Explicitly include excluded URLs in the result set; mark as excluded.

earwigbot/wiki/copyvios/result.py

@@ -40,6 +40,7 @@ class CopyvioSource(object):
     - :py:attr:`confidence`: the confidence of a violation, between 0 and 1
     - :py:attr:`chains`:     a 2-tuple of the source chain and the delta chain
     - :py:attr:`skipped`:    whether this URL was skipped during the check
+    - :py:attr:`excluded`:   whether this URL was in the exclusions list
     """
 
     def __init__(self, workspace, url, headers=None, timeout=5):
@@ -50,6 +51,7 @@ class CopyvioSource(object):
         self.confidence = 0.0
         self.chains = (EMPTY, EMPTY_INTERSECTION)
         self.skipped = False
+        self.excluded = False
 
         self._event1 = Event()
         self._event2 = Event()
@@ -57,11 +59,15 @@ class CopyvioSource(object):
 
     def __repr__(self):
         """Return the canonical string representation of the source."""
-        res = "CopyvioSource(url={0!r}, confidence={1!r}, skipped={2!r})"
-        return res.format(self.url, self.confidence, self.skipped)
+        res = ("CopyvioSource(url={0!r}, confidence={1!r}, skipped={2!r}, "
+               "excluded={3!r})")
+        return res.format(
+            self.url, self.confidence, self.skipped, self.excluded)
 
     def __str__(self):
         """Return a nice string representation of the source."""
+        if self.excluded:
+            return "<CopyvioSource ({0}, excluded)>".format(self.url)
         if self.skipped:
             return "<CopyvioSource ({0}, skipped)>".format(self.url)
         res = "<CopyvioSource ({0} with {1} conf)>"

earwigbot/wiki/copyvios/workers.py

@@ -311,11 +311,15 @@ class CopyvioWorkspace(object):
                 if url in self._handled_urls:
                     continue
                 self._handled_urls.add(url)
-                if exclude_check and exclude_check(url):
-                    continue
 
                 source = CopyvioSource(url=url, **self._source_args)
                 self.sources.append(source)
+
+                if exclude_check and exclude_check(url):
+                    self._logger.debug(u"enqueue(): exclude {0}".format(url))
+                    source.excluded = True
+                    source.skip()
+                    continue
                 if self._short_circuit and self.finished:
                     self._logger.debug(u"enqueue(): auto-skip {0}".format(url))
                     source.skip()
@@ -371,6 +375,8 @@ class CopyvioWorkspace(object):
         def cmpfunc(s1, s2):
             if s2.confidence != s1.confidence:
                 return 1 if s2.confidence > s1.confidence else -1
+            if s2.excluded != s1.excluded:
+                return 1 if s1.excluded else -1
             return int(s1.skipped) - int(s2.skipped)
 
         self.sources.sort(cmpfunc)