[utils] Make value optional for find_xpath_attr

This allows selecting particular attributes by name but without specifying the value and similar to xpath syntax `[@attrib]`

diff --git a/test/test_utils.py b/test/test_utils.py
index 65692a9fbb2bdc0dd6dbc5203439f1ca12bf5c46..a759b2da99ea9f2c7676171d3bc0d5d93aa2149d 100644 (file)
--- a/test/test_utils.py
+++ b/test/test_utils.py
@@ -235,12 +235,21 @@ class TestUtil(unittest.TestCase):
             <node x="a"/>
             <node x="a" y="c" />
             <node x="b" y="d" />
+            <node x="" />
         </root>'''
         doc = xml.etree.ElementTree.fromstring(testxml)
 
+        self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
+        self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
+        self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
+        self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
+        self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
+        self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
+        self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
+        self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
 
     def test_xpath_with_ns(self):
         testxml = '''<root xmlns:media="http://example.com/">

diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py
index 88f9f90707fbd04966638125e94b264af95fedcb..78dc2b449417c5a8f1b30db15fd7b7a74eb84745 100644 (file)
--- a/youtube_dl/utils.py
+++ b/youtube_dl/utils.py
@@ -139,21 +139,24 @@ def write_json_file(obj, fn):
 
 
 if sys.version_info >= (2, 7):
-    def find_xpath_attr(node, xpath, key, val):
+    def find_xpath_attr(node, xpath, key, val=None):
         """ Find the xpath xpath[@key=val] """
         assert re.match(r'^[a-zA-Z-]+$', key)
-        assert re.match(r'^[a-zA-Z0-9@\s:._-]*$', val)
-        expr = xpath + "[@%s='%s']" % (key, val)
+        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:
-    def find_xpath_attr(node, xpath, key, val):
+    def find_xpath_attr(node, xpath, key, val=None):
         # Here comes the crazy part: In 2.6, if the xpath is a unicode,
         # .//node does not match if a node is a direct child of . !
         if isinstance(xpath, compat_str):
             xpath = xpath.encode('ascii')
 
         for f in node.findall(xpath):
-            if f.attrib.get(key) == val:
+            if key not in f.attrib:
+                continue
+            if val is None or f.attrib.get(key) == val:
                 return f
         return None