Merge pull request #29515 from vespa-engine/arnej/parse-special-number-strings

add parsing of special strings for inf/nan cell values

2 files changed, 80 insertions, 8 deletions

diff --git a/vespajlib/src/main/java/com/yahoo/tensor/serialization/JsonFormat.java b/vespajlib/src/main/java/com/yahoo/tensor/serialization/JsonFormat.java
index 28f14c8d7ca..8cf36cbd038 100644
--- a/vespajlib/src/main/java/com/yahoo/tensor/serialization/JsonFormat.java
+++ b/vespajlib/src/main/java/com/yahoo/tensor/serialization/JsonFormat.java
@@ -234,10 +234,11 @@ public class JsonFormat {
         TensorAddress address = decodeAddress(cell.field("address"), builder.type());
 
         Inspector value = cell.field("value");
-        if (value.type() != Type.LONG && value.type() != Type.DOUBLE)
+        if (value.valid()) {
+            builder.cell(address, decodeNumeric(value));
+        } else {
             throw new IllegalArgumentException("Excepted a cell to contain a numeric value called 'value'");
-
-        builder.cell(address, value.asDouble());
+        }
     }
 
     private static void decodeSingleDimensionCell(String key, Inspector value, Tensor.Builder builder) {
@@ -268,8 +269,8 @@ public class JsonFormat {
         values.traverse((ArrayTraverser) (__, value) -> {
             if (value.type() == Type.ARRAY)
                 decodeNestedValues(value, builder, index);
-            else if (value.type() == Type.LONG || value.type() == Type.DOUBLE)
-                indexedBuilder.cellByDirectIndex(index.next(), value.asDouble());
+            else if (value.type() == Type.LONG || value.type() == Type.DOUBLE || value.type() == Type.STRING || value.type() == Type.NIX)
+                indexedBuilder.cellByDirectIndex(index.next(), decodeNumeric(value));
             else
                 throw new IllegalArgumentException("Excepted the values array to contain numbers or nested arrays, not " + value.type());
         });
@@ -446,9 +447,33 @@ public class JsonFormat {
     }
 
     private static double decodeNumeric(Inspector numericField) {
-        if (numericField.type() != Type.LONG && numericField.type() != Type.DOUBLE)
-            throw new IllegalArgumentException("Excepted a number, not " + numericField.type());
-        return numericField.asDouble();
+        if (numericField.type() == Type.DOUBLE || numericField.type() == Type.LONG) {
+            return numericField.asDouble();
+        }
+        if (numericField.type() == Type.STRING) {
+            return decodeNumberString(numericField.asString());
+        }
+        if (numericField.type() == Type.NIX) {
+            return Double.NaN;
+        }
+        throw new IllegalArgumentException("Excepted a number, not " + numericField.type());
+    }
+
+    public static double decodeNumberString(String input) {
+        String s = input.toLowerCase();
+        if (s.equals("infinity") || s.equals("+infinity") || s.equals("inf") || s.equals("+inf")) {
+            return Double.POSITIVE_INFINITY;
+        }
+        if (s.equals("-infinity") || s.equals("-inf")) {
+            return Double.NEGATIVE_INFINITY;
+        }
+        if (s.equals("nan") || s.equals("+nan")) {
+            return Double.NaN;
+        }
+        if (s.equals("-nan")) {
+            return Math.copySign(Double.NaN, -1.0); // or Double.longBitsToDouble(0xfff8000000000000L);
+        }
+        throw new NumberFormatException("Excepted a number, got string '" + input + "'");
     }
 
 }
diff --git a/vespajlib/src/test/java/com/yahoo/tensor/serialization/JsonFormatTestCase.java b/vespajlib/src/test/java/com/yahoo/tensor/serialization/JsonFormatTestCase.java
index d95396aca50..52f44f0ac42 100644
--- a/vespajlib/src/test/java/com/yahoo/tensor/serialization/JsonFormatTestCase.java
+++ b/vespajlib/src/test/java/com/yahoo/tensor/serialization/JsonFormatTestCase.java
@@ -669,6 +669,53 @@ public class JsonFormatTestCase {
                               "{\"type\":\"tensor<float>(x[1])\",\"values\":[0.3333333432674408]}");
     }
 
+    @Test
+    public void testSpecialNumberStrings() {
+        assertEquals(Double.POSITIVE_INFINITY, JsonFormat.decodeNumberString("Infinity"), 0.0);
+        assertEquals(Double.POSITIVE_INFINITY, JsonFormat.decodeNumberString("+Infinity"), 0.0);
+        assertEquals(Double.POSITIVE_INFINITY, JsonFormat.decodeNumberString("Inf"), 0.0);
+        assertEquals(Double.POSITIVE_INFINITY, JsonFormat.decodeNumberString("+Inf"), 0.0);
+        assertEquals(Double.POSITIVE_INFINITY, JsonFormat.decodeNumberString("infinity"), 0.0);
+        assertEquals(Double.NEGATIVE_INFINITY, JsonFormat.decodeNumberString("-Infinity"), 0.0);
+        assertEquals(Double.NEGATIVE_INFINITY, JsonFormat.decodeNumberString("-Inf"), 0.0);
+        assertEquals(Double.NEGATIVE_INFINITY, JsonFormat.decodeNumberString("-infinity"), 0.0);
+        assertEquals(Double.NEGATIVE_INFINITY, JsonFormat.decodeNumberString("-inf"), 0.0);
+        assertEquals(0x7FF8000000000000L, Double.doubleToRawLongBits(JsonFormat.decodeNumberString("nan")));
+        assertEquals(0x7FF8000000000000L, Double.doubleToRawLongBits(JsonFormat.decodeNumberString("NaN")));
+        assertEquals(0x7FF8000000000000L, Double.doubleToRawLongBits(JsonFormat.decodeNumberString("+NaN")));
+        assertEquals(0xFFF8000000000000L, Double.doubleToRawLongBits(JsonFormat.decodeNumberString("-nan")));
+        assertEquals(0xFFF8000000000000L, Double.doubleToRawLongBits(JsonFormat.decodeNumberString("-NaN")));
+    }
+
+    @Test
+    public void testWithNanVariants() {
+        TensorType x3 = TensorType.fromSpec("tensor(x[3])");
+        String json = "{\"cells\":[" +
+                      "{\"address\":{\"x\":\"0\"},\"value\":\"nan\"}," +
+                      "{\"address\":{\"x\":\"1\"},\"value\":null}," +
+                      "{\"address\":{\"x\":\"2\"},\"value\":\"+NaN\"}" +
+                      "]}";
+        var t = JsonFormat.decode(x3, json.getBytes(StandardCharsets.UTF_8));
+        checkThreeNans(t);
+        json = "['nan', null, '+NaN']";
+        t = JsonFormat.decode(x3, json.getBytes(StandardCharsets.UTF_8));
+        checkThreeNans(t);
+        json = "{'type':'tensor(x[3])','values':['nan', null, '+NaN']}";
+        t = JsonFormat.decode(x3, json.getBytes(StandardCharsets.UTF_8));
+        checkThreeNans(t);
+    }
+
+    private void checkThreeNans(Tensor t) {
+        final Double nan = Double.NaN;
+        int cnt = 0;
+        for (var iter = t.cellIterator(); iter.hasNext(); ) {
+            var cell = iter.next();
+            assertEquals(nan, cell.getValue());
+            ++cnt;
+        }
+        assertEquals(3, cnt);
+    }
+
     private void assertEncodeShortForm(String tensor, String expected) {
         assertEncodeShortForm(Tensor.from(tensor), expected);
     }