main.c

harvest是一个下载html网页得机器人

{
	int status, cpid;
	if ((cpid = waitpid(-1, &status, WNOHANG)) > 0) {
		Debug(77,1,("Child process exited: pid %d\n", cpid));
	}
#ifdef _HARVEST_SYSV_
	(void) signal(sig, sigreap);	/* reset signal handler */
#endif
}


/*===============================================================*/

/*
 * verify_path() - Verify that the absolute path is readable; if not, exit
 */
int verify_path(file)
char *file;
{
	if ((file[0] == '/') && (access(file, R_OK) < 0)) {
		Log("ERROR: %s is not readable.  Change your broker.conf file!\n", file);
		return ERROR;
	}
	return SUCCESS;
}

/*
 * verify_exe() - Verify that the absolute path is executable; if not, exit
 */
int verify_exe(file)
char *file;
{
	if ((file[0] == '/') && (access(file, X_OK) < 0)) {
		Log("ERROR: %s is not executable.  Change your broker.conf file!\n", file);
		return ERROR;
	}
	return SUCCESS;
}

/*===============================================================*/

/* Get Host Name */
int Initialize_Name()
{
	char *p = getfullhostname();

	if (p == NULL) {
		Log("WARNING: Internal Error: getfullhostname() failed...\n");
		return ERROR;
	}
	HName = xstrdup(p);
	Debug(77,1,("Broker Host is %s.\n", HName));
	return SUCCESS;
}

/* Initialize communication */
int Initialize_Com()
{
	static struct sockaddr_in brkr;
	int one = 1;

	if ((qsock = socket(AF_INET, SOCK_STREAM, 0)) < 0) {
		log_errno("socket");
		fatal("socket failed");
	}
    	if ((setsockopt(qsock, SOL_SOCKET, SO_REUSEADDR, (char*) &one,
	     sizeof(one))) < 0) {
		log_errno("setsockopt (SO_REUSEADDR = on)");
		fatal("setsockopt failed");
	}

	/* name socket using wildcards */
	brkr.sin_family = AF_INET;
	brkr.sin_addr.s_addr = htonl(INADDR_ANY);
	brkr.sin_port = htons(qport);
	if (bind(qsock, (struct sockaddr *) &brkr, sizeof brkr) < 0) {
		log_errno("bind");
		(void)close(qsock);
		fatal("Cannot not bind to port %d.  Exiting...\n", qport);
	}

	return SUCCESS;
}

/*
 *  Change global variables
 */
int set_str(tag, value)
char *tag;
char *value;
{
	unsigned int logf;
	int Set_Next_Collection();

	/* These are all paths that need to be verified */
	if (strcasecmp(tag, S_BRKDIR) == 0) {
		/* Assign Value */
		DIRpath = xstrdup(value);
		if (verify_path(DIRpath) == ERROR)
			return ERROR;
	} else if (strcasecmp(tag, S_BRKOFFLINE) == 0) {
		if (value != NULL && !strcasecmp(value, "yes"))
			broker_offline = 1;
	} else if (strcasecmp(tag, S_BRKHP) == 0) {
		/* Assign Value */
		BrkHomePage = xstrdup(value);
	} else if (strcasecmp(tag, S_COLCONF) == 0) {
		ColConfig = xstrdup(value);
		if (verify_path(ColConfig) == ERROR)
			return ERROR;
	} else if (strcasecmp(tag, S_GATHER) == 0) {
		Gather = xstrdup(value);
		if (verify_exe(Gather) == ERROR)
			return ERROR;
		/* The rest is not paths that need to be verified */
	} else if (strcasecmp(tag, S_WEBS) == 0) {
		WebServer = xstrdup(value);
	} else if (strcasecmp(tag, S_WEBPATH) == 0) {
		WebPath = xstrdup(value);
	} else if (strcasecmp(tag, S_PASSWD) == 0) {
		passwd = xstrdup(value);
	} else if (strcasecmp(tag, S_APROC) == 0) {
		aproc = xstrdup(value);
	} else if (strcasecmp(tag, S_DESC) == 0) {
		if (obj_desc != NULL)
			xfree(obj_desc);
		obj_desc = xstrdup(value);
		(void)COL_Normalize_Name(obj_desc);
		obj_desc_s = strlen(obj_desc);
	} else if (strcasecmp(tag, S_CLEANR) == 0) {
		if (sscanf(value, "%d", &clean_rate) < 1)
			errorlog("Could not convert clean rate in config. \n");
		if (clean_rate < 0) {
			errorlog("Invalid clean rate %d. Set to %d.\n", clean_rate, CLEAN_RATE);
			clean_rate = CLEAN_RATE;
		}
	} else if (strcasecmp(tag, S_COLLR) == 0) {
		if (sscanf(value, "%d", &collect_rate) < 1)
			errorlog("Could not convert collection rate in config. \n");
		if (collect_rate < 0) {
			errorlog("Invalid collection rate %d. Set to %d.\n", collect_rate, COLLECT_RATE);
			collect_rate = COLLECT_RATE;
		}
	} else if (strcasecmp(tag, S_RFR) == 0) {
		if (sscanf(value, "%d", &drefresh_rate) < 1)
			errorlog("Could not convert refresh rate in config. \n");
		if (drefresh_rate <= 0) {
			errorlog("Invalid refresh rate %d. Setting to %d.\n", drefresh_rate, REFRESH_RATE);
			drefresh_rate = REFRESH_RATE;
		}
	} else if (strcasecmp(tag, S_PORT) == 0) {
		if (sscanf(value, "%d", &qport) < 1)
			errorlog("Could not convert port in config. \n");
		if (qport < 1) {
			errorlog("Invalid port number %d. Set to %d.\n", qport, Q_PORT);
			qport = Q_PORT;
		}
	} else if (strcasecmp(tag, S_DEDLIM) == 0) {
		if (sscanf(value, "%d", &reg_limit) < 1)
			errorlog("Could not convert dead entry limit in config. \n");
		if (reg_limit < 0) {
			errorlog("Invalid dead entry limit %d. Set to %d.\n", reg_limit, MAX_DEAD);
			reg_limit = MAX_DEAD;
		}
	} else if (strcasecmp(tag, S_MAXEV) == 0) {
		if (sscanf(value, "%d", &max_events) < 1)
			errorlog("Could not convert event queue limit in config. \n");
		if (max_events < 1) {
			errorlog("Invalid event queue limit %d. Set to %d.\n", max_events, MAX_EVENTS);
			max_events = MAX_EVENTS;
		}
	} else if (strcasecmp(tag, S_LOGK) == 0) {
		if (sscanf(value, "%i", &logf) < 1)
			errorlog("Could not convert log key in config. \n");
		LOG_turn_on(logf);
		LogFlag = 1;
	} else if (strcasecmp(tag, S_FCOLL) == 0) {
		Tstr = xstrdup(value);
		Set_Next_Collection(Tstr);
	} else if (strcasecmp(tag, S_INDTP) == 0) {
		if (strcasecmp(value, FULLI) == 0)
			IndexType = I_FULL;
		else if (strcasecmp(value, INCRI) == 0)
			IndexType = I_INCR;
		else if (strcasecmp(value, PEROBJI) == 0)
			IndexType = I_PER_OBJ;
	} else if (strcasecmp(tag, S_INDEXER) == 0) {
		IndexerType = xstrdup(value);
		return (Indexer_Init());
	} else if (strcasecmp(tag, S_TERSELOG) == 0) {
		if (value != NULL && strcasecmp(value, "on") == 0)
			Log_Terse = 1;
		else
			Log_Terse = 0;
	} else if (strcasecmp(tag, S_FASTSTART) == 0) {
		if (value != NULL && strcasecmp(value, "on") == 0)
			do_fast_start = 1;
		else
			do_fast_start = 0;
	} else if (strcasecmp(tag, S_RDQTO) == 0) {
		if (value != NULL)
			ReadQueryTimeout = atoi(value);
		if (ReadQueryTimeout < 0)
			ReadQueryTimeout = READ_QUERY_TIMEOUT;
	} else {
		if (do_IND_config(value, tag) == ERROR)
			return ERROR;
	}

	return SUCCESS;
}

int Do_Collection()
{
	if (collect_flag == -1)
		Next_Collection = Next_Collection + collect_rate;
	if (collect_rate > 0)
		collect_flag = 0;
	return (COL_Do_Collection());
}

int Do_Compression()
{
	comflag = 0;
	return (RG_Compress());
}

int Do_Uniqify()
{
	comflag = 0;
	return (RG_Uniqify_by_URL());
}

int Do_Restart_Index_Server()
{
	IndexServer_ForceRestart = 1;
	return SUCCESS;
	/* The index server will be restarted after the next query */
}

int Do_Cleaning()
{
	RG_Cleaner();
	if (clean_rate > 0)
		Next_Clean = Cur_Time + clean_rate;
	return SUCCESS;
}

/* Configure the Broker according to a given config file */
int Config_Broker(cfile)
char *cfile;
{
	FILE *cfp;
	char dspace[BUFSIZ], tag[BUFSIZ], value[BUFSIZ];
	int llen, cc;
	char *vptr, *eptr;

	Debug(77,1,("Configuring Broker from the file: %s\n", cfile));

	/* assign defaults to non-critical globals */
	Gather = xstrdup("gather");
	obj_desc = xstrdup(OBJ_DESC);
	obj_desc_s = strlen(obj_desc);
	collect_rate = COLLECT_RATE;
	clean_rate = CLEAN_RATE;
	qport = Q_PORT;
	drefresh_rate = REFRESH_RATE;
	reg_limit = MAX_DEAD;
	max_events = MAX_EVENTS;
	WebPath = NULL;
	ColConfig = NULL;

	/* open config file */
	if ((cfp = fopen(cfile, "r")) == NULL) {
		Log("WARNING: Cannot read the Broker configuration file: %s\n", cfile);
		log_errno(cfile);
		return ERROR;
	}
	while (fgets(dspace, BUFSIZ, cfp) != NULL) {
		/* Parse config line */
		if (dspace[0] == '#') {
		        continue;
		} else if (sscanf(dspace, "%s %[^\n]", tag, value) < 2) {
			continue;
		} else {
			char *p = value+strlen(value)-1;
			while (isspace(*p)) *p-- = '\0';
			if (set_str(tag, value) == ERROR) {
                        	fclose(cfp);
                        	return ERROR;
                        }
                }
		memset(tag, 0, BUFSIZ);
		memset(value, 0, BUFSIZ);
		memset(dspace, 0, BUFSIZ);
	}

	fclose(cfp);

	/* if logging has not been set, turn on all logging. */
	if (LogFlag == 0)
		LOG_turn_on(~00);

	if (DIRpath == NULL) {
		Log("WARNING: Configuration file %s does not set the critical value: DIRpath.\n", cfile);
		return ERROR;
	} else if (WebServer == NULL) {
		Log("WARNING: Configuration file %s does not set the critical value: WebServer.\n", cfile);
		return ERROR;
	}
	return SUCCESS;
}


int Set_Next_Collection(mytime)
char *mytime;
{
	char *tt;
	int hr, min;
	struct tm *tmptr;
	time_t tmp, addt = 0;

	if ((tt = strchr(mytime, ':')) == NULL) {
		Log("WARNING: Bad Time Format for Next Collection.\n");
		Next_Collection = 0;
		return ERROR;
	}
	hr = atoi(mytime);
	min = atoi(tt + 1);

	tzset();
	tmp = UTIL_Get_Time();
	tmptr = localtime(&tmp);
	if (tmptr->tm_hour > hr)
		addt = 86400;	/* 24 hrs in sec. */
	else
		addt = 0;

	tmptr->tm_hour = hr;
	tmptr->tm_min = min;

#ifdef HAVE_MKTIME
	Next_Collection = mktime(tmptr) + addt;
#else
	Next_Collection = timelocal(tmptr) + addt;
#endif

	return SUCCESS;
}

static int Initialize_Broker(deffile)
char *deffile;
{

	Log("Using the configuration file: %s\n", deffile);
	harvest_add_broker_path();

	/* Randomize */
#if defined(HAVE_SRAND48)
	srand48(time(NULL));
#elif defined(HAVE_SRANDOM)
	(void) srandom(time(NULL));
#else
	srand(time(NULL));
#endif

	LOG_reset_master();
	(void)UTIL_Get_Time();	/* sets Cur_Time */

	if ((Initialize_Name() == ERROR) ||	/* sets current hostname */
	    (Indexer_Init() == ERROR) ||	/* indexer: before config */
	    (Config_Broker(deffile) == ERROR) ||/* must be first */
	    (LOG_Init() == ERROR) ||		/* log file */
	    (SM_Init() == ERROR) ||		/* storage manager */
	    (RG_Init() == ERROR) ||		/* registry depends on SM */
	    (EV_Init() == ERROR) ||		/* event list */
	    (Initialize_Com() == ERROR))	/* do port init last */
		return ERROR;

	/* do some misc initialization below */

	/* WebServer is non-NULL at this point */
	brk_obj_url = (char *) xmalloc(strlen(WebServer) + 100);
	if (WebPath) {
		sprintf(brk_obj_url, "http://%s%s/objects", WebServer, WebPath);
	} else {
		sprintf(brk_obj_url, "http://%s/BrkDir/objects", WebServer);
	}

	if (ColConfig == NULL)
		ColConfig = UTIL_make_admin_filename("Collection.conf");

	if (collect_rate == 0)
		collect_flag = 2;

	if (clean_rate == 0)
		Next_Clean = -1;

	AD_run_admin_process();
	return SUCCESS;
}

/*
 *  Initialize_Indexer_Routines - inits the run-time selectable indexing
 *  routines so that a user may switch indexers.
 */
static int Indexer_Init()
{
	int i;

	if (MAX_INDEXER_TYPES <= 0) {
		Log("ERROR: No Indexers defined for this broker!\n");
		return ERROR;
	}

	if (IndexerType != NULL) {
		if (!Valid_Indexer_Type(IndexerType)) {
			Log("ERROR: Invalid IndexerType: %s\n", IndexerType);
			return ERROR;
		}
	} else {
		/* default to first indexer in list, not necessarily Glimpse */
		IndexerType = xstrdup(Indexer_Routines[0].indexer_type);
	}
	for (i = 0; i < MAX_INDEXER_TYPES; i++) {
		if (!strcmp(Indexer_Routines[i].indexer_type, IndexerType))
			break;
	}
	if (i >= MAX_INDEXER_TYPES) {
		Log("ERROR: Invalid IndexerType: %s\n", IndexerType);
		return ERROR;
	}
	INDEXER = &Indexer_Routines[i];

	do_IND_initialize();
	return SUCCESS;
}

/*
 *  disconnect() - Disconnect the process from the controlling terminal.
 *  Adapted from Harvest's floodd daemon in the replicator.
 */
static void disconnect()
{
  	int pid, fd;

	/* ignore all job control signals */
#ifdef SIGTTOU
	signal (SIGTTOU, SIG_IGN);
#endif
#ifdef SIGTTIN
	signal (SIGTTIN, SIG_IGN);
#endif
#ifdef SIGTSTP
	signal (SIGTSTP, SIG_IGN);
#endif

#ifndef NO_FORK_FOR_DAEMON
	if ((pid = fork()) < 0) {
		log_errno("fork");
		return;
	}
	if (pid) {		/* parent */
		exit(0);	/* quietly exit to let child do the work */
	}
				/* child continues */
	(void) setsid();	/* Reset the session ID */
#endif

	/* Logging beneath here will not work */
	/* Close all file descriptors */
	close_all_fds(0);

  	/* Redirect the stdin to /dev/null */
  	if ((fd = open("/dev/null", O_RDONLY, 0)) < 0)
		exit(99);
  	(void) dup2(fd, 0);

  	/* Redirect the stdout and stderr to broker.out */
  	if ((fd = open("broker.out", O_WRONLY|O_CREAT|O_APPEND, 0640)) < 0)
		exit(99);

  	(void) dup2(fd, 1);
  	(void) dup2(fd, 2);
	/* stdout and stderr is now to broker.out for logging */
}