代码行统计

Python代码：

#encoding:utf-8

import os
import time
import traceback

totalFiles = 0
totalStatics = 0
totalError = 0
totalCount = 0
supportedTypes = (".c", ".cpp", ".h", ".cc", ".cxx", ".hpp")

def CountOneFile(filePath):
    fileCount = 0
    try:
        fin = open(filePath, "rt")
        content = fin.read()
        fileCount = content.count("\n")
        #print "[%5d] %s" %(fileCount, filePath)
    except:
        global totalError
        totalError += 1
        traceback.print_exc()
    return fileCount

def CountFiles(rootPath):
    if os.path.isfile(rootPath):
        global totalCount
        global totalFiles
        global totalStatics
        totalFiles += 1
        fileExten = os.path.splitext(rootPath)[1].lower()
        if fileExten in supportedTypes:
            totalStatics += 1
            totalCount += CountOneFile(rootPath)
    elif os.path.isdir(rootPath):
        childPaths = os.listdir(rootPath)
        for childPath in childPaths:
            CountFiles(os.path.join(rootPath, childPath))
    else:
        print "[error] %s" %rootPath

root = "D:\\Codes\\cmac_1225\\pub"

start = time.clock()
CountFiles(root)
end = time.clock()
print "Files: %d, Used: %d" %(totalFiles, totalStatics)
print "Lines: %d, Time: %d" %(totalCount, end-start)

C语言代码：

/* SourceCount.h */
#ifndef _SOURCE_COUNT_H
#define _SOURCE_COUNT_H

#ifndef NULL
#define NULL 0
#endif

#define MAX_CMD_LEN 2048
#define MAX_PATH_LEN 256
#define MAX_THREAD_NUM 20
#define DEFAULT_THREAD_NUM 2
#define EXTEN_NUM 6
#define READ_BUFF_SIZE 1024

#define CHECK_NULL_RET(var, ret) if(NULL == var) return ret;
#define PROC_THREAD_NUM(threadNum) ((threadNum > 0 && threadNum <= MAX_THREAD_NUM) ? threadNum : DEFAULT_THREAD_NUM)
#define RM_LINE_BREAK(path) if(strlen(path)>0 && path[strlen(path)-1] == '\n') path[strlen(path)-1] = '\0';
#define STR_LOWER(str) strlwr(str)

const char extensions[EXTEN_NUM][MAX_PATH_LEN] = {".c", ".cc", ".cxx", ".cpp", ".h", ".hpp"};

#endif

 

/* SourceCount.cpp */
#include <cstdlib>
#include <Windows.h>
#include <stdio.h>
#include <time.h>
#include "SourceCount.h"

HANDLE hMutex4File;
HANDLE hMutex4Stat;

FILE *pipe = NULL;
int totalCount = 0;

int CountFiles(const char *root, const int threadNum);
DWORD WINAPI ThreadProc(LPVOID lpParameter);
int CountOneFile(const char *filePath);
int CheckFileExten(const char *exten);

int main()
{
    char testRoot[] = "D:\\Codes\\cmac_1225";
    char testThreadNum = 12;

    clock_t start = clock();
    CountFiles(testRoot, testThreadNum);
    printf("Total: %d, time: %d\n", totalCount, clock()-start);
    system("pause");
    return 0;
}

int CountFiles(const char *root, const int threadNum)
{
    HANDLE pHandles[MAX_THREAD_NUM];
    char command[MAX_CMD_LEN];
    int handleNum;
    int loop;
    int ret;

    CHECK_NULL_RET(root, 0);
    
    sprintf(command, "dir \"%s\" /s /b", root);
    pipe = _popen(command, "rt"); // 执行命令，生成管道
    CHECK_NULL_RET(pipe, 0);

    // 创建互斥量
    hMutex4File = CreateMutex(NULL, false, NULL);
    hMutex4Stat = CreateMutex(NULL, false, NULL);

    handleNum = PROC_THREAD_NUM(threadNum); // 确定Thread数
    for(loop = 0; loop < handleNum; loop++)
    {
        pHandles[loop] = CreateThread(NULL, 0, ThreadProc, NULL, 0, NULL); // 创建线程
    }
    WaitForMultipleObjects(handleNum, pHandles, true, INFINITE); // 等待所有线程执行完毕
    for(loop = 0; loop < handleNum; loop++)
    {
        CloseHandle(pHandles[loop]); // 释放HANDLE资源
    }
    _pclose(pipe); // 关闭管道
    return totalCount;
}

/// 进程函数
DWORD WINAPI ThreadProc(LPVOID lpParameter)
{
    char filePath[MAX_PATH_LEN];
    char fileExten[MAX_PATH_LEN];
    int ret;
    int lines;

    while(true)
    {
        WaitForSingleObject(hMutex4File, INFINITE); // 获取pipe互斥权限
        if(feof(pipe))
        {
            ReleaseMutex(hMutex4File);
            return 0;
        }
        fgets(filePath, MAX_PATH_LEN, pipe);
        ReleaseMutex(hMutex4File);

        RM_LINE_BREAK(filePath);
        _splitpath(filePath, NULL, NULL, NULL, fileExten);
        STR_LOWER(fileExten);
        ret = CheckFileExten(fileExten);
        if(ret == 1)
        {
            lines = CountOneFile(filePath);
            WaitForSingleObject(hMutex4Stat, INFINITE); // 获取stat互斥权限
            totalCount += lines;
            ReleaseMutex(hMutex4Stat);
        }
    }
    return 0;
}

int CountOneFile(const char *filePath)
{
    FILE *fp = fopen(filePath, "rt");
    int lineCount = 0;
    int readBytes;
    int loop;
    unsigned char buff[READ_BUFF_SIZE];

    CHECK_NULL_RET(fp, 0);
    while(!feof(fp)) 
    {
        readBytes = fread(buff, 1, READ_BUFF_SIZE, fp);
        for(loop=0; loop<readBytes; loop++)
        {
            if(buff[loop] == 0x0a)
            {
                lineCount++;
            }
        }
    }
    fclose(fp);
    return lineCount;
}

int CheckFileExten(const char *exten)
{
    int loop;
    for(loop=0; loop<EXTEN_NUM; loop++)
    {
        if(strcmp(exten, extensions[loop]) == 0)
        {
            return 1;
        }
    }
    return 0;
}

 测试：

文件总数：21984

大小：991MB

满足条件文件总数：20274

代码总行数：6575561

Python用时：949s

C语言：

 4 线程，读取缓冲区 1024，用时80s
 6 线程，读取缓冲区 1024，用时66.7s
 8 线程，读取缓冲区 1024，用时53.5s
10 线程，读取缓冲区 1024，用时41.8s
10 线程，读取缓冲区 2048，用时16s
10 线程，读取缓冲区 4096，用时16s
12 线程，读取缓冲区 2048，用时14.9s

最佳：12线程，缓冲区2048

 

参数处理：

void PrintHelpInfo()
{
printf("代码行统计工具\n");
printf("SourceCount [drive:][path][filename] [ ...] [/E extension [ ...]] [/T threadnum]");
printf(" [/S buffsize] [/P [filename]]\n");
printf("\n");
printf(" [drive:][path][filename]\n");
printf(" 指定要统计的路径和/或文件。\n");
printf("\n");
printf(" /E extension 指定文件类型。\n");
printf(" /T threadnum 指定使用的线程数量。\n");
printf(" /S buffsize 指定文件读取缓冲区的大小。\n");
printf(" /P [filename] 输出文件统计结果[到文件]。\n");
printf("\n");
system("pause");
}