fix bug of matching

Author: cfhamlet

src/os_urlpattern/cmdline.py

@@ -15,7 +15,7 @@
                          IrregularURLException)
 from .formatter import FORMATTERS
 from .pattern_maker import PatternMaker
-from .pattern_matcher import PatternMatcher
+from .pattern_matcher import PatternMatcher, most_matched
 from .utils import LogSpeedAdapter, load_obj, pretty_counter
 
 
@@ -156,6 +156,12 @@ def add_argument(self, parser):
                             action='store',
                             dest='pattern_file')
 
+        parser.add_argument('-a', '--all_matched',
+                            help='all matched patterns',
+                            default=False,
+                            action='store_true',
+                            dest='all_matched')
+
     def _load(self, pattern_matcher, args):
         stats = Counter()
         io_input = args.pattern_file[0]
@@ -178,11 +184,13 @@ def _load(self, pattern_matcher, args):
         pattern_matcher.preprocess()
         self._logger.debug('[PREPROCESS] Finished')
 
-    def _match_result(self, pattern_matcher, raw_url):
+    def _match_result(self, pattern_matcher, raw_url, args):
         result = None
         try:
             url = raw_url.decode(DEFAULT_ENCODING)
             result = pattern_matcher.match(url)
+            if not args.all_matched:
+                result = most_matched(result)
             result = ", ".join([r.info['ptn'] for r in result])
         except (InvalidPatternException,
                 IrregularURLException,
@@ -201,7 +209,7 @@ def _match(self, pattern_matcher, args):
         for line in args.file[0]:
             speed_logger.debug('[MATCHING]')
             line = line.strip()
-            result = self._match_result(pattern_matcher, line)
+            result = self._match_result(pattern_matcher, line, args)
             if not result:
                 result = b'N'
             binary_stdout.write(result)

src/os_urlpattern/pattern_matcher.py

@@ -55,7 +55,7 @@ def match(self, parsed_piece):
             return []
         parsed_pieces = [EMPTY_PARSED_PIECE, ]
         parsed_pieces.extend(view.parsed_pieces)
-        return [n.info for n in self._nodes[view.view].match(parsed_pieces)]
+        return [n.info for n in self._nodes[view.view].match(parsed_pieces, match_all=True)]
 
 
 class PiecePatternViewMatcher(_ViewMatcher):
@@ -131,6 +131,7 @@ def match(self, parsed_piece):
 
 
 class PatternMatchNode(object):
+    __slot__ = ()
 
     def __init__(self, pattern, info=None):
         self._pattern = pattern
@@ -161,16 +162,21 @@ def preprocess(self):
     def iter_children(self):
         return itervalues(self._children)
 
-    def match(self, parsed_pieces, idx, matched_nodes):
+    def match(self, parsed_pieces, idx, matched_nodes, match_all=False):
         parsed_piece = parsed_pieces[idx]
         for matcher in itervalues(self._view_matchers):
             nodes = matcher.match(parsed_piece)
-            for node in nodes:
-                if node.leaf():
-                    matched_nodes.append(node)
+            self._deep_match(nodes, parsed_pieces, idx,
+                             matched_nodes, match_all)
+
+    def _deep_match(self, nodes, parsed_pieces, idx, matched_nodes, match_all):
+        for node in nodes:
+            if node.leaf():
+                matched_nodes.append(node)
+                if not match_all:
                     return
-                else:
-                    node.match(parsed_pieces, idx+1, matched_nodes)
+            else:
+                node.match(parsed_pieces, idx + 1, matched_nodes, match_all)
 
     @property
     def pattern(self):
@@ -218,9 +224,9 @@ def load_from_patterns(self, patterns, info):
     def root(self):
         return self._root
 
-    def match(self, parsed_pieces):
+    def match(self, parsed_pieces, match_all=False):
         matched_nodes = []
-        self._root.match(parsed_pieces, 0, matched_nodes)
+        self._root.match(parsed_pieces, 0, matched_nodes, match_all)
         return matched_nodes
 
     def preprocess(self):
@@ -253,3 +259,7 @@ def match(self, url):
         if sid in self._pattern_match_trees:
             return self._pattern_match_trees[sid].match(parsed_pieces)
         return []
+
+
+def most_matched(matched_nodes):
+    return matched_nodes[0:1]