|
| 1 | +from datetime import date, datetime |
| 2 | +import unittest, pytest |
| 3 | + |
| 4 | +from databricks.sql.utils import ParamEscaper, inject_parameters |
| 5 | + |
| 6 | +pe = ParamEscaper() |
| 7 | + |
| 8 | +class TestIndividualFormatters(object): |
| 9 | + |
| 10 | + # Test individual type escapers |
| 11 | + def test_escape_number_integer(self): |
| 12 | + """This behaviour falls back to Python's default string formatting of numbers |
| 13 | + """ |
| 14 | + assert pe.escape_number(100) == 100 |
| 15 | + |
| 16 | + def test_escape_number_float(self): |
| 17 | + """This behaviour falls back to Python's default string formatting of numbers |
| 18 | + """ |
| 19 | + assert pe.escape_number(100.1234) == 100.1234 |
| 20 | + |
| 21 | + def test_escape_string_normal(self): |
| 22 | + """ |
| 23 | + """ |
| 24 | + |
| 25 | + assert pe.escape_string("golly bob howdy") == "'golly bob howdy'" |
| 26 | + |
| 27 | + def test_escape_string_that_includes_special_characters(self): |
| 28 | + """Tests for how special characters are treated. |
| 29 | +
|
| 30 | + When passed a string, the `escape_string` method wraps it in single quotes |
| 31 | + and escapes any special characters with a back stroke (\) |
| 32 | +
|
| 33 | + Example: |
| 34 | +
|
| 35 | + IN : his name was 'robert palmer' |
| 36 | + OUT: 'his name was \'robert palmer\'' |
| 37 | + """ |
| 38 | + |
| 39 | + # Testing for the presence of these characters: '"/\😂 |
| 40 | + |
| 41 | + assert pe.escape_string("his name was 'robert palmer'") == r"'his name was \'robert palmer\''" |
| 42 | + |
| 43 | + # These tests represent the same user input in the several ways it can be written in Python |
| 44 | + # Each argument to `escape_string` evaluates to the same bytes. But Python lets us write it differently. |
| 45 | + assert pe.escape_string("his name was \"robert palmer\"") == "'his name was \"robert palmer\"'" |
| 46 | + assert pe.escape_string('his name was "robert palmer"') == "'his name was \"robert palmer\"'" |
| 47 | + assert pe.escape_string('his name was {}'.format('"robert palmer"')) == "'his name was \"robert palmer\"'" |
| 48 | + |
| 49 | + assert pe.escape_string("his name was robert / palmer") == r"'his name was robert / palmer'" |
| 50 | + |
| 51 | + # If you need to include a single backslash, use an r-string to prevent Python from raising a |
| 52 | + # DeprecationWarning for an invalid escape sequence |
| 53 | + assert pe.escape_string("his name was robert \\/ palmer") == r"'his name was robert \\/ palmer'" |
| 54 | + assert pe.escape_string("his name was robert \\ palmer") == r"'his name was robert \\ palmer'" |
| 55 | + assert pe.escape_string("his name was robert \\\\ palmer") == r"'his name was robert \\\\ palmer'" |
| 56 | + |
| 57 | + assert pe.escape_string("his name was robert palmer 😂") == r"'his name was robert palmer 😂'" |
| 58 | + |
| 59 | + # Adding the test from PR #56 to prove escape behaviour |
| 60 | + |
| 61 | + assert pe.escape_string("you're") == r"'you\'re'" |
| 62 | + |
| 63 | + # Adding this test from #51 to prove escape behaviour when the target string involves repeated SQL escape chars |
| 64 | + assert pe.escape_string("cat\\'s meow") == r"'cat\\\'s meow'" |
| 65 | + |
| 66 | + # Tests from the docs: https://docs.databricks.com/sql/language-manual/data-types/string-type.html |
| 67 | + |
| 68 | + assert pe.escape_string('Spark') == "'Spark'" |
| 69 | + assert pe.escape_string("O'Connell") == r"'O\'Connell'" |
| 70 | + assert pe.escape_string("Some\\nText") == r"'Some\\nText'" |
| 71 | + assert pe.escape_string("Some\\\\nText") == r"'Some\\\\nText'" |
| 72 | + assert pe.escape_string("서울시") == "'서울시'" |
| 73 | + assert pe.escape_string("\\\\") == r"'\\\\'" |
| 74 | + |
| 75 | + def test_escape_date_time(self): |
| 76 | + INPUT = datetime(1991,8,3,21,55) |
| 77 | + FORMAT = "%Y-%m-%d %H:%M:%S" |
| 78 | + OUTPUT = "'1991-08-03 21:55:00'" |
| 79 | + assert pe.escape_datetime(INPUT, FORMAT) == OUTPUT |
| 80 | + |
| 81 | + def test_escape_date(self): |
| 82 | + INPUT = date(1991,8,3) |
| 83 | + FORMAT = "%Y-%m-%d" |
| 84 | + OUTPUT = "'1991-08-03'" |
| 85 | + assert pe.escape_datetime(INPUT, FORMAT) == OUTPUT |
| 86 | + |
| 87 | + def test_escape_sequence_integer(self): |
| 88 | + assert pe.escape_sequence([1,2,3,4]) == "(1,2,3,4)" |
| 89 | + |
| 90 | + def test_escape_sequence_float(self): |
| 91 | + assert pe.escape_sequence([1.1,2.2,3.3,4.4]) == "(1.1,2.2,3.3,4.4)" |
| 92 | + |
| 93 | + def test_escape_sequence_string(self): |
| 94 | + assert pe.escape_sequence( |
| 95 | + ["his", "name", "was", "robert", "palmer"]) == \ |
| 96 | + "('his','name','was','robert','palmer')" |
| 97 | + |
| 98 | + def test_escape_sequence_sequence_of_strings(self): |
| 99 | + # This is not valid SQL. |
| 100 | + INPUT = [["his", "name"], ["was", "robert"], ["palmer"]] |
| 101 | + OUTPUT = "(('his','name'),('was','robert'),('palmer'))" |
| 102 | + |
| 103 | + assert pe.escape_sequence(INPUT) == OUTPUT |
| 104 | + |
| 105 | + |
| 106 | +class TestFullQueryEscaping(object): |
| 107 | + |
| 108 | + def test_simple(self): |
| 109 | + |
| 110 | + INPUT = """ |
| 111 | + SELECT |
| 112 | + field1, |
| 113 | + field2, |
| 114 | + field3 |
| 115 | + FROM |
| 116 | + table |
| 117 | + WHERE |
| 118 | + field1 = %(param1)s |
| 119 | + """ |
| 120 | + |
| 121 | + OUTPUT = """ |
| 122 | + SELECT |
| 123 | + field1, |
| 124 | + field2, |
| 125 | + field3 |
| 126 | + FROM |
| 127 | + table |
| 128 | + WHERE |
| 129 | + field1 = ';DROP ALL TABLES' |
| 130 | + """ |
| 131 | + |
| 132 | + args = {"param1": ";DROP ALL TABLES"} |
| 133 | + |
| 134 | + assert inject_parameters(INPUT, pe.escape_args(args)) == OUTPUT |
| 135 | + |
| 136 | + @unittest.skipUnless(False, "Thrift server supports native parameter binding.") |
| 137 | + def test_only_bind_in_where_clause(self): |
| 138 | + |
| 139 | + INPUT = """ |
| 140 | + SELECT |
| 141 | + %(field)s, |
| 142 | + field2, |
| 143 | + field3 |
| 144 | + FROM table |
| 145 | + """ |
| 146 | + |
| 147 | + args = {"field": "Some Value"} |
| 148 | + |
| 149 | + with pytest.raises(Exception): |
| 150 | + inject_parameters(INPUT, pe.escape_args(args)) |
0 commit comments