[test/test_utils] add more tests for update_url_query

[youtube-dl] / youtube_dl / utils.py

diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py
index be1f3b0d77a55a3245b5cff44982256e072d6c0e..31d60f3233bb3ea95775f390d4f4afc0681b52d3 100644 (file)
--- a/youtube_dl/utils.py
+++ b/youtube_dl/utils.py
@@ -160,8 +160,6 @@ if sys.version_info >= (2, 7):
     def find_xpath_attr(node, xpath, key, val=None):
         """ Find the xpath xpath[@key=val] """
         assert re.match(r'^[a-zA-Z_-]+$', key)
-        if val:
-            assert re.match(r'^[a-zA-Z0-9@\s:._-]*$', val)
         expr = xpath + ('[@%s]' % key if val is None else "[@%s='%s']" % (key, val))
         return node.find(expr)
 else:
@@ -1721,6 +1719,14 @@ def urlencode_postdata(*args, **kargs):
     return compat_urllib_parse.urlencode(*args, **kargs).encode('ascii')
 
 
+def update_url_query(url, query):
+    parsed_url = compat_urlparse.urlparse(url)
+    qs = compat_parse_qs(parsed_url.query)
+    qs.update(query)
+    return compat_urlparse.urlunparse(parsed_url._replace(
+        query=compat_urllib_parse.urlencode(qs, True)))
+
+
 def encode_dict(d, encoding='utf-8'):
     def encode(v):
         return v.encode(encoding) if isinstance(v, compat_basestring) else v
@@ -2621,9 +2627,17 @@ def ohdave_rsa_encrypt(data, exponent, modulus):
     return '%x' % encrypted
 
 
-def base_n(num, n, table):
+def encode_base_n(num, n, table=None):
+    FULL_TABLE = '0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ'
+    if not table:
+        table = FULL_TABLE[:n]
+
+    if n > len(table):
+        raise ValueError('base %d exceeds table length %d' % (n, len(table)))
+
     if num == 0:
-        return '0'
+        return table[0]
+
     ret = ''
     while num:
         ret = table[num % n] + ret
@@ -2631,9 +2645,21 @@ def base_n(num, n, table):
     return ret
 
 
-def base36(num):
-    return base_n(num, 36, '0123456789abcdefghijklmnopqrstuvwxyz')
+def decode_packed_codes(code):
+    mobj = re.search(
+        r"}\('(.+)',(\d+),(\d+),'([^']+)'\.split\('\|'\)",
+        code)
+    obfucasted_code, base, count, symbols = mobj.groups()
+    base = int(base)
+    count = int(count)
+    symbols = symbols.split('|')
+    symbol_table = {}
 
+    while count:
+        count -= 1
+        base_n_count = encode_base_n(count, base)
+        symbol_table[base_n_count] = symbols[count] or base_n_count
 
-def base62(num):
-    return base_n(num, 62, '0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ')
+    return re.sub(
+        r'\b(\w+)\b', lambda mobj: symbol_table[mobj.group(0)],
+        obfucasted_code)