Merge pull request #634 from scala/backport-lts-3.3-22945

Backport "Improve printing of strings" to 3.3 LTS

Author: tgodzik

compiler/src/dotty/tools/dotc/printing/PlainPrinter.scala

@@ -11,7 +11,7 @@ import ast.Trees.*
 import typer.Implicits.*
 import typer.ImportInfo
 import Variances.varianceSign
-import util.SourcePosition
+import util.{Chars, SourcePosition}
 import scala.util.control.NonFatal
 import scala.annotation.switch
 import config.{Config, Feature}
@@ -579,22 +579,10 @@ class PlainPrinter(_ctx: Context) extends Printer {
 
   def toText(denot: Denotation): Text = toText(denot.symbol) ~ "/D"
 
-  private def escapedChar(ch: Char): String = (ch: @switch) match {
-    case '\b' => "\\b"
-    case '\t' => "\\t"
-    case '\n' => "\\n"
-    case '\f' => "\\f"
-    case '\r' => "\\r"
-    case '"' => "\\\""
-    case '\'' => "\\\'"
-    case '\\' => "\\\\"
-    case _ => if ch.isControl then f"${"\\"}u${ch.toInt}%04x" else String.valueOf(ch).nn
-  }
-
   def toText(const: Constant): Text = const.tag match {
-    case StringTag => stringText("\"" + escapedString(const.value.toString) + "\"")
+    case StringTag => stringText(Chars.escapedString(const.value.toString, quoted = true))
     case ClazzTag => "classOf[" ~ toText(const.typeValue) ~ "]"
-    case CharTag => literalText(s"'${escapedChar(const.charValue)}'")
+    case CharTag => literalText(Chars.escapedChar(const.charValue))
     case LongTag => literalText(const.longValue.toString + "L")
     case DoubleTag => literalText(const.doubleValue.toString + "d")
     case FloatTag => literalText(const.floatValue.toString + "f")
@@ -612,7 +600,7 @@ class PlainPrinter(_ctx: Context) extends Printer {
     ~ (if param.isTypeParam then "" else ": ")
     ~ toText(param.paramInfo)
 
-  protected def escapedString(str: String): String = str flatMap escapedChar
+  protected final def escapedString(str: String): String = Chars.escapedString(str, quoted = false)
 
   def dclsText(syms: List[Symbol], sep: String): Text = Text(syms map dclText, sep)
 

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

@@ -3,7 +3,8 @@ package dotty.tools.dotc.util
 import scala.annotation.switch
 import Character.{LETTER_NUMBER, LOWERCASE_LETTER, OTHER_LETTER, TITLECASE_LETTER, UPPERCASE_LETTER}
 import Character.{MATH_SYMBOL, OTHER_SYMBOL}
-import Character.{isJavaIdentifierPart, isUnicodeIdentifierStart, isUnicodeIdentifierPart}
+import Character.{isISOControl as isControl, isJavaIdentifierPart, isUnicodeIdentifierStart, isUnicodeIdentifierPart}
+import java.lang.StringBuilder
 
 /** Contains constants and classifier methods for characters */
 object Chars:
@@ -110,3 +111,59 @@ object Chars:
 
   /** Would the character be encoded by `NameTransformer.encode`? */
   def willBeEncoded(c: Char): Boolean = !isJavaIdentifierPart(c)
+
+  private inline def requiresFormat(c: Char): Boolean = (c: @switch) match
+    case '\b' | '\t' | '\n' | '\f' | '\r' | '"' | '\'' | '\\' => true
+    case c => isControl(c)
+
+  def escapedString(text: String, quoted: Boolean): String =
+    inline def doBuild: String =
+      val b = StringBuilder(text.length + 16)
+      if quoted then
+        b.append('"')
+      var i = 0
+      while i < text.length do
+        escapedChar(b, text.charAt(i))
+        i += 1
+      if quoted then
+        b.append('"')
+      b.toString
+    var i = 0
+    while i < text.length do
+      if requiresFormat(text.charAt(i)) then return doBuild
+      i += 1
+    if quoted then "\"" + text + "\""
+    else text
+
+  def escapedChar(ch: Char): String =
+    if requiresFormat(ch) then
+      val b = StringBuilder().append('\'')
+      escapedChar(b, ch)
+      b.append('\'').toString
+    else
+      "'" + ch + "'"
+
+  private def escapedChar(b: StringBuilder, c: Char): Unit =
+    inline def quadNibble(x: Int, i: Int): Unit =
+      if i < 4 then
+        quadNibble(x >> 4, i + 1)
+        val n = x & 0xF
+        val c = if (n < 10) '0' + n else 'a' + (n - 10)
+        b.append(c.toChar)
+    val replace = (c: @switch) match
+      case '\b' => "\\b"
+      case '\t' => "\\t"
+      case '\n' => "\\n"
+      case '\f' => "\\f"
+      case '\r' => "\\r"
+      case '"'  => "\\\""
+      case '\'' => "\\\'"
+      case '\\' => "\\\\"
+      case c =>
+        if isControl(c) then
+          b.append("\\u")
+          quadNibble(c.toInt, 0)
+        else
+          b.append(c)
+        return
+    b.append(replace)

compiler/src/scala/quoted/runtime/impl/printers/SourceCode.scala

@@ -1,7 +1,12 @@
 package scala.quoted
 package runtime.impl.printers
 
+import dotty.tools.dotc.util.Chars
+
 import scala.annotation.switch
+import scala.collection.mutable
+
+import java.lang.StringBuilder
 
 /** Printer for fully elaborated representation of the source code */
 object SourceCode {
@@ -97,7 +102,7 @@ object SourceCode {
       this += lineBreak() += "}"
     }
 
-    def result(): String = sb.result()
+    def result(): String = sb.toString
 
     private def lineBreak(): String = "\n" + ("  " * indent)
     private def doubleLineBreak(): String = "\n\n" + ("  " * indent)
@@ -438,7 +443,7 @@ object SourceCode {
           case _ =>
             inParens {
               printTree(term)
-              this += (if (dotty.tools.dotc.util.Chars.isOperatorPart(sb.last)) " : " else ": ")
+              this += (if Chars.isOperatorPart(sb.charAt(sb.length - 1)) then " : " else ": ")
               def printTypeOrAnnots(tpe: TypeRepr): Unit = tpe match {
                 case AnnotatedType(tp, annot) if tp == term.tpe =>
                   printAnnotation(annot)
@@ -957,9 +962,6 @@ object SourceCode {
 
     }
 
-    inline private val qc  = '\''
-    inline private val qSc = '"'
-
     def printConstant(const: Constant): this.type = const match {
       case UnitConstant() => this += highlightLiteral("()")
       case NullConstant() => this += highlightLiteral("null")
@@ -970,8 +972,8 @@ object SourceCode {
       case LongConstant(v) => this += highlightLiteral(v.toString + "L")
       case FloatConstant(v) => this += highlightLiteral(v.toString + "f")
       case DoubleConstant(v) => this += highlightLiteral(v.toString)
-      case CharConstant(v) => this += highlightString(s"${qc}${escapedChar(v)}${qc}")
-      case StringConstant(v) => this += highlightString(s"${qSc}${escapedString(v)}${qSc}")
+      case CharConstant(v) => this += highlightString(Chars.escapedChar(v))
+      case StringConstant(v) => this += highlightString(Chars.escapedString(v, quoted = true))
       case ClassOfConstant(v) =>
         this += "classOf"
         inSquare(printType(v))
@@ -1444,22 +1446,8 @@ object SourceCode {
     private def +=(x: Char): this.type = { sb.append(x); this }
     private def +=(x: String): this.type = { sb.append(x); this }
 
-    private def escapedChar(ch: Char): String = (ch: @switch) match {
-      case '\b' => "\\b"
-      case '\t' => "\\t"
-      case '\n' => "\\n"
-      case '\f' => "\\f"
-      case '\r' => "\\r"
-      case '"' => "\\\""
-      case '\'' => "\\\'"
-      case '\\' => "\\\\"
-      case _ => if ch.isControl then f"${"\\"}u${ch.toInt}%04x" else String.valueOf(ch).nn
-    }
-
-    private def escapedString(str: String): String = str flatMap escapedChar
-
-    private[this] val names = collection.mutable.Map.empty[Symbol, String]
-    private[this] val namesIndex = collection.mutable.Map.empty[String, Int]
+    private val names = collection.mutable.Map.empty[Symbol, String]
+    private val namesIndex = collection.mutable.Map.empty[String, Int]
 
     private def splicedName(sym: Symbol): Option[String] = {
       if sym.owner.isClassDef then None

compiler/test/dotty/tools/dotc/printing/PrintingTest.scala

@@ -1,6 +1,7 @@
 package dotty
 package tools
 package dotc
+package printing
 
 import scala.language.unsafeNulls
 

tests/printing/untyped/strings.check

@@ -0,0 +1,8 @@
+[[syntax trees at end of                    parser]] // tests/printing/untyped/strings.scala
+package <empty> {
+  class C {
+    def chars = "\b\t\n\f\r\"\'\\a\u0003"
+    def greeting = "hello, world"
+  }
+}
+

tests/printing/untyped/strings.scala

@@ -0,0 +1,4 @@
+
+class C:
+  def chars = "\b\t\n\f\r\"\'\\\u0061\u0003"
+  def greeting = "hello, world"