用MFC写一个爬取ts视频的爬虫

首先，需要了解一下MFC的基础知识和网络编程知识。

MFC是Microsoft Foundation Classes的缩写，是微软提供的一套C++类库，用于开发Windows应用程序。网络编程是指在计算机网络上进行数据交换和通信的过程，包括Socket编程、HTTP协议、TCP/IP协议等。

接下来，我们可以按照以下步骤来编写一个爬取ts视频的爬虫：

1. 创建一个MFC窗口应用程序项目。
2. 添加一个对话框，用于输入要爬取的视频链接。
3. 在对话框类中添加一个按钮，用于触发爬取视频的操作。
4. 在按钮的单击事件中，使用WinHTTP库发送HTTP请求获取视频链接的源代码。
5. 在源代码中解析出视频的ts分片链接，并将这些链接存储到一个数组中。
6. 使用WinHTTP库下载每个分片链接对应的ts文件。
7. 将所有下载的ts文件合并为一个完整的视频文件。

这里需要用到WinHTTP库中的以下函数：

- WinHttpOpen: 打开一个HTTP会话。
- WinHttpConnect: 连接到指定的服务器。
- WinHttpOpenRequest: 创建一个HTTP请求。
- WinHttpSendRequest: 发送HTTP请求。
- WinHttpReceiveResponse: 接收HTTP响应。
- WinHttpQueryHeaders: 获取HTTP响应头信息。
- WinHttpReadData: 读取HTTP响应正文数据。

具体的实现可以参考以下代码示例：

void CMyDlg::OnBtnCrawl()
{
    CString strUrl;
    GetDlgItemText(IDC_EDIT_URL, strUrl);

    // 打开HTTP会话
    HINTERNET hSession = WinHttpOpen(L"CrawlTSVideo", WINHTTP_ACCESS_TYPE_DEFAULT_PROXY, WINHTTP_NO_PROXY_NAME, WINHTTP_NO_PROXY_BYPASS, 0);
    if (!hSession) {
        return;
    }

    // 连接到服务器
    URL_COMPONENTS urlComp = {};
    urlComp.dwStructSize = sizeof(urlComp);
    urlComp.lpszHostName = new TCHAR[INTERNET_MAX_HOST_NAME_LENGTH];
    urlComp.dwHostNameLength = INTERNET_MAX_HOST_NAME_LENGTH;
    urlComp.lpszUrlPath = new TCHAR[INTERNET_MAX_PATH_LENGTH];
    urlComp.dwUrlPathLength = INTERNET_MAX_PATH_LENGTH;
    urlComp.lpszScheme = new TCHAR[INTERNET_MAX_SCHEME_LENGTH];
    urlComp.dwSchemeLength = INTERNET_MAX_SCHEME_LENGTH;
    if (!WinHttpCrackUrl(strUrl, strUrl.GetLength(), 0, &urlComp)) {
        WinHttpCloseHandle(hSession);
        return;
    }
    HINTERNET hConnect = WinHttpConnect(hSession, urlComp.lpszHostName, urlComp.nPort, 0);
    if (!hConnect) {
        delete[] urlComp.lpszHostName;
        delete[] urlComp.lpszUrlPath;
        delete[] urlComp.lpszScheme;
        WinHttpCloseHandle(hSession);
        return;
    }

    // 创建HTTP请求
    HINTERNET hRequest = WinHttpOpenRequest(hConnect, L"GET", urlComp.lpszUrlPath, nullptr, WINHTTP_NO_REFERER, WINHTTP_DEFAULT_ACCEPT_TYPES, urlComp.nScheme == INTERNET_SCHEME_HTTPS ? WINHTTP_FLAG_SECURE : 0);
    if (!hRequest) {
        delete[] urlComp.lpszHostName;
        delete[] urlComp.lpszUrlPath;
        delete[] urlComp.lpszScheme;
        WinHttpCloseHandle(hConnect);
        WinHttpCloseHandle(hSession);
        return;
    }

    // 发送HTTP请求
    if (!WinHttpSendRequest(hRequest, WINHTTP_NO_ADDITIONAL_HEADERS, 0, WINHTTP_NO_REQUEST_DATA, 0, 0, 0)) {
        delete[] urlComp.lpszHostName;
        delete[] urlComp.lpszUrlPath;
        delete[] urlComp.lpszScheme;
        WinHttpCloseHandle(hRequest);
        WinHttpCloseHandle(hConnect);
        WinHttpCloseHandle(hSession);
        return;
    }

    // 接收HTTP响应
    if (!WinHttpReceiveResponse(hRequest, nullptr)) {
        delete[] urlComp.lpszHostName;
        delete[] urlComp.lpszUrlPath;
        delete[] urlComp.lpszScheme;
        WinHttpCloseHandle(hRequest);
        WinHttpCloseHandle(hConnect);
        WinHttpCloseHandle(hSession);
        return;
    }

    // 获取HTTP响应头信息
    DWORD dwHeaderLen = 0;
    WinHttpQueryHeaders(hRequest, WINHTTP_QUERY_RAW_HEADERS_CRLF, WINHTTP_HEADER_NAME_BY_INDEX, nullptr, &dwHeaderLen, WINHTTP_NO_HEADER_INDEX);
    if (GetLastError() == ERROR_INSUFFICIENT_BUFFER) {
        TCHAR* lpHeaders = new TCHAR[dwHeaderLen / sizeof(TCHAR)];
        if (WinHttpQueryHeaders(hRequest, WINHTTP_QUERY_RAW_HEADERS_CRLF, WINHTTP_HEADER_NAME_BY_INDEX, lpHeaders, &dwHeaderLen, WINHTTP_NO_HEADER_INDEX)) {
            // 解析出视频的ts分片链接
            CStringArray arrUrls;
            LPCTSTR lpHeader = lpHeaders;
            while (*lpHeader) {
                if (_tcsnicmp(lpHeader, _T("http"), 4) == 0) {
                    LPCTSTR lpEnd = _tcschr(lpHeader, _T('\r'));
                    if (lpEnd) {
                        arrUrls.Add(CString(lpHeader, lpEnd - lpHeader));
                        lpHeader = lpEnd;
                        continue;
                    }
                }
                lpHeader = _tcschr(lpHeader, _T('\r'));
                if (lpHeader) {
                    ++lpHeader;
                }
            }

            // 下载每个分片链接对应的ts文件
            for (int i = 0; i < arrUrls.GetCount(); ++i) {
                CString strFileName;
                strFileName.Format(_T("video_%d.ts"), i);
                DownloadFile(hSession, arrUrls[i], strFileName);
            }

            // 合并所有下载的ts文件
            MergeFiles(_T("video_%d.ts"), arrUrls.GetCount(), _T("video.ts"));
        }
        delete[] lpHeaders;
    }

    delete[] urlComp.lpszHostName;
    delete[] urlComp.lpszUrlPath;
    delete[] urlComp.lpszScheme;
    WinHttpCloseHandle(hRequest);
    WinHttpCloseHandle(hConnect);
    WinHttpCloseHandle(hSession);
}

void CMyDlg::DownloadFile(HINTERNET hSession, LPCTSTR lpUrl, LPCTSTR lpFileName)
{
    // 连接到服务器
    URL_COMPONENTS urlComp = {};
    urlComp.dwStructSize = sizeof(urlComp);
    urlComp.lpszHostName = new TCHAR[INTERNET_MAX_HOST_NAME_LENGTH];
    urlComp.dwHostNameLength = INTERNET_MAX_HOST_NAME_LENGTH;
    urlComp.lpszUrlPath = new TCHAR[INTERNET_MAX_PATH_LENGTH];
    urlComp.dwUrlPathLength = INTERNET_MAX_PATH_LENGTH;
    urlComp.lpszScheme = new TCHAR[INTERNET_MAX_SCHEME_LENGTH];
    urlComp.dwSchemeLength = INTERNET_MAX_SCHEME_LENGTH;
    if (!WinHttpCrackUrl(lpUrl, _tcslen(lpUrl), 0, &urlComp)) {
        return;
    }
    HINTERNET hConnect = WinHttpConnect(hSession, urlComp.lpszHostName, urlComp.nPort, 0);
    if (!hConnect) {
        delete[] urlComp.lpszHostName;
        delete[] urlComp.lpszUrlPath;
        delete[] urlComp.lpszScheme;
        return;
    }

    // 创建HTTP请求
    HINTERNET hRequest = WinHttpOpenRequest(hConnect, L"GET", urlComp.lpszUrlPath, nullptr, WINHTTP_NO_REFERER, WINHTTP_DEFAULT_ACCEPT_TYPES, urlComp.nScheme == INTERNET_SCHEME_HTTPS ? WINHTTP_FLAG_SECURE : 0);
    if (!hRequest) {
        delete[] urlComp.lpszHostName;
        delete[] urlComp.lpszUrlPath;
        delete[] urlComp.lpszScheme;
        WinHttpCloseHandle(hConnect);
        return;
    }

    // 发送HTTP请求
    if (!WinHttpSendRequest(hRequest, WINHTTP_NO_ADDITIONAL_HEADERS, 0, WINHTTP_NO_REQUEST_DATA, 0, 0, 0)) {
        delete[] urlComp.lpszHostName;
        delete[] urlComp.lpszUrlPath;
        delete[] urlComp.lpszScheme;
        WinHttpCloseHandle(hRequest);
        WinHttpCloseHandle(hConnect);
        return;
    }

    // 接收HTTP响应
    if (!WinHttpReceiveResponse(hRequest, nullptr)) {
        delete[] urlComp.lpszHostName;
        delete[] urlComp.lpszUrlPath;
        delete[] urlComp.lpszScheme;
        WinHttpCloseHandle(hRequest);
        WinHttpCloseHandle(hConnect);
        return;
    }

    // 将HTTP响应正文数据保存为文件
    HANDLE hFile = CreateFile(lpFileName, GENERIC_WRITE, 0, nullptr, CREATE_ALWAYS, FILE_ATTRIBUTE_NORMAL, nullptr);
    if (hFile != INVALID_HANDLE_VALUE) {
        BYTE buf[4096];
        DWORD dwReadLen = 0;
        while (WinHttpReadData(hRequest, buf, sizeof(buf), &dwReadLen) && dwReadLen > 0) {
            DWORD dwWrittenLen = 0;
            WriteFile(hFile, buf, dwReadLen, &dwWrittenLen, nullptr);
        }
        CloseHandle(hFile);
    }

    delete[] urlComp.lpszHostName;
    delete[] urlComp.lpszUrlPath;
    delete[] urlComp.lpszScheme;
    WinHttpCloseHandle(hRequest);
    WinHttpCloseHandle(hConnect);
}

void CMyDlg::MergeFiles(LPCTSTR lpFileNameFmt, int nFileCount, LPCTSTR lpMergedFileName)
{
    HANDLE hMergedFile = CreateFile(lpMergedFileName, GENERIC_WRITE, 0, nullptr, CREATE_ALWAYS, FILE_ATTRIBUTE_NORMAL, nullptr);
    if (hMergedFile != INVALID_HANDLE_VALUE) {
        for (int i = 0; i < nFileCount; ++i) {
            CString strFileName;
            strFileName.Format(lpFileNameFmt, i);
            HANDLE hFile = CreateFile(strFileName, GENERIC_READ, FILE_SHARE_READ, nullptr, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, nullptr);
            if (hFile != INVALID_HANDLE_VALUE) {
                BYTE buf[4096];
                DWORD dwReadLen = 0;
                while (ReadFile(hFile, buf, sizeof(buf), &dwReadLen, nullptr) && dwReadLen > 0) {
                    DWORD dwWrittenLen = 0;
                    WriteFile(hMergedFile, buf, dwReadLen, &dwWrittenLen, nullptr);
                }
                CloseHandle(hFile);
            }
        }
        CloseHandle(hMergedFile);
    }

    for (int i = 0; i < nFileCount; ++i) {
        CString strFileName;
        strFileName.Format(lpFileNameFmt, i);
        DeleteFile(strFileName);
    }
}

上述代码中，DownloadFile函数用于下载一个ts文件，MergeFiles函数用于合并多个ts文件。

需要注意的是，这里只是一个简单的示例代码，并没有考虑到网络异常、文件存储路径等问题，实际应用中需要根据需求进行完善。