def test_split_file(tmp_path: Path):
file = tmp_path / "test.txt"
content = "Hello\nWorld\n"
with open(file, "w") as o:
o.write(content)
with jsonql.SplitFile(file, chunk=0, n_chunks=2) as f:
assert f.readlines() == ["Hello\n"]
with jsonql.SplitFile(file, chunk=1, n_chunks=2) as f:
assert f.readlines() == ["World\n"]
def test_split_file(self):
tmp = self.get_tmpdir()
file = tmp("test.txt")
content = "Hello\nWorld\n"
with open(file, "w") as o:
o.write(content)
with jsonql.SplitFile(file, chunk=0, n_chunks=2) as f:
self.assertEqual(f.readlines(), ["Hello\n"])
with jsonql.SplitFile(file, chunk=1, n_chunks=2) as f:
self.assertEqual(f.readlines(), ["World\n"])
def test_split_file_middle_of_char(tmp_path: Path):
file = tmp_path / "test.txt"
content = "Hello\U0001F40D\nWorld\n"
# split is here ^^
with open(file, "w") as o:
o.write(content)
with jsonql.SplitFile(file, chunk=0, n_chunks=2) as f:
assert f.readlines() == ["Hello🐍\n"]
with jsonql.SplitFile(file, chunk=1, n_chunks=2) as f:
assert f.readlines() == ["World\n"]
def test_split_file_middle_of_char(self):
tmp = self.get_tmpdir()
file = tmp("test.txt")
content = "Hello\U0001F40D\nWorld\n"
# split is here ^^
with open(file, "w") as o:
o.write(content)
with jsonql.SplitFile(file, chunk=0, n_chunks=2) as f:
self.assertEqual(f.readlines(), ["Hello🐍\n"])
with jsonql.SplitFile(file, chunk=1, n_chunks=2) as f:
self.assertEqual(f.readlines(), ["World\n"])
def split_file(file, n_splits):
for i in range(n_splits):
yield jsonql.SplitFile(file, i, n_splits)