ClamAV学习【3】——scanmanager函数浏览

吃饱饭继续浏览Manager.c的scanmanager函数，这个函数的功能吧，暂时理解如下。

接收一个命令行参数（经过处理的optstruct结构指针）。

然后根据选项判断文件类型种类，还有一些扫描选项。看到了AC-Only选项，可是没有BM-Only选项，不解。然后就进行了病毒库的加载和引擎的初始化，接着进行了一些扫描限制（扫描文件大小、数量、递归深度等）设定，还有扫描的文件类型（貌似上面的是用来构建引擎，现在的是用来设置扫描）。然后就是重点，针对文件名进行处理，判断是否是目录还是单个文件，调用不同的扫描函数。

代码注释如下：（应该不是完全正确的）

int scanmanager(const struct optstruct *opt)
{
	mode_t fmode;
	int ret = 0, extunpacker = 0, fmodeint, i, x;
	unsigned int options = 0, dboptions = 0;
	struct cl_engine *engine = NULL;
	struct cl_limits limits;
	struct passwd *user = NULL;
	struct stat sb;
	char *fullpath = NULL, cwd[1024];

	//扫描文件类型判断
    if(opt_check(opt, "unzip") || opt_check(opt, "unrar") || opt_check(opt, "arj") ||
       opt_check(opt, "unzoo") || opt_check(opt, "jar") || opt_check(opt, "lha") ||
       opt_check(opt, "tar") || opt_check(opt, "tgz") || opt_check(opt, "deb"))
	    extunpacker = 1;

	//获取CLAMAVUSER的用户信息，识别码和根目录
	//参考passwd结构
/* njh@bandsman.co.uk: BeOS */
#if !defined(C_CYGWIN) && !defined(C_OS2) && !defined(C_BEOS) && !defined(C_WINDOWS)
    if(extunpacker && !geteuid()) {
	if((user = getpwnam(CLAMAVUSER)) == NULL) {
	    logg("!Can't get information about user "CLAMAVUSER" (required to run external unpackers)\n");
	    exit(60); /* this is critical problem, so we just exit here */
	}
    }
#endif

	//添加扫描选项
    if(!opt_check(opt, "no-phishing-sigs"))
	dboptions |= CL_DB_PHISHING;

    if(!opt_check(opt,"no-phishing-scan-urls"))
	dboptions |= CL_DB_PHISHING_URLS;
    if(opt_check(opt,"phishing-ssl")) {
	options |= CL_SCAN_PHISHING_BLOCKSSL;
    }
    if(opt_check(opt,"phishing-cloak")) {
	options |= CL_SCAN_PHISHING_BLOCKCLOAK;
    }

    if(opt_check(opt, "dev-ac-only"))
	dboptions |= CL_DB_ACONLY;

	//这个选项需要学习下，深度搜查用AC?
    if(opt_check(opt, "dev-ac-depth"))
	cli_ac_setdepth(AC_DEFAULT_MIN_DEPTH, atoi(opt_arg(opt, "dev-ac-depth")));

    if(opt_check(opt, "detect-pua"))
	dboptions |= CL_DB_PUA;

	//传说中的病毒库加载
    if(opt_check(opt, "database")) {
	if((ret = cl_load(opt_arg(opt, "database"), &engine, &info.sigs, dboptions))) {
	    logg("!%s\n", cl_strerror(ret));
	    return 50;
	}

    } else {
	    char *dbdir = freshdbdir();

	//病毒库更新后重新加载
	if((ret = cl_load(dbdir, &engine, &info.sigs, dboptions))) {
	    logg("!%s\n", cl_strerror(ret));
	    free(dbdir);
	    return 50;
	}
	free(dbdir);
    }

    if(!engine) {
	logg("!Can't initialize the virus database\n");
	return 50;
    }

	//引擎构建
    if((ret = cl_build(engine)) != 0) {
	logg("!Database initialization error: %s\n", cl_strerror(ret));;
	return 50;
    }

    /* set limits */
	//开始设置限制，应该是扫描大小等
    memset(&limits, 0, sizeof(struct cl_limits));

	//最大的扫描大小
    if(opt_check(opt, "max-scansize")) {
	char *cpy, *ptr;
	ptr = opt_arg(opt, "max-scansize");
	if(tolower(ptr[strlen(ptr) - 1]) == 'm') {
	    cpy = calloc(strlen(ptr), 1);
	    strncpy(cpy, ptr, strlen(ptr) - 1);
	    cpy[strlen(ptr)-1]='\0';
	    limits.maxscansize = atoi(cpy) * 1024 * 1024;
	    free(cpy);
	} else
	    limits.maxscansize = atoi(ptr) * 1024;
    } else
	limits.maxscansize = 104857600;

	//最大文件大小
    if(opt_check(opt, "max-filesize")) {
	char *cpy, *ptr;
	ptr = opt_arg(opt, "max-filesize");
	if(tolower(ptr[strlen(ptr) - 1]) == 'm') {
	    cpy = calloc(strlen(ptr), 1);
	    strncpy(cpy, ptr, strlen(ptr) - 1);
	    cpy[strlen(ptr)-1]='\0';
	    limits.maxfilesize = atoi(cpy) * 1024 * 1024;
	    free(cpy);
	} else
	    limits.maxfilesize = atoi(ptr) * 1024;
    } else
	limits.maxfilesize = 26214400;

	//最大文件数
    if(opt_check(opt, "max-files"))
	limits.maxfiles = atoi(opt_arg(opt, "max-files"));
    else
        limits.maxfiles = 10000;

	//最大递归深度
    if(opt_check(opt, "max-recursion"))
        limits.maxreclevel = atoi(opt_arg(opt, "max-recursion"));
    else
        limits.maxreclevel = 16;

    /* set options */
	//设置选项，扫描文件类型相关
    if(opt_check(opt, "disable-archive") || opt_check(opt, "no-archive"))
	options &= ~CL_SCAN_ARCHIVE;
    else
	options |= CL_SCAN_ARCHIVE;

    if(opt_check(opt, "detect-broken"))
	options |= CL_SCAN_BLOCKBROKEN;

    if(opt_check(opt, "block-encrypted"))
	options |= CL_SCAN_BLOCKENCRYPTED;

    if(opt_check(opt, "no-pe"))
	options &= ~CL_SCAN_PE;
    else
	options |= CL_SCAN_PE;

    if(opt_check(opt, "no-elf"))
	options &= ~CL_SCAN_ELF;
    else
	options |= CL_SCAN_ELF;

    if(opt_check(opt, "no-ole2"))
	options &= ~CL_SCAN_OLE2;
    else
	options |= CL_SCAN_OLE2;

    if(opt_check(opt, "no-pdf"))
	options &= ~CL_SCAN_PDF;
    else
	options |= CL_SCAN_PDF;

    if(opt_check(opt, "no-html"))
	options &= ~CL_SCAN_HTML;
    else
	options |= CL_SCAN_HTML;

    if(opt_check(opt, "no-mail")) {
	options &= ~CL_SCAN_MAIL;
    } else {
	options |= CL_SCAN_MAIL;

	if(opt_check(opt, "mail-follow-urls"))
	    options |= CL_SCAN_MAILURL;
    }

    if(opt_check(opt, "no-algorithmic"))
	options &= ~CL_SCAN_ALGORITHMIC;
    else
	options |= CL_SCAN_ALGORITHMIC;

	//Linux下，需要获取设备号
	//具体参见文件属性stat结构
#ifdef C_LINUX
    procdev = (dev_t) 0;
    if(stat("/proc", &sb) != -1 && !sb.st_size)
	procdev = sb.st_dev;
#endif

    /* check filetype */
	//遇到扫描目录的情况
    if(opt->filename == NULL || strlen(opt->filename) == 0) {

	/* we need full path for some reasons (eg. archive handling) */
	//获取当前工作目录
	//获取到后调用目录扫描函数scandirs
	if(!getcwd(cwd, sizeof(cwd))) {
	    logg("!Can't get absolute pathname of current working directory\n");
	    ret = 57;
	} else
	    ret = scandirs(cwd, engine, user, opt, &limits, options);

    } else if(!strcmp(opt->filename, "-")) { /* read data from stdin */
	ret = scanstdin(engine, &limits, options);

    } else {
    //扫描文件的情况
	char *thefilename;
	for (x = 0; (thefilename = cli_strtok(opt->filename, x, "\t")) != NULL; x++) {
		//fileinfo函数在clamscan的others.c里定义
		//2代表获取文件权限
		//有访问权限后找出最右边/的个数
	    if((fmodeint = fileinfo(thefilename, 2)) == -1) {
		logg("^Can't access file %s\n", thefilename);
		perror(thefilename);
		ret = 56;
	    } else {
		int slash = 1;
		for(i = strlen(thefilename) - 1; i > 0 && slash; i--) {
		    if(thefilename[i] == '/')
			thefilename[i] = 0;
		    else
			slash = 0;
		}

		fmode = (mode_t) fmodeint;

		//下面依旧是处理文件名的
		if(extunpacker && (thefilename[0] != '/' && thefilename[0] != '\\' && thefilename[1] != ':')) {
		    /* we need to complete the path */
		    if(!getcwd(cwd, sizeof(cwd))) {
			logg("!Can't get absolute pathname of current working directory\n");
			return 57;
		    } else {
			fullpath = malloc(512);
#ifdef NO_SNPRINTF
			sprintf(fullpath, "%s/%s", cwd, thefilename);
#else
			snprintf(fullpath, 512, "%s/%s", cwd, thefilename);
#endif
			logg("*Full path: %s\n", fullpath);
		    }
		} else
		    fullpath = thefilename;

		//开始区别扫描文件和文件夹
		switch(fmode & S_IFMT) {
		    case S_IFREG:
			ret = scanfile(fullpath, engine, user, opt, &limits, options);
			break;

		    case S_IFDIR:
			ret = scandirs(fullpath, engine, user, opt, &limits, options);
			break;

		    default:
			logg("!Not supported file type (%s)\n", thefilename);
			ret = 52;
		}

		if(extunpacker && (thefilename[0] != '/' && thefilename[0] != '\\' && thefilename[1] != ':')) {
		    free(fullpath);
		    fullpath = NULL;
		}
	    }
	    free(thefilename);
	}
    }

    /* free the engine */
    cl_free(engine);

    /* overwrite return code */
    if(info.ifiles)
	ret = 1;
    else if(ret < 50) /* hopefully no error detected */
	ret = 0; /* just make sure it's 0 */

    return ret;
}

原文：　http://blog.csdn.net/betabin/article/details/7422236