From e4f967730af7d5ea59f6a05d03c779df16a24bd3 Mon Sep 17 00:00:00 2001 From: Leonard Richardson Date: Wed, 25 Jan 2023 14:36:30 -0500 Subject: Removed very old tests that were imported as part of the bzr import but not removed. --- tests/test_tree.py | 982 ----------------------------------------------------- 1 file changed, 982 deletions(-) delete mode 100644 tests/test_tree.py (limited to 'tests/test_tree.py') diff --git a/tests/test_tree.py b/tests/test_tree.py deleted file mode 100644 index 8d0027c..0000000 --- a/tests/test_tree.py +++ /dev/null @@ -1,982 +0,0 @@ -# -*- coding: utf-8 -*- -"""Tests for Beautiful Soup's tree traversal methods. - -The tree traversal methods are the main advantage of using Beautiful -Soup over other parsers. - -Different parsers will build different Beautiful Soup trees given the -same markup, but all Beautiful Soup trees can be traversed with the -methods tested here. -""" - -import copy -import pickle -import re -from bs4 import BeautifulSoup -from bs4.builder import builder_registry -from bs4.element import CData, SoupStrainer, Tag -from bs4.testing import SoupTest - -class TreeTest(SoupTest): - - def assertSelects(self, tags, should_match): - """Make sure that the given tags have the correct text. - - This is used in tests that define a bunch of tags, each - containing a single string, and then select certain strings by - some mechanism. - """ - self.assertEqual([tag.string for tag in tags], should_match) - - def assertSelectsIDs(self, tags, should_match): - """Make sure that the given tags have the correct IDs. - - This is used in tests that define a bunch of tags, each - containing a single string, and then select certain strings by - some mechanism. - """ - self.assertEqual([tag['id'] for tag in tags], should_match) - - -class TestFind(TreeTest): - """Basic tests of the find() method. - - find() just calls find_all() with limit=1, so it's not tested all - that thouroughly here. - """ - - def test_find_tag(self): - soup = self.soup("1234") - self.assertEqual(soup.find("b").string, "2") - - def test_unicode_text_find(self): - soup = self.soup(u'

Räksmörgås

') - self.assertEqual(soup.find(text=u'Räksmörgås'), u'Räksmörgås') - - -class TestFindAll(TreeTest): - """Basic tests of the find_all() method.""" - - def test_find_all_text_nodes(self): - """You can search the tree for text nodes.""" - soup = self.soup("Foobar\xbb") - # Exact match. - self.assertEqual(soup.find_all(text="bar"), [u"bar"]) - # Match any of a number of strings. - self.assertEqual( - soup.find_all(text=["Foo", "bar"]), [u"Foo", u"bar"]) - # Match a regular expression. - self.assertEqual(soup.find_all(text=re.compile('.*')), - [u"Foo", u"bar", u'\xbb']) - # Match anything. - self.assertEqual(soup.find_all(text=True), - [u"Foo", u"bar", u'\xbb']) - - def test_find_all_limit(self): - """You can limit the number of items returned by find_all.""" - soup = self.soup("12345") - self.assertSelects(soup.find_all('a', limit=3), ["1", "2", "3"]) - self.assertSelects(soup.find_all('a', limit=1), ["1"]) - self.assertSelects( - soup.find_all('a', limit=10), ["1", "2", "3", "4", "5"]) - - # A limit of 0 means no limit. - self.assertSelects( - soup.find_all('a', limit=0), ["1", "2", "3", "4", "5"]) - -class TestFindAllByName(TreeTest): - """Test ways of finding tags by tag name.""" - - def setUp(self): - super(TreeTest, self).setUp() - self.tree = self.soup("""First tag. - Second tag. - Third Nested tag. tag.""") - - def test_find_all_by_tag_name(self): - # Find all the tags. - self.assertSelects( - self.tree.find_all('a'), ['First tag.', 'Nested tag.']) - - def test_find_all_on_non_root_element(self): - # You can call find_all on any node, not just the root. - self.assertSelects(self.tree.c.find_all('a'), ['Nested tag.']) - - def test_calling_element_invokes_find_all(self): - self.assertSelects(self.tree('a'), ['First tag.', 'Nested tag.']) - - def test_find_all_by_tag_strainer(self): - self.assertSelects( - self.tree.find_all(SoupStrainer('a')), - ['First tag.', 'Nested tag.']) - - def test_find_all_by_tag_names(self): - self.assertSelects( - self.tree.find_all(['a', 'b']), - ['First tag.', 'Second tag.', 'Nested tag.']) - - def test_find_all_by_tag_dict(self): - self.assertSelects( - self.tree.find_all({'a' : True, 'b' : True}), - ['First tag.', 'Second tag.', 'Nested tag.']) - - def test_find_all_by_tag_re(self): - self.assertSelects( - self.tree.find_all(re.compile('^[ab]$')), - ['First tag.', 'Second tag.', 'Nested tag.']) - - def test_find_all_with_tags_matching_method(self): - # You can define an oracle method that determines whether - # a tag matches the search. - def id_matches_name(tag): - return tag.name == tag.get('id') - - tree = self.soup("""Match 1. - Does not match. - Match 2.""") - - self.assertSelects( - tree.find_all(id_matches_name), ["Match 1.", "Match 2."]) - - -class TestFindAllByAttribute(TreeTest): - - def test_find_all_by_attribute_name(self): - # You can pass in keyword arguments to find_all to search by - # attribute. - tree = self.soup(""" - Matching a. - - Non-matching Matching b.a. - """) - self.assertSelects(tree.find_all(id='first'), - ["Matching a.", "Matching b."]) - - def test_find_all_by_attribute_dict(self): - # You can pass in a dictionary as the argument 'attrs'. This - # lets you search for attributes like 'name' (a fixed argument - # to find_all) and 'class' (a reserved word in Python.) - tree = self.soup(""" - Name match. - Class match. - Non-match. - A tag called 'name1'. - """) - - # This doesn't do what you want. - self.assertSelects(tree.find_all(name='name1'), - ["A tag called 'name1'."]) - # This does what you want. - self.assertSelects(tree.find_all(attrs={'name' : 'name1'}), - ["Name match."]) - - # Passing class='class2' would cause a syntax error. - self.assertSelects(tree.find_all(attrs={'class' : 'class2'}), - ["Class match."]) - - def test_find_all_by_class(self): - # Passing in a string to 'attrs' will search the CSS class. - tree = self.soup(""" - Class 1. - Class 2. - Class 1. - Class 3 and 4. - """) - self.assertSelects(tree.find_all('a', '1'), ['Class 1.']) - self.assertSelects(tree.find_all(attrs='1'), ['Class 1.', 'Class 1.']) - self.assertSelects(tree.find_all('c', '3'), ['Class 3 and 4.']) - self.assertSelects(tree.find_all('c', '4'), ['Class 3 and 4.']) - - def test_find_all_by_attribute_soupstrainer(self): - tree = self.soup(""" - Match. - Non-match.""") - - strainer = SoupStrainer(attrs={'id' : 'first'}) - self.assertSelects(tree.find_all(strainer), ['Match.']) - - def test_find_all_with_missing_atribute(self): - # You can pass in None as the value of an attribute to find_all. - # This will match tags that do not have that attribute set. - tree = self.soup("""ID present. - No ID present. - ID is empty.""") - self.assertSelects(tree.find_all('a', id=None), ["No ID present."]) - - def test_find_all_with_defined_attribute(self): - # You can pass in None as the value of an attribute to find_all. - # This will match tags that have that attribute set to any value. - tree = self.soup("""ID present. - No ID present. - ID is empty.""") - self.assertSelects( - tree.find_all(id=True), ["ID present.", "ID is empty."]) - - def test_find_all_with_numeric_attribute(self): - # If you search for a number, it's treated as a string. - tree = self.soup("""Unquoted attribute. - Quoted attribute.""") - - expected = ["Unquoted attribute.", "Quoted attribute."] - self.assertSelects(tree.find_all(id=1), expected) - self.assertSelects(tree.find_all(id="1"), expected) - - def test_find_all_with_list_attribute_values(self): - # You can pass a list of attribute values instead of just one, - # and you'll get tags that match any of the values. - tree = self.soup("""1 - 2 - 3 - No ID.""") - self.assertSelects(tree.find_all(id=["1", "3", "4"]), - ["1", "3"]) - - def test_find_all_with_regular_expression_attribute_value(self): - # You can pass a regular expression as an attribute value, and - # you'll get tags whose values for that attribute match the - # regular expression. - tree = self.soup("""One a. - Two as. - Mixed as and bs. - One b. - No ID.""") - - self.assertSelects(tree.find_all(id=re.compile("^a+$")), - ["One a.", "Two as."]) - - -class TestIndex(TreeTest): - """Test Tag.index""" - def test_index(self): - tree = self.soup(""" - Identical - Not identical - Identical - - Identical with child - Also not identical - Identical with child - """) - wrap = tree.wrap - for i, element in enumerate(wrap.contents): - self.assertEqual(i, wrap.index(element)) - self.assertRaises(ValueError, tree.index, 1) - - -class TestParentOperations(TreeTest): - """Test navigation and searching through an element's parents.""" - - def setUp(self): - super(TestParentOperations, self).setUp() - self.tree = self.soup(''' -