Math emulation improvements (#10176)

Fixes #9711 by using native emulation of `log1p`, `expm1` and `hypot`.
Fixes #9663 by adding the implementation of `IEEEremainder`
Adds `absExact`, `multiplyFull`, `ulp`, and `getExponent` + missing
variations of `multiplyExact`, `floorDiv`, `floorMod` for completeness.

Only missing methods for JVM 17 compatibility are `multiplyHigh` and
`fma` which might be better handled in #9909

Author: zbynek

user/super/com/google/gwt/emul/java/lang/Math.java

@@ -17,6 +17,7 @@
 
 import static javaemul.internal.InternalPreconditions.checkCriticalArithmetic;
 
+import javaemul.internal.JsUtils;
 import jsinterop.annotations.JsMethod;
 import jsinterop.annotations.JsPackage;
 import jsinterop.annotations.JsType;
@@ -25,13 +26,6 @@
  * Math utility methods and constants.
  */
 public final class Math {
-  // The following methods are not implemented because JS doesn't provide the
-  // necessary pieces:
-  //   public static double ulp (double x)
-  //   public static float ulp (float x)
-  //   public static int getExponent (double d)
-  //   public static int getExponent (float f)
-  //   public static double IEEEremainder(double f1, double f2)
 
   public static final double E = 2.7182818284590452354;
   public static final double PI = 3.14159265358979323846;
@@ -52,6 +46,16 @@ public static long abs(long x) {
     return x < 0 ? -x : x;
   }
 
+  public static int absExact(int v) {
+    checkCriticalArithmetic(v != Integer.MIN_VALUE);
+    return abs(v);
+  }
+
+  public static long absExact(long v) {
+    checkCriticalArithmetic(v != Long.MIN_VALUE);
+    return abs(v);
+  }
+
   @JsMethod(namespace = JsPackage.GLOBAL, name = "Math.acos")
   public static native double acos(double x);
 
@@ -77,9 +81,8 @@ public static long addExact(long x, long y) {
   @JsMethod(namespace = JsPackage.GLOBAL, name = "Math.atan2")
   public static native double atan2(double y, double x);
 
-  public static double cbrt(double x) {
-    return x == 0 || !Double.isFinite(x) ? x : pow(x, 1.0 / 3.0);
-  }
+  @JsMethod(namespace = JsPackage.GLOBAL, name = "Math.cbrt")
+  public static native double cbrt(double x);
 
   @JsMethod(namespace = JsPackage.GLOBAL, name = "Math.ceil")
   public static native double ceil(double x);
@@ -99,9 +102,8 @@ public static float copySign(float magnitude, float sign) {
   @JsMethod(namespace = JsPackage.GLOBAL, name = "Math.cos")
   public static native double cos(double x);
 
-  public static double cosh(double x) {
-    return (exp(x) + exp(-x)) / 2;
-  }
+  @JsMethod(namespace = JsPackage.GLOBAL, name = "Math.cosh")
+  public static native double cosh(double x);
 
   public static int decrementExact(int x) {
     checkCriticalArithmetic(x != Integer.MIN_VALUE);
@@ -116,9 +118,8 @@ public static long decrementExact(long x) {
   @JsMethod(namespace = JsPackage.GLOBAL, name = "Math.exp")
   public static native double exp(double x);
 
-  public static double expm1(double d) {
-    return d == 0 ? d : exp(d) - 1;
-  }
+  @JsMethod(namespace = JsPackage.GLOBAL, name = "Math.expm1")
+  public static native double expm1(double d);
 
   @JsMethod(namespace = JsPackage.GLOBAL, name = "Math.floor")
   public static native double floor(double x);
@@ -135,6 +136,10 @@ public static long floorDiv(long dividend, long divisor) {
     return ((dividend ^ divisor) >= 0 ? dividend / divisor : ((dividend + 1) / divisor) - 1);
   }
 
+  public static long floorDiv(long dividend, int divisor) {
+    return floorDiv(dividend, (long) divisor);
+  }
+
   public static int floorMod(int dividend, int divisor) {
     checkCriticalArithmetic(divisor != 0);
     return ((dividend % divisor) + divisor) % divisor;
@@ -145,11 +150,53 @@ public static long floorMod(long dividend, long divisor) {
     return ((dividend % divisor) + divisor) % divisor;
   }
 
-  public static double hypot(double x, double y) {
-    return Double.isInfinite(x) || Double.isInfinite(y) ?
-        Double.POSITIVE_INFINITY : sqrt(x * x + y * y);
+  public static long floorMod(long dividend, int divisor) {
+    return floorMod(dividend, (long) divisor);
+  }
+
+  @SuppressWarnings("CheckStyle.MethodName")
+  public static double IEEEremainder(double v, double m) {
+    double ratio = v / m;
+    double closest = Math.ceil(ratio);
+    double frac = Math.abs(closest - ratio);
+    if (frac > 0.5 || frac == 0.5 && (closest % 2 != 0)) {
+      closest = Math.floor(ratio);
+    }
+    // if closest == 0 and m == inf, avoid multiplication
+    return closest == 0 ? v : v - m * closest;
+  }
+
+  public static int getExponent(double v) {
+    int[] intBits = JsUtils.doubleToRawIntBits(v);
+    return ((intBits[1] >> 20) & 2047) - Double.MAX_EXPONENT;
+  }
+
+  public static int getExponent(float v) {
+    return ((JsUtils.floatToRawIntBits(v) >> 23) & 255) - Float.MAX_EXPONENT;
+  }
+
+  public static double ulp(double v) {
+    if (!Double.isFinite(v)) {
+      return Math.abs(v);
+    }
+    int exponent = Math.getExponent(v);
+    if (exponent == -1023) {
+      return Double.MIN_VALUE;
+    }
+    return Math.pow(2, exponent - 52);
+  }
+
+  public static float ulp(float v) {
+    int exponent = Math.getExponent(v);
+    if (exponent == -Float.MAX_EXPONENT) {
+      return Float.MIN_VALUE;
+    }
+    return (float) Math.pow(2, exponent - 23);
   }
 
+  @JsMethod(namespace = JsPackage.GLOBAL, name = "Math.hypot")
+  public static native double hypot(double x, double y);
+
   public static int incrementExact(int x) {
     checkCriticalArithmetic(x != Integer.MAX_VALUE);
     return x + 1;
@@ -163,13 +210,11 @@ public static long incrementExact(long x) {
   @JsMethod(namespace = JsPackage.GLOBAL, name = "Math.log")
   public static native double log(double x);
 
-  public static double log10(double x) {
-    return log(x) * NativeMath.LOG10E;
-  }
+  @JsMethod(namespace = JsPackage.GLOBAL, name = "Math.log10")
+  public static native double log10(double x);
 
-  public static double log1p(double x) {
-    return x == 0 ? x : log(x + 1);
-  }
+  @JsMethod(namespace = JsPackage.GLOBAL, name = "Math.log1p")
+  public static native double log1p(double x);
 
   @JsMethod(namespace = JsPackage.GLOBAL, name = "Math.max")
   public static native double max(double x, double y);
@@ -197,12 +242,28 @@ public static long min(long x, long y) {
     return x < y ? x : y;
   }
 
+  public static long multiplyFull(int x, int y) {
+    return (long) x * (long) y;
+  }
+
   public static int multiplyExact(int x, int y) {
     double r = (double) x * (double) y;
     checkCriticalArithmetic(isSafeIntegerRange(r));
     return (int) r;
   }
 
+  public static long multiplyExact(long x, int y) {
+    if (y == -1) {
+      return negateExact(x);
+    }
+    if (y == 0) {
+      return 0;
+    }
+    long r = x * y;
+    checkCriticalArithmetic(r / y == x);
+    return r;
+  }
+
   public static long multiplyExact(long x, long y) {
     if (y == -1) {
       return negateExact(x);
@@ -284,13 +345,8 @@ public static float scalb(float f, int scaleFactor) {
     return (float) scalb((double) f, scaleFactor);
   }
 
-  public static double signum(double d) {
-    if (d == 0 || Double.isNaN(d)) {
-      return d;
-    } else {
-      return d < 0 ? -1 : 1;
-    }
-  }
+  @JsMethod(namespace = JsPackage.GLOBAL, name = "Math.sign")
+  public static native double signum(double d);
 
   public static float signum(float f) {
     return (float) signum((double) f);
@@ -299,28 +355,17 @@ public static float signum(float f) {
   @JsMethod(namespace = JsPackage.GLOBAL, name = "Math.sin")
   public static native double sin(double x);
 
-  public static double sinh(double x) {
-    return x == 0.0 ? x : (exp(x) - exp(-x)) / 2;
-  }
+  @JsMethod(namespace = JsPackage.GLOBAL, name = "Math.sinh")
+  public static native double sinh(double x);
 
   @JsMethod(namespace = JsPackage.GLOBAL, name = "Math.sqrt")
   public static native double sqrt(double x);
 
   @JsMethod(namespace = JsPackage.GLOBAL, name = "Math.tan")
   public static native double tan(double x);
 
-  public static double tanh(double x) {
-    if (x == 0.0) {
-      // -0.0 should return -0.0.
-      return x;
-    }
-
-    double e2x = exp(2 * x);
-    if (Double.isInfinite(e2x)) {
-      return 1;
-    }
-    return (e2x - 1) / (e2x + 1);
-  }
+  @JsMethod(namespace = JsPackage.GLOBAL, name = "Math.tanh")
+  public static native double tanh(double x);
 
   public static double toDegrees(double x) {
     return x * PI_UNDER_180;
@@ -411,7 +456,6 @@ private static boolean isSafeIntegerRange(double value) {
 
   @JsType(isNative = true, name = "Math", namespace = JsPackage.GLOBAL)
   private static class NativeMath {
-    public static double LOG10E;
     public static native double round(double x);
   }
 }

user/super/com/google/gwt/emul/javaemul/internal/JsUtils.java

@@ -123,10 +123,14 @@ public static float intBitsToFloat(int value) {
     return JsUtils.<float[]>uncheckedCast(new Float32Array(buf))[0];
   }
 
-  public static long doubleToRawLongBits(double value) {
+  public static int[] doubleToRawIntBits(double value) {
     ArrayBuffer buf = new ArrayBuffer(8);
     JsUtils.<double[]>uncheckedCast(new Float64Array(buf))[0] = value;
-    int[] intBits = JsUtils.<int[]>uncheckedCast(new Uint32Array(buf));
+    return JsUtils.<int[]>uncheckedCast(new Uint32Array(buf));
+  }
+
+  public static long doubleToRawLongBits(double value) {
+    int[] intBits = doubleToRawIntBits(value);
     return LongUtils.fromBits(intBits[0] | 0, intBits[1] | 0);
   }
 

user/test-super/com/google/gwt/emultest/super/com/google/gwt/emultest/java17/lang/MathTest.java

@@ -0,0 +1,55 @@
+/*
+ * Copyright 2025 GWT Project Authors
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License"); you may not
+ * use this file except in compliance with the License. You may obtain a copy of
+ * the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
+ * License for the specific language governing permissions and limitations under
+ * the License.
+ */
+package com.google.gwt.emultest.java17.lang;
+
+import com.google.gwt.emultest.java.util.EmulTestBase;
+
+/**
+ * Tests for java.lang.String Java 12 API emulation.
+ */
+public class MathTest extends EmulTestBase {
+
+  public void testAbsExact() {
+    assertEquals(Integer.MAX_VALUE, Math.absExact(hideFromCompiler(Integer.MAX_VALUE)));
+    assertEquals(0, Math.absExact(hideFromCompiler(0)));
+    assertEquals(1, Math.absExact(hideFromCompiler(-1)));
+    assertThrowsArithmetic(() -> Math.absExact(hideFromCompiler(Integer.MIN_VALUE)));
+    assertEquals(Long.MAX_VALUE, Math.absExact(hideFromCompiler(Long.MAX_VALUE)));
+    assertEquals(0, Math.absExact(hideFromCompiler(0l)));
+    assertEquals(1, Math.absExact(hideFromCompiler(-1l)));
+    assertThrowsArithmetic(() -> Math.absExact(hideFromCompiler(Long.MIN_VALUE)));
+  }
+
+  public void testAbsExactWithFolding() {
+    assertEquals(Integer.MAX_VALUE, Math.absExact(Integer.MAX_VALUE));
+    assertEquals(0, Math.absExact(0));
+    assertEquals(1, Math.absExact(-1));
+    assertThrowsArithmetic(() -> Math.absExact(Integer.MIN_VALUE));
+    assertEquals(Long.MAX_VALUE, Math.absExact(Long.MAX_VALUE));
+    assertEquals(0, Math.absExact(0L));
+    assertEquals(1, Math.absExact(-1L));
+    assertThrowsArithmetic(() -> Math.absExact(Long.MIN_VALUE));
+  }
+
+  private void assertThrowsArithmetic(Runnable check) {
+    try {
+      check.run();
+      fail("Should have failed");
+    } catch (ArithmeticException ex) {
+      // good
+    }
+  }
+}

user/test/com/google/gwt/emultest/EmulJava17Suite.java

@@ -16,6 +16,7 @@
 package com.google.gwt.emultest;
 
 import com.google.gwt.emultest.java17.lang.CharSequenceTest;
+import com.google.gwt.emultest.java17.lang.MathTest;
 import com.google.gwt.emultest.java17.lang.StringTest;
 import com.google.gwt.emultest.java17.util.MapEntryTest;
 import com.google.gwt.emultest.java17.util.stream.CollectorsTest;
@@ -32,6 +33,7 @@
 @Suite.SuiteClasses({
     CharSequenceTest.class,
     StringTest.class,
+    MathTest.class,
     CollectorsTest.class,
     StreamTest.class,
     DoubleStreamTest.class,

user/test/com/google/gwt/emultest/java/lang/DoubleTest.java

@@ -161,10 +161,8 @@ public void testDoubleConstants() {
     assertTrue(Double.MIN_VALUE < Double.MAX_VALUE);
     assertFalse(Double.NaN == Double.NaN);
     assertEquals(64, Double.SIZE);
-    // jdk1.6 assertEquals(Math.getExponent(Double.MAX_VALUE),
-    // Double.MAX_EXPONENT);
-    // jdk1.6 assertEquals(Math.getExponent(Double.MIN_NORMAL),
-    // Double.MIN_EXPONENT);
+    assertEquals(Math.getExponent(Double.MAX_VALUE), Double.MAX_EXPONENT);
+    assertEquals(Math.getExponent(Double.MIN_NORMAL), Double.MIN_EXPONENT);
     // issue 8073 - used to fail in prod mode
     assertFalse(Double.isInfinite(Double.NaN));
   }

user/test/com/google/gwt/emultest/java17/lang/MathTest.java

@@ -0,0 +1,35 @@
+/*
+ * Copyright 2025 GWT Project Authors
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License"); you may not
+ * use this file except in compliance with the License. You may obtain a copy of
+ * the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
+ * License for the specific language governing permissions and limitations under
+ * the License.
+ */
+package com.google.gwt.emultest.java17.lang;
+
+import com.google.gwt.dev.util.arg.SourceLevel;
+import com.google.gwt.emultest.java.util.EmulTestBase;
+import com.google.gwt.junit.JUnitShell;
+
+public class MathTest extends EmulTestBase {
+
+  public void testAbsExact() {
+    assertFalse(isGwtSourceLevel17());
+  }
+
+  public void testAbsExactWithFolding() {
+    assertFalse(isGwtSourceLevel17());
+  }
+
+  private boolean isGwtSourceLevel17() {
+    return JUnitShell.getCompilerOptions().getSourceLevel().compareTo(SourceLevel.JAVA17) >= 0;
+  }
+}