From d18cebdeffafb4e4c9c6239da2b3ae94ae5a42e8 Mon Sep 17 00:00:00 2001 From: Leonard Richardson Date: Wed, 29 Dec 2010 11:52:00 -0500 Subject: Ported some tests from the old system to the new smoke test class. --- src/beautifulsoup/builder/__init__.py | 2 ++ src/beautifulsoup/element.py | 4 +-- src/beautifulsoup/tests/helpers.py | 50 ++++++++++++++++++++++++++++++--- src/beautifulsoup/tests/test_soup.py | 53 ----------------------------------- 4 files changed, 50 insertions(+), 59 deletions(-) (limited to 'src') diff --git a/src/beautifulsoup/builder/__init__.py b/src/beautifulsoup/builder/__init__.py index b7db8db..8294c0c 100644 --- a/src/beautifulsoup/builder/__init__.py +++ b/src/beautifulsoup/builder/__init__.py @@ -11,6 +11,8 @@ class TreeBuilder(Entities): assume_html = False smart_quotes_to = Entities.XML_ENTITIES + convert_html_entities = True + convert_xml_entities = True def __init__(self): self.soup = None diff --git a/src/beautifulsoup/element.py b/src/beautifulsoup/element.py index b5b8e84..7649b4c 100644 --- a/src/beautifulsoup/element.py +++ b/src/beautifulsoup/element.py @@ -383,10 +383,10 @@ class Tag(PageElement, Entities): entities are being converted, any unrecognized entities are escaped.""" x = match.group(1) - if builder.convertHTMLEntities and x in name2codepoint: + if builder.convert_html_entities and x in name2codepoint: return unichr(name2codepoint[x]) elif x in self.XML_ENTITIES_TO_SPECIAL_CHARS: - if builder.convertXMLEntities: + if builder.convert_xml_entities: return self.XML_ENTITIES_TO_SPECIAL_CHARS[x] else: return u'&%s;' % x diff --git a/src/beautifulsoup/tests/helpers.py b/src/beautifulsoup/tests/helpers.py index a4156cd..fbdcedb 100644 --- a/src/beautifulsoup/tests/helpers.py +++ b/src/beautifulsoup/tests/helpers.py @@ -66,15 +66,43 @@ class BuilderSmokeTest(SoupTest): self.assertEqual(blockquote.p.b.string, 'Foo') self.assertEqual(blockquote.b.string, 'Foo') + def test_collapsed_whitespace(self): + """In most tags, whitespace is collapsed.""" + self.assertSoupEquals("

", "

") + + def test_preserved_whitespace_in_pre_and_textarea(self): + """In 
 and ")
+
+
+    def test_single_quote_attribute_values_become_double_quotes(self):
+        self.assertSoupEquals("",
+                              '')
+
+    def test_attribute_values_with_nested_quotes_are_left_alone(self):
+        text = """a"""
+        self.assertSoupEquals(text)
+
+    def test_attribute_values_with_double_nested_quotes_get_quoted(self):
+        text = """a"""
+        soup = self.soup(text)
+        soup.foo['attr'] = 'Brawls happen at "Bob\'s Bar"'
+        self.assertSoupEquals(
+            soup.foo.decode(),
+            """a""")
+
+    def test_ampersand_in_attribute_value_gets_quoted(self):
+        self.assertSoupEquals('',
+                              '')
+
 
 class BuilderInvalidMarkupSmokeTest(SoupTest):
     """Tests of invalid markup.
 
     These are very likely to give different results for different tree
-    builders.
-
-    It's not required that a tree builder handle invalid markup at
-    all.
+    builders. It's not required that a tree builder handle invalid
+    markup at all.
     """
 
     def test_unclosed_block_level_elements(self):
@@ -82,3 +110,17 @@ class BuilderInvalidMarkupSmokeTest(SoupTest):
         self.assertSoupEquals(
             '
Foo
Bar',
             '
Foo
Bar
')
+
+    def test_fake_self_closing_tag(self):
+        # If a self-closing tag presents as a normal tag, the 'open'
+        # tag is treated as an instance of the self-closing tag and
+        # the 'close' tag is ignored.
+        self.assertSoupEquals(
+            "http://foo.com/",
+            "http://foo.com/")
+
+    def test_boolean_attribute_with_no_value_gets_empty_value(self):
+        soup = self.soup("
foo
")
+        self.assertEquals(soup.table.td['nowrap'], '')
+
+
diff --git a/src/beautifulsoup/tests/test_soup.py b/src/beautifulsoup/tests/test_soup.py
index c35d10b..5724247 100644
--- a/src/beautifulsoup/tests/test_soup.py
+++ b/src/beautifulsoup/tests/test_soup.py
@@ -439,25 +439,6 @@ class TheManWithoutAttributes(SoupTest):
         text = ""
         self.assertTrue(BeautifulSoup(text).foo.has_key('attr'))
 
-class QuoteMeOnThat(SoupTest):
-    "Test quoting"
-    def testQuotedAttributeValues(self):
-        self.assertSoupEquals("",
-                              '')
-
-        text = """a"""
-        soup = BeautifulSoup(text)
-        self.assertEquals(soup.decode(), text)
-
-        soup.foo['attr'] = 'Brawls happen at "Bob\'s Bar"'
-        newText = """a"""
-        self.assertSoupEquals(soup.decode(), newText)
-
-        self.assertSoupEquals('',
-                              '')
-
-
-
 class YoureSoLiteral(SoupTest):
     "Test literal mode."
     def testLiteralMode(self):
@@ -536,16 +517,6 @@ class CleanupOnAisleFour(SoupTest):
         self.assertEqual(soup.decode(),
                          '
test1test2
')
 
-    def testSelfClosingTagOrNot(self):
-        text = "http://foo.com/"
-        self.assertEqual(BeautifulStoneSoup(text).decode(), text)
-        self.assertEqual(BeautifulSoup(text).decode(),
-                         'http://foo.com/')
-
-    def testBooleanAttributes(self):
-        text = "foo"
-        self.assertSoupEquals(text, text)
-
     def testCData(self):
         xml = "foobar"
         self.assertSoupEquals(xml, xml)
@@ -825,29 +796,5 @@ class EncodeRed(SoupTest):
         self.assertSoupEquals(utf_8, encoding='utf-8')
 
 
-class Whitewash(SoupTest):
-    """Test whitespace preservation."""
-
-    def testPreservedWhitespace(self):
-        self.assertSoupEquals("
   
")
-        self.assertSoupEquals("
 woo  
")
-
-    def testCollapsedWhitespace(self):
-        self.assertSoupEquals("
   
", "
 
")
-
-
-class AlternateBuilders(SoupTest):
-    """Test alternate builders."""
-
-    def testICantBelieveItsValidHTML(self):
-        builder = ICantBelieveItsValidHTMLTreeBuilder()
-        markup = "FooBar"
-
-        soup = BeautifulSoup(markup)
-        self.assertEquals(soup.decode(), "FooBar")
-
-        soup = BeautifulSoup(markup, builder=builder)
-        self.assertEquals(soup.decode(), markup)
-
 if __name__ == '__main__':
     unittest.main()
-- 
cgit v1.2.3