X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fzebraapi.c;h=12bbae06a51751e2ae9f83c1201035f6d1cc2ec7;hb=46c0e649af38cec11668a4a15ab10915b06ccbc1;hp=a1673f73544c4bddbc2690bd982fbe0e65730fa0;hpb=4478d785b7769691261005c98063b98a5a5971b3;p=idzebra-moved-to-github.git diff --git a/index/zebraapi.c b/index/zebraapi.c index a1673f7..12bbae0 100644 --- a/index/zebraapi.c +++ b/index/zebraapi.c @@ -1,5 +1,5 @@ -/* $Id: zebraapi.c,v 1.224 2006-08-14 10:40:15 adam Exp $ - Copyright (C) 1995-2006 +/* $Id: zebraapi.c,v 1.250 2007-03-13 13:46:11 adam Exp $ + Copyright (C) 1995-2007 Index Data ApS This file is part of the Zebra server. @@ -84,9 +84,10 @@ static ZEBRA_RES zebra_flush_reg (ZebraHandle zh) ZEBRA_CHECK_HANDLE(zh); yaz_log(log_level, "zebra_flush_reg"); zebraExplain_flush (zh->reg->zei, zh); - - extract_flushWriteKeys (zh, 1 /* final */); - zebra_index_merge (zh ); + + key_block_flush(zh->reg->key_block, 1); + + zebra_index_merge(zh); return ZEBRA_OK; } @@ -140,6 +141,10 @@ ZebraHandle zebra_open(ZebraService zs, Res res) zh->shadow_enable = 1; zh->m_staticrank = 0; + zh->m_segment_indexing = 0; + + zh->break_handler_func = 0; + zh->break_handler_data = 0; default_encoding = res_get_def(zh->session_res, "encoding", "ISO-8859-1"); @@ -172,12 +177,12 @@ ZebraHandle zebra_open(ZebraService zs, Res res) return zh; } -ZebraService zebra_start (const char *configName) +ZebraService zebra_start(const char *configName) { return zebra_start_res(configName, 0, 0); } -ZebraService zebra_start_res (const char *configName, Res def_res, Res over_res) +ZebraService zebra_start_res(const char *configName, Res def_res, Res over_res) { Res res; @@ -207,7 +212,18 @@ ZebraService zebra_start_res (const char *configName, Res def_res, Res over_res) res_close(res); return 0; } + if (zebra_check_res(res)) + { + yaz_log(YLOG_FATAL, "Configuration error(s) for %s", + configName); + return 0; + } } + else + { + zebra_check_res(res); + } + zh = xmalloc(sizeof(*zh)); zh->global_res = res; zh->sessions = 0; @@ -245,6 +261,7 @@ ZebraService zebra_start_res (const char *configName, Res def_res, Res over_res) } } + zh->timing = yaz_timing_create(); zh->path_root = res_get (zh->global_res, "root"); zh->nmem = nmem_create(); zh->record_classes = recTypeClass_create (zh->global_res, zh->nmem); @@ -301,6 +318,7 @@ struct zebra_register *zebra_register_open(ZebraService zs, const char *name, const char *recordCompression = 0; const char *profilePath; char cwd[1024]; + int sort_type = ZEBRA_SORT_TYPE_FLAT; ZEBRA_RES ret = ZEBRA_OK; ASSERTZS; @@ -346,8 +364,7 @@ struct zebra_register *zebra_register_open(ZebraService zs, const char *name, } getcwd(cwd, sizeof(cwd)-1); - profilePath = res_get_def(res, "profilePath", DEFAULT_PROFILE_PATH); - yaz_log(YLOG_DEBUG, "profilePath=%s cwd=%s", profilePath, cwd); + profilePath = res_get_def(res, "profilePath", 0); data1_set_tabpath (reg->dh, profilePath); data1_set_tabroot (reg->dh, reg_path); @@ -366,22 +383,19 @@ struct zebra_register *zebra_register_open(ZebraService zs, const char *name, } reg->rank_classes = NULL; - reg->key_buf = 0; - + reg->key_block = 0; reg->keys = zebra_rec_keys_open(); reg->sortKeys = zebra_rec_keys_open(); reg->records = 0; reg->dict = 0; - reg->sortIdx = 0; + reg->sort_index = 0; reg->isams = 0; reg->matchDict = 0; reg->isamc = 0; reg->isamb = 0; reg->zei = 0; - reg->key_file_no = 0; - reg->ptr_i = 0; /* installing rank classes */ zebraRankInstall (reg, rank_1_class); @@ -418,9 +432,22 @@ struct zebra_register *zebra_register_open(ZebraService zs, const char *name, yaz_log (YLOG_WARN, "dict_open failed"); ret = ZEBRA_FAIL; } - if (!(reg->sortIdx = sortIdx_open (reg->bfs, rw))) + + + if (res_get_match (res, "sortindex", "f", "f")) + sort_type = ZEBRA_SORT_TYPE_FLAT; + else if (res_get_match (res, "sortindex", "i", "f")) + sort_type = ZEBRA_SORT_TYPE_ISAMB; + else + { + yaz_log (YLOG_WARN, "bad_value for 'sortindex'"); + ret = ZEBRA_FAIL; + } + + + if (!(reg->sort_index = zebra_sort_open(reg->bfs, rw, sort_type))) { - yaz_log (YLOG_WARN, "sortIdx_open failed"); + yaz_log (YLOG_WARN, "zebra_sort_open failed"); ret = ZEBRA_FAIL; } if (res_get_match (res, "isam", "s", ISAM_DEFAULT)) @@ -531,7 +558,7 @@ static void zebra_register_close(ZebraService zs, struct zebra_register *reg) dict_close (reg->dict); if (reg->matchDict) dict_close (reg->matchDict); - sortIdx_close (reg->sortIdx); + zebra_sort_close(reg->sort_index); if (reg->isams) isams_close (reg->isams); if (reg->isamc) @@ -549,7 +576,7 @@ static void zebra_register_close(ZebraService zs, struct zebra_register *reg) zebra_rec_keys_close(reg->keys); zebra_rec_keys_close(reg->sortKeys); - xfree(reg->key_buf); + key_block_destroy(®->key_block); xfree(reg->name); xfree(reg); } @@ -558,8 +585,6 @@ ZEBRA_RES zebra_stop(ZebraService zs) { if (!zs) return ZEBRA_OK; - yaz_log (log_level, "zebra_stop"); - while (zs->sessions) { zebra_close (zs->sessions); @@ -573,6 +598,15 @@ ZEBRA_RES zebra_stop(ZebraService zs) recTypeClass_destroy(zs->record_classes); nmem_destroy(zs->nmem); res_close (zs->global_res); + + yaz_timing_stop(zs->timing); + yaz_log (YLOG_LOG, "zebra_stop: %4.2f %4.2f %4.2f", + yaz_timing_get_real(zs->timing), + yaz_timing_get_user(zs->timing), + yaz_timing_get_sys(zs->timing)); + + + yaz_timing_destroy(&zs->timing); xfree(zs); return ZEBRA_OK; } @@ -767,6 +801,15 @@ static void zebra_select_register (ZebraHandle zh, const char *new_reg) if (res_get_int(zh->res, "staticrank", &zh->m_staticrank) == ZEBRA_OK) yaz_log(YLOG_LOG, "static rank set and is %d", zh->m_staticrank); } + if (zh->res) + { + if (res_get_int(zh->res, "segment", &zh->m_segment_indexing) == + ZEBRA_OK) + { + yaz_log(YLOG_DEBUG, "segment indexing set and is %d", + zh->m_segment_indexing); + } + } } void map_basenames_func (void *vp, const char *name, const char *value) @@ -975,8 +1018,25 @@ ZEBRA_RES zebra_set_approx_limit(ZebraHandle zh, zint approx_limit) return ZEBRA_OK; } -ZEBRA_RES zebra_search_RPN(ZebraHandle zh, ODR o, Z_RPNQuery *query, - const char *setname, zint *hits) +void zebra_set_partial_result(ZebraHandle zh) +{ + zh->partial_result = 1; +} + + +ZEBRA_RES zebra_set_break_handler(ZebraHandle zh, + int (*f)(void *client_data), + void *client_data) +{ + zh->break_handler_func = f; + zh->break_handler_data = client_data; + return ZEBRA_OK; +} + +ZEBRA_RES zebra_search_RPN_x(ZebraHandle zh, ODR o, Z_RPNQuery *query, + const char *setname, zint *hits, + int *estimated_hit_count, + int *partial_resultset) { ZEBRA_RES r; @@ -987,19 +1047,31 @@ ZEBRA_RES zebra_search_RPN(ZebraHandle zh, ODR o, Z_RPNQuery *query, assert(hits); assert(setname); yaz_log(log_level, "zebra_search_rpn"); - zh->hits = 0; - *hits = 0; + + zh->partial_result = 0; if (zebra_begin_read(zh) == ZEBRA_FAIL) return ZEBRA_FAIL; r = resultSetAddRPN(zh, odr_extract_mem(o), query, - zh->num_basenames, zh->basenames, setname); + zh->num_basenames, zh->basenames, setname, + hits, estimated_hit_count); + + *partial_resultset = zh->partial_result; zebra_end_read(zh); - *hits = zh->hits; return r; } +ZEBRA_RES zebra_search_RPN(ZebraHandle zh, ODR o, Z_RPNQuery *query, + const char *setname, zint *hits) +{ + int estimated_hit_count; + int partial_resultset; + return zebra_search_RPN_x(zh, o, query, setname, hits, + &estimated_hit_count, + &partial_resultset); +} + ZEBRA_RES zebra_records_retrieve(ZebraHandle zh, ODR stream, const char *setname, Z_RecordComposition *comp, @@ -1055,12 +1127,15 @@ ZEBRA_RES zebra_records_retrieve(ZebraHandle zh, ODR stream, else if (poset[i].sysno) { char *buf; - int len; + int len = 0; zebra_snippets *hit_snippet = zebra_snippets_create(); + /* we disable hit snippets for now. It does not work well + and it slows retrieval down a lot */ +#if 0 zebra_snippets_hit_vector(zh, setname, poset[i].sysno, hit_snippet); - +#endif recs[i].errCode = zebra_record_fetch(zh, poset[i].sysno, poset[i].score, hit_snippet, @@ -1167,10 +1242,10 @@ ZEBRA_RES zebra_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, return ZEBRA_FAIL; } } - res = rpn_scan (zh, stream, zapt, attributeset, - zh->num_basenames, zh->basenames, position, - num_entries, entries, is_partial, limit_rset, 0); - zebra_end_read (zh); + res = rpn_scan(zh, stream, zapt, attributeset, + zh->num_basenames, zh->basenames, position, + num_entries, entries, is_partial, limit_rset); + zebra_end_read(zh); return res; } @@ -1308,7 +1383,7 @@ ZEBRA_RES zebra_admin_import_end (ZebraHandle zh) ZEBRA_RES zebra_admin_import_segment (ZebraHandle zh, Z_Segment *segment) { ZEBRA_RES res = ZEBRA_OK; - SYSNO sysno; + zint sysno; int i; ZEBRA_CHECK_HANDLE(zh); yaz_log(log_level, "zebra_admin_import_segment"); @@ -1350,7 +1425,7 @@ ZEBRA_RES zebra_admin_exchange_record(ZebraHandle zh, /* 4 = update. Insert/replace */ { ZEBRA_RES res; - SYSNO sysno = 0; + zint sysno = 0; char *rinfo = 0; char recid_z[256]; int db_ord; @@ -1407,7 +1482,7 @@ ZEBRA_RES zebra_admin_exchange_record(ZebraHandle zh, &sysno, 0, /* match */ 0, /* fname */ - 0, /* force update */ + 1, /* force update */ 1 /* allow update */ ); if (res == ZEBRA_FAIL) @@ -1475,12 +1550,20 @@ ZEBRA_RES zebra_drop_database(ZebraHandle zh, const char *db) if (zh->reg->isamb) { int db_ord; - zebraExplain_curDatabase (zh->reg->zei, db); - db_ord = zebraExplain_get_database_ord(zh->reg->zei); - dict_delete_subtree_ord(zh->reg->matchDict, db_ord, - 0 /* handle */, 0 /* func */); - zebraExplain_trav_ord(zh->reg->zei, zh, delete_SU_handle); - zebraExplain_removeDatabase(zh->reg->zei, zh); + if (zebraExplain_curDatabase (zh->reg->zei, db)) + { + zebra_setError(zh, YAZ_BIB1_DATABASE_DOES_NOT_EXIST, db); + ret = ZEBRA_FAIL; + } + else + { + db_ord = zebraExplain_get_database_ord(zh->reg->zei); + dict_delete_subtree_ord(zh->reg->matchDict, db_ord, + 0 /* handle */, 0 /* func */); + zebraExplain_trav_ord(zh->reg->zei, zh, delete_SU_handle); + zebraExplain_removeDatabase(zh->reg->zei, zh); + zebra_remove_file_match(zh); + } } else { @@ -1546,6 +1629,16 @@ int zebra_string_norm (ZebraHandle zh, unsigned reg_id, return wrbuf_len(wrbuf); } +/** \brief set register state (state*.LCK) + \param zh Zebra handle + \param val state + \param seqno sequence number + + val is one of: + d=writing to shadow(shadow enabled); writing to register (shadow disabled) + o=reading only + c=commit (writing to register, reading from shadow, shadow mode only) +*/ static void zebra_set_state (ZebraHandle zh, int val, int seqno) { char state_fname[256]; @@ -1623,7 +1716,7 @@ static void read_res_for_transaction(ZebraHandle zh) v = res_get_prefix(zh->res, "openRW", group, "1"); zh->m_flag_rw = atoi(v); - v = res_get_prefix(zh->res, "fileVerboseLimit", group, "100000"); + v = res_get_prefix(zh->res, "fileVerboseLimit", group, "1000"); zh->m_file_verbose_limit = atoi(v); } @@ -1655,7 +1748,6 @@ ZEBRA_RES zebra_begin_trans(ZebraHandle zh, int rw) assert (zh->res); if (rw) { - int pass; int seqno = 0; char val = '?'; const char *rval = 0; @@ -1683,6 +1775,7 @@ ZEBRA_RES zebra_begin_trans(ZebraHandle zh, int rw) zh->records_updated = 0; zh->records_deleted = 0; zh->records_processed = 0; + zh->records_skipped = 0; #if HAVE_SYS_TIMES_H times (&zh->tms1); @@ -1691,51 +1784,45 @@ ZEBRA_RES zebra_begin_trans(ZebraHandle zh, int rw) if (zh->shadow_enable) rval = res_get (zh->res, "shadow"); - for (pass = 0; pass < 2; pass++) + if (rval) { - if (rval) + zebra_lock_r(zh->lock_normal); + zebra_lock_w(zh->lock_shadow); + } + else + { + zebra_lock_w(zh->lock_normal); + zebra_lock_w(zh->lock_shadow); + } + zebra_get_state (zh, &val, &seqno); + if (val != 'o') + { + /* either we didn't finish commit or shadow is dirty */ + if (!rval) { - zebra_lock_r (zh->lock_normal); - zebra_lock_w (zh->lock_shadow); + yaz_log(YLOG_WARN, "previous transaction did not finish " + "(shadow disabled)"); } - else + zebra_unlock (zh->lock_shadow); + zebra_unlock (zh->lock_normal); + if (zebra_commit (zh)) { - zebra_lock_w (zh->lock_normal); - zebra_lock_w (zh->lock_shadow); + zh->trans_no--; + zh->trans_w_no = 0; + return ZEBRA_FAIL; } - - zebra_get_state (zh, &val, &seqno); - if (val == 'c') + if (rval) { - yaz_log (YLOG_WARN, "previous transaction didn't finish commit"); - zebra_unlock (zh->lock_shadow); - zebra_unlock (zh->lock_normal); - zebra_commit (zh); - continue; + zebra_lock_r(zh->lock_normal); + zebra_lock_w(zh->lock_shadow); } - else if (val == 'd') + else { - if (rval) - { - BFiles bfs = bfs_create (res_get (zh->res, "shadow"), - zh->path_reg); - yaz_log (YLOG_WARN, "previous transaction didn't reach commit"); - bf_commitClean (bfs, rval); - bfs_destroy (bfs); - } - else - { - yaz_log (YLOG_WARN, "your previous transaction didn't finish"); - } + zebra_lock_w(zh->lock_normal); + zebra_lock_w(zh->lock_shadow); } - break; - } - if (pass == 2) - { - yaz_log (YLOG_FATAL, "zebra_begin_trans couldn't finish commit"); - abort(); - return ZEBRA_FAIL; } + zebra_set_state (zh, 'd', seqno); zh->reg = zebra_register_open(zh->service, zh->reg_name, @@ -1970,6 +2057,8 @@ static ZEBRA_RES zebra_commit_ex(ZebraHandle zh, int clean_only) char val; const char *rval; BFiles bfs; + ZEBRA_RES res = ZEBRA_OK; + ASSERTZH; zebra_select_default_database(zh); @@ -1978,60 +2067,74 @@ static ZEBRA_RES zebra_commit_ex(ZebraHandle zh, int clean_only) zh->errCode = YAZ_BIB1_DATABASE_UNAVAILABLE; return ZEBRA_FAIL; } - rval = res_get (zh->res, "shadow"); + rval = res_get(zh->res, "shadow"); if (!rval) { yaz_log (YLOG_WARN, "Cannot perform commit - No shadow area defined"); return ZEBRA_OK; } - zebra_lock_w (zh->lock_normal); - zebra_lock_r (zh->lock_shadow); + zebra_lock_w(zh->lock_normal); + zebra_lock_r(zh->lock_shadow); - bfs = bfs_create (res_get (zh->res, "register"), zh->path_reg); + bfs = bfs_create(res_get (zh->res, "register"), zh->path_reg); if (!bfs) { zebra_unlock(zh->lock_shadow); zebra_unlock(zh->lock_normal); return ZEBRA_FAIL; } - zebra_get_state (zh, &val, &seqno); + zebra_get_state(zh, &val, &seqno); + + if (val == 'd') + { + /* shadow area is dirty and so we must throw it away */ + yaz_log(YLOG_WARN, "previous transaction didn't reach commit"); + clean_only = 1; + } if (rval && *rval) bf_cache (bfs, rval); if (bf_commitExists (bfs)) { if (clean_only) - zebra_set_state (zh, 'd', seqno); + zebra_set_state(zh, 'd', seqno); else { - zebra_set_state (zh, 'c', seqno); + zebra_set_state(zh, 'c', seqno); - yaz_log (YLOG_DEBUG, "commit start"); - bf_commitExec (bfs); -#ifndef WIN32 - sync (); -#endif + yaz_log(YLOG_DEBUG, "commit start"); + if (bf_commitExec (bfs)) + res = ZEBRA_FAIL; + } + if (res == ZEBRA_OK) + { + seqno++; + zebra_set_state(zh, 'o', seqno); + + zebra_unlock(zh->lock_shadow); + zebra_unlock(zh->lock_normal); + + zebra_lock_w(zh->lock_shadow); + bf_commitClean(bfs, rval); + zebra_unlock(zh->lock_shadow); + } + else + { + zebra_unlock(zh->lock_shadow); + zebra_unlock(zh->lock_normal); + yaz_log(YLOG_WARN, "zebra_commit: failed"); } - seqno++; - zebra_set_state (zh, 'o', seqno); - - zebra_unlock (zh->lock_shadow); - zebra_unlock (zh->lock_normal); - - zebra_lock_w(zh->lock_shadow); - bf_commitClean (bfs, rval); - zebra_unlock (zh->lock_shadow); } else { zebra_unlock(zh->lock_shadow); zebra_unlock(zh->lock_normal); - yaz_log (log_level, "nothing to commit"); + yaz_log(log_level, "nothing to commit"); } - bfs_destroy (bfs); + bfs_destroy(bfs); - return ZEBRA_OK; + return res; } ZEBRA_RES zebra_clean(ZebraHandle zh) @@ -2202,12 +2305,12 @@ void zebra_set_shadow_enable (ZebraHandle zh, int value) ZEBRA_RES zebra_add_record(ZebraHandle zh, const char *buf, int buf_size) { - return zebra_update_record(zh, 0, 0 /* sysno */, 0, 0, buf, buf_size, 0); + return zebra_update_record(zh, 0, 0 /* sysno */, 0, 0, buf, buf_size, 1); } ZEBRA_RES zebra_insert_record(ZebraHandle zh, const char *recordType, - SYSNO *sysno, const char *match, + zint *sysno, const char *match, const char *fname, const char *buf, int buf_size, int force_update) { @@ -2228,7 +2331,7 @@ ZEBRA_RES zebra_insert_record(ZebraHandle zh, recordType, sysno, match, fname, - 0, + 1, 0); /* allow_update */ if (zebra_end_trans(zh) != ZEBRA_OK) { @@ -2238,12 +2341,12 @@ ZEBRA_RES zebra_insert_record(ZebraHandle zh, return res; } -ZEBRA_RES zebra_update_record (ZebraHandle zh, - const char *recordType, - SYSNO* sysno, const char *match, - const char *fname, - const char *buf, int buf_size, - int force_update) +ZEBRA_RES zebra_update_record(ZebraHandle zh, + const char *recordType, + zint *sysno, const char *match, + const char *fname, + const char *buf, int buf_size, + int force_update) { ZEBRA_RES res; @@ -2275,12 +2378,12 @@ ZEBRA_RES zebra_update_record (ZebraHandle zh, return res; } -ZEBRA_RES zebra_delete_record (ZebraHandle zh, - const char *recordType, - SYSNO *sysno, const char *match, - const char *fname, - const char *buf, int buf_size, - int force_update) +ZEBRA_RES zebra_delete_record(ZebraHandle zh, + const char *recordType, + zint *sysno, const char *match, + const char *fname, + const char *buf, int buf_size, + int force_update) { ZEBRA_RES res; @@ -2323,6 +2426,7 @@ ZEBRA_RES zebra_search_PQF(ZebraHandle zh, const char *pqf_query, Z_RPNQuery *query; ODR odr; + ZEBRA_CHECK_HANDLE(zh); odr = odr_createmem(ODR_ENCODE); @@ -2437,11 +2541,11 @@ void zebra_setError_zint(ZebraHandle zh, int code, zint i) zh->errString = nmem_strdup(zh->nmem_error, vstr); } -void zebra_lock_prefix (Res res, char *path) +void zebra_lock_prefix(Res res, char *path) { const char *lock_dir = res_get_def (res, "lockDir", ""); - - strcpy (path, lock_dir); + + strcpy(path, lock_dir); if (*path && path[strlen(path)-1] != '/') strcat (path, "/"); }