Patrick Williams | c124f4f | 2015-09-15 14:41:29 -0500 | [diff] [blame] | 1 | """Tests of the builder registry.""" |
| 2 | |
| 3 | import unittest |
Patrick Williams | c0f7c04 | 2017-02-23 20:41:17 -0600 | [diff] [blame] | 4 | import warnings |
Patrick Williams | c124f4f | 2015-09-15 14:41:29 -0500 | [diff] [blame] | 5 | |
| 6 | from bs4 import BeautifulSoup |
| 7 | from bs4.builder import ( |
| 8 | builder_registry as registry, |
| 9 | HTMLParserTreeBuilder, |
| 10 | TreeBuilderRegistry, |
| 11 | ) |
| 12 | |
| 13 | try: |
| 14 | from bs4.builder import HTML5TreeBuilder |
| 15 | HTML5LIB_PRESENT = True |
| 16 | except ImportError: |
| 17 | HTML5LIB_PRESENT = False |
| 18 | |
| 19 | try: |
| 20 | from bs4.builder import ( |
| 21 | LXMLTreeBuilderForXML, |
| 22 | LXMLTreeBuilder, |
| 23 | ) |
| 24 | LXML_PRESENT = True |
| 25 | except ImportError: |
| 26 | LXML_PRESENT = False |
| 27 | |
| 28 | |
| 29 | class BuiltInRegistryTest(unittest.TestCase): |
| 30 | """Test the built-in registry with the default builders registered.""" |
| 31 | |
| 32 | def test_combination(self): |
| 33 | if LXML_PRESENT: |
| 34 | self.assertEqual(registry.lookup('fast', 'html'), |
| 35 | LXMLTreeBuilder) |
| 36 | |
| 37 | if LXML_PRESENT: |
| 38 | self.assertEqual(registry.lookup('permissive', 'xml'), |
| 39 | LXMLTreeBuilderForXML) |
| 40 | self.assertEqual(registry.lookup('strict', 'html'), |
| 41 | HTMLParserTreeBuilder) |
| 42 | if HTML5LIB_PRESENT: |
| 43 | self.assertEqual(registry.lookup('html5lib', 'html'), |
| 44 | HTML5TreeBuilder) |
| 45 | |
| 46 | def test_lookup_by_markup_type(self): |
| 47 | if LXML_PRESENT: |
| 48 | self.assertEqual(registry.lookup('html'), LXMLTreeBuilder) |
| 49 | self.assertEqual(registry.lookup('xml'), LXMLTreeBuilderForXML) |
| 50 | else: |
| 51 | self.assertEqual(registry.lookup('xml'), None) |
| 52 | if HTML5LIB_PRESENT: |
| 53 | self.assertEqual(registry.lookup('html'), HTML5TreeBuilder) |
| 54 | else: |
| 55 | self.assertEqual(registry.lookup('html'), HTMLParserTreeBuilder) |
| 56 | |
| 57 | def test_named_library(self): |
| 58 | if LXML_PRESENT: |
| 59 | self.assertEqual(registry.lookup('lxml', 'xml'), |
| 60 | LXMLTreeBuilderForXML) |
| 61 | self.assertEqual(registry.lookup('lxml', 'html'), |
| 62 | LXMLTreeBuilder) |
| 63 | if HTML5LIB_PRESENT: |
| 64 | self.assertEqual(registry.lookup('html5lib'), |
| 65 | HTML5TreeBuilder) |
| 66 | |
| 67 | self.assertEqual(registry.lookup('html.parser'), |
| 68 | HTMLParserTreeBuilder) |
| 69 | |
| 70 | def test_beautifulsoup_constructor_does_lookup(self): |
Patrick Williams | c0f7c04 | 2017-02-23 20:41:17 -0600 | [diff] [blame] | 71 | |
| 72 | with warnings.catch_warnings(record=True) as w: |
| 73 | # This will create a warning about not explicitly |
| 74 | # specifying a parser, but we'll ignore it. |
| 75 | |
| 76 | # You can pass in a string. |
| 77 | BeautifulSoup("", features="html") |
| 78 | # Or a list of strings. |
| 79 | BeautifulSoup("", features=["html", "fast"]) |
Patrick Williams | c124f4f | 2015-09-15 14:41:29 -0500 | [diff] [blame] | 80 | |
| 81 | # You'll get an exception if BS can't find an appropriate |
| 82 | # builder. |
| 83 | self.assertRaises(ValueError, BeautifulSoup, |
| 84 | "", features="no-such-feature") |
| 85 | |
| 86 | class RegistryTest(unittest.TestCase): |
| 87 | """Test the TreeBuilderRegistry class in general.""" |
| 88 | |
| 89 | def setUp(self): |
| 90 | self.registry = TreeBuilderRegistry() |
| 91 | |
| 92 | def builder_for_features(self, *feature_list): |
| 93 | cls = type('Builder_' + '_'.join(feature_list), |
| 94 | (object,), {'features' : feature_list}) |
| 95 | |
| 96 | self.registry.register(cls) |
| 97 | return cls |
| 98 | |
| 99 | def test_register_with_no_features(self): |
| 100 | builder = self.builder_for_features() |
| 101 | |
| 102 | # Since the builder advertises no features, you can't find it |
| 103 | # by looking up features. |
| 104 | self.assertEqual(self.registry.lookup('foo'), None) |
| 105 | |
| 106 | # But you can find it by doing a lookup with no features, if |
| 107 | # this happens to be the only registered builder. |
| 108 | self.assertEqual(self.registry.lookup(), builder) |
| 109 | |
| 110 | def test_register_with_features_makes_lookup_succeed(self): |
| 111 | builder = self.builder_for_features('foo', 'bar') |
| 112 | self.assertEqual(self.registry.lookup('foo'), builder) |
| 113 | self.assertEqual(self.registry.lookup('bar'), builder) |
| 114 | |
| 115 | def test_lookup_fails_when_no_builder_implements_feature(self): |
| 116 | builder = self.builder_for_features('foo', 'bar') |
| 117 | self.assertEqual(self.registry.lookup('baz'), None) |
| 118 | |
| 119 | def test_lookup_gets_most_recent_registration_when_no_feature_specified(self): |
| 120 | builder1 = self.builder_for_features('foo') |
| 121 | builder2 = self.builder_for_features('bar') |
| 122 | self.assertEqual(self.registry.lookup(), builder2) |
| 123 | |
| 124 | def test_lookup_fails_when_no_tree_builders_registered(self): |
| 125 | self.assertEqual(self.registry.lookup(), None) |
| 126 | |
| 127 | def test_lookup_gets_most_recent_builder_supporting_all_features(self): |
| 128 | has_one = self.builder_for_features('foo') |
| 129 | has_the_other = self.builder_for_features('bar') |
| 130 | has_both_early = self.builder_for_features('foo', 'bar', 'baz') |
| 131 | has_both_late = self.builder_for_features('foo', 'bar', 'quux') |
| 132 | lacks_one = self.builder_for_features('bar') |
| 133 | has_the_other = self.builder_for_features('foo') |
| 134 | |
| 135 | # There are two builders featuring 'foo' and 'bar', but |
| 136 | # the one that also features 'quux' was registered later. |
| 137 | self.assertEqual(self.registry.lookup('foo', 'bar'), |
| 138 | has_both_late) |
| 139 | |
| 140 | # There is only one builder featuring 'foo', 'bar', and 'baz'. |
| 141 | self.assertEqual(self.registry.lookup('foo', 'bar', 'baz'), |
| 142 | has_both_early) |
| 143 | |
| 144 | def test_lookup_fails_when_cannot_reconcile_requested_features(self): |
| 145 | builder1 = self.builder_for_features('foo', 'bar') |
| 146 | builder2 = self.builder_for_features('foo', 'baz') |
| 147 | self.assertEqual(self.registry.lookup('bar', 'baz'), None) |