NaN and Inf in loads - Port of Pandas #30295

lib/ultrajson.h

@@ -166,6 +166,8 @@ enum JSTYPES
   JT_ARRAY,     // Array structure
   JT_OBJECT,    // Key/Value structure
   JT_INVALID,   // Internal, do not return nor expect
+  JT_POS_INF,   // Positive infinity
+  JT_NEG_INF,   // Negative infinity
 };
 
 typedef void * JSOBJ;
@@ -321,6 +323,8 @@ typedef struct __JSONObjectDecoder
   JSOBJ (*newTrue)(void *prv);
   JSOBJ (*newFalse)(void *prv);
   JSOBJ (*newNull)(void *prv);
+  JSOBJ (*newPosInf)(void *prv);
+  JSOBJ (*newNegInf)(void *prv);
   JSOBJ (*newObject)(void *prv);
   JSOBJ (*newArray)(void *prv);
   JSOBJ (*newInt)(void *prv, JSINT32 value);

lib/ultrajsondec.c

@@ -98,10 +98,16 @@ static FASTCALL_ATTR JSOBJ FASTCALL_MSVC decode_numeric (struct DecoderState *ds
 
   JSUINT64 overflowLimit = LLONG_MAX;
 
-  if (*(offset) == '-')
-  {
-    offset ++;
+  if (*(offset) == 'I') {
+      goto DECODE_INF;
+  } else if (*(offset) == 'N') {
+      goto DECODE_NAN;
+  } else if (*(offset) == '-') {
+      offset++;
       intNeg = -1;
+      if (*(offset) == 'I') {
+          goto DECODE_INF;
+      }
     overflowLimit = LLONG_MIN;
   }
 
@@ -189,6 +195,48 @@ BREAK_INT_LOOP:
   {
     return ds->dec->newInt(ds->prv, (JSINT32) (intValue * intNeg));
   }
+
+DECODE_NAN:
+    offset++;
+    if (*(offset++) != 'a') goto SET_NAN_ERROR;
+    if (*(offset++) != 'N') goto SET_NAN_ERROR;
+
+    ds->lastType = JT_NULL;
+    ds->start = offset;
+    return ds->dec->newNull(ds->prv);
+
+SET_NAN_ERROR:
+    return SetError(ds, -1, "Unexpected character found when decoding 'NaN'");
+
+DECODE_INF:
+    offset++;
+    if (*(offset++) != 'n') goto SET_INF_ERROR;
+    if (*(offset++) != 'f') goto SET_INF_ERROR;
+    if (*(offset++) != 'i') goto SET_INF_ERROR;
+    if (*(offset++) != 'n') goto SET_INF_ERROR;
+    if (*(offset++) != 'i') goto SET_INF_ERROR;
+    if (*(offset++) != 't') goto SET_INF_ERROR;
+    if (*(offset++) != 'y') goto SET_INF_ERROR;
+
+    ds->start = offset;
+
+    if (intNeg == 1) {
+      ds->lastType = JT_POS_INF;
+      return ds->dec->newPosInf(ds->prv);
+    } else {
+      ds->lastType = JT_NEG_INF;
+      return ds->dec->newNegInf(ds->prv);
+    }
+
+SET_INF_ERROR:
+    if (intNeg == 1) {
+      const char *msg = "Unexpected character found when decoding 'Infinity'";
+      return SetError(ds, -1, msg);
+    } else {
+      const char *msg = "Unexpected character found when decoding '-Infinity'";
+      return SetError(ds, -1, msg);
+    }
+
 }
 
 static FASTCALL_ATTR JSOBJ FASTCALL_MSVC decode_true ( struct DecoderState *ds)
@@ -732,6 +780,8 @@ static FASTCALL_ATTR JSOBJ FASTCALL_MSVC decode_any(struct DecoderState *ds)
       case '7':
       case '8':
       case '9':
+      case 'I':
+      case 'N':
       case '-':
         return decode_numeric (ds);
 

python/JSONtoObj.c

@@ -79,6 +79,16 @@ static JSOBJ Object_newNull(void *prv)
   Py_RETURN_NONE;
 }
 
+JSOBJ Object_newPosInf(void *prv)
+{
+    return PyFloat_FromDouble(Py_HUGE_VAL);
+}
+
+JSOBJ Object_newNegInf(void *prv)
+{
+    return PyFloat_FromDouble(-Py_HUGE_VAL);
+}
+
 static JSOBJ Object_newObject(void *prv)
 {
   return PyDict_New();
@@ -129,6 +139,8 @@ PyObject* JSONToObj(PyObject* self, PyObject *args, PyObject *kwargs)
     Object_newTrue,
     Object_newFalse,
     Object_newNull,
+    Object_newPosInf,
+    Object_newNegInf,
     Object_newObject,
     Object_newArray,
     Object_newInteger,

tests/test_ujson.py

@@ -670,6 +670,14 @@ def test_encode_raises_allow_nan(test_input, expected_exception):
         ujson.dumps(test_input, allow_nan=False)
 
 
+def test_nan_inf_support():
+    text = '["a", NaN, "NaN", Infinity, "Infinity", -Infinity, "-Infinity"]'
+    data = ujson.loads(text)
+    expected = ["a", float('nan'), "NaN", float('inf'), "Infinity", -float('inf'), "-Infinity"]
+    for a, b in zip(data, expected):
+        assert a == b or a is b
+
+
 @pytest.mark.parametrize(
     "test_input",
     [