Fix infinit loop when filename contains multiple languages ISO codes

This also adds ice as a common english word, avoiding detected ice as icelandic language. Close #170
guessit-io · Nov 10, 2014 · 4d1d40e · 4d1d40e
1 parent 7166dd7
commit 4d1d40e
Show file tree

Hide file tree

Showing 3 changed files with 33 additions and 9 deletions.
diff --git a/guessit/language.py b/guessit/language.py
@@ -169,7 +169,7 @@ def reverse(self, name):
     'no', 'non', 'war', 'min', 'new', 'car', 'day', 'bad', 'bat', 'fan',
     'fry', 'cop', 'zen', 'gay', 'fat', 'one', 'cherokee', 'got', 'an', 'as',
     'cat', 'her', 'be', 'hat', 'sun', 'may', 'my', 'mr', 'rum', 'pi', 'bb', 'bt',
-    'tv', 'aw', 'by', 'md', 'mp', 'cd', 'lt', 'gt', 'in', 'ad',
+    'tv', 'aw', 'by', 'md', 'mp', 'cd', 'lt', 'gt', 'in', 'ad', 'ice',
     # french words
     'bas', 'de', 'le', 'son', 'ne', 'ca', 'ce', 'et', 'que',
     'mal', 'est', 'vol', 'or', 'mon', 'se', 'je', 'tu', 'me',

diff --git a/guessit/matcher.py b/guessit/matcher.py
@@ -199,6 +199,15 @@ def log_found_guess(guess, logger=None):
                               (k, v, guess.raw(k), guess.confidence(k)))
 
 
+def _get_split_spans(node, span):
+    partition_spans = node.get_partition_spans(span)
+    for to_remove_span in partition_spans:
+        if to_remove_span[0] == span[0] and to_remove_span[1] in [span[1], span[1] + 1]:
+            partition_spans.remove(to_remove_span)
+            break
+    return partition_spans
+
+
 class GuessFinder(object):
     def __init__(self, guess_func, confidence=None, logger=None, options=None):
         self.guess_func = guess_func
@@ -243,13 +252,16 @@ def process_node(self, node, iterative=True, partial_span=None):
                     for skip_node in skip_nodes:
                         if skip_node.parent.node_idx == node.node_idx[:len(skip_node.parent.node_idx)] and\
                             skip_node.span == span or\
-                            skip_node.span == (span[0] + skip_node.offset, span[1] + skip_node.offset):
-                            partition_spans = node.get_partition_spans(skip_node.span)
-                            for to_remove_span in partition_spans:
-                                if to_remove_span[0] == skip_node.span[0] and to_remove_span[1] in [skip_node.span[1], skip_node.span[1] + 1]:
-                                    partition_spans.remove(to_remove_span)
-                                    break
-                            #break
+                                skip_node.span == (span[0] + skip_node.offset, span[1] + skip_node.offset):
+                            if partition_spans is None:
+                                partition_spans = _get_split_spans(node, skip_node.span)
+                            else:
+                                new_partition_spans = []
+                                for partition_span in partition_spans:
+                                    tmp_node = MatchTree(value, span=partition_span, parent=node)
+                                    tmp_partitions_spans = _get_split_spans(tmp_node, skip_node.span)
+                                    new_partition_spans.extend(tmp_partitions_spans)
+                                partition_spans.extend(new_partition_spans)
 
                 if not partition_spans:
                     # restore sentinels compensation

diff --git a/guessit/test/autodetect.yaml b/guessit/test/autodetect.yaml
@@ -458,4 +458,16 @@
   series: The Flash
   type: episode
   videoCodec: XviD
-  year: 2014
+  year: 2014
+
+? Ice.Lake.Rebels.S01E06.Ice.Lake.Games.720p.HDTV.x264-DHD
+: options: -n
+  episodeNumber: 6
+  format: HDTV
+  releaseGroup: DHD
+  screenSize: 720p
+  season: 1
+  series: Ice Lake Rebels
+  title: Ice Lake Games
+  type: episode
+  videoCodec: h264