Implemented macro expansion.

src/compiler/preprocessor/new/Diagnostics.h

@@ -33,6 +33,9 @@ class Diagnostics
         UNEXPECTED_TOKEN,
         MACRO_NAME_RESERVED,
         MACRO_REDEFINED,
+        MACRO_UNTERMINATED_INVOCATION,
+        MACRO_TOO_FEW_ARGS,
+        MACRO_TOO_MANY_ARGS,
         INVALID_EXTENSION_NAME,
         INVALID_EXTENSION_BEHAVIOR,
         INVALID_EXTENSION_DIRECTIVE,

src/compiler/preprocessor/new/DirectiveParser.cpp

@@ -120,11 +120,6 @@ void DirectiveParser::parseDirective(Token* token)
             parseLine(token);
     }
 
-    if ((token->type != '\n') && (token->type != 0))
-        mDiagnostics->report(Diagnostics::UNEXPECTED_TOKEN,
-                             token->location,
-                             token->value);
-
     while (token->type != '\n')
     {
         if (token->type == 0) {
@@ -172,7 +167,7 @@ void DirectiveParser::parseDefine(Token* token)
                 break;
             macro.parameters.push_back(token->value);
 
-            mTokenizer->lex(token);  // Get comma.
+            mTokenizer->lex(token);  // Get ','.
         } while (token->type == ',');
 
         if (token->type != ')')
@@ -182,6 +177,7 @@ void DirectiveParser::parseDefine(Token* token)
                                  token->value);
             return;
         }
+        mTokenizer->lex(token);  // Get ')'.
     }
 
     while ((token->type != '\n') && (token->type != Token::LAST))
@@ -193,6 +189,12 @@ void DirectiveParser::parseDefine(Token* token)
         macro.replacements.push_back(*token);
         mTokenizer->lex(token);
     }
+    if (!macro.replacements.empty())
+    {
+        // Whitespace preceding the replacement list is not considered part of
+        // the replacement list for either form of macro.
+        macro.replacements.front().setHasLeadingSpace(false);
+    }
 
     // Check for macro redefinition.
     MacroSet::const_iterator iter = mMacroSet->find(macro.name);

src/compiler/preprocessor/new/Macro.h

@@ -23,13 +23,17 @@ struct Macro
         kTypeObj,
         kTypeFunc
     };
+    typedef std::vector<std::string> Parameters;
+    typedef std::vector<Token> Replacements;
 
+    Macro() : disabled(false), type(kTypeObj) { }
     bool equals(const Macro& other) const;
 
+    mutable bool disabled;
     Type type;
     std::string name;
-    std::vector<std::string> parameters;
-    std::vector<Token> replacements;
+    Parameters parameters;
+    Replacements replacements;
 };
 
 typedef std::map<std::string, Macro> MacroSet;

src/compiler/preprocessor/new/MacroExpander.cpp

@@ -6,9 +6,46 @@
 
 #include "MacroExpander.h"
 
+#include <algorithm>
+#include <cassert>
+
+#include "Diagnostics.h"
+#include "Token.h"
+
 namespace pp
 {
 
+class TokenLexer : public Lexer
+{
+ public:
+    typedef std::vector<Token> TokenVector;
+
+    TokenLexer(TokenVector* tokens)
+    {
+        tokens->swap(mTokens);
+        mIter = mTokens.begin();
+    }
+
+    virtual void lex(Token* token)
+    {
+        if (mIter == mTokens.end())
+        {
+            token->reset();
+            token->type = Token::LAST;
+        }
+        else
+        {
+            *token = *mIter++;
+        }
+    }
+
+ private:
+    PP_DISALLOW_COPY_AND_ASSIGN(TokenLexer);
+
+    TokenVector mTokens;
+    TokenVector::const_iterator mIter;
+};
+
 MacroExpander::MacroExpander(Lexer* lexer,
                              MacroSet* macroSet,
                              Diagnostics* diagnostics) :
@@ -18,10 +55,292 @@ MacroExpander::MacroExpander(Lexer* lexer,
 {
 }
 
+MacroExpander::~MacroExpander()
+{
+    assert(!mReserveToken.get());
+    assert(mContextStack.empty());
+}
+
 void MacroExpander::lex(Token* token)
 {
-    // TODO(alokp): Implement me.
-    mLexer->lex(token);
+    while (true)
+    {
+        getToken(token);
+
+        if (token->type != Token::IDENTIFIER)
+            break;
+
+        if (token->expansionDisabled())
+            break;
+
+        MacroSet::const_iterator iter = mMacroSet->find(token->value);
+        if (iter == mMacroSet->end())
+            break;
+
+        const Macro& macro = iter->second;
+        if (macro.disabled)
+        {
+            // If a particular token is not expanded, it is never expanded.
+            token->setExpansionDisabled(true);
+            break;
+        }
+        if ((macro.type == Macro::kTypeFunc) && !isNextTokenLeftParen())
+        {
+            // If the token immediately after the macro name is not a '(',
+            // this macro should not be expanded.
+            break;
+        }
+
+        pushMacro(macro, *token);
+    }
+}
+
+void MacroExpander::getToken(Token* token)
+{
+    if (mReserveToken.get())
+    {
+        *token = *mReserveToken;
+        mReserveToken.reset();
+        return;
+    }
+
+    // First pop all empty macro contexts.
+    while (!mContextStack.empty() && mContextStack.back()->empty())
+    {
+        popMacro();
+    }
+
+    if (!mContextStack.empty())
+    {
+        *token = mContextStack.back()->get();
+    }
+    else
+    {
+        mLexer->lex(token);
+    }
+}
+
+void MacroExpander::ungetToken(const Token& token)
+{
+    if (!mContextStack.empty())
+    {
+        MacroContext* context = mContextStack.back();
+        context->unget();
+        assert(context->index >= 0);
+        assert(context->replacements[context->index] == token);
+    }
+    else
+    {
+        assert(!mReserveToken.get());
+        mReserveToken.reset(new Token(token));
+    }
+}
+
+bool MacroExpander::isNextTokenLeftParen()
+{
+    Token token;
+    getToken(&token);
+
+    bool lparen = token.type == '(';
+    ungetToken(token);
+
+    return lparen;
+}
+
+bool MacroExpander::pushMacro(const Macro& macro, const Token& identifier)
+{
+    assert(!macro.disabled);
+    assert(!identifier.expansionDisabled());
+    assert(identifier.type == Token::IDENTIFIER);
+    assert(identifier.value == macro.name);
+
+    std::vector<Token> replacements;
+    if (!expandMacro(macro, identifier, &replacements))
+        return false;
+
+    // Macro is disabled for expansion until it is popped off the stack.
+    macro.disabled = true;
+
+    MacroContext* context = new MacroContext;
+    context->macro = &macro;
+    context->replacements.swap(replacements);
+    mContextStack.push_back(context);
+    return true;
+}
+
+void MacroExpander::popMacro()
+{
+    assert(!mContextStack.empty());
+
+    MacroContext* context = mContextStack.back();
+    mContextStack.pop_back();
+
+    assert(context->empty());
+    assert(context->macro->disabled);
+    context->macro->disabled = false;
+    delete context;
+}
+
+bool MacroExpander::expandMacro(const Macro& macro,
+                                const Token& identifier,
+                                std::vector<Token>* replacements)
+{
+    if (macro.type == Macro::kTypeObj)
+    {
+        replacements->assign(macro.replacements.begin(),
+                             macro.replacements.end());
+    }
+    else
+    {
+        assert(macro.type == Macro::kTypeFunc);
+        std::vector<MacroArg> args;
+        args.reserve(macro.parameters.size());
+        if (!collectMacroArgs(macro, identifier, &args))
+            return false;
+
+        replaceMacroParams(macro, args, replacements);
+    }
+
+    for (size_t i = 0; i < replacements->size(); ++i)
+    {
+        Token& repl = replacements->at(i);
+        if (i == 0)
+        {
+            // The first token in the replacement list inherits the padding
+            // properties of the identifier token.
+            repl.setAtStartOfLine(identifier.atStartOfLine());
+            repl.setHasLeadingSpace(identifier.hasLeadingSpace());
+        }
+        repl.location = identifier.location;
+    }
+    return true;
+}
+
+bool MacroExpander::collectMacroArgs(const Macro& macro,
+                                     const Token& identifier,
+                                     std::vector<MacroArg>* args)
+{
+    Token token;
+    getToken(&token);
+    assert(token.type == '(');
+
+    args->push_back(MacroArg());
+    for (int openParens = 1; openParens != 0; )
+    {
+        getToken(&token);
+
+        if (token.type == Token::LAST)
+        {
+            mDiagnostics->report(Diagnostics::MACRO_UNTERMINATED_INVOCATION,
+                                 identifier.location, identifier.value);
+            // Do not lose EOF token.
+            ungetToken(token);
+            return false;
+        }
+
+        bool isArg = false; // True if token is part of the current argument.
+        switch (token.type)
+        {
+          case '(':
+            ++openParens;
+            isArg = true;
+            break;
+          case ')':
+            --openParens;
+            isArg = openParens != 0;
+            break;
+          case ',':
+            // The individual arguments are separated by comma tokens, but
+            // the comma tokens between matching inner parentheses do not
+            // seperate arguments.
+            if (openParens == 1) args->push_back(MacroArg());
+            isArg = openParens != 1;
+            break;
+          default:
+            isArg = true;
+            break;
+        }
+        if (isArg)
+        {
+            MacroArg& arg = args->back();
+            // Initial whitespace is not part of the argument.
+            if (arg.empty()) token.setHasLeadingSpace(false);
+            arg.push_back(token);
+        }
+    }
+
+    const Macro::Parameters& params = macro.parameters;
+    // If there is only one empty argument, it is equivalent to no argument.
+    if (params.empty() && (args->size() == 1) && args->front().empty())
+    {
+        args->clear();
+    }
+    // Validate the number of arguments.
+    if (args->size() != params.size())
+    {
+        Diagnostics::ID id = args->size() < macro.parameters.size() ?
+            Diagnostics::MACRO_TOO_FEW_ARGS :
+            Diagnostics::MACRO_TOO_MANY_ARGS;
+        mDiagnostics->report(id, identifier.location, identifier.value);
+        return false;
+    }
+
+    // Pre-expand each argument before substitution.
+    // This step expands each argument individually before they are
+    // inserted into the macro body.
+    for (size_t i = 0; i < args->size(); ++i)
+    {
+        MacroArg& arg = args->at(i);
+        TokenLexer lexer(&arg);
+        MacroExpander expander(&lexer, mMacroSet, mDiagnostics);
+
+        arg.clear();
+        expander.lex(&token);
+        while (token.type != Token::LAST)
+        {
+            arg.push_back(token);
+            expander.lex(&token);
+        }
+    }
+    return true;
+}
+
+void MacroExpander::replaceMacroParams(const Macro& macro,
+                                       const std::vector<MacroArg>& args,
+                                       std::vector<Token>* replacements)
+{
+    for (size_t i = 0; i < macro.replacements.size(); ++i)
+    {
+        const Token& repl = macro.replacements[i];
+        if (repl.type != Token::IDENTIFIER)
+        {
+            replacements->push_back(repl);
+            continue;
+        }
+
+        // TODO(alokp): Optimize this.
+        // There is no need to search for macro params every time.
+        // The param index can be cached with the replacement token.
+        Macro::Parameters::const_iterator iter = std::find(
+            macro.parameters.begin(), macro.parameters.end(), repl.value);
+        if (iter == macro.parameters.end())
+        {
+            replacements->push_back(repl);
+            continue;
+        }
+
+        size_t iArg = std::distance(macro.parameters.begin(), iter);
+        const MacroArg& arg = args[iArg];
+        if (arg.empty())
+        {
+            continue;
+        }
+        size_t iRepl = replacements->size();
+        replacements->insert(replacements->end(), arg.begin(), arg.end());
+        // The replacement token inherits padding properties from
+        // macro replacement token.
+        replacements->at(iRepl).setHasLeadingSpace(repl.hasLeadingSpace());
+    }
 }
 
 }  // namespace pp

src/compiler/preprocessor/new/MacroExpander.h

@@ -7,6 +7,8 @@
 #ifndef COMPILER_PREPROCESSOR_MACRO_EXPANDER_H_
 #define COMPILER_PREPROCESSOR_MACRO_EXPANDER_H_
 
+#include <vector>
+
 #include "Lexer.h"
 #include "Macro.h"
 #include "pp_utils.h"
@@ -20,15 +22,50 @@ class MacroExpander : public Lexer
 {
   public:
     MacroExpander(Lexer* lexer, MacroSet* macroSet, Diagnostics* diagnostics);
+    virtual ~MacroExpander();
 
     virtual void lex(Token* token);
 
   private:
     PP_DISALLOW_COPY_AND_ASSIGN(MacroExpander);
 
+    void getToken(Token* token);
+    void ungetToken(const Token& token);
+    bool isNextTokenLeftParen();
+
+    bool pushMacro(const Macro& macro, const Token& identifier);
+    void popMacro();
+
+    bool expandMacro(const Macro& macro,
+                     const Token& identifier,
+                     std::vector<Token>* replacements);
+
+    typedef std::vector<Token> MacroArg;
+    bool collectMacroArgs(const Macro& macro,
+                          const Token& identifier,
+                          std::vector<MacroArg>* args);
+    void replaceMacroParams(const Macro& macro,
+                            const std::vector<MacroArg>& args,
+                            std::vector<Token>* replacements);
+
+    struct MacroContext
+    {
+        const Macro* macro;
+        size_t index;
+        std::vector<Token> replacements;
+
+        MacroContext() : macro(0), index(0) { }
+        bool empty() const { return index == replacements.size(); }
+        const Token& get() { return replacements[index++]; }
+        void unget() { --index; }
+    };
+
     Lexer* mLexer;
     MacroSet* mMacroSet;
     Diagnostics* mDiagnostics;
+
+    std::auto_ptr<Token> mReserveToken;
+    std::vector<MacroContext*> mContextStack;
 };
 
 }  // namespace pp

src/compiler/preprocessor/new/Token.cpp

@@ -41,6 +41,14 @@ void Token::setHasLeadingSpace(bool space)
         flags &= ~HAS_LEADING_SPACE;
 }
 
+void Token::setExpansionDisabled(bool disable)
+{
+    if (disable)
+        flags |= EXPANSION_DISABLED;
+    else
+        flags &= ~EXPANSION_DISABLED;
+}
+
 std::ostream& operator<<(std::ostream& out, const Token& token)
 {
     if (token.hasLeadingSpace())

src/compiler/preprocessor/new/Token.h

@@ -50,8 +50,9 @@ struct Token
     };
     enum Flags
     {
-        AT_START_OF_LINE  = 1 << 0,
-        HAS_LEADING_SPACE = 1 << 1
+        AT_START_OF_LINE   = 1 << 0,
+        HAS_LEADING_SPACE  = 1 << 1,
+        EXPANSION_DISABLED = 1 << 2
     };
 
     Token() : type(0), flags(0) { }
@@ -67,8 +68,11 @@ struct Token
     bool hasLeadingSpace() const { return (flags & HAS_LEADING_SPACE) != 0; }
     void setHasLeadingSpace(bool space);
 
+    bool expansionDisabled() const { return (flags & EXPANSION_DISABLED) != 0; }
+    void setExpansionDisabled(bool disable);
+
     int type;
-    int flags;
+    unsigned int flags;
     SourceLocation location;
     std::string value;
 };

src/compiler/preprocessor/new/Tokenizer.cpp

@@ -2305,6 +2305,7 @@ bool Tokenizer::init(int count, const char* const string[], const int length[])
 void Tokenizer::lex(Token* token)
 {
     token->type = pplex(&token->value,&token->location,mHandle);
+    token->flags = 0;
 
     token->setAtStartOfLine(mContext.lineStart);
     mContext.lineStart = token->type == '\n';

src/compiler/preprocessor/new/Tokenizer.l

@@ -289,6 +289,7 @@ bool Tokenizer::init(int count, const char* const string[], const int length[])
 void Tokenizer::lex(Token* token)
 {
     token->type = yylex(&token->value, &token->location, mHandle);
+    token->flags = 0;
 
     token->setAtStartOfLine(mContext.lineStart);
     mContext.lineStart = token->type == '\n';

tests/build_tests.gyp

@@ -44,6 +44,7 @@
         '../third_party/googlemock/src/gmock_main.cc',
         'preprocessor_tests/char_test.cpp',
         'preprocessor_tests/comment_test.cpp',
+        'preprocessor_tests/define_test.cpp',
         'preprocessor_tests/error_test.cpp',
         'preprocessor_tests/extension_test.cpp',
         'preprocessor_tests/identifier_test.cpp',