Improved AST:

- Fixed TODO for ternary operator. Add parantheses between ? and : where necessary as a preparation for createAst()
- Improved AST validation:
-- Better message for binary operator with only one operand
-- Ensure ? has : as second operand (#7035)
This commit is contained in:
PKEuS 2015-10-12 18:14:56 +02:00
parent 97326fce13
commit 05b5275110
6 changed files with 106 additions and 52 deletions

View File

@ -1713,6 +1713,8 @@ bool Tokenizer::tokenize(std::istream &code,
if (simplifyTokenList1(FileName)) {
if (!noSymbolDB_AST) {
prepareTernaryOpForAST();
createSymbolDatabase();
// Use symbol database to identify rvalue references. Split && to & &. This is safe, since it doesn't delete any tokens (which might be referenced by symbol database)
@ -1729,10 +1731,13 @@ bool Tokenizer::tokenize(std::istream &code,
// Verify that ast looks ok
for (const Token *tok = list.front(); tok; tok = tok->next()) {
// internal error / syntax error if binary operator only has 1 operand
if ((tok->isAssignmentOp() || tok->isComparisonOp() || Token::Match(tok,"[|^/%]")) && tok->astOperand1() && !tok->astOperand2()) {
throw InternalError(tok, "ast", InternalError::INTERNAL);
}
// Syntax error if binary operator only has 1 operand
if ((tok->isAssignmentOp() || tok->isComparisonOp() || Token::Match(tok,"[|^/%]")) && tok->astOperand1() && !tok->astOperand2())
throw InternalError(tok, "Syntax Error: AST broken, binary operator has only one operand.", InternalError::SYNTAX);
// Syntax error if we encounter "?" with operand2 that is not ":"
if (tok->astOperand2() && tok->str() == "?" && tok->astOperand2()->str() != ":")
throw InternalError(tok, "Syntax Error: AST broken, ternary operator lacks ':'.", InternalError::SYNTAX);
}
SymbolDatabase::setValueTypeInTokenList(list.front());
@ -10140,6 +10145,41 @@ bool Tokenizer::simplifyStrlen()
return modified;
}
void Tokenizer::prepareTernaryOpForAST()
{
// http://en.cppreference.com/w/cpp/language/operator_precedence says about ternary operator:
// "The expression in the middle of the conditional operator (between ? and :) is parsed as if parenthesized: its precedence relative to ?: is ignored."
// The AST parser relies on this function to add such parantheses where necessary.
for (Token* tok = list.front(); tok; tok = tok->next()) {
if (tok->str() == "?") {
bool paranthesesNeeded = false;
unsigned int depth = 0;
Token* tok2 = tok->next();
for (; tok2; tok2 = tok2->next()) {
if (tok2->link() && Token::Match(tok2, "{|[|("))
tok2 = tok2->link();
else if (tok2->str() == ":") {
if (depth == 0)
break;
depth--;
} else if (tok2->str() == ";" || tok2->link())
break;
else if (tok2->str() == ",")
paranthesesNeeded = true;
else if (tok2->str() == "?") {
depth++;
paranthesesNeeded = true;
}
}
if (paranthesesNeeded && tok2 && tok2->str() == ":") {
tok->insertToken("(");
tok2->insertToken(")", true);
Token::createMutualLinks(tok->next(), tok2->previous());
}
}
}
}
void Tokenizer::reportError(const Token* tok, const Severity::SeverityType severity, const std::string& id, const std::string& msg, bool inconclusive) const
{
const std::list<const Token*> callstack(1, tok);

View File

@ -357,11 +357,6 @@ public:
*/
void simplifyTypedef();
/**
* Simplify float casts (float)1 => 1.0
*/
void simplifyFloatCasts();
/**
* Simplify casts
*/
@ -695,6 +690,11 @@ private:
* */
bool simplifyStrlen();
/**
* Prepare ternary operators with parantheses so that the AST can be created
* */
void prepareTernaryOpForAST();
/**
* check for duplicate enum definition
*/

View File

@ -852,14 +852,15 @@ static void compileAssignTernary(Token *&tok, AST_state& state)
{
compileLogicOr(tok, state);
while (tok) {
// TODO: http://en.cppreference.com/w/cpp/language/operator_precedence says:
// "The expression in the middle of the conditional operator (between ? and :) is parsed as if parenthesized: its precedence relative to ?: is ignored."
if (tok->isAssignmentOp()) {
state.assign++;
compileBinOp(tok, state, compileAssignTernary);
if (state.assign > 0U)
state.assign--;
} else if (tok->str() == "?") {
// http://en.cppreference.com/w/cpp/language/operator_precedence says about ternary operator:
// "The expression in the middle of the conditional operator (between ? and :) is parsed as if parenthesized: its precedence relative to ?: is ignored."
// Hence, we rely on Tokenizer::prepareTernaryOpForAST() to add such parantheses where necessary.
if (tok->strAt(1) == ":") {
state.op.push(0);
}
@ -895,7 +896,7 @@ static void compileExpression(Token *&tok, AST_state& state)
static Token * createAstAtToken(Token *tok, bool cpp)
{
if (Token::simpleMatch(tok,"for (")) {
if (Token::simpleMatch(tok, "for (")) {
Token *tok2 = tok->tokAt(2);
Token *init1 = nullptr;
const Token * const endPar = tok->next()->link();

View File

@ -187,6 +187,9 @@ private:
TEST_CASE(garbageCode135); // #4994
TEST_CASE(garbageCode136); // #7033
TEST_CASE(garbageCode137); // #7034
TEST_CASE(garbageCode138); // #6660
TEST_CASE(garbageCode139); // #6659
TEST_CASE(garbageCode140); // #7035
TEST_CASE(garbageValueFlow);
TEST_CASE(garbageSymbolDatabase);
@ -361,8 +364,8 @@ private:
}
void garbageCode6() { // #5214
checkCode("int b = ( 0 ? ? ) 1 : 0 ;");
checkCode("int a = int b = ( 0 ? ? ) 1 : 0 ;");
ASSERT_THROW(checkCode("int b = ( 0 ? ? ) 1 : 0 ;"), InternalError);
ASSERT_THROW(checkCode("int a = int b = ( 0 ? ? ) 1 : 0 ;"), InternalError);
}
void garbageCode7() {
@ -640,7 +643,7 @@ private:
}
void garbageCode56() { // #6713
checkCode("void foo() { int a = 0; int b = ???; }");
ASSERT_THROW(checkCode("void foo() { int a = 0; int b = ???; }"), InternalError);
}
void garbageCode57() { // #6731
@ -923,10 +926,9 @@ private:
}
void garbageCode121() { // #2585
checkCode("abcdef?""?<"
"123456?""?>"
"+?""?=");
ASSERT_EQUALS("", errout.str());
ASSERT_THROW(checkCode("abcdef?""?<"
"123456?""?>"
"+?""?="), InternalError);
}
void garbageCode122() { // #6303
@ -1083,6 +1085,30 @@ private:
checkCode("\" \" typedef signed char f; \" \"; void a() { f * s = () &[]; (; ) (; ) }");
}
void garbageCode138() { // #6660
checkCode("CS_PLUGIN_NAMESPACE_BEGIN(csparser)\n"
"{\n"
" struct foo\n"
" {\n"
" union\n"
" {};\n"
" } halo;\n"
"}\n"
"CS_PLUGIN_NAMESPACE_END(csparser)");
}
void garbageCode139() { // #6659 heap user after free: kernel: sm750_accel.c
ASSERT_THROW(checkCode("void hw_copyarea() {\n"
" de_ctrl = (nDirection == RIGHT_TO_LEFT) ?\n"
" ( (0 & ~(((1 << (1 - (0 ? DE_CONTROL_DIRECTION))) - 1) << (0 ? DE_CONTROL_DIRECTION))) )\n"
" : 42;\n"
"}"), InternalError);
}
void garbageCode140() { // #7035
ASSERT_THROW(checkCode("int foo(int align) { int off(= 0 % align; return off) ? \\ align - off : 0; \\ }"), InternalError);
}
void garbageValueFlow() {
// #6089

View File

@ -1881,13 +1881,11 @@ private:
}
{
const char code[] = "a ? b = c , d : e ;"; // do nothing
ASSERT_EQUALS(code, tok(code));
ASSERT_EQUALS("a ? ( b = c , d ) : e ;", tok("a ? b = c , d : e ;")); // Keep comma
}
{
const char code[] = "; return a ? b = c , d : e ;"; // do nothing
ASSERT_EQUALS(code, tok(code));
ASSERT_EQUALS("; return a ? ( b = c , d ) : e ;", tok("; return a ? b = c , d : e ;")); // Keep comma
}
{
@ -1983,7 +1981,7 @@ private:
{
const char code[] = "void f () { switch(n) { case 1?0?1:0:foo(): break; }}";
ASSERT_EQUALS("void f ( ) { switch ( n ) { case 0 : ; break ; } }", tok(code));
TODO_ASSERT_EQUALS("void f ( ) { switch ( n ) { case 0 : ; break ; } }", "void f ( ) { switch ( n ) { case ( 0 ) : ; break ; } }", tok(code));
}
{

View File

@ -385,8 +385,6 @@ private:
// a = b = 0;
TEST_CASE(multipleAssignment);
TEST_CASE(setVarId) // #6660 - crash
TEST_CASE(platformWin);
TEST_CASE(platformWin32);
TEST_CASE(platformWin32A);
@ -442,6 +440,8 @@ private:
TEST_CASE(compileLimits); // #5592 crash: gcc: testsuit: gcc.c-torture/compile/limits-declparen.c
TEST_CASE(prepareTernaryOpForAST);
// AST data
TEST_CASE(astexpr);
TEST_CASE(astpar);
@ -458,7 +458,6 @@ private:
TEST_CASE(removeMacroInClassDef); // #6058
TEST_CASE(sizeofAddParentheses);
TEST_CASE(incompleteTernary); // #6659
TEST_CASE(noreturn); // #5783
}
@ -7923,6 +7922,15 @@ private:
tokenizeAndStringify("[[deprecated]] int f();", false, true, Settings::Unspecified, "test.c", true));
}
void prepareTernaryOpForAST() {
ASSERT_EQUALS("a ? b : c ;", tokenizeAndStringify("a ? b : c;"));
ASSERT_EQUALS("a ? ( b , c ) : d ;", tokenizeAndStringify("a ? b , c : d;"));
ASSERT_EQUALS("a ? ( b , c ) : d ;", tokenizeAndStringify("a ? (b , c) : d;"));
ASSERT_EQUALS("a ? ( 1 ? ( a , b ) : 3 ) : d ;", tokenizeAndStringify("a ? 1 ? a, b : 3 : d;"));
}
std::string testAst(const char code[],bool verbose=false) {
// tokenize given code..
Tokenizer tokenList(&settings0, nullptr);
@ -7935,6 +7943,7 @@ private:
tokenList.createLinks2();
// Create AST..
tokenList.prepareTernaryOpForAST();
tokenList.list.createAst();
// Basic AST validation
@ -7996,8 +8005,12 @@ private:
// assignments are executed from right to left
ASSERT_EQUALS("abc==", testAst("a=b=c;"));
// assignment in ternary operator
// ternary operator
ASSERT_EQUALS("ab0=c1=:?", testAst("a?b=0:c=1;"));
ASSERT_EQUALS("fabc,d:?=e,", testAst("f = a ? b, c : d, e;"));
ASSERT_EQUALS("fabc,de,:?=", testAst("f = (a ? (b, c) : (d, e));"));
ASSERT_EQUALS("fabc,de,:?=", testAst("f = (a ? b, c : (d, e));"));
ASSERT_EQUALS("ab35,4:?foo(:?return", testAst("return (a ? b ? (3,5) : 4 : foo());"));
ASSERT_EQUALS("a\"\"=", testAst("a=\"\""));
ASSERT_EQUALS("a\'\'=", testAst("a=\'\'"));
@ -8293,30 +8306,6 @@ private:
ASSERT_EQUALS("f ( 0 , sizeof ( ptr . bar ) ) ;", tokenizeAndStringify("f(0, sizeof ptr->bar );"));
}
void setVarId() {
const char * code = "CS_PLUGIN_NAMESPACE_BEGIN(csparser)\n"
"{\n"
" struct foo\n"
" {\n"
" union\n"
" {};\n"
" } halo;\n"
"}\n"
"CS_PLUGIN_NAMESPACE_END(csparser)\n";
tokenizeAndStringify(code, true);
}
// #6659 heap user after free: kernel: sm750_accel.c
void incompleteTernary() {
const char * code = "void hw_copyarea() {\n"
" de_ctrl = (nDirection == RIGHT_TO_LEFT) ?\n"
" ( (0 & ~(((1 << (1 - (0 ? DE_CONTROL_DIRECTION))) - 1) << (0 ? DE_CONTROL_DIRECTION))) )\n"
" : 42;\n"
"}";
tokenizeAndStringify(code, true);
}
// see #5783
void noreturn() {
const char code[] = "void myassert() {\n"