def _test(self, lst):
exp = list(group_by_simple(lst))
def _(x):
return list(map(list_value, x))
nit1 = GroupByNestedIter(iter(lst))
res1 = _(nit1)
self.assertEqual(exp, res1, res1)
self.assertFalse(nit1.is_cached)
self.assertRaises(StopIteration, next, nit1)
nit2 = GroupByNestedIter(iter(lst))
res2 = list(nit2)
res2 = _(res2)
self.assertEqual(exp, res2, res2)
self.assertRaises(StopIteration, next, nit2)
if len(lst) > 1:
self.assertTrue(nit2.is_cached, res2)
if lst:
nit3 = GroupByNestedIter(iter(lst))
k1, vit1 = next(nit3)
res3 = _(nit3)
vs1 = list(vit1)
self.assertRaises(StopIteration, next, vit1)
res3 = [(k1, vs1)] + res3
self.assertEqual(exp, res3, res3)
def __iter__(self):
if not self.archives:
return six.iteritems(self.combined)
items = self.combined.items()
items.sort(key=itemgetter(0))
combined = items
self.archives.append(iter(combined))
iters = list(map(iter, self.archives))
if self.rddconf.is_groupby and self.rddconf.iter_group:
heap = HeapOnKey(key=lambda x: x[0], min_heap=True)
it = GroupByNestedIter(heap.merge(iters), "")
else:
it = heap_merged(iters, self._get_merge_function())
return it
def _merge_sorted(self, iters):
heap = HeapOnKey(key=lambda x: x[0], min_heap=True)
return GroupByNestedIter(heap.merge(iters), self.api_callsite)
def _(i):
v = list(range(i, 10, 3))
lst = [('k1', v)] * 2 + [('k2', list(v))]
return GroupByNestedIter(lst)