[PP] Obj-like macro define

src/exec/wasmvm-cc.cpp

@@ -135,17 +135,17 @@ int main(int argc, const char* argv[]){
         if(args["pp"]){
             for(std::filesystem::path source_path : source_files){
                 PreProcessor pp(source_path);
-                // if(args["output"]){
-                //     std::ofstream output_file(std::get<std::string>(args["output"].value()));
-                //     for(PreProcessor::PPToken tok = pp.get(); tok.has_value(); tok = pp.get()){
-                //         output_file << tok.value();
-                //     }
-                //     output_file.close();
-                // }else{
-                //     for(PreProcessor::PPToken tok = pp.get(); tok.has_value(); tok = pp.get()){
-                //         std::cout << tok.value();
-                //     }
-                // }
+                if(args["output"]){
+                    std::ofstream output_file(std::get<std::string>(args["output"].value()));
+                    for(std::optional<PreProcessor::Token> token = pp.get(); token.has_value(); token = pp.get()){
+                        output_file << token->text;
+                    }
+                    output_file.close();
+                }else{
+                    for(std::optional<PreProcessor::Token> token = pp.get(); token.has_value(); token = pp.get()){
+                        std::cout << token->text;
+                    }
+                }
             }
             return 0;
         }

src/include/PreProcessor.hpp

@@ -1,13 +1,13 @@
 #ifndef WVMCC_PreProcessor_DEF
 #define WVMCC_PreProcessor_DEF
 
-#include <istream>
-#include <sstream>
+#include <ostream>
 #include <filesystem>
 #include <stack>
 #include <deque>
 #include <string>
 #include <optional>
+#include <unordered_map>
 #include <common.hpp>
 
 namespace WasmVM {
@@ -22,13 +22,23 @@ struct PreProcessor {
         SourcePos pos;
     };
 
+    struct Macro {
+        std::string name;
+        std::optional<std::vector<std::string>> params;
+        std::vector<Token> replacement;
+        bool operator==(const Macro& op) const;
+    };
+
     PreProcessor(std::filesystem::path filepath);
     std::optional<Token> get();
 
 protected:
     struct Visitor;
     std::stack<std::shared_ptr<Visitor>> visitors;
     std::deque<Token> buffer;
+    std::unordered_map<std::string, Macro> macros;
+
+    friend struct Visitor;
 };
 
 } // namespace WasmVM

src/lib/PreProcessor/CMakeLists.txt

@@ -11,7 +11,6 @@ include_directories(
 
 add_library(pp
   PreProcessor.cpp
-  ErrorListener.cpp
   PPLexer.cpp
   PPParser.cpp
   Visitor.cpp

src/lib/PreProcessor/PP.g4

@@ -6,8 +6,15 @@ options {
 
 import PPToken;
 
-group : text_line | EOF;
+group : control_line | text_line;
 
-text_line : pp_token* NewLine?;
+control_line : define_obj
+    | Hash pp_token* line_end;
 
-pp_token : HeaderName | Identifier | PPNumber | CharConst | StringLiteral | Punctuator | BlockComment | LineComment | WhiteSpace;
+text_line : pp_token* line_end;
+
+define_obj : Hash WhiteSpace* 'define' WhiteSpace+ Identifier WhiteSpace+ pp_token* line_end;
+
+line_end : NewLine? EOF | NewLine;
+
+pp_token : HeaderName | Identifier | PPNumber | CharConst | StringLiteral | Punctuator | BlockComment | LineComment | WhiteSpace;

src/lib/PreProcessor/PPToken.g4

@@ -37,7 +37,9 @@ Escape : BSlash_ ([abfnrtv'"?\\] | BSlash_ | [0-7] [0-7]? [0-7]? | 'x' HexDigit
 StringLiteral : ([LuU] | 'u8')? '"' (~["\\\n] | Escape)* '"';
 
 /* Punctuator */
-Punctuator : '%:%:' | '<<=' | '>>=' | '...' | '%:' | '++' | '<'[:%<]? | [:%>]?'>' | ([*/%+\-&^|><!=] | Bar_ | Caret_)?'=' | '&&' | '##' | Hash_ Hash_ | '-' [\->] | BraceL_ | BraceR_ | BracketL_ | BracketR_ | (Bar_ Bar_?) | Caret_ | Tlide_ | Hash_ | [().&*+\-!/%?:;,];
+HashHash : '%:%:' | Hash_ Hash_;
+Hash : '%:' | Hash_;
+Punctuator : '<<=' | '>>=' | '...' | '++' | '<'[:%<]? | [:%>]?'>' | ([*/%+\-&^|><!=] | Bar_ | Caret_)?'=' | '&&' | '-' [\->] | BraceL_ | BraceR_ | BracketL_ | BracketR_ | (Bar_ Bar_?) | Caret_ | Tlide_ | [().&*+\-!/%?:;,];
 
 /* White space */
 WhiteSpace : [\t\f\r \u000B]+ ;

src/lib/PreProcessor/PreProcessor.cpp

@@ -36,4 +36,23 @@ std::optional<PreProcessor::Token> PreProcessor::get(){
     Token token = buffer.front();
     buffer.pop_front();
     return token;
+}
+
+bool PreProcessor::Macro::operator==(const PreProcessor::Macro& op) const {
+    if((op.name != name) || (op.params.has_value() != params.has_value()) || (op.replacement.size() != replacement.size())){
+        return false;
+    }
+    for(auto op_it = op.replacement.begin(), it = replacement.begin(); op_it != op.replacement.end(); op_it = std::next(op_it), it = std::next(it)){
+        if((op_it->type != it->type) || (op_it->text != it->text)){
+            return false;
+        }
+    }
+    if(op.params.has_value()){
+        for(auto op_it = op.params->begin(), it = params->begin(); op_it != op.params->end(); op_it = std::next(op_it), it = std::next(it)){
+            if(*op_it != *it){
+                return false;
+            }
+        }
+    }
+    return true;
 }

src/lib/PreProcessor/Visitor.cpp

@@ -15,6 +15,7 @@
 
 #include "Visitor.hpp"
 #include <Error.hpp>
+#include <exception.hpp>
 
 using namespace WasmVM;
 
@@ -25,24 +26,17 @@ PreProcessor::Visitor::Visitor(std::filesystem::path path, PreProcessor& pp) : p
     lexer = std::make_unique<PPLexer>(input.get());
     tokens = std::make_unique<antlr4::CommonTokenStream>(lexer.get());
     parser = std::make_unique<PPParser>(tokens.get());
-    parser->removeErrorListeners();
-    parser->addErrorListener(&parseErrorListener);
 }
 
 bool PreProcessor::Visitor::fetch(){
-    auto* group = parser->group();
-    visitGroup(group);
-    return !parser->isMatchedEOF();
+    return std::any_cast<bool>(visitGroup(parser->group()));
 }
 
 std::any PreProcessor::Visitor::visitGroup(PPParser::GroupContext *ctx){
     return visitChildren(ctx);
 }
 
-std::any PreProcessor::Visitor::visitText_line(PPParser::Text_lineContext *ctx){
-    for(auto pp_token : ctx->pp_token()){
-        pp.buffer.emplace_back(std::any_cast<Token>(visitPp_token(pp_token)));
-    }
+std::any PreProcessor::Visitor::visitLine_end(PPParser::Line_endContext* ctx){
     antlr4::tree::TerminalNode* newline = ctx->NewLine();
     if(newline){
         pp.buffer.emplace_back(Token {
@@ -54,10 +48,26 @@ std::any PreProcessor::Visitor::visitText_line(PPParser::Text_lineContext *ctx){
                 .col = newline->getSymbol()->getCharPositionInLine() + 1
             }
         });
-        return ctx->pp_token().size() + 1;
+    }
+    if(ctx->EOF()){
+        if(newline == nullptr){
+            Exception::Warning("no new-line in the end of file");
+        }
+        return false;
     }else{
-        return ctx->pp_token().size();
+        return true;
+    }
+}
+
+std::any PreProcessor::Visitor::visitText_line(PPParser::Text_lineContext *ctx){
+    for(auto pp_token : ctx->pp_token()){
+        pp.buffer.emplace_back(std::any_cast<Token>(visitPp_token(pp_token)));
     }
+    return visitLine_end(ctx->line_end());
+}
+
+std::any PreProcessor::Visitor::visitControl_line(PPParser::Control_lineContext *ctx){
+    return visitChildren(ctx);
 }
 
 std::any PreProcessor::Visitor::visitPp_token(PPParser::Pp_tokenContext *ctx){
@@ -81,6 +91,32 @@ std::any PreProcessor::Visitor::visitPp_token(PPParser::Pp_tokenContext *ctx){
         case PPLexer::WhiteSpace:
             return Token {.type = Token::WhiteSpace, .text = " ", .pos = pos};
         default:
-            throw Exception::SyntaxError(pos, std::string("unknown pre-processor token '") + token->getText() + "'");
+            throw Exception::SyntaxError(pos, "unknown pre-processor token '" + token->getText() + "'");
+    }
+}
+
+std::any PreProcessor::Visitor::visitDefine_obj(PPParser::Define_objContext *ctx){
+    std::string name = ctx->Identifier()->getText();
+    Macro macro {.name = name};
+    for(auto pp_token : ctx->pp_token()){
+        macro.replacement.emplace_back(std::any_cast<Token>(visitPp_token(pp_token)));
+    }
+    if(pp.macros.contains(name)){
+        if(pp.macros[name] != macro){
+            SourcePos pos {
+                .file = path,
+                .line = ctx->getStart()->getLine(),
+                .col = ctx->getStart()->getCharPositionInLine() + 1
+            };
+            throw Exception::SyntaxError(pos, "duplicated macro '" + name + "'");
+        }
+    }else{
+        pp.macros[name] = macro;
+        std::cout << "define macro [" << name << "]: ";
+        for(Token token : macro.replacement){
+            std::cout << token.text;
+        }
+        std::cout << std::endl;
     }
+    return visitLine_end(ctx->line_end());
 }

src/lib/PreProcessor/Visitor.hpp

@@ -6,8 +6,7 @@
 #include <PPLexer.h>
 #include <optional>
 #include <memory>
-
-#include "ErrorListener.hpp"
+#include <fstream>
 
 namespace WasmVM {
 
@@ -17,13 +16,15 @@ struct PreProcessor::Visitor : public PPVisitor {
     bool fetch(); // false if match EOF
 
     std::filesystem::path path;
-    ParseErrorListener parseErrorListener;
-
 
 protected:
     virtual std::any visitGroup(PPParser::GroupContext *) override;
     virtual std::any visitText_line(PPParser::Text_lineContext *) override;
     virtual std::any visitPp_token(PPParser::Pp_tokenContext *ctx) override;
+    virtual std::any visitLine_end(PPParser::Line_endContext* ctx) override;
+    virtual std::any visitControl_line(PPParser::Control_lineContext* ctx) override;
+    virtual std::any visitDefine_obj(PPParser::Define_objContext* ctx) override;
+
     std::ifstream fin;
     std::unique_ptr<antlr4::ANTLRInputStream> input;
     std::unique_ptr<PPLexer> lexer;

src/lib/PreProcessor/test.c

@@ -0,0 +1,5 @@
+#define foo const int test1
+foo = 1;
+#define foo2 test1 +
+#define val 4
+int test2 = foo2 val;

src/lib/PreProcessor/test.cpp

@@ -47,7 +47,7 @@ int main(int argc, char const *argv[])
         std::cerr << "Warning : " << message << std::endl;
     });
 
-    PreProcessor pp("test.txt");
+    PreProcessor pp("test.c");
 
     try{
         for(std::optional<PreProcessor::Token> token = pp.get(); token.has_value(); token = pp.get()){

test/CMakeLists.txt

@@ -31,5 +31,5 @@ macro(add_test_suite suite_name)
 endmacro(add_test_suite)
 
 add_test_suite(pp_token)
-# add_test_suite(pp_directive)
-# add_test_suite(pp_macro)
+add_test_suite(pp_macro)
+# add_test_suite(pp_directive)