Generalize tuple() to work nicely with iterators.

NEEDS DOC CHANGES.
This one surprised me!  While I expected tuple() to be a no-brainer, turns
out it's actually dripping with consequences:
1. It will *allow* the popular PySequence_Fast() to work with any iterable
   object (code for that not yet checked in, but should be trivial).
2. It caused two std tests to fail.  This because some places used
   PyTuple_Sequence() (the C spelling of tuple()) as an indirect way to test
   whether something *is* a sequence.  But tuple() code only looked for the
   existence of sq->item to determine that, and e.g. an instance passed
   that test whether or not it supported the other operations tuple()
   needed (e.g., __len__).  So some things the tests *expected* to fail
   with an AttributeError now fail with a TypeError instead.  This looks
   like an improvement to me; e.g., test_coercion used to produce 559
   TypeErrors and 2 AttributeErrors, and now they're all TypeErrors.  The
   error details are more informative too, because the places calling this
   were *looking* for TypeErrors in order to replace the generic tuple()
   "not a sequence" msg with their own more specific text, and
   AttributeErrors snuck by that.

Include/abstract.h

@@ -911,7 +911,7 @@ xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx*/
          tuple or list.  Use PySequence_Fast_GET_ITEM to access the
          members of this list.
 
-         Returns NULL on failure.  If the object is not a sequence,
+         Returns NULL on failure.  If the object does not support iteration,
          raises a TypeError exception with m as the message text.
        */
 

Lib/test/output/test_coercion

@@ -516,7 +516,7 @@ test_coercion
 [1] % None ... exceptions.TypeError
 [1] %= None ... exceptions.TypeError
 [1] + <MethodNumber 1> ... exceptions.TypeError
-[1] += <MethodNumber 1> ... exceptions.AttributeError
+[1] += <MethodNumber 1> ... exceptions.TypeError
 [1] - <MethodNumber 1> ... exceptions.TypeError
 [1] -= <MethodNumber 1> ... exceptions.TypeError
 [1] * <MethodNumber 1> = [1]
@@ -528,7 +528,7 @@ test_coercion
 [1] % <MethodNumber 1> ... exceptions.TypeError
 [1] %= <MethodNumber 1> ... exceptions.TypeError
 [1] + <CoerceNumber 2> ... exceptions.TypeError
-[1] += <CoerceNumber 2> ... exceptions.AttributeError
+[1] += <CoerceNumber 2> ... exceptions.TypeError
 [1] - <CoerceNumber 2> ... exceptions.TypeError
 [1] -= <CoerceNumber 2> ... exceptions.TypeError
 [1] * <CoerceNumber 2> = [1, 1]

Lib/test/test_extcall.py

@@ -58,20 +58,20 @@ g(1, 2, 3, *(4, 5))
 class Nothing: pass
 try:
     g(*Nothing())
-except AttributeError, attr:
+except TypeError, attr:
     pass
 else:
-    print "should raise AttributeError: __len__"
+    print "should raise TypeError"
 
 class Nothing:
     def __len__(self):
         return 5
 try:
     g(*Nothing())
-except AttributeError, attr:
+except TypeError, attr:
     pass
 else:
-    print "should raise AttributeError: __getitem__"
+    print "should raise TypeError"
 
 class Nothing:
     def __len__(self):

Lib/test/test_iter.py

@@ -275,6 +275,39 @@ class TestCase(unittest.TestCase):
             except OSError:
                 pass
 
+    # Test tuples()'s use of iterators.
+    def test_builtin_tuple(self):
+        self.assertEqual(tuple(SequenceClass(5)), (0, 1, 2, 3, 4))
+        self.assertEqual(tuple(SequenceClass(0)), ())
+        self.assertEqual(tuple([]), ())
+        self.assertEqual(tuple(()), ())
+        self.assertEqual(tuple("abc"), ("a", "b", "c"))
+
+        d = {"one": 1, "two": 2, "three": 3}
+        self.assertEqual(tuple(d), tuple(d.keys()))
+
+        self.assertRaises(TypeError, tuple, list)
+        self.assertRaises(TypeError, tuple, 42)
+
+        f = open(TESTFN, "w")
+        try:
+            for i in range(5):
+                f.write("%d\n" % i)
+        finally:
+            f.close()
+        f = open(TESTFN, "r")
+        try:
+            self.assertEqual(tuple(f), ("0\n", "1\n", "2\n", "3\n", "4\n"))
+            f.seek(0, 0)
+            self.assertEqual(tuple(f.xreadlines()),
+                             ("0\n", "1\n", "2\n", "3\n", "4\n"))
+        finally:
+            f.close()
+            try:
+                unlink(TESTFN)
+            except OSError:
+                pass
+
     # Test filter()'s use of iterators.
     def test_builtin_filter(self):
         self.assertEqual(filter(None, SequenceClass(5)), range(1, 5))

Misc/NEWS

@@ -24,9 +24,9 @@ Core
     min()
     reduce()
     XXX TODO string.join(), unicode.join()
-    XXX TODO tuple()
+    tuple()
     XXX TODO zip()
-    XXX TODO 'x in y' (!) (?)
+    XXX TODO 'x in y'
 
 What's New in Python 2.1 (final)?
 =================================

Objects/abstract.c

@@ -1176,61 +1176,68 @@ PySequence_DelSlice(PyObject *s, int i1, int i2)
 PyObject *
 PySequence_Tuple(PyObject *v)
 {
-	PySequenceMethods *m;
+	PyObject *it;  /* iter(v) */
+	int n;         /* guess for result tuple size */
+	PyObject *result;
+	int j;
 
 	if (v == NULL)
 		return null_error();
 
+	/* Special-case the common tuple and list cases, for efficiency. */
 	if (PyTuple_Check(v)) {
 		Py_INCREF(v);
 		return v;
 	}
-
 	if (PyList_Check(v))
 		return PyList_AsTuple(v);
 
-	/* There used to be code for strings here, but tuplifying strings is
-	   not a common activity, so I nuked it.  Down with code bloat! */
+	/* Get iterator. */
+	it = PyObject_GetIter(v);
+	if (it == NULL)
+		return type_error("tuple() argument must support iteration");
 
-	/* Generic sequence object */
-	m = v->ob_type->tp_as_sequence;
-	if (m && m->sq_item) {
-		int i;
-		PyObject *t;
-		int n = PySequence_Size(v);
-		if (n < 0)
-			return NULL;
-		t = PyTuple_New(n);
-		if (t == NULL)
-			return NULL;
-		for (i = 0; ; i++) {
-			PyObject *item = (*m->sq_item)(v, i);
-			if (item == NULL) {
-				if (PyErr_ExceptionMatches(PyExc_IndexError))
-					PyErr_Clear();
-				else {
-					Py_DECREF(t);
-					t = NULL;
-				}
-				break;
-			}
-			if (i >= n) {
-				if (n < 500)
-					n += 10;
-				else
-					n += 100;
-				if (_PyTuple_Resize(&t, n, 0) != 0)
-					break;
-			}
-			PyTuple_SET_ITEM(t, i, item);
+	/* Guess result size and allocate space. */
+	n = PySequence_Size(v);
+	if (n < 0) {
+		PyErr_Clear();
+		n = 10;  /* arbitrary */
+	}
+	result = PyTuple_New(n);
+	if (result == NULL)
+		goto Fail;
+
+	/* Fill the tuple. */
+	for (j = 0; ; ++j) {
+		PyObject *item = PyIter_Next(it);
+		if (item == NULL) {
+			if (PyErr_Occurred())
+				goto Fail;
+			break;
 		}
-		if (i < n && t != NULL)
-			_PyTuple_Resize(&t, i, 0);
-		return t;
+		if (j >= n) {
+			if (n < 500)
+				n += 10;
+			else
+				n += 100;
+			if (_PyTuple_Resize(&result, n, 0) != 0)
+				goto Fail;
+		}
+		PyTuple_SET_ITEM(result, j, item);
 	}
 
-	/* None of the above */
-	return type_error("tuple() argument must be a sequence");
+	/* Cut tuple back if guess was too large. */
+	if (j < n &&
+	    _PyTuple_Resize(&result, j, 0) != 0)
+		goto Fail;
+
+	Py_DECREF(it);
+	return result;
+
+Fail:
+	Py_XDECREF(result);
+	Py_DECREF(it);
+	return NULL;
 }
 
 PyObject *