start implementing more efficient querying of URL patterns

1 files changed, 62 insertions, 0 deletions

diff --git a/test/unit/test_patterns.py b/test/unit/test_patterns.py
index 4162fc0..4cfc10c 100644
--- a/test/unit/test_patterns.py
+++ b/test/unit/test_patterns.py
@@ -89,3 +89,65 @@ def test_regexes(execute_in_page, patterns_code):
     match = execute_in_page('returnval(ftp_regex.exec(arguments[0]));',
                             '@bad.url/')
     assert match is None
+
+def test_deconstruct_url(execute_in_page, patterns_code):
+    """
+    patterns.js contains deconstruct_url() function that handles URL parsing.
+    Verify it works properly.
+    """
+    execute_in_page(patterns_code, page='https://gotmyowndoma.in')
+
+    deco = execute_in_page('returnval(deconstruct_url(arguments[0]));',
+                           'https://eXaMpLe.com/a/b?ver=1.2.3#heading2')
+    assert deco
+    assert deco['trailing_dash'] == False
+    assert deco['proto']         == 'https'
+    assert deco['domain']        == ['example', 'com']
+    assert deco['path']          == ['a', 'b']
+
+    deco = execute_in_page('returnval(deconstruct_url(arguments[0]));',
+                           'http://**.example.com/')
+    assert deco
+    assert deco['trailing_dash'] == True
+    assert deco['proto']         == 'http'
+    assert deco['domain']        == ['**', 'example', 'com']
+    assert deco['path']          == []
+
+    deco = execute_in_page('returnval(deconstruct_url(arguments[0]));',
+                           'ftp://user@ftp.example.com/all///passwords.txt/')
+    assert deco
+    assert deco['trailing_dash'] == True
+    assert deco['proto']         == 'ftp'
+    assert deco['domain']        == ['ftp', 'example', 'com']
+    assert deco['path']          == ['all', 'passwords.txt']
+
+    deco = execute_in_page('returnval(deconstruct_url(arguments[0]));',
+                           'ftp://mirror.edu.pl.eu.org')
+    assert deco
+    assert deco['trailing_dash'] == False
+    assert deco['proto']         == 'ftp'
+    assert deco['domain']        == ['mirror', 'edu', 'pl', 'eu', 'org']
+    assert deco['path']          == []
+
+    deco = execute_in_page('returnval(deconstruct_url(arguments[0]));',
+                           'file:///mnt/parabola_chroot///etc/passwd')
+    assert deco
+    assert deco['trailing_dash'] == False
+    assert deco['proto']         == 'file'
+    assert deco['path']          == ['mnt', 'parabola_chroot', 'etc', 'passwd']
+
+    for bad_url in [
+            '://bad-url.missing/protocol',
+            'http:/example.com/a/b',
+            'unknown://example.com/a/b',
+            'idontfancypineapple',
+            'ftp://@example.org/',
+            'https:///some/path/',
+            'file://non-absolute/path'
+    ]:
+        with pytest.raises(Exception, match=r'Error in injected script'):
+            deco = execute_in_page('returnval(deconstruct_url(arguments[0]));',
+                                   bad_url)
+
+    # at some point we might also consider testing url deconstruction with
+    # length limits...