Code cleanup.

linkcheck/checker/__init__.py

@@ -27,6 +27,23 @@ from .. import strformat, url as urlutil, log, LOG_CHECK
 MAX_FILESIZE = 1024*1024*10 # 10MB
 
 
+def guess_url(url):
+    """Guess if URL is a http or ftp URL.
+    @param url: the URL to check
+    @ptype url: unicode
+    @return: url with http:// or ftp:// prepended if it's detected as
+      a http respective ftp URL.
+    @rtype: unicode
+    """
+    if url.lower().startswith("www."):
+        # syntactic sugar
+        return "http://%s" % url
+    elif url.lower().startswith("ftp."):
+        # syntactic sugar
+        return "ftp://%s" % url
+    return url
+
+
 def absolute_url (base_url, base_ref, parent_url):
     """
     Search for the absolute url to detect the link type. This does not

linkcheck/cmdline.py

@@ -115,11 +115,6 @@ class LCOptionParser (optparse.OptionParser, object):
 def aggregate_url (aggregate, url, err_exit_code=2):
     """Append given commandline URL to input queue."""
     get_url_from = checker.get_url_from
-    if url.lower().startswith("www."):
-        # syntactic sugar
-        url = "http://%s" % url
-    elif url.lower().startswith("ftp."):
-        # syntactic sugar
-        url = "ftp://%s" % url
+    url = checker.guess_url(url)
     url_data = get_url_from(url, 0, aggregate, extern=(0, 0))
     aggregate.urlqueue.put(url_data)

linkcheck/gui/__init__.py

@@ -436,11 +436,8 @@ Version 2 or later.
     def get_url (self):
         """Return URL to check from the urlinput widget."""
         url = strformat.stripurl(unicode(self.urlinput.text()))
-        if url.startswith(u"www."):
-            url = u"http://%s" % url
-        elif url.startswith(u"ftp."):
-            url = u"ftp://%s" % url
-        elif url and u":" not in url:
+        url = checker.guess_url(url)
+        if url and u":" not in url:
             # Look for local file, else assume it's an HTTP URL.
             if not os.path.exists(url):
                 url = u"http://%s" % url
@@ -459,14 +456,6 @@ Version 2 or later.
             return
         self.set_statusmsg(_("Checking '%s'.") % strformat.limit(url, 40))
         url_data = checker.get_url_from(url, 0, aggregate, extern=(0, 0))
-        # XXX
-        #try:
-        #    self.backup_config('internlinks')
-        #    add_intern_pattern(url_data, self.config)
-        #except UnicodeError:
-        #    self.set_statusmsg(_("Error, invalid URL `%s'.") %
-        #                          strformat.limit(url, 40))
-        #    return
         self.recent.add_document(url)
         aggregate.urlqueue.put(url_data)
         self.aggregate = aggregate