def test_link_encoding(self):
import mechanize
from mechanize._rfc3986 import clean_url
url = "http://example.com/"
for encoding in ["UTF-8", "latin-1"]:
encoding_decl = "; charset=%s" % encoding
b = TestBrowser()
r = MockResponse(url, """\
<a href="http://example.com/foo/bar——.html"
name="name0——">blah——</a>
""", {"content-type": "text/html%s" % encoding_decl})
b.add_handler(make_mock_handler()([("http_open", r)]))
r = b.open(url)
Link = mechanize.Link
mdashx2 = em_dash.encode('utf-8') * 2
qmdashx2 = clean_url(mdashx2, encoding)
# base_url, url, text, tag, attrs
exp = Link(
url, "http://example.com/foo/bar%s.html" % qmdashx2,
unicode_type("blah") + (em_dash * 2), "a",
[("href", unicode_type("http://example.com/foo/bar%s.html") % (
em_dash*2)),
("name", unicode_type("name0") + em_dash + em_dash)])
# nr
link = b.find_link()
# print
# print exp
# print link
self.assertEqual(link, exp)
def _test_link_encoding(self, factory):
import mechanize
from mechanize._rfc3986 import clean_url
url = "http://example.com/"
for encoding in ["UTF-8", "latin-1"]:
encoding_decl = "; charset=%s" % encoding
b = TestBrowser(factory=factory)
r = MockResponse(
url,
"""\
<a href="http://example.com/foo/bar——.html"
name="name0——">blah——</a>
""", #"
{"content-type": "text/html%s" % encoding_decl})
b.add_handler(make_mock_handler()([("http_open", r)]))
r = b.open(url)
Link = mechanize.Link
try:
mdashx2 = u"\u2014".encode(encoding) * 2
except UnicodeError:
mdashx2 = '——'
qmdashx2 = clean_url(mdashx2, encoding)
# base_url, url, text, tag, attrs
exp = Link(
url, "http://example.com/foo/bar%s.html" % qmdashx2,
"blah" + mdashx2, "a",
[("href", "http://example.com/foo/bar%s.html" % mdashx2),
("name", "name0%s" % mdashx2)])
# nr
link = b.find_link()
## print
## print exp
## print link
self.assertEqual(link, exp)
def test_link_whitespace(self):
from mechanize import Link
base_url = "http://example.com/"
url = " http://example.com/foo.html%20+ "
stripped_url = url.strip()
html = '<a href="%s"></a>' % url
b = TestBrowser()
r = MockResponse(base_url, html, {"content-type": "text/html"})
b.add_handler(make_mock_handler()([("http_open", r)]))
r = b.open(base_url)
link = b.find_link(nr=0)
self.assertEqual(
link, Link(base_url, stripped_url, "", "a", [("href", url)]))
def test_links(self):
import mechanize
from mechanize import Link
url = "http://example.com/"
b = TestBrowser()
r = MockResponse(url, """<html>
<head><title>Title</title></head>
<body>
<a href="http://example.com/foo/bar.html" name="apples"></a>
<a name="pears"></a>
<a href="spam" name="pears"></a>
<area href="blah" name="foo"></area>
<form name="form2">
<input type="submit" name="two">
</form>
<iframe name="name" href="href" src="src"></iframe>
<iframe name="name2" href="href" src="src"></iframe>
<a name="name3" href="one">yada yada</a>
<a name="pears" href="two" weird="stuff">rhubarb</a>
<a></a>
<iframe src="foo"></iframe>
</body>
</html>
""", {"content-type": "text/html"})
b.add_handler(make_mock_handler()([("http_open", r)]))
r = b.open(url)
exp_links = [
# base_url, url, text, tag, attrs
Link(url, "http://example.com/foo/bar.html", "", "a",
[("href", "http://example.com/foo/bar.html"),
("name", "apples")]),
Link(url, "spam", "", "a", [("href", "spam"), ("name", "pears")]),
Link(url, "blah", '', "area", [("href", "blah"), ("name", "foo")]),
Link(url, "src", '', "iframe",
[("name", "name"), ("href", "href"), ("src", "src")]),
Link(url, "src", '', "iframe",
[("name", "name2"), ("href", "href"), ("src", "src")]),
Link(url, "one", "yada yada", "a",
[("name", "name3"), ("href", "one")]),
Link(url, "two", "rhubarb", "a",
[("name", "pears"), ("href", "two"), ("weird", "stuff")]),
Link(url, "foo", '', "iframe", [("src", "foo")]),
]
links = list(b.links())
for got, expect in zip(links, exp_links):
self.assertEqual(got, expect)
self.assertEqual(len(links), len(exp_links))
# nr
lnk = b.find_link()
self.assertEqual(lnk.url, "http://example.com/foo/bar.html")
lnk = b.find_link(nr=1)
self.assertEqual(lnk.url, "spam")
# text
lnk = b.find_link(text="yada yada")
self.assertEqual(lnk.url, "one")
self.assertRaises(
mechanize.LinkNotFoundError, b.find_link, text="da ya")
lnk = b.find_link(text_regex=re.compile("da ya"))
self.assertEqual(lnk.url, "one")
lnk = b.find_link(text_regex="da ya")
self.assertEqual(lnk.url, "one")
# name
lnk = b.find_link(name="name3")
self.assertEqual(lnk.url, "one")
lnk = b.find_link(name_regex=re.compile("oo"))
self.assertEqual(lnk.url, "blah")
lnk = b.find_link(name_regex="oo")
self.assertEqual(lnk.url, "blah")
# url
lnk = b.find_link(url="spam")
self.assertEqual(lnk.url, "spam")
lnk = b.find_link(url_regex=re.compile("pam"))
self.assertEqual(lnk.url, "spam")
lnk = b.find_link(url_regex="pam")
self.assertEqual(lnk.url, "spam")
# tag
lnk = b.find_link(tag="area")
self.assertEqual(lnk.url, "blah")
# predicate
lnk = b.find_link(
predicate=lambda lnk: dict(lnk.attrs).get("weird") == "stuff")
self.assertEqual(lnk.url, "two")
# combinations
lnk = b.find_link(name="pears", nr=1)
self.assertEqual(lnk.text, "rhubarb")
lnk = b.find_link(url="src", nr=0, name="name2")
self.assertEqual(lnk.tag, "iframe")
self.assertEqual(lnk.url, "src")
self.assertRaises(
mechanize.LinkNotFoundError,
b.find_link,
url="src",
nr=1,
name="name2")
lnk = b.find_link(
tag="a", predicate=lambda lnk: dict(lnk.attrs).get(
"weird") == "stuff")
self.assertEqual(lnk.url, "two")
# .links()
self.assertEqual(
list(b.links(url="src")), [
Link(
url,
url="src",
text='',
tag="iframe",
attrs=[("name", "name"), ("href", "href"),
("src", "src")]),
Link(
url,
url="src",
text='',
tag="iframe",
attrs=[("name", "name2"), ("href", "href"),
("src", "src")]),
])