def test_comment2(self):
html = """
<html>
<head><title>Foo</title></head>
<body>
Some text.
<!-- HTML comment -->
Some more text.
<p>Text</p>
More text
</body>
</html>
"""
og = decode_and_calc_og(html, "http://example.com/test.html")
self.assertEqual(
og,
{
"og:title":
"Foo",
"og:description":
"Some text.\n\nSome more text.\n\nText\n\nMore text",
},
)
def test_missing_title(self):
html = """
<html>
<body>
Some text.
</body>
</html>
"""
og = decode_and_calc_og(html, "http://example.com/test.html")
self.assertEqual(og, {"og:title": None, "og:description": "Some text."})
def test_missing_title(self):
html = u"""
<html>
<body>
Some text.
</body>
</html>
"""
og = decode_and_calc_og(html, "http://example.com/test.html")
self.assertEquals(og, {u"og:title": None, u"og:description": u"Some text."})
def test_simple(self):
html = u"""
<html>
<head><title>Foo</title></head>
<body>
Some text.
</body>
</html>
"""
og = decode_and_calc_og(html, "http://example.com/test.html")
self.assertEquals(og, {u"og:title": u"Foo", u"og:description": u"Some text."})
def test_h1_as_title(self):
html = u"""
<html>
<meta property="og:description" content="Some text."/>
<body>
<h1>Title</h1>
</body>
</html>
"""
og = decode_and_calc_og(html, "http://example.com/test.html")
self.assertEquals(og, {u"og:title": u"Title", u"og:description": u"Some text."})
def test_invalid_encoding2(self):
"""A body which doesn't match the sent character encoding."""
# Note that this contains an invalid UTF-8 sequence in the title.
html = b"""
<html>
<head><title>\xff\xff Foo</title></head>
<body>
Some text.
</body>
</html>
"""
og = decode_and_calc_og(html, "http://example.com/test.html")
self.assertEqual(og, {"og:title": "ÿÿ Foo", "og:description": "Some text."})
def test_h1_as_title(self):
html = """
<html>
<meta property="og:description" content="Some text."/>
<body>
<h1>Title</h1>
</body>
</html>
"""
og = decode_and_calc_og(html, "http://example.com/test.html")
self.assertEqual(og, {"og:title": "Title", "og:description": "Some text."})
def test_simple(self):
html = """
<html>
<head><title>Foo</title></head>
<body>
Some text.
</body>
</html>
"""
og = decode_and_calc_og(html, "http://example.com/test.html")
self.assertEqual(og, {"og:title": "Foo", "og:description": "Some text."})
def test_invalid_encoding(self):
"""An invalid character encoding should be ignored and treated as UTF-8, if possible."""
html = """
<html>
<head><title>Foo</title></head>
<body>
Some text.
</body>
</html>
"""
og = decode_and_calc_og(
html, "http://example.com/test.html", "invalid-encoding"
)
self.assertEqual(og, {"og:title": "Foo", "og:description": "Some text."})
def test_missing_title_and_broken_h1(self):
html = u"""
<html>
<body>
<h1><a href="foo"/></h1>
Some text.
</body>
</html>
"""
og = decode_and_calc_og(html, "http://example.com/test.html")
self.assertEquals(og, {
u"og:title": None,
u"og:description": u"Some text."
})
def test_comment(self):
html = u"""
<html>
<head><title>Foo</title></head>
<body>
<!-- HTML comment -->
Some text.
</body>
</html>
"""
og = decode_and_calc_og(html, "http://example.com/test.html")
self.assertEquals(og, {
u"og:title": u"Foo",
u"og:description": u"Some text."
})
def test_script(self):
html = u"""
<html>
<head><title>Foo</title></head>
<body>
<script> (function() {})() </script>
Some text.
</body>
</html>
"""
og = decode_and_calc_og(html, "http://example.com/test.html")
self.assertEquals(og, {
u"og:title": u"Foo",
u"og:description": u"Some text."
})
def test_comment(self):
html = """
<html>
<head><title>Foo</title></head>
<body>
<!-- HTML comment -->
Some text.
</body>
</html>
"""
og = decode_and_calc_og(html, "http://example.com/test.html")
self.assertEquals(og, {
"og:title": "Foo",
"og:description": "Some text."
})
def test_script(self):
html = """
<html>
<head><title>Foo</title></head>
<body>
<script> (function() {})() </script>
Some text.
</body>
</html>
"""
og = decode_and_calc_og(html, "http://example.com/test.html")
self.assertEquals(og, {
"og:title": "Foo",
"og:description": "Some text."
})
def test_comment2(self):
html = u"""
<html>
<head><title>Foo</title></head>
<body>
Some text.
<!-- HTML comment -->
Some more text.
<p>Text</p>
More text
</body>
</html>
"""
og = decode_and_calc_og(html, "http://example.com/test.html")
self.assertEquals(og, {
u"og:title": u"Foo",
u"og:description": u"Some text.\n\nSome more text.\n\nText\n\nMore text"
})
def test_empty(self):
html = ""
og = decode_and_calc_og(html, "http://example.com/test.html")
self.assertEqual(og, {})
def test_no_tree(self):
"""A valid body with no tree in it."""
html = b"\x00"
og = decode_and_calc_og(html, "http://example.com/test.html")
self.assertEqual(og, {})
def test_empty(self):
"""Test a body with no data in it."""
html = b""
og = decode_and_calc_og(html, "http://example.com/test.html")
self.assertEqual(og, {})