文件监视开发技术小结

20 篇文章 1 订阅

最近在用java写监视文件系统的东东,特对C++和Java下的不同实现方法做一小结。

1.Java环境下

很多人都说用文件轮询HashTable,然后如何如何比较,这种方法效率极为低下,还会造成无谓的磁盘读写。好了JDK 7中提供了java.nio.file大家可以通过 WatchService 来实现对文件的事件的监听。千万记得在JDK 7下哈,现在的JDK7的预览版发布到 jdk-7-ea-bin-b84-windows-i586-18_feb_2010.exe了。有兴趣的哥们可以把这个程序拿下去学习学习。

该段代码来自于 Sun(准确的该说是Oracle了) http://java.sun.com/docs/books/tutorial/essential/io/notification.html

/*
 * Copyright 2008-2009 Sun Microsystems, Inc.  All Rights Reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 *
 *   - Redistributions of source code must retain the above copyright
 *     notice, this list of conditions and the following disclaimer.
 *
 *   - Redistributions in binary form must reproduce the above copyright
 *     notice, this list of conditions and the following disclaimer in the
 *     documentation and/or other materials provided with the distribution.
 *
 *   - Neither the name of Sun Microsystems nor the names of its
 *     contributors may be used to endorse or promote products derived
 *     from this software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS
 * IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
 * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
 * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT OWNER OR
 * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
 * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
 * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
 * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
 * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
 * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 */

import java.nio.file.*;
import static java.nio.file.StandardWatchEventKind.*;
import static java.nio.file.LinkOption.*;
import java.nio.file.attribute.*;
import java.io.*;
import java.util.*;

/**
 * Example to watch a directory (or tree) for changes to files.
 */

public class WatchDir {

    private final WatchService watcher;
    private final Map<WatchKey,Path> keys;
    private final boolean recursive;
    private boolean trace = false;

    @SuppressWarnings("unchecked")
    static <T> WatchEvent<T> cast(WatchEvent<?> event) {
        return (WatchEvent<T>)event;
    }

    /**
     * Register the given directory with the WatchService
     */
    private void register(Path dir) throws IOException {
        WatchKey key = dir.register(watcher, ENTRY_CREATE, ENTRY_DELETE, ENTRY_MODIFY);
        if (trace) {
            FileRef prev = keys.get(key);
            if (prev == null) {
                System.out.format("register: %s/n", dir);
            } else {
                if (!dir.equals(prev)) {
                    System.out.format("update: %s -> %s/n", prev, dir);
                }
            }
        }
        keys.put(key, dir);
    }

    /**
     * Register the given directory, and all its sub-directories, with the
     * WatchService.
     */
    private void registerAll(final Path start) throws IOException {
        // register directory and sub-directories
        Files.walkFileTree(start, new SimpleFileVisitor<Path>() {
            @Override
            public FileVisitResult preVisitDirectory(Path dir) {
                try {
                    register(dir);
                } catch (IOException x) {
                    throw new IOError(x);
                }
                return FileVisitResult.CONTINUE;
            }
        });
    }

    /**
     * Creates a WatchService and registers the given directory
     */
    WatchDir(Path dir, boolean recursive) throws IOException {
        this.watcher = FileSystems.getDefault().newWatchService();
        this.keys = new HashMap<WatchKey,Path>();
        this.recursive = recursive;

        if (recursive) {
            System.out.format("Scanning %s .../n", dir);
            registerAll(dir);
            System.out.println("Done.");
        } else {
            register(dir);
        }

        // enable trace after initial registration
        this.trace = true;
    }

    /**
     * Process all events for keys queued to the watcher
     */
    void processEvents() {
        for (;;) {

            // wait for key to be signalled
            WatchKey key;
            try {
                key = watcher.take();
            } catch (InterruptedException x) {
                return;
            }

            Path dir = keys.get(key);
            if (dir == null) {
                System.err.println("WatchKey not recognized!!");
                continue;
            }

            for (WatchEvent<?> event: key.pollEvents()) {
                WatchEvent.Kind kind = event.kind();

                // TBD - provide example of how OVERFLOW event is handled
                if (kind == OVERFLOW) {
                    continue;
                }

                // Context for directory entry event is the file name of entry
                WatchEvent<Path> ev = cast(event);
                Path name = ev.context();
                Path child = dir.resolve(name);

                // print out event
                System.out.format("%s: %s/n", event.kind().name(), child);

                // if directory is created, and watching recursively, then
                // register it and its sub-directories
                if (recursive && (kind == ENTRY_CREATE)) {
                    try {
                        if (Attributes.readBasicFileAttributes(child, NOFOLLOW_LINKS).isDirectory()) {
                            registerAll(child);
                        }
                    } catch (IOException x) {
                        // ignore to keep sample readbale
                    }
                }
            }

            // reset key and remove from set if directory no longer accessible
            boolean valid = key.reset();
            if (!valid) {
                keys.remove(key);

                // all directories are inaccessible
                if (keys.isEmpty()) {
                    break;
                }
            }
        }
    }

    static void usage() {
        System.err.println("usage: java WatchDir [-r] dir");
        System.exit(-1);
    }

    public static void main(String[] args) throws IOException {
        // parse arguments
        if (args.length == 0 || args.length > 2)
            usage();
        boolean recursive = false;
        int dirArg = 0;
        if (args[0].equals("-r")) {
            if (args.length < 2)
                usage();
            recursive = true;
            dirArg++;
        }

        // register directory and process its events
        Path dir = Paths.get(args[dirArg]);
        new WatchDir(dir, recursive).processEvents();
    }
}
2.C++环境下

其实从操作系统层面能够监听到文件操作事件是最好的了。

在windows环境下呢可以通过用拷贝钩子实现对文件夹的监控。

可参考网页 http://ccrun.com/article.asp?i=601&d=0z20f4

ICopyHook是一个用于创建拷贝钩子处理程序COM接口,它决定一个文件夹或者打印机对象是否可以被移动,拷贝,重命名或删除。Shell在执行这些操作之前,会调用ICopyHook接口的CopyCallback方法对它们进行验证。CopyCallback返回一个int值指示Shell是否应该继续执行这个操作。返回值IDYES表示继续,而返回值IDNO和IDCANCEL则表示终止。

一个文件夹对象可以安装多个拷贝钩子处理程序。如果出现这种情况,Shell会依次调用每个处理程序。只有当每个处理程序都返回IDYES时,Shell才真正执行用户请求的操作。

拷贝钩子处理程序的作用是在上述四种操作执行前对它们进行验证,但是Shell并不会把操作的结果通知给拷贝钩子处理程序。而windows提供的API函数FindFirstChangeNotification和FindNextChangeNotification却可以实现这个功能。因此,只有把这种两种方法结合起来,才能对一个文件夹的状态进行完全的监控。

拷贝钩子处理程序实现并不困难,首先创建一个作为进程内组件的COM对象,它只需要暴露一个ICopyHook接口(当然还有IUnknown)。然后用regsrv32.exe注册这个COM组件。最后一步是向Shell注册你的这个拷贝钩子处理程序,方法是在注册表HKEY_CLASSES_ROOT/Directory/Shellex/CopyHookHandlers下创建一个名称任意的sub key,在此sub key中创建一个类型为REG_SZ的项并将你的COM对象的CLSID作为它的默认值就可以了。

下面就是一个拷贝钩子的实现程序(注:以下代码经老妖改动并添加了详细操作过程,在BCB6中成功编译并通过测试)

2.1. 从ICopyHook接口创建TCopyHook,从IClassFactory接口创建TClassFactory:

// TCopyHook.h
// TCopyHook类实现了ICopyHook接口,TClassFactory实现了IClassFactory接口
//---------------------------------------------------------------------------
#define NO_WIN32_LEAN_AND_MEAN
#include <shlobj.h>
//---------------------------------------------------------------------------
class TCopyHook: public ICopyHook
{
public:
    TCopyHook():m_refcnt(0) {}
    STDMETHODIMP QueryInterface(REFIID iid,void **ppvObject);
    STDMETHODIMP_(ULONG) AddRef();
    STDMETHODIMP_(ULONG) Release();
    STDMETHODIMP_(UINT) CopyCallback(HWND hwnd, UINT wFunc, UINT wFlags,
            LPCTSTR pszSrcFile, DWORD dwSrcAttribs,
            LPCTSTR pszDestFile, DWORD dwDestAttribs);
private:
    int m_refcnt;
};
//---------------------------------------------------------------------------
class TClassFactory : public IClassFactory
{
public:
    TClassFactory():m_refcnt(0) {}
    STDMETHODIMP QueryInterface(REFIID iid, void **ppvObject);
    STDMETHODIMP_(ULONG) AddRef();
    STDMETHODIMP_(ULONG) Release();
    STDMETHODIMP CreateInstance(IUnknown *pUnkOuter, REFIID riid, void **ppvObject);
    STDMETHODIMP LockServer(BOOL fLock);
private:
    int m_refcnt;
};


// TCopyHook.cpp
// TCopyHook对象和TClassFactory对象的实现文件
#include <stdio.h>
#include "TCopyHook.h"
//---------------------------------------------------------------------------
extern LONG nLocks;          // 对象计数,用于DllCanUnloadNow
ULONG __stdcall TCopyHook::AddRef()
{
    if(m_refcnt == 0)
        nLocks++;
    m_refcnt++;
    return m_refcnt;
}
//---------------------------------------------------------------------------
ULONG __stdcall TCopyHook::Release()
{
    int nNewCnt = --m_refcnt;
    if(nNewCnt <= 0)
    {
        nLocks--;
        delete this;
    }
    return nNewCnt;
}
//---------------------------------------------------------------------------
HRESULT __stdcall TCopyHook::QueryInterface(REFIID dwIID, void **ppvObject)
{
    if(dwIID == IID_IUnknown)
        *ppvObject = static_cast<IUnknown*>(this);
    else
        if(dwIID == IID_IShellCopyHook)
            *ppvObject = static_cast<ICopyHook*>(this);
        else
            return E_NOINTERFACE;
    reinterpret_cast<IUnknown*>(*ppvObject)->AddRef();
    return S_OK;
}
//---------------------------------------------------------------------------
// 这就是CopyCallback方法,拷贝钩子的所有功能由它实现。参数的具体值参看MSDN
UINT __stdcall TCopyHook::CopyCallback(HWND hwnd, UINT wFunc, UINT wFlags,
        LPCTSTR pszSrcFile, DWORD dwSrcAttribs,
        LPCTSTR pszDestFile, DWORD dwDestAttribs)
{
    char szMessage[MAX_PATH+14];
    sprintf(szMessage, "对%s进行的操作,是否继续?", pszSrcFile);
    return MessageBox(NULL, szMessage, "确认", MB_YESNO | MB_ICONEXCLAMATION);
}
//---------------------------------------------------------------------------
ULONG __stdcall TClassFactory::AddRef()
{
    if(m_refcnt==0)
        nLocks++;
    m_refcnt++;
    return m_refcnt;
}
//---------------------------------------------------------------------------
ULONG __stdcall TClassFactory::Release()
{

    int nNewCnt = --m_refcnt;

    if(nNewCnt <= 0)
    {
        nLocks--;
        delete this;
    }
    return nNewCnt;
}
//---------------------------------------------------------------------------
HRESULT __stdcall TClassFactory::QueryInterface(REFIID dwIID, void **ppvObject)
{
    if(dwIID == IID_IUnknown)
        *ppvObject = static_cast<IUnknown*>(this);
    else
        if(dwIID == IID_IClassFactory)
            *ppvObject = static_cast<IClassFactory*>(this);
        else
            return E_NOINTERFACE;
    reinterpret_cast<IUnknown*>(*ppvObject)->AddRef();
    return S_OK;
}
//---------------------------------------------------------------------------
HRESULT __stdcall TClassFactory::CreateInstance(IUnknown* pUnkownOuter,
        REFIID riid, void** ppvObj)
{
    if(pUnkownOuter != NULL)
        return CLASS_E_NOAGGREGATION;
    TCopyHook *pObj = new TCopyHook;
    pObj->AddRef();
    HRESULT hr = pObj->QueryInterface(riid, ppvObj);
    pObj->Release();
    return hr;
}
//---------------------------------------------------------------------------
HRESULT __stdcall TClassFactory::LockServer(BOOL fLock)
{
    if(fLock)
        nLocks++;
    else
        nLocks--;
    return S_OK;
}

2.2. 在BCB中New-->ActiveX-->ActiveX Library,然后添加相应代码。
   
   以下是修改后的Project1.cpp,大家可以直接copy过去。:
  
//$$---- axlib proj source ---- (stAXLibProjectSource)
#define NO_WIN32_LEAN_AND_MEAN
#include <vcl.h>
#pragma hdrstop
#include <atl/atlvcl.h>
#include <objbase.h>
#include <olectl.h>
#include "TCopyHook.h"

#pragma package(smart_init)
TComModule  Project1Module;
TComModule &_Module = Project1Module;

// 这是要添加到注册表中的项,注意如果你要使用这段代码,应该用UUIDGEN.exe生成一
// 个新的CLSID。
const char* szRegTable[][3]=
{
    {"CLSID//{7e10a039-fe03-4f9c-b7e1-c5eeeaf53735}", 0, "CopyHook"},
    {"CLSID//{7e10a039-fe03-4f9c-b7e1-c5eeeaf53735}//InProcServer32", 0, (const char*)-1},
    {"CLSID//{7e10a039-fe03-4f9c-b7e1-c5eeeaf53735}//InProcServer32", "ThreadingModel", "Apartment"},
    {"CLSID//{7e10a039-fe03-4f9c-b7e1-c5eeeaf53735}//ProgID", 0, "webber84.CopyHook.1"},
    {"webber84.CopyHook.1", 0, "CopyHook"},
    {"webber84.CopyHook.1//CLSID", 0, "{7e10a039-fe03-4f9c-b7e1-c5eeeaf53735}"}
};

HMODULE hInstance = NULL;
LONG nLocks = 0;
//---------------------------------------------------------------------------
// The ATL Object map holds an array of _ATL_OBJMAP_ENTRY structures that
// described the objects of your OLE server. The MAP is handed to your
// project's CComModule-derived _Module object via the Init method.
BEGIN_OBJECT_MAP(ObjectMap)
END_OBJECT_MAP()
//---------------------------------------------------------------------------
// Entry point of your Server invoked by Windows for processes or threads are
// initialized or terminated.
int WINAPI DllEntryPoint(HINSTANCE hinst, unsigned long reason, void*)
{
    if(reason == DLL_PROCESS_ATTACH)
        hInstance = (HMODULE)hinst;
    return TRUE;
}
//---------------------------------------------------------------------------
// _Module.Term is typically invoked from the DLL_PROCESS_DETACH of your
// DllEntryPoint. However, this may result in an incorrect shutdown sequence.
// Instead an Exit routine is setup to invoke the cleanup routine
// CComModule::Term.
void ModuleTerm(void)
{
    _Module.Term();
}
#pragma exit ModuleTerm 63
//---------------------------------------------------------------------------
// Entry point of your Server invoked to inquire whether the DLL is no
// longer in use and should be unloaded.
STDAPI __export DllCanUnloadNow(void)
{
    return nLocks == 0? S_OK: S_FALSE;
}
//---------------------------------------------------------------------------
// Entry point of your Server allowing OLE to retrieve a class object from
// your Server
STDAPI __export DllGetClassObject(REFCLSID rclsid, REFIID riid, LPVOID* ppv)
{
    HRESULT hr = E_OUTOFMEMORY;
    *ppv = NULL;
    TClassFactory *pClassFactory = new TClassFactory;
    if(pClassFactory != NULL)
        hr = pClassFactory->QueryInterface(riid, ppv);
    return hr;
}
//---------------------------------------------------------------------------
// Entry point of your Server invoked to instruct the server to create
// registry entries for all classes supported by the module
STDAPI __export DllRegisterServer(void)
{
    HRESULT hr = S_OK;
    int nItems = sizeof(szRegTable) / sizeof(szRegTable[0]);
    char szDllPath[MAX_PATH];
    GetModuleFileName(hInstance, szDllPath, MAX_PATH);
    for(int i=0; i<nItems && SUCCEEDED(hr); i++)
    {
        const char *szKeyName = szRegTable[i][0];
        const char *szValueName = szRegTable[i][1];
        const char *szValue = szRegTable[i][2];
        if(szValue == (const char*) - 1)
            szValue = szDllPath;
        HKEY hKey;
        LONG lReturn = RegCreateKey(HKEY_CLASSES_ROOT, szKeyName, &hKey);
        if(lReturn == ERROR_SUCCESS)
        {
            RegSetValueEx(hKey, szValueName, 0, REG_SZ,
                    (const BYTE*)szValue, strlen(szValue)+1);
            RegCloseKey(hKey);
        }
        if(lReturn != ERROR_SUCCESS)
        {
            hr = SELFREG_E_CLASS;
            DllUnregisterServer();
        }
    }
    return hr;
}
//---------------------------------------------------------------------------
// Entry point of your Server invoked to instruct the server to remove
// all registry entries created through DllRegisterServer.
STDAPI __export DllUnregisterServer(void)
{
    HRESULT hr = S_OK;
    LONG lReturn = 0;
    int nItems = sizeof(szRegTable) / sizeof(szRegTable[0]);
    for(int i=nItems-1; i>=0; i--)
    {
        const char *szKeyName = szRegTable[i][0];
        if((i == nItems-1) || stricmp(szRegTable[i+1][0], szKeyName) != 0)
            lReturn = RegDeleteKey(HKEY_CLASSES_ROOT, szKeyName);
        if(lReturn != ERROR_SUCCESS)
            hr = SELFREG_E_CLASS;
    }
    return hr;
}
//---------------------------------------------------------------------------

2.3. 在BCB的IDE环境中,选择菜单的Project-->Add to Project-->找到刚才创建的TCopyHook.cpp-->OK
   编译工程。如果没有错误,将生成Project1.dll。

2.4. 修改注册表:
   在HKEY_CLASSES_ROOT/Directory/shellex/CopyHookHandlers/下新建一个项,命名为Test,更改其默认值为{7e10a039-fe03-4f9c-b7e1-c5eeeaf53735},老妖注:这里的默认值应该和project1.cpp中的相同注册项目相同。
  
2.5. 注册COM组件:
   运行regsvr32.exe 路径/project.dll,点击确定后不要理会再弹出的错误窗口。重新启动计算机,试着copy一个文件夹,当paste的时候,效果出来了,嘿嘿。自己试试吧。

源代码下载:
http://www.ccrun.com/cctools/MyCopyHook.rar

在Linux环境下呢可以通过用inotify 监视文件系统的变化。

可参考网页 http://blog.csdn.net/rstevens/archive/2008/05/12/2438567.aspx

可以用 Linux 2.6 kenel 提供的 inotify 机制来监控文件系统的变化,例如增加文件,删除文件,修改文件等

下面是一个例子程序,可用来监控一个目录,如果在此目录下进行创建文件、删除文件、修改文件的操作,会有提示信息打印到屏幕上

#include <sys/inotify.h>
#include <stdio.h>
#include <errno.h>
#include <sys/types.h>
#include <unistd.h>
#include <sys/ioctl.h>

#define BUF_LEN   4096

int main(int argc, char **argv)
{

    if(argc < 2) {
        printf("Usage: %s [directory] ", argv[0]);
        return -1;
    }

    int fd;

    fd = inotify_init();

    if(fd == -1) {
        printf("failed in inotify_init, %d ", errno);
        return -1;
    }

    int wd1, wd2;   
ADD_AGAIN:

    wd1 = inotify_add_watch(fd, argv[1], IN_DELETE|IN_MODIFY|IN_CREATE|IN_IGNORED);
    if(wd1 == -1){
        perror("inotify_add_watch");
        return -1;
    } else
        printf("new watch fd %d ", wd1);
/**//*
    wd2 = inotify_add_watch(fd, "/etc", IN_ACCESS|IN_MODIFY);
    if(wd2 == -1) {
        perror("inotify_add_watch");
        return -1;
    }
*/

    char buf[BUF_LEN] __attribute__ ((aligned(4)));
    ssize_t len, i = 0;

    unsigned int queue_len;
    int ret;

while (1){
    i = 0;
   
    ret = ioctl(fd, FIONREAD, &queue_len);
    if(ret < 0) {
        perror("ioctl");
    }

    len = read(fd, buf, BUF_LEN);

    while(i < len) {
        struct inotify_event *event = (struct inotify_event*)&buf[i];
       
        printf("wd=%d mask=%d cookie=%d len=%d dir=%s ",
            event->wd, event->mask, event->cookie, event->len, (event->mask & IN_ISDIR) ? "yes" : "no");
       
        if(event->len)
            printf("name=%s ", event->name);
       
        if(event->mask & IN_IGNORED) {
            printf("ignored, add again ");
            if(0 != inotify_rm_watch(fd, event->wd))
                perror("inotify_rm_watch");
            goto ADD_AGAIN;
        }
           
        i += sizeof(struct inotify_event) + event->len;
    }
}

    return 0;
}


3.小结

无论哪种语言和方法,都是借助系统级的事件通知来完成对文件系统的监视的,性能上自然很强了。


本文来自CSDN博客,转载请标明出处:http://blog.csdn.net/cnhome/archive/2010/03/03/5342920.aspx

评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值