masters-thesis/op-finder/OperationFinder.cpp

//
// Created by erki on 16.02.21.
//

#include "OperationFinder.hpp"

#include <iostream>

using namespace clang;
using namespace clang::ast_matchers;

namespace
{

StatementMatcher AssignmentMatcher =
    binaryOperation(isAssignmentOperator(),
                    hasLHS(expr().bind("lhs")),
                    hasRHS(expr().bind("rhs"))).bind("assignment");

StatementMatcher ArithmeticMatcher =
    binaryOperation(hasAnyOperatorName("+", "-", "/", "*", "<", ">",
                                       "<=", ">=", "==", "<<", ">>", "%"),
                    hasLHS(expr().bind("lhs")),
                    hasRHS(expr().bind("lhs"))).bind("arithmetic");

StatementMatcher UnaryArithmeticMatcher =
    unaryOperator(hasAnyOperatorName("!", "++", "--"),
                  hasUnaryOperand(expr().bind("lhs"))).bind("unary_arithmetic");

StatementMatcher CompoundMatcher =
    compoundStmt().bind("compound_stmt");

//StatementMatcher ForStatementMatcher =
//    forStmt(hasInitStatement(expr().bind("init"))).bind("for_stmt");

template<typename TOp>
StringRef getOpcode(const TOp *op)
{
  return op->getOpcodeStr(op->getOpcode());
}

SourceLocation ResolveOperationSourceLocation(const SourceManager& source_manager, const SourceLocation& original)
{
  if (source_manager.isMacroBodyExpansion(original))
  {
    return source_manager.getExpansionLoc(original);
  }

  return original;
}

template<typename TOp>
std::tuple<std::string, unsigned int, unsigned int> resolveLocations(const TOp* op, const SourceManager& source_manager)
{
  const auto& loc = ResolveOperationSourceLocation(source_manager, op->getBeginLoc());

  return {
          source_manager.getFilename(loc).str(),
          source_manager.getSpellingLineNumber(loc),
          source_manager.getSpellingColumnNumber(loc)
  };
}

}

OperationFinder::OperationFinder(IOperationOutput* storage)
  : _storage(storage)
{
  assert(storage);
}

void OperationFinder::addMatcher(MatchFinder &finder)
{
  finder.addMatcher(traverse(TK_IgnoreUnlessSpelledInSource, ArithmeticMatcher), this);
  finder.addMatcher(traverse(TK_IgnoreUnlessSpelledInSource, AssignmentMatcher), this);
  finder.addMatcher(traverse(TK_IgnoreUnlessSpelledInSource, UnaryArithmeticMatcher), this);
}

void OperationFinder::run(const MatchFinder::MatchResult& result)
{
  if (result.Nodes.getNodeAs<clang::BinaryOperator>("assignment"))
  {
    _processAssignment(result);
  }
  else if (result.Nodes.getNodeAs<clang::BinaryOperator>("arithmetic"))
  {
    _processArithmetic(result);
  }
  else if (result.Nodes.getNodeAs<clang::UnaryOperator>("unary_arithmetic"))
  {
    _processUnaryArithmetic(result);
  }
}

void OperationFinder::_processAssignment(const clang::ast_matchers::MatchFinder::MatchResult& result)
{
  const BinaryOperator* op = result.Nodes.getNodeAs<BinaryOperator>("assignment");

  assert(op);

  const auto [file_name, line_number, column_number] = resolveLocations(op, *result.SourceManager);

  llvm::outs()  << file_name << ":"
                << line_number << ":"
                << column_number << ":"
                << "Assignment: Assigned to: ";

  OperationLog log;
  log.line = line_number;
  log.operation = "=";

  _processExpressionTypes(log, op, op->getLHS(), op->getRHS());

  _storage->pushOperation(file_name, log);

  llvm::outs() << "\n";
}

void OperationFinder::_processArithmetic(const MatchFinder::MatchResult& result)
{
  const BinaryOperator* op = result.Nodes.getNodeAs<clang::BinaryOperator>("arithmetic");

  assert(op);

  const auto [file_name, line_number, column_number] = resolveLocations(op, *result.SourceManager);

  llvm::outs()  << file_name << ":"
                << line_number << ":"
                << column_number << ":"
                << "Binary arithmetic: Type: " << getOpcode(op) << " LHS: ";

  OperationLog log;
  log.line = line_number;
  log.operation = getOpcode(op).str();

  _processExpressionTypes(log, op, op->getLHS(), op->getRHS());

  _storage->pushOperation(file_name, log);

  llvm::outs() << "\n";
}

void OperationFinder::_processUnaryArithmetic(const MatchFinder::MatchResult &result)
{
  const UnaryOperator* op = result.Nodes.getNodeAs<UnaryOperator>("unary_arithmetic");
  const Expr* lhs = result.Nodes.getNodeAs<Expr>("lhs");

  assert(op);
  assert(lhs);

  const auto [file_name, line_number, column_number] = resolveLocations(op, *result.SourceManager);

  llvm::outs()  << file_name << ":"
                << line_number << ":"
                << column_number << ":"
                << "Unary arithmetic: Type: " << getOpcode(op) << " LHS: ";

  OperationLog log;
  log.line = line_number;
  log.operation = getOpcode(op).str();

  _processExpressionTypes(log, op, lhs, nullptr);

  _storage->pushOperation(file_name, log);

  llvm::outs() << "\n";
}

void OperationFinder::_processExpressionTypes(OperationLog& log, const Expr* source, const Expr* op1, const Expr* op2)
{

  auto printTypeName = [](const QualType& t) -> std::string
  {
    if (t->isTypedefNameType())
    {
      const TypedefType* tdt = cast<TypedefType>(t);
      assert(tdt);
      llvm::outs() << "Typedef type: " << t.getAsString()
                   << ", underlying: " << tdt->desugar().getAsString()
                   << ".";

      return tdt->desugar().getAsString();
    }
    else if (t->isBuiltinType())
    {
      llvm::outs() << "Builtin type: " << t.getAsString() << ".";
      return t.getAsString();
    }
    else
    {
      llvm::outs() << "Other type: " << t.getAsString() << ".";
      return t.getAsString();
    }
  };

  llvm::outs() << "Expression types:\n\tExpression eval type: ";
  log.operand_result = printTypeName(source->getType());

  if (op1)
  {
    llvm::outs() << "\n\tLHS eval type: ";
    log.operand_lhs = printTypeName(op1->getType());
  }
  if (op2)
  {
    llvm::outs() << "\n\tRHS eval type: ";
    log.operand_rhs = printTypeName(op2->getType());
  }
}