add __hash__ to complex, and fix the hash problem in long and str

src/runtime/complex.cpp

@@ -228,6 +228,33 @@ static void _addFunc(const char* name, ConcreteCompilerType* rtn_type, void* com
     complex_cls->giveAttr(name, new BoxedFunction(cl));
 }
 
+Box* complexHash(BoxedComplex* self) {
+    if (!isSubclass(self->cls, complex_cls))
+        raiseExcHelper(TypeError, "descriptor '__hash__' requires a 'complex' object but received a '%s'",
+                       getTypeName(self));
+    long hashreal, hashimag, combined;
+
+    hashreal = _Py_HashDouble(self->real);
+    if (hashreal == -1) {
+        throwCAPIException();
+    }
+
+    hashimag = _Py_HashDouble(self->imag);
+    if (hashimag == -1) {
+        throwCAPIException();
+    }
+    /* Note:  if the imaginary part is 0, hashimag is 0 now,
+     * so the following returns hashreal unchanged.  This is
+     * important because numbers of different types that
+     * compare equal must have the same hash value, so that
+     * hash(x + 0*j) must equal hash(x).
+     */
+    combined = hashreal + 1000003 * hashimag;
+    if (combined == -1)
+        combined = -2;
+    return boxInt(combined);
+}
+
 Box* complexStr(BoxedComplex* self) {
     assert(self->cls == complex_cls);
     return boxString(complexFmt(self->real, self->imag, 12, 'g'));
@@ -336,6 +363,7 @@ void setupComplex() {
              (void*)complexDiv);
 
     complex_cls->giveAttr("__pos__", new BoxedFunction(boxRTFunction((void*)complexPos, BOXED_COMPLEX, 1)));
+    complex_cls->giveAttr("__hash__", new BoxedFunction(boxRTFunction((void*)complexHash, BOXED_INT, 1)));
     complex_cls->giveAttr("__str__", new BoxedFunction(boxRTFunction((void*)complexStr, STR, 1)));
     complex_cls->giveAttr("__repr__", new BoxedFunction(boxRTFunction((void*)complexRepr, STR, 1)));
     complex_cls->giveAttr("real",

src/runtime/long.cpp

@@ -661,7 +661,7 @@ BoxedLong* _longNew(Box* val, Box* _base) {
             int r = mpz_init_set_str(rtn->n, s.data(), 10);
             RELEASE_ASSERT(r == 0, "");
         } else if (val->cls == float_cls) {
-            mpz_init_set_si(rtn->n, static_cast<BoxedFloat*>(val)->d);
+            mpz_init_set_d(rtn->n, static_cast<BoxedFloat*>(val)->d);
         } else {
             static BoxedString* long_str = internStringImmortal("__long__");
             CallattrFlags callattr_flags{.cls_only = true, .null_on_nonexistent = true, .argspec = ArgPassSpec(0) };

src/runtime/str.cpp

@@ -1526,6 +1526,10 @@ extern "C" size_t unicodeHashUnboxed(PyUnicodeObject* self) {
         return self->hash;
 
     Py_ssize_t len = PyUnicode_GET_SIZE(self);
+
+    if (len == 0)
+        return 0;
+
     Py_UNICODE* p = PyUnicode_AS_UNICODE(self);
     pyston::StringHash<Py_UNICODE> H;
     return H(p, len);
@@ -1534,6 +1538,10 @@ extern "C" size_t unicodeHashUnboxed(PyUnicodeObject* self) {
 extern "C" Box* strHash(BoxedString* self) {
     assert(PyString_Check(self));
 
+    // CPython set the hash empty string to 0 manually
+    if (self->size() == 0)
+        return boxInt(0);
+
     StringHash<char> H;
     return boxInt(H(self->data(), self->size()));
 }