Path: blob/master/venv/Lib/site-packages/bs4/tests/test_builder_registry.py
811 views
"""Tests of the builder registry."""12import unittest3import warnings45from bs4 import BeautifulSoup6from bs4.builder import (7builder_registry as registry,8HTMLParserTreeBuilder,9TreeBuilderRegistry,10)1112try:13from bs4.builder import HTML5TreeBuilder14HTML5LIB_PRESENT = True15except ImportError:16HTML5LIB_PRESENT = False1718try:19from bs4.builder import (20LXMLTreeBuilderForXML,21LXMLTreeBuilder,22)23LXML_PRESENT = True24except ImportError:25LXML_PRESENT = False262728class BuiltInRegistryTest(unittest.TestCase):29"""Test the built-in registry with the default builders registered."""3031def test_combination(self):32if LXML_PRESENT:33self.assertEqual(registry.lookup('fast', 'html'),34LXMLTreeBuilder)3536if LXML_PRESENT:37self.assertEqual(registry.lookup('permissive', 'xml'),38LXMLTreeBuilderForXML)39self.assertEqual(registry.lookup('strict', 'html'),40HTMLParserTreeBuilder)41if HTML5LIB_PRESENT:42self.assertEqual(registry.lookup('html5lib', 'html'),43HTML5TreeBuilder)4445def test_lookup_by_markup_type(self):46if LXML_PRESENT:47self.assertEqual(registry.lookup('html'), LXMLTreeBuilder)48self.assertEqual(registry.lookup('xml'), LXMLTreeBuilderForXML)49else:50self.assertEqual(registry.lookup('xml'), None)51if HTML5LIB_PRESENT:52self.assertEqual(registry.lookup('html'), HTML5TreeBuilder)53else:54self.assertEqual(registry.lookup('html'), HTMLParserTreeBuilder)5556def test_named_library(self):57if LXML_PRESENT:58self.assertEqual(registry.lookup('lxml', 'xml'),59LXMLTreeBuilderForXML)60self.assertEqual(registry.lookup('lxml', 'html'),61LXMLTreeBuilder)62if HTML5LIB_PRESENT:63self.assertEqual(registry.lookup('html5lib'),64HTML5TreeBuilder)6566self.assertEqual(registry.lookup('html.parser'),67HTMLParserTreeBuilder)6869def test_beautifulsoup_constructor_does_lookup(self):7071with warnings.catch_warnings(record=True) as w:72# This will create a warning about not explicitly73# specifying a parser, but we'll ignore it.7475# You can pass in a string.76BeautifulSoup("", features="html")77# Or a list of strings.78BeautifulSoup("", features=["html", "fast"])7980# You'll get an exception if BS can't find an appropriate81# builder.82self.assertRaises(ValueError, BeautifulSoup,83"", features="no-such-feature")8485class RegistryTest(unittest.TestCase):86"""Test the TreeBuilderRegistry class in general."""8788def setUp(self):89self.registry = TreeBuilderRegistry()9091def builder_for_features(self, *feature_list):92cls = type('Builder_' + '_'.join(feature_list),93(object,), {'features' : feature_list})9495self.registry.register(cls)96return cls9798def test_register_with_no_features(self):99builder = self.builder_for_features()100101# Since the builder advertises no features, you can't find it102# by looking up features.103self.assertEqual(self.registry.lookup('foo'), None)104105# But you can find it by doing a lookup with no features, if106# this happens to be the only registered builder.107self.assertEqual(self.registry.lookup(), builder)108109def test_register_with_features_makes_lookup_succeed(self):110builder = self.builder_for_features('foo', 'bar')111self.assertEqual(self.registry.lookup('foo'), builder)112self.assertEqual(self.registry.lookup('bar'), builder)113114def test_lookup_fails_when_no_builder_implements_feature(self):115builder = self.builder_for_features('foo', 'bar')116self.assertEqual(self.registry.lookup('baz'), None)117118def test_lookup_gets_most_recent_registration_when_no_feature_specified(self):119builder1 = self.builder_for_features('foo')120builder2 = self.builder_for_features('bar')121self.assertEqual(self.registry.lookup(), builder2)122123def test_lookup_fails_when_no_tree_builders_registered(self):124self.assertEqual(self.registry.lookup(), None)125126def test_lookup_gets_most_recent_builder_supporting_all_features(self):127has_one = self.builder_for_features('foo')128has_the_other = self.builder_for_features('bar')129has_both_early = self.builder_for_features('foo', 'bar', 'baz')130has_both_late = self.builder_for_features('foo', 'bar', 'quux')131lacks_one = self.builder_for_features('bar')132has_the_other = self.builder_for_features('foo')133134# There are two builders featuring 'foo' and 'bar', but135# the one that also features 'quux' was registered later.136self.assertEqual(self.registry.lookup('foo', 'bar'),137has_both_late)138139# There is only one builder featuring 'foo', 'bar', and 'baz'.140self.assertEqual(self.registry.lookup('foo', 'bar', 'baz'),141has_both_early)142143def test_lookup_fails_when_cannot_reconcile_requested_features(self):144builder1 = self.builder_for_features('foo', 'bar')145builder2 = self.builder_for_features('foo', 'baz')146self.assertEqual(self.registry.lookup('bar', 'baz'), None)147148149