nudt-compiler-cpp/test2_instcount.sh

#!/usr/bin/env bash
# test2_instcount.sh - Lab2(IR生成)指令数量统计脚本

set -u
set -o pipefail

RED='\033[0;31m'
GREEN='\033[0;32m'
YELLOW='\033[1;33m'
BLUE='\033[0;34m'
NC='\033[0m'

TEST_ROOT="./test_merged"
OUTPUT_DIR="./lab2_instcount_results"
COMPILER="./build/bin/compiler"
VERIFY_SCRIPT="./scripts/verify_ir.sh"
MAX_CASES=0
STOP_ON_FIRST_FAILURE=false
START_FROM=1
KEEP_OLD=true
RUN_TESTS=false

# 指令数量统计相关变量
total_orig_inst=0
total_opt_inst=0
inst_cases_count=0
total_reduction_sum=0
reduction_count=0
total_increase_sum=0
increase_count=0

show_help() {
  cat << 'EOF'
用法: ./test2_instcount.sh [选项]

说明:
  批量执行 Lab2(IR 生成)指令数量统计。
  比较优化前后的IR指令数量变化。
  结果保存在 lab2_incount_results 文件夹中。

选项:
  -m, --max N           最多执行 N 个用例（0 表示不限制，默认: 0）
  -x, --stop-on-fail    遇到第一个失败时停止
  -s, --start-from N    从第 N 个测试用例开始（默认: 1）
  -k, --keep            保留旧的输出目录，不删除
  -r, --run             运行测试用例验证正确性（默认不运行）
  -h, --help            显示帮助

示例:
  ./test2_instcount.sh                    # 仅统计指令数量
  ./test2_instcount.sh -r                 # 统计指令数量并运行验证
  ./test2_instcount.sh --max 10           # 只运行前10个
  ./test2_instcount.sh --start-from 50    # 从第50个开始
EOF
}

while [[ $# -gt 0 ]]; do
  case "$1" in
    -m|--max)
      MAX_CASES="$2"
      shift 2
      ;;
    -x|--stop-on-fail)
      STOP_ON_FIRST_FAILURE=true
      shift
      ;;
    -s|--start-from)
      START_FROM="$2"
      shift 2
      ;;
    -k|--keep)
      KEEP_OLD=true
      shift
      ;;
    -r|--run)
      RUN_TESTS=true
      shift
      ;;
    -h|--help)
      show_help
      exit 0
      ;;
    *)
      echo -e "${RED}错误: 未知选项 $1${NC}"
      show_help
      exit 1
      ;;
  esac
done

if ! [[ "$START_FROM" =~ ^[0-9]+$ ]] || [[ "$START_FROM" -lt 1 ]]; then
  echo -e "${RED}错误: --start-from 需要正整数${NC}"
  exit 1
fi

if [[ ! -d "$TEST_ROOT" ]]; then
  echo -e "${RED}错误: 测试目录不存在: $TEST_ROOT${NC}"
  exit 1
fi

if [[ ! -x "$COMPILER" ]]; then
  echo -e "${RED}错误: 编译器不可执行: $COMPILER${NC}"
  echo -e "${YELLOW}提示: 请先构建: cmake -S . -B build -DCMAKE_BUILD_TYPE=Release && cmake --build build -j \"\$(nproc)\"${NC}"
  exit 1
fi

if [[ "$RUN_TESTS" == "true" && ! -x "$VERIFY_SCRIPT" ]]; then
  echo -e "${RED}错误: verify脚本不可执行: $VERIFY_SCRIPT${NC}"
  exit 1
fi

if ! [[ "$MAX_CASES" =~ ^[0-9]+$ ]]; then
  echo -e "${RED}错误: --max 需要非负整数${NC}"
  exit 1
fi

mkdir -p "$OUTPUT_DIR"

# 删除旧的输出目录（除非指定了 --keep）
if [[ "$KEEP_OLD" != "true" ]]; then
  if [[ -d "$OUTPUT_DIR" ]]; then
    echo "正在删除旧的输出目录..."
    rm -rf "$OUTPUT_DIR"/*
    echo "已删除旧的输出目录"
  fi
fi

mkdir -p "$OUTPUT_DIR"

LOG_FILE="$OUTPUT_DIR/lab2_instcount.log"
FAIL_FILE="$OUTPUT_DIR/failed_cases.txt"
ERROR_LOG_FILE="$OUTPUT_DIR/error_log.txt"
INST_REPORT_FILE="$OUTPUT_DIR/instruction_report.txt"
INST_SUMMARY_FILE="$OUTPUT_DIR/instruction_summary.txt"

: > "$LOG_FILE"
: > "$FAIL_FILE"
: > "$ERROR_LOG_FILE"
: > "$INST_SUMMARY_FILE"

# 初始化指令报告文件
{
  echo "================================================"
  echo "Lab2 IR指令数量统计报告"
  echo "================================================"
  echo "测试模式: IR"
  echo "运行验证: ${RUN_TESTS}"
  echo "生成时间: $(date)"
  echo "================================================"
  echo ""
  printf "%-60s | %12s\n" "测试用例" "IR指令数"
  echo "------------------------------------------------"
} > "$INST_REPORT_FILE"

echo "Lab2 指令数量统计日志 - $(date)" >> "$LOG_FILE"
echo "TEST_ROOT=$TEST_ROOT" >> "$LOG_FILE"
echo "OUTPUT_DIR=$OUTPUT_DIR" >> "$LOG_FILE"
echo "MAX_CASES=$MAX_CASES" >> "$LOG_FILE"
echo "RUN_TESTS=$RUN_TESTS" >> "$LOG_FILE"
echo "================================================" >> "$LOG_FILE"

# 统计IR指令数量的函数
count_ir_instructions() {
  local ir_file="$1"
  if [[ ! -f "$ir_file" ]]; then
    echo "0"
    return
  fi
  
  local count=0
  local term_count=0
  local store_count=0
  local call_count=0
  
  # 统计赋值指令（包含变量定义）
  count=$(grep -cE '^\s*%?[a-zA-Z0-9_]+\s*=' "$ir_file" 2>/dev/null || echo "0")
  count=$(echo "$count" | tr -d '[:space:]')
  [[ ! "$count" =~ ^[0-9]+$ ]] && count=0
  
  # 统计终止指令（ret, br, switch）
  term_count=$(grep -cE '^\s*(ret|br|switch)' "$ir_file" 2>/dev/null || echo "0")
  term_count=$(echo "$term_count" | tr -d '[:space:]')
  [[ ! "$term_count" =~ ^[0-9]+$ ]] && term_count=0
  
  # 统计store指令
  store_count=$(grep -cE '^\s*store ' "$ir_file" 2>/dev/null || echo "0")
  store_count=$(echo "$store_count" | tr -d '[:space:]')
  [[ ! "$store_count" =~ ^[0-9]+$ ]] && store_count=0
  
  # 统计call/invoke指令
  call_count=$(grep -cE '^\s*(call|invoke)' "$ir_file" 2>/dev/null || echo "0")
  call_count=$(echo "$call_count" | tr -d '[:space:]')
  [[ ! "$call_count" =~ ^[0-9]+$ ]] && call_count=0
  
  echo $((count + term_count + store_count + call_count))
}

# 运行单个测试用例
run_single_test() {
  local file="$1"
  local case_num="$2"
  
  local rel_path="${file#$TEST_ROOT/}"
  local filename="$(basename "$file")"
  local base_name="${filename%.sy}"
  local rel_dir="$(dirname "$rel_path")"
  local case_out_dir="$OUTPUT_DIR/$rel_dir"
  
  mkdir -p "$case_out_dir"
  
  # 删除已有的输出文件，确保重新生成
  rm -f "$case_out_dir/${base_name}.ll"
  rm -f "$case_out_dir/${base_name}.o"
  rm -f "$case_out_dir/${base_name}"
  rm -f "$case_out_dir/${base_name}.stdout"
  rm -f "$case_out_dir/${base_name}.actual.out"
  rm -f "$case_out_dir/${base_name}_time.txt"
  
  echo -ne "${YELLOW}[$case_num] $filename ... ${NC}"
  
  local ir_file="$case_out_dir/${base_name}.ll"
  
  # 生成IR文件
  set +e
  "$COMPILER" --emit-ir "$file" > "$ir_file" 2>/dev/null
  local ir_code=$?
  set -e
  
  if [[ $ir_code -ne 0 ]]; then
    echo -e "${RED}编译失败${NC}"
    echo "[FAILED] $file (compile error: ir=$ir_code)" >> "$LOG_FILE"
    FAILED=$((FAILED + 1))
    echo "$file" >> "$FAIL_FILE"
    
    # 写入详细错误信息
    echo "========================================" >> "$ERROR_LOG_FILE"
    echo "编译失败: $file" >> "$ERROR_LOG_FILE"
    echo "退出码: ir=$ir_code" >> "$ERROR_LOG_FILE"
    echo "时间: $(date)" >> "$ERROR_LOG_FILE"
    echo "========================================" >> "$ERROR_LOG_FILE"
    
    if [[ "$STOP_ON_FIRST_FAILURE" = true ]]; then
      echo -e "${RED}========================================${NC}"
      echo -e "${RED}在第一个失败处停止测试${NC}"
      echo -e "${RED}失败文件: $file${NC}"
      echo -e "${RED}日志: $LOG_FILE${NC}"
      echo -e "${RED}错误日志: $ERROR_LOG_FILE${NC}"
      echo -e "${RED}========================================${NC}"
      return 1
    fi
    return 0
  fi
  
  # 如果需要运行验证
  if [[ "$RUN_TESTS" == "true" ]]; then
    "$VERIFY_SCRIPT" "$file" "$case_out_dir" --run > /dev/null 2>&1
    local run_code=$?
    
    if [[ $run_code -ne 0 ]]; then
      echo -e "${RED}运行失败${NC}"
      echo "[FAILED] $file (runtime error, exit=$run_code)" >> "$LOG_FILE"
      FAILED=$((FAILED + 1))
      echo "$file" >> "$FAIL_FILE"
      
      echo "========================================" >> "$ERROR_LOG_FILE"
      echo "运行失败: $file" >> "$ERROR_LOG_FILE"
      echo "退出码: $run_code" >> "$ERROR_LOG_FILE"
      echo "时间: $(date)" >> "$ERROR_LOG_FILE"
      echo "========================================" >> "$ERROR_LOG_FILE"
      
      if [[ "$STOP_ON_FIRST_FAILURE" = true ]]; then
        return 1
      fi
      return 0
    fi
  fi
  
  # 统计指令数量
  local ir_count=0
  
  ir_count=$(count_ir_instructions "$ir_file")
  
  SUCCESS=$((SUCCESS + 1))
  
  # 累加统计
  total_orig_inst=$((total_orig_inst + ir_count))
  inst_cases_count=$((inst_cases_count + 1))
  
  # 显示结果
  echo -ne "${GREEN}成功${NC} "
  echo -ne "(指令数: ${ir_count})${NC}"
  
  # 写入报告
  printf "%-60s | %12d\n" "$rel_path" "$ir_count" >> "$INST_REPORT_FILE"
  echo "$rel_path: inst=$ir_count" >> "$INST_SUMMARY_FILE"
  echo ""
  
  echo "[SUCCESS] $file (inst=$ir_count)" >> "$LOG_FILE"
  
  return 0
}

# 主函数
main() {
  mapfile -d '' -t CASES < <(find "$TEST_ROOT" -type f -name '*.sy' -print0 | sort -z)
  
  TOTAL_FOUND=${#CASES[@]}
  if [[ $TOTAL_FOUND -eq 0 ]]; then
    echo -e "${YELLOW}未找到任何 .sy 用例，请检查目录: $TEST_ROOT${NC}"
    exit 0
  fi
  
  TOTAL=0
  SUCCESS=0
  FAILED=0
  
  echo -e "${BLUE}========================================${NC}"
  echo -e "${BLUE}Lab2 IR指令数量统计开始${NC}"
  echo -e "${BLUE}测试根目录: $TEST_ROOT${NC}"
  echo -e "${BLUE}找到用例数: $TOTAL_FOUND${NC}"
  echo -e "${BLUE}输出目录: $OUTPUT_DIR${NC}"
  echo -e "${BLUE}运行验证: $RUN_TESTS${NC}"
  if [[ "$KEEP_OLD" == "true" ]]; then
    echo -e "${BLUE}模式: 保留旧输出${NC}"
  else
    echo -e "${BLUE}模式: 全新输出${NC}"
  fi
  if [[ "$START_FROM" -gt 1 ]]; then
    echo -e "${BLUE}起始用例: $START_FROM${NC}"
  fi
  echo -e "${BLUE}========================================${NC}"
  
  for file in "${CASES[@]}"; do
    TOTAL=$((TOTAL + 1))
    
    # 跳过起始用例之前的测试
    if [[ $TOTAL -lt $START_FROM ]]; then
      continue
    fi
    
    # 检查是否已达到最大用例数限制
    if [[ $MAX_CASES -gt 0 && $((TOTAL - START_FROM + 1)) -gt $MAX_CASES ]]; then
      break
    fi
    
    set +e
    run_single_test "$file" "$TOTAL"
    test_result=$?
    set -e
    
    if [[ $test_result -eq 1 ]] && [[ "$STOP_ON_FIRST_FAILURE" = true ]]; then
      break
    fi
  done
  
  # 生成指令报告摘要
  {
    echo ""
    echo "================================================"
    echo "指令数量统计摘要"
    echo "================================================"
    echo "总指令数: $total_orig_inst"
    echo "测试用例数: $inst_cases_count"
    
    if [[ $inst_cases_count -gt 0 ]]; then
      local avg_inst=$((total_orig_inst / inst_cases_count))
      echo "平均指令数: ${avg_inst}"
    fi
    
    echo "================================================"
  } >> "$INST_REPORT_FILE"
  
  # 计算成功率
  RATE="0.00"
  if [[ $TOTAL -gt 0 ]]; then
    RATE=$(awk -v s="$SUCCESS" -v t="$TOTAL" 'BEGIN { printf "%.2f", (s*100.0)/t }')
  fi
  
  echo -e "${BLUE}========================================${NC}"
  echo -e "${BLUE}Lab2 指令数量统计完成${NC}"
  echo -e "${BLUE}执行用例: $TOTAL${NC}"
  echo -e "${GREEN}成功: $SUCCESS${NC}"
  echo -e "${RED}失败: $FAILED${NC}"
  echo -e "${BLUE}成功率: ${RATE}%${NC}"
  
  echo ""
  echo -e "${BLUE}指令数量统计:${NC}"
  echo "  总指令数: $total_orig_inst"
  echo "  测试用例数: $inst_cases_count"
  
  if [[ $inst_cases_count -gt 0 ]]; then
    local avg_inst=$((total_orig_inst / inst_cases_count))
    echo -e "  平均指令数: ${avg_inst}"
  fi
  
  echo ""
  echo -e "${BLUE}详细报告: $INST_REPORT_FILE${NC}"
  echo -e "${BLUE}指令汇总: $INST_SUMMARY_FILE${NC}"
  echo -e "${BLUE}日志: $LOG_FILE${NC}"
  if [[ $FAILED -gt 0 ]]; then
    echo -e "${RED}失败清单: $FAIL_FILE${NC}"
    echo -e "${RED}错误日志: $ERROR_LOG_FILE${NC}"
  fi
  echo -e "${BLUE}========================================${NC}"
  
  exit 0
}

main