static int open_socket(DPS_AGENT *A, char *unix_socket) {
char unix_path[128];
struct sockaddr_un unix_addr;
int sockfd, saddrlen;
if (DpsRelVarName(A->Conf, unix_path, sizeof(unix_path), unix_socket) < 105) {
} else {
DpsLog(A, DPS_LOG_ERROR, "Unix socket name '%s' is too large", unix_path);
return(DPS_NET_CANT_CONNECT);
}
if ((sockfd = socket(AF_UNIX, SOCK_STREAM, 0)) < 0) {
DpsLog(A, DPS_LOG_ERROR, "unix socket() error %d", errno);
return(DPS_NET_CANT_CONNECT);
}
DpsSockOpt(A, sockfd);
bzero((void*)&unix_addr, sizeof(unix_addr));
unix_addr.sun_family = AF_UNIX;
dps_strncpy(unix_addr.sun_path, unix_path, sizeof(unix_addr.sun_path));
saddrlen = sizeof(unix_addr.sun_family) + dps_strlen(unix_addr.sun_path);
if(connect(sockfd, (struct sockaddr *)&unix_addr, sizeof (unix_addr))) {
dps_strerror(A, DPS_LOG_ERROR, "unix socket '%s' connect() error", unix_path);
return(DPS_NET_CANT_CONNECT);
}
return sockfd;
}
void DpsSockOpt(DPS_AGENT *A, int dps_socket) {
const int lowat = 1;
struct timeval so_tval;
so_tval.tv_sec = 300;
so_tval.tv_usec = 0;
#if !defined(sgi) && !defined(__sgi) && !defined(__irix__) && !defined(sun) && !defined(__sun) /* && !defined(__FreeBSD__)*/
if (setsockopt(dps_socket, SOL_SOCKET, SO_SNDTIMEO, (char *)&so_tval, sizeof(so_tval)) != 0) {
if (A)
DpsLog(A, DPS_LOG_EXTRA, "%s [%d] setsockopt error: %d (%s)\n", __FILE__, __LINE__, errno, strerror(errno));
else
fprintf(stderr, "%s [%d] setsockopt error: %d (%s)\n", __FILE__, __LINE__, errno, strerror(errno));
}
#endif
#if defined(SO_SNDLOWAT) && !defined(__linux__)
if (setsockopt(dps_socket, SOL_SOCKET, SO_SNDLOWAT, (char *)&lowat, sizeof(lowat)) != 0) {
if (A)
DpsLog(A, DPS_LOG_EXTRA, "%s [%d] setsockopt error: %d (%s)\n", __FILE__, __LINE__, errno, strerror(errno));
else
fprintf(stderr, "%s [%d] setsockopt error: %d (%s)\n", __FILE__, __LINE__, errno, strerror(errno));
}
#endif
#if defined(SO_RCVLOWAT) && (!defined(__linux__) || LINUX_VERSION_CODE >= 0x20400)
if (setsockopt(dps_socket, SOL_SOCKET, SO_RCVLOWAT, (char *)&lowat, sizeof(lowat)) != 0) {
if (A)
DpsLog(A, DPS_LOG_EXTRA, "%s [%d] setsockopt error: %d (%s)\n", __FILE__, __LINE__, errno, strerror(errno));
else
fprintf(stderr, "%s [%d] setsockopt error: %d (%s)\n", __FILE__, __LINE__, errno, strerror(errno));
}
#endif
}
__C_LINK int __DPSCALL DpsCacheMakeIndexes(DPS_AGENT *Indexer, DPS_DB *db) {
DPS_VARLIST *v = &Indexer->Conf->Vars;
size_t i, r;
char *ind;
r = (size_t) 'l';
for (i = 0; i < v->Root[r].nvars; i++) {
if (!strncasecmp("Limit-", v->Root[r].Var[i].name, 6)) {
ind = v->Root[r].Var[i].val;
if (!strcasecmp(ind, "category")) {
/* To see the URL being indexed in "ps" output on xBSD */
dps_setproctitle("[%d] Category index creation", Indexer->handle);
DpsLog(Indexer, DPS_LOG_EXTRA, "Creating category index");
MakeNestedIndex(Indexer, "Category", DPS_LIMFNAME_CAT, DPS_IFIELD_TYPE_HEX8STR, db);
} else if (!strcasecmp(ind, "tag")) {
/* To see the URL being indexed in "ps" output on xBSD */
dps_setproctitle("[%d] Tag index creation", Indexer->handle);
DpsLog(Indexer, DPS_LOG_EXTRA, "Creating tag index");
MakeLinearIndex(Indexer, "Tag", DPS_LIMFNAME_TAG, DPS_IFIELD_TYPE_STRCRC32, db);
} else if (!strcasecmp(ind, "link")) {
/* To see the URL being indexed in "ps" output on xBSD */
dps_setproctitle("[%d] Link index creation", Indexer->handle);
DpsLog(Indexer, DPS_LOG_EXTRA, "Creating link index");
MakeLinearIndex(Indexer, "link", DPS_LIMFNAME_LINK, DPS_IFIELD_TYPE_INT, db);
} else if (!strcasecmp(ind, "time")) {
/* To see the URL being indexed in "ps" output on xBSD */
dps_setproctitle("[%d] Time index creation", Indexer->handle);
DpsLog(Indexer, DPS_LOG_EXTRA, "Creating time index");
MakeLinearIndex(Indexer, "last_mod_time", DPS_LIMFNAME_TIME, DPS_IFIELD_TYPE_HOUR, db);
} else if (!strcasecmp(ind, "hostname")) {
/* To see the URL being indexed in "ps" output on xBSD */
dps_setproctitle("[%d] Hostname index creation", Indexer->handle);
DpsLog(Indexer, DPS_LOG_EXTRA, "Creating hostname index");
MakeLinearIndex(Indexer, "url", DPS_LIMFNAME_HOST, DPS_IFIELD_TYPE_HOSTNAME, db);
} else if (!strcasecmp(ind, "language")) {
/* To see the URL being indexed in "ps" output on xBSD */
dps_setproctitle("[%d] Language index creation", Indexer->handle);
DpsLog(Indexer, DPS_LOG_EXTRA, "Creating language index");
MakeLinearIndex(Indexer, "Content-Language", DPS_LIMFNAME_LANG, DPS_IFIELD_TYPE_STR2CRC32, db);
} else if (!strcasecmp(ind, "content")) {
/* To see the URL being indexed in "ps" output on xBSD */
dps_setproctitle("[%d] Content-Type index creation", Indexer->handle);
DpsLog(Indexer, DPS_LOG_EXTRA, "Creating Content-Type index");
MakeLinearIndex(Indexer, "Content-Type", DPS_LIMFNAME_CTYPE, DPS_IFIELD_TYPE_STRCRC32, db);
} else if (!strcasecmp(ind, "siteid")) {
/* To see the URL being indexed in "ps" output on xBSD */
dps_setproctitle("[%d] Site_id index creation", Indexer->handle);
DpsLog(Indexer, DPS_LOG_EXTRA, "Creating Site_id index");
MakeLinearIndex(Indexer, "site_id", DPS_LIMFNAME_SITE, DPS_IFIELD_TYPE_INT, db);
}
/* To see the URL being indexed in "ps" output on xBSD */
dps_setproctitle("[%d] Indexes done.", Indexer->handle);
DpsLog(Indexer, DPS_LOG_EXTRA, "Done");
}
}
return DPS_OK;
}
static int DoStore(DPS_AGENT *Agent, urlid_t rec_id, Byte *Doc, size_t DocSize, char *Client) {
z_stream zstream;
DPS_BASE_PARAM P;
int rc = DPS_OK;
Byte *CDoc = NULL;
size_t dbnum = ((size_t)rec_id) % ((Agent->flags & DPS_FLAG_UNOCON) ? Agent->Conf->dbl.nitems : Agent->dbl.nitems);
DPS_DB *db = (Agent->flags & DPS_FLAG_UNOCON) ? &Agent->Conf->dbl.db[dbnum] : &Agent->dbl.db[dbnum];
zstream.zalloc = Z_NULL;
zstream.zfree = Z_NULL;
zstream.opaque = Z_NULL;
zstream.next_in = Doc;
if (deflateInit2(&zstream, 9, Z_DEFLATED, 15, 9, Z_DEFAULT_STRATEGY) == Z_OK) {
zstream.avail_in = DocSize;
zstream.avail_out = 2 * DocSize;
CDoc = zstream.next_out = (Byte *) DpsMalloc(2 * DocSize + 1);
if (zstream.next_out == NULL) {
return DPS_ERROR;
}
deflate(&zstream, Z_FINISH);
deflateEnd(&zstream);
/* store operations */
bzero(&P, sizeof(P));
P.subdir = "store";
P.basename = "doc";
P.indname = "doc";
P.rec_id = rec_id;
P.mode = DPS_WRITE_LOCK;
P.NFiles = (db->StoredFiles) ? db->StoredFiles : DpsVarListFindInt(&Agent->Vars, "StoredFiles", 0x100);
P.vardir = (db->vardir) ? db->vardir : DpsVarListFindStr(&Agent->Vars, "VarDir", DPS_VAR_DIR);
P.A = Agent;
if (DpsBaseWrite(&P, CDoc, zstream.total_out) != DPS_OK) {
DpsLog(Agent, DPS_LOG_ERROR, "store/doc write error: %s", strerror(errno));
rc = DPS_ERROR;
}
DpsBaseClose(&P);
if (rc == DPS_OK) DpsLog(Agent, DPS_LOG_EXTRA, "[%s] Stored rec_id: %x Size: %d Ratio: %5.2f%%", Client,
rec_id, DocSize, 100.0 * zstream.total_out / DocSize);
if (Agent->Flags.OptimizeAtUpdate) {
DpsBaseOptimize(&P, ((int)rec_id) >> DPS_BASE_BITS);
}
void DpsCookiesClean(DPS_AGENT *A) {
char buf[256];
DPS_DB *db;
size_t i, dbfrom = 0, dbto;
int res;
if (A->Flags.robots_period == 0) return;
dps_snprintf(buf, sizeof(buf), "DELETE FROM cookies WHERE expires < %d", A->now);
if (A->flags & DPS_FLAG_UNOCON) DPS_GETLOCK(A, DPS_LOCK_CONF);
dbto = DPS_DBL_TO(A);
if (A->flags & DPS_FLAG_UNOCON) DPS_RELEASELOCK(A, DPS_LOCK_CONF);
for (i = dbfrom; i < dbto; i++) {
db = DPS_DBL_DB(A, i);
if (A->flags & DPS_FLAG_UNOCON) DPS_GETLOCK(A, DPS_LOCK_DB);
#ifdef HAVE_SQL
res = DpsSQLAsyncQuery(db, NULL, buf);
#endif
if (res != DPS_OK) {
DpsLog(A, DPS_LOG_ERROR, db->errstr);
}
if (A->flags & DPS_FLAG_UNOCON) DPS_RELEASELOCK(A, DPS_LOCK_DB);
if (res != DPS_OK) break;
}
}
static int proceedSTOP(DPS_AGENT *query, DPS_STACK_ITEM *res, DPS_STACK_ITEM *x, DPS_STACK_ITEM *stop) {
res->pbegin = res->pcur = (DPS_URL_CRD_DB*)DpsMalloc((x->count + stop->count + 1) * sizeof(DPS_URL_CRD_DB));
if (res->pbegin == NULL) {
DpsLog(query, DPS_LOG_ERROR, "Can't alloc %d bytes for %d results", (x->count + stop->count + 1) * sizeof(DPS_URL_CRD_DB),
(x->count + stop->count + 1));
return DPS_ERROR;
}
x->pcur = x->pbegin; x->plast = x->pbegin + x->count;
stop->pcur = stop->pbegin; stop->plast = stop->pbegin + stop->count;
if (stop->pcur < stop->plast) {
while (x->pcur < x->plast) {
while (stop->pcur < stop->plast && stop->pcur->url_id < x->pcur->url_id) stop->pcur++;
while (stop->pcur < stop->plast && DpsCmpUrlid(stop->pcur, x->pcur) <= 0) {
*res->pcur = *stop->pcur;
res->pcur++; stop->pcur++;
}
if (stop->pcur >= stop->plast) break;
while (x->pcur < x->plast && DpsCmpUrlid(x->pcur, stop->pcur) <= 0) {
*res->pcur = *x->pcur;
res->pcur++; x->pcur++;
}
}
}
while (x->pcur < x->plast) {
*res->pcur = *x->pcur;
res->pcur++; x->pcur++;
}
return DPS_OK;
}
static int proceedOR(DPS_AGENT *query, DPS_STACK_ITEM *res, DPS_STACK_ITEM *x1, DPS_STACK_ITEM *x2) {
res->pbegin = res->pcur = (DPS_URL_CRD_DB*)DpsMalloc((x1->count + x2->count + 1) * sizeof(DPS_URL_CRD_DB));
if (res->pbegin == NULL) {
DpsLog(query, DPS_LOG_ERROR, "Can't alloc %d bytes for %d results", (x1->count + x2->count + 1) * sizeof(DPS_URL_CRD_DB),
(x1->count + x2->count + 1));
return DPS_ERROR;
}
x1->pcur = x1->pbegin; x1->plast = x1->pbegin + x1->count;
x2->pcur = x2->pbegin; x2->plast = x2->pbegin + x2->count;
while (x1->pcur < x1->plast && x2->pcur < x2->plast) {
while((x1->pcur < x1->plast) && (DpsCmpUrlid(x1->pcur, x2->pcur) <= 0)) {
*res->pcur = *x1->pcur;
res->pcur++; x1->pcur++;
}
{
register DPS_STACK_ITEM *t = x1;
x1 = x2; x2 = t;
}
}
while (x1->pcur < x1->plast) {
*res->pcur = *x1->pcur;
res->pcur++; x1->pcur++;
}
while (x2->pcur < x2->plast) {
*res->pcur = *x2->pcur;
res->pcur++; x2->pcur++;
}
return DPS_OK;
}
__C_LINK int __DPSCALL DpsBaseRead(DPS_BASE_PARAM *P, void *buf, size_t len) {
int res = DPS_OK;
#ifdef HAVE_ZLIB
z_stream zstream;
Byte *CDoc = NULL;
#endif
if ((res = DpsBaseSeek(P, DPS_READ_LOCK)) != DPS_OK) return res;
if (P->Item.rec_id == P->rec_id) {
if (lseek(P->Sfd, (off_t)P->Item.offset, SEEK_SET) == (off_t)-1) {
DpsLog(P->A, DPS_LOG_ERROR, "[%s/%s.%d] %ld lseek error, rec_id: %x",
P->subdir, P->basename, P->FileNo, P->Item.offset, P->rec_id);
return DPS_ERROR;
}
if ((P->Item.orig_size ? P->Item.orig_size : P->Item.size) > len) {
DpsLog(P->A, DPS_LOG_ERROR, "[%s/%s] size %d->%d error, rec_id: %x",
P->subdir, P->basename, (P->Item.orig_size ? P->Item.orig_size : P->Item.size), len, P->rec_id);
return DPS_ERROR;
}
#ifdef HAVE_ZLIB
bzero(&zstream, sizeof(zstream));
if ((P->zlib_method == Z_DEFLATED) && (P->Item.orig_size != 0)) {
zstream.avail_in = (uInt)P->Item.size;
zstream.avail_out = (uInt)len;
zstream.next_out = (Byte *) buf;
CDoc = zstream.next_in = (Byte *) DpsMalloc(P->Item.size + 1);
if (CDoc == NULL) {
return DPS_ERROR;
}
zstream.zalloc = Z_NULL;
zstream.zfree = Z_NULL;
zstream.opaque = Z_NULL;
if (read(P->Sfd, CDoc, P->Item.size) != (ssize_t)P->Item.size) {
DpsLog(P->A, DPS_LOG_ERROR, "[%s/%s] %d read error, rec_id: %x -- %d", P->subdir, P->basename, P->Item.size, P->rec_id, __LINE__);
DPS_FREE(CDoc);
return DPS_ERROR;
}
inflateInit2(&zstream, P->zlib_windowBits);
inflate(&zstream, Z_FINISH);
inflateEnd(&zstream);
DPS_FREE(CDoc);
} else
#endif
if (read(P->Sfd, buf, P->Item.size) != (ssize_t)P->Item.size) {
DpsLog(P->A, DPS_LOG_ERROR, "[%s/%s] %d read error, rec_id: %x -- %d", P->subdir, P->basename, P->Item.size, P->rec_id, __LINE__);
return DPS_ERROR;
}
} else {
DpsLog(P->A, DPS_LOG_DEBUG, "%s:[%s/%s] Not found rec_id: %x", P->vardir, P->subdir, P->basename, P->rec_id);
return DPS_ERROR;
}
#ifdef DEBUG_SEARCH
DpsLog(P->A, DPS_LOG_DEBUG, "[%s/%s] Retrieved rec_id: %x Size: %d", P->subdir, P->basename, P->rec_id, P->Item.size);
#endif
return DPS_OK;
}
__C_LINK int __DPSCALL DpsBaseDelete(DPS_BASE_PARAM *P) {
int res = DPS_OK;
if ((res = DpsBaseSeek(P, DPS_WRITE_LOCK)) != DPS_OK) return res;
if (P->Item.rec_id == P->rec_id) {
P->Item.rec_id = 0;
if (lseek(P->Ifd, (off_t)P->CurrentItemPos, SEEK_SET) == (off_t)-1) {
DpsLog(P->A, DPS_LOG_ERROR, "Can't seek file %s {%s:%d}", P->Ifilename, __FILE__, __LINE__);
return DPS_ERROR;
}
if (write(P->Ifd, &P->Item, sizeof(DPS_BASEITEM)) != sizeof(DPS_BASEITEM)) {
DpsLog(P->A, DPS_LOG_ERROR, "Can't write hash chain for file %s (%s:%d)", P->Ifilename, __FILE__, __LINE__);
return DPS_ERROR;
}
#ifdef DEBUG_SEARCH
DpsLog(P->A, DPS_LOG_DEBUG, "[%s/%s] Deleted rec_id: %x", P->subdir, P->basename, P->rec_id);
#endif
} else {
DpsLog(P->A, DPS_LOG_DEBUG, "[%s/%s] rec_id: %x not found for delete", P->subdir, P->basename, P->rec_id);
}
return DPS_OK;
}
int __DPSCALL DpsSearchdCatAction(DPS_AGENT *A, DPS_CATEGORY *C, int cmd, void *db) {
DPS_DB *searchd = db;
DPS_SEARCHD_PACKET_HEADER hdr;
char *buf;
ssize_t nsent, nrecv;
int done = 0;
int rc=DPS_OK;
char *msg = NULL;
char *dinfo = NULL;
TRACE_IN(A, "DpsSearchdCatAction");
hdr.cmd = DPS_SEARCHD_CMD_CATINFO;
hdr.len = sizeof(int) + dps_strlen(C->addr) + 1;
if ((buf = (char*)DpsMalloc(hdr.len + 1)) == NULL) {
DpsLog(A, DPS_LOG_ERROR, "Out of memory");
TRACE_OUT(A);
return DPS_ERROR;
}
*((int*)buf) = cmd;
dps_strcpy(buf + sizeof(int), C->addr);
nsent = DpsSearchdSendPacket(searchd->searchd, &hdr, buf);
DPS_FREE(buf);
while(!done) {
char * tok, * lt;
nrecv = DpsRecvall(searchd->searchd, &hdr, sizeof(hdr), 360);
if(nrecv != sizeof(hdr)){
DpsLog(A, DPS_LOG_ERROR, "Received incomplete header from searchd (%d bytes)", (int)nrecv);
TRACE_OUT(A);
return(DPS_ERROR);
}else{
#ifdef DEBUG_SDP
DpsLog(A, DPS_LOG_ERROR, "Received header cmd=%d len=%d\n", hdr.cmd, hdr.len);
#endif
}
switch(hdr.cmd){
case DPS_SEARCHD_CMD_ERROR:
msg = (char*)DpsMalloc(hdr.len + 1);
if (msg == NULL) {
done=1;
break;
}
nrecv = DpsRecvall(searchd->searchd, msg, hdr.len, 360);
msg[(nrecv >= 0) ? nrecv : 0 ] = '\0';
sprintf(A->Conf->errstr, "Searchd error: '%s'", msg);
rc=DPS_ERROR;
DPS_FREE(msg);
done=1;
break;
case DPS_SEARCHD_CMD_MESSAGE:
msg=(char*)DpsMalloc(hdr.len+1);
if (msg == NULL) {
done=1;
break;
}
nrecv = DpsRecvall(searchd->searchd, msg, hdr.len, 360);
msg[(nrecv >= 0) ? nrecv : 0] = '\0';
#ifdef DEBUG_SDP
DpsLog(A, DPS_LOG_ERROR, "Message from searchd: '%s'\n",msg);
#endif
DPS_FREE(msg);
break;
case DPS_SEARCHD_CMD_CATINFO:
dinfo=(char*)DpsMalloc(hdr.len+1);
if (dinfo == NULL) {
done=1;
break;
}
nrecv = DpsRecvall(searchd->searchd, dinfo, hdr.len, 360);
dinfo[(nrecv >= 0) ? nrecv : 0] = '\0';
#ifdef DEBUG_SDP
DpsLog(A, DPS_LOG_ERROR, "Received CATINFO size=%d buf=%s\n",hdr.len,dinfo);
#endif
C->ncategories = 0;
tok = dps_strtok_r(dinfo, "\r\n", <, NULL);
while(tok){
DpsCatFromTextBuf(C, tok);
tok = dps_strtok_r(NULL, "\r\n", <, NULL);
}
DPS_FREE(dinfo);
done=1;
break;
default:
sprintf(A->Conf->errstr, "Unknown searchd response: cmd=%d len=%d", hdr.cmd, hdr.len);
rc=DPS_ERROR;
done = 1;
break;
}
}
TRACE_OUT(A);
return rc;
}
int DpsSearchdGetWordResponse(DPS_AGENT *query,DPS_RESULT *Res,DPS_DB *cl) {
DPS_URL_CRD_DB *wrd = NULL;
DPS_URLDATA *udt = NULL;
#ifdef WITH_REL_TRACK
DPS_URLTRACK *trk = NULL;
#endif
DPS_SEARCHD_PACKET_HEADER hdr;
ssize_t nrecv;
char *msg;
int done=0, rc = DPS_OK;
char *wbuf, *p;
DPS_WIDEWORDLIST_EX *wwl;
DPS_WIDEWORD *ww_ex;
DPS_WIDEWORD ww;
size_t i;
TRACE_IN(query, "DpsSearchdGetWordResponse");
Res->total_found=0;
while(!done){
nrecv = DpsRecvall(cl->searchd, &hdr, sizeof(hdr), 360);
if(nrecv!=sizeof(hdr)){
sprintf(query->Conf->errstr,"Received incomplete header from searchd (%d bytes,errno:%d)",(int)nrecv, errno);
TRACE_OUT(query);
return DPS_ERROR;;
}
#ifdef DEBUG_SDP
DpsLog(query, DPS_LOG_ERROR, "Received header cmd=%d len=%d\n",hdr.cmd,hdr.len);
#endif
switch(hdr.cmd){
case DPS_SEARCHD_CMD_ERROR:
msg=(char*)DpsMalloc(hdr.len+1);
if (msg == NULL) {
done = 1;
break;
}
nrecv = DpsRecvall(cl->searchd, msg, hdr.len, 360);
if (nrecv >= 0) {
msg[nrecv]='\0';
sprintf(query->Conf->errstr,"Searchd error: '%s',received:%d", msg, (int)nrecv);
}
rc = DPS_ERROR;
DPS_FREE(msg);
done=1;
break;
case DPS_SEARCHD_CMD_MESSAGE:
msg=(char*)DpsMalloc(hdr.len+1);
if (msg == NULL) {
done = 1;
break;
}
nrecv = DpsRecvall(cl->searchd, msg, hdr.len, 360);
msg[(nrecv >= 0) ? nrecv : 0] = '\0';
if (strncmp(msg, "Total_found", 11) == 0) {
Res->total_found = (size_t)DPS_ATOI(msg + 12);
Res->grand_total = (size_t)DPS_ATOI(strchr(msg + 12, (int)' ') + 1);
}
#ifdef DEBUG_SDP
DpsLog(query, DPS_LOG_ERROR, "Message from searchd: '%s'\n",msg);
#endif
DPS_FREE(msg);
break;
case DPS_SEARCHD_CMD_WORDS:
DPS_FREE(wrd);
wrd=(DPS_URL_CRD_DB*)DpsMalloc(hdr.len + 1);
if (wrd == NULL) {
done = 1;
break;
}
nrecv = DpsRecvall(cl->searchd, wrd, hdr.len, 360);
/*Res->total_found=hdr.len/sizeof(*wrd);*/
Res->num_rows = (nrecv >= 0) ? (size_t)nrecv / sizeof(*wrd) : 0;
#ifdef DEBUG_SDP
DpsLog(query, DPS_LOG_ERROR, "Received words size=%d nwrd=%d\n",hdr.len, Res->num_rows /*Res->total_found*/);
#endif
done=1;
break;
case DPS_SEARCHD_CMD_SUGGEST:
DPS_FREE(Res->Suggest);
Res->Suggest = (char*)DpsMalloc(hdr.len + 1);
if (Res->Suggest == NULL) {
done = 1; break;
}
nrecv = DpsRecvall(cl->searchd, Res->Suggest, hdr.len, 360);
Res->Suggest[(nrecv >=0) ? nrecv : 0] = '\0';
#ifdef DEBUG_SDP
DpsLog(query, DPS_LOG_ERROR, "Received Suggest size=%d\n", hdr.len);
#endif
break;
case DPS_SEARCHD_CMD_PERSITE:
Res->PerSite = (size_t*)DpsMalloc(hdr.len + 1);
if (Res->PerSite == NULL) {
done = 1;
break;
}
nrecv = DpsRecvall(cl->searchd, Res->PerSite, hdr.len, 360);
#ifdef DEBUG_SDP
DpsLog(query, DPS_LOG_ERROR, "Received PerSite size=%d nwrd=%d\n", nrecv, Res->num_rows/*Res->total_found*/);
#endif
break;
case DPS_SEARCHD_CMD_DATA:
udt = (DPS_URLDATA*)DpsMalloc(hdr.len + 1);
if (udt == NULL) {
done = 1;
break;
}
nrecv = DpsRecvall(cl->searchd, udt, hdr.len, 360);
#ifdef DEBUG_SDP
DpsLog(query, DPS_LOG_ERROR, "Received URLDATA size=%d nwrd=%d\n", nrecv, Res->num_rows);
#endif
break;
#ifdef WITH_REL_TRACK
case DPS_SEARCHD_CMD_TRACKDATA:
trk = (DPS_URLTRACK*)DpsMalloc(hdr.len + 1);
if (trk == NULL) {
done = 1;
break;
}
nrecv = DpsRecvall(cl->searchd, trk, hdr.len, 360);
#ifdef DEBUG_SDP
DpsLog(query, DPS_LOG_ERROR, "Received TRACKDATA size=%d nwrd=%d\n", nrecv, Res->num_rows);
#endif
break;
#endif
case DPS_SEARCHD_CMD_WITHOFFSET:
/* Res->offset = 1;*/
break;
case DPS_SEARCHD_CMD_QLC:
if ((p = (char *)DpsXmalloc(hdr.len + 1)) != NULL) {
if (DpsRecvall(cl->searchd, p, hdr.len, 360)) {
DpsVarListReplaceStr(&query->Vars, "q", p);
}
}
DPS_FREE(p);
break;
case DPS_SEARCHD_CMD_WWL:
Res->PerSite = NULL;
if ((wbuf = p = (char *)DpsXmalloc(hdr.len + 1)) != NULL)
if (DpsRecvall(cl->searchd, wbuf, hdr.len, 360)) {
wwl = (DPS_WIDEWORDLIST_EX *)p;
p += sizeof(DPS_WIDEWORDLIST_EX);
#ifdef DEBUG_SDP
DpsLog(query, DPS_LOG_ERROR, "wbuf :%x, wwl: %x, p: %x hdr.len:%d\n", wbuf, wwl, p, hdr.len);
DpsLog(query, DPS_LOG_ERROR, "Received WWL nwords=%d nuniq=%d\n", wwl->nwords, wwl->nuniq);
#endif
/* DpsWideWordListFree(&Res->WWList);*/
for(i = 0; i < wwl->nwords; i++) {
/* ww_ex = (DPS_WIDEWORD_EX *)((void*)&p[0]);*/
dps_memcpy((char*)&ww, p, sizeof(DPS_WIDEWORD_EX));
p += sizeof(DPS_WIDEWORD_EX);
/*
ww.order = ww_ex->order;
ww.order_inquery = ww_ex->order_inquery;
ww.count = ww_ex->count;
ww.len = ww_ex->len;
ww.ulen = ww_ex->ulen;
ww.origin = ww_ex->origin;
ww.crcword = ww_ex->crcword;
*/
ww.word = p;
#ifdef DEBUG_SDP
DpsLog(query, DPS_LOG_ERROR, "Word {%d}: %s\n", ww.len+1, ww.word);
#endif
p += ww.len + 1;
p += sizeof(dpsunicode_t) - ((SDPALIGN)p % sizeof(dpsunicode_t));
ww.uword = (dpsunicode_t*)p;
p += sizeof(dpsunicode_t) * (ww.ulen + 1);
DpsWideWordListAdd(&Res->WWList, &ww, DPS_WWL_STRICT);
}
Res->WWList.nuniq = wwl->nuniq;
DPS_FREE(wbuf);
}
break;
default:
sprintf(query->Conf->errstr,"Unknown searchd response: cmd=%d len=%d",hdr.cmd,hdr.len);
rc = DPS_ERROR;
done=1;
break;
}
}
Res->CoordList.Coords = wrd;
Res->CoordList.Data = udt;
#ifdef WITH_REL_TRACK
Res->CoordList.Track = trk;
#endif
TRACE_OUT(query);
return rc;
}
int __DPSCALL DpsResAddDocInfoSearchd(DPS_AGENT * query,DPS_DB *cl,DPS_RESULT * Res,size_t clnum){
DPS_SEARCHD_PACKET_HEADER hdr;
char * msg=NULL;
size_t i; /* num=0,curnum=0;*/
int done = 0;
ssize_t nsent,nrecv;
char * dinfo=NULL;
int rc=DPS_OK;
char *textbuf;
size_t dlen = 0;
TRACE_IN(query, "DpsResAddDocInfoSearchd");
if(!Res->num_rows) { TRACE_OUT(query); return(DPS_OK); }
for(i=0;i<Res->num_rows;i++){
size_t ulen;
size_t olen;
size_t nsec, r;
DPS_DOCUMENT *D=&Res->Doc[i];
r = (size_t) 's';
for(nsec = 0; nsec < D->Sections.Root[r].nvars; nsec++)
if (strcasecmp(D->Sections.Root[r].Var[nsec].name, "Score") == 0) D->Sections.Root[r].Var[nsec].section = 1;
#ifdef WITH_MULTIDBADDR
if (D->dbnum != cl->dbnum) continue;
#endif
textbuf = DpsDocToTextBuf(D, 1, 0);
if (textbuf == NULL) {TRACE_OUT(query); return DPS_ERROR;}
ulen = dps_strlen(textbuf)+2;
olen = dlen;
dlen = dlen + ulen;
dinfo = (char*)DpsRealloc(dinfo, dlen + 1);
if (dinfo == NULL) {
DpsFree(textbuf);
TRACE_OUT(query);
return DPS_ERROR;
}
dinfo[olen] = '\0';
sprintf(dinfo + olen, "%s\r\n", textbuf);
DpsFree(textbuf);
}
if (dinfo == NULL) {
TRACE_OUT(query);
return DPS_OK;
}
hdr.cmd=DPS_SEARCHD_CMD_DOCINFO;
hdr.len = dps_strlen(dinfo);
nsent = DpsSearchdSendPacket(cl->searchd, &hdr, dinfo);
#ifdef DEBUG_SDP
DpsLog(query, DPS_LOG_ERROR, "Sent DOCINFO size=%d buf=%s\n", hdr.len, dinfo);
#endif
while(!done){
char * tok, * lt;
nrecv = DpsRecvall(cl->searchd, &hdr, sizeof(hdr), 360);
if(nrecv!=sizeof(hdr)){
DpsLog(query, DPS_LOG_ERROR, "Received incomplete header from searchd (%d bytes, errno:%d)", (int)nrecv, errno);
TRACE_OUT(query);
return(DPS_ERROR);
}else{
#ifdef DEBUG_SDP
DpsLog(query, DPS_LOG_ERROR, "Received header cmd=%d len=%d\n",hdr.cmd,hdr.len);
#endif
}
switch(hdr.cmd){
case DPS_SEARCHD_CMD_ERROR:
msg=(char*)DpsMalloc(hdr.len+1);
if (msg == NULL) {
done = 1;
break;
}
nrecv = DpsRecvall(cl->searchd, msg, hdr.len, 360);
msg[(nrecv >= 0) ? nrecv : 0]='\0';
sprintf(query->Conf->errstr,"Searchd error: '%s'",msg);
rc=DPS_ERROR;
DPS_FREE(msg);
done=1;
break;
case DPS_SEARCHD_CMD_MESSAGE:
msg=(char*)DpsMalloc(hdr.len+1);
if (msg == NULL) {
done = 1;
break;
}
nrecv = DpsRecvall(cl->searchd, msg, hdr.len, 360);
msg[(nrecv >= 0) ? nrecv : 0]='\0';
#ifdef DEBUG_SDP
DpsLog(query, DPS_LOG_ERROR, "Message from searchd: '%s'\n",msg);
#endif
DPS_FREE(msg);
break;
case DPS_SEARCHD_CMD_DOCINFO:
dinfo = (char*)DpsRealloc(dinfo, hdr.len + 1);
if (dinfo == NULL) {
done=1;
break;
}
nrecv = DpsRecvall(cl->searchd, dinfo, hdr.len, 360);
dinfo[(nrecv > 0) ? nrecv : 0] = '\0';
#ifdef DEBUG_SDP
DpsLog(query, DPS_LOG_ERROR, "Received DOCINFO size=%d buf=%s\n",hdr.len,dinfo);
#endif
tok = dps_strtok_r(dinfo, "\r\n", <, NULL);
while(tok){
urlid_t Doc_url_id, Res_Doc_url_id;
DPS_DOCUMENT Doc;
DpsDocInit(&Doc);
DpsDocFromTextBuf(&Doc,tok);
Doc_url_id = (urlid_t)DpsVarListFindInt(&Doc.Sections, "DP_ID", 0);
for(i=0;i<Res->num_rows;i++){
#ifdef WITH_MULTIDBADDR
if (Res->Doc[i].dbnum != cl->dbnum) continue;
#endif
Res_Doc_url_id = (urlid_t)DpsVarListFindInt(&Res->Doc[i].Sections, "DP_ID", 0);
if (Res_Doc_url_id == Doc_url_id) {
DpsDocFromTextBuf(&Res->Doc[i], tok);
break;
}
}
tok = dps_strtok_r(NULL, "\r\n", <, NULL);
DpsDocFree(&Doc);
}
DPS_FREE(dinfo);
done=1;
break;
default:
sprintf(query->Conf->errstr,"Unknown searchd response: cmd=%d len=%d",hdr.cmd,hdr.len);
rc=DPS_ERROR;
done=1;
break;
}
}
TRACE_OUT(query);
return rc;
}
int main(int argc, char ** argv, char **envp) {
const char *env, *bcharset, *lcharset, *conf_dir;
char template_name[PATH_MAX+6]="";
char *template_filename = NULL;
char *query_string = NULL;
char self[1024]="";
char *url = NULL;
const char *ResultContentType;
int res,httpd=0;
size_t catcolumns = 0;
int page_size,page_number;
DPS_ENV *Env;
DPS_AGENT *Agent;
DPS_VARLIST query_vars;
/* Output Content-type if under HTTPD */
/* Some servers do not pass QUERY_STRING */
/* if the query was empty, so check */
/* REQUEST_METHOD too to be safe */
httpd=(getenv("QUERY_STRING")||getenv("REQUEST_METHOD"));
if (!(conf_dir=getenv("DPS_ETC_DIR")))
conf_dir=DPS_CONF_DIR;
DpsInit(argc, argv, envp);
Env=DpsEnvInit(NULL);
if (Env == NULL) {
if(httpd){
printf("Content-Type: text/plain\r\n\r\n");
}
printf("Can't alloc Env\n");
exit(0);
}
DpsVarListInit(&query_vars);
Agent = DpsAgentInit(NULL, Env, 0);
if (Agent == NULL) {
if(httpd){
printf("Content-Type: text/plain\r\n\r\n");
}
printf("Can't alloc Agent\n");
exit(0);
}
DpsVarListAddEnviron(&Env->Vars,"ENV");
/* Detect self and template name */
if((env = getenv("DPSEARCH_TEMPLATE")))
dps_strncpy(template_name, env, sizeof(template_name) - 1);
else if((env = getenv("PATH_INFO")) && env[0])
dps_strncpy(template_name, env + 1, sizeof(template_name) - 1);
if((env=getenv("DPSEARCH_SELF")))
dps_strncpy(self,env,sizeof(self)-1);
if((env=getenv("QUERY_STRING"))){
query_string = (char*)DpsRealloc(query_string, dps_strlen(env) + 2);
if (query_string == NULL) {
if(httpd){
printf("Content-Type: text/plain\r\n\r\n");
}
printf("Can't alloc query_string\n");
exit(0);
}
dps_strncpy(query_string, env, dps_strlen(env) + 1);
/* Hack for Russian Apache from apache.lexa.ru */
/* QUERY_STRING is already converted to server */
/* character set. We must print original query */
/* string instead however. Under usual apache */
/* we'll use QUERY_STRING. Note that query_vars */
/* list will contain not unescaped values, so */
/* we don't have to escape them when displaying */
env = getenv("CHARSET_SAVED_QUERY_STRING");
DpsParseQStringUnescaped(&query_vars,env?env:query_string);
/* Unescape and save variables from QUERY_STRING */
/* Env->Vars will have unescaped values however */
DpsParseQueryString(Agent,&Env->Vars,query_string);
template_filename = (char*)DpsStrdup(DpsVarListFindStr(&Env->Vars, "tmplt", ""));
if((env=getenv("REDIRECT_STATUS"))){
/* Check Apache internal redirect */
/* via "AddHandler" and "Action" */
if(!self[0]){
dps_strncpy(self,(env=getenv("REDIRECT_URL"))?env:"filler.cgi",sizeof(self)-1);
}
if(!template_name[0]){
dps_strncpy(template_name,(env=getenv("PATH_TRANSLATED"))?env:"",sizeof(template_name)-1);
}
if (*template_filename == '\0') {
DPS_FREE(template_filename);
template_filename = (char*)DpsStrdup("filler.htm");
}
}else{
/* CGI executed without Apache internal redirect */
/* Detect $Self variable with OS independant SLASHES */
if(!self[0]){
dps_strncpy(self,(env=getenv("SCRIPT_NAME"))?env:"filler.cgi",sizeof(self)-1);
}
if(!template_name[0]){
char *s,*e;
/*This is with OS specific SLASHES */
env=((env=getenv("SCRIPT_FILENAME"))?env:"filler.cgi");
if(strcmp(conf_dir,".")){
/* Take from the config directory */
dps_snprintf(template_name, sizeof(template_name)-1, "%s/%s",
conf_dir,(s=strrchr(env,DPSSLASH))?(s+1):(self));
}else{
/* Take from the current directory */
dps_strncpy(template_name,env,sizeof(template_name)-1);
}
/* Find right slash if it presents */
s=((s=strrchr(template_name,DPSSLASH))?s:template_name);
if (*template_filename == '\0') {
/* Find .cgi substring */
if ((e = strstr(s, ".cgi")) != NULL) {
/* Replace ".cgi" with ".htm" */
e[1]='h';e[2]='t';e[3]='m';
} else {
dps_strcat(s, ".htm");
}
e = strrchr(s, '/');
DPS_FREE(template_filename);
template_filename = (char*)DpsStrdup(e + 1);
} else {
dps_strncpy(s + 1, template_filename, sizeof(template_name) - (s - template_name) - 2);
}
}
}
}else{
/* Executed from command line */
/* or under server which does not */
/* pass an empty QUERY_STRING var */
if(argv[1]) {
query_string = (char*)DpsRealloc(query_string, dps_strlen(argv[1]) + 10);
if (query_string == NULL) {
if(httpd){
printf("Content-Type: text/plain\r\n\r\n");
}
printf("Can't realloc query_string\n");
exit(0);
}
sprintf(query_string, "q=%s", argv[1]);
} else {
query_string = (char*)DpsRealloc(query_string, 1024);
if (query_string == NULL) {
if(httpd){
printf("Content-Type: text/plain\r\n\r\n");
}
printf("Can't realloc query_string\n");
exit(0);
}
sprintf(query_string, "q=");
}
/* Hack for Russian Apache from apache.lexa.ru */
/* QUERY_STRING is already converted to server */
/* character set. We must print original query */
/* string instead however. Under usual apache */
/* we'll use QUERY_STRING. Note that query_vars */
/* list will contain not unescaped values, so */
/* we don't have to escape them when displaying */
env = getenv("CHARSET_SAVED_QUERY_STRING");
DpsParseQStringUnescaped(&query_vars,env?env:query_string);
/* Unescape and save variables from QUERY_STRING */
/* Env->Vars will have unescaped values however */
DpsParseQueryString(Agent,&Env->Vars,query_string);
DPS_FREE(template_filename);
template_filename = (char*)DpsStrdup(DpsVarListFindStr(&Env->Vars, "tmplt", ""));
if (*template_filename == '\0') {
DPS_FREE(template_filename); template_filename = (char*)DpsStrdup("filler.htm");
}
/*// Get template name from command line variable &tmplt */
if(!template_name[0])
dps_snprintf(template_name,sizeof(template_name),"%s/%s", conf_dir, template_filename);
}
DpsVarListReplaceStr(&Agent->Conf->Vars, "tmplt", template_filename);
DPS_FREE(template_filename);
Agent->tmpl.Env_Vars = &Env->Vars;
DpsURLNormalizePath(template_name);
if (strncmp(template_name, conf_dir, dps_strlen(conf_dir))
|| (res = DpsTemplateLoad(Agent, Env, &Agent->tmpl, template_name))) {
if (strcmp(template_name, "filler.htm")) { /* trying load default template */
fprintf(stderr, "Can't load template: '%s' %s\n", template_name, Env->errstr);
DPS_FREE(template_filename);
template_filename = (char*)DpsStrdup("filler.htm");
dps_snprintf(template_name, sizeof(template_name), "%s/%s", conf_dir, template_filename);
if ((res = DpsTemplateLoad(Agent, Env, &Agent->tmpl, template_name))) {
if(httpd)printf("Content-Type: text/plain\r\n\r\n");
printf("%s\n",Env->errstr);
DpsVarListFree(&query_vars);
DpsEnvFree(Env);
DPS_FREE(query_string);
DpsAgentFree(Agent);
return(0);
}
} else {
if(httpd)printf("Content-Type: text/plain\r\n\r\n");
printf("%s\n",Env->errstr);
DpsVarListFree(&query_vars);
DpsEnvFree(Env);
DPS_FREE(query_string);
DpsAgentFree(Agent);
return(0);
}
}
/* set locale if specified */
if ((url = DpsVarListFindStr(&Env->Vars, "Locale", NULL)) != NULL) {
setlocale(LC_ALL, url);
/*#ifdef HAVE_ASPELL*/
{ char *p;
if ((p = strchr(url, '.')) != NULL) {
*p = '\0';
DpsVarListReplaceStr(&Env->Vars, "g-lc", url);
*p = '.';
}
}
/*#endif*/
url = NULL;
}
/* Call again to load search Limits if need */
DpsParseQueryString(Agent, &Env->Vars, query_string);
Agent->Flags = Env->Flags;
Agent->flags |= DPS_FLAG_UNOCON;
Env->flags |= DPS_FLAG_UNOCON;
DpsSetLogLevel(NULL, DpsVarListFindInt(&Env->Vars, "LogLevel", 0));
DpsOpenLog("filler.cgi", Env, !strcasecmp(DpsVarListFindStr(&Env->Vars, "Log2stderr", (!httpd) ? "yes" : "no"), "yes"));
DpsLog(Agent,DPS_LOG_ERROR,"filler.cgi started with '%s'",template_name);
DpsLog(Agent, DPS_LOG_DEBUG, "VarDir: '%s'", DpsVarListFindStr(&Agent->Conf->Vars, "VarDir", DPS_VAR_DIR));
DpsLog(Agent, DPS_LOG_DEBUG, "Affixes: %d, Spells: %d, Synonyms: %d, Acronyms: %d, Stopwords: %d",
Env->Affixes.naffixes,Env->Spells.nspell,
Env->Synonyms.nsynonyms,
Env->Acronyms.nacronyms,
Env->StopWords.nstopwords);
DpsLog(Agent, DPS_LOG_DEBUG, "Chinese dictionary with %d entries", Env->Chi.nwords);
DpsLog(Agent, DPS_LOG_DEBUG, "Korean dictionary with %d entries", Env->Korean.nwords);
DpsLog(Agent, DPS_LOG_DEBUG, "Thai dictionary with %d entries", Env->Thai.nwords);
DpsVarListAddLst(&Agent->Vars, &Env->Vars, NULL, "*");
Agent->tmpl.Env_Vars = &Agent->Vars;
/* DpsVarListAddEnviron(&Agent->Vars, "ENV");*/
/****************************************************************************************************************************************/
/* This is for query tracking */
DpsVarListAddStr(&Agent->Vars, "QUERY_STRING", query_string);
DpsVarListAddStr(&Agent->Vars, "self", self);
env = getenv("HTTP_X_FORWARDER_FOR");
if (env) {
DpsVarListAddStr(&Agent->Vars, "IP", env);
} else {
env = getenv("REMOTE_ADDR");
DpsVarListAddStr(&Agent->Vars, "IP", env ? env : "localhost");
}
bcharset = DpsVarListFindStr(&Agent->Vars, "BrowserCharset", "iso-8859-1");
Env->bcs=DpsGetCharSet(bcharset);
lcharset = DpsVarListFindStr(&Agent->Vars, "LocalCharset", "iso-8859-1");
Env->lcs=DpsGetCharSet(lcharset);
ResultContentType = DpsVarListFindStr(&Agent->Vars, "ResultContentType", "text/html");
if(httpd){
if(!Env->bcs){
printf("Content-Type: text/plain\r\n\r\n");
printf("Unknown BrowserCharset '%s' in template '%s'\n",bcharset,template_name);
exit(0);
}else if(!Env->lcs){
printf("Content-Type: text/plain\r\n\r\n");
printf("Unknown LocalCharset '%s' in template '%s'\n",lcharset,template_name);
exit(0);
}else{
printf("Content-type: %s; charset=%s\r\n\r\n", ResultContentType, bcharset);
}
}else{
if(!Env->bcs){
printf("Unknown BrowserCharset '%s' in template '%s'\n",bcharset,template_name);
exit(0);
}
if(!Env->lcs){
printf("Unknown LocalCharset '%s' in template '%s'\n",lcharset,template_name);
exit(0);
}
}
/* These parameters taken from "variable section of template"*/
res = DpsVarListFindInt(&Agent->Vars, "ps", DPS_DEFAULT_PS);
page_size = dps_min(res, MAX_PS);
page_number = DpsVarListFindInt(&Agent->Vars, "p", 0);
if (page_number == 0) {
page_number = DpsVarListFindInt(&Agent->Vars, "np", 0);
DpsVarListReplaceInt(&Agent->Vars, "p", page_number + 1);
} else page_number--;
res = DpsVarListFindInt(&Agent->Vars, "np", 0) * page_size;
DpsVarListAddInt(&Agent->Vars, "pn", res);
catcolumns = (size_t)atoi(DpsVarListFindStr(&Agent->Vars, "CatColumns", ""));
DpsTemplatePrint(Agent, (DPS_OUTPUTFUNCTION)&fprintf, stdout, NULL, 0, &Agent->tmpl, "top");
DpsTemplatePrint(Agent, (DPS_OUTPUTFUNCTION)&fprintf, stdout, NULL, 0, &Agent->tmpl, "restop");
DpsTemplatePrint(Agent, (DPS_OUTPUTFUNCTION)&fprintf, stdout, NULL, 0, &Agent->tmpl, "res");
DpsTemplatePrint(Agent, (DPS_OUTPUTFUNCTION)&fprintf, stdout, NULL, 0, &Agent->tmpl, "resbot");
DpsTemplatePrint(Agent, (DPS_OUTPUTFUNCTION)&fprintf, stdout, NULL, 0, &Agent->tmpl, "bottom");
DpsVarListFree(&query_vars);
DpsAgentFree(Agent);
DpsEnvFree(Env);
DPS_FREE(query_string);
DPS_FREE(url);
if (httpd) fflush(NULL); else fclose(stdout);
#ifdef EFENCE
fprintf(stderr, "Memory leaks checking\n");
DpsEfenceCheckLeaks();
#endif
#ifdef FILENCE
fprintf(stderr, "FD leaks checking\n");
DpsFilenceCheckLeaks(NULL);
#endif
return DPS_OK;
}
static int MakeLinearIndex(DPS_AGENT *Indexer, const char *field, const char *lim_name, int type, DPS_DB *db) {
DPS_ENV *Conf = Indexer->Conf;
DPS_UINT4URLIDLIST L;
size_t k,prev;
urlid_t *data = NULL;
DPS_UINT4_POS_LEN *ind=NULL;
size_t mind=1000,nind=0;
char fname[PATH_MAX];
int dat_fd=0, ind_fd=0, rc;
const char *vardir = (db->vardir) ? db->vardir : DpsVarListFindStr(&Conf->Vars, "VarDir", DPS_VAR_DIR);
bzero(&L, sizeof(DPS_UINT4URLIDLIST));
rc = DpsLimit4(Indexer, &L, field, type, db);
if(rc != DPS_OK) {
DpsLog(Indexer, DPS_LOG_ERROR, "Error: %s [%s:%d]", DpsEnvErrMsg(Conf), __FILE__, __LINE__);
goto err1;
}
if(!L.Item)return(1);
if (L.nitems > 1) DpsSort(L.Item, L.nitems, sizeof(DPS_UINT4URLID), (qsort_cmp)cmp_ind4);
data = (urlid_t*)DpsMalloc((L.nitems + 1) * sizeof(*data));
if(!data) {
fprintf(stderr,"Error1: %s\n",strerror(errno));
goto err1;
}
ind=(DPS_UINT4_POS_LEN*)DpsMalloc(mind*sizeof(DPS_UINT4_POS_LEN));
if(!ind) {
fprintf(stderr,"Error2: %s\n",strerror(errno));
goto err1;
}
prev=0;
for(k=0; k<L.nitems; k++) {
data[k]=L.Item[k].url_id;
if((k==L.nitems-1) || (L.Item[k].val!=L.Item[prev].val)) {
if(nind==mind) {
mind+=1000;
ind=(DPS_UINT4_POS_LEN*)DpsRealloc(ind,mind*sizeof(DPS_UINT4_POS_LEN));
if(!ind) {
fprintf(stderr,"Error3: %s\n",strerror(errno));
goto err1;
}
}
/* Fill index */
ind[nind].val=L.Item[prev].val;
ind[nind].pos = prev * sizeof(*data);
if (k == L.nitems - 1) ind[nind].len = (k - prev + 1) * sizeof(*data);
else ind[nind].len = (k - prev) * sizeof(*data);
DpsLog(Indexer, DPS_LOG_DEBUG, "%d - pos:%x len:%d\n", ind[nind].val, (int)ind[nind].pos, ind[nind].len);
nind++;
prev=k;
}
}
if (L.mapped) {
#ifdef HAVE_SYS_MMAN_H
if (munmap(L.Item, (L.nitems + 1) * sizeof(DPS_UINT4URLID))) {
fprintf(stderr, "Can't shmdt '%s': %s\n", L.shm_name, strerror(errno));
}
#elif defined(HAVE_SYS_SHM_H)
if (shmdt(L.Item)) {
fprintf(stderr, "Can't shmdt '%s': %s\n", L.shm_name, strerror(errno));
}
#endif
unlink(L.shm_name);
} else {
DPS_FREE(L.Item);
}
dps_snprintf(fname,sizeof(fname),"%s%c%s%c%s.dat", vardir,DPSSLASH, DPS_TREEDIR, DPSSLASH, lim_name);
if((dat_fd = DpsOpen3(fname, O_CREAT | O_WRONLY | O_TRUNC | DPS_BINARY, DPS_IWRITE)) < 0) {
fprintf(stderr,"Can't open '%s': %s\n",fname,strerror(errno));
goto err1;
}
DpsWriteLock(dat_fd);
if((L.nitems * sizeof(*data)) != (size_t)write(dat_fd, data, L.nitems * sizeof(*data))) {
fprintf(stderr,"Can't write '%s': %s\n",fname,strerror(errno));
goto err1;
}
DpsUnLock(dat_fd);
DpsClose(dat_fd);
DPS_FREE(data);
dps_snprintf(fname,sizeof(fname),"%s%c%s%c%s.ind", vardir,DPSSLASH, DPS_TREEDIR, DPSSLASH, lim_name);
if((ind_fd = DpsOpen3(fname, O_CREAT | O_WRONLY | O_TRUNC | DPS_BINARY, DPS_IWRITE)) < 0) {
fprintf(stderr,"Can't open '%s': %s\n",fname,strerror(errno));
goto err1;
}
DpsWriteLock(ind_fd);
if((nind*sizeof(DPS_UINT4_POS_LEN)) != (size_t)write(ind_fd,ind,nind*sizeof(DPS_UINT4_POS_LEN))) {
fprintf(stderr,"Can't write '%s': %s\n",fname,strerror(errno));
goto err1;
}
DpsUnLock(ind_fd);
DpsClose(ind_fd);
DPS_FREE(ind);
return(0);
err1:
if (L.mapped) {
#ifdef HAVE_SYS_MMAN_H
if (munmap(L.Item, (L.nitems + 1) * sizeof(DPS_UINT4URLID))) {
fprintf(stderr, "Can't shmdt '%s': %s\n", L.shm_name, strerror(errno));
}
#elif defined(HAVE_SYS_SHM_H)
if (shmdt(L.Item)) {
fprintf(stderr, "Can't shmdt '%s': %s\n", L.shm_name, strerror(errno));
}
#endif
unlink(L.shm_name);
} else {
DPS_FREE(L.Item);
}
DPS_FREE(data);
DPS_FREE(ind);
if(dat_fd) DpsClose(dat_fd);
if(ind_fd) DpsClose(ind_fd);
return(1);
}
int main(int argc,char **argv, char **envp) {
int ch, sleeps = 1, optimize = 0, obi = 0;
unsigned int from = 0, to = 0xFFF, p_to = 0;
DPS_ENV * Env;
const char * config_name = DPS_CONF_DIR "/cached.conf";
DpsInit(argc, argv, envp); /* Initialize library */
DpsInitMutexes();
Env=DpsEnvInit(NULL);
if (Env == NULL) exit(1);
DpsSetLockProc(Env, DpsLockProc);
/*#ifndef HAVE_SETPROCTITLE*/
ARGV = argv;
ARGC = argc;
/*#endif*/
while ((ch = getopt(argc, argv, "blt:f:op:w:v:h?")) != -1){
switch (ch) {
case 'f':
sscanf(optarg, "%x", &from);
break;
case 't':
sscanf(optarg, "%x", &p_to);
break;
case 'w':
DpsVarListReplaceStr(&Env->Vars, "VarDir", optarg);
break;
case 'v': DpsSetLogLevel(NULL, atoi(optarg)); break;
case 'b': obi++; break;
case 'o': optimize++; break;
case 'p': sleeps = atoi(optarg); break;
case 'h':
case '?':
default:
usage();
DpsEnvFree(Env);
DpsDeInit();
DpsDestroyMutexes();
return 1;
break;
}
}
argc -= optind;
argv += optind;
if(argc > 1) {
usage();
DpsEnvFree(Env);
DpsDeInit();
DpsDestroyMutexes();
return 1;
} else if (argc == 1) {
config_name = argv[0];
}
{
DPS_LOGDEL *del_buf=NULL;
size_t del_count = 0, log, bytes, n = 0;
int dd, log_fd;
struct stat sb;
char dname[PATH_MAX] = "";
DPS_BASE_PARAM P;
DPS_LOGWORD *log_buf = NULL;
DPS_AGENT *Indexer = DpsAgentInit(NULL, Env, 0);
log2stderr = 1;
if (Indexer == NULL) {
fprintf(stderr, "Can't alloc Agent at %s:%d\n", __FILE__, __LINE__);
exit(DPS_ERROR);
}
if(DPS_OK != DpsEnvLoad(Indexer, config_name, (dps_uint8)0)){
fprintf(stderr, "%s\n", DpsEnvErrMsg(Env));
DpsEnvFree(Env);
DpsDeInit();
DpsDestroyMutexes();
return DPS_ERROR;
}
DpsOpenLog("splitter", Env, log2stderr);
Indexer->flags = Env->flags = DPS_FLAG_UNOCON;
DpsVarListAddLst(&Indexer->Vars, &Env->Vars, NULL, "*");
bzero(&P, sizeof(P));
P.subdir = DPS_TREEDIR;
P.basename = "wrd";
P.indname = "wrd";
P.mode = DPS_WRITE_LOCK;
P.NFiles = DpsVarListFindInt(&Indexer->Conf->Vars, "WrdFiles", 0x300);
P.vardir = DpsStrdup(DpsVarListFindStr(&Indexer->Conf->Vars, "VarDir", DPS_VAR_DIR));
P.A = Indexer;
if (p_to != 0) to = p_to;
else to = P.NFiles - 1;
#ifdef HAVE_ZLIB
P.zlib_method = Z_DEFLATED;
P.zlib_level = 9;
P.zlib_windowBits = DPS_BASE_WRD_WINDOWBITS;
P.zlib_memLevel = 9;
P.zlib_strategy = DPS_BASE_WRD_STRATEGY;
#endif
/* Open del log file */
dps_snprintf(dname,sizeof(dname),"%s%c%s%cdel-split.log", P.vardir, DPSSLASH, DPS_SPLDIR, DPSSLASH);
if((dd = DpsOpen2(dname, O_RDONLY | DPS_BINARY)) < 0) {
dps_strerror(NULL, 0, "Can't open del log '%s'", dname);
exit(DPS_ERROR);
}
DpsLog(Indexer, DPS_LOG_DEBUG, "VarDir: %s, WrdFiles: %d [%x]", P.vardir, P.NFiles, P.NFiles);
/* Allocate del buffer */
fstat(dd, &sb);
if (sb.st_size != 0) {
del_buf=(DPS_LOGDEL*)DpsMalloc((size_t)sb.st_size + 1);
if (del_buf == NULL) {
fprintf(stderr, "Can't alloc %d bytes at %s:%d\n", (int)sb.st_size, __FILE__, __LINE__);
exit(0);
}
del_count=read(dd,del_buf,(size_t)sb.st_size)/sizeof(DPS_LOGDEL);
}
DpsClose(dd);
/* Remove duplicates URLs in DEL log */
/* Keep only oldest records for each URL */
if (del_count > 0) {
DpsLog(Indexer, DPS_LOG_DEBUG, "Sorting del_buf: %d items", del_count);
if (del_count > 1) DpsSort(del_buf, (size_t)del_count, sizeof(DPS_LOGDEL), DpsCmpurldellog);
DpsLog(Indexer, DPS_LOG_DEBUG, "Removing DelLogDups");
del_count = DpsRemoveDelLogDups(del_buf, del_count);
}
DpsLog(Indexer, DPS_LOG_DEBUG, "Processing Bufs from %d [%x] to %d [%x]", from, from, to, to);
for(log = from; log <= to; log++) {
/* Open log file */
dps_snprintf(dname, sizeof(dname), "%s%c%s%c%03X-split.log", P.vardir, DPSSLASH, DPS_SPLDIR, DPSSLASH, log);
if((log_fd = DpsOpen2(dname, O_RDWR|DPS_BINARY)) < 0){
if (errno == ENOENT) {
dps_strerror(Indexer, DPS_LOG_DEBUG, "Can't open '%s'", dname);
n = 0;
/* continue;*/
} else {
dps_strerror(Indexer, DPS_LOG_ERROR, "Can't open '%s'", dname);
continue;
}
} else {
DpsWriteLock(log_fd);
DpsLog(Indexer, DPS_LOG_DEBUG, "Processing Log: %x", log);
fstat(log_fd, &sb);
log_buf = (sb.st_size > 0) ? (DPS_LOGWORD*)DpsMalloc((size_t)sb.st_size + 1) : NULL;
if (log_buf != NULL) {
unlink(dname);
bytes = read(log_fd,log_buf,(size_t)sb.st_size);
(void)ftruncate(log_fd, (off_t)0);
DpsUnLock(log_fd);
DpsClose(log_fd);
n = bytes / sizeof(DPS_LOGWORD);
DpsLog(Indexer, DPS_LOG_DEBUG, "Sorting log_buf: %d items", n);
if (n > 1) DpsSort(log_buf, n, sizeof(DPS_LOGWORD), (qsort_cmp)DpsCmplog);
DpsLog(Indexer, DPS_LOG_DEBUG, "Removing OldWords");
n = DpsRemoveOldWords(log_buf, n, del_buf, del_count);
if (n > 1) DpsSort(log_buf, n, sizeof(DPS_LOGWORD), (qsort_cmp)DpsCmplog_wrd);
} else {
n = 0;
DpsUnLock(log_fd);
DpsClose(log_fd);
}
}
DpsLog(Indexer, DPS_LOG_DEBUG, "Processing Buf, optimize: %d", optimize);
if (obi) DpsBaseOptimize(&P, log);
DpsProcessBuf(Indexer, &P, log, log_buf, n, del_buf, del_count);
if (optimize) DpsBaseOptimize(&P, log);
DpsBaseClose(&P);
DPS_FREE(log_buf);
DpsLog(Indexer, DPS_LOG_DEBUG, "pas done: %d from %d to %d", log, from, to);
DPSSLEEP(sleeps);
}
DPS_FREE(del_buf);
DpsAgentFree(Indexer);
DPS_FREE(P.vardir);
}
fprintf(stderr, "Splitting done.\n");
DpsEnvFree(Env);
DpsDeInit();
DpsDestroyMutexes();
#ifdef EFENCE
fprintf(stderr, "Memory leaks checking\n");
DpsEfenceCheckLeaks();
#endif
#ifdef FILENCE
fprintf(stderr, "FD leaks checking\n");
DpsFilenceCheckLeaks(NULL);
#endif
return 0;
}
/* Compute one operation and store result */
static int perform(DPS_AGENT *query, DPS_RESULT *Res, DPS_BOOLSTACK *s, int com) {
DPS_STACK_ITEM res, *x1, *x2;
int rc = DPS_OK, found, flag1;
bzero(&res, sizeof(res));
switch(com){
case DPS_STACK_PHRASE_LEFT:
x1 = POPARG(s);
if (x1 == NULL) {
bzero(&res, sizeof(res));
} else {
res = *x1; /* FIXME: add checking ? */
if (res.order_from != res.order_to) {
DPS_URL_CRD_DB *w;
dps_uint4 *pos_real, *order_ideal, *order_real, *gap_ahead/*, *gap_back*/;
urlid_t curlid;
size_t nwords = res.order_to - res.order_from + 1, nonstop_words;
size_t p_cmp, p_ins;
res.plast = res.pbegin + res.count;
w = res.pcur = res.pchecked = res.pbegin;
if ((pos_real = (dps_uint4*)DpsMalloc(5 * nwords * sizeof(dps_uint4) + 1)) == NULL) {
DpsLog(query, DPS_LOG_ERROR, "Can't alloc %d bytes %s:%d",(5 * nwords * sizeof(dps_uint4) + 1), __FILE__, __LINE__);
return DPS_ERROR;
}
order_real = pos_real + nwords;
gap_ahead = order_real + nwords;
order_ideal = gap_ahead + nwords;
/* gap_back = order_ideal + nwords;*/
nonstop_words = 0;
{ register size_t tt;
for (tt = res.order_from; tt <= res.order_to; tt++) {
#ifdef DEBUG_BOOL
#endif
if ((Res->items[tt].order_origin & DPS_WORD_ORIGIN_STOP) == 0) {
order_ideal[nonstop_words] = tt;
gap_ahead[nonstop_words] = (tt == res.order_to) ? 0 : 1;
{ register size_t zz;
for (zz = tt + 1; zz <= res.order_to; zz++) {
if (Res->items[zz].order_origin & DPS_WORD_ORIGIN_STOP) gap_ahead[nonstop_words]++;
else break;
}
}
nonstop_words++;
}
}
}
#ifdef DEBUG_BOOL
DpsLog(query, DPS_LOG_EXTRA, "nonstopwords: %d nwords:%d", nonstop_words, nwords);
{ register size_t tt;
for (tt = 0; tt < nonstop_words; tt++) {
DpsLog(query, DPS_LOG_EXTRA, "%d:order_ideal:%d gap_ahead:%d", tt, order_ideal[tt], gap_ahead[tt]);
}
}
#endif
if (nonstop_words != 0) {
while (res.pcur < res.plast) {
register size_t tt;
curlid = res.pcur->url_id;
found = 0;
p_ins = 0;
p_cmp = nwords - nonstop_words;
res.pchecked = res.pcur; /******* ? *****/
for (tt = 0; tt < nwords - nonstop_words; tt++) pos_real[tt] = 0;
for (tt = 0; (tt < nonstop_words) && (res.pcur < res.plast) && (res.pcur->url_id == curlid) ; tt++) {
order_real[p_ins] = Res->WWList.Word[DPS_WRDNUM(res.pcur->coord)].order;
pos_real[p_ins] = DPS_WRDPOS(res.pcur->coord);
while((res.pcur < res.plast) && (pos_real[p_ins] == DPS_WRDPOS(res.pcur->coord))) res.pcur++;
p_ins++; p_cmp++;
p_ins %= nwords;
p_cmp %= nwords;
}
if (tt == nonstop_words) {
/* [[[[[[ */
found = 1;
for (tt = 0; tt < nonstop_words; tt++) {
if (order_real[(p_cmp + tt) % nwords] != order_ideal[tt]) {
found = 0; break;
}
if (gap_ahead[tt] && (tt + gap_ahead[tt] < nwords) &&
(pos_real[(p_cmp + tt) % nwords] + gap_ahead[tt] != pos_real[(p_cmp + tt + 1) % nwords])) {
found = 0; break;
}
}
if (found) {
while((res.pchecked < res.pcur) /*&& (res.pchecked->url_id == curlid)*/) {
*w = *res.pchecked;
w++; res.pchecked++;
}
res.pcur = res.pchecked;
} else {
res.pchecked = res.pcur;
}
/* ]]]]]] */
}
while (/*(found == 0) &&*/ (res.pcur < res.plast) && (res.pcur->url_id == curlid)) {
order_real[p_ins] = Res->WWList.Word[DPS_WRDNUM(res.pcur->coord)].order;
pos_real[p_ins] = DPS_WRDPOS(res.pcur->coord);
while((res.pcur < res.plast) && (pos_real[p_ins] == DPS_WRDPOS(res.pcur->coord))) res.pcur++;
p_ins++; p_cmp++;
p_ins %= nwords;
p_cmp %= nwords;
/* [[[[[[ */
found = 1;
for (tt = 0; tt < nonstop_words; tt++) {
if (order_real[(p_cmp + tt) % nwords] != order_ideal[tt]) {
found = 0; break;
}
if (gap_ahead[tt] && (tt + gap_ahead[tt] < nwords) &&
(pos_real[(p_cmp + tt) % nwords] + gap_ahead[tt] != pos_real[(p_cmp + tt + 1) % nwords])) {
found = 0; break;
}
}
if (found) {
while((res.pchecked < res.pcur) /*&& (res.pchecked->url_id == curlid)*/) {
*w = *res.pchecked;
w++; res.pchecked++;
}
res.pcur = res.pchecked;
} else {
res.pchecked = res.pcur;
}
/* ]]]]]] */
}
}
}
res.count = w - res.pbegin;
DPS_FREE(pos_real);
}
}
#ifdef DEBUG_BOOL
DpsLog(query, DPS_LOG_EXTRA,
"Perform <{%d}:%d:%d> ->{%d}", x1 ? x1->count : 0, x1 ? x1->order_from : 0, x1 ? x1->order_to : 0, res.count);
#endif
rc = PUSHARG(s, &res);
break;
case DPS_STACK_OR:
x1 = POPARG(s);
x2 = POPARG(s);
if (x2 == NULL || x1 == NULL) {
if (x1 != NULL) { res = *x1; x1 = NULL; }
if (x2 != NULL) { res = *x2; x2 = NULL; }
} else {
#ifdef DEBUG_BOOL
/* printBoolRes(query, x1);*/
DpsLog(query, DPS_LOG_EXTRA, "^^^");
/* printBoolRes(query, x2);*/
#endif
res.order_from = (x1->order_from <= x2->order_from) ? x1->order_from : x2->order_from;
res.order_to = (x1->order_to >= x2->order_to) ? x1->order_to : x2->order_to;
if (DPS_OK != proceedOR(query, &res, x1, x2)) return DPS_ERROR;
DpsStackItemFree(x1); DpsStackItemFree(x2);
res.count = res.pcur - res.pbegin;
{ register size_t tt; int x1origin=0, x2origin=0;
for (tt = x1->order_from; tt <= x1->order_to; tt++) {
#ifdef DEBUG_BOOL
DpsLog(query, DPS_LOG_EXTRA, "\t\t\t\tx1order_origin[%d].%x ", tt, Res->items[tt].origin /*order_origin*/);
#endif
if (Res->items[tt].origin /*order_origin*/ & DPS_WORD_ORIGIN_STOP) {
x1origin = Res->items[tt].origin /*order_origin*/;
break;
}
}
for (tt = x2->order_from; tt <= x2->order_to; tt++) {
#ifdef DEBUG_BOOL
DpsLog(query, DPS_LOG_EXTRA, "\t\t\t\tx2order_origin[%d].%x ", tt, Res->items[tt].origin /*order_origin*/);
#endif
if (Res->items[tt].origin /*order_origin*/ & DPS_WORD_ORIGIN_STOP) {
x2origin = Res->items[tt].origin /*order_origin*/;
break;
}
}
x1origin = x1->origin; x2origin = x2->origin;
#ifdef DEBUG_BOOL
DpsLog(query, DPS_LOG_EXTRA, "\t\t\t\tx1origin.%x x2origin.%x", x1origin, x2origin);
#endif
if ((x1origin & (DPS_WORD_ORIGIN_STOP | DPS_WORD_ORIGIN_QUERY)) == (DPS_WORD_ORIGIN_STOP|DPS_WORD_ORIGIN_QUERY)) {
res.origin = x1origin;
} else
if ((x2origin & (DPS_WORD_ORIGIN_STOP | DPS_WORD_ORIGIN_QUERY)) == (DPS_WORD_ORIGIN_STOP|DPS_WORD_ORIGIN_QUERY)) {
res.origin = x2origin;
} else
if (((x1origin & DPS_WORD_ORIGIN_STOP) && (x2origin & DPS_WORD_ORIGIN_STOP)) ||
( (res.count == 0) &&
((x1origin & DPS_WORD_ORIGIN_STOP) || (x2origin & DPS_WORD_ORIGIN_STOP))))
res.origin = DPS_WORD_ORIGIN_STOP;
if ((x1origin & DPS_WORD_ORIGIN_ACRONYM) && (x1origin & DPS_WORD_ORIGIN_ACRONYM))
res.origin |= DPS_WORD_ORIGIN_ACRONYM;
}
}
#ifdef DEBUG_BOOL
DpsLog(query, DPS_LOG_EXTRA, "Perform {%d}.%x | {%d}.%x -> {%d}.%x",
(x1) ? x1->count:-1, (x1)?x1->origin:-1, (x2)?x2->count : -1, (x2) ? x2->origin : -1, res.count, res.origin);
/* printBoolRes(query, &res);*/
DpsLog(query, DPS_LOG_EXTRA, "===");
#endif
rc = PUSHARG(s, &res);
break;
case DPS_STACK_NEAR:
x1 = POPARG(s);
x2 = POPARG(s);
if (x2 == NULL || x1 == NULL) {
if (x1 != NULL) { res = *x1; x1 = NULL; }
if (x2 != NULL) { res = *x2; x2 = NULL; }
} else {
res.order_from = (x1->order_from <= x2->order_from) ? x1->order_from : x2->order_from;
res.order_to = (x1->order_to >= x2->order_to) ? x1->order_to : x2->order_to;
if ((x1->origin & DPS_WORD_ORIGIN_STOP) && (x2->origin & DPS_WORD_ORIGIN_STOP) ) {
if (DPS_OK != proceedOR(query, &res, x1, x2)) return DPS_ERROR;
res.origin = DPS_WORD_ORIGIN_STOP;
} else if (x2->origin & DPS_WORD_ORIGIN_STOP) {
if (DPS_OK != proceedSTOP(query, &res, x1, x2)) return DPS_ERROR;
} else if (x1->origin & DPS_WORD_ORIGIN_STOP ) {
if (DPS_OK != proceedSTOP(query, &res, x2, x1)) return DPS_ERROR;
} else if (!((x1->cmd & DPS_STACK_WORD_NOT) && (x2->cmd & DPS_STACK_WORD_NOT))) {
#ifdef DEBUG_BOOL
/* printBoolRes(query, x1);*/
DpsLog(query, DPS_LOG_EXTRA, "^^^");
DpsLog(query, DPS_LOG_DEBUG, "x1.NOT: %d x2.NOT: %d", x1->cmd & DPS_STACK_WORD_NOT, x2->cmd & DPS_STACK_WORD_NOT);
/* printBoolRes(query, x2);*/
#endif
res.pbegin = res.pcur = (DPS_URL_CRD_DB*)DpsMalloc((x1->count + x2->count + 1) * sizeof(DPS_URL_CRD_DB));
if (res.pbegin == NULL) return DPS_ERROR;
x1->pcur = x1->pbegin; x1->plast = x1->pbegin + x1->count;
x2->pcur = x2->pbegin; x2->plast = x2->pbegin + x2->count;
if (x1->cmd & DPS_STACK_WORD_NOT) {
register DPS_STACK_ITEM *t = x1;
x1 = x2; x2 = t;
}
while (x1->pcur < x1->plast && x2->pcur < x2->plast) {
while ((x2->pcur < x2->plast) && (x2->pcur->url_id < x1->pcur->url_id)) x2->pcur++;
if (x2->pcur >= x2->plast) break;
if (x2->pcur->url_id == x1->pcur->url_id) {
dps_uint4 pos1 = DPS_WRDPOS(x1->pcur->coord);
dps_uint4 pos2 = DPS_WRDPOS(x2->pcur->coord);
register urlid_t curlid = x1->pcur->url_id;
if (pos1 > pos2) { found = ((pos2 + 16) >= pos1);
} else { found = ((pos1 + 16) >= pos2);
}
x1->pchecked = x1->pcur; x2->pchecked = x2->pcur;
while ((!found) && (x1->pchecked < x1->plast) && (x2->pchecked < x2->plast)
/*&& (x1->pchecked->url_id == x2->pchecked->url_id)*/ ) {
if (x1->pchecked->coord <= x2->pchecked->coord) {
x1->pchecked++;
pos1 = DPS_WRDPOS(x1->pchecked->coord);
if (x1->pchecked->url_id != curlid) break;
} else {
x2->pchecked++;
pos2 = DPS_WRDPOS(x2->pchecked->coord);
if (x2->pchecked->url_id != curlid) break;
}
if (pos1 > pos2) { found = ((pos2 + 16) >= pos1);
} else { found = ((pos1 + 16) >= pos2);
}
}
if (x2->cmd & DPS_STACK_WORD_NOT || x1->cmd & DPS_STACK_WORD_NOT) found = !found;
if (found) {
while ((x1->pcur < x1->plast) && (x2->pcur < x2->plast) /*&& (x1->pcur->url_id == x2->pcur->url_id)*/) {
if (x1->pcur->coord <= x2->pcur->coord) {
*res.pcur = *x1->pcur;
res.pcur++; x1->pcur++; if (x1->pcur->url_id != curlid) break;
} else {
*res.pcur = *x2->pcur;
res.pcur++; x2->pcur++; if (x2->pcur->url_id != curlid) break;
}
}
while ((x1->pcur < x1->plast) && (x1->pcur->url_id == curlid)) {
*res.pcur = *x1->pcur;
res.pcur++; x1->pcur++;
}
while ((x2->pcur < x2->plast) && (x2->pcur->url_id == curlid)) {
*res.pcur = *x2->pcur;
res.pcur++; x2->pcur++;
}
} else {
x1->pcur = x1->pchecked; x2->pcur = x2->pchecked;
while ((x1->pcur < x1->plast) && (x1->pcur->url_id == curlid)) x1->pcur++;
while ((x2->pcur < x2->plast) && (x2->pcur->url_id == curlid)) x2->pcur++;
}
} else {
register DPS_STACK_ITEM *t = x1;
x1 = x2; x2 = t;
}
}
}
}
DpsStackItemFree(x1); DpsStackItemFree(x2);
res.count = res.pcur - res.pbegin;
#ifdef DEBUG_BOOL
DpsLog(query, DPS_LOG_EXTRA,"Perform {%d}.%x NEAR {%d}.%x - > %d.%d",
(x1)?x1->count:-1, (x1)?x1->origin:-1, (x2) ? x2->count : -1, (x2) ? x2->origin: - 1, res.count, res.origin);
/* printBoolRes(query, &res);*/
DpsLog(query, DPS_LOG_EXTRA, "===");
#endif
rc = PUSHARG(s, &res);
break;
case DPS_STACK_ANYWORD:
x1 = POPARG(s);
x2 = POPARG(s); flag1 = 0;
if (x2 == NULL || x1 == NULL) {
if (x1 != NULL) { res = *x1; x1 = NULL; }
if (x2 != NULL) { res = *x2; x2 = NULL; }
} else {
res.order_from = (x1->order_from <= x2->order_from) ? x1->order_from : x2->order_from;
res.order_to = (x1->order_to >= x2->order_to) ? x1->order_to : x2->order_to;
if ((x1->origin & DPS_WORD_ORIGIN_STOP) && (x2->origin & DPS_WORD_ORIGIN_STOP) ) {
if (DPS_OK != proceedOR(query, &res, x1, x2)) return DPS_ERROR;
res.origin = DPS_WORD_ORIGIN_STOP;
} else if (x2->origin & DPS_WORD_ORIGIN_STOP) {
if (DPS_OK != proceedSTOP(query, &res, x1, x2)) return DPS_ERROR;
} else if (x1->origin & DPS_WORD_ORIGIN_STOP ) {
if (DPS_OK != proceedSTOP(query, &res, x2, x1)) return DPS_ERROR;
} else if (!((x1->cmd & DPS_STACK_WORD_NOT) && (x2->cmd & DPS_STACK_WORD_NOT))) {
res.pbegin = res.pcur = (DPS_URL_CRD_DB*)DpsMalloc((x1->count + x2->count + 1) * sizeof(DPS_URL_CRD_DB));
if (res.pbegin == NULL) return DPS_ERROR;
x1->pcur = x1->pbegin; x1->plast = x1->pbegin + x1->count;
x2->pcur = x2->pbegin; x2->plast = x2->pbegin + x2->count;
if (x1->cmd & DPS_STACK_WORD_NOT) {
register DPS_STACK_ITEM *t = x1;
x1 = x2; x2 = t; flag1 = !flag1;
}
while (x1->pcur < x1->plast && x2->pcur < x2->plast) {
while ((x2->pcur < x2->plast) && (x2->pcur->url_id < x1->pcur->url_id)) x2->pcur++;
if (x2->pcur >= x2->plast) break;
if (x2->pcur->url_id == x1->pcur->url_id) {
dps_int4 pos1 = (dps_int4)DPS_WRDPOS(x1->pcur->coord);
dps_int4 pos2 = (dps_int4)DPS_WRDPOS(x2->pcur->coord);
register urlid_t curlid = x1->pcur->url_id;
found = ((flag1) ? ((pos1 + 2) == pos2) : ((pos2 + 2) == pos1));
x1->pchecked = x1->pcur; x2->pchecked = x2->pcur;
while ((!found) && (x1->pchecked < x1->plast) && (x2->pchecked < x2->plast)
&& (x1->pchecked->url_id == x2->pchecked->url_id)) {
if (x1->pchecked->coord <= x2->pchecked->coord) {
x1->pchecked++;
pos1 = (dps_int4)DPS_WRDPOS(x1->pchecked->coord);
} else {
x2->pchecked++;
pos2 = (dps_int4)DPS_WRDPOS(x1->pchecked->coord);
}
found = ((flag1) ? ((pos1 + 2) == pos2) : ((pos2 + 2) == pos1));
}
if (x2->cmd & DPS_STACK_WORD_NOT || x1->cmd & DPS_STACK_WORD_NOT) found = !found;
if (found) {
while ((x1->pcur < x1->plast) && (x2->pcur < x2->plast) && (x1->pcur->url_id == x2->pcur->url_id)) {
if (x1->pcur->coord <= x2->pcur->coord) {
*res.pcur = *x1->pcur;
res.pcur++; x1->pcur++;
} else {
*res.pcur = *x2->pcur;
res.pcur++; x2->pcur++;
}
}
while ((x1->pcur < x1->plast) && (x1->pcur->url_id == curlid)) {
*res.pcur = *x1->pcur;
res.pcur++; x1->pcur++;
}
while ((x2->pcur < x2->plast) && (x2->pcur->url_id == curlid)) {
*res.pcur = *x2->pcur;
res.pcur++; x2->pcur++;
}
} else {
x1->pcur = x1->pchecked; x2->pcur = x2->pchecked;
while ((x1->pcur < x1->plast) && (x1->pcur->url_id == curlid)) x1->pcur++;
while ((x2->pcur < x2->plast) && (x2->pcur->url_id == curlid)) x2->pcur++;
}
} else {
register DPS_STACK_ITEM *t = x1;
x1 = x2; x2 = t; flag1 = !flag1;
}
}
}
}
DpsStackItemFree(x1); DpsStackItemFree(x2);
res.count = res.pcur - res.pbegin;
#ifdef DEBUG_BOOL
DpsLog(query, DPS_LOG_EXTRA, "Perform {%d} ANYWORD {%d} - > %d",
(x1) ? x1->count : -1, (x2) ? x2->count : -1, res.count);
#endif
rc = PUSHARG(s, &res);
break;
case DPS_STACK_AND:
x1 = POPARG(s);
x2 = POPARG(s);
if (x2 == NULL || x1 == NULL) {
if (x1 != NULL) { res = *x1; x1 = NULL; }
if (x2 != NULL) { res = *x2; x2 = NULL; }
} else {
res.order_from = (x1->order_from <= x2->order_from) ? x1->order_from : x2->order_from;
res.order_to = (x1->order_to >= x2->order_to) ? x1->order_to : x2->order_to;
if ((x1->origin & DPS_WORD_ORIGIN_STOP) && (x2->origin & DPS_WORD_ORIGIN_STOP) ) {
if (DPS_OK != proceedOR(query, &res, x1, x2)) return DPS_ERROR;
res.origin = DPS_WORD_ORIGIN_STOP;
} else if (x2->origin & DPS_WORD_ORIGIN_STOP) {
if (DPS_OK != proceedSTOP(query, &res, x1, x2)) return DPS_ERROR;
} else if (x1->origin & DPS_WORD_ORIGIN_STOP ) {
if (DPS_OK != proceedSTOP(query, &res, x2, x1)) return DPS_ERROR;
} else if (!((x1->cmd & DPS_STACK_WORD_NOT) && (x2->cmd & DPS_STACK_WORD_NOT))) {
res.pbegin = res.pcur = (DPS_URL_CRD_DB*)DpsMalloc((x1->count + x2->count + 1) * sizeof(DPS_URL_CRD_DB));
if (res.pbegin == NULL) return DPS_ERROR;
x1->pcur = x1->pbegin; x1->plast = x1->pbegin + x1->count;
x2->pcur = x2->pbegin; x2->plast = x2->pbegin + x2->count;
if (x1->cmd & DPS_STACK_WORD_NOT) {
register DPS_STACK_ITEM *t = x1;
x1 = x2; x2 = t;
}
if (x2->cmd & DPS_STACK_WORD_NOT) {
while (x1->pcur < x1->plast && x2->pcur < x2->plast) {
while ((x1->pcur < x1->plast) && (x1->pcur->url_id < x2->pcur->url_id)) {
*res.pcur = *x1->pcur;
res.pcur++; x1->pcur++;
}
while ((x2->pcur < x2->plast) && (x2->pcur->url_id < x1->pcur->url_id)) x2->pcur++;
if (x2->pcur->url_id == x1->pcur->url_id) {
register urlid_t curlid = x1->pcur->url_id;
while ((x1->pcur < x1->plast) && (x1->pcur->url_id == curlid)) x1->pcur++;
while ((x2->pcur < x2->plast) && (x2->pcur->url_id == curlid)) x2->pcur++;
}
}
while (x1->pcur < x1->plast) {
*res.pcur = *x1->pcur;
res.pcur++; x1->pcur++;
}
} else {
#if 0
{
DPS_URL_CRD_DB *w;
for (w = x1->pcur; w < x1->plast; w++) {
fprintf(stderr, "x1.url_id:%d .coord:%d\n", w->url_id, w->coord);
}
for (w = x2->pcur; w < x2->plast; w++) {
fprintf(stderr, "x2.url_id:%d .coord:%d\n", w->url_id, w->coord);
}
}
#endif
while (x1->pcur < x1->plast && x2->pcur < x2->plast) {
while ((x2->pcur < x2->plast) && (x2->pcur->url_id < x1->pcur->url_id)) x2->pcur++;
if (x2->pcur >= x2->plast) break;
if (x2->pcur->url_id == x1->pcur->url_id) {
register urlid_t curlid = x1->pcur->url_id;
while ((x1->pcur < x1->plast) && (x2->pcur < x2->plast) && (x1->pcur->url_id == x2->pcur->url_id)) {
if (x1->pcur->coord <= x2->pcur->coord) {
*res.pcur = *x1->pcur;
res.pcur++; x1->pcur++;
} else {
*res.pcur = *x2->pcur;
res.pcur++; x2->pcur++;
}
}
while ((x1->pcur < x1->plast) && (x1->pcur->url_id == curlid)) {
*res.pcur = *x1->pcur;
res.pcur++; x1->pcur++;
}
while ((x2->pcur < x2->plast) && (x2->pcur->url_id == curlid)) {
*res.pcur = *x2->pcur;
res.pcur++; x2->pcur++;
}
} else {
register DPS_STACK_ITEM *t = x1;
x1 = x2; x2 = t;
}
}
}
}
}
DpsStackItemFree(x1); DpsStackItemFree(x2);
res.count = res.pcur - res.pbegin;
#ifdef DEBUG_BOOL
#if 0
{
DPS_URL_CRD_DB *w = res.pbegin;
size_t q;
for (q = 0; q < res.count; q++) {
fprintf(stderr, "res.url_id:%d .coord:%d\n", w[q].url_id, w[q].coord);
}
}
#endif
DpsLog(query, DPS_LOG_EXTRA, "Perform {%d}.%x & {%d}.%x - > {%d}.%x",
(x1) ? x1->count : -1, (x1) ? x1->origin : -1, (x2) ? x2->count : -1 , (x2) ? x2->origin : -1, res.count, res.origin);
/* printBoolRes(query, &res);*/
DpsLog(query, DPS_LOG_EXTRA, "===");
#endif
rc = PUSHARG(s, &res);
break;
case DPS_STACK_NOT:
x1 = POPARG(s);
/* res = x1 ? 0 : 1; */
if (x1 != NULL) {
x1->cmd ^= DPS_STACK_WORD_NOT;
rc = PUSHARG(s, x1);
}
#ifdef DEBUG_BOOL
DpsLog(query, DPS_LOG_EXTRA, "Perform ~ {%d}", (x1) ? x1->count : -1);
#endif
break;
}
return rc;
}
static void printBoolRes(DPS_AGENT *query, DPS_STACK_ITEM *res) {
size_t i;;
for (i = 0; i < res->count; i++) {
DpsLog(query, DPS_LOG_EXTRA, "url_id: %03x coord: %08x", res->pbegin[i].url_id, res->pbegin[i].coord);
}
}
int DpsSEAMake(DPS_AGENT *Indexer, DPS_DOCUMENT *Doc, DPS_DSTR *excerpt,
const char *content_lang, size_t *indexed_size, size_t *indexed_limit,
size_t max_word_len, size_t min_word_len, int crossec, int seasec
#ifdef HAVE_ASPELL
, int have_speller, AspellSpeller *speller
#endif
) {
DPS_SENTENCELIST List;
DPS_MAPSTAT MapStat;
DPS_TEXTITEM Item;
DPS_VAR *Sec;
dpsunicode_t *sentence, *lt, savec;
double *links, *lang_cs, w;
double delta, pdiv, cur_div;
size_t l, sent_len, order;
size_t min_len = 10000000, min_pos = 0;
int it;
register size_t i, j;
#ifdef DEBUG
char lcstr[4096];
#endif
TRACE_IN(Indexer, "DpsSEAMake");
if((Sec = DpsVarListFind(&Doc->Sections, "sea"))) { /* set SEA section to NULL */
DPS_FREE(Sec->val);
DPS_FREE(Sec->txt_val);
Sec->curlen = 0;
}
bzero(&List, sizeof(List));
order = 0;
sentence = DpsUniStrTok_SEA((dpsunicode_t*)excerpt->data, <);
while(sentence) {
if (lt != NULL) { savec = *lt; *lt = 0; }
#ifdef DEBUG
DpsConv(&Indexer->uni_lc, lcstr, sizeof(lcstr), (char*)sentence, sizeof(dpsunicode_t) * (DpsUniLen(sentence) + 1));
fprintf(stderr, "Sentence.%d: %s\n", List.nitems, lcstr);
#endif
if ((sent_len = DpsUniLen(sentence)) >= Indexer->Flags.SEASentenceMinLength) {
j = 1;
for (i = 0; i < List.nitems; i++) {
if (DpsUniStrCmp(sentence, List.Sent[i].sentence) == 0) {
j = 0; break;
}
}
if (j) {
if ( List.nitems < Indexer->Flags.SEASentences ) {
if (List.nitems == List.mitems) {
List.mitems += 16;
List.Sent = (DPS_SENTENCE*)DpsRealloc(List.Sent, List.mitems * sizeof(DPS_SENTENCE));
if (List.Sent == NULL) { TRACE_OUT(Indexer); return DPS_ERROR;}
}
List.Sent[List.nitems].sentence = DpsUniDup(sentence);
List.Sent[List.nitems].len = sent_len;
List.Sent[List.nitems].order = order++;
sentence = DpsUniDup(sentence);
DpsUniStrToLower(sentence);
bzero(&List.Sent[List.nitems].LangMap, sizeof(DPS_LANGMAP));
DpsBuildLangMap(&List.Sent[List.nitems].LangMap, (char*)sentence, sent_len * sizeof(dpsunicode_t), 0, 0);
if (sent_len < min_len) { min_len = sent_len; min_pos = List.nitems; }
List.nitems++;
DPS_FREE(sentence);
} else if (sent_len > min_len) {
DPS_FREE(List.Sent[min_pos].sentence);
List.Sent[min_pos].sentence = DpsUniDup(sentence);
List.Sent[min_pos].len = sent_len;
List.Sent[min_pos].order = order++;
sentence = DpsUniDup(sentence);
DpsUniStrToLower(sentence);
bzero(&List.Sent[min_pos].LangMap, sizeof(DPS_LANGMAP));
DpsBuildLangMap(&List.Sent[min_pos].LangMap, (char*)sentence, sent_len * sizeof(dpsunicode_t), 0, 0);
DPS_FREE(sentence);
min_len = List.Sent[0].len; min_pos = 0;
for(i = 1; i < List.nitems; i++) if (List.Sent[i].len < min_len) { min_len = List.Sent[i].len; min_pos = i; }
}
}
}
#ifdef DEBUG
fprintf(stderr, "Sent. len.:%d, Min.allowed: %d\n", sent_len, Indexer->Flags.SEASentenceMinLength);
#endif
if (lt != NULL) *lt = savec;
sentence = DpsUniStrTok_SEA(NULL, <);
}
DpsLog(Indexer, DPS_LOG_DEBUG, "SEA sentences: %d", List.nitems);
if (List.nitems < 4) {
for (i = 0; i < List.nitems; i++) DPS_FREE(List.Sent[i].sentence);
DPS_FREE(List.Sent);
TRACE_OUT(Indexer);
return DPS_OK;
}
links = (double*)DpsMalloc(sizeof(double) * List.nitems * List.nitems);
lang_cs = (double*)DpsMalloc(sizeof(double) * List.nitems);
/*
k ot
links[i * List.nitems + j]
*/
if (links != NULL && lang_cs != NULL) {
for (i = 0; i < List.nitems; i++) {
DpsPrepareLangMap(&List.Sent[i].LangMap);
}
for (i = 0; i < List.nitems; i++) {
List.Sent[i].Oi = List.Sent[i].di = 0.5;
if (Doc->lang_cs_map == NULL) {
links[i * List.nitems + i] = 0.0;
} else {
MapStat.map = &List.Sent[i].LangMap;
DpsCheckLangMap6(Doc->lang_cs_map, &List.Sent[i].LangMap, &MapStat, DPS_LM_TOPCNT * DPS_LM_TOPCNT, 2 * DPS_LM_TOPCNT);
links[i * List.nitems + i] = (double)MapStat.hits / (2.0 * DPS_LM_TOPCNT) / (List.nitems + 1);
}
#ifdef DEBUG
DpsLog(Indexer, DPS_LOG_INFO, "Link %u->%u: %f [hits:%d miss:%d]", i, i, links[i * List.nitems + i], MapStat.hits, MapStat.miss);
#endif
for (j = 0; j < List.nitems; j++) {
if (j == i) continue;
MapStat.map = &List.Sent[j].LangMap;
DpsCheckLangMap6(&List.Sent[j].LangMap, &List.Sent[i].LangMap, &MapStat, DPS_LM_TOPCNT * DPS_LM_TOPCNT, 2 * DPS_LM_TOPCNT);
links[i * List.nitems + j] = (double)MapStat.hits / (2.0 * DPS_LM_TOPCNT) / (List.nitems + 1);
#ifdef DEBUG
DpsLog(Indexer, DPS_LOG_INFO, "Link %u->%u: %f [hits:%d miss:%d]", i, j, links[i * List.nitems + j], MapStat.hits, MapStat.miss);
#endif
}
}
for (l = 0; l < List.nitems; l++) {
w = 0.0;
for (i = 0; i < List.nitems; i++) {
w += links[l * List.nitems + i] * List.Sent[i].Oi;
}
w = f(w);
if (w < LOW_BORDER_EPS2) w = LOW_BORDER_EPS2;
else if (w > HI_BORDER_EPS2) w = HI_BORDER_EPS2;
List.Sent[l].di = w;
}
DpsSort(List.Sent, List.nitems, sizeof(DPS_SENTENCE), (qsort_cmp)SentCmp);
#ifdef DEBUG
DpsConv(&Indexer->uni_lc, lcstr, sizeof(lcstr), (char*)List.Sent[0].sentence, sizeof(dpsunicode_t) * (DpsUniLen(List.Sent[0].sentence) + 1));
fprintf(stderr, "Sent.0: %f %f -- %s\n", List.Sent[0].di, List.Sent[0].Oi, lcstr);
DpsConv(&Indexer->uni_lc, lcstr, sizeof(lcstr), (char*)List.Sent[1].sentence, sizeof(dpsunicode_t) * (DpsUniLen(List.Sent[1].sentence) + 1));
fprintf(stderr, "Sent.1: %f %f -- %s\n", List.Sent[1].di, List.Sent[1].Oi, lcstr);
DpsConv(&Indexer->uni_lc, lcstr, sizeof(lcstr), (char*)List.Sent[2].sentence, sizeof(dpsunicode_t) * (DpsUniLen(List.Sent[2].sentence) + 1));
fprintf(stderr, "Sent.2: %f %f -- %s\n", List.Sent[2].di, List.Sent[2].Oi, lcstr);
DpsConv(&Indexer->uni_lc, lcstr, sizeof(lcstr), (char*)List.Sent[3].sentence, sizeof(dpsunicode_t) * (DpsUniLen(List.Sent[3].sentence) + 1));
fprintf(stderr, "Sent.3: %f %f -- %s\n", List.Sent[3].di, List.Sent[3].Oi, lcstr);
DpsConv(&Indexer->uni_lc, lcstr, sizeof(lcstr), (char*)List.Sent[4].sentence, sizeof(dpsunicode_t) * (DpsUniLen(List.Sent[4].sentence) + 1));
fprintf(stderr, "Sent.4: %f %f -- %s\n", List.Sent[4].di, List.Sent[4].Oi, lcstr);
#endif
DpsSort(List.Sent, TOP_SENTENCES, sizeof(DPS_SENTENCE), (qsort_cmp)SentOrderCmp);
bzero(&Item, sizeof(Item));
Item.section = seasec;
Item.href = NULL;
Item.section_name = "sea";
for (i = 0; i < TOP_SENTENCES; i++) {
dpsunicode_t *UStr = DpsUniDup(List.Sent[i].sentence);
DpsPrepareItem(Indexer, Doc, &Item, List.Sent[i].sentence, UStr, content_lang, indexed_size, indexed_limit,
max_word_len, min_word_len, crossec
#ifdef HAVE_ASPELL
, have_speller, speller, NULL
#endif
);
DPS_FREE(UStr);
}
}
DPS_FREE(lang_cs);
DPS_FREE(links);
for (i = 0; i < List.nitems; i++) DPS_FREE(List.Sent[i].sentence);
DPS_FREE(List.Sent);
TRACE_OUT(Indexer);
return DPS_OK;
}
extern __C_LINK int __DPSCALL DpsBaseOptimize(DPS_BASE_PARAM *P, int sbase) {
struct stat sb;
urlid_t base, base_from, base_to;
long unsigned ActualSize, OriginalSize, i, nitems;
off_t pos, posold, NewItemPos, SSize;
dps_uint8 diff, gain;
double dr = 0.0, cr = 0.0;
ssize_t nread; size_t rsize;
ssize_t wr;
int OptimizeRatio, res, error_cnt;
char buffer[BUFSIZ];
DPS_BASEITEM *hTable;
DPS_SORTBASEITEM *si = NULL;
OptimizeRatio = DpsVarListFindInt(&P->A->Vars, "OptimizeRatio", 15);
P->mode = DPS_WRITE_LOCK;
if (sbase == -1) {
base_from = 0; base_to = (urlid_t)P->NFiles;
} else {
base_from = sbase; base_to = sbase + 1;
}
for (base = base_from; base < base_to; base++) {
error_cnt = 0;
gain = (dps_uint8)0;
P->rec_id = ((base & DPS_BASE_MASK) << DPS_BASE_BITS);
if (DpsBaseOpen(P, DPS_WRITE_LOCK) != DPS_OK) {
DpsLog(P->A, DPS_LOG_ERROR, "Can't open base %s/%s {%s:%d}", P->subdir, P->basename, __FILE__, __LINE__);
DpsBaseClose(P);
return DPS_ERROR;
}
if (lseek(P->Ifd, (off_t)0, SEEK_SET) == (off_t)-1) {
DpsLog(P->A, DPS_LOG_ERROR, "Can't seek %s {%s:%d}", P->Ifilename, __FILE__, __LINE__);
DpsBaseClose(P);
return DPS_ERROR;
}
if (fstat(P->Sfd, &sb) == 0) {
SSize = sb.st_size;
} else {
if ((SSize = (off_t)lseek(P->Sfd, (off_t)0, SEEK_END)) == (off_t)-1) {
DpsLog(P->A, DPS_LOG_ERROR, "Can't seek %s {%s:%d}", P->Sfilename, __FILE__, __LINE__);
DpsBaseClose(P);
return DPS_ERROR;
}
}
nitems = 0;
ActualSize = 0;
OriginalSize = 0;
while(read(P->Ifd, &P->Item, sizeof(DPS_BASEITEM)) == sizeof(DPS_BASEITEM)) {
nitems++;
if ((P->Item.rec_id != 0) && ((dps_uint8)P->Item.offset < (dps_uint8)SSize) && (P->Item.size > 0)) {
ActualSize += (long unsigned)P->Item.size;
OriginalSize += (long unsigned)(P->Item.orig_size ? P->Item.orig_size : P->Item.size);
}
}
if (ftruncate(P->Ifd, (off_t)(nitems * sizeof(DPS_BASEITEM))) != 0) {
dps_strerror(P->A, DPS_LOG_EXTRA, "ftruncate error (pos:%ld) [%s:%d]", (off_t)(nitems * sizeof(DPS_BASEITEM)), __FILE__, __LINE__);
}
dr = (nitems) ? fabs(100.0 * ((long unsigned)SSize - ActualSize) / ((double)SSize + 1.0)) : 0.0;
cr = (nitems) ? fabs(100.0 * ActualSize / (OriginalSize + 1)) : 0.0;
DpsLog(P->A, DPS_LOG_EXTRA, "Optimize: %s/%s base 0x%X, %ld recs defrag: %.2f%% Ratio: %.2f%% Data: %ld File: %ld",
P->subdir, P->basename, P->FileNo, nitems, dr, cr, ActualSize, (long)SSize);
if ((dr >= (double)OptimizeRatio) || (ActualSize == 0 && SSize != 0)) {
si = (DPS_SORTBASEITEM*)DpsMalloc((nitems + 1) * sizeof(DPS_SORTBASEITEM));
if (si == NULL) {
DpsLog(P->A, DPS_LOG_ERROR, "Can't alloc si (%d bytes) at {%s:%d}", (nitems + 1) * sizeof(DPS_SORTBASEITEM), __FILE__, __LINE__);
DpsBaseClose(P);
return DPS_ERROR;
}
if (lseek(P->Ifd, (off_t)0, SEEK_SET) == (off_t)-1) {
DpsLog(P->A, DPS_LOG_ERROR, "Can't seek %s {%s:%d}", P->Ifilename, __FILE__, __LINE__);
DpsBaseClose(P);
DPS_FREE(si);
return DPS_ERROR;
}
for (i = 0; (i < nitems) && (read(P->Ifd, &si[i].Item, sizeof(DPS_BASEITEM)) == sizeof(DPS_BASEITEM)); ) {
if(si[i].Item.rec_id != 0 && ((dps_uint8)si[i].Item.offset < (dps_uint8)SSize) && (si[i].Item.size > 0) && (si[i].Item.size < ActualSize) ) {
i++;
}
}
if (i < nitems) nitems = i;
if (nitems > 1) DpsSort((void*)si, (size_t)nitems, sizeof(DPS_SORTBASEITEM), cmpsi);
gain = (dps_uint8)0;
pos = (off_t)0;
posold = (off_t)0;
if (nitems > 0) {
if ((long unsigned)si[0].Item.offset < (long unsigned)SSize) {
posold = (off_t)si[0].Item.offset;
} else {
si[0].Item.offset = (off_t)0;
si[0].Item.size = 0;
}
}
if (nitems > 1) {
if (si[0].Item.size > (rsize = (size_t)(si[1].Item.offset - si[0].Item.offset))) {
DpsLog(P->A, DPS_LOG_ERROR, "si[0] size adjusted by offset: %ld -> %ld", (long)si[0].Item.size, (long)rsize);
si[0].Item.size = rsize;
error_cnt++;
}
}
if ((diff = (dps_uint8)posold) > 0) {
for(
lseek(P->Sfd, posold, SEEK_SET), rsize = 0;
(rsize < si[0].Item.size) && ((nread = read(P->Sfd, buffer,
(rsize + BUFSIZ < si[0].Item.size) ? BUFSIZ : (si[0].Item.size - rsize) )) > 0);
lseek(P->Sfd, posold, SEEK_SET)
) {
lseek(P->Sfd, pos, SEEK_SET);
(void)write(P->Sfd, buffer, (size_t)nread);
rsize += (size_t)nread;
posold += (off_t)nread;
pos += (off_t)nread;
}
si[0].Item.offset = 0;
if (rsize != si[0].Item.size) {
DpsLog(P->A, DPS_LOG_ERROR, "si[0] size adjusted by size: %ld -> %ld", (long)si[0].Item.size, (long)rsize);
si[0].Item.size = rsize;
error_cnt++;
}
gain += diff;
}
if (nitems > 0)
for (i = 0; i < nitems - 1; i++) {
if ((long unsigned)si[i + 1].Item.offset > (long unsigned)SSize) {
DpsLog(P->A, DPS_LOG_ERROR, "si[%ld] too long offset: %ld > %ld, removing", i , (long)si[i + 1].Item.offset, (long)SSize);
si[i + 1].Item.size = 0;
si[i + 1].Item.offset = si[i].Item.offset + si[i].Item.size;
error_cnt++;
} else {
pos = (off_t)(si[i].Item.offset + si[i].Item.size);
posold = (off_t)si[i + 1].Item.offset;
if (i < nitems - 2) {
if (si[i + 1].Item.size > (rsize = (size_t)(si[i + 2].Item.offset - si[i + 1].Item.offset))) {
DpsLog(P->A, DPS_LOG_ERROR, "si[%ld] size adjusted by offset: %ld -> %ld", i + 1, (long)si[i + 1].Item.size, (long)rsize );
si[i + 1].Item.size = rsize;
error_cnt++;
}
}
if ((diff = (dps_uint8)posold - (dps_uint8)pos) > 0) {
for(
lseek(P->Sfd, posold, SEEK_SET), rsize = 0;
(rsize < si[i + 1].Item.size) && ((nread = read(P->Sfd, buffer,
(rsize + BUFSIZ < si[i + 1].Item.size) ? BUFSIZ : (si[i + 1].Item.size - rsize) )) > 0);
lseek(P->Sfd, posold, SEEK_SET)
) {
lseek(P->Sfd, pos, SEEK_SET);
(void)write(P->Sfd, buffer, (size_t)nread);
rsize += (size_t)nread;
posold += (off_t)nread;
pos += (off_t)nread;
}
if (rsize != si[i + 1].Item.size) {
DpsLog(P->A, DPS_LOG_ERROR, "si[%ld] size adjusted by size: %ld -> %ld", i + 1, (long)si[i + 1].Item.size, (long)rsize);
si[i + 1].Item.size = rsize;
error_cnt++;
}
si[i + 1].Item.offset = si[i].Item.offset + si[i].Item.size;
gain += diff;
}
}
}
posold = SSize;
pos = (nitems) ? (off_t)(si[nitems - 1].Item.offset + si[nitems - 1].Item.size) : (off_t)0;
if (ftruncate(P->Sfd, (off_t)(pos)) != 0) {
dps_strerror(P->A, DPS_LOG_ERROR, "ftruncate error (pos:%ld) [%s:%d]", pos, __FILE__, __LINE__);
}
SSize = pos;
if (posold > pos) {
gain += ((dps_uint8)posold - (dps_uint8)pos);
}
/*if (gain != 0 || OptimizeRatio == 0 || error_cnt > 0)*/ {
posold = lseek(P->Ifd, (off_t)0, SEEK_END);
(void)ftruncate(P->Ifd, (off_t)0);
lseek(P->Ifd, (off_t)0, SEEK_SET);
if ((hTable = (DPS_BASEITEM *)DpsXmalloc(sizeof(DPS_BASEITEM) * DPS_HASH_PRIME)) == NULL) {
DpsLog(P->A, DPS_LOG_ERROR, "Memory alloc error hTable: %d bytes", sizeof(DPS_BASEITEM) * DPS_HASH_PRIME);
DpsBaseClose(P);
DPS_FREE(si);
return DPS_ERROR;
}
if ( (wr = write(P->Ifd, hTable, sizeof(DPS_BASEITEM) * DPS_HASH_PRIME)) != sizeof(DPS_BASEITEM) * DPS_HASH_PRIME) {
dps_strerror(P->A, DPS_LOG_ERROR, "[%s:%d] Can't set new index for file %s\nwritten %d bytes of %d",
__FILE__, __LINE__, P->Ifilename, wr, sizeof(DPS_BASEITEM) * DPS_HASH_PRIME);
DPS_FREE(hTable);
DpsBaseClose(P);
DPS_FREE(si);
return DPS_ERROR;
}
DPS_FREE(hTable);
for (i = 0; i < nitems; i++) {
if (si[i].Item.rec_id == 0 || si[i].Item.size == 0) continue;
if ((long)si[i].Item.offset > (long)SSize) {
DpsLog(P->A, DPS_LOG_ERROR, "si[%ld] too long offset: %ld > %ld, removing", i , (long)si[i].Item.offset, (long)SSize);
error_cnt++;
continue;
}
P->rec_id = si[i].Item.rec_id;
if ((res = DpsBaseSeek(P, DPS_WRITE_LOCK)) != DPS_OK) {
DpsBaseClose(P);
DPS_FREE(si);
return res;
}
if (P->Item.rec_id != P->rec_id) {
if (P->mishash && P->Item.rec_id != 0) {
if ((P->Item.next = (dps_uint8)(NewItemPos = lseek(P->Ifd, (off_t)0, SEEK_END))) == (dps_uint8)-1) {
DpsBaseClose(P);
DPS_FREE(si);
return DPS_ERROR;
}
if (lseek(P->Ifd, (off_t)P->CurrentItemPos, SEEK_SET) == (off_t)-1) {
DpsBaseClose(P);
DPS_FREE(si);
return DPS_ERROR;
}
if (write(P->Ifd, &P->Item, sizeof(DPS_BASEITEM)) != sizeof(DPS_BASEITEM)) {
DpsBaseClose(P);
DPS_FREE(si);
return DPS_ERROR;
}
P->CurrentItemPos = (dps_uint8)NewItemPos;
}
}
P->Item = si[i].Item;
P->Item.next = (off_t)0;
if (lseek(P->Ifd, (off_t)P->CurrentItemPos, SEEK_SET) == (off_t)-1) {
DpsLog(P->A, DPS_LOG_ERROR, "Can't seek %s {%s:%d}", P->Ifilename, __FILE__, __LINE__);
DpsBaseClose(P);
DPS_FREE(si);
return DPS_ERROR;
}
if (write(P->Ifd, &P->Item, sizeof(DPS_BASEITEM)) != sizeof(DPS_BASEITEM)) {
DpsLog(P->A, DPS_LOG_ERROR, "Can't write index for file %s {%s:%d}", P->Ifilename, __FILE__, __LINE__);
DpsBaseClose(P);
DPS_FREE(si);
return DPS_ERROR;
}
}
pos = lseek(P->Ifd, (off_t)0, SEEK_END);
gain += ((dps_uint8)posold - (dps_uint8)pos);
DpsLog(P->A, DPS_LOG_DEBUG, "Optimize: %s/%s base 0x%X cleaned, %ld bytes freed", P->subdir, P->basename, base, gain);
}
DPS_FREE(si);
}
if (error_cnt) base--;
DpsBaseClose(P);
}
return DPS_OK;
}
int __DPSCALL DpsSearchdURLAction(DPS_AGENT *A, DPS_DOCUMENT *D, int cmd, void *db) {
DPS_DB *searchd = db;
DPS_SEARCHD_PACKET_HEADER hdr;
char *buf;
ssize_t nsent, nrecv;
int done = 0;
char *msg = NULL;
char *dinfo = NULL;
int rc=DPS_OK;
TRACE_IN(A, "DpsSearchdURLAction");
if (cmd != DPS_URL_ACTION_DOCCOUNT) {
DpsLog(A, DPS_LOG_ERROR, "searchd: unsupported URL action");
TRACE_OUT(A);
return DPS_ERROR;
}
hdr.cmd = DPS_SEARCHD_CMD_URLACTION;
hdr.len = sizeof(int);
if ((buf = (char*)DpsMalloc(hdr.len + 1)) == NULL) {
DpsLog(A, DPS_LOG_ERROR, "Out of memory");
TRACE_OUT(A);
return DPS_ERROR;
}
*((int*)buf) = cmd;
nsent = DpsSearchdSendPacket(searchd->searchd, &hdr, buf);
DPS_FREE(buf);
while(!done) {
nrecv = DpsRecvall(searchd->searchd, &hdr, sizeof(hdr), 360);
if(nrecv != sizeof(hdr)){
DpsLog(A, DPS_LOG_ERROR, "Received incomplete header from searchd (%d bytes)", (int)nrecv);
TRACE_OUT(A);
return(DPS_ERROR);
}else{
#ifdef DEBUG_SDP
DpsLog(A, DPS_LOG_ERROR, "Received header cmd=%d len=%d\n", hdr.cmd, hdr.len);
#endif
}
switch(hdr.cmd){
case DPS_SEARCHD_CMD_ERROR:
msg = (char*)DpsMalloc(hdr.len + 1);
if (msg == NULL) {
done=1;
break;
}
nrecv = DpsRecvall(searchd->searchd, msg, hdr.len, 360);
msg[(nrecv >= 0) ? nrecv : 0] = '\0';
sprintf(A->Conf->errstr, "Searchd error: '%s'", msg);
rc=DPS_OK;
DPS_FREE(msg);
done=1;
break;
case DPS_SEARCHD_CMD_MESSAGE:
msg=(char*)DpsMalloc(hdr.len+1);
if (msg == NULL) {
done=1;
break;
}
nrecv = DpsRecvall(searchd->searchd, msg, hdr.len, 360);
msg[(nrecv >= 0) ? nrecv : 0] = '\0';
#ifdef DEBUG_SDP
DpsLog(A, DPS_LOG_ERROR, "Message from searchd: '%s'\n",msg);
#endif
DPS_FREE(msg);
break;
case DPS_SEARCHD_CMD_DOCCOUNT:
dinfo=(char*)DpsMalloc(hdr.len+1);
if (dinfo == NULL) {
done=1;
break;
}
nrecv = DpsRecvall(searchd->searchd, dinfo, hdr.len, 360);
dinfo[(nrecv >= 0) ? nrecv : 0] = '\0';
A->doccount += *((int *)dinfo);
#ifdef DEBUG_SDP
DpsLog(A, DPS_LOG_DEBUG, "Received DOCCOUNT size=%d doccount=%d(+%s)\n", hdr.len, A->doccount, dinfo);
#endif
DPS_FREE(dinfo);
done=1;
break;
default:
sprintf(A->Conf->errstr, "Unknown searchd response: cmd=%d len=%d", hdr.cmd, hdr.len);
rc=DPS_ERROR;
done = 1;
break;
}
}
TRACE_OUT(A);
return rc;
}
int DpsCloneListSearchd(DPS_AGENT *Indexer, DPS_DOCUMENT *Doc, DPS_RESULT *Res, DPS_DB *db) {
DPS_SEARCHD_PACKET_HEADER hdr;
ssize_t nsent,nrecv;
char *msg = NULL, *dinfo = NULL;
char *tok, *lt;
char buf[128];
int done = 0;
int rc = DPS_OK;
TRACE_IN(Indexer, "DpsCloneListSearchd");
dps_snprintf(buf, 128, "%s", DpsVarListFindStr(&Doc->Sections, "DP_ID", "0"));
hdr.cmd = DPS_SEARCHD_CMD_CLONES;
hdr.len = dps_strlen(buf);
nsent = DpsSearchdSendPacket(db->searchd, &hdr, buf);
while(!done){
nrecv = DpsRecvall(db->searchd, &hdr, sizeof(hdr), 360);
if(nrecv != sizeof(hdr)){
DpsLog(Indexer, DPS_LOG_ERROR, "Received incomplete header from searchd (%d bytes)", (int)nrecv);
TRACE_OUT(Indexer);
return(DPS_ERROR);
}else{
#ifdef DEBUG_SDP
DpsLog(Indexer, DPS_LOG_DEBUG, "Received header cmd=%d len=%d\n", hdr.cmd, hdr.len);
#endif
}
switch(hdr.cmd){
case DPS_SEARCHD_CMD_ERROR:
msg = (char*)DpsMalloc(hdr.len + 1);
if (msg == NULL) {
done=1;
break;
}
nrecv = DpsRecvall(db->searchd, msg, hdr.len, 360);
msg[(nrecv >= 0) ? nrecv : 0] = '\0';
sprintf(Indexer->Conf->errstr, "Searchd error: '%s'", msg);
rc = DPS_ERROR;
DPS_FREE(msg);
done = 1;
break;
case DPS_SEARCHD_CMD_DOCINFO:
dinfo = (char*)DpsMalloc(hdr.len + 1);
if (dinfo == NULL) {
done=1;
break;
}
nrecv = DpsRecvall(db->searchd, dinfo, hdr.len, 360);
dinfo[(nrecv >= 0) ? nrecv : 0] = '\0';
#ifdef DEBUG_SDP
DpsLog(Indexer, DPS_LOG_DEBUG, "Received DOCINFO size=%d buf=%s\n", hdr.len, dinfo);
#endif
if (strcasecmp(dinfo, "nocloneinfo") != 0) {
tok = dps_strtok_r(dinfo, "\r\n", <, NULL);
while(tok){
DPS_DOCUMENT *D;
size_t nd = Res->num_rows++;
Res->Doc = (DPS_DOCUMENT*)DpsRealloc(Res->Doc, (Res->num_rows + 1) * sizeof(DPS_DOCUMENT));
if (Res->Doc == NULL) {
sprintf(Indexer->Conf->errstr, "Realloc error");
rc = DPS_ERROR;
break;
}
D = &Res->Doc[nd];
DpsDocInit(D);
DpsDocFromTextBuf(D, tok);
tok = dps_strtok_r(NULL, "\r\n", <, NULL);
}
}
DPS_FREE(dinfo);
done = 1;
break;
default:
sprintf(Indexer->Conf->errstr, "Unknown searchd response: cmd=%d len=%d", hdr.cmd, hdr.len);
rc = DPS_ERROR;
done = 1;
break;
}
}
TRACE_OUT(Indexer);
return rc;
}
int DpsAddStackItem(DPS_AGENT *query, DPS_RESULT *Res, DPS_PREPARE_STATE *state, char *word, dpsunicode_t *uword) {
int origin;
size_t i;
size_t wlen = (uword == NULL) ? 0 : DpsUniLen(uword);
dpshash32_t crcword = (word == NULL) ? 0 : DpsStrHash32(word);
#ifdef DEBUG_BOOL
DpsLog(query, DPS_LOG_EXTRA, "0[%d].%x %c -- %s [%x] .secno:%d\n", state->order, state->origin, item_type(state->cmd),
(word == NULL) ? "<NULL>" : word, crcword, state->secno[state->p_secno]);
#endif
if((uword != NULL) && ( DpsStopListFind(&query->Conf->StopWords, uword, state->qlang) ||
(query->WordParam.min_word_len > wlen) ||
(query->WordParam.max_word_len < wlen)) ) {
origin = state->origin | DPS_WORD_ORIGIN_STOP;
} else {
origin = state->origin;
}
if (state->cmd == DPS_STACK_WORD && !(origin & DPS_WORD_ORIGIN_QUERY)) {
for (i = 0; i < Res->nitems; i++) {
if ((Res->items[i].order == state->order) && (Res->items[i].crcword == crcword)) return DPS_OK;
}
}
if (Res->nitems >= Res->mitems - 2) {
Res->mitems += DPS_MAXSTACK;
Res->items = (DPS_STACK_ITEM*)DpsRealloc(Res->items, Res->mitems * sizeof(DPS_STACK_ITEM));
if (Res->items == NULL) {
DpsLog(query, DPS_LOG_ERROR, "Can't alloc %d bytes for %d mitems", Res->mitems * sizeof(DPS_STACK_ITEM), Res->mitems);
return DPS_ERROR;
}
}
if (Res->nitems > 0) {
if (state->cmd == DPS_STACK_OR || state->cmd == DPS_STACK_AND || state->cmd == DPS_STACK_NEAR || state->cmd == DPS_STACK_ANYWORD) {
if (Res->items[Res->nitems-1].cmd == DPS_STACK_AND || Res->items[Res->nitems-1].cmd == DPS_STACK_OR
|| Res->items[Res->nitems-1].cmd == DPS_STACK_NEAR || Res->items[Res->nitems-1].cmd == DPS_STACK_ANYWORD) {
return DPS_OK;
}
}
if ((Res->nitems > 0) && (state->cmd == DPS_STACK_WORD)
&& (
(Res->items[Res->nitems-1].cmd == DPS_STACK_WORD)
|| (Res->items[Res->nitems-1].cmd == DPS_STACK_RIGHT)
|| (Res->items[Res->nitems-1].cmd == DPS_STACK_PHRASE_RIGHT)
)) {
Res->items[Res->nitems].cmd = DPS_STACK_OR;
Res->items[Res->nitems].order = 0;
Res->items[Res->nitems].origin = 0;
Res->items[Res->nitems].count = 0;
Res->items[Res->nitems].len = 0;
Res->items[Res->nitems].crcword = 0;
Res->items[Res->nitems].word = NULL;
Res->items[Res->nitems].ulen = 0;
Res->items[Res->nitems].uword = NULL;
Res->items[Res->nitems].pbegin = NULL;
Res->items[Res->nitems].order_origin = 0;
Res->items[Res->nitems].secno = state->secno[state->p_secno];
Res->nitems++;
Res->ncmds++;
#ifdef DEBUG_BOOL
DpsLog(query, DPS_LOG_EXTRA, "1[%d].%x %c -- %s", 0, 0, item_type(DPS_STACK_OR), "<NULL>");
#endif
}
if ((Res->nitems > 0) && (state->cmd == DPS_STACK_LEFT)
&& (
(Res->items[Res->nitems-1].cmd == DPS_STACK_RIGHT)
|| (Res->items[Res->nitems-1].cmd == DPS_STACK_PHRASE_RIGHT)
)) {
Res->items[Res->nitems].cmd = state->add_cmd;
Res->items[Res->nitems].order = 0;
Res->items[Res->nitems].origin = 0;
Res->items[Res->nitems].count = 0;
Res->items[Res->nitems].len = 0;
Res->items[Res->nitems].crcword = 0;
Res->items[Res->nitems].word = NULL;
Res->items[Res->nitems].ulen = 0;
Res->items[Res->nitems].uword = NULL;
Res->items[Res->nitems].pbegin = NULL;
Res->items[Res->nitems].order_origin = 0;
Res->items[Res->nitems].secno = state->secno[state->p_secno];
Res->nitems++;
Res->ncmds++;
#ifdef DEBUG_BOOL
DpsLog(query, DPS_LOG_EXTRA, "1[%d].%x %c -- %s", 0, 0, item_type(state->add_cmd), "<NULL>");
#endif
}
}
Res->items[Res->nitems].cmd = state->cmd;
Res->items[Res->nitems].order = state->order;
Res->items[Res->nitems].order_inquery = state->order_inquery;
Res->items[Res->nitems].origin = origin;
Res->items[Res->nitems].count = 0;
Res->items[Res->nitems].len = (word == NULL) ? 0 : dps_strlen(word);
Res->items[Res->nitems].crcword = crcword;
Res->items[Res->nitems].word = (word == NULL) ? NULL : DpsStrdup(word);
Res->items[Res->nitems].ulen = wlen;
Res->items[Res->nitems].uword = (uword == NULL) ? NULL : DpsUniDup(uword);
Res->items[Res->nitems].pbegin = NULL;
Res->items[Res->nitems].order_origin = 0;
Res->items[Res->nitems].wordnum = Res->nitems;
Res->items[Res->nitems].secno = state->secno[state->p_secno];
Res->nitems++;
if (state->cmd != DPS_STACK_WORD) {
Res->ncmds++;
} else {
Res->items[state->order].order_origin |= origin;
if (state->order > Res->max_order) Res->max_order = state->order;
if (state->order_inquery > Res->max_order_inquery) Res->max_order_inquery = state->order;
}
/* if ((state->cmd == DPS_STACK_WORD) && state->order > Res->max_order) Res->max_order = state->order;*/
#ifdef DEBUG_BOOL
DpsLog(query, DPS_LOG_EXTRA, "1[%d,%d].%x %c -- %s", state->order, state->order_inquery, state->origin, item_type(state->cmd),
(word == NULL) ? "<NULL>" : word);
#endif
return DPS_OK;
}
void RelLink(DPS_AGENT *Indexer, DPS_URL *curURL, DPS_URL *newURL, char **str, int ReverseAliasFlag) {
const char *schema = newURL->schema ? newURL->schema : curURL->schema;
const char *hostname = newURL->hostname ? newURL->hostname : curURL->hostname;
const char *auth = newURL->auth ? newURL->auth : curURL->auth;
const char *path = (newURL->path && newURL->path[0]) ? newURL->path : curURL->path;
const char *fname = ((newURL->filename && newURL->filename[0]) || (newURL->path && newURL->path[0])) ?
newURL->filename : curURL->filename;
const char *query_string = newURL->query_string;
char *pathfile = (char*)DpsMalloc(dps_strlen(DPS_NULL2EMPTY(path)) + dps_strlen(DPS_NULL2EMPTY(fname)) +
dps_strlen(DPS_NULL2EMPTY(query_string)) + 5);
int cascade;
DPS_MATCH *Alias;
char *alias = NULL;
size_t aliassize, nparts = 10;
DPS_MATCH_PART Parts[10];
if (newURL->hostinfo == NULL) newURL->charset_id = curURL->charset_id;
if (pathfile == NULL) return;
/* sprintf(pathfile, "/%s%s%s", DPS_NULL2EMPTY(path), DPS_NULL2EMPTY(fname), DPS_NULL2EMPTY(query_string));*/
pathfile[0] = '/';
dps_strcpy(pathfile + 1, DPS_NULL2EMPTY(path)); dps_strcat(pathfile, DPS_NULL2EMPTY(fname)); dps_strcat(pathfile, DPS_NULL2EMPTY(query_string));
DpsURLNormalizePath(pathfile);
if (!strcasecmp(DPS_NULL2EMPTY(schema), "mailto")
|| !strcasecmp(DPS_NULL2EMPTY(schema), "javascript")
|| !strcasecmp(DPS_NULL2EMPTY(schema), "feed")
) {
*str = (char*)DpsMalloc(dps_strlen(DPS_NULL2EMPTY(schema)) + dps_strlen(DPS_NULL2EMPTY(newURL->specific)) + 4);
if (*str == NULL) return;
/* sprintf(*str, "%s:%s", DPS_NULL2EMPTY(schema), DPS_NULL2EMPTY(newURL->specific));*/
dps_strcpy(*str, DPS_NULL2EMPTY(schema)); dps_strcat(*str, ":"); dps_strcat(*str, DPS_NULL2EMPTY(newURL->specific));
} else if(/*!strcasecmp(DPS_NULL2EMPTY(schema), "file") ||*/ !strcasecmp(DPS_NULL2EMPTY(schema), "htdb")) {
*str = (char*)DpsMalloc(dps_strlen(DPS_NULL2EMPTY(schema)) + dps_strlen(pathfile) + 4);
if (*str == NULL) return;
/* sprintf(*str, "%s:%s", DPS_NULL2EMPTY(schema), pathfile);*/
dps_strcpy(*str, DPS_NULL2EMPTY(schema)); dps_strcat(*str, ":"); dps_strcat(*str, pathfile);
}else{
*str = (char*)DpsMalloc(dps_strlen(DPS_NULL2EMPTY(schema)) + dps_strlen(pathfile) + dps_strlen(DPS_NULL2EMPTY(hostname)) + dps_strlen(DPS_NULL2EMPTY(auth)) + 8);
if (*str == NULL) return;
/* sprintf(*str, "%s://%s%s", DPS_NULL2EMPTY(schema), DPS_NULL2EMPTY(hostinfo), pathfile);*/
dps_strcpy(*str, DPS_NULL2EMPTY(schema)); dps_strcat(*str, "://");
if (auth) {
dps_strcat(*str, auth); dps_strcat(*str,"@");
}
dps_strcat(*str, DPS_NULL2EMPTY(hostname)); dps_strcat(*str, pathfile);
}
if(!strncmp(*str, "ftp://", 6) && (strstr(*str, ";type=")))
*(strstr(*str, ";type")) = '\0';
DPS_FREE(pathfile);
if (ReverseAliasFlag) {
const char *alias_prog = DpsVarListFindStr(&Indexer->Vars, "ReverseAliasProg", NULL);
if (alias_prog) {
int result;
aliassize = 256 + 2 * dps_strlen(*str);
alias = (char*)DpsRealloc(alias, aliassize);
if (alias == NULL) {
DpsLog(Indexer, DPS_LOG_ERROR, "No memory (%d bytes). %s line %d", aliassize, __FILE__, __LINE__);
goto ret;
}
alias[0] = '\0';
result = DpsAliasProg(Indexer, alias_prog, *str, alias, aliassize - 1);
DpsLog(Indexer, DPS_LOG_EXTRA, "ReverseAliasProg result: '%s'", alias);
if(result != DPS_OK) goto ret;
DPS_FREE(*str);
*str = (char*)DpsStrdup(alias);
}
for(cascade = 0; ((Alias=DpsMatchListFind(&Indexer->Conf->ReverseAliases,*str,nparts,Parts))) && (cascade < 1024); cascade++) {
aliassize = dps_strlen(Alias->arg) + dps_strlen(Alias->pattern) + dps_strlen(*str) + 128;
alias = (char*)DpsRealloc(alias, aliassize);
if (alias == NULL) {
DpsLog(Indexer, DPS_LOG_ERROR, "No memory (%d bytes). %s line %d", aliassize, __FILE__, __LINE__);
goto ret;
}
DpsMatchApply(alias,aliassize,*str,Alias->arg,Alias,nparts,Parts);
if(alias[0]){
DpsLog(Indexer,DPS_LOG_DEBUG,"ReverseAlias%d: pattern:%s, arg:%s -> '%s'", cascade, Alias->pattern, Alias->arg, alias);
DPS_FREE(*str);
*str = (char*)DpsStrdup(alias);
} else break;
if (Alias->last) break;
}
}
ret:
DPS_FREE(alias);
}
__C_LINK int __DPSCALL DpsCacheMakeIndexes(DPS_AGENT *Indexer, DPS_DB *db) {
DPS_UINT8URLIDLIST L8;
DPS_UINT4URLIDLIST L4;
DPS_VARLIST *v = &Indexer->Conf->Vars;
size_t i, r;
char *ind, *nm, *lfname;
bzero(&L4, sizeof(DPS_UINT4URLIDLIST));
bzero(&L8, sizeof(DPS_UINT8URLIDLIST));
r = (size_t) 'l';
for (i = 0; i < v->Root[r].nvars; i++) {
if (!strncasecmp("Limit-", v->Root[r].Var[i].name, 6)) {
ind = v->Root[r].Var[i].val;
lfname = v->Root[r].Var[i].name;
nm = lfname + 6;
if (!strcasecmp(ind, "category")) {
/* To see the URL being indexed in "ps" output on xBSD */
dps_setproctitle("[%d] Category index creation", Indexer->handle);
DpsLog(Indexer, DPS_LOG_EXTRA, "Creating category index");
if (DPS_OK == DpsLimit8(Indexer, &L8, "Category", DPS_IFIELD_TYPE_HEX8STR, db)) {
MakeNestedIndex(Indexer, &L8, DPS_LIMFNAME_CAT, db);
}
} else if (!strcasecmp(ind, "tag")) {
/* To see the URL being indexed in "ps" output on xBSD */
dps_setproctitle("[%d] Tag index creation", Indexer->handle);
DpsLog(Indexer, DPS_LOG_EXTRA, "Creating tag index");
if (DPS_OK == DpsLimit4(Indexer, &L4, "Tag", DPS_IFIELD_TYPE_STRCRC32, db)) {
MakeLinearIndex(Indexer, &L4, DPS_LIMFNAME_TAG, db);
}
} else if (!strcasecmp(ind, "link")) {
/* To see the URL being indexed in "ps" output on xBSD */
dps_setproctitle("[%d] Link index creation", Indexer->handle);
DpsLog(Indexer, DPS_LOG_EXTRA, "Creating link index");
if (DPS_OK == DpsLimit4(Indexer, &L4, "link", DPS_IFIELD_TYPE_INT, db)) {
MakeLinearIndex(Indexer, &L4, DPS_LIMFNAME_LINK, db);
}
} else if (!strcasecmp(ind, "time")) {
/* To see the URL being indexed in "ps" output on xBSD */
dps_setproctitle("[%d] Time index creation", Indexer->handle);
DpsLog(Indexer, DPS_LOG_EXTRA, "Creating time index");
if (DPS_OK == DpsLimit4(Indexer, &L4, "last_mod_time", DPS_IFIELD_TYPE_HOUR, db)) {
MakeLinearIndex(Indexer, &L4, DPS_LIMFNAME_TIME, db);
}
} else if (!strcasecmp(ind, "hostname")) {
/* To see the URL being indexed in "ps" output on xBSD */
dps_setproctitle("[%d] Hostname index creation", Indexer->handle);
DpsLog(Indexer, DPS_LOG_EXTRA, "Creating hostname index");
if (DPS_OK == DpsLimit4(Indexer, &L4, "url", DPS_IFIELD_TYPE_HOSTNAME, db)) {
MakeLinearIndex(Indexer, &L4, DPS_LIMFNAME_HOST, db);
}
} else if (!strcasecmp(ind, "language")) {
/* To see the URL being indexed in "ps" output on xBSD */
dps_setproctitle("[%d] Language index creation", Indexer->handle);
DpsLog(Indexer, DPS_LOG_EXTRA, "Creating language index");
if (DPS_OK == DpsLimit4(Indexer, &L4, "Content-Language", DPS_IFIELD_TYPE_STR2CRC32, db)) {
MakeLinearIndex(Indexer, &L4, DPS_LIMFNAME_LANG, db);
}
} else if (!strcasecmp(ind, "content")) {
/* To see the URL being indexed in "ps" output on xBSD */
dps_setproctitle("[%d] Content-Type index creation", Indexer->handle);
DpsLog(Indexer, DPS_LOG_EXTRA, "Creating Content-Type index");
if (DPS_OK == DpsLimit4(Indexer, &L4, "Content-Type", DPS_IFIELD_TYPE_STRCRC32, db)) {
MakeLinearIndex(Indexer, &L4, DPS_LIMFNAME_CTYPE, db);
}
} else if (!strcasecmp(ind, "siteid")) {
/* To see the URL being indexed in "ps" output on xBSD */
dps_setproctitle("[%d] Site_id index creation", Indexer->handle);
DpsLog(Indexer, DPS_LOG_EXTRA, "Creating Site_id index");
if (DPS_OK == DpsLimit4(Indexer, &L4, "site_id", DPS_IFIELD_TYPE_INT, db)) {
MakeLinearIndex(Indexer, &L4, DPS_LIMFNAME_SITE, db);
}
} else {
char *buf, *req, *dbaddr;
DPS_DB ldb, *pdb = &ldb;
size_t buf_len = dps_strlen(nm) + 16;
if ((buf = (char*) DpsMalloc(buf_len * sizeof(char))) == NULL) {
DpsLog(Indexer, DPS_LOG_ERROR, "Can't alloc %d chars at %s:%d", buf_len, __FILE__, __LINE__);
return DPS_ERROR;
}
dps_setproctitle("[%d] %s index creation", Indexer->handle, nm);
DpsLog(Indexer, DPS_LOG_EXTRA, "Creating %s index", nm);
dps_snprintf(buf, buf_len, "Req-%s", nm);
req = DpsVarListFindStr(&Indexer->Conf->Vars, buf, NULL);
if (req != NULL) {
dps_snprintf(buf, buf_len, "dbaddr-%s", nm);
dbaddr = DpsVarListFindStr(&Indexer->Conf->Vars, buf, NULL);
if (dbaddr != NULL) {
DpsDBSetAddr(pdb, dbaddr, DPS_OPEN_MODE_READ);
} else {
pdb = db;
}
if (!strcasecmp(ind, "nex8str")) {
if (DPS_OK == DpsSQLLimit8(Indexer, &L8, req, DPS_IFIELD_TYPE_HEX8STR, pdb)) {
MakeNestedIndex(Indexer, &L8, lfname, pdb);
}
} else {
int field_type = DPS_IFIELD_TYPE_INT;
if (!strcasecmp(ind, "strcrc32")) field_type = DPS_IFIELD_TYPE_STRCRC32;
else if (!strcasecmp(ind, "hour")) field_type = DPS_IFIELD_TYPE_HOUR;
else if (!strcasecmp(ind, "hostname")) field_type = DPS_IFIELD_TYPE_HOSTNAME;
else if (!strcasecmp(ind, "char2")) field_type = DPS_IFIELD_TYPE_STR2CRC32;
else if (!strcasecmp(ind, "int")) field_type = DPS_IFIELD_TYPE_INT;
if (DPS_OK == DpsSQLLimit4(Indexer, &L4, req, field_type, db)) {
MakeLinearIndex(Indexer, &L4, lfname, db);
}
}
}
}
/* To see the URL being indexed in "ps" output on xBSD */
dps_setproctitle("[%d] Indexes done.", Indexer->handle);
DpsLog(Indexer, DPS_LOG_EXTRA, "Done");
}
}
return DPS_OK;
}
int DpsChineseListLoad(DPS_AGENT *Agent, DPS_CHINALIST *List, const char *charset, const char *fname) {
struct stat sb;
char *str, *data = NULL, *cur_n = NULL;
DPS_CHINAWORD chinaword;
char word[PATH_MAX];
dpsunicode_t uword[256];
DPS_CHARSET *sys_int, *fcs;
DPS_CONV to_uni;
int fd;
char savebyte;
sys_int = DpsGetCharSet("sys-int");
if (!(fcs = DpsGetCharSet(charset))) {
if (Agent->Conf->is_log_open) DpsLog(Agent, DPS_LOG_ERROR, "Charset '%s' not found or not supported", charset);
else fprintf(stderr, "Charset '%s' not found or not supported", charset);
return DPS_ERROR;
}
DpsConvInit(&to_uni, fcs, sys_int, Agent->Conf->CharsToEscape, DPS_RECODE_HTML);
if (*fname != '/') {
dps_snprintf(word, sizeof(word), "%s/%s", DpsVarListFindStr(&Agent->Conf->Vars, "EtcDir", DPS_CONF_DIR), fname);
fname = word;
}
if (stat(fname, &sb)) {
if (Agent->Conf->is_log_open)
DpsLog(Agent, DPS_LOG_ERROR, "Unable to stat FreqDic file '%s': %s", fname, strerror(errno));
else fprintf(stderr, "Unable to stat FrecDic file '%s': %s", fname, strerror(errno));
return DPS_ERROR;
}
if ((fd = open(fname, O_RDONLY)) <= 0) {
if (Agent->Conf->is_log_open)
DpsLog(Agent, DPS_LOG_ERROR, "Unable to open FreqDic file '%s': %s", fname, strerror(errno));
else fprintf(stderr, "Unable to open FreqDic file '%s': %s", fname, strerror(errno));
return DPS_ERROR;
}
if ((data = (char*)DpsMalloc(sb.st_size + 1)) == NULL) {
if (Agent->Conf->is_log_open)
DpsLog(Agent, DPS_LOG_ERROR, "Unable to alloc %d bytes", sb.st_size);
else fprintf(stderr, "Unable to alloc %ld bytes", (long)sb.st_size);
close(fd);
return DPS_ERROR;
}
if (read(fd, data, sb.st_size) != (ssize_t)sb.st_size) {
if (Agent->Conf->is_log_open)
DpsLog(Agent, DPS_LOG_ERROR, "Unable to read FreqDic file '%s': %s", fname, strerror(errno));
else fprintf(stderr, "Unable to read FreqDic file '%s': %s", fname, strerror(errno));
DPS_FREE(data);
close(fd);
return DPS_ERROR;
}
data[sb.st_size] = '\0';
str = data;
cur_n = strchr(str, NL_INT);
if (cur_n != NULL) {
cur_n++;
savebyte = *cur_n;
*cur_n = '\0';
}
close(fd);
bzero((void*)&chinaword, sizeof(chinaword));
chinaword.word = uword;
while(str != NULL) {
if(!str[0]) goto loop_continue;
if(str[0]=='#') goto loop_continue;
sscanf(str, "%d %63s ", &chinaword.freq, word );
DpsConv(&to_uni, (char*)uword, sizeof(uword), word, sizeof(word));
DpsChineseListAdd(List, &chinaword);
loop_continue:
str = cur_n;
if (str != NULL) {
*str = savebyte;
cur_n = strchr(str, NL_INT);
if (cur_n != NULL) {
cur_n++;
savebyte = *cur_n;
*cur_n = '\0';
}
}
}
DPS_FREE(data);
DpsChineseListSort(List);
{ register size_t i, j = 0;
for (i = 1; i < List->nwords; i++) {
if (cmpchinese(&List->ChiWord[j], &List->ChiWord[i]) == 0) {
List->ChiWord[j].freq += List->ChiWord[i].freq;
} else { j++;
}
}
for (i = j + 1; i < List->nwords; i++) {
DPS_FREE(List->ChiWord[i].word);
}
List->nwords = j + 1;
}
return DPS_OK;
}
__C_LINK int __DPSCALL DpsBaseCheckup(DPS_BASE_PARAM *P, int (*checkrec) (DPS_AGENT *A, const urlid_t rec_id)) {
int found;
urlid_t i;
size_t z;
urlid_t *todel = (int*)DpsMalloc(128 * sizeof(urlid_t));
size_t ndel = 0, mdel = 128, totaldel = 0;
if (todel == NULL) return DPS_ERROR;
for (i = 0; i < (urlid_t)P->NFiles; i++) {
if (have_sigterm || have_sigint || have_sigalrm) {
DpsLog(P->A, DPS_LOG_EXTRA, "%s signal received. Exiting chackup", (have_sigterm) ? "SIGTERM" :
(have_sigint) ? "SIGINT" : "SIGALRM");
DpsBaseClose(P);
DPS_FREE(todel);
return DPS_OK;
}
P->rec_id = i << DPS_BASE_BITS;
if (DpsBaseOpen(P, DPS_READ_LOCK) != DPS_OK) {
DpsBaseClose(P);
continue;
}
if (lseek(P->Ifd, (off_t)0, SEEK_SET) == (off_t)-1) {
DpsLog(P->A, DPS_LOG_ERROR, "Can't seeek for file %s", P->Ifilename);
DpsBaseClose(P);
DPS_FREE(todel);
return DPS_ERROR;
}
while (read(P->Ifd, &P->Item, sizeof(DPS_BASEITEM)) == sizeof(DPS_BASEITEM)) {
if (P->Item.rec_id != 0) {
found = checkrec(P->A, P->Item.rec_id);
if (found == 0) {
if (ndel >= mdel) {
mdel += 128;
todel = (urlid_t*)DpsRealloc(todel, mdel * sizeof(urlid_t));
if (todel == NULL) {
DpsBaseClose(P);
DpsLog(P->A, DPS_LOG_ERROR, "Can't realloc %d bytes %s:%d", mdel * sizeof(urlid_t), __FILE__, __LINE__);
DPS_FREE(todel);
return DPS_ERROR;
}
}
todel[ndel++] = P->Item.rec_id;
}
}
}
DpsBaseClose(P);
for (z = 0; z < ndel; z++) {
DpsLog(P->A, DPS_LOG_DEBUG, "Base %s/%s store %03X: deleting url_id: %X", P->subdir, P->basename, i, todel[z]);
P->rec_id = todel[z];
DpsBaseDelete(P);
}
DpsBaseClose(P);
DpsLog(P->A, DPS_LOG_INFO, "Base %s/%s store %03X, %d lost records deleted", P->subdir, P->basename, i, ndel);
totaldel += ndel;
ndel = 0;
}
DPS_FREE(todel);
DpsLog(P->A, DPS_LOG_EXTRA, "Total lost record(s) deleted: %d\n", totaldel);
return DPS_OK;
}
/* Main function to calculate items sequence */
int DpsCalcBoolItems(DPS_AGENT *query, DPS_RESULT *Res) {
DPS_BOOLSTACK *s = DpsBoolStackInit(NULL);
DPS_STACK_ITEM *items = Res->items, *res;
DPS_WIDEWORD Word;
size_t nitems = Res->nitems;
size_t i, j;
int first_time = (Res->WWList.nwords == 0);
if (s == NULL) return DPS_STACK_ERR;
if (nitems == 0) {
Res->CoordList.Coords = Res->items[0].pbegin;
Res->items[0].pbegin = NULL;
Res->items[0].count = 0;
/* Res->CoordList.ncoords = Res->items[0]->count;*/
DpsBoolStackFree(s);
return DPS_OK;
}
for (i = 0; i < nitems; i++) {
if (items[i].cmd != DPS_STACK_WORD) continue;
if ((items[i].pbegin == NULL) && ((items[i].origin & DPS_WORD_ORIGIN_STOP) == 0)) {
for(j = 0; j < i; j++)
if (items[j].crcword == items[i].crcword) break;
if (j < i) {
items[i].count = items[j].count;
items[i].pbegin = (DPS_URL_CRD_DB*)DpsMalloc((items[j].count + 1) * sizeof(DPS_URL_CRD_DB));
if (items[i].pbegin == NULL) {
DpsLog(query, DPS_LOG_ERROR, "Can't alloc %d bytes %s:%d", (items[j].count + 1) * sizeof(DPS_URL_CRD_DB), __FILE__, __LINE__);
DpsBoolStackFree(s);
return DPS_STACK_ERR;
}
{
register size_t z;
for (z = 0; z < items[i].count; z++) {
items[i].pbegin[z] = items[j].pbegin[z];
items[i].pbegin[z].coord &= 0xFFFFFF00;
items[i].pbegin[z].coord += (items[i].wordnum & 0xFF);
}
}
}
}
if (first_time) {
Word.order = items[i].order;
Word.count = items[i].count;
Word.crcword = items[i].crcword;
Word.word = items[i].word;
Word.uword = items[i].uword;
Word.origin = items[i].origin;
DpsWideWordListAdd(&Res->WWList, &Word, DPS_WWL_LOOSE);
}
}
#ifdef DEBUG_BOOL
DpsLog(query, DPS_LOG_EXTRA, "--------");
for(i=0;i<nitems;i++){
DpsLog(query, DPS_LOG_EXTRA, "[%d].%d %c : %d -- %s",
i, items[i].wordnum, item_type(items[i].cmd), items[i].count, (items[i].word == NULL) ? "<NULL>" : items[i].word);
}
DpsLog(query, DPS_LOG_EXTRA, "--------");
#endif
for(i=0;i<nitems;i++){
int c;
#ifdef DEBUG_BOOL
DpsLog(query, DPS_LOG_EXTRA,
".[%d].%d %c : %d -- %s, (order_origin:%d)", i, items[i].wordnum, item_type(items[i].cmd), items[i].count, (items[i].word == NULL) ? "<NULL>" : items[i].word, items[i].order_origin);
#endif
switch(c=items[i].cmd){
case DPS_STACK_RIGHT:
/* Perform till LEFT bracket */
while((TOPCMD(s) != DPS_STACK_LEFT) && (TOPCMD(s) != DPS_STACK_BOT))
if (DPS_OK != perform(query, Res, s, POPCMD(s))) {
DpsBoolStackFree(s);
return DPS_STACK_ERR;
}
/* Pop LEFT bracket itself */
if(TOPCMD(s) == DPS_STACK_LEFT)
POPCMD(s);
break;
case DPS_STACK_OR:
case DPS_STACK_AND:
case DPS_STACK_NEAR:
case DPS_STACK_ANYWORD:
if (s->nastack > 1)
while(c <= TOPCMD(s)) {
if (DPS_OK != perform(query, Res, s, POPCMD(s))) {
DpsBoolStackFree(s);
return DPS_STACK_ERR;
}
}
/* IMPORTANT! No break here! That's OK*/
/* Так надо ! */
case DPS_STACK_LEFT:
case DPS_STACK_PHRASE_LEFT:
case DPS_STACK_NOT:
if (PUSHCMD(s,c) != DPS_OK) {
DpsBoolStackFree(s);
return DPS_STACK_ERR;
}
break;
case DPS_STACK_PHRASE_RIGHT:
/* perform till RIGHT phrase quote */
while((TOPCMD(s) != DPS_STACK_PHRASE_LEFT) && (TOPCMD(s) != DPS_STACK_BOT))
if (DPS_OK != perform(query, Res, s, POPCMD(s))) {
DpsBoolStackFree(s);
return DPS_STACK_ERR;
}
if (TOPCMD(s) == DPS_STACK_PHRASE_LEFT) perform(query, Res, s, POPCMD(s));
#ifdef DEBUG_BOOL
DpsLog(query, DPS_LOG_EXTRA, "[%d] %c", i, item_type(items[i].cmd));
#endif
break;
case DPS_STACK_WORD:
items[i].order_from = items[i].order_to = items[i].order;
default:
if (DPS_OK != PUSHARG(s, &items[i])) {
DpsBoolStackFree(s);
return DPS_STACK_ERR;
}
items[i].pbegin = items[i].plast = NULL; /*items[i].word = NULL; items[i].uword = NULL;*/
/* items[i].count = 0;*/
/* DpsStackItemFree(&items[i]);*/
/* if (DPS_OK != PUSHARG(s, (count[items[i].order]) ? 1UL : 0UL)) {
DpsBoolStackFree(s);
return DPS_STACK_ERR;
}*/
break;
}
}
while(TOPCMD(s) != DPS_STACK_BOT) {
if (DPS_OK != perform(query, Res, s, POPCMD(s))) {
DpsBoolStackFree(s);
return DPS_STACK_ERR;
}
}
res = POPARG(s);
if (res != NULL) {
Res->CoordList.Coords = res->pbegin;
Res->CoordList.ncoords = res->count;
res->pbegin = res->plast = NULL;
/* res->count = 0;*/
DpsStackItemFree(res);
}
#ifdef DEBUG_BOOL
DpsLog(query, DPS_LOG_EXTRA, "result: %x", res);
#endif
DpsBoolStackFree(s);
return DPS_OK;
}
int DpsCarryLimitWrite(DPS_AGENT *Indexer, FILE *f, urlid_t *data, size_t num) {
unsigned char *bits;
urlid_t curr = 0;
urlid_t *a;
size_t i, n, z;
int j;
size_t avail, elems;
unsigned size, max_bits;
unsigned char *table, *base;
unsigned global_max, global_sum, global_n;
/* allocating mem for bits[i] - minimal bits needed to code a[i] */
if (! (bits = (unsigned char*)malloc(ELEMS_PER_BLOCK * sizeof(unsigned char)))) {
DpsLog(Indexer, DPS_LOG_ERROR, "Out of memory [%s:%d]", __FILE__, __LINE__);
return DPS_ERROR;
}
if (! (a = (urlid_t*) malloc(ELEMS_PER_BLOCK * sizeof(urlid_t)))) {
DpsLog(Indexer, DPS_LOG_ERROR, "Out of memory [%s:%d]", __FILE__, __LINE__);
DPS_FREE(bits);
return DPS_ERROR;
}
CARRY_ENCODE_START(f);
size = TRANS_TABLE_STARTER;
global_max = global_sum = global_n = 0;
for (z = 0; z < num; ) {
for (n = 0; (n < ELEMS_PER_BLOCK) && (z < num); n++,z++) {
a[n] = data[z] - curr;
curr = data[z];
}
max_bits = CalcMinBits(a, bits, n, &global_max, &global_sum, &global_n);
CARRY_BLOCK_ENCODE_START(n, max_bits);
for (i=0; i<n; )
{
avail = GET_AVAILABLE_BITS;
table = GET_TRANS_TABLE(avail);
base= table+(size<<2); /* row in trans table */
/* 1. Modeling: Find j= the first-fit column in base */
for (j=0; j<4; j++)
{
size = base[j];
if (size >avail) /* must use next word for data */
{
avail=32;
j=-1;
continue;
}
if ( elems=elems_coded(avail,size,bits,i,n-1) )
break;
}
/* 2. Coding: Code elements using row "base" & column "j" */
WORD_ENCODE(j+1,2); /* encoding column */
for ( ; elems ; elems--, i++) /* encoding d-gaps */
WORD_ENCODE(a[i],size);
}
}
CARRY_ENCODE_END;
DPS_FREE(a);
DPS_FREE(bits);
return DPS_OK;
}
static int MakeNestedIndex(DPS_AGENT *Indexer, DPS_UINT8URLIDLIST *L, const char *lim_name, DPS_DB *db) {
DPS_ENV *Conf = Indexer->Conf;
size_t k, prev;
urlid_t *data=NULL;
DPS_UINT8_POS_LEN *ind=NULL;
size_t mind=1000, nind=0, ndata;
char fname[PATH_MAX];
int dat_fd=0, ind_fd=0;
int rc=DPS_OK;
const char *vardir = (db->vardir) ? db->vardir : DpsVarListFindStr(&Conf->Vars, "VarDir", DPS_VAR_DIR);
if(!L->Item)return(1);
if (L->nitems > 1) DpsSort(L->Item, L->nitems, sizeof(DPS_UINT8URLID), (qsort_cmp)cmp_ind8);
data = (urlid_t*)DpsMalloc((L->nitems + 1) * sizeof(urlid_t));
if(!data){
DpsLog(Indexer, DPS_LOG_ERROR, "Can't alloc %d bytes [%s:%d]", (L->nitems + 1) * sizeof(urlid_t), __FILE__, __LINE__);
goto err1;
}
ind=(DPS_UINT8_POS_LEN*)DpsMalloc(mind*sizeof(DPS_UINT8_POS_LEN));
if(!ind){
DpsLog(Indexer, DPS_LOG_ERROR, "Can't alloc %d bytes [%s:%d]", mind * sizeof(DPS_UINT8_POS_LEN), __FILE__, __LINE__);
goto err1;
}
prev=0;
for(k=0; k < L->nitems; k++) {
data[k] = L->Item[k].url_id;
if((k == L->nitems-1) || (L->Item[k].hi != L->Item[prev].hi) || (L->Item[k].lo != L->Item[prev].lo)) {
if(nind==mind){
mind+=1000;
ind=(DPS_UINT8_POS_LEN*)DpsRealloc(ind,mind*sizeof(DPS_UINT8_POS_LEN));
if(!ind) {
DpsLog(Indexer, DPS_LOG_ERROR, "Can't alloc %d bytes [%s:%d]", mind * sizeof(DPS_UINT8_POS_LEN), __FILE__, __LINE__);
goto err1;
}
}
/* Fill index */
ind[nind].hi = L->Item[prev].hi;
ind[nind].lo = L->Item[prev].lo;
ind[nind].pos = prev * sizeof(*data);
if (k == L->nitems - 1) ind[nind].len = (k - prev + 1) * sizeof(*data);
else ind[nind].len = (k - prev) * sizeof(*data);
DpsLog(Indexer, DPS_LOG_DEBUG, "%08X%08X - %d %d\n", ind[nind].hi, ind[nind].lo, (int)ind[nind].pos, ind[nind].len);
nind++;
prev=k;
}
}
ndata = L->nitems;
ClearIndex8(L);
dps_snprintf(fname,sizeof(fname)-1,"%s%c%s%c%s.dat", vardir,DPSSLASH, DPS_TREEDIR,DPSSLASH, lim_name);
if((dat_fd = DpsOpen3(fname, O_CREAT | O_WRONLY | O_TRUNC | DPS_BINARY, DPS_IWRITE)) < 0) {
DpsLog(Indexer, DPS_LOG_ERROR, "Can't open '%s': %s [%s:%d]", fname, strerror(errno), __FILE__, __LINE__);
goto err1;
}
DpsWriteLock(dat_fd);
if((ndata * sizeof(*data)) != (size_t)write(dat_fd, data, ndata * sizeof(*data))) {
DpsLog(Indexer, DPS_LOG_ERROR, "Can't write '%s': %s [%s:%d]", fname, strerror(errno), __FILE__, __LINE__);
goto err1;
}
DpsUnLock(dat_fd);
DpsClose(dat_fd);
DPS_FREE(data);
dps_snprintf(fname,sizeof(fname)-1,"%s%c%s%c%s.ind", vardir, DPSSLASH,DPS_TREEDIR, DPSSLASH, lim_name);
if((ind_fd = DpsOpen3(fname, O_CREAT | O_WRONLY | O_TRUNC | DPS_BINARY, DPS_IWRITE)) < 0) {
DpsLog(Indexer, DPS_LOG_ERROR, "Can't open '%s': %s [%s:%d]", fname, strerror(errno), __FILE__, __LINE__);
goto err1;
}
DpsWriteLock(ind_fd);
if((nind*sizeof(DPS_UINT8_POS_LEN)) != (size_t)write(ind_fd,ind,nind*sizeof(DPS_UINT8_POS_LEN))){
DpsLog(Indexer, DPS_LOG_ERROR, "Can't write '%s': %s [%s:%d]", fname, strerror(errno), __FILE__, __LINE__);
goto err1;
}
DpsUnLock(ind_fd);
DpsClose(ind_fd);
DPS_FREE(ind);
return(0);
err1:
ClearIndex8(L);
DPS_FREE(data);
DPS_FREE(ind);
if(dat_fd) DpsClose(dat_fd);
if(ind_fd) DpsClose(ind_fd);
return(1);
}
void DpsParseHTTPResponse(DPS_AGENT *Indexer, DPS_DOCUMENT *Doc) {
char *token, *lt, *headers, savec;
int status, oldstatus;
DPS_DSTR header;
time_t now, last_mod_time;
Doc->Buf.content=NULL;
oldstatus = DpsVarListFindInt(&Doc->Sections, "Status", 0);
DpsVarListReplaceInt(&Doc->Sections, "ResponseSize", (int)Doc->Buf.size);
DpsVarListDel(&Doc->Sections, "Content-Length");
/* DpsVarListDel(&Doc->Sections, "Last-Modified");*/ /* if it's not deleted Lat-Modified equals to the first appearance in db */
if (Doc->Buf.buf == NULL) return;
/* Cut HTTP response header first */
for(token=Doc->Buf.buf;*token;token++){
if(!strncmp(token,"\r\n\r\n",4)){
if (token <= Doc->Buf.buf + Doc->Buf.size - 4) {
*token='\0';
Doc->Buf.content = token + 4;
}
break;
} else if(!strncmp(token,"\n\n",2)){
if (token <= Doc->Buf.buf + Doc->Buf.size - 2) {
*token='\0';
Doc->Buf.content = token + 2;
}
break;
}
}
/* Bad response, return */
if(!Doc->Buf.content) {
if (token <= Doc->Buf.buf + Doc->Buf.size - 4) {
if (token[2] == CR_CHAR) Doc->Buf.content = token + 4;
else Doc->Buf.content = token + 2;
}
}
/* Copy headers not to break them */
headers = (char*)DpsStrdup(Doc->Buf.buf);
/* Now lets parse response header lines */
token = dps_strtok_r(headers, "\r\n", <, &savec);
if(!token) {
DpsFree(headers);
return;
}
if(!strncmp(token,"HTTP/",5)){
status = atoi(token + 8);
DpsVarListReplaceStr(&Doc->Sections,"ResponseLine",token);
DpsVarListReplaceInt(&Doc->Sections, "Status", (oldstatus > status) ? oldstatus : status );
}else{
DpsFree(headers);
return;
}
token = dps_strtok_r(NULL, "\r\n", <, &savec);
DpsDSTRInit(&header, 128);
while(token){
if(strchr(token,':')) {
if (header.data_size) {
DpsParseHTTPHeader(Indexer, Doc, &header);
DpsDSTRFree(&header);
DpsDSTRInit(&header, 128);
}
}
DpsDSTRAppendStr(&header, token);
token = dps_strtok_r(NULL, "\r\n", <, &savec);
}
if (header.data_size) {
DpsParseHTTPHeader(Indexer, Doc, &header);
}
DpsDSTRFree(&header);
DPS_FREE(headers);
{
now = Indexer->now;
last_mod_time = DpsHttpDate2Time_t(DpsVarListFindStr(&Doc->Sections, "Last-Modified", ""));
if (last_mod_time > now + 3600 * 4) { /* we have a document with Last-Modified time in the future */
DpsLog(Indexer, DPS_LOG_EXTRA, "Last-Modified date is deep in future (%d>%d), dropping it.", last_mod_time, now);
DpsVarListDel(&Doc->Sections, "Last-Modified");
}
}
/* Bad response, return */
if(!Doc->Buf.content) {
return;
}
DpsVarListReplaceInt(&Doc->Sections,"Content-Length", Doc->Buf.buf-Doc->Buf.content+(int)Doc->Buf.size + DpsVarListFindInt(&Doc->Sections,"Content-Length", 0));
}