void DpsAppendTarget(DPS_AGENT *Indexer, const char *url, const char *lang, const int hops, int parent) {
DPS_DOCUMENT *Doc, *Save;
size_t i;
TRACE_IN(Indexer, "AppendTarget");
DPS_GETLOCK(Indexer, DPS_LOCK_THREAD);
DPS_GETLOCK(Indexer, DPS_LOCK_CONF);
if (Indexer->Conf->Targets.num_rows > 0) {
for (i = Indexer->Conf->Targets.num_rows - 1; i > 0; i--) {
Doc = &Indexer->Conf->Targets.Doc[i];
if ((strcasecmp(DpsVarListFindStr(&Doc->Sections, "URL", ""), url) == 0)
&& (strcmp(DpsVarListFindStr(&Doc->RequestHeaders, "Accept-Language", ""), lang) == 0)) {
DPS_RELEASELOCK(Indexer, DPS_LOCK_CONF);
DPS_RELEASELOCK(Indexer, DPS_LOCK_THREAD);
TRACE_OUT(Indexer);
return;
}
}
}
if ((Indexer->Conf->Targets.Doc =
DpsRealloc(Save = Indexer->Conf->Targets.Doc, (Indexer->Conf->Targets.num_rows + 1) * sizeof(DPS_DOCUMENT))) == NULL) {
Indexer->Conf->Targets.Doc = Save;
DPS_RELEASELOCK(Indexer, DPS_LOCK_CONF);
DPS_RELEASELOCK(Indexer, DPS_LOCK_THREAD);
TRACE_OUT(Indexer);
return;
}
Doc = &Indexer->Conf->Targets.Doc[Indexer->Conf->Targets.num_rows];
DpsDocInit(Doc);
DpsVarListAddStr(&Doc->Sections, "URL", url);
DpsVarListAddInt(&Doc->Sections, "Hops", hops);
DpsVarListDel(&Doc->Sections, "URL_ID");
DpsVarListReplaceInt(&Doc->Sections, "Referrer-ID", parent);
if (*lang != '\0') DpsVarListAddStr(&Doc->RequestHeaders, "Accept-Language", lang);
if (DPS_OK == DpsURLAction(Indexer, Doc, DPS_URL_ACTION_FINDBYURL)) {
urlid_t url_id = DpsVarListFindInt(&Doc->Sections, "DP_ID", 0);
if (url_id != 0) Indexer->Conf->Targets.num_rows++;
else DpsDocFree(Doc);
}
/* fprintf(stderr, "-- AppandTarget: url:%s URL_ID:%d\n", url, DpsStrHash32(url));*/
DPS_RELEASELOCK(Indexer, DPS_LOCK_CONF);
DpsURLAction(Indexer, Doc, DPS_URL_ACTION_ADD);
DPS_RELEASELOCK(Indexer, DPS_LOCK_THREAD);
TRACE_OUT(Indexer);
return;
}
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);
}
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);
}
__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;
}
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 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;
}
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;
}
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);
}
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 __DPSCALL DpsFindWordsSearchd(DPS_AGENT *query, DPS_RESULT *Res, DPS_DB *searchd) {
size_t maxlen = 1024;
char *request, *edf = NULL, *e_empty = NULL;
const char *df = DpsVarListFindStr(&query->Vars, "DateFormat", NULL);
const char *empty = DpsVarListFindStr(&query->Vars, "empty", NULL);
const char *qs = DpsVarListFindStr(&query->Vars, "QUERY_STRING", "");
const char *tmplt = DpsVarListFindStr(&query->Vars, "tmplt", "");
int res=DPS_OK;
TRACE_IN(query, "DpsFindWordsSearchd");
if (df) {
edf = (char*)DpsMalloc(dps_strlen(df) * 10 + 1);
if (edf == NULL) {
sprintf(query->Conf->errstr,"Can't allocate memory");
TRACE_OUT(query);
return DPS_ERROR;
}
DpsEscapeURL(edf, df);
maxlen += dps_strlen(edf);
}
if (empty) {
e_empty = (char*)DpsMalloc(dps_strlen(empty) * 10 + 1);
if (e_empty == NULL) {
sprintf(query->Conf->errstr, "Can't allocate memory");
TRACE_OUT(query);
return DPS_ERROR;
}
DpsEscapeURL(e_empty, empty);
maxlen += dps_strlen(e_empty);
}
maxlen += dps_strlen(qs) + dps_strlen(tmplt) + 64;
if (NULL==(request=(char*)DpsMalloc(maxlen))) {
sprintf(query->Conf->errstr,"Can't allocate memory");
DPS_FREE(edf);
TRACE_OUT(query);
return DPS_ERROR;
}
dps_snprintf(request, maxlen, "%s&BrowserCharset=%s&IP=%s&g-lc=%s&ExcerptSize=%s&ExcerptPadding=%s&DoExcerpt=%s&tmplt=%s%s%s%s%s%s%s&sp=%s&sy=%s&s=%s",
qs,
DpsVarListFindStr(&query->Vars, "BrowserCharset", "iso-8859-1"),
DpsVarListFindStr(&query->Vars, "IP", "localhost"),
DpsVarListFindStr(&query->Vars, "g-lc", "en"),
DpsVarListFindStr(&query->Vars, "ExcerptSize", "256"),
DpsVarListFindStr(&query->Vars, "ExcerptPadding", "40"),
(query->Flags.do_excerpt) ? "yes" : "no",
tmplt,
(edf) ? "&DateFormat=" : "", (edf) ? edf : "",
(e_empty) ? "&empty=" : "", (e_empty) ? e_empty : "",
(searchd->label) ? "&label=" : "", (searchd->label) ? searchd->label : "",
DpsVarListFindStr(&query->Vars, "sp", "1"),
DpsVarListFindStr(&query->Vars, "sy", "1"),
DpsVarListFindStr(&query->Vars, "s", "RP")
);
DPS_FREE(edf);
DPS_FREE(e_empty);
request[maxlen-1]='\0';
res = DpsSearchdSendWordRequest(query, searchd, request);
DPS_FREE(request);
if (DPS_OK != res) {
TRACE_OUT(query);
return res;
}
/* res = DpsSearchdGetWordResponse(query, Res, searchd); called later from DpsFind */
TRACE_OUT(query);
return res;
}
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));
}
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;
}
extern __C_LINK int __DPSCALL DpsBaseRelocate(DPS_AGENT *Agent, int base_type) {
DPS_BASE_PARAM O, N;
DPS_BASE_PARAM *Old = &O, *New = &N;
size_t base, i, ndel, mdel = 128, data_len;
urlid_t *todel = (int*)DpsMalloc(128 * sizeof(urlid_t));
void *data;
bzero(Old, sizeof(O));
bzero(New, sizeof(N));
switch(base_type) {
case 0: /* stored */
Old->subdir = "store";
Old->basename = "doc";
Old->indname = "doc";
Old->mode = DPS_WRITE_LOCK;
Old->NFiles = (size_t)DpsVarListFindInt(&Agent->Vars, "OldStoredFiles", 0x100);
Old->vardir = DpsVarListFindStr(&Agent->Vars, "VarDir", DPS_VAR_DIR);
Old->A = Agent;
New->subdir = "store";
New->basename = "doc";
New->indname = "doc";
New->mode = DPS_WRITE_LOCK;
New->NFiles = (size_t)DpsVarListFindInt(&Agent->Vars, "StoredFiles", 0x100);
New->vardir = DpsVarListFindStr(&Agent->Vars, "VarDir", DPS_VAR_DIR);
New->A = Agent;
DpsLog(Agent, DPS_LOG_INFO, "Relocating stored database");
break;
case 1: /* URL data */
Old->subdir = DPS_URLDIR;
Old->basename = "info";
Old->indname = "info";
Old->mode = DPS_WRITE_LOCK;
Old->NFiles = (size_t)DpsVarListFindInt(&Agent->Vars, "OldURLDataFiles", 0x300);
Old->vardir = DpsVarListFindStr(&Agent->Vars, "VarDir", DPS_VAR_DIR);
Old->A = Agent;
#ifdef HAVE_ZLIB
O.zlib_method = Z_DEFLATED;
O.zlib_level = 9;
O.zlib_windowBits = DPS_BASE_INFO_WINDOWBITS;
O.zlib_memLevel = 9;
O.zlib_strategy = DPS_BASE_INFO_STRATEGY;
#endif
New->subdir = DPS_URLDIR;
New->basename = "info";
New->indname = "info";
New->mode = DPS_WRITE_LOCK;
New->NFiles = (size_t)DpsVarListFindInt(&Agent->Vars, "URLDataFiles", 0x300);
New->vardir = DpsVarListFindStr(&Agent->Vars, "VarDir", DPS_VAR_DIR);
New->A = Agent;
#ifdef HAVE_ZLIB
N.zlib_method = Z_DEFLATED;
N.zlib_level = 9;
N.zlib_windowBits = DPS_BASE_INFO_WINDOWBITS;
N.zlib_memLevel = 9;
N.zlib_strategy = DPS_BASE_INFO_STRATEGY;
#endif
DpsLog(Agent, DPS_LOG_INFO, "Relocating URLData database");
break;
case 2: /* tree wrd */
Old->subdir = DPS_TREEDIR;
Old->basename = "wrd";
Old->indname = "wrd";
Old->mode = DPS_WRITE_LOCK;
Old->NFiles = (size_t)DpsVarListFindInt(&Agent->Vars, "OldWrdFiles", 0x300);
Old->vardir = DpsVarListFindStr(&Agent->Vars, "VarDir", DPS_VAR_DIR);
Old->A = Agent;
#ifdef HAVE_ZLIB
O.zlib_method = Z_DEFLATED;
O.zlib_level = 9;
O.zlib_windowBits = DPS_BASE_WRD_WINDOWBITS;
O.zlib_memLevel = 9;
O.zlib_strategy = DPS_BASE_WRD_STRATEGY;
#endif
New->subdir = DPS_TREEDIR;
New->basename = "wrd";
New->indname = "wrd";
New->mode = DPS_WRITE_LOCK;
New->NFiles = (size_t)DpsVarListFindInt(&Agent->Vars, "WrdFiles", 0x300);
New->vardir = DpsVarListFindStr(&Agent->Vars, "VarDir", DPS_VAR_DIR);
New->A = Agent;
#ifdef HAVE_ZLIB
N.zlib_method = Z_DEFLATED;
N.zlib_level = 9;
N.zlib_windowBits = DPS_BASE_WRD_WINDOWBITS;
N.zlib_memLevel = 9;
N.zlib_strategy = DPS_BASE_WRD_STRATEGY;
#endif
DpsLog(Agent, DPS_LOG_INFO, "Relocating Wrd database");
break;
default:
DPS_FREE(todel);
return DPS_OK;
}
for (base = 0; base < O.NFiles; base++) {
ndel = 0;
if (have_sigterm || have_sigint || have_sigalrm) {
DpsLog(Agent, DPS_LOG_EXTRA, "%s signal received. Exiting chackup", (have_sigterm) ? "SIGTERM" :
(have_sigint) ? "SIGINT" : "SIGALRM");
DpsBaseClose(Old);
DpsBaseClose(New);
DPS_FREE(todel);
return DPS_OK;
}
Old->rec_id = (urlid_t)(base << DPS_BASE_BITS);
if (DpsBaseOpen(Old, DPS_READ_LOCK) != DPS_OK) {
DpsBaseClose(Old);
DpsBaseClose(New);
continue;
}
if (lseek(O.Ifd, (off_t)0, SEEK_SET) == (off_t)-1) {
DpsLog(Agent, DPS_LOG_ERROR, "Can't seeek for file %s", Old->Ifilename);
DpsBaseClose(Old);
DpsBaseClose(New);
DPS_FREE(todel);
return DPS_ERROR;
}
while (read(Old->Ifd, &Old->Item, sizeof(DPS_BASEITEM)) == sizeof(DPS_BASEITEM)) {
if (Old->Item.rec_id != 0) {
if (ndel >= mdel) {
mdel += 128;
todel = (urlid_t*)DpsRealloc(todel, mdel * sizeof(urlid_t));
if (todel == NULL) {
DpsBaseClose(Old);
DpsBaseClose(New);
DpsLog(Agent, DPS_LOG_ERROR, "Can't realloc %d bytes %s:%d", mdel * sizeof(urlid_t), __FILE__, __LINE__);
DPS_FREE(todel);
return DPS_ERROR;
}
}
todel[ndel++] = Old->Item.rec_id;
}
}
DpsBaseClose(Old);
for (i = 0; i < ndel; i++) {
Old->rec_id = todel[i];
data = DpsBaseARead(Old, &data_len);
if (data == NULL) continue;
DpsBaseDelete(Old);
DpsBaseClose(Old);
New->rec_id = todel[i];
DpsBaseWrite(New, data, data_len);
DpsBaseClose(New);
DPS_FREE(data);
}
DpsLog(Agent, DPS_LOG_EXTRA, "\tbase: %d [0x%x], %d records relocated", base, base, ndel);
}
DPS_FREE(todel);
for (base = N.NFiles; base < O.NFiles; base++) {
Old->rec_id = (urlid_t)(base << DPS_BASE_BITS);
if (DpsBaseOpen(Old, DPS_READ_LOCK) != DPS_OK) {
DpsBaseClose(Old);
continue;
}
unlink(O.Ifilename);
unlink(O.Sfilename);
DpsBaseClose(Old);
}
return DPS_OK;
}
void DpsCookiesFind(DPS_AGENT *Indexer, DPS_SERVER *Server, DPS_DOCUMENT *Doc, const char *hostinfo) {
#ifdef HAVE_SQL
DPS_DSTR cookie;
DPS_COOKIES *Cookies = &Indexer->Cookies;
DPS_COOKIE *Coo;
size_t i, blen = dps_strlen(hostinfo), slen;
int have_no_cookies = DpsVarListFindInt(&Doc->Sections, "have_no_cookies", 1);
#ifdef WITH_PARANOIA
void *paran = DpsViolationEnter(paran);
#endif
TRACE_IN(Indexer, "DpsCookiesFind");
DpsDSTRInit(&cookie, 1024);
for(i = 0; i < Cookies->ncookies; i++) {
Coo = &Cookies->Cookie[i];
slen = dps_strlen(Coo->domain);
if (slen > blen) continue;
if (Coo->secure == 'y' && strcasecmp(Doc->CurURL.schema, "https")) continue;
if (strncasecmp(Coo->path, Doc->CurURL.path, dps_strlen(Coo->path))) continue;
if (strcasecmp(Coo->domain, hostinfo + (blen - slen))) continue;
if (Coo->from_config != 1) have_no_cookies = 0;
if (Coo->name[0] == '\0' && Coo->value[0] == '\0') continue;
if (cookie.data_size)
DpsDSTRAppend(&cookie, "; ", 2);
DpsDSTRAppendStr(&cookie, Coo->name);
DpsDSTRAppend(&cookie, "=", 1);
DpsDSTRAppendStr(&cookie, Coo->value);
}
if (have_no_cookies) {
char buf[2*PATH_MAX];
dpshash32_t url_id;
DPS_DB *db;
DPS_SQLRES Res;
size_t rows;
int rc;
if (Server != NULL) {
char *PingData = DpsVarListFindStr(&Server->Vars, "AuthPing", NULL);
if (PingData != NULL) {
char *AuthPing = DpsStrdup(DpsTrim(PingData, " \t\r\n"));
int method = DPS_METHOD_GET;
dps_base64_decode(AuthPing, PingData, dps_strlen(PingData));
if (!strncasecmp(AuthPing, "GET", 3)) {
method = DPS_METHOD_GET;
PingData = DpsTrim(AuthPing + 3, " \t\r\n");
} else if (!strncasecmp(AuthPing, "POST", 4)) {
method = DPS_METHOD_POST;
PingData = DpsTrim(AuthPing + 4, " \t\r\n");
} else {
DpsLog(Indexer, DPS_LOG_ERROR, "AuthPing should be GET or POST: %s", AuthPing);
PingData = NULL;
}
if (PingData != NULL) {
size_t size = dps_strlen(PingData);
{
char PingURL[size + 2];
char PingBody[size];
DPS_DOCUMENT *rDoc;
int result;
rDoc = DpsDocInit(NULL);
DpsSpiderParamInit(&rDoc->Spider);
DpsVarList2Doc(rDoc, Server);
rDoc->Buf.max_size = (size_t)DpsVarListFindInt(&Indexer->Vars, "MaxDocSize", DPS_MAXDOCSIZE);
rDoc->Buf.allocated_size = DPS_NET_BUF_SIZE;
if ((rDoc->Buf.buf = (char*)DpsMalloc(rDoc->Buf.allocated_size + 1)) == NULL) {
DpsDocFree(rDoc);
TRACE_OUT(Indexer);
return;
}
rDoc->Buf.buf[0]='\0';
rDoc->subdoc = Indexer->Flags.SubDocLevel + 1;
#if 1
dps_snprintf(buf, sizeof(buf), "%s://%s/", DPS_NULL2EMPTY(Doc->CurURL.schema), DPS_NULL2EMPTY(Doc->CurURL.hostinfo));
DpsVarListReplaceStr(&rDoc->Sections, "URL", buf);
DpsURLParse(&rDoc->CurURL, buf);
DpsLog(Indexer, DPS_LOG_INFO, "HOME: %s", buf);
rDoc->method = DPS_METHOD_HEAD;
/* DpsVarListFree(&rDoc->RequestHeaders);*/
if (Doc != NULL) {
DpsVarListReplaceLst(&rDoc->RequestHeaders, &Doc->RequestHeaders, NULL, "*");
}
DpsVarListReplaceStr(&rDoc->Sections, "have_no_cookies", "0");
DpsDocAddDocExtraHeaders(Indexer, Server, rDoc);
DpsDocAddConfExtraHeaders(Indexer->Conf, rDoc);
DpsVarListReplaceLst(&rDoc->Sections, &Server->Vars, NULL, "*");
DpsDocAddServExtraHeaders(Server, rDoc);
DpsVarListLog(Indexer, &rDoc->RequestHeaders, DPS_LOG_DEBUG, "HOME.Request");
if (Doc == NULL || Indexer->Flags.cmd == DPS_IND_FILTER) {
DpsDocLookupConn(Indexer, rDoc);
} else {
DPS_FREE(rDoc->connp.connp);
rDoc->connp = Doc->connp;
}
result = DpsGetURL(Indexer, rDoc, NULL); /* Just get headers from the home as we need only Cookies from it */
DpsDocProcessResponseHeaders(Indexer, rDoc);
DpsVarListLog(Indexer, &rDoc->Sections, DPS_LOG_DEBUG, "HOME.Response");
#endif
sscanf(PingData, "%s %s", PingURL, PingBody);
if (rDoc->method == DPS_METHOD_GET) {
dps_strcat(PingURL, "?");
dps_strcat(PingURL, PingBody);
} else {
DpsVarListReplaceStr(&rDoc->Sections, "body", PingBody);
}
DpsVarListReplaceStr(&rDoc->Sections, "URL", PingURL);
DpsURLParse(&rDoc->CurURL, PingURL);
DpsLog(Indexer, DPS_LOG_INFO, "AUTH.PING: %s", PingURL);
rDoc->method = method;
DpsVarListFree(&rDoc->RequestHeaders);
DpsVarListReplaceStr(&rDoc->Sections, "have_no_cookies", "0");
DpsDocAddDocExtraHeaders(Indexer, Server, rDoc);
DpsDocAddConfExtraHeaders(Indexer->Conf, rDoc);
DpsVarListReplaceLst(&rDoc->Sections, &Server->Vars, NULL, "*");
DpsDocAddServExtraHeaders(Server, rDoc);
if (method == DPS_METHOD_POST) {
dps_snprintf(buf, sizeof(buf), "application/x-www-form-urlencoded; charset=%s", DpsVarListFindStr(&Indexer->Conf->Vars, "LocalCharset", "iso-8859-1"));
DpsVarListReplaceStr(&rDoc->RequestHeaders, "Content-Type", buf);
dps_snprintf(buf, sizeof(buf), "%d", dps_strlen(PingBody));
DpsVarListReplaceStr(&rDoc->RequestHeaders, "Content-Length", buf);
}
DpsVarListLog(Indexer, &rDoc->RequestHeaders, DPS_LOG_DEBUG, "AUTHPING.Request");
#if 0
if (Doc == NULL || Indexer->Flags.cmd == DPS_IND_FILTER) {
DpsDocLookupConn(Indexer, rDoc);
} else {
DPS_FREE(rDoc->connp.connp);
rDoc->connp = Doc->connp;
}
#endif
result = DpsGetURL(Indexer, rDoc, NULL); /* Just get it as we need only Cookies from the headers */
DpsDocProcessResponseHeaders(Indexer, rDoc);
DpsVarListDel(&rDoc->Sections, "body");
DpsVarListLog(Indexer, &rDoc->Sections, DPS_LOG_DEBUG, "AUTHPING.Response");
if (Doc != NULL) bzero(&rDoc->connp, sizeof(rDoc->connp));
DpsDocFree(rDoc);
}
}
DpsFree(AuthPing);
}
}
while(hostinfo != NULL) {
url_id = DpsStrHash32(hostinfo);
DpsSQLResInit(&Res);
dps_snprintf(buf, sizeof(buf), "SELECT name,value,path,secure FROM cookies WHERE domain='%s'", hostinfo);
if (Indexer->flags & DPS_FLAG_UNOCON) {
DPS_GETLOCK(Indexer, DPS_LOCK_DB);
db = Indexer->Conf->dbl.db[url_id % Indexer->Conf->dbl.nitems];
} else {
db = Indexer->dbl.db[url_id % Indexer->dbl.nitems];
}
if(DPS_OK == (rc = DpsSQLQuery(db, &Res, buf))) {
rows = DpsSQLNumRows(&Res);
for(i = 0; i < rows; i++) {
DpsCookiesAdd(Indexer, hostinfo, DpsSQLValue(&Res, i, 2), DpsSQLValue(&Res, i, 0), DpsSQLValue(&Res, i, 1),
*DpsSQLValue(&Res, i, 3), 0, 0, 0);
if (*DpsSQLValue(&Res, i, 3) == 'y' && strcasecmp(Doc->CurURL.schema, "https")) continue;
if (strncasecmp(DpsSQLValue(&Res, i, 2), Doc->CurURL.path, dps_strlen(DpsSQLValue(&Res, i, 2)))) continue;
if (cookie.data_size)
DpsDSTRAppend(&cookie, "; ", 2);
DpsDSTRAppendStr(&cookie, DpsSQLValue(&Res, i, 0));
DpsDSTRAppend(&cookie, "=", 1);
DpsDSTRAppendStr(&cookie, DpsSQLValue(&Res, i, 1));
}
if (rows == 0) {
DpsCookiesAdd(Indexer, hostinfo, "/", "", "", 'n', 0, 0, 0);
}
}
DpsSQLFree(&Res);
if (Indexer->flags & DPS_FLAG_UNOCON) {
DPS_RELEASELOCK(Indexer, DPS_LOCK_DB);
}
hostinfo = strchr(hostinfo, '.');
if (hostinfo != NULL) hostinfo++;
}
}
if (cookie.data_size) {
DpsVarListReplaceStr(&Doc->RequestHeaders, "Cookie", cookie.data);
}
DpsDSTRFree(&cookie);
#endif
TRACE_OUT(Indexer);
return;
}