GumTree Serialization

Author: cedricrupb

code_diff/gumtree/__init__.py

@@ -2,28 +2,40 @@
 from .editmap  import gumtree_editmap
 from .chawathe import compute_chawathe_edit_script
 from .ops      import (Update, Insert, Delete, Move)
+from .ops      import serialize_script
 
 # Edit script ----------------------------------------------------------------
 
-def compute_edit_script(source_ast, target_ast, min_height = 1):
+def compute_edit_script(source_ast, target_ast, min_height = 1, max_size = 1000, min_dice = 0.5):
 
     # If source_ast and target_ast only leaves
     if len(source_ast.children) == 0 and len(target_ast.children) == 0:
-        return [_update_leaf(source_ast, target_ast)]
+        return EditScript([_update_leaf(source_ast, target_ast)])
 
     isomap = gumtree_isomap(source_ast, target_ast, min_height)
 
     while len(isomap) == 0 and min_height > 0:
         min_height -= 1
         isomap = gumtree_isomap(source_ast, target_ast, min_height)
 
-    editmap = gumtree_editmap(isomap, source_ast, target_ast)
+    editmap = gumtree_editmap(isomap, source_ast, target_ast, max_size, min_dice)
     editscript = compute_chawathe_edit_script(editmap, source_ast, target_ast)
 
-    return editscript
-    
+    return EditScript(editscript)
 
+    
 # Update leaf ----------------------------------------------------------------
 
 def _update_leaf(source_ast, target_ast):
-    return Update(source_ast, target_ast.text)
+    return Update(source_ast, target_ast.text)
+
+
+# Edit script ----------------------------------------------------------------
+
+class EditScript(list):
+
+    def __init__(self, operations):
+        super().__init__(operations)
+
+    def __repr__(self):
+        return serialize_script(self)

code_diff/gumtree/chawathe.py

@@ -33,7 +33,8 @@ def compute_chawathe_edit_script(editmap, source, target):
             op = Insert(
                     parent_partner.delegate,
                     (target_node.type, target_node.text),
-                    k
+                    k,
+                    -1
                 )
             edit_script.append(op)
             node = parent_partner.apply(op)
@@ -150,10 +151,15 @@ def _partner_child(c, o, src_partner = False):
 
 class InsertNode:
 
+    INSERT_COUNT = 0
+
     def __init__(self, type, text = None, children = None):
         self.type = type
         self.text = text
 
+        self.node_id = InsertNode.INSERT_COUNT
+        InsertNode.INSERT_COUNT += 1
+
         self.parent = None 
         self.children = children if children is not None else []
 
@@ -226,6 +232,7 @@ def apply(self, operation):
 
         if isinstance(operation, Insert):
             node = InsertNode(*operation.node)
+            operation.insert_id = node.node_id
             wn   = self.src._access_wn(node)
             self.children.insert(operation.position, wn)
 

code_diff/gumtree/ops.py

@@ -13,6 +13,7 @@ class Update(EditOperation):
 class Insert(EditOperation):
     node: Tuple[str, Any]
     position: int
+    insert_id: int # This is necessary to keep track of nodes (TODO: Better solution?)
 
 @dataclass
 class Move(EditOperation):
@@ -21,4 +22,70 @@ class Move(EditOperation):
 
 @dataclass
 class Delete(EditOperation):
-    pass
+    pass
+
+
+# Serialization --------------------------------
+
+import json
+
+
+def _serialize_new_node(new_node_index, node):
+    
+    if node.node_id not in new_node_index:
+        new_node_index[node.node_id] = len(new_node_index)
+
+    return "N%d" % new_node_index[node.node_id]
+
+def _serialize_ast_node(node):
+   position  = node.position
+   node_text = node.type
+
+   if node.text: node_text += ":" + node.text
+   
+   return "(%s, line %d:%d - %d:%d)" % (node_text, position[0][0], position[0][1], position[1][0], position[1][1]) 
+
+
+def _serialize_node(new_node_index, node):
+    
+    if hasattr(node, 'node_id'):
+        return _serialize_new_node(new_node_index, node)
+    
+    return _serialize_ast_node(node)
+
+
+def serialize_script(edit_script):
+    
+    sedit_script = []
+    new_node_index = {}
+
+    for operation in edit_script:
+
+        operation_name = operation.__class__.__name__
+        target_node_str = _serialize_node(new_node_index, operation.target_node)
+
+        if operation_name == "Update":
+            sedit_script.append("%s(%s, %s)" % (operation_name, target_node_str, operation.value))
+        
+        elif operation_name == "Insert":
+            
+            new_node = operation.node
+
+            if new_node[1] is None:
+                new_node_index[operation.insert_id] = len(new_node_index)
+                new_node_str = "(%s, %s)" % (new_node[0], "N%d" % new_node_index[operation.insert_id])
+            else: # Leaf node
+                new_node_str = "%s:%s" % new_node
+
+            sedit_script.append("%s(%s, %s, %d)" % (operation_name, target_node_str, new_node_str, operation.position))
+
+        elif operation_name == "Move":
+
+            new_node_str = _serialize_node(new_node_index, operation.node)
+
+            sedit_script.append("%s(%s, %s, %d)" % (operation_name, target_node_str, new_node_str, operation.position))
+
+        elif operation_name == "Delete":
+            sedit_script.append("%s(%s)" % (operation_name, target_node_str))
+
+    return json.dumps(sedit_script, indent=2)