filter duplicate URLs (#36)

Duplicate URLs might occur if, for example,  an artist adds another
image to his gallery while an extractor is running and images are being
downloaded on sites like pixiv/nijie/hentaifoundry.
The next image on the next page will have already been downloaded and
will cause a premature end if '--abort-on-skip' is being used.

gallery_dl/job.py

@@ -25,17 +25,26 @@ class Job():
         self.extractor.log.debug(
             "Using %s for %s", self.extractor.__class__.__name__, url)
 
+        # url predicates
+        predicates = [util.UniquePredicate()]
+
         items = config.get(("images",))
         if items:
             pred = util.RangePredicate(items)
             if pred.lower > 1:
                 pred.index += self.extractor.skip(pred.lower - 1)
-            self.pred_url = pred
-        else:
-            self.pred_url = True
+            predicates.append(pred)
+
+        self.pred_url = util.build_predicate(predicates)
+
+        # queue predicates
+        predicates = []
 
         items = config.get(("chapters",))
-        self.pred_queue = util.RangePredicate(items) if items else True
+        if items:
+            predicates.append(util.RangePredicate(items))
+
+        self.pred_queue = util.build_predicate(predicates)
 
     def run(self):
         """Execute or run the job"""
@@ -73,16 +82,17 @@ class Job():
     def dispatch(self, msg):
         """Call the appropriate message handler"""
         if msg[0] == Message.Url:
-            if self.pred_url:
-                self.update_kwdict(msg[2])
-                self.handle_url(msg[1], msg[2])
+            _, url, kwds = msg
+            if self.pred_url(url, kwds):
+                self.update_kwdict(kwds)
+                self.handle_url(url, kwds)
 
         elif msg[0] == Message.Directory:
             self.update_kwdict(msg[1])
             self.handle_directory(msg[1])
 
         elif msg[0] == Message.Queue:
-            if self.pred_queue:
+            if self.pred_queue(msg[1], None):
                 self.handle_queue(msg[1])
 
         elif msg[0] == Message.Version:

gallery_dl/util.py

@@ -139,8 +139,17 @@ CODES = {
 SPECIAL_EXTRACTORS = ("oauth", "recursive", "test")
 
 
+def build_predicate(predicates):
+    if not predicates:
+        return lambda url, kwds: True
+    elif len(predicates) == 1:
+        return predicates[0]
+    else:
+        return ChainPredicate(predicates)
+
+
 class RangePredicate():
-    """Predicate; is True if the current index is in the given range"""
+    """Predicate; True if the current index is in the given range"""
     def __init__(self, rangespec):
         self.ranges = optimize_range(parse_range(rangespec))
         self.index = 0
@@ -149,7 +158,7 @@ class RangePredicate():
         else:
             self.lower, self.upper = 0, 0
 
-    def __bool__(self):
+    def __call__(self, url, kwds):
         self.index += 1
 
         if self.index > self.upper:
@@ -161,6 +170,30 @@ class RangePredicate():
         return False
 
 
+class UniquePredicate():
+    """Predicate; True if given URL has not been encountered before"""
+    def __init__(self):
+        self.urls = set()
+
+    def __call__(self, url, kwds):
+        if url not in self.urls:
+            self.urls.add(url)
+            return True
+        return False
+
+
+class ChainPredicate():
+    """Predicate; True if all of its predicates return True"""
+    def __init__(self, predicates):
+        self.predicates = predicates
+
+    def __call__(self, url, kwds):
+        for pred in self.predicates:
+            if not pred(url, kwds):
+                return False
+        return True
+
+
 class PathFormat():
 
     def __init__(self, extractor):