Add hex/binary/octal support to String.toInt()

pkl-core/src/main/java/org/pkl/core/ast/builder/AstBuilder.java

@@ -572,35 +572,18 @@ public ExpressionNode visitBoolLiteralExpr(BoolLiteralExpr expr) {
     }
   }
 
-  private <T> T parseNumber(IntLiteralExpr expr, BiFunction<String, Integer, T> parser) {
-    var text = remove_(expr.getNumber());
-
-    var radix = 10;
-    if (text.startsWith("0x") || text.startsWith("0b") || text.startsWith("0o")) {
-      radix =
-          switch (text.charAt(1)) {
-            case 'x' -> 16;
-            case 'b' -> 2;
-            default -> 8;
-          };
-
-      text = text.substring(2);
-    }
-
-    // relies on grammar rule nesting depth, but a breakage won't go unnoticed by tests
-    if (expr.parent() instanceof UnaryMinusExpr) {
-      // handle negation here to make parsing of base.MinInt work
-      // also moves negation from runtime to parse time
-      text = "-" + text;
-    }
-    return parser.apply(text, radix);
+  private <T> T parseInteger(IntLiteralExpr expr, BiFunction<String, Integer, T> parser) {
+    // negate relies on grammar rule nesting depth, but a breakage won't go unnoticed by tests
+    // handle negation here to make parsing of base.MinInt work
+    // also moves negation from runtime to parse time
+    return VmUtils.parseInteger(expr.getNumber(), expr.parent() instanceof UnaryMinusExpr, parser);
   }
 
   @Override
   public IntLiteralNode visitIntLiteralExpr(IntLiteralExpr expr) {
     var section = createSourceSection(expr);
     try {
-      var num = parseNumber(expr, Long::parseLong);
+      var num = parseInteger(expr, Long::parseLong);
       return new IntLiteralNode(section, num);
     } catch (NumberFormatException e) {
       var text = expr.getNumber();
@@ -611,7 +594,7 @@ public IntLiteralNode visitIntLiteralExpr(IntLiteralExpr expr) {
   @Override
   public FloatLiteralNode visitFloatLiteralExpr(FloatLiteralExpr expr) {
     var section = createSourceSection(expr);
-    var text = remove_(expr.getNumber());
+    var text = VmUtils.removeUnderscoresFromNumber(expr.getNumber(), false);
     // relies on grammar rule nesting depth, but a breakage won't go unnoticed by tests
     if (expr.parent() instanceof UnaryMinusExpr) {
       // handle negation here for consistency with visitIntegerLiteral
@@ -627,16 +610,6 @@ public FloatLiteralNode visitFloatLiteralExpr(FloatLiteralExpr expr) {
     }
   }
 
-  private static String remove_(String number) {
-    var builder = new StringBuilder(number.length());
-    for (var i = 0; i < number.length(); i++) {
-      var ch = number.charAt(i);
-      if (ch == '_') continue;
-      builder.append(ch);
-    }
-    return builder.toString();
-  }
-
   @Override
   public ExpressionNode visitThrowExpr(ThrowExpr expr) {
     return ThrowNodeGen.create(createSourceSection(expr), visitExpr(expr.getExpr()));
@@ -1314,7 +1287,7 @@ private Pair<ExpressionNode[], Boolean> createCollectionArgumentBytesNodes(Argum
       var expr = args.get(i);
       if (expr instanceof IntLiteralExpr intLiteralExpr && isAllByteLiterals) {
         try {
-          var byt = parseNumber(intLiteralExpr, Byte::parseByte);
+          var byt = parseInteger(intLiteralExpr, Byte::parseByte);
           expressionNodes[i] = new ByteConstantValueNode(byt);
         } catch (NumberFormatException e) {
           // proceed with initializing a constant value node; we'll throw an error inside

pkl-core/src/main/java/org/pkl/core/runtime/VmUtils.java

@@ -31,6 +31,7 @@
 import java.text.DecimalFormat;
 import java.text.DecimalFormatSymbols;
 import java.util.*;
+import java.util.function.BiFunction;
 import java.util.function.Function;
 import java.util.regex.Pattern;
 import java.util.regex.PatternSyntaxException;
@@ -1092,4 +1093,51 @@ public static boolean isPklBug(VmStackOverflowException e) {
     var truffleStackTraceElements = TruffleStackTrace.getStackTrace(e);
     return truffleStackTraceElements != null && truffleStackTraceElements.size() < 100;
   }
+
+  public static <T> T parseInteger(
+      String raw, boolean negate, BiFunction<String, Integer, T> parser)
+      throws NumberFormatException {
+    var text = raw;
+
+    var radix = 10;
+    if (text.length() >= 2 && text.charAt(0) == '0') {
+      radix =
+          switch (text.charAt(1)) {
+            case 'x' -> 16;
+            case 'o' -> 8;
+            case 'b' -> 2;
+            default -> 10;
+          };
+      if (radix != 10) {
+        text = text.substring(2);
+      }
+    }
+
+    text = removeUnderscoresFromNumber(text, radix == 16);
+    if (negate) {
+      text = "-" + text;
+    }
+    return parser.apply(text, radix);
+  }
+
+  /** Removes `_` from numbers to be parsed. Returns the string unmodified if it's invalid. */
+  public static String removeUnderscoresFromNumber(String number, boolean isHex) {
+    if (number.indexOf('_') < 0) return number;
+
+    var builder = new StringBuilder();
+    var numberStart = true;
+    for (var i = 0; i < number.length(); i++) {
+      var c = number.charAt(i);
+      if (c != '_') {
+        builder.append(c);
+      } else if (numberStart) {
+        // invalid: _ at start or after [.eE]
+        return number;
+      }
+
+      numberStart = c == '.' || (!isHex && (c == 'e' || c == 'E'));
+    }
+
+    return builder.toString();
+  }
 }

pkl-core/src/main/java/org/pkl/core/stdlib/base/StringNodes.java

@@ -835,7 +835,8 @@ public abstract static class toInt extends ExternalMethod0Node {
     @Specialization
     protected long eval(String self) {
       try {
-        return Long.parseLong(removeUnderlinesFromNumber(self));
+        var negate = self.charAt(0) == '-';
+        return VmUtils.parseInteger(negate ? self.substring(1) : self, negate, Long::parseLong);
       } catch (NumberFormatException e) {
         throw exceptionBuilder()
             .evalError("cannotParseStringAs", "Int")
@@ -850,7 +851,8 @@ public abstract static class toIntOrNull extends ExternalMethod0Node {
     @Specialization
     protected Object eval(String self) {
       try {
-        return Long.parseLong(removeUnderlinesFromNumber(self));
+        var negate = self.charAt(0) == '-';
+        return VmUtils.parseInteger(negate ? self.substring(1) : self, negate, Long::parseLong);
       } catch (NumberFormatException e) {
         return VmNull.withoutDefault();
       }
@@ -862,7 +864,7 @@ public abstract static class toFloat extends ExternalMethod0Node {
     @Specialization
     protected double eval(String self) {
       try {
-        return Double.parseDouble(removeUnderlinesFromNumber(self));
+        return Double.parseDouble(VmUtils.removeUnderscoresFromNumber(self, false));
       } catch (NumberFormatException e) {
         throw exceptionBuilder()
             .evalError("cannotParseStringAs", "Float")
@@ -877,7 +879,7 @@ public abstract static class toFloatOrNull extends ExternalMethod0Node {
     @Specialization
     protected Object eval(String self) {
       try {
-        return Double.parseDouble(removeUnderlinesFromNumber(self));
+        return Double.parseDouble(VmUtils.removeUnderscoresFromNumber(self, false));
       } catch (NumberFormatException e) {
         return VmNull.withoutDefault();
       }
@@ -1031,23 +1033,4 @@ private static String applyMapper(
     var replacement = applyNode.executeString(mapper, regexMatch);
     return Matcher.quoteReplacement(replacement);
   }
-
-  /**
-   * Removes `_` from numbers to be parsed to be compatible with how Pkl parses numbers. Will return
-   * the string unmodified if it's invalid.
-   */
-  private static String removeUnderlinesFromNumber(String number) {
-    var builder = new StringBuilder();
-    var numberStart = true;
-    for (var i = 0; i < number.length(); i++) {
-      var c = number.charAt(i);
-      if (c != '_') {
-        builder.append(c);
-      } else if (numberStart) return number;
-
-      numberStart = c == '.' || c == 'e' || c == 'E';
-    }
-
-    return builder.toString();
-  }
 }

pkl-core/src/test/files/LanguageSnippetTests/input/api/string.pkl

@@ -270,11 +270,34 @@ examples {
     "-1_2__3___".toInt()
     "0".toInt()
     "-0".toInt()
+
+    "0x123".toInt()
+    "-0x123".toInt()
+    "0x1_2__3___".toInt()
+    "-0x1_2__3___".toInt()
+    "0x0".toInt()
+    "-0x0".toInt()
+
+    "0b101".toInt()
+    "-0b101".toInt()
+    "0b1_0__1___".toInt()
+    "-0b1_0__1___".toInt()
+    "0b0".toInt()
+    "-0b0".toInt()
+
+    "0o123".toInt()
+    "-0o123".toInt()
+    "0o1_2__3___".toInt()
+    "-0o1_2__3___".toInt()
+    "0o0".toInt()
+    "-0o0".toInt()
+
     module.catch(() -> "1.2".toInt())
     module.catch(() -> "9223372036854775808".toInt())
     module.catch(() -> "-9223372036854775809".toInt())
     module.catch(() -> "abc".toInt())
     module.catch(() -> "_1_000".toInt())
+    module.catch(() -> "0p0".toInt())
   }
 
   ["toIntOrNull()"] {

pkl-core/src/test/files/LanguageSnippetTests/output/api/reflectedDeclaration.pcf

@@ -2255,9 +2255,15 @@ alias {
             displayUri = "https://github.com/apple/pkl/blob/$commitId/stdlib/base.pkl#LXXXX"
           }
           docComment = """
-            Parses this string as a signed decimal (base 10) integer.
+            Parses this string as a signed integer.
 
-            Throws if this string cannot be parsed as a signed decimal integer,
+            Supports integer formats supported by Pkl:
+            * Decimal (base 10)
+            * Hexadecimal (base 16) with prefix `0x` or `-0x`
+            * Binary (base 2) with prefix `0b` or `-0b`
+            * Octal (base 8) with prefix `0o` or `-0o`
+
+            Throws if this string cannot be parsed,
             or if the integer is too large to fit into [Int].
             """
           annotations = List()
@@ -2272,9 +2278,15 @@ alias {
             displayUri = "https://github.com/apple/pkl/blob/$commitId/stdlib/base.pkl#LXXXX"
           }
           docComment = """
-            Parses this string as a signed decimal (base 10) integer.
+            Parses this string as a signed integer.
+
+            Supports integer formats supported by Pkl:
+            * Decimal (base 10)
+            * Hexadecimal (base 16) with prefix `0x` or `-0x`
+            * Binary (base 2) with prefix `0b` or `-0b`
+            * Octal (base 8) with prefix `0o` or `-0o`
 
-            Returns [null] if this string cannot be parsed as a signed decimal integer,
+            Returns [null] if this string cannot be parsed,
             or if the integer is too large to fit into [Int].
             """
           annotations = List()
@@ -3192,9 +3204,15 @@ alias {
             displayUri = "https://github.com/apple/pkl/blob/$commitId/stdlib/base.pkl#LXXXX"
           }
           docComment = """
-            Parses this string as a signed decimal (base 10) integer.
+            Parses this string as a signed integer.
 
-            Throws if this string cannot be parsed as a signed decimal integer,
+            Supports integer formats supported by Pkl:
+            * Decimal (base 10)
+            * Hexadecimal (base 16) with prefix `0x` or `-0x`
+            * Binary (base 2) with prefix `0b` or `-0b`
+            * Octal (base 8) with prefix `0o` or `-0o`
+
+            Throws if this string cannot be parsed,
             or if the integer is too large to fit into [Int].
             """
           annotations = List()
@@ -3209,9 +3227,15 @@ alias {
             displayUri = "https://github.com/apple/pkl/blob/$commitId/stdlib/base.pkl#LXXXX"
           }
           docComment = """
-            Parses this string as a signed decimal (base 10) integer.
+            Parses this string as a signed integer.
+
+            Supports integer formats supported by Pkl:
+            * Decimal (base 10)
+            * Hexadecimal (base 16) with prefix `0x` or `-0x`
+            * Binary (base 2) with prefix `0b` or `-0b`
+            * Octal (base 8) with prefix `0o` or `-0o`
 
-            Returns [null] if this string cannot be parsed as a signed decimal integer,
+            Returns [null] if this string cannot be parsed,
             or if the integer is too large to fit into [Int].
             """
           annotations = List()

pkl-core/src/test/files/LanguageSnippetTests/output/api/string.pcf

@@ -220,11 +220,30 @@ examples {
     -123
     0
     0
+    291
+    -291
+    291
+    -291
+    0
+    0
+    5
+    -5
+    5
+    -5
+    0
+    0
+    83
+    -83
+    83
+    -83
+    0
+    0
     "Cannot parse string as `Int`. String: \"1.2\""
     "Cannot parse string as `Int`. String: \"9223372036854775808\""
     "Cannot parse string as `Int`. String: \"-9223372036854775809\""
     "Cannot parse string as `Int`. String: \"abc\""
     "Cannot parse string as `Int`. String: \"_1_000\""
+    "Cannot parse string as `Int`. String: \"0p0\""
   }
   ["toIntOrNull()"] {
     123

stdlib/base.pkl

@@ -1640,15 +1640,27 @@ external class String extends Any {
   /// ```
   external function decapitalize(): String
 
-  /// Parses this string as a signed decimal (base 10) integer.
+  /// Parses this string as a signed integer.
   ///
-  /// Throws if this string cannot be parsed as a signed decimal integer,
+  /// Supports integer formats supported by Pkl:
+  /// * Decimal (base 10)
+  /// * Hexadecimal (base 16) with prefix `0x` or `-0x`
+  /// * Binary (base 2) with prefix `0b` or `-0b`
+  /// * Octal (base 8) with prefix `0o` or `-0o`
+  ///
+  /// Throws if this string cannot be parsed,
   /// or if the integer is too large to fit into [Int].
   external function toInt(): Int
 
-  /// Parses this string as a signed decimal (base 10) integer.
+  /// Parses this string as a signed integer.
+  ///
+  /// Supports integer formats supported by Pkl:
+  /// * Decimal (base 10)
+  /// * Hexadecimal (base 16) with prefix `0x` or `-0x`
+  /// * Binary (base 2) with prefix `0b` or `-0b`
+  /// * Octal (base 8) with prefix `0o` or `-0o`
   ///
-  /// Returns [null] if this string cannot be parsed as a signed decimal integer,
+  /// Returns [null] if this string cannot be parsed,
   /// or if the integer is too large to fit into [Int].
   external function toIntOrNull(): Int?