#!/usr/bin/env python # -*- coding: utf-8 -*- """ OpenGEODE SDL92 parser This library builds the SDL AST (described in ogAST.py) The AST can then be used to build SDL backends such as the diagram editor (placing symbols in a graphical canvas for editition) or code generators, etc. The AST build is based on the ANTLR-grammar and generated lexer and parser (the grammar is in the file sdl92.g and requires antlr 3.1.3 for Python to be compiled and used). During the build of the AST this library makes a number of semantic checks on the SDL input mode. Copyright (c) 2012-2013 European Space Agency Designed and implemented by Maxime Perrotin Contact: maxime.perrotin@esa.int """ __author__ = 'Maxime Perrotin' import sys import os import logging import traceback from itertools import chain, permutations, combinations from collections import defaultdict import antlr3 import antlr3.tree import sdl92Lexer as lexer from sdl92Parser import sdl92Parser import samnmax import ogAST from Asn1scc import parse_asn1, ASN1 LOG = logging.getLogger(__name__) EXPR_NODE = { lexer.PLUS: ogAST.ExprPlus, lexer.ASTERISK: ogAST.ExprMul, lexer.IMPLIES: ogAST.ExprImplies, lexer.DASH: ogAST.ExprMinus, lexer.OR: ogAST.ExprOr, lexer.AND: ogAST.ExprAnd, lexer.XOR: ogAST.ExprXor, lexer.EQ: ogAST.ExprEq, lexer.NEQ: ogAST.ExprNeq, lexer.GT: ogAST.ExprGt, lexer.GE: ogAST.ExprGe, lexer.LT: ogAST.ExprLt, lexer.LE: ogAST.ExprLe, lexer.DIV: ogAST.ExprDiv, lexer.MOD: ogAST.ExprMod, lexer.APPEND: ogAST.ExprAppend, lexer.IN: ogAST.ExprIn, lexer.REM: ogAST.ExprRem, lexer.NOT: ogAST.ExprNot, lexer.NEG: ogAST.ExprNeg, lexer.PRIMARY: ogAST.Primary, } # Insert current path in the search list for importing modules sys.path.insert(0, '.') DV = None # Code generator backends may need some intemediate variables to process # expressions. For convenience and to avoid multiple pass parsing, the parser # tries to guess where they may be useful, and adds a hint in the AST. TMPVAR = 0 # ASN.1 types used to support the signature of special operators INTEGER = type('IntegerType', (object,), {'kind': 'IntegerType', 'Min': str(-(2 ** 63)), 'Max': str(2 ** 63 - 1)}) INT32 = type('Integer32Type', (object,), {'kind': 'Integer32Type', 'Min': '-2147483648', 'Max': '2147483647'}) NUMERICAL = type('NumericalType', (object,), {'kind': 'Numerical'}) TIMER = type('TimerType', (object,), {'kind': 'TimerType'}) REAL = type('RealType', (object,), {'kind': 'RealType', 'Min': str(1e-308), 'Max': str(1e308)}) LIST = type('ListType', (object,), {'kind': 'ListType'}) ANY_TYPE = type('AnyType', (object,), {'kind': 'AnyType'}) CHOICE = type('ChoiceType', (object,), {'kind': 'ChoiceType'}) BOOLEAN = type('BooleanType', (object,), {'kind': 'BooleanType'}) RAWSTRING = type('RawString', (object,), {'kind': 'StandardStringType'}) OCTETSTRING = type('OctetString', (object,), {'kind': 'OctetStringType'}) ENUMERATED = type('EnumeratedType', (object,), {'kind': 'EnumeratedType'}) UNKNOWN_TYPE = type('UnknownType', (object,), {'kind': 'UnknownType'}) # Special SDL operators and signature SPECIAL_OPERATORS = {'length': [LIST], 'write': [ANY_TYPE], 'writeln': [ANY_TYPE], 'present': [CHOICE], 'set_timer': [INTEGER, TIMER], 'reset_timer': [TIMER], 'abs': [NUMERICAL], 'num': [ENUMERATED], 'float': [NUMERICAL], 'fix': [NUMERICAL], 'power': [NUMERICAL, INTEGER]} # Container to keep a list of types mapped from ANTLR Tokens # (Used with singledispatch/visitor pattern) ANTLR_TOKEN_TYPES = {a: type(a, (antlr3.tree.CommonTree,), {}) for a, b in lexer.__dict__.viewitems() if type(b) == int} # Shortcut to create a new referenced ASN.1 type new_ref_type = lambda refname: \ type(str(refname), (object,), {'kind': 'ReferenceType', 'ReferencedTypeName': refname.replace('_', '-')}) # Shortcut to return a type name (Reference name or basic type) type_name = lambda t: \ t.kind if t.kind != 'ReferenceType' else t.ReferencedTypeName types = lambda: getattr(DV, 'types', {}) def is_integer(ty): ''' Return true if a type is an Integer Type ''' return find_basic_type(ty).kind in ( 'IntegerType', 'Integer32Type' ) def is_numeric(ty): ''' Return true if a type is a Numeric Type ''' return find_basic_type(ty).kind in ( 'IntegerType', 'Integer32Type', 'Numerical', 'RealType' ) def is_string(ty): ''' Return true if a type is a String Type ''' return find_basic_type(ty).kind in ( 'StandardStringType', 'OctetStringType' ) def sdl_to_asn1(sort): ''' Convert case insensitive type reference to the actual type as found in the ASN.1 datamodel ''' for asn1_type in types().viewkeys(): if sort.replace('_', '-').lower() == asn1_type.lower(): break else: raise TypeError('Type {} not found in ASN.1 model'.format(sort)) return new_ref_type(asn1_type) def node_filename(node): ''' Return the filename associated to the stream of this node ''' parent = node while parent: try: return parent.getToken().getInputStream().fileName except AttributeError: parent = parent.getParent() return None def token_stream(node): ''' Return the token stream associated to a tree node It is set at the root of the tree by the parser ''' parent = node while parent: try: return parent.token_stream except AttributeError: parent = parent.getParent() def signals_in_system(ast): ''' Recursively find signal definitions in a nested SDL model ''' all_signals = [] for block in ast.blocks: all_signals.extend(signals_in_system(block)) all_signals.extend(ast.signals) return all_signals def find_process_declaration(ast, process_name): ''' Recursively search for a process declaration in a nested SDL model ''' for block in ast.blocks: result = find_process_declaration(block, process_name) if result: return result try: for process in ast.processes: if process.processName == process_name: return process except AttributeError: return None return None def valid_output(scope): ''' Yields the output, procedures, and operators names, that is all the elements that can be valid in an OUTPUT symbol (does not mean it IS valid - caller still has to check it) ''' for out_sig in scope.output_signals: yield out_sig['name'].lower() for proc in scope.procedures: yield proc.inputString.lower() for special_op in SPECIAL_OPERATORS: yield special_op.lower() def get_interfaces(ast, process_name): ''' Search for the list of input and output signals (async PI/RI) and procedures (sync RI) of a process in a given top-level AST ''' all_signals = [] async_signals = [] system = None # Move up to the system level, in case process is nested in a block # and not defined at root level as it is the case when it is referenced system = ast while hasattr(system, 'parent'): system = system.parent # If we are at AST level, check in all systems, otherwise in current one iterator = ast.systems if hasattr(ast, 'systems') else (system,) for system in iterator: all_signals.extend(signals_in_system(system)) process_ref = find_process_declaration(system, process_name) if process_ref: # Go to the block where the process is defined process_parent = process_ref.parent break else: if isinstance(ast, ogAST.Block): process_parent = ast else: raise TypeError('Process ' + process_name + ' is defined but not not declared in a system') # Find in and out signals names using the signalroutes for signalroute in process_parent.signalroutes: for route in signalroute['routes']: if route['source'] == process_name: direction = 'out' elif route['dest'] == process_name: direction = 'in' else: continue for sig_id in route['signals']: # Copy the signal to the result dict found, = [dict(sig) for sig in all_signals if sig['name'] == sig_id] found['direction'] = direction async_signals.append(found) return async_signals, system.procedures def get_input_string(root): ''' Return the input string of a tree node ''' return token_stream(root).toString(root.getTokenStartIndex(), root.getTokenStopIndex()) def error(root, msg): ''' Return an error message ''' return '{} - "{}"'.format(msg, get_input_string(root)) def warning(root, msg): ''' Return a warning message ''' return '{} - "{}"'.format(msg, get_input_string(root)) def tmp(): ''' Return a temporary variable name ''' global TMPVAR varname = TMPVAR TMPVAR += 1 return varname def get_state_list(process_root): ''' Return the list of states of a process ''' # 1) get all STATE statements states = (child for child in process_root.getChildren() if child.type == lexer.STATE) # 2) keep only the ones containing a STATELIST token (i.e. no ASTERISK) relevant = (child for state in states for child in state.getChildren() if child.type == lexer.STATELIST) # 3) extract the state list from each of them state_list = [s.text.lower() for r in relevant for s in r.getChildren()] # state_list.append('START') # 4) create a set to remove duplicates return set(state_list) def find_basic_type(a_type): ''' Return the ASN.1 basic type of a_type ''' basic_type = a_type or UNKNOWN_TYPE while basic_type.kind == 'ReferenceType': # Find type with proper case in the data view for typename in types().viewkeys(): if typename.lower() == basic_type.ReferencedTypeName.lower(): basic_type = types()[typename].type break else: raise TypeError('Type "' + type_name(basic_type) + '" was not found in Dataview') return basic_type def is_constant(var): ''' Check in ASN.1 modules if var (Primary) is declared as a constant ''' if var is None: return False if isinstance(var, ogAST.PrimConstant): return True if DV and isinstance(var, ogAST.PrimVariable): for mod in DV.asn1Modules: for constant in DV.exportedVariables[mod]: if(constant.lower() == var.value[0].lower().replace('_', '-')): LOG.debug('Constant ' + var.inputString + ' found') return True return False def fix_special_operators(op_name, expr_list, context): ''' Verify/fix type of special operators parameters ''' if op_name.lower() in ('length', 'present', 'abs', 'float', 'fix', 'num'): if len(expr_list) != 1: raise AttributeError('Only one parameter for the {} operator' .format(op_name)) expr = expr_list[0] if expr.exprType is UNKNOWN_TYPE: expr.exprType = find_variable(expr.value[0], context) # XXX should change type to PrimVariable basic = find_basic_type(expr.exprType) if op_name.lower() == 'length' and basic.kind != 'SequenceOfType' \ and not is_string(basic): raise TypeError('Length operator works only on strings/lists') elif op_name.lower() == 'present' and basic.kind != 'ChoiceType': raise TypeError('Present operator works only on CHOICE types') elif op_name.lower() in ('abs', 'float', 'fix') \ and not is_numeric(basic): raise TypeError('"{}" operator needs a numerical parameter'.format( op_name)) elif op_name.lower() == 'num' and basic.kind != 'EnumeratedType': raise TypeError('Num operaror works only with enumerations') elif op_name.lower() == 'power': if len(expr_list) != 2: raise AttributeError('The "power" operator takes two parameters') for idx, expr in enumerate(expr_list): if expr.exprType is UNKNOWN_TYPE: expr.exprType = find_variable(expr.value[0], context) # XXX should change type to PrimVariable if idx == 0 and not is_numeric(expr.exprType): raise TypeError('First parameter of power must be numerical') elif idx == 1 and not is_integer(expr.exprType): raise TypeError('Second parameter of power must be integer') elif op_name.lower() in ('write', 'writeln'): for param in expr_list: if param.exprType is UNKNOWN_TYPE: for each in (INTEGER, REAL, BOOLEAN, RAWSTRING, OCTETSTRING): try: check_type_compatibility(param, each, context) param.exprType = each break except TypeError: continue else: # Type not found among supported types # Has to be a variable...otherwise, error! try: param.exprType = find_variable(param.value[0], context) except (KeyError, AttributeError): raise TypeError('Could not determine type of argument' ' "{}"'.format(param.inputString)) basic = find_basic_type(param.exprType) if basic.kind not in ('IntegerType', 'Integer32Type', 'RealType', 'BooleanType') \ and not basic.kind.endswith('StringType'): # Currently supported printable types raise TypeError('Write operator does not support type') elif op_name.lower() == 'set_timer': if len(expr_list) != 2: raise TypeError('SET_TIMER has 2 parameters: (int, timer_name)') basic = find_basic_type(expr_list[0].exprType) if not basic.kind.startswith('Integer'): raise TypeError('SET_TIMER first parameter is not an integer') timer = expr_list[1].inputString for each in chain(context.timers, context.global_timers): if each.lower() == timer.lower(): break else: raise TypeError('Timer {} is not defined'.format(timer)) elif op_name.lower == 'reset_timer': if len(expr_list) != 1: raise TypeError('RESET_TIMER has 1 parameter: timer_name') timer = expr_list[0].inputString for each in context.timers: if each.lower() == timer.lower(): break else: raise TypeError('Timer {} is not defined'.format(timer)) else: # TODO: other operators return def check_and_fix_op_params(op_name, expr_list, context): ''' Verify and/or set the type of a procedure/output parameters TODO: when supported, add operators ''' # (1) Find the signature of the function # signature will hold the list of parameters for the function LOG.debug('[check_and_fix_op_params] ' + op_name + ' - ' + str(expr_list)) signature = [] key = '' for out_sig in context.output_signals: if out_sig['name'].lower() == op_name.lower(): if out_sig.get('type'): # output signals: one single parameter signature = [{'type': out_sig.get('type'), 'name': out_sig.get('param_name' or ''), 'direction': 'in'}] break else: # Procedures (inner and external) for inner_proc in context.procedures: key = inner_proc.inputString if key.lower() == op_name.lower(): signature = inner_proc.fpar break else: if op_name.lower() not in SPECIAL_OPERATORS: raise AttributeError('Operator/output/procedure not found: ' + op_name) else: # Special operators: parameters are context dependent fix_special_operators(op_name, expr_list, context) return # (2) Check that the number of given parameters matches the signature if signature is not None and len(signature) != len(expr_list): raise TypeError('Wrong number of parameters') # (3) Check each individual parameter type for idx, param in enumerate(expr_list): if signature is None: break # Get parameter type name from the function signature: param_type = type_name(signature[idx].get('type')) # Retrieve the type (or None if it is a sepecial operator) dataview_entry = types().get(param_type) or UNKNOWN_TYPE if dataview_entry is not UNKNOWN_TYPE: dataview_type = new_ref_type(param_type) else: dataview_type = UNKNOWN_TYPE expr = ogAST.ExprAssign() expr.left = ogAST.PrimVariable() expr.left.exprType = dataview_type expr.right = param fix_expression_types(expr, context) expr_list[idx] = expr.right if signature[idx].get('direction') != 'in' \ and not isinstance(expr.right, ogAST.PrimVariable): raise TypeError('OUT parameter "{}" is not a variable' .format(expr.right.inputString)) def check_range(typeref, type_to_check): ''' Verify the that the Min/Max bounds of two types are compatible Called to test that assignments are withing allowed range both types assumed to be basic types ''' try: if float(type_to_check.Min) < float(typeref.Min) \ or float(type_to_check.Max) > float(typeref.Max): raise TypeError('Expression evaluation in range [{}..{}], ' 'outside expected range [{}..{}]' .format(type_to_check.Min, type_to_check.Max, typeref.Min, typeref.Max)) except (AttributeError, ValueError): raise TypeError('Missing range') def check_type_compatibility(primary, typeRef, context): ''' Check if an ogAST.Primary (raw value, enumerated, ASN.1 Value...) is compatible with a given type (typeRef is an ASN1Scc type) Does not return anything if OK, otherwise raises TypeError ''' assert typeRef is not None if typeRef is UNKNOWN_TYPE: raise TypeError('Type reference is unknown') if isinstance(primary, ogAST.PrimConstant): # ASN.1 constants type is unknown (Asn1 backend to be completed) return actual_type = find_basic_type(typeRef) LOG.debug("[check_type_compatibility] " "checking if {value} is of type {typeref}" .format(value=primary.inputString, typeref=type_name(typeRef))) if (isinstance(primary, ogAST.PrimEnumeratedValue) and actual_type.kind.endswith('EnumeratedType')): # If type ref is an enumeration, check that the value is valid # Note, when using the "present" operator of a CHOICE type, the # resulting value is actually an EnumeratedType enumerant = primary.inputString.replace('_', '-') corr_type = actual_type.EnumValues.get(enumerant) if corr_type: return else: err = ('Value "' + primary.inputString + '" not in this enumeration: ' + str(actual_type.EnumValues.keys())) raise TypeError(err) elif isinstance(primary, ogAST.PrimConditional): then_expr = primary.value['then'] else_expr = primary.value['else'] for expr in (then_expr, else_expr): if expr.is_raw: check_type_compatibility(expr, typeRef, context) return elif isinstance(primary, ogAST.PrimVariable): try: compare_types(primary.exprType, typeRef) except TypeError as err: raise TypeError('{expr} should be of type {ty} - {err}' .format(expr=primary.inputString, ty=type_name(typeRef), err=str(err))) return elif isinstance(primary, ogAST.PrimInteger) \ and actual_type.kind.startswith('Integer'): return elif isinstance(primary, ogAST.PrimReal) \ and actual_type.kind.startswith('Real'): return elif isinstance(primary, ogAST.PrimBoolean) \ and actual_type.kind.startswith('Boolean'): return elif (isinstance(primary, ogAST.PrimEmptyString) and actual_type.kind == 'SequenceOfType'): if int(actual_type.Min) == 0: return else: raise TypeError('SEQUENCE OF has a minimum size of ' + actual_type.Min + ')') elif isinstance(primary, ogAST.PrimSequenceOf) \ and actual_type.kind == 'SequenceOfType': if (len(primary.value) < int(actual_type.Min) or len(primary.value) > int(actual_type.Max)): raise TypeError(str(len(primary.value)) + ' elements in SEQUENCE OF, while constraint is [' + str(actual_type.Min) + '..' + str(actual_type.Max) + ']') for elem in primary.value: check_type_compatibility(elem, actual_type.type, context) return elif isinstance(primary, ogAST.PrimSequence) \ and actual_type.kind == 'SequenceType': user_nb_elem = len(primary.value.keys()) type_nb_elem = len(actual_type.Children.keys()) if user_nb_elem != type_nb_elem: raise TypeError('Wrong number of fields in SEQUENCE of type {}' .format(type_name(typeRef))) else: for field, fd_data in actual_type.Children.viewitems(): ufield = field.replace('-', '_') if ufield not in primary.value: raise TypeError('Missing field {field} in SEQUENCE' ' of type {t1} ' .format(field=ufield, t1=type_name(typeRef))) else: # If the user field is a raw value if primary.value[ufield].is_raw: check_type_compatibility(primary.value[ufield], fd_data.type, context) else: # Compare the types for semantic equivalence try: compare_types( primary.value[ufield].exprType, fd_data.type) except TypeError as err: raise TypeError('Field ' + ufield + ' is not of the proper type, i.e. ' + type_name(fd_data.type) + ' - ' + str(err)) return elif isinstance(primary, ogAST.PrimChoiceItem) \ and actual_type.kind.startswith('Choice'): for choicekey, choice in actual_type.Children.viewitems(): if choicekey.lower() == primary.value['choice'].lower(): break else: raise TypeError('Non-existent choice "{choice}" in type {t1}' .format(choice=primary.value['choice'], t1=type_name(typeRef))) # compare primary.value['value'] # with actual_type['Children'][primary.choiceItem['choice']] value = primary.value['value'] choice_field_type = choice.type # if the user field is a raw value: if value.is_raw: check_type_compatibility(value, choice_field_type, context) # Compare the types for semantic equivalence: else: try: compare_types(value.exprType, choice_field_type) except TypeError as err: raise TypeError( 'Field {field} in CHOICE is not of type {t1} - {e}' .format(field=primary.value['choice'], t1=type_name(choice_field_type), e=str(err))) value.exprType = choice_field_type # XXX return elif isinstance(primary, ogAST.PrimChoiceDeterminant) \ and actual_type.kind.startswith('Choice'): for choicekey, choice in actual_type.EnumValues.viewitems(): if choicekey.replace('-', '_').lower() == \ primary.inputString.lower(): break else: raise TypeError('Non-existent choice "{choice}" in type {t1}' .format(choice=primary.inputString, t1=type_name(typeRef))) elif isinstance(primary, ogAST.PrimStringLiteral): # Octet strings basic_type = find_basic_type(typeRef) if basic_type.kind == 'StandardStringType': return elif basic_type.kind.endswith('StringType'): if int(basic_type.Min) <= len( primary.value[1:-1]) <= int(basic_type.Max): return else: raise TypeError('Invalid string literal - check that length is' 'within the bound limits {Min}..{Max}'.format (Min=str(basic_type.Min), Max=str(basic_type.Max))) else: raise TypeError('String literal not expected') elif (isinstance(primary, ogAST.PrimMantissaBaseExp) and actual_type.kind == 'RealType'): LOG.debug('PROBABLY (it is a float but I did not check' 'if values are compatible)') return else: raise TypeError('{prim} does not match type {t1}' .format(prim=primary.inputString, t1=type_name(typeRef))) def compare_types(type_a, type_b): ''' Compare two types, return if they are semantically equivalent, otherwise raise TypeError ''' LOG.debug('[compare_types]' + str(type_a) + ' and ' + str(type_b) + ': ') type_a = find_basic_type(type_a) type_b = find_basic_type(type_b) if type_a == type_b: return # Check if both types have basic compatibility simple_types = ( 'IntegerType', 'BooleanType', 'RealType', 'StringType', 'SequenceOfType', 'Integer32Type', 'OctetStringType' ) for ty in (type_a, type_b): if ty.kind not in simple_types: raise TypeError('Type {} is not a basic type'.format(type_name(ty))) if type_a.kind == type_b.kind: if type_a.kind == 'SequenceOfType': if type_a.Min == type_b.Min and type_a.Max == type_b.Max: compare_types(type_a.type, type_b.type) return else: raise TypeError('Incompatible arrays') # TODO: Check that OctetString types have compatible range return elif is_string(type_a) and is_string(type_b): return elif is_integer(type_a) and is_integer(type_b): return else: raise TypeError('Incompatible types {} and {}'.format( type_name(type_a), type_name(type_b) )) def find_variable(var, context): ''' Look for a variable name in the context and return its type ''' result = UNKNOWN_TYPE LOG.debug('[find_variable] checking if ' + str(var) + ' is defined') # all DCL-variables all_visible_variables = dict(context.global_variables) all_visible_variables.update(context.variables) # First check locally, i.e. in FPAR try: for variable in context.fpar: if variable['name'].lower() == var.lower(): LOG.debug(str(var) + ' is defined') return variable['type'] except AttributeError: # No FPAR section pass for varname, (vartype, _) in all_visible_variables.viewitems(): # Case insensitive comparison with variables if var.lower() == varname.lower(): result = vartype LOG.debug(str(var) + ' is defined') return result for timer in chain(context.timers, context.global_timers): if var.lower() == timer.lower(): LOG.debug(str(var) + ' is defined') return result LOG.debug('[find_variable] result: not found, raising exception') raise AttributeError('Variable {var} not defined'.format(var=var)) def fix_enumerated_and_choice(expr_enum, context): ''' If left side of the expression is of Enumerated or Choice type, check if right side is a literal of that sort, and update type ''' kind = find_basic_type(expr_enum.left.exprType).kind if kind == 'EnumeratedType': prim = ogAST.PrimEnumeratedValue(primary=expr_enum.right) elif kind == 'ChoiceEnumeratedType': prim = ogAST.PrimChoiceDeterminant(primary=expr_enum.right) try: check_type_compatibility(prim, expr_enum.left.exprType, context) expr_enum.right = prim expr_enum.right.exprType = expr_enum.left.exprType except (UnboundLocalError, AttributeError, TypeError): pass else: LOG.debug('Fixed enumerated/choice: {}'.format(expr_enum.inputString)) def fix_expression_types(expr, context): ''' Check/ensure type consistency in binary expressions ''' for _ in range(2): # Check if an raw enumerated value is of a reference type fix_enumerated_and_choice(expr, context) expr.right, expr.left = expr.left, expr.right # for side in permutations(('left', 'right')): # side_type = find_basic_type(getattr(expr, side[0]).exprType).kind # if side_type == 'EnumeratedType': # prim = ogAST.PrimEnumeratedValue(primary=getattr(expr, side[1])) # elif side_type == 'ChoiceEnumeratedType': # prim = ogAST.PrimChoiceDeterminant(primary=getattr(expr, side[1])) # try: # check_type_compatibility(prim, getattr(expr, side[0]).exprType, # context) # setattr(expr, side[1], prim) # getattr(expr, side[1]).exprType = getattr(expr, side[0]).exprType # except (UnboundLocalError, AttributeError, TypeError): # pass # If a side type remains unknown, check if it is an ASN.1 constant for side in permutations(('left', 'right')): value = getattr(expr, side[0]) if value.exprType == UNKNOWN_TYPE and is_constant(value): setattr(expr, side[0], ogAST.PrimConstant(primary=value)) getattr(expr, side[0]).exprType = getattr(expr, side[1]).exprType for side in (expr.right, expr.left): if side.is_raw: raw_expr = side else: typed_expr = side ref_type = typed_expr.exprType # If a side is a raw Sequence Of with unknown type, try to resolve it for side in permutations(('left', 'right')): value = getattr(expr, side[0]) # get expr.left then expr.right if not isinstance(value, ogAST.PrimSequenceOf): continue other = getattr(expr, side[1]) # other side basic = find_basic_type(value.exprType) if basic.kind == 'SequenceOfType' and basic.type == UNKNOWN_TYPE: asn_type = find_basic_type(other.exprType) if asn_type.kind == 'SequenceOfType': asn_type = asn_type.type for idx, elem in enumerate(value.value): check_expr = ogAST.ExprAssign() check_expr.left = ogAST.PrimVariable() check_expr.left.exprType = asn_type check_expr.right = elem fix_expression_types(check_expr, context) value.value[idx] = check_expr.right # the type of the raw PrimSequenceOf can be set now value.exprType.type = asn_type if isinstance(expr, ogAST.ExprIn): return if not expr.right.is_raw and not expr.left.is_raw: unknown = [uk_expr for uk_expr in expr.right, expr.left if uk_expr.exprType == UNKNOWN_TYPE] if unknown: #print traceback.print_stack() raise TypeError('Cannot resolve type of "{}"' .format(unknown[0].inputString)) # In Sequence, Choice and SEQUENCE OF expressions, # we must fix missing inner types # (due to similarities, the following should be refactored FIXME) if isinstance(expr.right, ogAST.PrimSequence): # left side must have a known type asn_type = find_basic_type(expr.left.exprType) if asn_type.kind != 'SequenceType': raise TypeError('left side must be a SEQUENCE type') for field, fd_expr in expr.right.value.viewitems(): if fd_expr.exprType == UNKNOWN_TYPE: try: expected_type = asn_type.Children.get( field.replace('_', '-')).type except AttributeError: raise TypeError('Field not found: ' + field) check_expr = ogAST.ExprAssign() check_expr.left = ogAST.PrimVariable() check_expr.left.exprType = expected_type check_expr.right = fd_expr fix_expression_types(check_expr, context) # Id of fd_expr may have changed (enumerated, choice) expr.right.value[field] = check_expr.right elif isinstance(expr.right, ogAST.PrimChoiceItem): asn_type = find_basic_type(expr.left.exprType) field = expr.right.value['choice'].replace('_', '-') if asn_type.kind != 'ChoiceType' \ or field.lower() not in [key.lower() for key in asn_type.Children.viewkeys()]: raise TypeError('Field is not valid in CHOICE:' + field) key, = [key for key in asn_type.Children.viewkeys() if key.lower() == field.lower()] if expr.right.value['value'].exprType == UNKNOWN_TYPE: try: expected_type = asn_type.Children.get(key).type except AttributeError: raise TypeError('Field not found in CHOICE: ' + field) check_expr = ogAST.ExprAssign() check_expr.left = ogAST.PrimVariable() check_expr.left.exprType = expected_type check_expr.right = expr.right.value['value'] fix_expression_types(check_expr, context) expr.right.value['value'] = check_expr.right elif isinstance(expr.right, ogAST.PrimConditional): for det in ('then', 'else'): # Recursively fix possibly missing types in the expression check_expr = ogAST.ExprAssign() check_expr.left = ogAST.PrimVariable() check_expr.left.exprType = expr.left.exprType check_expr.right = expr.right.value[det] fix_expression_types(check_expr, context) expr.right.value[det] = check_expr.right if expr.right.is_raw != expr.left.is_raw: check_type_compatibility(raw_expr, ref_type, context) if not raw_expr.exprType.kind.startswith(('Integer', 'Real')): # Raw int/real must keep their type because of the range # that can be computed raw_expr.exprType = ref_type else: compare_types(expr.left.exprType, expr.right.exprType) def expression_list(root, context): ''' Parse a list of expression parameters ''' errors = [] warnings = [] result = [] for param in root.getChildren(): exp, err, warn = expression(param, context) errors.extend(err) warnings.extend(warn) result.append(exp) return result, errors, warnings def primary_variable(root, context): ''' Primary Variable analysis ''' lexeme = root.children[0].text # Differentiate DCL and FPAR variables Prim = ogAST.PrimVariable if isinstance(context, ogAST.Procedure): for each in context.fpar: if each['name'].lower() == lexeme.lower(): Prim = ogAST.PrimFPAR break prim, errors, warnings = Prim(), [], [] prim.value = [root.children[0].text] prim.exprType = UNKNOWN_TYPE prim.inputString = get_input_string(root) prim.tmpVar = tmp() try: prim.exprType = find_variable(lexeme, context) except AttributeError: pass return prim, errors, warnings def binary_expression(root, context): ''' Binary expression analysis ''' errors, warnings = [], [] ExprNode = EXPR_NODE[root.type] expr = ExprNode( get_input_string(root), root.getLine(), root.getCharPositionInLine() ) expr.exprType = UNKNOWN_TYPE expr.tmpVar = tmp() left, right = root.children expr.left, err_left, warn_left = expression(left, context) expr.right, err_right, warn_right = expression(right, context) errors.extend(err_left) warnings.extend(warn_left) errors.extend(err_right) warnings.extend(warn_right) try: fix_expression_types(expr, context) except (AttributeError, TypeError) as err: errors.append(error(root, str(err))) return expr, errors, warnings def unary_expression(root, context): ''' Unary expression analysys ''' ExprNode = EXPR_NODE[root.type] expr = ExprNode( get_input_string(root), root.getLine(), root.getCharPositionInLine() ) expr.exprType = UNKNOWN_TYPE expr.tmpVar = tmp() expr.expr, errors, warnings = expression(root.children[0], context) return expr, errors, warnings def expression(root, context): ''' Expression analysis (e.g. 5+5*hello(world)!foo) ''' logic = (lexer.OR, lexer.AND, lexer.XOR) arithmetic = (lexer.PLUS, lexer.ASTERISK, lexer.DASH, lexer.DIV, lexer.MOD, lexer.REM) relational = (lexer.EQ, lexer.NEQ, lexer.GT, lexer.GE, lexer.LT, lexer.LE) if root.type in logic: return logic_expression(root, context) elif root.type in arithmetic: return arithmetic_expression(root, context) elif root.type in relational: return relational_expression(root, context) elif root.type == lexer.IN: return in_expression(root, context) elif root.type == lexer.APPEND: return append_expression(root, context) elif root.type == lexer.NOT: return not_expression(root, context) elif root.type == lexer.NEG: return neg_expression(root, context) elif root.type == lexer.PAREN: return expression(root.children[0], context) elif root.type == lexer.CONDITIONAL: return conditional_expression(root, context) elif root.type == lexer.PRIMARY: return primary(root.children[0], context) elif root.type == lexer.CALL: return call_expression(root, context) elif root.type == lexer.SELECTOR: return selector_expression(root, context) else: raise NotImplementedError def logic_expression(root, context): ''' Logic expression analysis ''' shortcircuit = '' # detect optional THEN in AND/OR expressions, indicating that the # short-circuit version of the operator is needed, to prevent the # evaluation of the right part if the left part does not evaluate # to true. if root.type in (lexer.OR, lexer.AND) and len(root.children) == 3: if root.children[1].type in (lexer.THEN, lexer.ELSE): root.children.pop(1) shortcircuit = ' else' if root.type == lexer.OR else ' then' expr, errors, warnings = binary_expression(root, context) expr.shortcircuit = shortcircuit left_bty = find_basic_type(expr.left.exprType) right_bty = find_basic_type(expr.right.exprType) for bty in left_bty, right_bty: if shortcircuit and bty.kind != 'BooleanType': msg = 'Shortcircuit operators only work with type Boolean' errors.append(error(root, msg)) break if bty.kind == 'BooleanType': continue elif bty.kind == 'BitStringType' and bty.Min == bty.Max: continue elif bty.kind == 'SequenceOfType' and bty.Min == bty.Max \ and find_basic_type(bty.type).kind == 'BooleanType': continue else: msg = 'Bitwise operators only work with Booleans, ' \ 'fixed size SequenceOf Booleans or fixed size BitStrings' errors.append(error(root, msg)) break if left_bty.kind == right_bty.kind == 'BooleanType': expr.exprType = BOOLEAN else: expr.exprType = expr.left.exprType return expr, errors, warnings def arithmetic_expression(root, context): ''' Arithmetic expression analysis ''' expr, errors, warnings = binary_expression(root, context) # Expressions returning a numerical type must have their range defined # accordingly with the kind of opration used between operand: basic = find_basic_type(expr.left.exprType) left = find_basic_type(expr.left.exprType) right = find_basic_type(expr.right.exprType) try: if isinstance(expr, ogAST.ExprPlus): attrs = {'Min': str(float(left.Min) + float(right.Min)), 'Max': str(float(left.Max) + float(right.Max))} expr.exprType = type('Plus', (basic,), attrs) elif isinstance(expr, ogAST.ExprMul): attrs = {'Min': str(float(left.Min) * float(right.Min)), 'Max': str(float(left.Max) * float(right.Max))} expr.exprType = type('Mul', (basic,), attrs) elif isinstance(expr, ogAST.ExprMinus): attrs = {'Min': str(float(left.Min) - float(right.Min)), 'Max': str(float(left.Max) - float(right.Max))} expr.exprType = type('Minus', (basic,), attrs) elif isinstance(expr, ogAST.ExprDiv): attrs = {'Min': str(float(left.Min) / (float(right.Min) or 1)), 'Max': str(float(left.Max) / (float(right.Max) or 1))} expr.exprType = type('Div', (basic,), attrs) elif isinstance(expr, (ogAST.ExprMod, ogAST.ExprRem)): attrs = {'Min': right.Min, 'Max': right.Max} expr.exprType = type('Mod', (basic,), attrs) except (ValueError, AttributeError): msg = 'Check that all your numerical data types '\ 'have a range constraint' errors.append(error(root, msg)) if root.type in (lexer.REM, lexer.MOD): for ty in (expr.left.exprType, expr.right.exprType): if not is_integer(ty): msg = 'Mod/Rem expressions can only applied to Integer types' errors.append(error(root, msg)) break return expr, errors, warnings def relational_expression(root, context): ''' Relational expression analysys ''' expr, errors, warnings = binary_expression(root, context) if root.type not in (lexer.EQ, lexer.NEQ): for ty in (expr.left.exprType, expr.right.exprType): if not is_numeric(ty): errors.append(error(root, 'Operands in relational expressions must be numerical')) break expr.exprType = BOOLEAN return expr, errors, warnings def in_expression(root, context): ''' In expression analysis ''' # Left and right are reversed for IN operator root.children[0], root.children[1] = root.children[1], root.children[0] expr, errors, warnings = binary_expression(root, context) expr.exprType = BOOLEAN # check that left part is a SEQUENCE OF or a string left_type = expr.left.exprType container_basic_type = find_basic_type(expr.left.exprType) if container_basic_type.kind == 'SequenceOfType': ref_type = container_basic_type.type elif container_basic_type.kind.endswith('StringType'): ref_type = container_basic_type else: msg = 'IN expression: right part must be a list' errors.append(error(root, msg)) return expr, errors, warnings expr.left.exprType = ref_type if find_basic_type(ref_type).kind == 'EnumeratedType': fix_enumerated_and_choice(expr, context) expr.left.exprType = left_type try: compare_types(expr.right.exprType, ref_type) except TypeError as err: errors.append(error(root, str(err))) if expr.right.is_raw and expr.left.is_raw: # If both sides are raw (e.g. "3 in {1,2,3}"), evaluate expression bool_expr = ogAST.PrimBoolean() bool_expr.inputString = expr.inputString bool_expr.line = expr.line bool_expr.charPositionInLine = expr.charPositionInLine bool_expr.exprType = type('PrBool', (object,), {'kind': 'BooleanType'}) if expr.right.value in [each.value for each in expr.left.value]: bool_expr.value = ['true'] warnings.append('Expression {} is always true' .format(expr.inputString)) else: bool_expr.value = ['false'] warnings.append('Expression {} is always false' .format(expr.inputString)) expr = bool_expr return expr, errors, warnings def append_expression(root, context): ''' Append expression analysis ''' expr, errors, warnings = binary_expression(root, context) for bty in (find_basic_type(expr.left.exprType), find_basic_type(expr.right.exprType)): if bty.kind != 'SequenceOfType' and not is_string(bty): msg = 'Append can only be applied to types SequenceOf or String' errors.append(error(root, msg)) break expr.exprType = expr.left.exprType return expr, errors, warnings def not_expression(root, context): ''' Not expression analysis ''' expr, errors, warnings = unary_expression(root, context) bty = find_basic_type(expr.expr.exprType) if bty.kind in ('BooleanType', ): expr.exprType = BOOLEAN elif bty.kind == 'BitStringType': expr.exprType = expr.expr.exprType elif bty.kind == 'SequenceOfType' and bty.type.kind == 'BooleanType': expr.exprType = expr.expr.exprType else: msg = 'Bitwise operators only work with booleans '\ 'and arrays of booleans' errors.append(error(root, msg)) return expr, errors, warnings def neg_expression(root, context): ''' Negative expression analysis ''' expr, errors, warnings = unary_expression(root, context) basic = find_basic_type(expr.expr.exprType) if not is_numeric(basic): msg = 'Negative expressions can only be applied to numeric types' errors.append(error(root, msg)) return expr, errors, warnings try: attrs = {'Min': str(-float(basic.Max)), 'Max': str(-float(basic.Min))} expr.exprType = type('Neg', (basic,), attrs) except (ValueError, AttributeError): msg = 'Check that all your numerical data types '\ 'have a range constraint' errors.append(error(root, msg)) return expr, errors, warnings def conditional_expression(root, context): ''' Conditional expression analysis ''' errors, warnings = [], [] expr = ogAST.PrimConditional( get_input_string(root), root.getLine(), root.getCharPositionInLine() ) expr.exprType = UNKNOWN_TYPE expr.tmpVar = tmp() if_part, then_part, else_part = root.getChildren() if_expr, err, warn = expression(if_part, context) errors.extend(err) warnings.extend(warn) then_expr, err, warn = expression(then_part, context) errors.extend(err) warnings.extend(warn) else_expr, err, warn = expression(else_part, context) errors.extend(err) warnings.extend(warn) if find_basic_type(if_expr.exprType).kind != 'BooleanType': msg = 'Conditions in conditional expressions must be of type Boolean' errors.append(error(root, msg)) # TODO: Refactor this try: expr.left = then_expr expr.right = else_expr fix_expression_types(expr, context) expr.exprType = then_expr.exprType except (AttributeError, TypeError) as err: if UNKNOWN_TYPE not in (then_expr.exprType, else_expr.exprType): errors.append(error(root, str(err))) expr.value = { 'if': if_expr, 'then': then_expr, 'else': else_expr, 'tmpVar': expr.tmpVar } return expr, errors, warnings def call_expression(root, context): ''' Call expression analysis ''' errors, warnings = [], [] if root.children[0].type == lexer.PRIMARY: primary = root.children[0] if primary.children[0].type == lexer.VARIABLE: variable = primary.children[0] ident = variable.children[0].text.lower() proc_list = [proc.inputString.lower() for proc in context.procedures] if ident in (SPECIAL_OPERATORS.keys() + proc_list): return primary_call(root, context) num_params = len(root.children[1].children) if num_params == 1: return primary_index(root, context) elif num_params == 2: return primary_substring(root, context) else: node = ogAST.PrimCall() # Use error node instead? node.inputString = get_input_string(root) node.exprType = UNKNOWN_TYPE errors.append(error(root, 'Wrong number of parameters')) return node, errors, warnings def primary_call(root, context): ''' Primary call analysis ''' node, errors, warnings = ogAST.PrimCall(), [], [] node.exprType = UNKNOWN_TYPE node.inputString = get_input_string(root) node.tmpVar = tmp() ident = root.children[0].children[0].children[0].text.lower() params, params_errors, param_warnings = \ expression_list(root.children[1], context) errors.extend(params_errors) warnings.extend(param_warnings) try: fix_special_operators(ident, params, context) except (AttributeError, TypeError) as err: errors.append(error(root, str(err))) node.value = [ident, {'procParams': params}] params_bty = [find_basic_type(p.exprType) for p in params] if ident == 'present': try: node.exprType = type('present', (object,), { 'kind': 'ChoiceEnumeratedType', 'EnumValues': params_bty[0].Children }) except AttributeError: errors.append(error(root, 'Parameter type is not a CHOICE')) elif ident in ('length', 'fix'): # result is an integer type with range of the param type try: node.exprType = type('fix', (INTEGER,), { 'Min': params_bty[0].Min, 'Max': params_bty[0].Max }) except AttributeError: errors.append(error(root, 'Parameter type has no range')) elif ident == 'float': try: node.exprType = type('float_op', (REAL,), { 'Min': params_bty[0].Min, 'Max': params_bty[0].Max }) except AttributeError: errors.append(error(root, 'Parameter type has no range')) elif ident == 'power': try: node.exprType = type('Power', (params_bty[0],), { 'Min': str(pow(float(params_bty[0].Min), float(params_bty[1].Min))), 'Max': str(pow(float(params_bty[0].Max), float(params_bty[1].Max))) }) except OverflowError: errors.append(error(root, 'Result can exceeds 64-bits')) except AttributeError: errors.append(error(root, 'Parameter type has no range')) elif ident == 'abs': try: node.exprType = type('Abs', (params_bty[0],), { 'Min': str(max(float(params_bty[0].Min), 0)), 'Max': str(max(float(params_bty[0].Max), 0)) }) except AttributeError: errors.append(error(root, '"Abs" parameter type has no range')) elif ident == 'num': try: enum_values = [int(each.IntValue) for each in params_bty[0].EnumValues.viewvalues()] node.exprType = type('Num', (INTEGER,), { 'Min': str(min(enum_values)), 'Max': str(max(enum_values)) }) except AttributeError: errors.append(error(root, '"Num" parameter error')) return node, errors, warnings def primary_index(root, context): ''' Primary index analysis ''' node, errors, warnings = ogAST.PrimIndex(), [], [] node.exprType = UNKNOWN_TYPE node.inputString = get_input_string(root) node.tmpVar = tmp() receiver, receiver_err, receiver_warn = \ expression(root.children[0], context) receiver_bty = find_basic_type(receiver.exprType) errors.extend(receiver_err) warnings.extend(receiver_warn) params, params_errors, param_warnings = \ expression_list(root.children[1], context) errors.extend(params_errors) warnings.extend(param_warnings) node.value = [receiver, {'index': params}] if receiver_bty.kind == 'SequenceOfType': node.exprType = receiver_bty.type else: msg = 'Index can only be applied to type SequenceOf' errors.append(error(root, msg)) return node, errors, warnings def primary_substring(root, context): ''' Primary substring analysis ''' node, errors, warnings = ogAST.PrimSubstring(), [], [] node.exprType = UNKNOWN_TYPE node.inputString = get_input_string(root) node.tmpVar = tmp() receiver, receiver_err, receiver_warn = \ expression(root.children[0], context) receiver_bty = find_basic_type(receiver.exprType) errors.extend(receiver_err) warnings.extend(receiver_warn) params, params_errors, param_warnings = \ expression_list(root.children[1], context) errors.extend(params_errors) warnings.extend(param_warnings) node.value = [receiver, {'substring': params, 'tmpVar': tmp()}] if receiver_bty.kind == 'SequenceOfType' or \ receiver_bty.kind.endswith('StringType'): node.exprType = receiver.exprType # Check bounds basic = find_basic_type(node.exprType) min0 = find_basic_type(params[0].exprType).Min min1 = find_basic_type(params[1].exprType).Min max0 = find_basic_type(params[0].exprType).Max max1 = find_basic_type(params[1].exprType).Max if int(min0) > int(min1) or int(max0) > int(max1): msg = 'Substring bounds are invalid' errors.append(error(root, msg)) if int(min0) > int(basic.Max) \ or int(max1) > int(basic.Max): msg = 'Substring bounds [{}..{}] outside range [{}..{}]'.format( min0, max1, basic.Min, basic.Max) errors.append(error(root, msg)) else: msg = 'Substring can only be applied to types SequenceOf or String' errors.append(error(root, msg)) return node, errors, warnings def selector_expression(root, context): ''' Selector expression analysis ''' errors, warnings = [], [] node = ogAST.PrimSelector() node.exprType = UNKNOWN_TYPE node.inputString = get_input_string(root) node.tmpVar = tmp() receiver, receiver_err, receiver_warn = \ expression(root.children[0], context) receiver_bty = find_basic_type(receiver.exprType) errors.extend(receiver_err) warnings.extend(receiver_warn) field_name = root.children[1].text.replace('_', '-').lower() for n, f in receiver_bty.Children.viewitems(): if n.lower() == field_name: node.exprType = f.type break else: msg = 'Field "{}" not found in expression {}'.format(field_name) errors.append(error(root, msg)) node.value = [receiver, field_name.replace('-', '_').lower()] return node, errors, warnings def primary(root, context): ''' Literal expression analysis ''' prim, errors, warnings = None, [], [] if root.type == lexer.VARIABLE: return primary_variable(root, context) elif root.type == lexer.INT: prim = ogAST.PrimInteger() prim.value = [root.text.lower()] prim.exprType = type('PrInt', (object,), { 'kind': 'IntegerType', 'Min': root.text, 'Max': root.text }) elif root.type in (lexer.TRUE, lexer.FALSE): prim = ogAST.PrimBoolean() prim.value = [root.text.lower()] prim.exprType = type('PrBool', (object,), {'kind': 'BooleanType'}) elif root.type == lexer.FLOAT: prim = ogAST.PrimReal() prim.value = [root.text] prim.exprType = type('PrReal', (object,), { 'kind': 'RealType', 'Min': prim.value[0], 'Max': prim.value[0] }) elif root.type == lexer.STRING: prim = ogAST.PrimStringLiteral() prim.value = root.text prim.exprType = type('PrStr', (object,), { 'kind': 'StringType', 'Min': str(len(prim.value) - 2), 'Max': str(len(prim.value) - 2) }) elif root.type == lexer.FLOAT2: prim = ogAST.PrimMantissaBaseExp() mant = float(root.getChild(0).toString()) base = int(root.getChild(1).toString()) exp = int(root.getChild(2).toString()) # Compute mantissa * base**exponent to get the value type range value = float(mant * pow(base, exp)) prim.value = {'mantissa': mant, 'base': base, 'exponent': exp} prim.exprType = type('PrMantissa', (object,), { 'kind': 'RealType', 'Min': str(value), 'Max': str(value) }) elif root.type == lexer.EMPTYSTR: # Empty SEQUENCE OF (i.e. "{}") prim = ogAST.PrimEmptyString() prim.exprType = type('PrES', (object,), { 'kind': 'SequenceOfType', 'Min': '0', 'Max': '0' }) elif root.type == lexer.CHOICE: prim = ogAST.PrimChoiceItem() choice = root.getChild(0).toString() expr, err, warn = expression(root.getChild(1), context) errors.extend(err) warnings.extend(warn) prim.value = {'choice': choice, 'value': expr} prim.exprType = UNKNOWN_TYPE elif root.type == lexer.SEQUENCE: prim = ogAST.PrimSequence() prim.value = {} for elem in root.getChildren(): if elem.type == lexer.ID: field_name = elem.text else: prim.value[field_name], err, warn = (expression(elem, context)) errors.extend(err) warnings.extend(warn) prim.exprType = UNKNOWN_TYPE elif root.type == lexer.SEQOF: prim = ogAST.PrimSequenceOf() prim.value = [] for elem in root.getChildren(): # SEQUENCE OF elements cannot have fieldnames/indexes prim_elem, prim_elem_errors, prim_elem_warnings = \ primary(elem, context) errors += prim_elem_errors warnings += prim_elem_warnings prim_elem.inputString = get_input_string(elem) prim_elem.line = elem.getLine() prim_elem.charPositionInLine = elem.getCharPositionInLine() prim.value.append(prim_elem) prim.exprType = type('PrSO', (object,), { 'kind': 'SequenceOfType', 'Min': str(len(root.children)), 'Max': str(len(root.children)), 'type': UNKNOWN_TYPE }) elif root.type == lexer.BITSTR: prim = ogAST.PrimBitStringLiteral() warnings.append(warning(root, 'Bit string literal not supported yet')) elif root.type == lexer.OCTSTR: prim = ogAST.PrimOctetStringLiteral() warnings.append( warning(root, 'Octet string literal not supported yet')) else: # TODO: return error message raise NotImplementedError prim.inputString = get_input_string(root) prim.tmpVar = tmp() return prim, errors, warnings def variables(root, ta_ast, context): ''' Process declarations of variables (dcl a,b Type := 5) ''' var = [] errors = [] warnings = [] asn1_sort, def_value = UNKNOWN_TYPE, None for child in root.getChildren(): if child.type == lexer.ID: var.append(child.text) elif child.type == lexer.SORT: sort = child.getChild(0).text # Find corresponding type in ASN.1 model try: asn1_sort = sdl_to_asn1(sort) except TypeError as err: errors.append(error(root, str(err))) elif child.type == lexer.GROUND: # Default value for a variable - needs to be a ground expression def_value, err, warn = expression(child.getChild(0), context) errors.extend(err) warnings.extend(warn) expr = ogAST.ExprAssign() expr.left = ogAST.PrimVariable() expr.left.inputString = var[-1] expr.left.exprType = asn1_sort expr.right = def_value try: fix_expression_types(expr, context) def_value = expr.right except(AttributeError, TypeError) as err: #print (traceback.format_exc()) errors.append('Types are incompatible in DCL assignment: ' 'left (' + expr.left.inputString + ', type= ' + type_name(expr.left.exprType) + '), right (' + expr.right.inputString + ', type= ' + type_name(expr.right.exprType) + ') ' + str(err)) else: def_value.exprType = asn1_sort if not def_value.is_raw and not is_constant(def_value): errors.append('In variable declaration {}: default' ' value is not a valid ground expression'. format(var[-1])) else: warnings.append('Unsupported variables construct type: ' + str(child.type)) for variable in var: # Add to the context and text area AST entries context.variables[variable] = (asn1_sort, def_value) ta_ast.variables[variable] = (asn1_sort, def_value) if not DV: errors.append('Cannot do semantic checks on variable declarations') return errors, warnings def dcl(root, ta_ast, context): ''' Process a set of variable declarations ''' errors = [] warnings = [] for child in root.getChildren(): if child.type == lexer.VARIABLES: err, warn = variables(child, ta_ast, context) errors.extend(err) warnings.extend(warn) else: warnings.append( 'Unsupported dcl construct, type: ' + str(child.type)) return errors, warnings def fpar(root): ''' Process a formal parameter declaration ''' errors = [] warnings = [] params = [] asn1_sort = UNKNOWN_TYPE for param in root.getChildren(): param_names = [] sort = '' direction = 'in' assert param.type == lexer.PARAM for child in param.getChildren(): if child.type == lexer.INOUT: direction = 'out' elif child.type == lexer.IN: pass elif child.type == lexer.ID: # variable name param_names.append(child.text) elif child.type == lexer.SORT: sort = child.getChild(0).text try: asn1_sort = sdl_to_asn1(sort) except TypeError as err: errors.append(str(err) + '(line ' + str(child.getLine()) + ')') for name in param_names: params.append({'name': name, 'direction': direction, 'type': asn1_sort}) else: warnings.append( 'Unsupported construct in FPAR, type: ' + str(child.type)) return params, errors, warnings def composite_state(root, parent=None, context=None): ''' Parse a composite state definition ''' comp = ogAST.CompositeState() errors, warnings = [], [] # Create a list of all inherited data try: comp.global_variables = dict(context.variables) comp.global_variables.update(context.global_variables) comp.global_timers = list(context.timers) comp.global_timers.extend(list(context.global_timers)) comp.input_signals = context.input_signals comp.output_signals = context.output_signals comp.procedures = context.procedures comp.operators = dict(context.operators) except AttributeError: LOG.debug('Procedure context is undefined') # Gather the list of states defined in the composite state # and map a list of transitionsi to each state comp.mapping = {name: [] for name in get_state_list(root)} inner_composite, states, floatings, starts = [], [], [], [] for child in root.getChildren(): if child.type == lexer.ID: comp.line = child.getLine() comp.charPositionInLine = child.getCharPositionInLine() comp.statename = child.toString().lower() elif child.type == lexer.COMMENT: comp.comment, _, _ = end(child) elif child.type == lexer.IN: # state entry point for point in child.getChildren(): comp.state_entrypoints.append(point.toString().lower()) elif child.type == lexer.OUT: # state exit point for point in child.getChildren(): comp.state_exitpoints.append(point.toString().lower()) elif child.type == lexer.TEXTAREA: textarea, err, warn = text_area(child, context=comp) errors.extend(err) warnings.extend(warn) comp.content.textAreas.append(textarea) elif child.type == lexer.PROCEDURE: new_proc, err, warn = procedure(child, context=comp) errors.extend(err) warnings.extend(warn) if new_proc.inputString.strip().lower() == 'entry': comp.entry_procedure = new_proc elif new_proc.inputString.strip().lower() == 'exit': comp.exit_procedure = new_proc comp.content.inner_procedures.append(new_proc) # Add procedure to the context, to make it visible at scope level context.procedures.append(new_proc) elif child.type == lexer.COMPOSITE_STATE: inner_composite.append(child) elif child.type == lexer.STATE: states.append(child) elif child.type == lexer.FLOATING_LABEL: floatings.append(child) elif child.type == lexer.START: starts.append(child) else: warnings.append( 'Unsupported construct in nested state, type: ' + str(child.type) + ' - line ' + str(child.getLine()) + ' - state name: ' + str(comp.statename)) for each in inner_composite: # Parse inner composite states after the text areas to make sure # that all variables are propagated to the the inner scope inner, err, warn = composite_state(each, parent=None, context=comp) errors.extend(err) warnings.extend(warn) comp.composite_states.append(inner) # Parse other elements after the nested states for each in starts: # START transition (fills the mapping structure) st, err, warn = start(each, context=comp) errors.extend(err) warnings.extend(warn) if st.inputString: comp.content.named_start.append(st) elif not comp.content.start: comp.content.start = st else: errors.append('Only one unnamed START transition is allowed') for each in floatings: lab, err, warn = floating_label(each, parent=None, context=comp) errors.extend(err) warnings.extend(warn) comp.content.floating_labels.append(lab) for each in states: # And parse the states after inner states to make sure all CONNECTS # are properly defined. # Fill up the 'mapping' structure. newstate, err, warn = state(each, parent=None, context=comp) errors.extend(err) warnings.extend(warn) comp.content.states.append(newstate) # Post-processing: check that all NEXTSTATEs have a corresponding STATE for ns in [t.inputString.lower() for t in comp.terminators if t.kind == 'next_state']: if not ns in [s.lower() for s in comp.mapping.viewkeys()] + ['-']: errors.append('In composite state "{}": missing definition ' 'of substate "{}"'.format(comp.statename, ns.upper())) return comp, errors, warnings def procedure(root, parent=None, context=None): ''' Parse a procedure definition ''' proc = ogAST.Procedure() errors = [] warnings = [] # Create a list of all inherited data try: proc.global_variables = dict(context.variables) proc.global_variables.update(context.global_variables) proc.global_timers = list(context.timers) proc.global_timers.extend(list(context.global_timers)) proc.input_signals = context.input_signals proc.output_signals = context.output_signals proc.procedures = context.procedures proc.operators = dict(context.operators) except AttributeError: LOG.debug('Procedure context is undefined') # Gather the list of states defined in the procedure # and create a mapping of transitions to each state # (Note, procedures in OG currently do NOT support states) proc.mapping = {name: [] for name in get_state_list(root)} for child in root.getChildren(): if child.type == lexer.CIF: # Get symbol coordinates proc.pos_x, proc.pos_y, proc.width, proc.height = cif(child) elif child.type == lexer.ID: proc.line = child.getLine() proc.charPositionInLine = child.getCharPositionInLine() proc.inputString = child.toString() elif child.type == lexer.COMMENT: proc.comment, _, ___ = end(child) elif child.type == lexer.TEXTAREA: textarea, err, warn = text_area(child, context=proc) errors.extend(err) warnings.extend(warn) proc.content.textAreas.append(textarea) elif child.type == lexer.PROCEDURE: new_proc, err, warn = procedure(child, context=proc) errors.extend(err) warnings.extend(warn) proc.content.inner_procedures.append(new_proc) # Add procedure to the context, to make it visible at scope level context.procedures.append(new_proc) elif child.type == lexer.EXTERNAL: proc.external = True elif child.type == lexer.FPAR: params, err, warn = fpar(child) errors.extend(err) warnings.extend(warn) proc.fpar = params elif child.type == lexer.START: # START transition (fills the mapping structure) proc.content.start, err, warn = start(child, context=proc) errors.extend(err) warnings.extend(warn) elif child.type == lexer.STATE: # STATE - fills up the 'mapping' structure. newstate, err, warn = state(child, parent=None, context=proc) errors.extend(err) warnings.extend(warn) proc.content.states.append(newstate) elif child.type == lexer.FLOATING_LABEL: lab, err, warn = floating_label(child, parent=None, context=proc) errors.extend(err) warnings.extend(warn) proc.content.floating_labels.append(lab) else: warnings.append( 'Unsupported construct in procedure, type: ' + str(child.type) + ' - line ' + str(child.getLine()) + ' - string: ' + str(proc.inputString)) for each in proc.terminators: # check that RETURN statements type is correct if not proc.return_type and each.return_expr: errors.append('No return value expected in procedure ' + proc.inputString) elif proc.return_type and each.return_expr: check_expr = ogAST.ExprAssign() check_expr.left = ogAST.PrimVariable() check_expr.left.exprType = proc.return_type check_expr.right = each.return_expr try: fix_expression_types(check_expr, context) except (TypeError, AttributeError) as err: errors.append(str(err)) # Id of fd_expr may have changed (enumerated, choice) each.return_expr = check_expr.right elif proc.return_type and not each.return_expr: errors.append('Missing return value in procedure ' + proc.inputString) else: continue return proc, errors, warnings def floating_label(root, parent, context): ''' Floating label: name and optional transition ''' errors = [] warnings = [] lab = ogAST.Floating_label() # Keep track of the number of terminator statements following the label # useful if we want to render graphs from the SDL model terminators = len(context.terminators) for child in root.getChildren(): if child.type == lexer.ID: lab.inputString = child.text lab.line = child.getLine() lab.charPositionInLine = child.getCharPositionInLine() elif child.type == lexer.CIF: # Get symbol coordinates lab.pos_x, lab.pos_y, lab.width, lab.height = cif(child) elif child.type == lexer.HYPERLINK: lab.hyperlink = child.getChild(0).text[1:-1] elif child.type == lexer.TRANSITION: trans, err, warn = transition( child, parent=lab, context=context) errors.extend(err) warnings.extend(warn) lab.transition = trans else: warnings.append( 'Unsupported construct in floating label: ' + str(child.type)) if not lab.transition: warnings.append('Floating labels not followed by a transition: ' + str(lab.inputString)) # At the end of the label parsing, get the the list of terminators that # the transition contains by making a diff with the list at context # level (we counted the number of terminators before parsing the item) lab.terminators = list(context.terminators[terminators:]) return lab, errors, warnings def newtype_gettype(root, ta_ast, context): ''' Returns the name of the new type created by a NEWTYPE construction ''' errors = [] warnings = [] newtypename = "" if (root.getChild(0).type != lexer.SORT): warnings.append("Expected SORT in newtype identifier, got type:" + str(root.type)) return newtypename, errors, warnings newtypename = root.getChild(0).getChild(0).text return newtypename, errors, warnings def get_array_type(root): ''' Returns the subtype associated to an NEWTYPE ARRAY construction ''' # indexSort = root.getChild(0).text typeSort = root.getChild(1).text typeSortLine = root.getChild(1).getLine() typeSortChar = root.getChild(1).getCharPositionInLine() # Constructing ASN.1 AST subtype newtype = type("SeqOf_type", (object,), { "Line": typeSortLine, "CharPositionInLine": typeSortChar, "Kind": "ReferenceType", "ReferencedTypeName": typeSort }) return newtype def get_struct_children(root): ''' Returns the fields of a STRUCT as a dictionary ''' children = {} fieldlist = root.getChild(0) fieldname = "" typename = "" if (fieldlist.type != lexer.FIELDS): return children for field in fieldlist.getChildren(): if (field.type == lexer.FIELD): fieldname = field.getChild(0).text typename = field.getChild(1).getChild(0).text line = field.getChild(0).getLine() charpos = field.getChild(0).getCharPositionInLine() children[fieldname] = type(str(fieldname), (object,), { "Optional": "False", "Line": line, "CharPositionInLine": charpos, "type": type(str(fieldname + "_type"), (object,), { "Line": line, "CharPositionInLine": charpos, "kind": "ReferenceType", "ReferencedTypeName": typename }) }) return children def syntype(root, ta_ast, context): ''' Parse a SYNTYPE definition and inject it in ASN1 AST''' errors = [] warnings = [] newtype = "" reftype = "" global DV newtypename = root.getChild(0).getChild(0).text # reftypename = root.getChild(1).getChild(0).text newtype = type(str(newtypename), (object,), { "Line": root.getChild(0).getLine(), "CharPositionInLine": root.getChild(0).getCharPositionInLine(), }) newtype.type = type(str(newtypename) + "_type", (object,), { "Line": root.getChild(1).getLine(), "CharPositionInLine": root.getChild(1).getCharPositionInLine(), "kind": reftype + "Type" }) DV.types[str(newtypename)] = newtype LOG.debug("Found new SYNTYPE " + newtypename) return errors, warnings def newtype(root, ta_ast, context): ''' Parse a NEWTYPE definition and inject it in ASN1 AST''' errors = [] warnings = [] global DV newtypename, errors, warnings = newtype_gettype(root, ta_ast, context) if (newtypename == ""): return errors, warnings newtype = type(str(newtypename), (object,), { "Line": root.getLine(), "CharPositionInLine": root.getCharPositionInLine()}) if (root.getChild(1).type == lexer.ARRAY): newtype.kind = "SequenceOfType" newtype.type = get_array_type(root.getChild(1)) newtype.Min = "Min" newtype.Max = "Max" DV.types[str(newtypename)] = newtype LOG.debug("Found new ARRAY type " + newtypename) elif (root.getChild(1).type == lexer.STRUCT): newtype.kind = "SequenceType" newtype.Children = get_struct_children(root.getChild(1)) DV.types[str(newtypename)] = newtype LOG.debug("Found new STRUCT type " + newtypename) else: warnings.append( 'Unsupported type definition in newtype, type: ' + str(root.type)) # STRUCT CASE return errors, warnings def synonym(root, ta_ast, context): ''' Parse a SYNONYM definition and inject it in ASN1 exported variables''' errors = [] warnings = [] global DV if not "SDL-Constants" in DV.asn1Modules: DV.asn1Modules.append("SDL-Constants") DV.exportedVariables["SDL-Constants"] = [] for child in root.getChildren(): if child.getChild(0).type == lexer.SORT: DV.exportedVariables["SDL-Constants"].append( child.getChild(0).getChild(0).text) return errors, warnings def text_area_content(root, ta_ast, context): ''' Content of a text area: DCL, NEWTYPES, SYNTYPES, SYNONYMS, operators, procedures ''' errors = [] warnings = [] for child in root.getChildren(): if child.type == lexer.DCL: err, warn = dcl(child, ta_ast, context) errors.extend(err) warnings.extend(warn) elif child.type == lexer.SYNONYM_LIST: err, warn = synonym(child, ta_ast, context) errors.extend(err) warnings.extend(warn) elif child.type == lexer.NEWTYPE: err, warn = newtype(child, ta_ast, context) errors.extend(err) warnings.extend(warn) elif child.type == lexer.SYNTYPE: err, warn = syntype(child, ta_ast, context) errors.extend(err) warnings.extend(warn) elif child.type == lexer.PROCEDURE: proc, err, warn = procedure(child, context=context) errors.extend(err) warnings.extend(warn) # Add procedure to the container (process or procedure) context.content.inner_procedures.append(proc) # Add to context to make it visible at scope level context.procedures.append(proc) elif child.type == lexer.FPAR: params, err, warn = fpar(child) errors.extend(err) warnings.extend(warn) try: if context.fpar: errors.append('Duplicate declaration of FPAR section') else: context.fpar = params except AttributeError: errors.append('Only procedures can have an FPAR section') elif child.type == lexer.TIMER: context.timers.extend(timer.text.lower() for timer in child.children) else: warnings.append( 'Unsupported construct in text area content, type: ' + str(child.type)) return errors, warnings def text_area(root, parent=None, context=None): ''' Process a text area (DCL, procedure, operators declarations ''' errors = [] warnings = [] ta = ogAST.TextArea() coord = False for child in root.getChildren(): if child.type == lexer.CIF: userTextStartIndex = child.getTokenStopIndex() + 1 ta.pos_x, ta.pos_y, ta.width, ta.height = cif(child) coord = True elif child.type == lexer.TEXTAREA_CONTENT: ta.line = child.getLine() ta.charPositionInLine = child.getCharPositionInLine() # Go update the process-level list of variables # (TODO: also ops and procedures) err, warn = text_area_content(child, ta, context) errors.extend(err) warnings.extend(warn) elif child.type == lexer.ENDTEXT: userTextStopIndex = child.getTokenStartIndex() - 1 ta.inputString = token_stream(child).toString( userTextStartIndex, userTextStopIndex).strip() elif child.type == lexer.HYPERLINK: ta.hyperlink = child.getChild(0).toString()[1:-1] else: warnings.append('Unsupported construct in text area, type: ' + str(child.type)) # Report errors with symbol coordinates if coord: errors = [[e, [ta.pos_x, ta.pos_y]] for e in errors] warnings = [[w, [ta.pos_x, ta.pos_y]] for w in warnings] return ta, errors, warnings def signal(root): ''' SIGNAL definition: name and optional list of types ''' errors, warnings = [], [] new_signal = {} for child in root.getChildren(): if child.type == lexer.ID: new_signal['name'] = child.text elif child.type == lexer.PARAMNAMES: try: param_name, = [par.text for par in child.getChildren()] new_signal['param_name'] = param_name except ValueError: # Will be raised and reported at PARAMS token pass elif child.type == lexer.PARAMS: try: param, = [par.text for par in child.getChildren()] new_signal['type'] = sdl_to_asn1(param) except ValueError: errors.append(new_signal['name'] + ' cannot have more' + ' than one parameter. Check signal declaration.') except TypeError as err: errors.append(str(err)) return new_signal, errors, warnings def single_route(root): ''' Route (from id to id with [signal id] ''' route = {'source': root.getChild(0).text, 'dest': root.getChild(1).text, 'signals': [sig.text for sig in root.getChildren()[2:]]} return route def channel_signalroute(root): ''' Channel/signalroute definition (connections) ''' # no AST entry for edges - a simple dict is sufficient # (name, [route]) edge = {'routes': []} for child in root.getChildren(): if child.type == lexer.ID: edge['name'] = child.text elif child.type == lexer.ROUTE: edge['routes'].append(single_route(child)) return edge def block_definition(root, parent): ''' BLOCK entity definition ''' errors, warnings = [], [] block = ogAST.Block() block.parent = parent parent.blocks.append(block) for child in root.getChildren(): if child.type == lexer.ID: block.name = child.text elif child.type == lexer.SIGNAL: sig, err, warn = signal(child) errors.extend(err) warnings.extend(warn) block.signals.append(sig) elif child.type == lexer.CONNECTION: block.connections.append({'channel': cnx[0].text, 'signalroute': cnx[1].text} for cnx in child.getChildren()) elif child.type == lexer.BLOCK: block, err, warn = block_definition(child, parent=block) errors.extend(err) warnings.extend(warn) elif child.type == lexer.PROCESS: proc, err, warn = process_definition(child, parent=block) block.processes.append(proc) errors.extend(err) warnings.extend(warn) elif child.type == lexer.SIGNALROUTE: sigroute = channel_signalroute(child) block.signalroutes.append(sigroute) else: warnings.append('Unsupported block child type: ' + str(child.type)) return block, errors, warnings def system_definition(root, parent): ''' SYSTEM part - contains blocks, procedures and channels ''' errors, warnings = [], [] system = ogAST.System() # Store the name of the file where the system is defined system.filename = node_filename(root) system.ast = parent for child in root.getChildren(): if child.type == lexer.ID: system.name = child.text LOG.debug('System name: ' + system.name) elif child.type == lexer.SIGNAL: sig, err, warn = signal(child) errors.extend(err) warnings.extend(warn) system.signals.append(sig) LOG.debug('Found signal: ' + str(sig)) elif child.type == lexer.PROCEDURE: LOG.debug('procedure declaration') proc, err, warn = procedure( child, parent=None, context=system) errors.extend(err) warnings.extend(warn) system.procedures.append(proc) LOG.debug('Added procedure: ' + proc.inputString) elif child.type == lexer.CHANNEL: LOG.debug('channel declaration') channel = channel_signalroute(child) system.channels.append(channel) elif child.type == lexer.BLOCK: LOG.debug('block declaration') block, err, warn = block_definition(child, parent=system) errors.extend(err) warnings.extend(warn) else: warnings.append('Unsupported construct in system: ' + str(child.type)) return system, errors, warnings def process_definition(root, parent=None, context=None): ''' Process definition analysis ''' errors = [] warnings = [] process = ogAST.Process() process.filename = node_filename(root) process.parent = parent coord = False # Prepare the transition/state mapping process.mapping = {name: [] for name in get_state_list(root)} for child in root.getChildren(): if child.type == lexer.CIF: # Get symbol coordinates process.pos_x, process.pos_y, process.width, process.height =\ cif(child) coord = True elif child.type == lexer.ID: # Get process (taste function) name process.processName = child.text try: # Retrieve process interface (PI/RI) async_signals, procedures = get_interfaces(parent, child.text) process.input_signals.extend([sig for sig in async_signals if sig['direction'] == 'in']) process.output_signals.extend([sig for sig in async_signals if sig['direction'] == 'out']) process.procedures.extend(procedures) except AttributeError as err: # No interface because process is defined standalone LOG.debug('Discarding process ' + child.text + ' ' + str(err)) except TypeError as error: LOG.debug(str(error)) errors.append(str(error)) if coord: errors = [[e, [process.pos_x, process.pos_y]] for e in errors] warnings = [[w, [process.pos_x, process.pos_y]] for w in warnings] elif child.type == lexer.TEXTAREA: # Text zone where variables and operators are declared textarea, err, warn = text_area(child, context=process) errors.extend(err) warnings.extend(warn) process.content.textAreas.append(textarea) elif child.type == lexer.START: # START transition (fills the mapping structure) process.content.start, err, warn = start( child, context=process) errors.extend(err) warnings.extend(warn) elif child.type == lexer.STATE: # STATE - fills up the 'mapping' structure. statedef, err, warn = state( child, parent=None, context=process) errors.extend(err) warnings.extend(warn) process.content.states.append(statedef) elif child.type == lexer.NUMBER_OF_INSTANCES: # Number of instances - discarded (working on a single process) pass elif child.type == lexer.PROCEDURE: proc, err, warn = procedure( child, parent=None, context=process) errors.extend(err) warnings.extend(warn) process.content.inner_procedures.append(proc) # Add it at process level so that it is in the scope process.procedures.append(proc) elif child.type == lexer.FLOATING_LABEL: lab, err, warn = floating_label( child, parent=None, context=process) errors.extend(err) warnings.extend(warn) process.content.floating_labels.append(lab) elif child.type == lexer.COMPOSITE_STATE: comp, err, warn = composite_state(child, parent=None, context=process) errors.extend(err) warnings.extend(warn) process.composite_states.append(comp) elif child.type == lexer.REFERENCED: process.referenced = True elif child.type == lexer.COMMENT: process.comment, _, _ = end(child) else: warnings.append('Unsupported process definition child: ' + sdl92Parser.tokenNames[child.type] + ' - line ' + str(child.getLine())) return process, errors, warnings def input_part(root, parent, context): ''' Parse an INPUT - set of TASTE provided interfaces ''' i = ogAST.Input() warnings, errors = [], [] coord = False # Keep track of the number of terminator statements follow the input # useful if we want to render graphs from the SDL model terminators = len(context.terminators) for child in root.getChildren(): if child.type == lexer.CIF: # Get symbol coordinates i.pos_x, i.pos_y, i.width, i.height = cif(child) coord = True elif child.type == lexer.INPUTLIST: i.inputString = get_input_string(child) i.line = child.getLine() i.charPositionInLine = child.getCharPositionInLine() # get input name and parameters (support for one parameter only) sig_param_type, user_param_type = None, None inputnames = [c for c in child.getChildren() if c.type != lexer.PARAMS] for inputname in inputnames: for inp_sig in context.input_signals: if inp_sig['name'].lower() == inputname.text.lower(): i.inputlist.append(inp_sig['name']) sig_param_type = inp_sig.get('type') break else: for timer in chain(context.timers, context.global_timers): if timer.lower() == inputname.text.lower(): i.inputlist.append(timer.lower()) break else: errors.append('Input signal or timer not declared: ' + inputname.toString() + ' (line ' + str(i.line) + ')') if len(inputnames) > 1 and sig_param_type is not None: errors.append('Inputs in a list shall not expect parameters') # Parse all parameters (then check that there is only one) inputparams = [c.getChildren() for c in child.getChildren() if c.type == lexer.PARAMS] if len(inputparams) > 1: # user entered e.g. INPUT a(x), b(y) instead of INPUT a,b errors.append('Only one input can have a parameter') elif len(inputparams) == 1 and sig_param_type is not None and \ len(inputparams[0]) == 1: user_param, = inputparams[0] try: user_param_type = find_variable(user_param.text, context) try: compare_types(sig_param_type, user_param_type) except TypeError as err: errors.append('Parameter type does not match with ' 'signal declaration (expecting ' 'a variable of type ' + sig_param_type.ReferencedTypeName + ') -' + str(err)) else: # Store parameter only if everything is OK i.parameters = [user_param.text.lower()] except AttributeError as err: errors.append(str(err)) elif inputparams or sig_param_type: errors.append('Wrong number of parameters or type mismatch') # Report errors with symbol coordinates if coord: errors = [[e, [i.pos_x, i.pos_y]] for e in errors] warnings = [[w, [i.pos_x, i.pos_y]] for w in warnings] elif child.type == lexer.ASTERISK: # Asterisk means: all inputs not processed explicitely # Here we do not set the input list - it is set after # all other inputs are processed (see state function) i.inputString = get_input_string(child) i.line = child.getLine() i.charPositionInLine = child.getCharPositionInLine() elif child.type == lexer.PROVIDED: warnings.append('"PROVIDED" expressions not supported') i.provided = 'Provided' elif child.type == lexer.TRANSITION: trans, err, warn = transition( child, parent=i, context=context) errors.extend(err) warnings.extend(warn) i.transition = trans # Associate a reference to the transition to the list of inputs # The reference is an index to process.transitions table context.transitions.append(trans) i.transition_id = len(context.transitions) - 1 elif child.type == lexer.COMMENT: i.comment, _, _ = end(child) elif child.type == lexer.HYPERLINK: i.hyperlink = child.getChild(0).toString()[1:-1] else: warnings.append('Unsupported INPUT child type: ' + str(child.type)) # At the end of the input parsing, get the the list of terminators that # follow the input transition by making a diff with the list at process # level (we counted the number of terminators before parsing the input) i.terminators = list(context.terminators[terminators:]) return i, errors, warnings def state(root, parent, context): ''' Parse a STATE. "parent" is used to compute absolute coordinates "context" is the AST used to store global data (process/procedure) ''' errors = [] warnings = [] state_def = ogAST.State() asterisk_state = False asterisk_input = None for child in root.getChildren(): if child.type == lexer.CIF: # Get symbol coordinates (state_def.pos_x, state_def.pos_y, state_def.width, state_def.height) = cif(child) elif child.type == lexer.STATELIST: # State name(state_def) state_def.inputString = get_input_string(child) state_def.line = child.getLine() state_def.charPositionInLine = child.getCharPositionInLine() for statename in child.getChildren(): state_def.statelist.append(statename.toString()) elif child.type == lexer.ASTERISK: asterisk_state = True state_def.inputString = get_input_string(child) state_def.line = child.getLine() state_def.charPositionInLine = child.getCharPositionInLine() exceptions = [c.toString() for c in child.getChildren()] for st in context.mapping: if st not in (exceptions, 'START'): state_def.statelist.append(st) elif child.type == lexer.INPUT: # A transition triggered by an INPUT inp, err, warn = \ input_part(child, parent=state_def, context=context) errors.extend(err) warnings.extend(warn) try: for statename in state_def.statelist: context.mapping[statename.lower()].append(inp) except KeyError: warnings.append('State definition missing') state_def.inputs.append(inp) if inp.inputString.strip() == '*': if asterisk_input: errors.append('Multiple asterisk inputs under state ' + str(state_def.inputString)) else: asterisk_input = inp elif child.type == lexer.CONNECT: comp_states = (comp.statename for comp in context.composite_states) if asterisk_state or len(state_def.statelist) != 1 \ or state_def.statelist[0].lower() not in comp_states: errors.append('State {} is not a composite state and cannot ' 'be followed by a connect statement' .format(state_def.statelist[0])) conn_part, err, warn = connect_part(child, state_def, context) state_def.connects.append(conn_part) warnings.extend(warn) errors.extend(err) elif child.type == lexer.COMMENT: state_def.comment, _, _ = end(child) elif child.type == lexer.HYPERLINK: state_def.hyperlink = child.getChild(0).toString()[1:-1] else: warnings.append('Unsupported STATE definition child type: ' + str(child.type)) # post-processing: if state is followed by an ASTERISK input, the exact # list of inputs can be updated. Possible only if context has signals if context.input_signals and asterisk_input: explicit_inputs = set() for inp in state_def.inputs: explicit_inputs |= set(inp.inputlist) # Keep only inputs that are not explicitely defined input_signals = (sig['name'] for sig in context.input_signals) remaining_inputs = set(input_signals) - explicit_inputs asterisk_input.inputlist = list(remaining_inputs) # post-processing: if state is composite, add link to the content if len(state_def.statelist) == 1 and not asterisk_state: for each in context.composite_states: if each.statename.lower() == state_def.statelist[0].lower(): state_def.composite = each return state_def, errors, warnings def connect_part(root, parent, context): ''' Connection of a nested state exit point with a transition Very similar to INPUT ''' errors, warnings = [], [] coord = False conn = ogAST.Connect() try: statename = parent.statelist[0].lower() except AttributeError: # Ignore missing parent/statelist to allow local syntax check statename = '' id_token = [] # Keep track of the number of terminator statements follow the input # useful if we want to render graphs from the SDL model terms = len(context.terminators) # Retrieve composite state try: nested, = (comp for comp in context.composite_states if comp.statename.lower() == statename.lower()) except ValueError: # Ignore unexisting state - to allow local syntax check nested = ogAST.CompositeState() for child in root.getChildren(): if child.type == lexer.CIF: # Get symbol coordinates conn.pos_x, conn.pos_y, conn.width, conn.height = cif(child) coord = True elif child.type == lexer.ID: id_token.append(child) conn.connect_list.append(child.toString().lower()) elif child.type == lexer.ASTERISK: id_token.append(child) conn.connect_list = nested.state_exitpoints elif child.type == lexer.TRANSITION: trans, err, warn = transition(child, parent=conn, context=context) errors.extend(err) warnings.extend(warn) context.transitions.append(trans) trans_id = len(context.transitions) - 1 conn.transition_id = trans_id conn.transition = trans elif child.type == lexer.HYPERLINK: conn.hyperlink = child.getChild(0).toString()[1:-1] elif child.type == lexer.COMMENT: conn.comment, _, _ = end(child) else: warnings.append('Unsupported CONNECT PART child type: ' + sdl92Parser.tokenNames[child.type]) if not conn.connect_list: conn.connect_list.append('') if not id_token: conn.inputString = '' conn.line = root.getLine() conn.charPositionInLine = root.getCharPositionInLine() else: conn.line = id_token[0].getLine() conn.charPositionInLine = id_token[0].getCharPositionInLine() conn.inputString = token_stream(id_token[0]).toString( id_token[0].getTokenStartIndex(), id_token[-1].getTokenStopIndex()) for exitp in conn.connect_list: if exitp != '' and not exitp in nested.state_exitpoints: errors.append('Exit point {ep} not defined in state {st}' .format(ep=exitp, st=statename)) terminators = [term for term in nested.terminators if term.kind == 'return' and term.inputString.lower() == exitp] if not terminators: errors.append('No {rs} return statement in nested state {st}' .format(rs=exitp, st=statename)) for each in terminators: # Set next transition, exact id to be found in postprocessing each.next_trans = trans # Set list of terminators conn.terminators = list(context.terminators[terms:]) # Report errors with symbol coordinates if coord: errors = [[e, [conn.pos_x, conn.pos_y]] for e in errors] warnings = [[w, [conn.pos_x, conn.pos_y]] for w in warnings] return conn, errors, warnings def cif(root): ''' Return the CIF coordinates ''' result = [] for child in root.getChildren(): if child.type == lexer.DASH: val = -int(child.getChild(0).toString()) else: val = int(child.toString()) result.append(val) return result def start(root, parent=None, context=None): ''' Parse the START transition ''' errors = [] warnings = [] if isinstance(context, ogAST.Procedure): s = ogAST.Procedure_start() elif isinstance(context, ogAST.CompositeState): s = ogAST.CompositeState_start() else: s = ogAST.Start() # Keep track of the number of terminator statements following the start # useful if we want to render graphs from the SDL model terminators = len(context.terminators) for child in root.getChildren(): if child.type == lexer.CIF: # Get symbol coordinates s.pos_x, s.pos_y, s.width, s.height = cif(child) elif child.type == lexer.ID: # in nested states, START can be followed by the entry point name s.inputString = child.toString().lower() + '_START' elif child.type == lexer.TRANSITION: s.transition, err, warn = transition( child, parent=s, context=context) errors.extend(err) warnings.extend(warn) context.transitions.append(s.transition) context.mapping[s.inputString or 'START'] = \ len(context.transitions) - 1 elif child.type == lexer.COMMENT: s.comment, _, _ = end(child) elif child.type == lexer.HYPERLINK: s.hyperlink = child.getChild(0).toString()[1:-1] else: warnings.append('START unsupported child type: ' + str(child.type)) # At the end of the START parsing, get the the list of terminators that # follow the START transition by making a diff with the list at process # level (we counted the number of terminators before parsing the START) s.terminators = list(context.terminators[terminators:]) return s, errors, warnings def end(root, parent=None, context=None): ''' Parse a comment symbol ''' c = ogAST.Comment() c.line = root.getLine() c.charPositionInLine = root.getCharPositionInLine() for child in root.getChildren(): if child.type == lexer.CIF: # Get symbol coordinates c.pos_x, c.pos_y, c.width, c.height = cif(child) elif child.type == lexer.STRING: c.inputString = child.toString()[1:-1] elif child.type == lexer.HYPERLINK: c.hyperlink = child.getChild(0).toString()[1:-1] return c, [], [] def procedure_call(root, parent, context): ''' Parse a PROCEDURE CALL (synchronous required interface) ''' # Same as OUTPUT for external procedures out_ast = ogAST.ProcedureCall() _, err, warn = output(root, parent, out_ast, context) return out_ast, err, warn def outputbody(root, context): ''' Parse an output body (the content excluding the CIF statement) ''' errors = [] warnings = [] body = {'outputName': '', 'params': []} for child in root.getChildren(): if child.type == lexer.ID: body['outputName'] = child.text if child.text.lower() not in valid_output(context): errors.append('"' + child.text + '" is not defined in the current scope') elif child.type == lexer.PARAMS: body['params'], err, warn = expression_list( child, context) errors.extend(err) warnings.extend(warn) elif child.type == lexer.TO: pass # TODO: better support of TO primitive else: warnings.append('Unsupported output body type:' + str(child.type)) # Check/set the type of each param try: check_and_fix_op_params(body.get('outputName', ''), body.get('params', []), context) except (AttributeError, TypeError) as op_err: errors.append(str(op_err) + ' - ' + get_input_string(root)) LOG.debug('[outputbody] call check_and_fix_op_params : ' + get_input_string(root) + str(op_err)) LOG.debug(str(traceback.format_exc())) if body['params']: body['tmpVars'] = [] for _ in body['params']: body['tmpVars'].append(tmp()) return body, errors, warnings def output(root, parent, out_ast=None, context=None): ''' Parse an OUTPUT : set of asynchronous required interface(s) ''' errors = [] warnings = [] coord = False out_ast = out_ast or ogAST.Output() # syntax checker passes no ast for child in root.getChildren(): if child.type == lexer.CIF: # Get symbol coordinates out_ast.pos_x, out_ast.pos_y, out_ast.width, out_ast.height = \ cif(child) coord = True elif child.type == lexer.OUTPUT_BODY: out_ast.inputString = get_input_string(child) out_ast.line = child.getLine() out_ast.charPositionInLine = child.getCharPositionInLine() body, err, warn = outputbody(child, context) errors.extend(err) warnings.extend(warn) out_ast.output.append(body) elif child.type == lexer.COMMENT: out_ast.comment, _, _ = end(child) elif child.type == lexer.HYPERLINK: out_ast.hyperlink = child.getChild(0).toString()[1:-1] else: warnings.append('Unsupported output child type: ' + str(child.type)) # Report errors with symbol coordinates if coord: errors = [[e, [out_ast.pos_x, out_ast.pos_y]] for e in errors] warnings = [[w, [out_ast.pos_x, out_ast.pos_y]] for w in warnings] return out_ast, errors, warnings def alternative_part(root, parent, context): ''' Parse a decision answer ''' errors = [] warnings = [] ans = ogAST.Answer() coord = False for child in root.getChildren(): if child.type == lexer.CIF: # Get symbol coordinates ans.pos_x, ans.pos_y, ans.width, ans.height = cif(child) coord = True elif child.type == lexer.CLOSED_RANGE: ans.kind = 'closed_range' cl0, err0, warn0 = expression(child.getChild(0), context) cl1, err1, warn1 = expression(child.getChild(1), context) errors.extend(err0) errors.extend(err1) warnings.extend(warn0) warnings.extend(warn1) ans.closedRange = [cl0, cl1] elif child.type == lexer.CONSTANT: ans.kind = 'constant' ans.constant, err, warn = expression( child.getChild(0), context) errors.extend(err) warnings.extend(warn) ans.openRangeOp = ogAST.ExprEq elif child.type == lexer.OPEN_RANGE: ans.kind = 'open_range' for c in child.getChildren(): if c.type == lexer.CONSTANT: ans.constant, err, warn = expression( c.getChild(0), context) errors.extend(err) warnings.extend(warn) if not ans.openRangeOp: ans.openRangeOp = ogAST.ExprEq else: ans.openRangeOp = EXPR_NODE[c.type] elif child.type == lexer.INFORMAL_TEXT: ans.kind = 'informal_text' ans.informalText = child.getChild(0).toString()[1:-1] elif child.type == lexer.TRANSITION: ans.transition, err, warn = transition( child, parent=ans, context=context) errors.extend(err) warnings.extend(warn) elif child.type == lexer.HYPERLINK: ans.hyperlink = child.getChild(0).toString()[1:-1] else: warnings.append('Unsupported answer type: ' + str(child.type)) if child.type in (lexer.CLOSED_RANGE, lexer.CONSTANT, lexer.OPEN_RANGE, lexer.INFORMAL_TEXT): ans.inputString = get_input_string(child) ans.line = child.getLine() ans.charPositionInLine = child.getCharPositionInLine() # Report errors with symbol coordinates if coord: errors = [[e, [ans.pos_x, ans.pos_y]] for e in errors] warnings = [[w, [ans.pos_x, ans.pos_y]] for w in warnings] return ans, errors, warnings def decision(root, parent, context): ''' Parse a DECISION ''' errors = [] warnings = [] dec = ogAST.Decision() dec.tmpVar = tmp() has_else = False for child in root.getChildren(): if child.type == lexer.CIF: # Get symbol coordinates dec.pos_x, dec.pos_y, dec.width, dec.height = cif(child) elif child.type == lexer.QUESTION: dec.kind = 'question' decisionExpr, err, warn = expression( child.getChild(0), context) for e in err: errors.append([e, [dec.pos_x, dec.pos_y]]) for w in warn: warnings.append([w, [dec.pos_x, dec.pos_y]]) dec.question = decisionExpr dec.inputString = dec.question.inputString dec.line = dec.question.line dec.charPositionInLine = dec.question.charPositionInLine elif child.type == lexer.INFORMAL_TEXT: dec.kind = 'informal_text' dec.inputString = get_input_string(child) dec.informalText = child.getChild(0).toString()[1:-1] dec.line = child.getLine() dec.charPositionInLine = child.getCharPositionInLine() elif child.type == lexer.ANY: dec.kind = 'any' elif child.type == lexer.COMMENT: dec.comment, _, _ = end(child) elif child.type == lexer.HYPERLINK: dec.hyperlink = child.getChild(0).toString()[1:-1] elif child.type == lexer.ANSWER: ans, err, warn = alternative_part(child, parent, context) errors.extend(err) warnings.extend(warn) dec.answers.append(ans) elif child.type == lexer.ELSE: a = ogAST.Answer() a.inputString = child.toString() for c in child.getChildren(): if c.type == lexer.CIF: a.pos_x, a.pos_y, a.width, a.height = cif(c) elif c.type == lexer.TRANSITION: a.transition, err, warn = transition( c, parent=a, context=context) errors.extend(err) warnings.extend(warn) elif child.type == lexer.HYPERLINK: a.hyperlink = child.getChild(0).toString()[1:-1] a.kind = 'else' dec.answers.append(a) has_else = True else: warnings.append(['Unsupported DECISION child type: ' + str(child.type), [dec.pos_x, dec.pos_y]]) # Make type checks to be sure that question and answers are compatible covered_ranges = defaultdict(list) qmin, qmax = 0, 0 need_else = False for ans in dec.answers: if ans.kind in ('constant', 'open_range'): expr = ans.openRangeOp() expr.left = dec.question expr.right = ans.constant try: fix_expression_types(expr, context) if dec.question.exprType == UNKNOWN_TYPE: dec.question = expr.left ans.constant = expr.right q_basic = find_basic_type(dec.question.exprType) a_basic = find_basic_type(ans.constant.exprType) if not q_basic.kind.startswith('Integer'): continue # numeric type -> find the range covered by this answer if a_basic.Min != a_basic.Max: # Not a constant or a raw number, range is not fix need_else = True continue val_a = int(float(a_basic.Min)) qmin, qmax = int(float(q_basic.Min)), int(float(q_basic.Max)) # Check the operator to compute the range reachable = True if ans.openRangeOp == ogAST.ExprLe: if qmin <= val_a: covered_ranges[ans].append((qmin, val_a)) else: reachable = False elif ans.openRangeOp == ogAST.ExprLt: if qmin < val_a: covered_ranges[ans].append((qmin, val_a - 1)) else: reachable = False elif ans.openRangeOp == ogAST.ExprGt: if qmax > val_a: covered_ranges[ans].append((val_a + 1, qmax)) else: reachable = False elif ans.openRangeOp == ogAST.ExprGe: if qmax >= val_a: covered_ranges[ans].append((val_a, qmax)) else: reachable = False elif ans.openRangeOp == ogAST.ExprEq: if qmin <= val_a <= qmax: covered_ranges[ans].append((val_a, val_a)) else: reachable = False elif ans.openRangeOp == ogAST.ExprNeq: if qmin == val_a: covered_ranges[ans].append((qmin + 1, qmax)) elif qmax == val_a: covered_ranges[ans].append((qmin, qmax - 1)) elif qmin < val_a < qmax: covered_ranges[ans].append((qmin, val_a - 1)) covered_ranges[ans].append((val_a + 1, qmax)) else: warnings.append('Condition is always true: {} /= {}' .format(dec.inputString, ans.inputString)) else: warnings.append('Unsupported range expression') if not reachable: warnings.append('Decision "{}": ' 'Unreachable branch "{}"' .format(dec.inputString, ans.inputString)) except (AttributeError, TypeError) as err: errors.append('Types are incompatible in DECISION: ' 'question (' + expr.left.inputString + ', type= ' + type_name(expr.left.exprType) + '), answer (' + expr.right.inputString + ', type= ' + type_name(expr.right.exprType) + ') ' + str(err)) elif ans.kind == 'closed_range': if not is_numeric(dec.question.exprType): errors.append('Closed range are only for numerical types') continue for ast_type, idx in zip((ogAST.ExprGe, ogAST.ExprLe), (0, 1)): expr = ast_type() expr.left = dec.question expr.right = ans.closedRange[idx] try: fix_expression_types(expr, context) if dec.question.exprType == UNKNOWN_TYPE: dec.question = expr.left ans.closedRange[idx] = expr.right except (AttributeError, TypeError) as err: errors.append('Types are incompatible in DECISION: ' 'question (' + expr.left.inputString + ', type= ' + type_name(expr.left.exprType) + '), answer (' + expr.right.inputString + ', type= ' + type_name(expr.right.exprType) + ') ' + str(err)) q_basic = find_basic_type(dec.question.exprType) if not q_basic.kind.startswith('Integer'): continue # numeric type -> find the range covered by this answer a0_basic = find_basic_type(ans.closedRange[0].exprType) a1_basic = find_basic_type(ans.closedRange[1].exprType) if a0_basic.Min != a0_basic.Max or a1_basic.Min != a1_basic.Max: # Not a constant or a raw number, range is not fix need_else = True continue qmin, qmax = int(float(q_basic.Min)), int(float(q_basic.Max)) a0_val = int(float(a0_basic.Min)) a1_val = int(float(a1_basic.Max)) if a0_val < qmin: warnings.append('Decision "{dec}": ' 'Range [{a0} .. {qmin}] is unreachable' .format(a0=a0_val, qmin=qmin - 1, dec=dec.inputString)) if a1_val > qmax: warnings.append('Decision "{dec}": ' 'Range [{qmax} .. {a1}] is unreachable' .format(qmax=qmax + 1, a1=a1_val, dec=dec.inputString)) if (a0_val < qmin and a1_val < qmin) or (a0_val > qmax and a1_val > qmax): warnings.append('Decision "{dec}": Unreachable branch {l}:{h}' .format(dec=dec.inputString, l=a0_val, h=a1_val)) covered_ranges[ans].append((int(float(a0_basic.Min)), int(float(a1_basic.Max)))) # Check the following: # (1) no overlap between covered ranges in decision answers # (2) no gap in the coverage of the decision possible values # (3) ELSE branch, if present, can be reached # (4) if an answer uses a non-ground expression an ELSE is there q_ranges = [(qmin, qmax)] if is_numeric(dec.question.exprType) else [] for each in combinations(covered_ranges.viewitems(), 2): for comb in combinations( chain.from_iterable(val[1] for val in each), 2): comb_overlap = (max(comb[0][0], comb[1][0]), min(comb[0][1], comb[1][1])) if comb_overlap[0] <= comb_overlap[1]: # (1) - check for overlaps errors.append('Decision "{d}": answers {a1} and {a2} ' 'are overlapping in range [{o1} .. {o2}]' .format(d=dec.inputString, a1=each[0][0].inputString, a2=each[1][0].inputString, o1=comb_overlap[0], o2=comb_overlap[1])) new_q_ranges = [] # for minq, maxq in q_ranges: # (2) Check that decision range is fully covered for ans_ref, ranges in covered_ranges.viewitems(): for mina, maxa in ranges: for minq, maxq in q_ranges: left = (minq, min(maxq, mina - 1)) right = (max(minq, maxa + 1), maxq) if mina > minq and maxa < maxq: new_q_ranges.extend([left, right]) elif mina <= minq and maxa >= maxq: pass elif mina <= minq: new_q_ranges.append(right) elif maxa >= maxq: new_q_ranges.append(left) q_ranges, new_q_ranges = new_q_ranges, [] if not has_else: for minq, maxq in q_ranges: errors.append('Decision "{}": No answer to cover range [{} .. {}]' .format(dec.inputString, minq, maxq)) elif has_else and is_numeric(dec.question.exprType) and not q_ranges: # (3) Check that ELSE branch is reachable warnings.append('Decision "{}": ELSE branch is unreachable' .format(dec.inputString)) if need_else and not has_else: # (4) Answers use non-ground expression -> there should be an ELSE warnings.append('Decision "{}": Missing ELSE branch' .format(dec.inputString)) return dec, errors, warnings def nextstate(root, context): ''' Parse a NEXTSTATE [VIA State_Entry_Point] - detect various kinds of errors when trying to enter a nested state ''' next_state_id, via, entrypoint = '', None, None errors = [] for child in root.getChildren(): if child.type == lexer.ID: next_state_id = child.text elif child.type == lexer.DASH: next_state_id = '-' elif child.type == lexer.VIA: if next_state_id.strip() != '-': via = get_input_string(root).replace( 'NEXTSTATE', '', 1).strip() entrypoint = child.getChild(0).text try: composite, = (comp for comp in context.composite_states if comp.statename.lower() == next_state_id.lower()) except ValueError: errors.append('State {} is not a composite state' .format(next_state_id)) else: if entrypoint.lower() not in composite.state_entrypoints: errors.append('State {s} has no "{p}" entrypoint' .format(s=next_state_id, p=entrypoint)) for each in composite.content.named_start: if each.inputString == entrypoint.lower() + '_START': break else: errors.append('Entrypoint {p} in state {s} is ' 'declared but not defined'.format (s=next_state_id, p=entrypoint)) else: errors.append('"History" NEXTSTATE cannot have a "via" clause') else: errors.append('NEXTSTATE undefined construct') if not via: # check that if the nextstate is nested, it has a START symbol try: composite, = (comp for comp in context.composite_states if comp.statename.lower() == next_state_id.lower()) if not composite.content.start: errors.append('Composite state "{}" has no unnamed ' 'START symbol'.format(composite.statename)) except ValueError: pass return next_state_id, via, entrypoint, errors def terminator_statement(root, parent, context): ''' Parse a terminator (NEXTSTATE, JOIN, STOP) ''' errors = [] warnings = [] t = ogAST.Terminator() coord = False for term in root.getChildren(): if term.type == lexer.CIF: t.pos_x, t.pos_y, t.width, t.height = cif(term) coord = True elif term.type == lexer.LABEL: lab, err, warn = label(term, parent=parent) errors.extend(err) warnings.extend(warn) t.label = lab context.labels.append(lab) lab.terminators = [t] elif term.type == lexer.NEXTSTATE: t.kind = 'next_state' t.inputString, t.via, t.entrypoint, err = nextstate(term, context) if err: errors.extend(err) t.line = term.getChild(0).getLine() t.charPositionInLine = term.getChild(0).getCharPositionInLine() # Add next state infos at process level # Used in rendering backends to merge a NEXTSTATE with a STATE context.terminators.append(t) # post-processing: if nextatate is nested, add link to the content # (normally handled at state level, but if state is not defined # standalone, the nextstate must hold the composite content) if t.inputString != '-': for each in context.composite_states: if each.statename.lower() == t.inputString.lower(): t.composite = each elif term.type == lexer.JOIN: t.kind = 'join' t.inputString = term.getChild(0).toString() t.line = term.getChild(0).getLine() t.charPositionInLine = term.getChild(0).getCharPositionInLine() context.terminators.append(t) elif term.type == lexer.STOP: t.kind = 'stop' context.terminators.append(t) elif term.type == lexer.RETURN: t.kind = 'return' if term.children: t.return_expr, err, warn = expression( term.getChild(0), context) t.inputString = t.return_expr.inputString errors.extend(err) warnings.extend(warn) context.terminators.append(t) elif term.type == lexer.COMMENT: t.comment, _, _ = end(term) elif term.type == lexer.HYPERLINK: t.hyperlink = term.getChild(0).toString()[1:-1] else: warnings.append('Unsupported terminator type: ' + str(term.type)) # Report errors with symbol coordinates if coord: errors = [[e, [t.pos_x, t.pos_y]] for e in errors] warnings = [[w, [t.pos_x, t.pos_y]] for w in warnings] return t, errors, warnings def transition(root, parent, context): ''' Parse a transition ''' errors = [] warnings = [] trans = ogAST.Transition() # Used to list terminators in this transition terminators = {'trans': len(context.terminators)} #terminators = len(context.terminators) for child in root.getChildren(): if child.type == lexer.PROCEDURE_CALL: proc_call, err, warn = procedure_call( child, parent=parent, context=context) errors.extend(err) warnings.extend(warn) trans.actions.append(proc_call) parent = proc_call elif child.type == lexer.LABEL: term_count = len(context.terminators) lab, err, warn = label(child, parent=parent) terminators[lab] = term_count errors.extend(err) warnings.extend(warn) trans.actions.append(lab) parent = lab context.labels.append(lab) elif child.type == lexer.TASK: tas, err, warn = task( child, parent=parent, context=context) errors.extend(err) warnings.extend(warn) trans.actions.append(tas) parent = tas elif child.type == lexer.TASK_BODY: t, err, warn = task_body(child, context) t.inputString = get_input_string(child) t.line = child.getLine() t.charPositionInLine = child.getCharPositionInLine() errors.extend(err) warnings.extend(warn) trans.actions.append(t) parent = t elif child.type == lexer.OUTPUT: out_ast = ogAST.Output() _, err, warn = output(child, parent=parent, out_ast=out_ast, context=context) errors.extend(err) warnings.extend(warn) trans.actions.append(out_ast) parent = out_ast elif child.type == lexer.DECISION: dec, err, warn = decision( child, parent=parent, context=context) errors.extend(err) warnings.extend(warn) trans.actions.append(dec) parent = dec elif child.type == lexer.TERMINATOR: term, err, warn = terminator_statement(child, parent=parent, context=context) errors.extend(err) warnings.extend(warn) trans.terminator = term else: warnings.append('Unsupported symbol in transition, type: ' + str(child.type)) # At the end of the transition parsing, get the the list of terminators # the transition contains by making a diff with the list at context # level (we counted the number of terminators before parsing the item) trans.terminators = list(context.terminators[terminators.pop('trans'):]) # Also update the list of terminators of each label in the transition for lab, term_count in terminators.viewitems(): lab.terminators = list(context.terminators[term_count:]) return trans, errors, warnings def assign(root, context): ''' Parse an assignation (a := b) in a task symbol ''' errors = [] warnings = [] expr = ogAST.ExprAssign( get_input_string(root), root.getLine(), root.getCharPositionInLine() ) expr.kind = 'assign' if root.children[0].type == lexer.CALL: expr.left, err, warn = call_expression(root.children[0], context) elif root.children[0].type == lexer.SELECTOR: expr.left, err, warn = selector_expression(root.children[0], context) else: expr.left, err, warn = primary_variable(root.children[0], context) warnings.extend(warn) errors.extend(err) expr.right, err, warn = expression(root.children[1], context) errors.extend(err) warnings.extend(warn) try: fix_expression_types(expr, context) # Assignment with numerical value: check range basic = find_basic_type(expr.left.exprType) if basic.kind.startswith(('Integer', 'Real')): check_range(basic, find_basic_type(expr.right.exprType)) except(AttributeError, TypeError) as err: errors.append('Types are incompatible in assignment: left (' + expr.left.inputString + ', type= ' + type_name(expr.left.exprType) + '), right (' + expr.right.inputString + ', type= ' + (type_name(expr.right.exprType) if expr.right.exprType else 'Unknown') + ') ' + str(err)) else: expr.right.exprType = expr.left.exprType return expr, errors, warnings def for_range(root, context): ''' Parse a RANGE statement (in a FOR loop) ''' errors = [] warnings = [] # start and stop are expressions result = {'start': None, 'stop': None, 'step': 1} expr = [] for child in root.getChildren(): if child.type == lexer.GROUND: ground, err, warn = expression(child.getChild(0), context) errors.extend(err) warnings.extend(warn) expr.append(ground) elif child.type == lexer.INT: result['step'] = int(child.text) else: warnings.append('Unsupported child type in RANGE: ' + str(child.type)) for range_item in expr: if not range_item: errors.append('Range must use a ground expression: ' + range_item.inputString) if len(expr) == 2: result['start'], result['stop'] = expr elif len(expr) == 1: result['stop'] = expr[0] else: errors.append('Incorrect range expression') return result, errors, warnings def for_loop(root, context): ''' Parse a FOR LOOP in a task symbol ''' errors = [] warnings = [] forloop = {'var': '', 'type': None, 'list': '', 'range': None, 'transition': None} for child in root.getChildren(): if child.type == lexer.ID: forloop['var'] = child.text # Implicit variable declaration for the iterator context_scope = dict(context.variables) elif child.type == lexer.VARIABLE: forloop['list'], err, warn = primary_variable(child, context) warnings.extend(warn) errors.extend(err) elif child.type == lexer.CALL: forloop['list'], err, warn = call_expression(child, context) warnings.extend(warn) errors.extend(err) elif child.type == lexer.SELECTOR: forloop['list'], err, warn = selector_expression(child, context) warnings.extend(warn) errors.extend(err) elif child.type == lexer.RANGE: forloop['range'], err, warn = for_range(child, context) errors.extend(err) warnings.extend(warn) elif child.type == lexer.TRANSITION: # First we need to define the type of the iterator (and check it) if forloop['list']: basic_type = find_basic_type(forloop['list'].exprType) if basic_type.kind != 'SequenceOfType': errors.append('Variable {} is not iterable' .format(forloop['list'].inputString)) else: forloop['type'] = basic_type.type # Set the type of the iterator context.variables[forloop['var']] = (forloop['type'], 0) else: # Using a range - set type of iterator to standard integer start_expr, stop_expr = forloop['range']['start'], \ forloop['range']['stop'] if not start_expr: r_min = '0' else: basic = find_basic_type(start_expr.exprType) r_min = basic.Min if basic != UNKNOWN_TYPE else '0' basic = find_basic_type(stop_expr.exprType) r_max = basic.Max if basic != UNKNOWN_TYPE else '4294967295' # basic may be UNKNOWN_TYPE if the expression is a # reference to an ASN.1 constant - their values are not # currently visible to the SDL parser result_type = type('for_range', (INT32,), {'Min': r_min, 'Max': r_max}) context.variables[forloop['var']] = (result_type, 0) forloop['transition'], err, warn = transition( child, parent=for_loop, context=context) errors.extend(err) warnings.extend(warn) else: warnings.append('Unsupported child type in FOR body' + str(child.type)) context.variables = context_scope return forloop, errors, warnings def task_body(root, context): ''' Parse the body of a task (excluding CIF and TASK tokens) ''' errors = [] warnings = [] body = None for child in root.getChildren(): if child.type == lexer.ASSIGN: if not body: body = ogAST.TaskAssign() assig, err, warn = assign(child, context) errors.extend(err) warnings.extend(warn) body.elems.append(assig) elif child.type == lexer.INFORMAL_TEXT: if not body: body = ogAST.TaskInformalText() body.elems.append(child.getChild(0).toString()[1:-1]) elif child.type == lexer.FOR: if not body: body = ogAST.TaskForLoop() forloop, err, warn = for_loop(child, context) errors.extend(err) warnings.extend(warn) body.elems.append(forloop) else: warnings.append('Unsupported child type in task body: ' + str(child.type)) if not body: body = ogAST.TaskAssign() return body, errors, warnings def task(root, parent=None, context=None): ''' Parse a TASK symbol (assignation or informal text) ''' errors = [] warnings = [] coord = False comment, body = None, None for child in root.getChildren(): if child.type == lexer.CIF: # Get symbol coordinates pos_x, pos_y, width, height = cif(child) coord = True elif child.type == lexer.TASK_BODY: body, task_err, task_warn = task_body(child, context) body.inputString = get_input_string(child) body.line = child.getLine() body.charPositionInLine = child.getCharPositionInLine() errors.extend(task_err) warnings.extend(task_warn) elif child.type == lexer.COMMENT: comment, _, _ = end(child) elif child.type == lexer.HYPERLINK: body.hyperlink = child.getChild(0).toString()[1:-1] else: warnings.append('Unsupported child type in task definition: ' + str(child.type)) # Report errors with symbol coordinates if coord and body: body.pos_x, body.pos_y, body.width, body.height = \ pos_x, pos_y, width, height errors = [[e, [pos_x, pos_y]] for e in errors] warnings = [[w, [pos_x, pos_y]] for w in warnings] if body: body.comment = comment else: warnings.append('TASK missing content') body = ogAST.TaskAssign() return body, errors, warnings def label(root, parent, context=None): ''' Parse a LABEL symbol ''' errors = [] warnings = [] lab = ogAST.Label() coord = False for child in root.getChildren(): if child.type == lexer.CIF: # Get symbol coordinates lab.pos_x, lab.pos_y, lab.width, lab.height = cif(child) coord = True elif child.type == lexer.ID: lab.inputString = get_input_string(child) lab.line = child.getLine() lab.charPositionInLine = child.getCharPositionInLine() elif child.type == lexer.HYPERLINK: lab.hyperlink = child.getChild(0).toString()[1:-1] else: warnings.append( 'Unsupported child type in label definition: ' + str(child.type)) # Report errors with symbol coordinates if coord: errors = [[e, [lab.pos_x, lab.pos_y]] for e in errors] warnings = [[w, [lab.pos_x, lab.pos_y]] for w in warnings] return lab, errors, warnings def pr_file(root): ''' Complete PR model - can be made up from several files/strings ''' errors = [] warnings = [] ast = ogAST.AST() global DV # In case no ASN.1 files are parsed, the DV structure is pre-initialised # This to allow SDL types injection in ASN1 ASTs DV = type("ASNParseTree", (object, ), {"types": {}, "exportedVariables": {}, "asn1Modules": []}) # Re-order the children of the AST to make sure system and use clauses # are parsed before process definition - to get signal definitions # and data typess references. processes, uses, systems = [], [], [] for child in root.getChildren(): ast.pr_files.add(node_filename(child)) if child.type == lexer.PROCESS: processes.append(child) elif child.type == lexer.USE: uses.append(child) elif child.type == lexer.SYSTEM: systems.append(child) else: LOG.error('Unsupported construct in PR:' + str(child.type)) for child in uses: LOG.debug('USE clause') # USE clauses can contain a CIF comment with the ASN.1 filename use_clause_subs = child.getChildren() asn1_filename = None for clause in use_clause_subs: if clause.type == lexer.ASN1: asn1_filename = clause.getChild(0).text[1:-1] ast.asn1_filenames.append(asn1_filename) else: ast.use_clauses.append(clause.text) # if not asn1_filename: # # Look for case insentitive pr file and add it to AST # search = fnmatch.translate(clause.text + '.pr') # searchobj = re.compile(search, re.IGNORECASE) # for each in os.listdir('.'): # if searchobj.match(each): # print 'found', each try: DV = parse_asn1(tuple(ast.asn1_filenames), ast_version=ASN1.UniqueEnumeratedNames, flags=[ASN1.AstOnly]) ast.asn1Modules = DV.asn1Modules # Add constants defined in the ASN.1 modules (for visibility) for mod in ast.asn1Modules: ast.asn1_constants.extend(DV.exportedVariables[mod]) except (ImportError, NameError) as err: # Can happen if DataView.py is not there LOG.info('USE Clause did not contain ASN.1 filename') LOG.debug(str(err)) for child in systems: LOG.debug('found SYSTEM') system, err, warn = system_definition(child, parent=ast) errors.extend(err) warnings.extend(warn) ast.systems.append(system) def find_processes(block): ''' Recursively find processes in a system ''' try: result = [proc for proc in block.processes if not proc.referenced] except AttributeError: result = [] for nested in block.blocks: result.extend(find_processes(nested)) return result ast.processes.extend(find_processes(system)) for child in processes: # process definition at root level (must be referenced in a system) LOG.debug('found PROCESS') process, err, warn = process_definition(child, parent=ast) ast.processes.append(process) process.dataview = types() process.asn1Modules = ast.asn1Modules errors.extend(err) warnings.extend(warn) LOG.debug('all files: ' + str(ast.pr_files)) # Since SDL type declarations are injected in ASN.1 ast, # The ASN.1 ASTs needs to be copied at the end of PR parsing process # and not just after the ASN1 specific parsing ast.dataview = types() for mod in DV.exportedVariables: ast.asn1_constants.extend(DV.exportedVariables[mod]) return ast, errors, warnings def add_to_ast(ast, filename=None, string=None): ''' Parse a file or string and add it to an AST ''' errors, warnings = [], [] try: parser = parser_init(filename=filename, string=string) except IOError: LOG.error('parser_init failed') raise # Use Sam & Max output capturer to get errors from ANTLR parser with samnmax.capture_ouput() as (stdout, stderr): tree_rule_return_scope = parser.pr_file() for e in stderr: errors.append([e.strip()]) for w in stdout: warnings.append([w.strip()]) # Root of the AST is of type antlr3.tree.CommonTree # Add it as a child of the common tree subtree = tree_rule_return_scope.tree token_str = parser.getTokenStream() children_before = set(ast.children) # addChild does not simply add the subtree - it flattens it if necessary # this means that possibly SEVERAL subtrees can be added. We must set # the token_stream reference to all of them. ast.addChild(subtree) for tree in set(ast.children) - children_before: tree.token_stream = token_str return errors, warnings def parse_pr(files=None, string=None): ''' Parse SDL files (.pr) and/or string ''' warnings = [] errors = [] files = files or [] # define a common tree to combine several PR inputs common_tree = antlr3.tree.CommonTree(None) for filename in files: sys.path.insert(0, os.path.dirname(filename)) for filename in files: err, warn = add_to_ast(common_tree, filename=filename) errors.extend(err) warnings.extend(warn) if string: err, warn = add_to_ast(common_tree, string=string) errors.extend(err) warnings.extend(warn) # At the end when common tree is complete, perform the parsing og_ast, err, warn = pr_file(common_tree) for error in err: errors.append([error] if type(error) is not list else error) for warning in warn: warnings.append([warning] if type(warning) is not list else warning) # Post-parsing: additional semantic checks # check that all NEXTSTATEs have a correspondingly defined STATE # (except the '-' state, which means "stay in the same state') for process in og_ast.processes: for ns in [t.inputString.lower() for t in process.terminators if t.kind == 'next_state']: if not ns in [s.lower() for s in process.mapping.viewkeys()] + ['-']: errors.append(['State definition missing: ' + ns.upper()]) # TODO: do the same with JOIN/LABEL return og_ast, warnings, errors def parseSingleElement(elem='', string=''): ''' Parse any symbol and return syntax error and AST entry Used for on-the-fly checks when user edits text and for copy/cut to create a new object ''' assert(elem in ('input_part', 'output', 'decision', 'alternative_part', 'terminator_statement', 'label', 'task', 'procedure_call', 'end', 'text_area', 'state', 'start', 'procedure', 'floating_label', 'connect_part', 'process_definition', 'proc_start', 'state_start')) # Create a dummy context, needed to place context data if elem == 'proc_start': elem = 'start' context = ogAST.Procedure() elif elem == 'state_start': elem = 'start' context = ogAST.CompositeState() else: context = ogAST.Process() LOG.debug('Parsing string: ' + string + ' with elem ' + elem) parser = parser_init(string=string) parser_ptr = getattr(parser, elem) assert parser_ptr is not None syntax_errors = [] semantic_errors = [] warnings = [] t = None if parser: with samnmax.capture_ouput() as (stdout, stderr): r = parser_ptr() for e in stderr: syntax_errors.append(e.strip()) for w in stdout: syntax_errors.append(w.strip()) # Get the root of the Antlr-AST to build our own AST entry root = r.tree root.token_stream = parser.getTokenStream() backend_ptr = eval(elem) try: t, semantic_errors, warnings = backend_ptr( root=root, parent=None, context=context) except AttributeError as err: print str(err) print (traceback.format_exc()) # Syntax checker has no visibility on variables and types # so we have to discard exceptions sent by e.g. find_variable pass return(t, syntax_errors, semantic_errors, warnings, context.terminators) def parser_init(filename=None, string=None): ''' Initialize the parser (to be called first) ''' try: char_stream = antlr3.ANTLRFileStream(filename, encoding='utf-8') except (IOError, TypeError): try: char_stream = antlr3.ANTLRStringStream(string) except TypeError as err: raise IOError('Could not parse input' + str(err)) lex = lexer.sdl92Lexer(char_stream) tokens = antlr3.CommonTokenStream(lex) parser = sdl92Parser(tokens) return parser if __name__ == '__main__': print 'This module is not callable'