def flatten_interfaces(node):
interfaces = ASTNode('Interfaces')
# Find all InterfaceType nodes.
int_types = flatten(node, 'Interfaces', 'InterfaceType')
# Flatten each InterfaceType name.
for int_type in int_types.children:
typ = ASTNode('Type', None,
[flatten(int_type, 'ReferenceType', 'Identifier')])
interfaces.add(typ)
return interfaces
def build_ast(parse_tree):
ast = ASTNode('Root')
# Start with CompilationUnit.
unit = parse_tree.find_child('CompilationUnit')
if unit is not None:
cu = ASTNode('CompilationUnit')
ast.add(build_top_level(ASTNode('CompilationUnit'), unit))
# Make sure that we use AST nodes.
# ensure_ast_children(ast)
return ast
def flatten(node, root_name, leaf_name):
root = ASTNode(root_name)
stack = [node]
while len(stack) > 0:
n = stack.pop()
if n.name == leaf_name:
root.add(n)
else:
stack.extend(n.children)
root.children.reverse()
return root
def flatten_leaves(node, root_name):
root = ASTNode(root_name)
stack = node.children
while len(stack) > 0:
n = stack.pop()
if len(n.children) == 0:
root.add(n)
else:
stack.extend(n.children)
root.children.reverse()
return root
def build_statement_wts(node):
if node[0].name == 'Block':
return build_block(node[0])
elif node[0].name == 'EmptyStatement':
return ASTNode('EmptyStatement')
# ExpressionStatement => StatementExpression.
elif node[0].name == 'ExpressionStatement':
expr_stmt = ASTNode('ExpressionStatement')
expr_stmt.add(build_expr(node[0][0]))
return expr_stmt
else: # ReturnStatement
return_stmt = ASTNode('ReturnStatement')
if node[0][1].name == 'Expression':
return_stmt.add(build_expr(node[0][1]))
return return_stmt
def build_block(node):
block = ASTNode('Block')
# If no statements, return empty block.
if node[1].name != 'BlockStatements':
return block
blk_stmts = flatten(node[1], 'Statements', 'BlockStatement')
for blk_stmt in blk_stmts.children:
# Local variable declaration.
if blk_stmt[0].name == 'LocalVariableDeclarationStatement':
block.add(build_local_variable_declaration(blk_stmt[0][0]))
else: # Statement
block.add(build_statement(blk_stmt[0]))
return block
def build_local_variable_declaration(node):
local_var = ASTNode('LocalVariableDeclaration')
local_var.add(build_type(node[0]))
# Extract name and initializer.
var_declr = node[1][0]
local_var.add(var_declr[0][0])
if len(var_declr.children) > 1:
initializer = build_expr(var_declr[2][0])
local_var.add(ASTNode('Initializer', None, [initializer]))
else:
local_var.add(ASTNode('Initializer')) # No initializer
return local_var
def build_array_access(node):
array_access = ASTNode('ArrayAccess')
# Determine the receiver (i.e., array being accessed).
array_receiver = ASTNode('ArrayReceiver')
if node[0].name == 'Name':
array_receiver.add(flatten(node[0], 'Name', 'Identifier'))
else:
array_receiver.add(build_primary(node[0]))
array_access.add(array_receiver)
# Get the accessing expression.
array_access.add(build_expr(node[2]))
return array_access
def build_field_access(node):
field_access = ASTNode('FieldAccess')
field_access.add(ASTNode('FieldName', None, [node[2]]))
if node[0].name == 'Primary':
field_access.add(ASTNode('FieldReceiver', None, [build_primary(node[0])]))
else: # Super
field_access.add(ASTNode('FieldReceiver', None, [node[0]]))
return field_access
def build_type(node):
typ = ASTNode('Type')
# return node
if node[0].name == 'PrimitiveType':
typ.add(flatten_leaves(node[0], 'PrimitiveType'))
elif node[0][0].name == 'ClassOrInterfaceType':
typ.add(flatten(node[0][0], 'ReferenceType', 'Identifier'))
elif node[0][0][0].name == 'PrimitiveType': # ArrayType of PrimitiveType
typ.add(ASTNode('ArrayType', None,
[flatten_leaves(node[0][0][0], 'PrimitiveType')]))
else: # ArrayType of ClassOrInterfaceType
typ.add(ASTNode('ArrayType', None,
[flatten(node[0][0][0], 'ReferenceType', 'Identifier')]))
return typ
def build_if_statement(node):
stmt = ASTNode('IfStatement')
stmt.add(build_expr(node[2])) # Condition
stmt.add(build_statement(node[4]))
if len(node.children) > 5: # Else
stmt.add(build_statement(node[6]))
return stmt
def build_parameters(node):
params = ASTNode('Parameters')
# Get a list of FormalParameter.
param_decls = flatten(node, 'Parameters', 'FormalParameter')
for param_decl in param_decls.children:
param = ASTNode('Parameter')
param.add(build_type(param_decl[0]))
param.add(param_decl[1][0]) # Name
params.add(param)
return params
def build_interface_structure(node):
decl_node = ASTNode('InterfaceDeclaration')
# Add modifiers and name (same for classes and interfaces).
decl_node.add(flatten_leaves(node[0], 'Modifiers'))
decl_node.add(ASTNode('InterfaceName', None, [node[2]]))
# Extract interface extends.
interfaces = ASTNode('Interfaces')
decl_ints = node.find_child('ExtendsInterfaces')
if decl_ints is not None:
interfaces = flatten_interfaces(decl_ints)
decl_node.add(interfaces)
body = node.find_child('InterfaceBody')
if body is None:
logging.error("AST: missing InterfaceBody")
sys.exit(1)
# Joos only allows methods to be in interfaces.
methods = flatten(body, 'Methods', 'InterfaceMemberDeclaration')
decl_node.add(build_methods(methods))
return decl_node
def build_assignment(node):
assignment = ASTNode('Assignment')
assignment.add(build_left_hand_side(node[0]))
assignment.add(build_expr(node[2]))
return assignment
def build_arguments(node):
arguments = ASTNode('Arguments')
args = flatten(node, 'Arguments', 'Expression')
for arg in args:
arguments.add(build_expr(arg))
return arguments
def build_constructors(node):
constructors = ASTNode('Constructors')
for cons_decl in node.children:
cons = ASTNode('ConstructorDeclaration')
if hasattr(cons_decl, 'decl_order'):
cons.decl_order = cons_decl.decl_order
else:
cons.decl_order = -1
# Extract modifiers.
cons.add(flatten_leaves(cons_decl[0], 'Modifiers'))
# Extract name.
cons.add(cons_decl[1][0][0]) # Name
# Extract parameters.
if cons_decl[1][2].name == 'FormalParameterList':
cons.add(build_parameters(cons_decl[1][2]))
else:
cons.add(ASTNode('Parameters'))
# Extract body.
cons.add(build_block(cons_decl[2]))
constructors.add(cons)
return constructors
def build_cast_expression(node):
cast_expr = ASTNode('CastExpression')
typ = ASTNode('Type')
if node[2].name == 'Dims': # array type
if node[1].name == 'PrimitiveType':
typ.add(ASTNode('ArrayType', None,
[flatten_leaves(node[1], 'PrimitiveType')]))
else: # Name
typ.add(ASTNode('ArrayType', None,
[flatten(node[1], 'ReferenceType', 'Identifier')]))
elif node[1].name == 'Expression': # Actually a name.
typ.add(flatten(node[1], 'ReferenceType', 'Identifier'))
else: # PrimitiveType
typ.add(flatten_leaves(node[1], 'PrimitiveType'))
cast_expr.add(typ)
# Last child is the thing we're casting.
cast_expr.add(build_expr(node[-1]))
return cast_expr
def build_while_statement(node):
stmt = ASTNode('WhileStatement')
stmt.add(build_expr(node[2])) # Condition
stmt.add(build_statement(node[4]))
return stmt
def build_for_statement(node):
stmt = ASTNode('ForStatement')
for_init = node.find_child('ForInit')
for_cond = node.find_child('Expression')
for_updt = node.find_child('ForUpdate')
for_body = node.find_child('StatementNoShortIf')
if for_body is None:
for_body = node.find_child('Statement')
init_node = ASTNode('ForInit')
if for_init is not None:
if for_init[0].name == 'StatementExpressionList':
init_node.add(build_expr(for_init[0][0]))
else: # LocalVariableDeclaration
init_node.add(build_local_variable_declaration(for_init[0]))
stmt.add(init_node)
cond_node = ASTNode('ForCondition')
if for_cond is not None:
cond_node.add(build_expr(for_cond))
stmt.add(cond_node)
updt_node = ASTNode('ForUpdate')
if for_updt is not None:
updt_node.add(build_expr(for_updt[0][0]))
stmt.add(updt_node)
body_node = ASTNode('ForBody')
if for_body is not None:
body_node.add(build_statement(for_body))
stmt.add(body_node)
return stmt
def build_fields(node):
fields = ASTNode('Fields')
for field_decl in node.children:
field = ASTNode('FieldDeclaration')
if hasattr(field_decl, 'decl_order'):
field.decl_order = field_decl.decl_order
else:
field.decl_order = -1
# Extract modifiers.
field.add(flatten_leaves(field_decl[0], 'Modifiers'))
# Extract type.
field.add(build_type(field_decl[1]))
# Extract name.
field.add(field_decl[2][0][0][0])
# Extract initializer.
var_declr = field_decl[2][0] # VariableDeclarator
if len(var_declr.children) > 1:
initializer = build_expr(var_declr[2][0])
field.add(ASTNode('Initializer', None, [initializer]))
else:
field.add(ASTNode('Initializer')) # No initializer
fields.add(field)
return fields
def build_creation_expression(node):
creation = ASTNode('CreationExpression')
# We do some funky stuff to make sure that it matches 'Type'.
typ = ASTNode('Type')
# Note: both creation expression types have ClassType at second index.
# For ArrayCreationExpression, we actually need to nest the type.
if node.name == 'ArrayCreationExpression':
# Type => ArrayType => PrimitiveType
if node[1].name == 'PrimitiveType':
typ.add(ASTNode('ArrayType', None,
[flatten_leaves(node[1], 'PrimitiveType')]))
# Type => ArrayType => ReferenceType
else:
typ.add(ASTNode('ArrayType', None,
[flatten(node[1], 'ReferenceType', 'Identifier')]))
else: # ClassInstanceCreationExpression
if node[1].name == 'PrimitiveType':
typ.add(flatten_leaves(node[1], 'PrimitiveType'))
else:
typ.add(flatten(node[1], 'ReferenceType', 'Identifier'))
creation.add(typ)
# Get the arguments.
args = ASTNode('Arguments')
if node.name == 'ArrayCreationExpression':
if node[2].name == 'DimExprs':
args.add(build_expr(node[2][0][1]))
else: # ClassInstanceCreationExpression
if node[3].name == 'ArgumentList':
args = build_arguments(node[3])
creation.add(args)
return creation
def build_methods(node):
methods = ASTNode('Methods')
for method_decl in node.children:
method = ASTNode('MethodDeclaration')
if hasattr(method_decl, 'decl_order'):
method.decl_order = method_decl.decl_order
else:
method.decl_order = -1
# Extract modifiers.
method.add(flatten_leaves(method_decl[0][0], 'Modifiers'))
# Extract return type.
if method_decl[0][1].name == 'Void':
method.add(ASTNode('Type', None, [method_decl[0][1]]))
else:
method.add(build_type(method_decl[0][1])) # Non-void.
# Name.
method.add(method_decl[0][2][0])
# Extract parameters.
if method_decl[0][2][2].name == 'FormalParameterList':
method.add(build_parameters(method_decl[0][2][2]))
else:
method.add(ASTNode('Parameters'))
# Extract body.
# We have two levels, for differentiating between:
# 1. Methods with no body (i.e., abstract)
# 2. Methods with an empty body (i.e., {})
# 3. Methods with a nonempty body
body = ASTNode('MethodBody')
if method_decl[1].name != 'SemiColon' and \
method_decl[1][0].name != 'SemiColon':
body.add(build_block(method_decl[1][0]))
method.add(body)
methods.add(method)
return methods
def build_class_structure(node):
decl_node = ASTNode('ClassDeclaration')
# Add modifiers and name (same for classes and interfaces).
decl_node.add(flatten_leaves(node[0], 'Modifiers'))
decl_node.add(ASTNode('ClassName', None, [node[2]]))
# Extract superclass.
superstuff = node.find_child('SuperStuff')
if superstuff is None:
decl_node.add(ASTNode('Superclass'))
else:
typ = ASTNode('Type', None,
[flatten(superstuff, 'ReferenceType', 'Identifier')])
decl_node.add(ASTNode('Superclass', None, [typ]))
# Extract interface implements.
interfaces = ASTNode('Interfaces')
decl_ints = node.find_child('Interfaces')
if decl_ints is not None:
interfaces = flatten_interfaces(decl_ints)
decl_node.add(interfaces)
body = node.find_child('ClassBody')
if body is None:
logging.error("AST: missing ClassBody")
sys.exit(1)
# Extract fields, constructors, and methods.
members = flatten(body, 'Members', 'ClassBodyDeclaration')
fields = flatten(members, 'Fields', 'FieldDeclaration')
decl_node.add(build_fields(fields))
constructors = flatten(members, 'Constructors', 'ConstructorDeclaration')
decl_node.add(build_constructors(constructors))
methods = flatten(members, 'Methods', 'MethodDeclaration')
decl_node.add(build_methods(methods))
return decl_node
def build_method_invocation(node):
method_invo = ASTNode('MethodInvocation')
# First, extract the method name and receiver (i.e., the "thing we're
# calling the method on").
method_receiver = ASTNode('MethodReceiver')
if node[0].name == 'Name':
qualified_name = flatten(node[0], 'MethodName', 'Identifier')
# If the name is qualified (i.e., more than 1 Identifier), we take the
# last Identifier to be the method name. There is a method receiver
# only if there are at least 2 identifiers.
# Note that the latter case of no method receivers declared is possible
# since Joos allows "implicit this for methods".
if len(qualified_name.children) == 1:
method_invo.add(qualified_name)
else:
method_name = qualified_name.children.pop()
method_invo.add(ASTNode('MethodName', None, [method_name]))
qualified_name.name = 'Name'
method_receiver.add(qualified_name)
else: # Name is at position 2, receiver is the Primary at position 0.
method_invo.add(ASTNode('MethodName', None, [node[2]]))
method_receiver.add(build_primary(node[0]))
method_invo.add(method_receiver)
method_args = node.find_child('ArgumentList')
if method_args is not None:
method_invo.add(build_arguments(method_args))
else:
method_invo.add(ASTNode('Arguments'))
return method_invo