|
| 1 | +import pytest |
| 2 | +from core_codemods.lxml_safe_parsing import LxmlSafeParsing |
| 3 | +from tests.codemods.base_codemod_test import BaseSemgrepCodemodTest |
| 4 | + |
| 5 | +each_func = pytest.mark.parametrize("func", ["parse", "fromstring"]) |
| 6 | + |
| 7 | + |
| 8 | +class TestLxmlSafeParsing(BaseSemgrepCodemodTest): |
| 9 | + codemod = LxmlSafeParsing |
| 10 | + |
| 11 | + def test_name(self): |
| 12 | + assert self.codemod.name() == "safe-lxml-parsing" |
| 13 | + |
| 14 | + @each_func |
| 15 | + def test_import(self, tmpdir, func): |
| 16 | + input_code = f"""import lxml.etree |
| 17 | +
|
| 18 | +lxml.etree.{func}("path_to_file") |
| 19 | +var = "hello" |
| 20 | +""" |
| 21 | + expexted_output = f"""import lxml.etree |
| 22 | +
|
| 23 | +lxml.etree.{func}("path_to_file", parser=lxml.etree.XMLParser(resolve_entities=False)) |
| 24 | +var = "hello" |
| 25 | +""" |
| 26 | + |
| 27 | + self.run_and_assert(tmpdir, input_code, expexted_output) |
| 28 | + |
| 29 | + @each_func |
| 30 | + def test_from_import(self, tmpdir, func): |
| 31 | + input_code = f"""from lxml.etree import {func} |
| 32 | +
|
| 33 | +{func}("path_to_file") |
| 34 | +var = "hello" |
| 35 | +""" |
| 36 | + expexted_output = f"""from lxml.etree import {func} |
| 37 | +import lxml.etree |
| 38 | +
|
| 39 | +{func}("path_to_file", parser=lxml.etree.XMLParser(resolve_entities=False)) |
| 40 | +var = "hello" |
| 41 | +""" |
| 42 | + |
| 43 | + self.run_and_assert(tmpdir, input_code, expexted_output) |
| 44 | + |
| 45 | + @each_func |
| 46 | + def test_from_import_module(self, tmpdir, func): |
| 47 | + input_code = f"""from lxml import etree |
| 48 | +
|
| 49 | +etree.{func}("path_to_file") |
| 50 | +var = "hello" |
| 51 | +""" |
| 52 | + expexted_output = f"""from lxml import etree |
| 53 | +import lxml.etree |
| 54 | +
|
| 55 | +etree.{func}("path_to_file", parser=lxml.etree.XMLParser(resolve_entities=False)) |
| 56 | +var = "hello" |
| 57 | +""" |
| 58 | + |
| 59 | + self.run_and_assert(tmpdir, input_code, expexted_output) |
| 60 | + |
| 61 | + @each_func |
| 62 | + def test_import_alias(self, tmpdir, func): |
| 63 | + input_code = f"""from lxml.etree import {func} as func |
| 64 | +
|
| 65 | +func("path_to_file") |
| 66 | +var = "hello" |
| 67 | +""" |
| 68 | + expexted_output = f"""from lxml.etree import {func} as func |
| 69 | +import lxml.etree |
| 70 | +
|
| 71 | +func("path_to_file", parser=lxml.etree.XMLParser(resolve_entities=False)) |
| 72 | +var = "hello" |
| 73 | +""" |
| 74 | + |
| 75 | + self.run_and_assert(tmpdir, input_code, expexted_output) |
| 76 | + |
| 77 | + @pytest.mark.parametrize( |
| 78 | + "input_args,expected_args", |
| 79 | + [ |
| 80 | + ( |
| 81 | + "'str', parser=None", |
| 82 | + "'str', parser=lxml.etree.XMLParser(resolve_entities=False)", |
| 83 | + ), |
| 84 | + ( |
| 85 | + "source, base_url='url', parser=None", |
| 86 | + "source, base_url='url', parser=lxml.etree.XMLParser(resolve_entities=False)", |
| 87 | + ), |
| 88 | + ( |
| 89 | + "source, parser=lxml.etree.XMLParser(resolve_entities=False)", |
| 90 | + "source, parser=lxml.etree.XMLParser(resolve_entities=False)", |
| 91 | + ), |
| 92 | + # This case would be changed by `safe-lxml-parser-defaults` codemod |
| 93 | + ( |
| 94 | + "source, parser=lxml.etree.XMLParser()", |
| 95 | + "source, parser=lxml.etree.XMLParser()", |
| 96 | + ), |
| 97 | + ], |
| 98 | + ) |
| 99 | + @each_func |
| 100 | + def test_verify_variations(self, tmpdir, func, input_args, expected_args): |
| 101 | + input_code = f"""import lxml.etree |
| 102 | +lxml.etree.{func}({input_args}) |
| 103 | +var = "hello" |
| 104 | +""" |
| 105 | + expexted_output = f"""import lxml.etree |
| 106 | +lxml.etree.{func}({expected_args}) |
| 107 | +var = "hello" |
| 108 | +""" |
| 109 | + self.run_and_assert(tmpdir, input_code, expexted_output) |
0 commit comments