split and fix initialization of UrlData objects

git-svn-id: https://linkchecker.svn.sourceforge.net/svnroot/linkchecker/trunk/linkchecker@2622 e7d03fd6-7b0d-0410-9947-9c21f3af8025

linkcheck/checker/fileurl.py

@@ -83,18 +83,16 @@ class FileUrl (urlbase.UrlBase):
     Url link with file scheme.
     """
 
-    def __init__ (self, base_url, recursion_level, consumer,
-                  parent_url = None,
-                  base_ref = None, line=0, column=0, name=u""):
+    def init (self, base_ref, base_url, parent_url, recursion_level,
+              consumer, line, column, name):
         """
         Besides the usual initialization the URL is normed according
         to the platform:
          - the base URL is made an absolute file:// URL
          - under Windows platform the drive specifier is normed
         """
-        super(FileUrl, self).__init__(base_url, recursion_level, consumer,
-             parent_url=parent_url, base_ref=base_ref,
-             line=line, column=column, name=name)
+        super(FileUrl, self).init(base_ref, base_url, parent_url,
+                               recursion_level, consumer, line, column, name)
         base_url = self.base_url
         if not (parent_url or base_ref or base_url.startswith("file:")):
             base_url = os.path.expanduser(base_url)

linkcheck/checker/ftpurl.py

@@ -38,15 +38,11 @@ class FtpUrl (internpaturl.InternPatternUrl, proxysupport.ProxySupport):
     Url link with ftp scheme.
     """
 
-    def __init__ (self, base_url, recursion_level, consumer,
-                  parent_url = None,
-                  base_ref = None, line=0, column=0, name=u""):
+    def reset (self):
         """
         Initialize FTP url data.
         """
-        super(FtpUrl, self).__init__(base_url, recursion_level, consumer,
-             parent_url=parent_url, base_ref=base_ref,
-             line=line, column=column, name=name)
+        super(FtpUrl, self).reset()
         # list of files for recursion
         self.files = []
         # last part of URL filename

linkcheck/checker/httpurl.py

@@ -51,14 +51,11 @@ class HttpUrl (internpaturl.InternPatternUrl, proxysupport.ProxySupport):
     Url link with http scheme.
     """
 
-    def __init__ (self, base_url, recursion_level, consumer,
-                  parent_url=None, base_ref=None, line=0, column=0, name=u""):
+    def reset (self):
         """
-        Initialize basic url data and HTTP specific variables.
+        Initialize HTTP specific variables.
         """
-        super(HttpUrl, self).__init__(base_url, recursion_level, consumer,
-               parent_url=parent_url, base_ref=base_ref, line=line,
-               column=column, name=name)
+        super(HttpUrl, self).reset()
         self.max_redirects = 5
         self.has301status = False
         self.no_anchor = False # remove anchor in request url

linkcheck/checker/urlbase.py

@@ -112,6 +112,17 @@ class UrlBase (object):
         @param column: column number of url in parent content
         @param name: name of url or empty
         """
+        self.init(base_ref, base_url, parent_url, recursion_level,
+                  consumer, line, column, name)
+        self.reset()
+        self.check_syntax()
+
+
+    def init (self, base_ref, base_url, parent_url, recursion_level,
+              consumer, line, column, name):
+        """
+        Initialize internal data.
+        """
         self.base_ref = base_ref
         # note that self.base_url must not be modified
         self.base_url = base_url
@@ -131,6 +142,7 @@ class UrlBase (object):
         # assume file link if no scheme is found
         self.scheme = url.split(":", 1)[0] or "file"
 
+    def reset (self):
         # self.url is constructed by self.build_url() out of base_url
         # and (base_ref or parent) as absolute and normed url.
         # This the real url we use when checking so it also referred to
@@ -166,7 +178,6 @@ class UrlBase (object):
         # cache keys, are set by build_url() calling set_cache_keys()
         self.cache_url_key = None
         self.cache_content_key = None
-        self.check_syntax()
 
     def set_result (self, msg, valid=True):
         """