audk/BaseTools/Source/Python/Common/Expression.py

## @file
# This file is used to parse and evaluate expression in directive or PCD value.
#
# Copyright (c) 2011, Intel Corporation. All rights reserved.<BR>
# This program and the accompanying materials
# are licensed and made available under the terms and conditions of the BSD License
# which accompanies this distribution.    The full text of the license may be found at
# http://opensource.org/licenses/bsd-license.php
#
# THE PROGRAM IS DISTRIBUTED UNDER THE BSD LICENSE ON AN "AS IS" BASIS,
# WITHOUT WARRANTIES OR REPRESENTATIONS OF ANY KIND, EITHER EXPRESS OR IMPLIED.

## Import Modules
#
from Common.GlobalData import *
from CommonDataClass.Exceptions import BadExpression
from CommonDataClass.Exceptions import WrnExpression
from Misc import GuidStringToGuidStructureString

ERR_STRING_EXPR         = 'This operator cannot be used in string expression: [%s].'
ERR_SNYTAX              = 'Syntax error, the rest of expression cannot be evaluated: [%s].'
ERR_MATCH               = 'No matching right parenthesis.'
ERR_STRING_TOKEN        = 'Bad string token: [%s].'
ERR_MACRO_TOKEN         = 'Bad macro token: [%s].'
ERR_EMPTY_TOKEN         = 'Empty token is not allowed.'
ERR_PCD_RESOLVE         = 'PCD token cannot be resolved: [%s].'
ERR_VALID_TOKEN         = 'No more valid token found from rest of string: [%s].'
ERR_EXPR_TYPE           = 'Different types found in expression.'
ERR_OPERATOR_UNSUPPORT  = 'Unsupported operator: [%s]'
ERR_REL_NOT_IN          = 'Expect "IN" after "not" operator.'
WRN_BOOL_EXPR           = 'Operand of boolean type cannot be used in arithmetic expression.'
WRN_EQCMP_STR_OTHERS    = '== Comparison between Operand of string type and Boolean/Number Type always return False.'
WRN_NECMP_STR_OTHERS    = '!= Comparison between Operand of string type and Boolean/Number Type always return True.'
ERR_RELCMP_STR_OTHERS   = 'Operator taking Operand of string type and Boolean/Number Type is not allowed: [%s].'
ERR_STRING_CMP          = 'Unicode string and general string cannot be compared: [%s %s %s]'
ERR_ARRAY_TOKEN         = 'Bad C array or C format GUID token: [%s].'
ERR_ARRAY_ELE           = 'This must be HEX value for NList or Array: [%s].'
ERR_EMPTY_EXPR          = 'Empty expression is not allowed.'

## SplitString
#  Split string to list according double quote
#  For example: abc"de\"f"ghi"jkl"mn will be: ['abc', '"de\"f"', 'ghi', '"jkl"', 'mn']
#
def SplitString(String):
    # There might be escaped quote: "abc\"def\\\"ghi"
    Str = String.replace('\\\\', '//').replace('\\\"', '\\\'')
    RetList = []
    InQuote = False
    Item = ''
    for i, ch in enumerate(Str):
        if ch == '"':
            InQuote = not InQuote
            if not InQuote:
                Item += String[i]
                RetList.append(Item)
                Item = ''
                continue
            if Item:
                RetList.append(Item)
                Item = ''
        Item += String[i]
    if InQuote:
        raise BadExpression(ERR_STRING_TOKEN % Item)
    if Item:
        RetList.append(Item)
    return RetList

## ReplaceExprMacro
#
def ReplaceExprMacro(String, Macros, ExceptionList = None):
    StrList = SplitString(String)
    for i, String in enumerate(StrList):
        InQuote = False
        if String.startswith('"'):
            InQuote = True
        MacroStartPos = String.find('$(')
        if MacroStartPos < 0:
            continue
        RetStr = ''
        while MacroStartPos >= 0:
            RetStr = String[0:MacroStartPos]
            MacroEndPos = String.find(')', MacroStartPos)
            if MacroEndPos < 0:
                raise BadExpression(ERR_MACRO_TOKEN % String[MacroStartPos:])
            Macro = String[MacroStartPos+2:MacroEndPos]
            if Macro not in Macros:
                # From C reference manual:
                # If an undefined macro name appears in the constant-expression of
                # !if or !elif, it is replaced by the integer constant 0.
                RetStr += '0'
            elif not InQuote and ExceptionList and Macro in ExceptionList:
                # Make sure the macro in exception list is encapsulated by double quote
                # For example: DEFINE ARCH = IA32 X64
                # $(ARCH) is replaced with "IA32 X64"
                RetStr += '"' + Macros[Macro] + '"'
            else:
                if Macros[Macro].strip() != "":
                    RetStr += Macros[Macro]
                else:
                    RetStr += '""'
            RetStr += String[MacroEndPos+1:]
            String = RetStr
            MacroStartPos = String.find('$(')
        StrList[i] = RetStr
    return ''.join(StrList)

class ValueExpression(object):
    # Logical operator mapping
    LogicalOperators = {
        '&&' : 'and', '||' : 'or',
        '!'  : 'not', 'AND': 'and',
        'OR' : 'or' , 'NOT': 'not',
        'XOR': '^'  , 'xor': '^',
        'EQ' : '==' , 'NE' : '!=',
        'GT' : '>'  , 'LT' : '<',
        'GE' : '>=' , 'LE' : '<=',
        'IN' : 'in'
    }

    NonLetterOpLst = ['+', '-', '&', '|', '^', '!', '=', '>', '<']

    PcdPattern = re.compile(r'[_a-zA-Z][0-9A-Za-z_]*\.[_a-zA-Z][0-9A-Za-z_]*$')
    HexPattern = re.compile(r'0[xX][0-9a-fA-F]+$')
    RegGuidPattern = re.compile(r'[0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{12}')

    SymbolPattern = re.compile("("
                                 "\$\([A-Z][A-Z0-9_]*\)|\$\(\w+\.\w+\)|\w+\.\w+|"
                                 "&&|\|\||!(?!=)|"
                                 "(?<=\W)AND(?=\W)|(?<=\W)OR(?=\W)|(?<=\W)NOT(?=\W)|(?<=\W)XOR(?=\W)|"
                                 "(?<=\W)EQ(?=\W)|(?<=\W)NE(?=\W)|(?<=\W)GT(?=\W)|(?<=\W)LT(?=\W)|(?<=\W)GE(?=\W)|(?<=\W)LE(?=\W)"
                               ")")

    @staticmethod
    def Eval(Operator, Oprand1, Oprand2 = None):
        WrnExp = None

        if Operator not in ["==", "!=", ">=", "<=", ">", "<", "in", "not in"] and \
            (type(Oprand1) == type('') or type(Oprand2) == type('')):
            raise BadExpression(ERR_STRING_EXPR % Operator)

        TypeDict = {
            type(0)  : 0,
            type(0L) : 0,
            type('') : 1,
            type(True) : 2
        }

        EvalStr = ''
        if Operator in ["!", "NOT", "not"]:
            if type(Oprand1) == type(''):
                raise BadExpression(ERR_STRING_EXPR % Operator)
            EvalStr = 'not Oprand1'
        else:
            if Operator in ["+", "-"] and (type(True) in [type(Oprand1), type(Oprand2)]):
                # Boolean in '+'/'-' will be evaluated but raise warning
                WrnExp = WrnExpression(WRN_BOOL_EXPR)
            elif type('') in [type(Oprand1), type(Oprand2)] and type(Oprand1)!= type(Oprand2):
                # == between string and number/boolean will always return False, != return True
                if Operator == "==":
                    WrnExp = WrnExpression(WRN_EQCMP_STR_OTHERS)
                    WrnExp.result = False
                    raise WrnExp
                elif Operator == "!=":
                    WrnExp = WrnExpression(WRN_NECMP_STR_OTHERS)
                    WrnExp.result = True
                    raise WrnExp
                else:
                    raise BadExpression(ERR_RELCMP_STR_OTHERS % Operator)
            elif TypeDict[type(Oprand1)] != TypeDict[type(Oprand2)]:
                if Operator in ["==", "!=", ">=", "<=", ">", "<"] and set((TypeDict[type(Oprand1)], TypeDict[type(Oprand2)])) == set((TypeDict[type(True)], TypeDict[type(0)])):
                    # comparison between number and boolean is allowed
                    pass
                elif Operator in ['&', '|', '^', "and", "or"] and set((TypeDict[type(Oprand1)], TypeDict[type(Oprand2)])) == set((TypeDict[type(True)], TypeDict[type(0)])):
                    # bitwise and logical operation between number and boolean is allowed
                    pass
                else:
                    raise BadExpression(ERR_EXPR_TYPE)
            if type(Oprand1) == type('') and type(Oprand2) == type(''):
                if (Oprand1.startswith('L"') and not Oprand2.startswith('L"')) or \
                    (not Oprand1.startswith('L"') and Oprand2.startswith('L"')):
                    raise BadExpression(ERR_STRING_CMP % (Oprand1, Operator, Oprand2))
            if 'in' in Operator and type(Oprand2) == type(''):
                Oprand2 = Oprand2.split()
            EvalStr = 'Oprand1 ' + Operator + ' Oprand2'

        # Local symbols used by built in eval function
        Dict = {
            'Oprand1' : Oprand1,
            'Oprand2' : Oprand2
        }
        try:
            Val = eval(EvalStr, {}, Dict)
        except Exception, Excpt:
            raise BadExpression(str(Excpt))

        if Operator in ['and', 'or']:
            if Val:
                Val = True
            else:
                Val = False

        if WrnExp:
            WrnExp.result = Val
            raise WrnExp
        return Val

    def __init__(self, Expression, SymbolTable={}):
        self._NoProcess = False
        if type(Expression) != type(''):
            self._Expr = Expression
            self._NoProcess = True
            return

        self._Expr = ReplaceExprMacro(Expression.strip(),
                                  SymbolTable,
                                  ['TARGET', 'TOOL_CHAIN_TAG', 'ARCH'])

        if not self._Expr.strip():
            raise BadExpression(ERR_EMPTY_EXPR)

        #
        # The symbol table including PCD and macro mapping
        #
        self._Symb = SymbolTable
        self._Symb.update(self.LogicalOperators)
        self._Idx = 0
        self._Len = len(self._Expr)
        self._Token = ''
        self._WarnExcept = None

        # Literal token without any conversion
        self._LiteralToken = ''

    # Public entry for this class
    #   @param RealValue: False: only evaluate if the expression is true or false, used for conditional expression
    #                     True : return the evaluated str(value), used for PCD value
    #
    #   @return: True or False if RealValue is False
    #            Evaluated value of string format if RealValue is True
    #
    def __call__(self, RealValue=False):
        if self._NoProcess:
            return self._Expr

        self._Expr = self._Expr.strip()
        if RealValue:
            self._Token = self._Expr
            if self.__IsNumberToken():
                return self._Expr

            Token = self._GetToken()
            if type(Token) == type('') and Token.startswith('{') and Token.endswith('}') and self._Idx >= self._Len:
                return self._Expr

            self._Idx = 0
            self._Token = ''

        Val = self._OrExpr()
        RealVal = Val
        if type(Val) == type(''):
            if Val == 'L""':
                Val = False
            elif not Val:
                Val = False
                RealVal = '""'
            elif not Val.startswith('L"') and not Val.startswith('{'):
                Val = True
                RealVal = '"' + RealVal + '"'

        # The expression has been parsed, but the end of expression is not reached
        # It means the rest does not comply EBNF of <Expression>
        if self._Idx != self._Len:
            raise BadExpression(ERR_SNYTAX % self._Expr[self._Idx:])

        if RealValue:
            RetVal = str(RealVal)
        elif Val:
            RetVal = True
        else:
            RetVal = False

        if self._WarnExcept:
            self._WarnExcept.result = RetVal
            raise self._WarnExcept
        else:
            return RetVal

    # Template function to parse binary operators which have same precedence
    # Expr [Operator Expr]*
    def _ExprFuncTemplate(self, EvalFunc, OpLst):
        Val = EvalFunc()
        while self._IsOperator(OpLst):
            Op = self._Token
            try:
                Val = self.Eval(Op, Val, EvalFunc())
            except WrnExpression, Warn:
                self._WarnExcept = Warn
                Val = Warn.result
        return Val

    # A [|| B]*
    def _OrExpr(self):
        return self._ExprFuncTemplate(self._AndExpr, ["OR", "or", "||"])

    # A [&& B]*
    def _AndExpr(self):
        return self._ExprFuncTemplate(self._BitOr, ["AND", "and", "&&"])

    # A [ | B]*
    def _BitOr(self):
        return self._ExprFuncTemplate(self._BitXor, ["|"])

    # A [ ^ B]*
    def _BitXor(self):
        return self._ExprFuncTemplate(self._BitAnd, ["XOR", "xor", "^"])

    # A [ & B]*
    def _BitAnd(self):
        return self._ExprFuncTemplate(self._EqExpr, ["&"])

    # A [ == B]*
    def _EqExpr(self):
        Val = self._RelExpr()
        while self._IsOperator(["==", "!=", "EQ", "NE", "IN", "in", "!", "NOT", "not"]):
            Op = self._Token
            if Op in ["!", "NOT", "not"]:
                if not self._IsOperator(["IN", "in"]):
                    raise BadExpression(ERR_REL_NOT_IN)
                Op += ' ' + self._Token
            try:
                Val = self.Eval(Op, Val, self._RelExpr())
            except WrnExpression, Warn:
                self._WarnExcept = Warn
                Val = Warn.result
        return Val

    # A [ > B]*
    def _RelExpr(self):
        return self._ExprFuncTemplate(self._AddExpr, ["<=", ">=", "<", ">", "LE", "GE", "LT", "GT"])

    # A [ + B]*
    def _AddExpr(self):
        return self._ExprFuncTemplate(self._UnaryExpr, ["+", "-"])

    # [!]*A
    def _UnaryExpr(self):
        if self._IsOperator(["!", "NOT", "not"]):
            Val = self._UnaryExpr()
            try:
                return self.Eval('not', Val)
            except WrnExpression, Warn:
                self._WarnExcept = Warn
                return Warn.result
        return self._IdenExpr()

    # Parse identifier or encapsulated expression
    def _IdenExpr(self):
        Tk = self._GetToken()
        if Tk == '(':
            Val = self._OrExpr()
            try:
                # _GetToken may also raise BadExpression
                if self._GetToken() != ')':
                    raise BadExpression(ERR_MATCH)
            except BadExpression:
                raise BadExpression(ERR_MATCH)
            return Val
        return Tk

    # Skip whitespace or tab
    def __SkipWS(self):
        for Char in self._Expr[self._Idx:]:
            if Char not in ' \t':
                break
            self._Idx += 1

    # Try to convert string to number
    def __IsNumberToken(self):
        Radix = 10
        if self._Token.lower()[0:2] == '0x' and len(self._Token) > 2:
            Radix = 16
        try:
            self._Token = int(self._Token, Radix)
            return True
        except ValueError:
            return False
        except TypeError:
            return False

    # Parse array: {...}
    def __GetArray(self):
        Token = '{'
        self._Idx += 1
        self.__GetNList(True)
        Token += self._LiteralToken
        if self._Idx >= self._Len or self._Expr[self._Idx] != '}':
            raise BadExpression(ERR_ARRAY_TOKEN % Token)
        Token += '}'

        # All whitespace and tabs in array are already stripped.
        IsArray = IsGuid = False
        if len(Token.split(',')) == 11 and len(Token.split(',{')) == 2 \
            and len(Token.split('},')) == 1:
            HexLen = [11,6,6,5,4,4,4,4,4,4,6]
            HexList= Token.split(',')
            if HexList[3].startswith('{') and \
                not [Index for Index, Hex in enumerate(HexList) if len(Hex) > HexLen[Index]]:
                IsGuid = True
        if Token.lstrip('{').rstrip('}').find('{') == -1:
            if not [Hex for Hex in Token.lstrip('{').rstrip('}').split(',') if len(Hex) > 4]:
                IsArray = True
        if not IsArray and not IsGuid:
            raise BadExpression(ERR_ARRAY_TOKEN % Token)
        self._Idx += 1
        self._Token = self._LiteralToken = Token
        return self._Token

    # Parse string, the format must be: "..."
    def __GetString(self):
        Idx = self._Idx

        # Skip left quote
        self._Idx += 1

        # Replace escape \\\", \"
        Expr = self._Expr[self._Idx:].replace('\\\\', '//').replace('\\\"', '\\\'')
        for Ch in Expr:
            self._Idx += 1
            if Ch == '"':
                break
        self._Token = self._LiteralToken = self._Expr[Idx:self._Idx]
        if not self._Token.endswith('"'):
            raise BadExpression(ERR_STRING_TOKEN % self._Token)
        self._Token = self._Token[1:-1]
        return self._Token

    # Get token that is comprised by alphanumeric, underscore or dot(used by PCD)
    # @param IsAlphaOp: Indicate if parsing general token or script operator(EQ, NE...)
    def __GetIdToken(self, IsAlphaOp = False):
        IdToken = ''
        for Ch in self._Expr[self._Idx:]:
            if not self.__IsIdChar(Ch):
                break
            self._Idx += 1
            IdToken += Ch

        self._Token = self._LiteralToken = IdToken
        if not IsAlphaOp:
            self.__ResolveToken()
        return self._Token

    # Try to resolve token
    def __ResolveToken(self):
        if not self._Token:
            raise BadExpression(ERR_EMPTY_TOKEN)

        # PCD token
        if self.PcdPattern.match(self._Token):
            if self._Token not in self._Symb:
                raise BadExpression(ERR_PCD_RESOLVE % self._Token)
            self._Token = ValueExpression(self._Symb[self._Token], self._Symb)(True)
            if type(self._Token) != type(''):
                self._LiteralToken = hex(self._Token)
                return

        if self._Token.startswith('"'):
            self._Token = self._Token[1:-1]
        elif self._Token in ["FALSE", "false", "False"]:
            self._Token = False
        elif self._Token in ["TRUE", "true", "True"]:
            self._Token = True
        else:
            self.__IsNumberToken()

    def __GetNList(self, InArray=False):
        self._GetSingleToken()
        if not self.__IsHexLiteral():
            if InArray:
                raise BadExpression(ERR_ARRAY_ELE % self._Token)
            return self._Token

        self.__SkipWS()
        Expr = self._Expr[self._Idx:]
        if not Expr.startswith(','):
            return self._Token

        NList = self._LiteralToken
        while Expr.startswith(','):
            NList += ','
            self._Idx += 1
            self.__SkipWS()
            self._GetSingleToken()
            if not self.__IsHexLiteral():
                raise BadExpression(ERR_ARRAY_ELE % self._Token)
            NList += self._LiteralToken
            self.__SkipWS()
            Expr = self._Expr[self._Idx:]
        self._Token = self._LiteralToken = NList
        return self._Token

    def __IsHexLiteral(self):
        if self._LiteralToken.startswith('{') and \
            self._LiteralToken.endswith('}'):
            return True

        if self.HexPattern.match(self._LiteralToken):
            Token = self._LiteralToken[2:]
            Token = Token.lstrip('0')
            if not Token:
                self._LiteralToken = '0x0'
            else:
                self._LiteralToken = '0x' + Token.lower()
            return True
        return False

    def _GetToken(self):
        return self.__GetNList()

    @staticmethod
    def __IsIdChar(Ch):
        return Ch in '._/:' or Ch.isalnum()

    # Parse operand
    def _GetSingleToken(self):
        self.__SkipWS()
        Expr = self._Expr[self._Idx:]
        if Expr.startswith('L"'):
            # Skip L
            self._Idx += 1
            UStr = self.__GetString()
            self._Token = 'L"' + UStr + '"'
            return self._Token

        self._Token = ''
        if Expr:
            Ch = Expr[0]
            Match = self.RegGuidPattern.match(Expr)
            if Match and not Expr[Match.end():Match.end()+1].isalnum() \
                and Expr[Match.end():Match.end()+1] != '_':
                self._Idx += Match.end()
                self._Token = ValueExpression(GuidStringToGuidStructureString(Expr[0:Match.end()]))(True)
                return self._Token
            elif self.__IsIdChar(Ch):
                return self.__GetIdToken()
            elif Ch == '"':
                return self.__GetString()
            elif Ch == '{':
                return self.__GetArray()
            elif Ch == '(' or Ch == ')':
                self._Idx += 1
                self._Token = Ch
                return self._Token

        raise BadExpression(ERR_VALID_TOKEN % Expr)

    # Parse operator
    def _GetOperator(self):
        self.__SkipWS()
        LegalOpLst = ['&&', '||', '!=', '==', '>=', '<='] + self.NonLetterOpLst

        self._Token = ''
        Expr = self._Expr[self._Idx:]

        # Reach end of expression
        if not Expr:
            return ''

        # Script operator: LT, GT, LE, GE, EQ, NE, and, or, xor, not
        if Expr[0].isalpha():
            return self.__GetIdToken(True)

        # Start to get regular operator: +, -, <, > ...
        if Expr[0] not in self.NonLetterOpLst:
            return ''

        OpToken = ''
        for Ch in Expr:
            if Ch in self.NonLetterOpLst:
                if '!' == Ch and OpToken:
                    break
                self._Idx += 1
                OpToken += Ch
            else:
                break

        if OpToken not in LegalOpLst:
            raise BadExpression(ERR_OPERATOR_UNSUPPORT % OpToken)
        self._Token = OpToken
        return OpToken

    # Check if current token matches the operators given from OpList
    def _IsOperator(self, OpList):
        Idx = self._Idx
        self._GetOperator()
        if self._Token in OpList:
            if self._Token in self.LogicalOperators:
                self._Token = self.LogicalOperators[self._Token]
            return True
        self._Idx = Idx
        return False

if __name__ == '__main__':
    pass
    while True:
        input = raw_input('Input expr: ')
        if input in 'qQ':
            break
        try:
            print ValueExpression(input)(True)
            print ValueExpression(input)(False)
        except WrnExpression, Ex:
            print Ex.result
            print str(Ex)
        except Exception, Ex:
            print str(Ex)