,11 @@ lazy val libraries = new {
   val munitCatsEffect3 = "org.typelevel" %% "munit-cats-effect-3" % versions.munitCatsEffect3
   val scalaTest = "org.scalatest" %% "scalatest" % versions.scalaTest
   val scalaCheck = "org.scalacheck" %% "scalacheck" % versions.scalaCheck
-  val scalaTestPlusScalaCheck = "org.scalatestplus" %% "scalacheck-1-14" % versions.scalaTestPlusScalaCheck
+  val scalaTestPlusScalaCheck = "org.scalatestplus" %% "scalacheck-1-17" % versions.scalaTestPlusScalaCheck
 }
 
 lazy val dependencies =
   List(
-    libraries.acyclic % "provided",
     libraries.cats,
     libraries.commonsIo,
     libraries.commonsMath3,
@@ -130,13 +125,9 @@ lazy val poolq = project
     name := "poolq3",
     organization := "org.broadinstitute.gpp",
     libraryDependencies := dependencies,
-    scalacOptions ++= List("-P:acyclic:force", "-Xsource:3"),
     buildInfoKeys := Seq[BuildInfoKey](name, version),
     buildInfoPackage := "org.broadinstitute.gpp.poolq3",
-    addCompilerPlugin(libraries.acyclic),
-    addCompilerPlugin(libraries.betterMonadicFor),
     testFrameworks += new TestFramework("munit.Framework"),
-    scalacOptions += "-Yrangepos", // ensure munit clues work
     // Tests pass in parallel, but SLF4J logging behaves weirdly. Disable this flag to examine test
     // log output; leave this enabled for very fast test execution.
     Test / parallelExecution := true

diff --git a/src/main/scala/org/broadinstitute/gpp/poolq3/PoolQ.scala b/src/main/scala/org/broadinstitute/gpp/poolq3/PoolQ.scala
@@ -9,7 +9,7 @@ import java.nio.file.{Files, Path}
 
 import scala.util.{Failure, Success, Try, Using}
 
-import cats.syntax.all._
+import cats.syntax.all.*
 import org.broadinstitute.gpp.poolq3.PoolQConfig.synthesizeArgs
 import org.broadinstitute.gpp.poolq3.barcode.{BarcodePolicy, Barcodes, barcodeSource}
 import org.broadinstitute.gpp.poolq3.parser.{BarcodeSet, CloseableIterable, ReferenceData}
@@ -122,7 +122,7 @@ object PoolQ {
       barcodeSource(config.input, rowBarcodePolicy, revRowBarcodePolicyOpt, colBarcodePolicyOrLength, umiInfo.map(_._2))
 
     lazy val unexpectedSequenceCacheDir: Option[Path] =
-      if (config.skipUnexpectedSequenceReport) None
+      if config.skipUnexpectedSequenceReport then None
       else {
         val ret = config.unexpectedSequenceCacheDir.map(Files.createDirectories(_)).orElse {
           val ret: Path = Files.createTempDirectory("unexpected-sequence-cache")
@@ -133,7 +133,7 @@ object PoolQ {
       }
 
     val consumer =
-      if (config.noopConsumer) new NoOpConsumer
+      if config.noopConsumer then new NoOpConsumer
       else
         new ScoringConsumer(
           rowReference,
@@ -145,7 +145,7 @@ object PoolQ {
           config.isPairedEnd
         )
 
-    for {
+    for
       runSummary <- runProcess(barcodes, consumer)
       state = runSummary.state
       counts = state.known
@@ -195,7 +195,7 @@ object PoolQ {
               globalReference
             )
             .as(UnexpectedSequencesFileType.some)
-        if (config.removeUnexpectedSequenceCache) {
+        if config.removeUnexpectedSequenceCache then {
           log.info(s"Removing unexpected sequence cache ${config.unexpectedSequenceCacheDir}")
           UnexpectedSequenceWriter.removeCache(dir)
         }
@@ -204,7 +204,7 @@ object PoolQ {
       _ = log.info(s"Writing run info ${config.output.unexpectedSequencesFile}")
       _ <- RunInfoWriter.write(config.output.runInfoFile, config)
       _ = log.info("PoolQ complete")
-    } yield PoolQSummary(runSummary, AlwaysWrittenFiles ++ Set(cfto, usfto).flatten)
+    yield PoolQSummary(runSummary, AlwaysWrittenFiles ++ Set(cfto, usfto).flatten)
   }
 
   def runProcess(barcodes: CloseableIterable[Barcodes], consumer: Consumer): Try[PoolQRunSummary] =

diff --git a/src/main/scala/org/broadinstitute/gpp/poolq3/PoolQConfig.scala b/src/main/scala/org/broadinstitute/gpp/poolq3/PoolQConfig.scala
@@ -11,7 +11,7 @@ import java.nio.file.{Files, Path, Paths}
 import scala.collection.mutable
 
 import cats.data.{NonEmptyList => Nel}
-import cats.syntax.all._
+import cats.syntax.all.*
 import org.broadinstitute.gpp.poolq3.PoolQConfig.DefaultPath
 import org.broadinstitute.gpp.poolq3.reports.{GctDialect, PoolQ2Dialect, PoolQ3Dialect, ReportsDialect}
 import org.broadinstitute.gpp.poolq3.types.{PoolQException, ReadIdCheckPolicy}
@@ -42,7 +42,7 @@ final case class PoolQInput(
 
     case (Some(rr), None, Some(cr), None, false) =>
       val rs = ReadsSource.Split(Nel(cr, addlColReads), Nel(rr._2, addlRowReads.view.map(_._2).toList))
-      if (rs.forward.length == rs.index.length) Right(rs)
+      if rs.forward.length == rs.index.length then Right(rs)
       else Left(PoolQException("Number of row, column, and reverse reads files must match"))
 
     case (Some(rr), Some(rrr), Some(cr), None, false) =>
@@ -51,15 +51,15 @@ final case class PoolQInput(
         Nel(rr._2, addlRowReads.view.map(_._2).toList),
         Nel(rrr._2, addlReverseRowReads.view.map(_._2).toList)
       )
-      if (rs.forward.length == rs.index.length && rs.forward.length == rs.reverse.length) Right(rs)
+      if rs.forward.length == rs.index.length && rs.forward.length == rs.reverse.length then Right(rs)
       else Left(PoolQException("Number of row and column reads files must match"))
 
     case (Some(rr), None, None, None, true) =>
       Right(ReadsSource.Dmuxed(Nel(rr, addlRowReads)))
 
     case (Some(rr), Some(rrr), None, None, true) =>
       val rs = ReadsSource.DmuxedPairedEnd(Nel(rr, addlRowReads), Nel(rrr, addlReverseRowReads))
-      if (rs.read1.map(_._1) == rs.read2.map(_._1)) Right(rs)
+      if rs.read1.map(_._1) == rs.read2.map(_._1) then Right(rs)
       else Left(PoolQException("Row and column reads files must match"))
 
     case _ => Left(PoolQException("Conflicting input options"))
@@ -147,8 +147,8 @@ object PoolQConfig {
 
     val parser: OptionParser[PoolQConfig] = new OptionParser[PoolQConfig]("poolq") {
       private[this] def existsAndIsReadable(f: Path): Either[String, Unit] =
-        if (!Files.exists(f)) failure(s"Could not find ${f.toAbsolutePath}")
-        else if (!Files.isReadable(f)) failure(s"Could not read ${f.toAbsolutePath}")
+        if !Files.exists(f) then failure(s"Could not find ${f.toAbsolutePath}")
+        else if !Files.isReadable(f) then failure(s"Could not read ${f.toAbsolutePath}")
         else success
 
       locally {
@@ -283,7 +283,7 @@ object PoolQConfig {
           .valueName("<number>")
           .action((n, c) => c.copy(unexpectedSequencesToReport = n))
           .validate { n =>
-            if (n > 0) success
+            if n > 0 then success
             else failure(s"Unexpected sequence threshold must be greater than 0, got: $n")
           }
 
@@ -388,7 +388,7 @@ object PoolQConfig {
     // run control
     args += (("row-matcher", config.rowMatchFn))
     args += (("col-matcher", config.colMatchFn))
-    if (config.countAmbiguous) {
+    if config.countAmbiguous then {
       args += (("count-ambiguous", ""))
     }
     args += (("row-barcode-policy", config.rowBarcodePolicyStr))
@@ -397,18 +397,18 @@ object PoolQConfig {
     umiInfo.map(_._2).foreach(str => args += (("umi-barcode-policy", str)))
 
     // deal with the unexpected sequence options
-    if (config.skipUnexpectedSequenceReport) {
+    if config.skipUnexpectedSequenceReport then {
       args += (("skip-unexpected-sequence-report", ""))
     } else {
       // give whatever path we were given here - this _may_ not need to be included at all, honestly
       config.unexpectedSequenceCacheDir.foreach(file => args += (("unexpected-sequence-cache", file.toString)))
       args += (("unexpected-sequence-threshold", config.unexpectedSequencesToReport.toString))
     }
 
-    if (config.skipShortReads) {
+    if config.skipShortReads then {
       args += (("skip-short-reads", ""))
     }
-    if (config.reportsDialect == PoolQ2Dialect) {
+    if config.reportsDialect == PoolQ2Dialect then {
       args += (("compat", ""))
     }
 
@@ -424,10 +424,10 @@ object PoolQConfig {
       output.umiCountsFilesDir.foreach(d => args += (("umi-counts-dir", d.toString)))
       output.umiBarcodeCountsFilesDir.foreach(d => args += (("umi-barcode-counts-dir", d.toString)))
     }
-    if (!config.skipUnexpectedSequenceReport) {
+    if !config.skipUnexpectedSequenceReport then {
       args += (("unexpected-sequences", output.unexpectedSequencesFile.getFileName.toString))
     }
-    if (config.alwaysCountColumnBarcodes) {
+    if config.alwaysCountColumnBarcodes then {
       args += (("always-count-col-barcodes", ""))
     }
     args += (("run-info", output.runInfoFile.getFileName.toString))

diff --git a/src/main/scala/org/broadinstitute/gpp/poolq3/barcode/BarcodePolicy.scala b/src/main/scala/org/broadinstitute/gpp/poolq3/barcode/BarcodePolicy.scala
@@ -40,8 +40,8 @@ final case class FixedOffsetPolicy(startPos0: Int, length: Int, skipShortReads:
   private[this] val endPos0: Int = startPos0 + length
 
   override def find(read: Read): Option[FoundBarcode] =
-    if (read.seq.length < minLength) {
-      if (skipShortReads) None
+    if read.seq.length < minLength then {
+      if skipShortReads then None
       else throw ReadTooShortException(read.seq, startPos0, minLength)
     } else Some(FoundBarcode(copyOfRange(read.seq.toCharArray, startPos0, endPos0), startPos0))
 
@@ -83,9 +83,9 @@ final case class IndexOfKnownPrefixPolicy(
   override def find(read: Read): Option[FoundBarcode] = {
     val index = read.seq.indexOf(prefix, minPrefixStartPosInt)
     val barcodeStart = index + prefixLength
-    if (index < 0) None
-    else if (index > maxPrefixStartPosInt) None
-    else if ((read.seq.length - barcodeStart) < length) None
+    if index < 0 then None
+    else if index > maxPrefixStartPosInt then None
+    else if (read.seq.length - barcodeStart) < length then None
     else Some(FoundBarcode(copyOfRange(read.seq.toCharArray, barcodeStart, barcodeStart + length), barcodeStart))
   }
 
@@ -141,15 +141,15 @@ object TemplatePolicy {
       case Regex1(ctx, minStr, maxStr) =>
         ctx match {
           case Regex2(p1, b1, gap, p2, b2) =>
-            if ((b1.length + b2.length) != refBarcodeLength) {
+            if (b1.length + b2.length) != refBarcodeLength then {
               throw new IllegalArgumentException(s"$s is not compatible with the provided reference file")
             }
             val min = Option(minStr).map(_.toInt)
             val max = Option(maxStr).map(_.tail.toInt)
             SplitBarcodePolicy(p1.toUpperCase, b1.length, gap.length, p2.toUpperCase, b2.length, min, max)
           case _ =>
             val km = KeyMask(ctx)
-            if (km.keyLengthInBases != refBarcodeLength) {
+            if km.keyLengthInBases != refBarcodeLength then {
               throw new IllegalArgumentException(s"$s is not compatible with the provided reference file")
             }
             val min = Option(minStr).map(_.toInt)
@@ -200,8 +200,8 @@ object TemplatePolicy {
 
   final def satisfies(template: Array[Char], seq: String, seqOffset: Int): Boolean = {
     var i = 0
-    while (i < template.length) {
-      if (!compatible(template(i), seq(seqOffset + i))) return false
+    while i < template.length do {
+      if !compatible(template(i), seq(seqOffset + i)) then return false
       i += 1
     }
     true
@@ -230,8 +230,8 @@ final case class GeneralTemplatePolicy(template: KeyMask, minStartPos: Option[In
 
     @tailrec
     def find(i: Int): Option[Int] =
-      if (i > maxPos) None
-      else if (TemplatePolicy.satisfies(templateChars, read.seq, i)) Some(i)
+      if i > maxPos then None
+      else if TemplatePolicy.satisfies(templateChars, read.seq, i) then Some(i)
       else find(i + 1)
 
     find(minStartPosInt).map(i => extract(read, i))
@@ -273,13 +273,13 @@ final case class SplitBarcodePolicy(
     @tailrec
     def loop(start: Int): Option[FoundBarcode] = {
       val e = math.min(maxPrefix1StartPosInt, read.seq.length - patternLength)
-      if (start > e) None
+      if start > e then None
       else {
         indexOf(prefix1, read.seq, start, e) match {
           case None => None
           case Some(p1Index) =>
             val p2Index = p1Index + expectedP2Offset
-            if (matches(prefix2, read.seq, p2Index)) {
+            if matches(prefix2, read.seq, p2Index) then {
               val dest = Array.ofDim[Char](length)
               // copy in the the barcodes
               read.seq.getChars(p1Index + p1Length, p1Index + p1Length + b1Length, dest, 0)
@@ -306,10 +306,10 @@ object SplitBarcodePolicy {
   final private[barcode] def indexOf(needle: String, haystack: String, start: Int, end: Int): Option[Int] = {
     @tailrec
     def loop(i: Int): Option[Int] =
-      if (i > math.min(haystack.length - needle.length, end)) {
+      if i > math.min(haystack.length - needle.length, end) then {
         None
       } else {
-        if (matches(needle, haystack, i)) Some(i)
+        if matches(needle, haystack, i) then Some(i)
         else loop(i + 1)
       }
     loop(start)
@@ -318,9 +318,9 @@ object SplitBarcodePolicy {
   final private def matches(needle: String, haystack: String, haystackOffset: Int): Boolean = {
     @tailrec
     def loop(i: Int): Boolean =
-      if (i >= needle.length) true
+      if i >= needle.length then true
       else {
-        if (needle(i) != haystack(haystackOffset + i)) false
+        if needle(i) != haystack(haystackOffset + i) then false
         else loop(i + 1)
       }
     loop(0)

diff --git a/src/main/scala/org/broadinstitute/gpp/poolq3/barcode/KeyMask.scala b/src/main/scala/org/broadinstitute/gpp/poolq3/barcode/KeyMask.scala
@@ -40,8 +40,7 @@ object KeyMask {
     mergedRanges.length match {
       case 1 =>
         val r = mergedRanges.head
-        if (r.start0 == 0 && r.length == pattern.length)
-          KeyMask0(pattern)
+        if r.start0 == 0 && r.length == pattern.length then KeyMask0(pattern)
         else KeyMask1(pattern, r)
       case 2 =>
         KeyMask2(pattern, mergedRanges(0), mergedRanges(1))
@@ -64,8 +63,7 @@ object KeyMask {
       acc match {
         case Nil => current :: Nil
         case head :: tail =>
-          if (head.end0 >= current.start0 - 1)
-            KeyRange(head.start0, current.end0) :: tail
+          if head.end0 >= current.start0 - 1 then KeyRange(head.start0, current.end0) :: tail
           else current :: acc
       }
     // but use an IndexedSeq for efficiency later

diff --git a/src/main/scala/org/broadinstitute/gpp/poolq3/barcode/KeyRange.scala b/src/main/scala/org/broadinstitute/gpp/poolq3/barcode/KeyRange.scala
@@ -16,7 +16,7 @@ final case class KeyRange(start0: Int, end0: Int) {
 
   def end1: Int = end0 + 1
 
-  override def toString: String = if (length == 1) start1.toString else s"$start1..$end1"
+  override def toString: String = if length == 1 then start1.toString else s"$start1..$end1"
 
 }
 

diff --git a/src/main/scala/org/broadinstitute/gpp/poolq3/barcode/KnuthMorrisPratt.scala b/src/main/scala/org/broadinstitute/gpp/poolq3/barcode/KnuthMorrisPratt.scala
@@ -17,13 +17,13 @@ final class KnuthMorrisPratt(word: String) {
     val wordLength = word.length
     var m = fromIndex
     var i = 0
-    while ((m + i) < toIndex) {
-      if (word.charAt(i) == text.charAt(m + i)) {
-        if (i == (wordLength - 1)) return Some(m)
+    while (m + i) < toIndex do {
+      if word.charAt(i) == text.charAt(m + i) then {
+        if i == (wordLength - 1) then return Some(m)
         i += 1
       } else {
         val fi = f(i)
-        if (fi > -1) {
+        if fi > -1 then {
           m = m + i - fi
           i = fi
         } else {
@@ -51,12 +51,12 @@ object KnuthMorrisPratt {
     var wi = 2 // the word index
     var si = 0 // the substring index
 
-    while (wi < word.length) {
-      if (word(wi - 1) == word(si)) {
+    while wi < word.length do {
+      if word(wi - 1) == word(si) then {
         f(wi) = si + 1
         si = si + 1
         wi = wi + 1
-      } else if (si > 0) {
+      } else if si > 0 then {
         si = f(si)
       } else {
         f(wi) = 0