Merge pull request #1772 from palaviv/set-fixes

Set fixes

Lib/test/test_set.py

@@ -172,7 +172,6 @@ class TestJointOps:
         else:
             self.fail("s-t did not screen-out general iterables")
 
-    @unittest.skip("TODO: RUSTPYTHON")
     def test_symmetric_difference(self):
         i = self.s.symmetric_difference(self.otherword)
         for c in self.letters:
@@ -187,7 +186,6 @@ class TestJointOps:
             self.assertEqual(self.thetype('abcba').symmetric_difference(C('ccb')), set('a'))
             self.assertEqual(self.thetype('abcba').symmetric_difference(C('ef')), set('abcef'))
 
-    @unittest.skip("TODO: RUSTPYTHON")
     def test_xor(self):
         i = self.s.symmetric_difference(self.otherword)
         self.assertEqual(self.s ^ set(self.otherword), i)
@@ -207,7 +205,6 @@ class TestJointOps:
         self.assertNotEqual(self.s, frozenset(self.otherword))
         self.assertEqual(self.s != self.word, True)
 
-    @unittest.skip("TODO: RUSTPYTHON")
     def test_setOfFrozensets(self):
         t = map(frozenset, ['abcdef', 'bcd', 'bdcb', 'fed', 'fedccba'])
         s = self.thetype(t)
@@ -515,7 +512,6 @@ class TestSet(TestJointOps, unittest.TestCase):
         for c in (self.word + self.otherword):
             self.assertIn(c, self.s)
 
-    @unittest.skip("TODO: RUSTPYTHON")
     def test_intersection_update(self):
         retval = self.s.intersection_update(self.otherword)
         self.assertEqual(retval, None)
@@ -582,7 +578,6 @@ class TestSet(TestJointOps, unittest.TestCase):
             else:
                 self.assertNotIn(c, self.s)
 
-    @unittest.skip("TODO: RUSTPYTHON")
     def test_symmetric_difference_update(self):
         retval = self.s.symmetric_difference_update(self.otherword)
         self.assertEqual(retval, None)
@@ -713,7 +708,6 @@ class TestFrozenSet(TestJointOps, unittest.TestCase):
         t = self.thetype(s)
         self.assertEqual(id(s), id(t))
 
-    @unittest.skip("TODO: RUSTPYTHON")
     def test_hash(self):
         self.assertEqual(hash(self.thetype('abcdeb')),
                          hash(self.thetype('ebecda')))
@@ -732,7 +726,6 @@ class TestFrozenSet(TestJointOps, unittest.TestCase):
         dup = self.s.copy()
         self.assertEqual(id(self.s), id(dup))
 
-    @unittest.skip("TODO: RUSTPYTHON")
     def test_frozen_as_dictkey(self):
         seq = list(range(10)) + list('abcdefg') + ['apple']
         key1 = self.thetype(seq)

vm/src/obj/objset.rs

@@ -9,6 +9,7 @@ use super::objlist::PyListIterator;
 use super::objtype::{self, PyClassRef};
 use crate::dictdatatype;
 use crate::function::{Args, OptionalArg};
+use crate::pyhash;
 use crate::pyobject::{
     PyClassImpl, PyContext, PyIterable, PyObjectRef, PyRef, PyResult, PyValue, TryFromObject,
     TypeProtocol,
@@ -206,8 +207,11 @@ impl PySetInner {
     fn symmetric_difference(&self, other: PyIterable, vm: &VirtualMachine) -> PyResult<PySetInner> {
         let mut new_inner = self.clone();
 
-        for item in other.iter(vm)? {
-            new_inner.content.delete_or_insert(vm, &item?, ())?
+        // We want to remove duplicates in other
+        let other_set = Self::new(other, vm)?;
+
+        for item in other_set.content.keys() {
+            new_inner.content.delete_or_insert(vm, &item, ())?
         }
 
         Ok(new_inner)
@@ -294,7 +298,7 @@ impl PySetInner {
         others: Args<PyIterable>,
         vm: &VirtualMachine,
     ) -> PyResult<()> {
-        let temp_inner = self.copy();
+        let mut temp_inner = self.copy();
         self.clear();
         for iterable in others {
             for item in iterable.iter(vm)? {
@@ -303,6 +307,7 @@ impl PySetInner {
                     self.add(&obj, vm)?;
                 }
             }
+            temp_inner = self.copy()
         }
         Ok(())
     }
@@ -322,12 +327,18 @@ impl PySetInner {
         vm: &VirtualMachine,
     ) -> PyResult<()> {
         for iterable in others {
-            for item in iterable.iter(vm)? {
-                self.content.delete_or_insert(vm, &item?, ())?;
+            // We want to remove duplicates in iterable
+            let iterable_set = Self::new(iterable, vm)?;
+            for item in iterable_set.content.keys() {
+                self.content.delete_or_insert(vm, &item, ())?;
             }
         }
         Ok(())
     }
+
+    fn hash(&self, vm: &VirtualMachine) -> PyResult<pyhash::PyHash> {
+        pyhash::hash_iter_unordered(self.content.keys(), vm)
+    }
 }
 
 macro_rules! try_set_cmp {
@@ -780,6 +791,11 @@ impl PyFrozenSet {
         };
         Ok(vm.new_str(s))
     }
+
+    #[pymethod(name = "__hash__")]
+    fn hash(&self, vm: &VirtualMachine) -> PyResult<pyhash::PyHash> {
+        self.inner.hash(vm)
+    }
 }
 
 struct SetIterable {

vm/src/pyhash.rs

@@ -84,6 +84,19 @@ pub fn hash_iter<'a, I: std::iter::Iterator<Item = &'a PyObjectRef>>(
     Ok(hasher.finish() as PyHash)
 }
 
+pub fn hash_iter_unordered<I: std::iter::Iterator<Item = PyObjectRef>>(
+    iter: I,
+    vm: &VirtualMachine,
+) -> PyResult<PyHash> {
+    let mut hash: PyHash = 0;
+    for element in iter {
+        let item_hash = vm._hash(&element)?;
+        // xor is commutative and hash should be independent of order
+        hash ^= item_hash;
+    }
+    Ok(hash)
+}
+
 pub fn hash_bigint(value: &BigInt) -> PyHash {
     match value.to_i64() {
         Some(i64_value) => (i64_value % MODULUS as i64),