// An macro token was encountered and it expects a parameter list.
// The routine has to check if there is a following parenthesis
// and eventually skip it but it MUST NOT parse the next token
// if it is not a parenthesis. This is because the macro token
// may have a replacement and is that one that has to be returned
// back to the caller from cxxParserParseNextToken().
static bool cxxParserParseNextTokenSkipMacroParenthesis(CXXToken ** ppChain)
{
CXX_DEBUG_ENTER();
CXX_DEBUG_ASSERT(ppChain,"ppChain should not be null here");
cxxParserSkipToNonWhiteSpace();
if(g_cxx.iChar != '(')
{
*ppChain = NULL;
return true; // no parenthesis
}
if(!cxxParserParseNextToken())
{
CXX_DEBUG_LEAVE_TEXT("No next token after ignored identifier");
return false;
}
if(!cxxTokenTypeIs(g_cxx.pToken,CXXTokenTypeOpeningParenthesis))
{
CXX_DEBUG_ASSERT(false,"Should have found an open parenthesis token here!");
CXX_DEBUG_LEAVE_TEXT("Internal error");
return false;
}
if(!cxxParserParseAndCondenseCurrentSubchain(
CXXTokenTypeOpeningParenthesis |
CXXTokenTypeOpeningSquareParenthesis |
CXXTokenTypeOpeningBracket,
false
))
{
CXX_DEBUG_LEAVE_TEXT("Failed to parse and condense subchains");
return false;
}
CXX_DEBUG_ASSERT(
cxxTokenTypeIs(g_cxx.pToken,CXXTokenTypeParenthesisChain),
"Should have a parenthesis chain as last token!"
);
// Now just kill the chain.
*ppChain = cxxTokenChainTakeLast(g_cxx.pTokenChain);
CXX_DEBUG_LEAVE();
return true;
}
bool cxxParserParseNextToken(void)
{
CXXToken * t = cxxTokenCreate();
// The token chain should not be allowed to grow arbitrairly large.
// The token structures are quite big and it's easy to grow up to
// 5-6GB or memory usage. However this limit should be large enough
// to accomodate all the reasonable statements that could have some
// information in them. This includes multiple function prototypes
// in a single statement (ImageMagick has some examples) but probably
// does NOT include large data tables.
if(g_cxx.pTokenChain->iCount > 16384)
cxxTokenChainDestroyLast(g_cxx.pTokenChain);
cxxTokenChainAppend(g_cxx.pTokenChain,t);
g_cxx.pToken = t;
cxxParserSkipToNonWhiteSpace();
// FIXME: this cpp handling is kind of broken:
// it works only because the moon is in the correct phase.
cppBeginStatement();
// This must be done after getting char from input
t->iLineNumber = getInputLineNumber();
t->oFilePosition = getInputFilePosition();
if(g_cxx.iChar == EOF)
{
t->eType = CXXTokenTypeEOF;
t->bFollowedBySpace = false;
return false;
}
unsigned int uInfo = UINFO(g_cxx.iChar);
//fprintf(stderr,"Char %c %02x info %u\n",g_cxx.iChar,g_cxx.iChar,uInfo);
if(uInfo & CXXCharTypeStartOfIdentifier)
{
// word
t->eType = CXXTokenTypeIdentifier;
t->bFollowedBySpace = false;
vStringPut(t->pszWord,g_cxx.iChar);
// special case for tile, which may actually be an operator
if(g_cxx.iChar == '~')
{
// may be followed by space!
g_cxx.iChar = cppGetc();
if(isspace(g_cxx.iChar))
{
t->bFollowedBySpace = true;
g_cxx.iChar = cppGetc();
while(isspace(g_cxx.iChar))
g_cxx.iChar = cppGetc();
}
// non space
uInfo = UINFO(g_cxx.iChar);
if(!(uInfo & CXXCharTypeStartOfIdentifier))
{
// this is not an identifier after all
t->eType = CXXTokenTypeOperator;
if((!t->bFollowedBySpace) && g_cxx.iChar == '=')
{
// make ~= single token so it's not handled as
// a separate assignment
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
t->bFollowedBySpace = isspace(g_cxx.iChar);
}
return true;
}
} else {
g_cxx.iChar = cppGetc();
}
for(;;)
{
uInfo = UINFO(g_cxx.iChar);
if(!(uInfo & CXXCharTypePartOfIdentifier))
break;
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
}
int iCXXKeyword = lookupKeyword(t->pszWord->buffer,g_cxx.eLanguage);
if(iCXXKeyword >= 0)
{
if(
(
(iCXXKeyword == CXXKeywordFINAL) &&
(!g_cxx.bParsingClassStructOrUnionDeclaration)
) || (
(
(iCXXKeyword == CXXKeywordPUBLIC) ||
(iCXXKeyword == CXXKeywordPROTECTED) ||
(iCXXKeyword == CXXKeywordPRIVATE)
) &&
(!g_cxx.bEnablePublicProtectedPrivateKeywords)
)
)
{
t->eType = CXXTokenTypeIdentifier;
} else {
t->eType = CXXTokenTypeKeyword;
t->eKeyword = (enum CXXKeyword)iCXXKeyword;
if(iCXXKeyword == CXXKeyword__ATTRIBUTE__)
{
// special handling for __attribute__
return cxxParserParseNextTokenCondenseAttribute();
}
}
} else {
const cppMacroInfo * pMacro = cppFindMacro(vStringValue(t->pszWord));
if(pMacro)
{
CXX_DEBUG_PRINT("Macro %s",vStringValue(t->pszWord));
cxxTokenChainDestroyLast(g_cxx.pTokenChain);
CXXToken * pParameterChain = NULL;
if(pMacro->hasParameterList)
{
CXX_DEBUG_PRINT("Macro has parameter list");
if(!cxxParserParseNextTokenSkipMacroParenthesis(&pParameterChain))
return false;
}
// This is used to avoid infinite recursion in substitution
// (things like -D foo=foo or similar)
static int iReplacementRecursionCount = 0;
if(pMacro->replacements)
{
CXX_DEBUG_PRINT("The token has replacements: applying");
if(iReplacementRecursionCount < 1024)
{
// unget last char
cppUngetc(g_cxx.iChar);
// unget the replacement
cxxParserParseNextTokenApplyReplacement(
pMacro,
pParameterChain
);
g_cxx.iChar = cppGetc();
}
}
if(pParameterChain)
cxxTokenDestroy(pParameterChain);
iReplacementRecursionCount++;
// Have no token to return: parse it
CXX_DEBUG_PRINT("Parse inner token");
bool bRet = cxxParserParseNextToken();
CXX_DEBUG_PRINT("Parsed inner token: %s type %d",g_cxx.pToken->pszWord->buffer,g_cxx.pToken->eType);
iReplacementRecursionCount--;
return bRet;
}
}
t->bFollowedBySpace = isspace(g_cxx.iChar);
return true;
}
if(g_cxx.iChar == '-')
{
// special case for pointer
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
if(g_cxx.iChar == '>')
{
t->eType = CXXTokenTypePointerOperator;
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
} else {
t->eType = CXXTokenTypeOperator;
if(g_cxx.iChar == '-')
{
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
}
}
t->bFollowedBySpace = isspace(g_cxx.iChar);
return true;
}
#if 0
// As long as we use cppGetc() we don't need this
if(g_cxx.iChar == '"')
{
// special case for strings
t->eType = CXXTokenTypeStringConstant;
vStringPut(t->pszWord,g_cxx.iChar);
// We don't even care of storing the other chars: we don't need
// them for parsing
// FIXME: We might need them in signature:() tag.. maybe add
// them up to a certain length only?
for(;;)
{
g_cxx.iChar = cppGetc();
if(g_cxx.iChar == EOF)
{
t->bFollowedBySpace = false;
return true;
}
if(g_cxx.iChar == '\\')
{
// escape
g_cxx.iChar = cppGetc();
if(g_cxx.iChar == EOF)
{
t->bFollowedBySpace = false;
return true;
}
} else if(g_cxx.iChar == '"')
{
g_cxx.iChar = cppGetc();
break;
}
}
t->bFollowedBySpace = isspace(g_cxx.iChar);
return true;
}
#else
if(g_cxx.iChar == STRING_SYMBOL)
{
t->eType = CXXTokenTypeStringConstant;
vStringPut(t->pszWord,'"');
vStringPut(t->pszWord,'"');
g_cxx.iChar = cppGetc();
t->bFollowedBySpace = isspace(g_cxx.iChar);
return true;
}
#endif
#if 0
// As long as we use cppGetc() we don't need this
if(g_cxx.iChar == '\'')
{
// special case for strings
t->eType = CXXTokenTypeCharacterConstant;
vStringPut(t->pszWord,g_cxx.iChar);
// We don't even care storing the other chars: we don't
// need them for parsing
for(;;)
{
g_cxx.iChar = cppGetc();
if(g_cxx.iChar == EOF)
{
t->bFollowedBySpace = false;
return true;
}
if(g_cxx.iChar == '\\')
{
// escape
g_cxx.iChar = cppGetc();
if(g_cxx.iChar == EOF)
{
t->bFollowedBySpace = false;
return true;
}
} else if(g_cxx.iChar == '\'')
{
g_cxx.iChar = cppGetc();
break;
}
}
t->bFollowedBySpace = isspace(g_cxx.iChar);
return true;
}
#else
if(g_cxx.iChar == CHAR_SYMBOL)
{
t->eType = CXXTokenTypeCharacterConstant;
vStringPut(t->pszWord,'\'');
vStringPut(t->pszWord,'\'');
g_cxx.iChar = cppGetc();
t->bFollowedBySpace = isspace(g_cxx.iChar);
return true;
}
#endif
if(uInfo & CXXCharTypeDecimalDigit)
{
// number
t->eType = CXXTokenTypeNumber;
vStringPut(t->pszWord,g_cxx.iChar);
for(;;)
{
g_cxx.iChar = cppGetc();
uInfo = UINFO(g_cxx.iChar);
if(!(uInfo & CXXCharTypeValidInNumber))
break;
vStringPut(t->pszWord,g_cxx.iChar);
}
t->bFollowedBySpace = isspace(g_cxx.iChar);
return true;
}
if(uInfo & CXXCharTypeNamedSingleOrRepeatedCharToken)
{
t->eType = g_aCharTable[g_cxx.iChar].uSingleTokenType;
vStringPut(t->pszWord,g_cxx.iChar);
int iChar = g_cxx.iChar;
g_cxx.iChar = cppGetc();
if(g_cxx.iChar == iChar)
{
t->eType = g_aCharTable[g_cxx.iChar].uMultiTokenType;
// We could signal a syntax error with more than two colons
// or equal signs...but we're tolerant
do {
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
} while(g_cxx.iChar == iChar);
}
t->bFollowedBySpace = isspace(g_cxx.iChar);
return true;
}
if(uInfo & CXXCharTypeNamedSingleOrOperatorToken)
{
t->eType = g_aCharTable[g_cxx.iChar].uSingleTokenType;
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
uInfo = UINFO(g_cxx.iChar);
if(uInfo & (CXXCharTypeOperator | CXXCharTypeNamedSingleOrOperatorToken))
{
t->eType = CXXTokenTypeOperator;
do {
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
uInfo = UINFO(g_cxx.iChar);
} while(
uInfo &
(CXXCharTypeOperator | CXXCharTypeNamedSingleOrOperatorToken)
);
}
t->bFollowedBySpace = isspace(g_cxx.iChar);
return true;
}
if(uInfo & CXXCharTypeNamedSingleCharToken)
{
t->eType = g_aCharTable[g_cxx.iChar].uSingleTokenType;
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
t->bFollowedBySpace = isspace(g_cxx.iChar);
return true;
}
if(uInfo & CXXCharTypeOperator)
{
t->eType = CXXTokenTypeOperator;
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
uInfo = UINFO(g_cxx.iChar);
while(uInfo & CXXCharTypeOperator)
{
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
uInfo = UINFO(g_cxx.iChar);
}
t->bFollowedBySpace = isspace(g_cxx.iChar);
return true;
}
t->eType = CXXTokenTypeUnknown;
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
t->bFollowedBySpace = isspace(g_cxx.iChar);
return true;
}
boolean cxxParserParseNextToken(void)
{
CXXToken * t = cxxTokenCreate();
// The token chain should not be allowed to grow arbitrairly large.
// The token structures are quite big and it's easy to grow up to
// 5-6GB or memory usage. However this limit should be large enough
// to accomodate all the reasonable statements that could have some
// information in them. This includes multiple function prototypes
// in a single statement (ImageMagick has some examples) but probably
// does NOT include large data tables.
if(g_cxx.pTokenChain->iCount > 16384)
cxxTokenChainDestroyLast(g_cxx.pTokenChain);
cxxTokenChainAppend(g_cxx.pTokenChain,t);
g_cxx.pToken = t;
cxxParserSkipToNonWhiteSpace();
// FIXME: this cpp handling is kind of broken:
// it works only because the moon is in the correct phase.
cppBeginStatement();
// This must be done after getting char from input
t->iLineNumber = getInputLineNumber();
t->oFilePosition = getInputFilePosition();
if(g_cxx.iChar == EOF)
{
t->eType = CXXTokenTypeEOF;
t->bFollowedBySpace = FALSE;
return FALSE;
}
unsigned int uInfo = (g_cxx.iChar < 0x80) ? g_aCharTable[g_cxx.iChar].uType : 0;
//printf("Char %c %02x info %u\n",g_cxx.iChar,g_cxx.iChar,uInfo);
if(uInfo & CXXCharTypeStartOfIdentifier)
{
// word
t->eType = CXXTokenTypeIdentifier;
t->bFollowedBySpace = FALSE;
vStringPut(t->pszWord,g_cxx.iChar);
// special case for tile, which may actually be an operator
if(g_cxx.iChar == '~')
{
// may be followed by space!
g_cxx.iChar = cppGetc();
if(isspace(g_cxx.iChar))
{
t->bFollowedBySpace = TRUE;
g_cxx.iChar = cppGetc();
while(isspace(g_cxx.iChar))
g_cxx.iChar = cppGetc();
}
// non space
uInfo = (g_cxx.iChar < 0x80) ? g_aCharTable[g_cxx.iChar].uType : 0;
if(!(uInfo & CXXCharTypeStartOfIdentifier))
{
// this is not an identifier after all
t->eType = CXXTokenTypeOperator;
if((!t->bFollowedBySpace) && g_cxx.iChar == '=')
{
// make ~= single token so it's not handled as
// a separate assignment
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
t->bFollowedBySpace = isspace(g_cxx.iChar);
}
return TRUE;
}
} else {
g_cxx.iChar = cppGetc();
}
for(;;)
{
uInfo = (g_cxx.iChar < 0x80) ? g_aCharTable[g_cxx.iChar].uType : 0;
if(!(uInfo & CXXCharTypePartOfIdentifier))
break;
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
}
int iCXXKeyword = lookupKeyword(t->pszWord->buffer,g_cxx.eLanguage);
if(iCXXKeyword >= 0)
{
if(
(
(iCXXKeyword == CXXKeywordFINAL) &&
(!g_cxx.bParsingClassStructOrUnionDeclaration)
) || (
(
(iCXXKeyword == CXXKeywordPUBLIC) ||
(iCXXKeyword == CXXKeywordPROTECTED) ||
(iCXXKeyword == CXXKeywordPRIVATE)
) &&
(!g_cxx.bEnablePublicProtectedPrivateKeywords)
)
)
{
t->eType = CXXTokenTypeIdentifier;
} else {
t->eType = CXXTokenTypeKeyword;
t->eKeyword = (enum CXXKeyword)iCXXKeyword;
if(iCXXKeyword == CXXKeyword__ATTRIBUTE__)
{
// special handling for __attribute__
return cxxParserParseNextTokenCondenseAttribute();
}
}
} else {
boolean bIgnoreParens = FALSE;
const char * szReplacement = NULL;
if(isIgnoreToken(
vStringValue(t->pszWord),
&bIgnoreParens,
&szReplacement
))
{
CXX_DEBUG_PRINT("Ignore token %s",vStringValue(t->pszWord));
// FIXME: Handle ignore parens!
if(szReplacement && *szReplacement)
{
vStringClear(t->pszWord);
vStringCatS(t->pszWord,szReplacement);
} else {
// skip
cxxTokenChainDestroyLast(g_cxx.pTokenChain);
return cxxParserParseNextToken();
}
}
}
t->bFollowedBySpace = t->bFollowedBySpace | isspace(g_cxx.iChar);
return TRUE;
}
if(g_cxx.iChar == '-')
{
// special case for pointer
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
if(g_cxx.iChar == '>')
{
t->eType = CXXTokenTypePointerOperator;
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
} else {
t->eType = CXXTokenTypeOperator;
if(g_cxx.iChar == '-')
{
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
}
}
t->bFollowedBySpace = isspace(g_cxx.iChar);
return TRUE;
}
#if 0
// As long as we use cppGetc() we don't need this
if(g_cxx.iChar == '"')
{
// special case for strings
t->eType = CXXTokenTypeStringConstant;
vStringPut(t->pszWord,g_cxx.iChar);
// We don't even care of storing the other chars: we don't need
// them for parsing
// FIXME: We might need them in signature:() tag.. maybe add
// them up to a certain length only?
for(;;)
{
g_cxx.iChar = cppGetc();
if(g_cxx.iChar == EOF)
{
t->bFollowedBySpace = FALSE;
return TRUE;
}
if(g_cxx.iChar == '\\')
{
// escape
g_cxx.iChar = cppGetc();
if(g_cxx.iChar == EOF)
{
t->bFollowedBySpace = FALSE;
return TRUE;
}
} else if(g_cxx.iChar == '"')
{
g_cxx.iChar = cppGetc();
break;
}
}
t->bFollowedBySpace = isspace(g_cxx.iChar);
return TRUE;
}
#else
if(g_cxx.iChar == STRING_SYMBOL)
{
t->eType = CXXTokenTypeStringConstant;
vStringPut(t->pszWord,'"');
vStringPut(t->pszWord,'"');
g_cxx.iChar = cppGetc();
t->bFollowedBySpace = isspace(g_cxx.iChar);
return TRUE;
}
#endif
#if 0
// As long as we use cppGetc() we don't need this
if(g_cxx.iChar == '\'')
{
// special case for strings
t->eType = CXXTokenTypeCharacterConstant;
vStringPut(t->pszWord,g_cxx.iChar);
// We don't even care storing the other chars: we don't
// need them for parsing
for(;;)
{
g_cxx.iChar = cppGetc();
if(g_cxx.iChar == EOF)
{
t->bFollowedBySpace = FALSE;
return TRUE;
}
if(g_cxx.iChar == '\\')
{
// escape
g_cxx.iChar = cppGetc();
if(g_cxx.iChar == EOF)
{
t->bFollowedBySpace = FALSE;
return TRUE;
}
} else if(g_cxx.iChar == '\'')
{
g_cxx.iChar = cppGetc();
break;
}
}
t->bFollowedBySpace = isspace(g_cxx.iChar);
return TRUE;
}
#else
if(g_cxx.iChar == CHAR_SYMBOL)
{
t->eType = CXXTokenTypeCharacterConstant;
vStringPut(t->pszWord,'\'');
vStringPut(t->pszWord,'\'');
g_cxx.iChar = cppGetc();
t->bFollowedBySpace = isspace(g_cxx.iChar);
return TRUE;
}
#endif
if(uInfo & CXXCharTypeDecimalDigit)
{
// number
t->eType = CXXTokenTypeNumber;
vStringPut(t->pszWord,g_cxx.iChar);
for(;;)
{
g_cxx.iChar = cppGetc();
uInfo = (g_cxx.iChar < 0x80) ? g_aCharTable[g_cxx.iChar].uType : 0;
if(!(uInfo & CXXCharTypeValidInNumber))
break;
vStringPut(t->pszWord,g_cxx.iChar);
}
t->bFollowedBySpace = isspace(g_cxx.iChar);
return TRUE;
}
if(uInfo & CXXCharTypeNamedSingleOrRepeatedCharToken)
{
t->eType = g_aCharTable[g_cxx.iChar].uSingleTokenType;
vStringPut(t->pszWord,g_cxx.iChar);
int iChar = g_cxx.iChar;
g_cxx.iChar = cppGetc();
if(g_cxx.iChar == iChar)
{
t->eType = g_aCharTable[g_cxx.iChar].uMultiTokenType;
// We could signal a syntax error with more than two colons
// or equal signs...but we're tolerant
do {
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
} while(g_cxx.iChar == iChar);
}
t->bFollowedBySpace = isspace(g_cxx.iChar);
return TRUE;
}
if(uInfo & CXXCharTypeNamedSingleOrOperatorToken)
{
t->eType = g_aCharTable[g_cxx.iChar].uSingleTokenType;
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
uInfo = (g_cxx.iChar < 0x80) ? g_aCharTable[g_cxx.iChar].uType : 0;
if(uInfo & (CXXCharTypeOperator | CXXCharTypeNamedSingleOrOperatorToken))
{
t->eType = CXXTokenTypeOperator;
do {
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
uInfo = (g_cxx.iChar < 0x80) ? g_aCharTable[g_cxx.iChar].uType : 0;
} while(
uInfo &
(CXXCharTypeOperator | CXXCharTypeNamedSingleOrOperatorToken)
);
}
t->bFollowedBySpace = isspace(g_cxx.iChar);
return TRUE;
}
if(uInfo & CXXCharTypeNamedSingleCharToken)
{
t->eType = g_aCharTable[g_cxx.iChar].uSingleTokenType;
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
t->bFollowedBySpace = isspace(g_cxx.iChar);
return TRUE;
}
if(uInfo & CXXCharTypeOperator)
{
t->eType = CXXTokenTypeOperator;
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
uInfo = (g_cxx.iChar < 0x80) ? g_aCharTable[g_cxx.iChar].uType : 0;
while(uInfo & CXXCharTypeOperator)
{
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
uInfo = (g_cxx.iChar < 0x80) ? g_aCharTable[g_cxx.iChar].uType : 0;
}
t->bFollowedBySpace = isspace(g_cxx.iChar);
return TRUE;
}
t->eType = CXXTokenTypeUnknown;
vStringPut(t->pszWord,g_cxx.iChar);
g_cxx.iChar = cppGetc();
t->bFollowedBySpace = isspace(g_cxx.iChar);
return TRUE;
}
