Fix bug with link spanning on several lines

pydoctor/epydoc/markup/epytext.py

@@ -1194,7 +1194,8 @@ def _colorize_link(link: Element, token: Token, end: int, errors: List[ParseErro
         # Here we used to process the target in order to remove arg lists for functions
         # and validate it. But now this happens in node2stan.parse_reference().
         # The target is not validated anymore since the intersphinx taget names can contain any kind of text.
-        pass
+        # We simply normalize it.
+        target = re.sub(r'\s', ' ', target)
 
     # Construct the target element.
     target_elt = Element('target', target, lineno=str(token.startline))

pydoctor/node2stan.py

@@ -80,11 +80,13 @@ def parse_reference(node:nodes.Node) -> Reference:
         label, target = node.children, node.attributes['refuri']
     else:
         # RST parsed.
-        m = _TARGET_RE.match(node.astext())
+        # Sanitize links spaning over multiple lines
+        node_text = re.sub(r'\s', ' ', node.astext())
+        m = _TARGET_RE.match(node_text)
         if m:
             label, target = m.groups()
         else:
-            label = target = node.astext()
+            label = target = node_text
     # Support linking to functions and methods with parameters
     try:
         begin_parameters = target.index('(')