[utils] Improve xpath_text

[youtube-dl] / youtube_dl / utils.py

diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py
index de5069c7b39834666209fa2dfb7ab38db9a803f1..cc792be6ab2fb077d0f530396cd8fdc9072d8df4 100644 (file)
--- a/youtube_dl/utils.py
+++ b/youtube_dl/utils.py
@@ -141,7 +141,7 @@ def write_json_file(obj, fn):
 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-]+$', key)
+        assert re.match(r'^[a-zA-Z_-]+$', 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))
@@ -181,7 +181,7 @@ def xpath_element(node, xpath, name=None, fatal=False, default=NO_DEFAULT):
         xpath = xpath.encode('ascii')
 
     n = node.find(xpath)
-    if n is None or n.text is None:
+    if n is None:
         if default is not NO_DEFAULT:
             return default
         elif fatal:
@@ -193,7 +193,18 @@ def xpath_element(node, xpath, name=None, fatal=False, default=NO_DEFAULT):
 
 
 def xpath_text(node, xpath, name=None, fatal=False, default=NO_DEFAULT):
-    return xpath_element(node, xpath, name, fatal=fatal, default=default).text
+    n = xpath_element(node, xpath, name, fatal=fatal, default=default)
+    if n is None or n == default:
+        return n
+    if n.text is None:
+        if default is not NO_DEFAULT:
+            return default
+        elif fatal:
+            name = xpath if name is None else name
+            raise ExtractorError('Could not find XML element\'s text %s' % name)
+        else:
+            return None
+    return n.text
 
 
 def xpath_attr(node, xpath, key, name=None, fatal=False, default=NO_DEFAULT):