query_man.c

harvest是一个下载html网页得机器人

{
	if (qptr) {
		switch (qptr->type) {
		case LOGICAL:
		Debug (75, 4, ("node is LOGICAL; llist=%x  rlist=%x\n",
			qptr->llist, qptr->rlist));
			if (qptr->rlist)
				QM_free_qlist((qlist_t *) qptr->rlist);
			if (qptr->llist)
				QM_free_qlist((qlist_t *) qptr->llist);
			break;
		case FUNCTION:
		Debug (75, 4, ("node is FUNCTION; llist=%x  rlist=%x\n",
			qptr->llist, qptr->rlist));
			if (qptr->rlist)
				QM_free_args((arg_t *) qptr->rlist);
			if (qptr->llist)
				QM_free_qlist((qlist_t *) qptr->llist);
			break;
		case SELECT:
		Debug (75, 4, ("node is SELECT; llist=%x  rlist=%x\n",
			qptr->llist, qptr->rlist));
			if (qptr->rlist) {
				xfree(qptr->rlist);
				qptr->rlist = NULL;
			}
			if (qptr->llist) {
				xfree(qptr->llist);
				qptr->llist = NULL;
			}
			break;
		}
		xfree(qptr);
		qptr = NULL;
	}
	return SUCCESS;
}

#define BulkBeginStr "@DELETE{  }\n@REFRESH{  }\n@UPDATE{\n"
static int QM_bulk_count;

int QM_send_bulk_begin(sock)
int sock;
{
	Debug(75, 1, ("Starting a Broker-to-Broker transfer...\n"));
	SWRITE(sock, BULK_SUC, strlen(BULK_SUC));
	SWRITE(sock, BulkBeginStr, strlen(BulkBeginStr));
	QM_bulk_count = 0;
	return SUCCESS;
}

int QM_send_bulk_end(sock)
int sock;
{
	SWRITE(sock, "}\n", 2);	/* close the @UPDATE */
	SWRITE(sock, BULK_END, strlen(BULK_END));
	Log("Bulk transfer returned %d objects.\n", QM_bulk_count);
	Debug(75, 1, ("Finished Broker-to-Broker transfer...\n"));
	return SUCCESS;
}


/* ----------------------------------------------------------------- 
   QM_send_bulk_fd -- send a file to the socket ofile, based on FD
   ----------------------------------------------------------------- */
int QM_send_bulk_fd(FD, ofile, tmp)
fd_t FD;
FILE *ofile;
reg_t *tmp;	/* needed for logging */
{
	FILE *ifile;
	char buf[BUFSIZ];
	int n;

	if (tmp == NULL)
		return ERROR;

	LOGBULK_R(tmp);

	if ((ifile = SM_Read_Obj(FD)) == NULL) {
		return ERROR;
	}
	while ((n = fread(buf, 1, BUFSIZ, ifile)) > 0) {
		fwrite(buf, 1, n, ofile);
	}
	fclose(ifile);
	fflush(ofile);
	QM_bulk_count++;
	return SUCCESS;
}

/*
 *  QM_return_attributes - Selects and returns the desired attributes
 *  for FD to rsock socket.
 */
void QM_return_attributes(FD, rsock)
fd_t FD;
int rsock;
{
	FILE *fp;
	Template *T = NULL;
	AVPair *pair;
	int i;
	char buf[BUFSIZ];

	if (QM_nattlist <= 0)	/* any attributes to return? */
		return;

	/* Parse the SOIF */
	if ((fp = SM_Read_Obj(FD)) != NULL) {
		init_parse_template_file(fp);
		T = parse_template();
		finish_parse_template();
		fclose(fp);
	}
	
	/* Valid SOIF? */
	if (T == NULL)
		return;

	/* Now return the attributes */
	for (i = 0; i < QM_nattlist; i++) {
		if ((pair = extract_AVPair(T->list, QM_attlist[i])) != NULL) {
			sprintf(buf, "%03d - %s %ld\n",
				BR_ATTRIBUTE,
				QM_attlist[i],
				(long) pair->vsize);
			(void) write(rsock, buf, strlen(buf));
			(void) write(rsock, pair->value, pair->vsize);
		}
	}
	free_template(T);	/* clean up */
}

#ifdef QM_RET_EMBED_ATT
/*
 *  QM_return_embed_attrs - Selects and returns the desired attributes
 *  of the form embed<nn>-attr for matched lines for FD to rsock socket.
 *
 *  Experimental --DW
 */
void QM_return_embed_attrs(FD, rsock, num)
fd_t FD;
int rsock;
int num;
{
        FILE *fp;
        Template *T = NULL;
        AVPair *pair;
        int i;
        char buf[BUFSIZ];
        char embedattr[128];

        if (QM_nattlist <= 0)   /* any attributes to return? */
                return;

        /* Parse the SOIF */
        if ((fp = SM_Read_Obj(FD)) != NULL) {
                init_parse_template_file(fp);
                T = parse_template();
                finish_parse_template();
                fclose(fp);
        }

        /* Valid SOIF? */
        if (T == NULL)
                return;

        /* Now return the attributes */
        for (i = 0; i < QM_nattlist; i++) {
                sprintf (embedattr, "embed<%d>-%s", num, QM_attlist[i]);
                if ((pair = extract_AVPair(T->list, embedattr)) != NULL) {
                        sprintf(buf, "%03d - %s %ld\n",
				BR_ATTRIBUTE,
                                QM_attlist[i],
				(long) pair->vsize);
                        (void) write(rsock, buf, strlen(buf));
                        (void) write(rsock, pair->value, pair->vsize);
                }
        }
        free_template(T);       /* clean up */
}

/*
** QM_find_embed_attrs
**
** Searches for 'embed<num>-' on a matched line (Glimpse only).
** If found, passes 'num' to QM_return_embed_attrs()
*/
void QM_find_embed_attrs (FD, rsock, ml)
fd_t FD;
int rsock;
char *ml;
{
	int num;
	char jnk[64];

	if (sscanf (ml, "Matched line:  embed<%d>-", &num) == 1) {
		QM_return_embed_attrs(FD, rsock, num);
	} else
	if (sscanf (ml, "Matched line:  %[^#]# embed<%d>-", jnk, &num) == 2) {
		QM_return_embed_attrs(FD, rsock, num);
	}
}

#endif



/* ----------------------------------------------------------------- *
   QM_user_object -- send data about an object back to user.
   * ----------------------------------------------------------------- */
int QM_user_object(rsock, fd, opsize, opdata)
int rsock;
fd_t fd;
int opsize;
char *opdata[];
{
	reg_t *entry;
	int i;
	char buf[BUFSIZ], *s;
	int len;

	/* Be careful not to sprintf something too large! */

	if ((entry = RG_Get_Entry(fd)) == NULL)
		return FAIL;

	/* ----- Show URL ----- */
	sprintf(buf, "%03d - %s\n", BR_URL, entry->url);
	SWRITE(rsock, buf, strlen(buf));

	/* ----- Show Description ----- */
	if (QM_descflag == 1 && entry->desc != NULL && entry->descs > 0) {
		sprintf(buf, "%03d - %d\n", BR_DESCRIPTION, entry->descs);
		SWRITE(rsock, buf, strlen(buf));
		SWRITE(rsock, entry->desc, entry->descs);
	}
	/* ----- Pointer to Summary Object ----- */
	s = SM_Get_Obj_URL(fd);
	sprintf(buf, "%03d - %s\n", BR_SUMMARY_URL, s);
	xfree(s); s = NULL;
	SWRITE(rsock, buf, strlen(buf));

	/* ----- Opaque Data ----- */
	if (QM_opaqueflag == 1) { for (i = 0; i < opsize; i++) {
			if (opdata[i] && (len = strlen(opdata[i])) > 0) {
				if (len > (BUFSIZ-15))
					*(opdata[i]+BUFSIZ-15) = '\0';
				sprintf(buf, "%03d - %s\n",
					BR_OPAQUE_DATA,
					opdata[i]);
				SWRITE(rsock, buf, strlen(buf));
#ifdef QM_RET_EMBED_ATT
				if (QM_nattlist > 0)
				QM_find_embed_attrs (fd, rsock, opdata[i]);
#endif
			}
		}
	}
	/* ----- Attribute Requests ----- */
	QM_return_attributes(fd, rsock);

	/* ----- Object End ----- */
	sprintf (buf, "%d - object end\n", BR_END_OBJECT);
	SWRITE(rsock, buf, strlen(buf));

	LOGQUERY_R(entry);
	return SUCCESS;
}


/* ----------------------------------------------------------------- 
   QM_user_done -- send trailer information from user query.
   ----------------------------------------------------------------- */
int QM_user_done(rsock, count)
int rsock, count;
{
	char ret[BUFSIZ];
	extern char *BrkHomePage, *WebServer;

	QM_free_attributes();	/* do this first */
	if (BrkHomePage != NULL && WebServer != NULL) {
		sprintf(ret, "%03d - http://%s%s\n",
			BR_BROKER_URL, WebServer, BrkHomePage);
		SWRITE(rsock, ret, strlen(ret));
	}
	/* End the Query results with 103 */
	sprintf(ret, "%03d - %d Returned Object(s).\n",
		BR_END_RESULTS, count);
	Log("Query returned %d objects.\n", count);
	SWRITE(rsock, ret, strlen(ret));
	return SUCCESS;
}

/*
 *  QM_add_attribute() - Adds the given #attribute argument to the
 *  query.  Must normalize the attribute name.
 */
int QM_add_attribute(att)
char *att;
{
	QM_attlist[QM_nattlist] = xstrdup(att);
	(void)COL_Normalize_Name(QM_attlist[QM_nattlist]);
	Debug(75,5,("QM: Adding attribute: %s\n",QM_attlist[QM_nattlist]));
	QM_nattlist++;
	return SUCCESS;
}

int QM_free_attributes()
{
	int i;
	for (i = 0; i < QM_nattlist; i++) {
		if (QM_attlist[i] != NULL) {
			xfree(QM_attlist[i]);
			QM_attlist[i] = NULL;
		}
	}
	QM_nattlist = 0;
	return SUCCESS;
}

int QM_user_warning(rsock, message)
int rsock;
char *message;
{
	char ret[BUFSIZ];

	sprintf(ret, "%03d - %s\n", BR_USER_MSG, message);
	SWRITE(rsock, ret, strlen(ret));

	return SUCCESS;
}

int QM_send_bulk_err(sock)
int sock;
{
	Debug(75, 1, ("Error occurred in Broker-to-Broker transfer.\n"));
	SWRITE(sock, BULK_ERR, strlen(BULK_ERR));
	return SUCCESS;
}

static void admin_log(n,s)
int n;
char *s;
{
	char *p, *q, *w, *t, buf[BUFSIZ];

	/* remove password from broker.out */
	for (t = buf, w = s; *w; ) *t++ = tolower(*w++);
	if ((p = (char *) strstr(buf, "#password")) != NULL) {
		p += strlen("#password");
		q = strchr(p+1, '#');
		q = (q == NULL) ? t : q - 1;
		for (t = buf, w = s; *w; t++, w++) *t = (t < p || t > q) ? *w : 'X';
		*t = '\0';
		Log("Processing Admin Command %d: %s\n", n, buf);
	} else {
		Log("Processing Admin Command %d: %s\n", n, s);
	}
}