Fix columnation and forward port improvements

Avoid boxed ints when creating line index table.
Reject more bad inputs, but keep the convention
that the current line at EOF is the last line
(whether or not the last line is empty).

Author: som-snytt

compiler/src/dotty/tools/dotc/util/SourceFile.scala

@@ -11,8 +11,7 @@ import core.Contexts._
 import scala.io.Codec
 import Chars._
 import scala.annotation.internal.sharable
-import scala.collection.mutable
-import scala.collection.mutable.ArrayBuffer
+import scala.collection.mutable.ArrayBuilder
 import scala.util.chaining.given
 
 import java.io.File.separator
@@ -95,8 +94,8 @@ class SourceFile(val file: AbstractFile, computeContent: => Array[Char]) extends
   def apply(idx: Int): Char = content().apply(idx)
 
   /** length of the original source file
-   * Note that when the source is from Tasty, content() could be empty even though length > 0.
-   * Use content().length to determine the length of content(). */
+   *  Note that when the source is from Tasty, content() could be empty even though length > 0.
+   *  Use content().length to determine the length of content(). */
   def length: Int =
     if lineIndicesCache ne null then lineIndicesCache.last
     else content().length
@@ -122,23 +121,25 @@ class SourceFile(val file: AbstractFile, computeContent: => Array[Char]) extends
   def positionInUltimateSource(position: SourcePosition): SourcePosition =
     SourcePosition(underlying, position.span shift start)
 
-  private def calculateLineIndicesFromContents() = {
+  private def calculateLineIndicesFromContents(): Array[Int] =
     val cs = content()
-    val buf = new ArrayBuffer[Int]
-    buf += 0
+    val buf = new ArrayBuilder.ofInt
+    buf.sizeHint(cs.length / 30)       // guesstimate line count
+    buf.addOne(0)
     var i = 0
     while i < cs.length do
       val isLineBreak =
         val ch = cs(i)
         // don't identify the CR in CR LF as a line break, since LF will do.
         if ch == CR then i + 1 == cs.length || cs(i + 1) != LF
         else isLineBreakChar(ch)
-      if isLineBreak then buf += i + 1
+      if isLineBreak then buf.addOne(i + 1)
       i += 1
-    buf += cs.length // sentinel, so that findLine below works smoother
-    buf.toArray
-  }
+    buf.addOne(cs.length)              // sentinel, so that findLine below works smoother
+    buf.result()
 
+  // offsets of lines, plus a sentinel which is the content length.
+  // if the last line is empty (end of context is NL) then the penultimate index == sentinel.
   private var lineIndicesCache: Array[Int] = _
   private def lineIndices: Array[Int] =
     if lineIndicesCache eq null then
@@ -159,7 +160,9 @@ class SourceFile(val file: AbstractFile, computeContent: => Array[Char]) extends
     lineIndicesCache = indices
 
   /** Map line to offset of first character in line */
-  def lineToOffset(index: Int): Int = lineIndices(index)
+  def lineToOffset(index: Int): Int =
+    if index < 0 || index >= lineIndices.length - 1 then throw new IndexOutOfBoundsException(index.toString)
+    lineIndices(index)
 
   /** Like `lineToOffset`, but doesn't crash if the index is out of bounds. */
   def lineToOffsetOpt(index: Int): Option[Int] =
@@ -171,24 +174,28 @@ class SourceFile(val file: AbstractFile, computeContent: => Array[Char]) extends
   /** A cache to speed up offsetToLine searches to similar lines */
   private var lastLine = 0
 
-  /** Convert offset to line in this source file
-   *  Lines are numbered from 0
+  /** Convert offset to line in this source file.
+   *  Lines are numbered from 0. Conventionally, a final empty line begins at EOF.
+   *  For simplicity, the line at EOF is the last line (possibly non-empty).
    */
-  def offsetToLine(offset: Int): Int = {
+  def offsetToLine(offset: Int): Int =
+    //if lineIndices.isEmpty || offset < lineIndices.head || offset > lineIndices.last then
+    //  throw new IndexOutOfBoundsException(offset.toString)
     lastLine = Util.bestFit(lineIndices, lineIndices.length, offset, lastLine)
     if (offset >= length) lastLine -= 1 // compensate for the sentinel
     lastLine
-  }
 
   /** The index of the first character of the line containing position `offset` */
   def startOfLine(offset: Int): Int = {
     require(offset >= 0)
     lineToOffset(offsetToLine(offset))
   }
 
-  /** The start index of the line following the one containing position `offset` */
+  /** The start index of the line following the one containing position `offset` or `length` at last line */
   def nextLine(offset: Int): Int =
-    lineToOffset(offsetToLine(offset) + 1 min lineIndices.length - 1)
+    val next = offsetToLine(offset) + 1
+    if next >= lineIndices.length - 1 then lineIndices.last
+    else lineToOffset(next)
 
   /** The content of the line containing position `offset` */
   def lineContent(offset: Int): String =
@@ -198,10 +205,10 @@ class SourceFile(val file: AbstractFile, computeContent: => Array[Char]) extends
   def column(offset: Int): Int = {
     var idx = startOfLine(offset)
     var col = 0
-    while (idx != offset) {
-      col += (if (idx < content().length && content()(idx) == '\t') (tabInc - col) % tabInc else 1)
+    while idx != offset do
+      if idx < content().length && content()(idx) == '\t' then col += tabInc - col % tabInc
+      else col += 1
       idx += 1
-    }
     col
   }
 

compiler/test/dotty/tools/dotc/util/SourceFileTests.scala

@@ -0,0 +1,112 @@
+
+package dotty.tools
+package dotc.util
+
+import Spans.*
+
+import org.junit.Assert.{assertThrows as _, *}
+import org.junit.Test
+
+class SourceFileTests:
+  @Test def `i15209 source column handles tabs`: Unit =
+    val text = "\ta\n \tb\n \t c\n"
+    val f = SourceFile.virtual("batch", text)
+    assertEquals(8, f.column(text.indexOf('a')))
+    assertEquals(8, f.column(text.indexOf('b')))
+    assertEquals(9, f.column(text.indexOf('c')))
+
+  def lineContentOf(code: String, offset: Int) = SourceFile.virtual("batch", code).lineContent(offset)
+
+  @Test def si8205_lineToString: Unit =
+    assertEquals("", lineContentOf("", 0))
+    assertEquals("abc", lineContentOf("abc", 0))
+    assertEquals("abc", lineContentOf("abc", 3))
+    assertEquals("code no newline", lineContentOf("code no newline", 1))
+    assertEquals("\n", lineContentOf("\n", 0))
+    assertEquals("abc\n", lineContentOf("abc\ndef", 0))
+    assertEquals("abc\n", lineContentOf("abc\ndef", 3))
+    assertEquals("def", lineContentOf("abc\ndef", 4))
+    assertEquals("def", lineContentOf("abc\ndef", 6))
+    assertEquals("def\n", lineContentOf("abc\ndef\n", 7))
+
+  @Test def CRisEOL: Unit =
+    assertEquals("\r", lineContentOf("\r", 0))
+    assertEquals("abc\r", lineContentOf("abc\rdef", 0))
+    assertEquals("abc\r", lineContentOf("abc\rdef", 3))
+    assertEquals("def", lineContentOf("abc\rdef", 4))
+    assertEquals("def", lineContentOf("abc\rdef", 6))
+    assertEquals("def\r", lineContentOf("abc\rdef\r", 7))
+
+  @Test def CRNLisEOL(): Unit =
+    assertEquals("\r\n", lineContentOf("\r\n", 0))
+    assertEquals("abc\r\n", lineContentOf("abc\r\ndef", 0))
+    assertEquals("abc\r\n", lineContentOf("abc\r\ndef", 3))
+    assertEquals("abc\r\n", lineContentOf("abc\r\ndef", 4))
+    assertEquals("def", lineContentOf("abc\r\ndef", 5))
+    assertEquals("def", lineContentOf("abc\r\ndef", 7))
+    assertEquals("def", lineContentOf("abc\r\ndef", 8))
+    assertEquals("def\r\n", lineContentOf("abc\r\ndef\r\n", 9))
+
+  @Test def `t9885 lineToOffset throws on bad line`: Unit =
+    val text = "a\nb\nc\n"
+    val f = SourceFile.virtual("batch", text)
+    // was: EOL is line terminator, not line separator, so there is not an empty 4th line
+    val splitsville = text.split("\n")
+    assertEquals(List(0, 2, 4, 6), (0 to splitsville.nn.length).toList.map(f.lineToOffset))
+    assertThrows[IndexOutOfBoundsException] {
+      f.lineToOffset(4)   // was: 3 in Scala 2 (no empty line), 5 in Scala 3 (sentinel induces off-by-one)
+    }
+
+    // Position and SourceFile used to count differently
+    val p = SourcePosition(f, Span(text.length - 1))
+    val q = SourcePosition(f, Span(f.lineToOffset(p.line - 1)))
+    assertEquals(2, p.line)
+    assertEquals(p.line - 1, q.line)
+    assertEquals(p.column, q.column + 1)
+    assertEquals(f.startOfLine(p.span.start), SourcePosition(f, Span(f.lineToOffset(p.line))).span.start)
+
+  @Test def `t9885 lineToOffset ignores lack of final EOL`: Unit =
+    val text = "a\nb\nc"
+    val f = SourceFile.virtual("batch", text)
+    assertThrows[IndexOutOfBoundsException] {
+      f.lineToOffset(3)
+    }
+    assertEquals(4, f.lineToOffset(2))
+    assertEquals(2, f.offsetToLine(text.length))
+
+  @Test def `t11572 offsetToLine throws on bad offset`: Unit =
+    val text = "a\nb\nc\n"
+    val f = SourceFile.virtual("batch", text)
+    /* current code requires offsets untethered from source
+    assertThrows[IndexOutOfBoundsException] {
+      f.offsetToLine(-1)  // was: -1
+    }
+    assertThrows[IndexOutOfBoundsException] {
+      f.offsetToLine(7)   // was: 3
+    }
+    */
+    assertEquals(0, f.offsetToLine(0))
+    assertEquals(0, f.offsetToLine(1))
+    assertEquals(1, f.offsetToLine(2))
+    assertEquals(2, f.offsetToLine(4))
+    assertEquals(2, f.offsetToLine(5))
+    assertEquals(3, f.offsetToLine(6))
+
+  @Test def `t11572b offsetToLine throws on bad offset`: Unit =
+    val text = "a\nb\nc\nd"
+    val f = SourceFile.virtual("batch", text)
+    /*
+    assertThrows[IndexOutOfBoundsException] {
+      f.offsetToLine(-1)
+    }
+    assertThrows[IndexOutOfBoundsException] {
+      f.offsetToLine(8)
+    }
+    */
+    assertEquals(0, f.offsetToLine(0))
+    assertEquals(0, f.offsetToLine(1))
+    assertEquals(1, f.offsetToLine(2))
+    assertEquals(2, f.offsetToLine(4))
+    assertEquals(2, f.offsetToLine(5))
+    assertEquals(3, f.offsetToLine(6))
+    assertEquals(3, f.offsetToLine(7))

compiler/test/dotty/tools/utils.scala

@@ -37,6 +37,8 @@ def readFile(f: File): String                       = withFile(f)(_.mkString)
 
 private object Unthrown extends ControlThrowable
 
+def assertThrows[T <: Throwable: ClassTag](body: => Any): Unit = assertThrows[T](_ => true)(body)
+
 def assertThrows[T <: Throwable: ClassTag](p: T => Boolean)(body: => Any): Unit =
   try
     body
@@ -68,4 +70,4 @@ def toolArgsParse(lines: List[String]): List[String] = {
     // but avoid picking up comments like "% scalac ./a.scala" and "$ scalac a.scala"
   }.map(stripped).headOption
   args.map(dotc.config.CommandLineParser.tokenize).getOrElse(Nil)
-}
+}