Merge pull request #140 from kaorahi/old_sgf

accept old SGF

katrain/core/sgf_parser.py

@@ -132,7 +132,9 @@ class SGFNode:
 
     def add_list_property(self, property: str, values: List):
         """Add some values to the property list."""
-        self.properties[property] += values
+        # SiZe[19] ==> SZ[19] etc. for old SGF
+        normalized_property = re.sub("[a-z]", "", property)
+        self.properties[normalized_property] += values
 
     def get_list_property(self, property, default=None) -> Any:
         """Get the list of values for a property."""
@@ -323,11 +325,14 @@ class SGF:
     _NODE_CLASS = SGFNode  # Class used for SGF Nodes, can change this to something that inherits from SGFNode
     # https://xkcd.com/1171/
     SGFPROP_PAT = re.compile(r"\s*(?:\(|\)|;|(\w+)((\s*\[([^\]\\]|\\.)*\])+))", flags=re.DOTALL)
+    SGF_PAT = re.compile(r"\(;.*\)", flags=re.DOTALL)
 
     @classmethod
     def parse_sgf(cls, input_str) -> SGFNode:
         """Parse a string as SGF."""
-        return cls(input_str).root
+        match = re.search(cls.SGF_PAT, input_str)
+        clipped_str = match.group() if match else input_str
+        return cls(clipped_str).root
 
     @classmethod
     def parse_file(cls, filename, encoding=None) -> SGFNode:
@@ -375,7 +380,10 @@ class SGF:
             if matched_item == "(":
                 self._parse_branch(self._NODE_CLASS(parent=current_move))
             elif matched_item == ";":
-                if not current_move.empty:  # ignore ; that generate empty nodes
+                # ignore ;) for old SGF
+                useless = self.ix < len(self.contents) and self.contents[self.ix] == ")"
+                # ignore ; that generate empty nodes
+                if not (current_move.empty or useless):
                     current_move = self._NODE_CLASS(parent=current_move)
             else:
                 property, value = match[1], match[2].strip()[1:-1]

tests/data/xmgt97.sgf

@@ -0,0 +1,133 @@
+(
+;
+FileFormat[1]GaMe[1]
+VieW[]
+SiZe[9]
+PlayerBlack[hi]
+PlayerWhite[ka]
+KoMi[-1.5]
+Black[ff]
+;
+White[gd]
+;
+Black[fd]
+;
+White[fe]
+;
+Black[ge]
+;
+White[ee]
+;
+Black[gc]
+;
+White[hd]
+;
+Black[hc]
+;
+White[he]
+;
+Black[gf]
+;
+White[hf]
+;
+Black[ed]
+;
+White[ef]
+;
+Black[eg]
+;
+White[dg]
+;
+Black[fg]
+;
+White[gh]
+;
+Black[cg]
+;
+White[dh]
+;
+Black[df]
+;
+White[de]
+;
+Black[cf]
+;
+White[dd]
+;
+Black[dc]
+;
+White[ec]
+;
+Black[fc]
+;
+White[cc]
+;
+Black[eb]
+;
+White[be]
+;
+Black[bf]
+;
+White[ce]
+;
+Black[bd]
+;
+White[ae]
+;
+Black[af]
+;
+White[db]
+;
+Black[ec]
+;
+White[bc]
+;
+Black[eh]
+;
+White[gg]
+;
+Black[ch]
+;
+White[fh]
+;
+Black[di]
+;
+White[ic]
+;
+Black[cb]
+;
+White[bb]
+;
+Black[da]
+;
+White[hb]
+;
+Black[gb]
+;
+White[ba]
+;
+Black[ca]
+;
+White[fi]
+;
+Black[id]
+;
+White[ib]
+;
+Black[ha]
+;
+White[ie]
+;
+Black[ga]
+;
+White[hg]
+;
+Black[ia]
+;
+Comment[Black: 13 + 4 = 17
+
+White: 13 + 2 + -1.5 = 13.5
+
+Black wins by 3.5]
+White[id]
+)

tests/test_parser.py

@@ -59,7 +59,7 @@ def test_pandanet():
         "GM",
         "EV",
         "US",
-        "CoPyright",
+        "CP",
         "GN",
         "RE",
         "PW",
@@ -91,6 +91,16 @@ def test_pandanet():
     assert move is root
 
 
+def test_old_long_properties():
+    file = os.path.join(os.path.dirname(__file__), "data/xmgt97.sgf")
+    SGF.parse_file(file)
+
+
+def test_old_server_style():
+    input_sgf = "... 01:23:45 +0900 (JST) ... (;SZ[19];B[aa];W[ba];)"
+    SGF.parse_sgf(input_sgf)
+
+
 def test_ogs():
     file = os.path.join(os.path.dirname(__file__), "data/ogs.sgf")
     tree = SGF.parse_file(file)