Merge pull request 'bnk' (#1) from bnk into master

Reviewed-on: pabgh52us/nudt-compiler-cpp#1

.gitignore

@@ -68,3 +68,8 @@ Thumbs.db
 # Project outputs
 # =========================
 test/test_result/
+
+
+# Added by cargo
+
+/target

CMakeLists.txt

@@ -2,6 +2,8 @@ cmake_minimum_required(VERSION 3.20)
 
 project(compiler LANGUAGES C CXX)
 
+find_package(Java REQUIRED COMPONENTS Runtime)
+
 # 统一 C++ 标准（按实验环境可调整）
 set(CMAKE_CXX_STANDARD 17)
 set(CMAKE_CXX_STANDARD_REQUIRED ON)
@@ -39,6 +41,11 @@ endif()
 
 option(COMPILER_PARSE_ONLY "Build only the frontend parser pipeline" OFF)
 
+set(ANTLR4_JAR "${PROJECT_SOURCE_DIR}/third_party/antlr-4.13.2-complete.jar")
+if(NOT EXISTS "${ANTLR4_JAR}")
+  message(FATAL_ERROR "ANTLR jar not found: ${ANTLR4_JAR}")
+endif()
+
 # 使用仓库内 third_party 提供的 ANTLR4 C++ runtime（较新版本）
 # third_party 目录结构以仓库为准：runtime 源码位于 third_party/antlr4-runtime-4.13.2/runtime/src
 set(ANTLR4_RUNTIME_SRC_DIR "${PROJECT_SOURCE_DIR}/third_party/antlr4-runtime-4.13.2/runtime/src")

Cargo.lock

@@ -0,0 +1,7 @@
+# This file is automatically @generated by Cargo.
+# It is not intended for manual editing.
+version = 4
+
+[[package]]
+name = "nudt-compiler-cpp"
+version = "0.1.0"

Cargo.toml

@@ -0,0 +1,6 @@
+[package]
+name = "nudt-compiler-cpp"
+version = "0.1.0"
+edition = "2024"
+
+[dependencies]

src/antlr4/SysY.g4

@@ -1,67 +1,145 @@
-// SysY 子集语法：支持形如
-// int main() { int a = 1; int b = 2; return a + b; }
-// 的最小返回表达式编译。
-
-// 后续需要自行添加
-
 grammar SysY;
 
 /*===-------------------------------------------===*/
 /* Lexer rules                                     */
 /*===-------------------------------------------===*/
 
+CONST: 'const';
 INT: 'int';
+FLOAT: 'float';
+VOID: 'void';
+IF: 'if';
+ELSE: 'else';
+WHILE: 'while';
+BREAK: 'break';
+CONTINUE: 'continue';
 RETURN: 'return';
 
 ASSIGN: '=';
+EQ: '==';
+NE: '!=';
+LT: '<';
+GT: '>';
+LE: '<=';
+GE: '>=';
+
 ADD: '+';
+SUB: '-';
+MUL: '*';
+DIV: '/';
+MOD: '%';
+NOT: '!';
+LAND: '&&';
+LOR: '||';
 
 LPAREN: '(';
 RPAREN: ')';
 LBRACE: '{';
 RBRACE: '}';
+LBRACK: '[';
+RBRACK: ']';
+COMMA: ',';
 SEMICOLON: ';';
 
+FLOAT_CONST
+    : DEC_FLOAT_CONST
+    | HEX_FLOAT_CONST
+    ;
+
+INT_CONST
+    : HEX_PREFIX HEX_DIGIT+
+    | '0' [0-7]+
+    | '0'
+    | [1-9] DIGIT*
+    ;
+
 ID: [a-zA-Z_][a-zA-Z_0-9]*;
-ILITERAL: [0-9]+;
 
 WS: [ \t\r\n] -> skip;
 LINECOMMENT: '//' ~[\r\n]* -> skip;
 BLOCKCOMMENT: '/*' .*? '*/' -> skip;
 
+fragment DEC_FLOAT_CONST
+    : DIGIT+ '.' DIGIT* EXP_PART?
+    | '.' DIGIT+ EXP_PART?
+    | DIGIT+ EXP_PART
+    ;
+
+fragment HEX_FLOAT_CONST
+    : HEX_PREFIX HEX_DIGIT+ '.' HEX_DIGIT* BIN_EXP_PART
+    | HEX_PREFIX '.' HEX_DIGIT+ BIN_EXP_PART
+    | HEX_PREFIX HEX_DIGIT+ BIN_EXP_PART
+    ;
+
+fragment EXP_PART: [eE] [+-]? DIGIT+;
+fragment BIN_EXP_PART: [pP] [+-]? DIGIT+;
+fragment HEX_PREFIX: '0' [xX];
+fragment HEX_DIGIT: [0-9a-fA-F];
+fragment DIGIT: [0-9];
+
 /*===-------------------------------------------===*/
 /* Syntax rules                                    */
 /*===-------------------------------------------===*/
 
 compUnit
-    : funcDef EOF
+    : (decl | funcDef)+ EOF
     ;
 
 decl
-    : btype varDef SEMICOLON
+    : constDecl
+    | varDecl
     ;
 
-btype
+constDecl
+    : CONST bType constDef (COMMA constDef)* SEMICOLON
+    ;
+
+varDecl
+    : bType varDef (COMMA varDef)* SEMICOLON
+    ;
+
+bType
     : INT
+    | FLOAT
+    ;
+
+constDef
+    : ID (LBRACK constExp RBRACK)* ASSIGN constInitVal
+    ;
+
+constInitVal
+    : constExp
+    | LBRACE (constInitVal (COMMA constInitVal)*)? RBRACE
     ;
 
 varDef
-    : lValue (ASSIGN initValue)?
+    : ID (LBRACK constExp RBRACK)* (ASSIGN initVal)?
     ;
 
-initValue
+initVal
     : exp
+    | LBRACE (initVal (COMMA initVal)*)? RBRACE
     ;
 
 funcDef
-    : funcType ID LPAREN RPAREN blockStmt
+    : funcType ID LPAREN funcFParams? RPAREN block
     ;
 
 funcType
-    : INT
+    : VOID
+    | INT
+    | FLOAT
+    ;
+
+funcFParams
+    : funcFParam (COMMA funcFParam)*
+    ;
+
+funcFParam
+    : bType ID (LBRACK RBRACK (LBRACK exp RBRACK)*)?
     ;
 
-blockStmt
+block
     : LBRACE blockItem* RBRACE
     ;
 
@@ -71,28 +149,80 @@ blockItem
     ;
 
 stmt
-    : returnStmt
+    : lVal ASSIGN exp SEMICOLON
+    | exp? SEMICOLON
+    | block
+    | IF LPAREN cond RPAREN stmt (ELSE stmt)?
+    | WHILE LPAREN cond RPAREN stmt
+    | BREAK SEMICOLON
+    | CONTINUE SEMICOLON
+    | RETURN exp? SEMICOLON
     ;
 
-returnStmt
+exp
-    : RETURN exp SEMICOLON
+    : addExp
     ;
 
-exp
+cond
-    : LPAREN exp RPAREN          # parenExp
+    : lOrExp
-    | var                        # varExp
-    | number                     # numberExp
-    | exp ADD exp                # additiveExp
     ;
 
-var
+lVal
-    : ID
+    : ID (LBRACK exp RBRACK)*
     ;
 
-lValue
+primaryExp
-    : ID
+    : LPAREN exp RPAREN
+    | lVal
+    | number
     ;
 
 number
-    : ILITERAL
+    : INT_CONST
+    | FLOAT_CONST
+    ;
+
+unaryExp
+    : primaryExp
+    | ID LPAREN funcRParams? RPAREN
+    | unaryOp unaryExp
+    ;
+
+unaryOp
+    : ADD
+    | SUB
+    | NOT
+    ;
+
+funcRParams
+    : exp (COMMA exp)*
+    ;
+
+mulExp
+    : unaryExp ((MUL | DIV | MOD) unaryExp)*
+    ;
+
+addExp
+    : mulExp ((ADD | SUB) mulExp)*
+    ;
+
+relExp
+    : addExp ((LT | GT | LE | GE) addExp)*
+    ;
+
+eqExp
+    : relExp ((EQ | NE) relExp)*
+    ;
+
+lAndExp
+    : eqExp (LAND eqExp)*
+    ;
+
+lOrExp
+    : lAndExp (LOR lAndExp)*
     ;
+
+constExp
+    : addExp
+    ;
+

src/frontend/CMakeLists.txt

@@ -3,15 +3,44 @@ add_library(frontend STATIC
   SyntaxTreePrinter.cpp
 )
 
+set(ANTLR4_GRAMMAR "${PROJECT_SOURCE_DIR}/src/antlr4/SysY.g4")
+set(ANTLR4_GENERATED_FILES
+  "${ANTLR4_GENERATED_DIR}/SysYLexer.cpp"
+  "${ANTLR4_GENERATED_DIR}/SysYLexer.h"
+  "${ANTLR4_GENERATED_DIR}/SysYLexer.interp"
+  "${ANTLR4_GENERATED_DIR}/SysYLexer.tokens"
+  "${ANTLR4_GENERATED_DIR}/SysYParser.cpp"
+  "${ANTLR4_GENERATED_DIR}/SysYParser.h"
+  "${ANTLR4_GENERATED_DIR}/SysY.interp"
+  "${ANTLR4_GENERATED_DIR}/SysY.tokens"
+  "${ANTLR4_GENERATED_DIR}/SysYBaseVisitor.h"
+  "${ANTLR4_GENERATED_DIR}/SysYVisitor.h"
+)
+
+add_custom_command(
+  OUTPUT ${ANTLR4_GENERATED_FILES}
+  COMMAND ${CMAKE_COMMAND} -E make_directory "${ANTLR4_GENERATED_DIR}"
+  COMMAND ${Java_JAVA_EXECUTABLE} -jar "${ANTLR4_JAR}"
+          -Dlanguage=Cpp
+          -visitor
+          -no-listener
+          -Xexact-output-dir
+          -o "${ANTLR4_GENERATED_DIR}"
+          "${ANTLR4_GRAMMAR}"
+  DEPENDS "${ANTLR4_GRAMMAR}" "${ANTLR4_JAR}"
+  COMMENT "Generating ANTLR4 parser sources from SysY.g4"
+  VERBATIM
+)
+
+add_custom_target(antlr4_generated DEPENDS ${ANTLR4_GENERATED_FILES})
+add_dependencies(frontend antlr4_generated)
+
+target_sources(frontend PRIVATE
+  "${ANTLR4_GENERATED_DIR}/SysYLexer.cpp"
+  "${ANTLR4_GENERATED_DIR}/SysYParser.cpp"
+)
+
 target_link_libraries(frontend PUBLIC
   build_options
   ${ANTLR4_RUNTIME_TARGET}
 )
-
-# 自动纳入构建目录中的 Lexer/Parser 生成源码（若存在）
-file(GLOB_RECURSE ANTLR4_GENERATED_SOURCES CONFIGURE_DEPENDS
-  "${ANTLR4_GENERATED_DIR}/*.cpp"
-)
-if(ANTLR4_GENERATED_SOURCES)
-  target_sources(frontend PRIVATE ${ANTLR4_GENERATED_SOURCES})
-endif()

src/main.rs

@@ -0,0 +1,3 @@
+fn main() {
+    println!("Hello, world!");
+}

test_parse.sh

@@ -0,0 +1,78 @@
+#!/bin/bash
+# ================================================
+# SysY 编译器 Lab1 批量解析测试脚本
+# 文件名：scripts/test_parse.sh
+# 适用环境：Arch Linux（bash 原生支持，无需额外安装）
+# 功能：
+#   - 遍历 test/test_case 下所有 .sy 文件（functional + performance）
+#   - 执行 --emit-parse-tree 检查是否能成功解析
+#   - 输出简洁的 PASS/FAIL 结果 + 统计
+#   - 错误文件会自动打印最后 10 行报错信息（方便调试）
+#   - 所有结果保存到 test/test_result/parse_test.log
+# ================================================
+
+set -u  # 遇到未定义变量直接报错
+
+# ================== 配置 ==================
+COMPILER="./build/bin/compiler"
+TEST_DIR="test/test_case"
+LOG_FILE="test/test_result/parse_test.log"
+MAX_ERROR_LINES=10
+
+# 检查编译器是否存在
+if [[ ! -x "$COMPILER" ]]; then
+    echo "❌ 错误：找不到编译器 $COMPILER"
+    echo "   请先执行 Lab1 构建命令："
+    echo "   cmake -S . -B build -DCMAKE_BUILD_TYPE=Release -DCOMPILER_PARSE_ONLY=ON"
+    echo "   cmake --build build -j \"\$(nproc)\""
+    exit 1
+fi
+
+# 创建日志目录（如果不存在）
+mkdir -p "$(dirname "$LOG_FILE")"
+> "$LOG_FILE"   # 清空日志
+
+echo "开始 Lab1 批量语法树测试..." | tee -a "$LOG_FILE"
+echo "测试目录：$TEST_DIR" | tee -a "$LOG_FILE"
+echo "编译器：$COMPILER" | tee -a "$LOG_FILE"
+echo "========================================" | tee -a "$LOG_FILE"
+
+pass=0
+fail=0
+total=0
+
+# 遍历所有 .sy 文件（支持子目录）
+while IFS= read -r -d '' sy_file; do
+    ((total++))
+    echo -n "[$total] 测试: $sy_file ... " | tee -a "$LOG_FILE"
+
+    # 执行解析（把输出丢到 /dev/null，防止刷屏）
+    if "$COMPILER" --emit-parse-tree "$sy_file" > /dev/null 2>&1; then
+        echo "✅PASS" | tee -a "$LOG_FILE"
+        ((pass++))
+    else
+        echo "FAIL" | tee -a "$LOG_FILE"
+        ((fail++))
+        
+        # 打印错误信息到日志（最后几行）
+        echo "   └── 错误详情（最后 $MAX_ERROR_LINES 行）:" >> "$LOG_FILE"
+        "$COMPILER" --emit-parse-tree "$sy_file" 2>&1 | tail -n "$MAX_ERROR_LINES" >> "$LOG_FILE"
+        echo "" >> "$LOG_FILE"
+    fi
+done < <(find "$TEST_DIR" -name "*.sy" -print0 | sort -z)
+
+# ================== 总结 ==================
+echo "========================================" | tee -a "$LOG_FILE"
+echo "测试完成！" | tee -a "$LOG_FILE"
+echo "总文件数 : $total" | tee -a "$LOG_FILE"
+echo "通过   : $pass" | tee -a "$LOG_FILE"
+echo "失败   : $fail" | tee -a "$LOG_FILE"
+
+if [[ $fail -eq 0 ]]; then
+    echo "恭喜！Lab1 语法树构建全部通过！可以进入 Lab2 啦～" | tee -a "$LOG_FILE"
+else
+    echo "有 $fail 个文件解析失败，请检查 SysY.g4 或报错日志" | tee -a "$LOG_FILE"
+    echo "   日志文件：$LOG_FILE" | tee -a "$LOG_FILE"
+fi
+
+echo "========================================" | tee -a "$LOG_FILE"