I found it a good idea to add to aspseek the ability,
to index just one specific site on demand.
The code is pretty straightforward.
Added the -b flag to denote site_id as an command line option.
[root@achill aspseek-1.0.2]# diff include/database.h include/database.h.orig
109c109
< int MarkForReindex(int SID);
---
> int MarkForReindex();
[root@achill aspseek-1.0.2]#
[root@achill aspseek-1.0.2]# diff src/index.cpp src/index.cpp.orig
54,56d53
<
< int achSiteId;
<
103d99
< fprintf(stderr, " -b siteid Index Only urlz from this site\n");
343c339
< result = database->MarkForReindex(achSiteId);
---
> result = database->MarkForReindex();
382c378
< if (database->MarkForReindex(achSiteId) == IND_ERROR) return 1;
---
> if (database->MarkForReindex() == IND_ERROR) return 1;
843a840
>
879c876
< achSiteId = 0;
---
>
881c878
< while ((ch = getopt(argc, argv,
"FUMVEKLCSaheomdqiwlDB?t:u:s:n:L:D:N:R:f:c:r:P:T:g:A:b:")) != -1)
---
> while ((ch = getopt(argc, argv,
>"FUMVEKLCSaheomdqiwlDB?t:u:s:n:L:D:N:R:f:c:r:P:T:g:A:")) != -1)
939d935
< case 'b': achSiteId = atoi(optarg); break;
[root@achill aspseek-1.0.2]#
[root@achill aspseek-1.0.2]# diff src/squeue.cpp src/squeue.cpp.orig
37,38d36
< extern int achSiteId;
<
214,218d211
< char achClause[200];
< if (achSiteId)
< sprintf(achClause,"AND site_id=%d",achSiteId);
< else
< sprintf(achClause,"");
225c218
< sprintf(qSel, "SELECT url_id,site_id,next_index_time FROM urlword
WHERE deleted=0 AND next_index_time<=%d AND next_index_time > %i %s %s ORDER BY
next_index_time LIMIT %lu", (int)now(), m_maxtime, Limit.c_str(), achClause,
NextDocLimit);
---
> sprintf(qSel, "SELECT url_id,site_id,next_index_time FROM urlword
>WHERE deleted=0 AND next_index_time<=%d AND next_index_time > %i %s ORDER BY
>next_index_time LIMIT %lu", (int)now(), m_maxtime, Limit.c_str(), NextDocLimit);
[root@achill aspseek-1.0.2]#
[root@achill aspseek-1.0.2]#
Plz let me know if you plan to support this in a later version.
Thanx
Achilleus
