X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fzebraapi.c;h=f74eae4434fd4f399adbbdc02ec7ffd4c100b54e;hb=810bce66201f40acfd7e8577d3997e6ea385f1cf;hp=d2f6d9b2a77cc40a6a87ada1237f545afd88f24c;hpb=1ccf2613ceef2359f589cb3dd7e72a899c618b2f;p=idzebra-moved-to-github.git diff --git a/index/zebraapi.c b/index/zebraapi.c index d2f6d9b..d536064 100644 --- a/index/zebraapi.c +++ b/index/zebraapi.c @@ -1,152 +1,28 @@ -/* - * Copyright (C) 1995-2000, Index Data - * All rights reserved. - * - * $Log: zebraapi.c,v $ - * Revision 1.42 2000-12-05 10:01:44 adam - * Fixed bug regarding user-defined attribute sets. - * - * Revision 1.41 2000/12/01 17:59:08 adam - * Fixed bug regarding online updates on WIN32. - * When zebra.cfg is not available the server will not abort. - * - * Revision 1.40 2000/11/29 15:21:31 adam - * Fixed problem with passwd db. - * - * Revision 1.39 2000/11/29 14:24:01 adam - * Script configure uses yaz pthreads options. Added locking for - * zebra_register_{lock,unlock}. - * - * Revision 1.38 2000/11/08 13:46:58 adam - * Fixed scan: server could break if bad attribute/database was selected. - * Work on remote update. - * - * Revision 1.37 2000/10/17 12:37:09 adam - * Fixed notification of live-updates. Fixed minor problem with mf_init - * where it didn't handle shadow area file names correctly. - * - * Revision 1.36 2000/09/06 08:59:36 adam - * Using read-only (for now) for server. - * - * Revision 1.35 2000/07/07 12:49:20 adam - * Optimized resultSetInsert{Rank,Sort}. - * - * Revision 1.34 2000/06/09 13:56:38 ian - * Added some logging on Authentication and searches. - * - * Revision 1.33 2000/05/18 12:01:36 adam - * System call times(2) used again. More 64-bit fixes. - * - * Revision 1.32 2000/04/19 14:35:59 adam - * WIN32 update (this version is known not to work on Windows). - * - * Revision 1.31 2000/04/05 10:07:02 adam - * Minor zebra compile fix. - * - * Revision 1.30 2000/04/05 09:49:35 adam - * On Unix, zebra/z'mbol uses automake. - * - * Revision 1.29 2000/03/20 19:08:36 adam - * Added remote record import using Z39.50 extended services and Segment - * Requests. - * - * Revision 1.28 2000/03/15 15:00:30 adam - * First work on threaded version. - * - * Revision 1.27 2000/02/24 12:31:17 adam - * Added zebra_string_norm. - * - * Revision 1.26 1999/11/30 13:48:03 adam - * Improved installation. Updated for inclusion of YAZ header files. - * - * Revision 1.25 1999/11/04 15:00:45 adam - * Implemented delete result set(s). - * - * Revision 1.24 1999/10/14 14:33:50 adam - * Added truncation 5=106. - * - * Revision 1.23 1999/09/07 11:36:32 adam - * Minor changes. - * - * Revision 1.22 1999/08/02 10:13:47 adam - * Fixed bug regarding zebra_hits. - * - * Revision 1.21 1999/07/14 10:59:26 adam - * Changed functions isc_getmethod, isams_getmethod. - * Improved fatal error handling (such as missing EXPLAIN schema). - * - * Revision 1.20 1999/07/06 12:28:04 adam - * Updated record index structure. Format includes version ID. Compression - * algorithm ID is stored for each record block. - * - * Revision 1.19 1999/05/26 07:49:13 adam - * C++ compilation. - * - * Revision 1.18 1999/05/15 14:36:38 adam - * Updated dictionary. Implemented "compression" of dictionary. - * - * Revision 1.17 1999/05/12 13:08:06 adam - * First version of ISAMS. - * - * Revision 1.16 1999/02/19 10:38:30 adam - * Implemented chdir-setting. - * - * Revision 1.15 1999/02/17 12:18:12 adam - * Fixed zebra_close so that a NULL pointer is ignored. - * - * Revision 1.14 1999/02/02 14:51:11 adam - * Updated WIN32 code specific sections. Changed header. - * - * Revision 1.13 1998/12/16 12:23:30 adam - * Added facility for database name mapping using resource mapdb. - * - * Revision 1.12 1998/11/16 10:18:10 adam - * Better error reporting for result sets. - * - * Revision 1.11 1998/10/16 08:14:34 adam - * Updated record control system. - * - * Revision 1.10 1998/09/22 10:03:42 adam - * Changed result sets to be persistent in the sense that they can - * be re-searched if needed. - * Fixed memory leak in rsm_or. - * - * Revision 1.9 1998/09/02 13:53:17 adam - * Extra parameter decode added to search routines to implement - * persistent queries. - * - * Revision 1.8 1998/08/24 17:29:23 adam - * Minor changes. - * - * Revision 1.7 1998/06/24 12:16:13 adam - * Support for relations on text operands. Open range support in - * DFA module (i.e. [-j], [g-]). - * - * Revision 1.6 1998/06/22 11:36:47 adam - * Added authentication check facility to zebra. - * - * Revision 1.5 1998/06/13 00:14:08 adam - * Minor changes. - * - * Revision 1.4 1998/06/12 12:22:12 adam - * Work on Zebra API. - * - * Revision 1.3 1998/05/27 16:57:44 adam - * Zebra returns surrogate diagnostic for single records when - * appropriate. - * - * Revision 1.2 1998/05/20 10:12:19 adam - * Implemented automatic EXPLAIN database maintenance. - * Modified Zebra to work with ASN.1 compiled version of YAZ. - * - * Revision 1.1 1998/03/05 08:45:13 adam - * New result set model and modular ranking system. Moved towards - * descent server API. System information stored as "SGML" records. - * - */ +/* $Id: zebraapi.c,v 1.175 2005-06-09 10:39:53 adam Exp $ + Copyright (C) 1995-2005 + Index Data ApS + +This file is part of the Zebra server. + +Zebra is free software; you can redistribute it and/or modify it under +the terms of the GNU General Public License as published by the Free +Software Foundation; either version 2, or (at your option) any later +version. + +Zebra is distributed in the hope that it will be useful, but WITHOUT ANY +WARRANTY; without even the implied warranty of MERCHANTABILITY or +FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +for more details. + +You should have received a copy of the GNU General Public License +along with Zebra; see the file LICENSE.zebra. If not, write to the +Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA +02111-1307, USA. +*/ #include #include +#include #ifdef WIN32 #include #include @@ -156,15 +32,32 @@ #endif #include -#include "zserver.h" +#include +#include +#include "index.h" #include +#include + +/* simple asserts to validate the most essential input args */ +#define ASSERTZH assert(zh && zh->service) +#define ASSERTZHRES assert(zh && zh->service && zh->res) +#define ASSERTZS assert(zs) + +static int log_level = 0; +static int log_level_initialized = 0; + +static Res zebra_open_res (ZebraHandle zh); +static void zebra_close_res (ZebraHandle zh); -static void zebra_chdir (ZebraService zh) +static void zebra_chdir (ZebraService zs) { - const char *dir = res_get (zh->res, "chdir"); + const char *dir ; + ASSERTZS; + yaz_log(log_level, "zebra_chdir"); + dir = res_get (zs->global_res, "chdir"); if (!dir) return; - logf (LOG_DEBUG, "chdir %s", dir); + yaz_log (YLOG_DEBUG, "chdir %s", dir); #ifdef WIN32 _chdir(dir); #else @@ -172,114 +65,80 @@ static void zebra_chdir (ZebraService zh) #endif } -static int extract_rec_in_mem (ZebraHandle zh, const char *recordType, - const char *buf, size_t buf_size, - const char *databaseName, int delete_flag, - int test_mode, int *sysno, - int store_keys, int store_data, - const char *match_criteria); - -static int explain_extract (void *handle, Record rec, data1_node *n); -static void extract_index (ZebraHandle zh); - -static void zebra_register_unlock (ZebraHandle zh); - -static int zebra_register_activate (ZebraService zh, int rw); -static int zebra_register_deactivate (ZebraService zh); - -static int zebra_register_lock (ZebraHandle zh, int rw) +static void zebra_flush_reg (ZebraHandle zh) { - time_t lastChange; - int state; - zh->errCode = 0; - zh->errString = 0; - if (!zh->service->active) - { - zh->errCode = 1019; - return 1; - } + ASSERTZH; + yaz_log(log_level, "zebra_flush_reg"); + zebra_clearError(zh); + zebraExplain_flush (zh->reg->zei, zh); -#if HAVE_SYS_TIMES_H - times (&zh->tms1); -#endif - - zebra_mutex_cond_lock (&zh->service->session_lock); - - state = zebra_server_lock_get_state(zh->service, &lastChange); - - zebra_server_lock (zh->service, state); - - switch (state) - { - case 'c': - state = 1; - break; - default: - state = 0; - } - if (rw) - logf (LOG_LOG, "Register in read/write mode"); - else if (zh->service->registerState == state) - { - logf (LOG_LOG, "registerChange = %ld lastChange = %ld", - (long) zh->service->registerChange, (long)lastChange); - if (zh->service->registerChange >= lastChange) - { - return 0; - } - logf (LOG_LOG, "Register completely updated since last access"); - } - else if (zh->service->registerState == -1) - logf (LOG_LOG, "Reading register using state %d pid=%ld", state, - (long) getpid()); - else - logf (LOG_LOG, "Register has changed state from %d to %d", - zh->service->registerState, state); - zh->service->registerChange = lastChange; - - zebra_register_deactivate (zh->service); - - zh->service->registerState = state; - - zebra_register_activate (zh->service, rw); - return 0; + extract_flushWriteKeys (zh, 1 /* final */); + zebra_index_merge (zh ); } - -static void zebra_register_unlock (ZebraHandle zh) -{ - if (zh->service->registerState != -1) - zebra_server_unlock (zh->service, zh->service->registerState); - zebra_mutex_cond_unlock (&zh->service->session_lock); -#if HAVE_SYS_TIMES_H - times (&zh->tms2); - logf (LOG_LOG, "user/system: %ld/%ld", - (long) (zh->tms2.tms_utime - zh->tms1.tms_utime), - (long) (zh->tms2.tms_stime - zh->tms1.tms_stime)); - -#endif -} +static struct zebra_register *zebra_register_open (ZebraService zs, + const char *name, + int rw, int useshadow, + Res res, + const char *reg_path); +static void zebra_register_close (ZebraService zs, struct zebra_register *reg); ZebraHandle zebra_open (ZebraService zs) { ZebraHandle zh; + const char *default_encoding; + if (!log_level_initialized) + { + log_level = yaz_log_module_level("zebraapi"); + log_level_initialized = 1; + } - assert (zs); - if (zs->stop_flag) - return 0; + yaz_log(log_level, "zebra_open"); + + if (!zs) + return 0; - zh = (ZebraHandle) xmalloc (sizeof(*zh)); - yaz_log (LOG_LOG, "zebra_open zs=%p returns %p", zs, zh); + zh = (ZebraHandle) xmalloc(sizeof(*zh)); + yaz_log (YLOG_DEBUG, "zebra_open zs=%p returns %p", zs, zh); zh->service = zs; + zh->reg = 0; /* no register attached yet */ zh->sets = 0; zh->destroyed = 0; zh->errCode = 0; zh->errString = 0; + zh->res = 0; + zh->user_perm = 0; + + zh->reg_name = xstrdup (""); + zh->path_reg = 0; + zh->num_basenames = 0; + zh->basenames = 0; + + zh->approx_limit = 1000000000; + zh->trans_no = 0; + zh->trans_w_no = 0; + + zh->lock_normal = 0; + zh->lock_shadow = 0; + + zh->shadow_enable = 1; + + default_encoding = res_get_def(zs->global_res, "encoding", "ISO-8859-1"); + + zh->iconv_to_utf8 = + yaz_iconv_open ("UTF-8", default_encoding); + if (zh->iconv_to_utf8 == 0) + yaz_log (YLOG_WARN, "iconv: %s to UTF-8 unsupported", + default_encoding); + zh->iconv_from_utf8 = + yaz_iconv_open (default_encoding, "UTF-8"); + if (zh->iconv_to_utf8 == 0) + yaz_log (YLOG_WARN, "iconv: UTF-8 to %s unsupported", + default_encoding); + + zh->record_encoding = 0; - zh->key_buf = 0; - zh->admin_databaseName = 0; - zebra_mutex_cond_lock (&zs->session_lock); zh->next = zs->sessions; @@ -287,256 +146,406 @@ ZebraHandle zebra_open (ZebraService zs) zebra_mutex_cond_unlock (&zs->session_lock); + zh->store_data_buf = 0; + + zh->m_limit = zebra_limit_create(1, 0); + + zh->nmem_error = nmem_create(); + return zh; } - ZebraService zebra_start (const char *configName) { - ZebraService zh = xmalloc (sizeof(*zh)); - - yaz_log (LOG_LOG, "zebra_start %s", configName); - - zh->configName = xstrdup(configName); - zh->sessions = 0; - zh->stop_flag = 0; - zh->active = 1; + return zebra_start_res(configName, 0, 0); +} - zh->registerState = -1; - zh->registerChange = 0; +ZebraService zebra_start_res (const char *configName, Res def_res, Res over_res) +{ + Res res; - if (!(zh->res = res_open (zh->configName))) + if (!log_level_initialized) { - logf (LOG_WARN, "Failed to read resources `%s'", zh->configName); -// return zh; + log_level = yaz_log_module_level("zebraapi"); + log_level_initialized = 1; } - zebra_chdir (zh); - zebra_server_lock_init (zh); - zebra_mutex_cond_init (&zh->session_lock); - if (!res_get (zh->res, "passwd")) - zh->passwd_db = NULL; - else + + yaz_log(YLOG_LOG, "zebra_start %s %s",configName, ZEBRAVER); + assert(configName); + + if ((res = res_open (configName, def_res, over_res))) { - zh->passwd_db = passwd_db_open (); - if (!zh->passwd_db) - logf (LOG_WARN|LOG_ERRNO, "passwd_db_open failed"); - else - passwd_db_file (zh->passwd_db, res_get (zh->res, "passwd")); + const char *passwd_plain = 0; + const char *passwd_encrypt = 0; + ZebraService zh = xmalloc(sizeof(*zh)); + + yaz_log (YLOG_DEBUG, "Read resources `%s'", configName); + + zh->global_res = res; + zh->configName = xstrdup(configName); + zh->sessions = 0; + + zebra_chdir (zh); + + zebra_mutex_cond_init (&zh->session_lock); + passwd_plain = res_get (zh->global_res, "passwd"); + passwd_encrypt = res_get (zh->global_res, "passwd.c"); + + if (!passwd_plain && !passwd_encrypt) + zh->passwd_db = NULL; + else + { + zh->passwd_db = passwd_db_open(); + if (!zh->passwd_db) + yaz_log (YLOG_WARN|YLOG_ERRNO, "passwd_db_open failed"); + else + { + if (passwd_plain) + passwd_db_file_plain(zh->passwd_db, passwd_plain); + if (passwd_encrypt) + passwd_db_file_crypt(zh->passwd_db, passwd_encrypt); + } + } + zh->path_root = res_get (zh->global_res, "root"); + zh->nmem = nmem_create(); + zh->record_classes = recTypeClass_create (zh->global_res, zh->nmem); + return zh; } + return 0; +} - return zh; +void zebra_filter_info(ZebraService zs, void *cd, + void (*cb)(void *cd, const char *name)) +{ + ASSERTZS; + assert(cb); + recTypeClass_info(zs->record_classes, cd, cb); +} + +void zebra_pidfname(ZebraService zs, char *path) +{ + ASSERTZS; + zebra_lock_prefix (zs->global_res, path); + strcat(path, "zebrasrv.pid"); } -static int zebra_register_activate (ZebraService zh, int rw) +Dict dict_open_res (BFiles bfs, const char *name, int cache, int rw, + int compact_flag, Res res) { - if (zh->active > 1) + int page_size = 4096; + char resource_str[200]; + const char *v; + sprintf (resource_str, "dict.%.100s.pagesize", name); + assert(bfs); + assert(name); + + v = res_get(res, resource_str); + if (v) { - yaz_log (LOG_LOG, "zebra_register_activate (ignored since active=%d)", - zh->active); - return 0; + page_size = atoi(v); + yaz_log(YLOG_LOG, "Using custom dictionary page size %d for %s", + page_size, name); } - yaz_log (LOG_LOG, "zebra_register_activate shadow=%s", - zh->registerState ? "yes" : "no"); + return dict_open(bfs, name, cache, rw, compact_flag, page_size); +} - zh->dh = data1_create (); - if (!zh->dh) - return -1; - zh->bfs = bfs_create (res_get (zh->res, "register")); - if (!zh->bfs) + +static +struct zebra_register *zebra_register_open (ZebraService zs, const char *name, + int rw, int useshadow, Res res, + const char *reg_path) +{ + struct zebra_register *reg; + int record_compression = REC_COMPRESS_NONE; + const char *recordCompression = 0; + const char *profilePath; + char cwd[1024]; + + ASSERTZS; + + reg = xmalloc(sizeof(*reg)); + + assert (name); + reg->name = xstrdup (name); + + reg->seqno = 0; + reg->last_val = 0; + + assert (res); + + yaz_log (YLOG_DEBUG, "zebra_register_open rw=%d useshadow=%d p=%p n=%s rp=%s", + rw, useshadow, reg, name, reg_path ? reg_path : "(none)"); + + reg->dh = data1_createx (DATA1_FLAG_XML); + if (!reg->dh) { - data1_destroy(zh->dh); - return -1; + xfree(reg); + return 0; } - bf_lockDir (zh->bfs, res_get (zh->res, "lockDir")); - bf_cache (zh->bfs, zh->registerState ? res_get (zh->res, "shadow") : NULL); - data1_set_tabpath (zh->dh, res_get(zh->res, "profilePath")); - zh->recTypes = recTypes_init (zh->dh); - recTypes_default_handlers (zh->recTypes); - - zh->records = NULL; - zh->zebra_maps = zebra_maps_open (zh->res); - zh->rank_classes = NULL; - - zh->records = 0; - zh->dict = 0; - zh->sortIdx = 0; - zh->isams = 0; -#if ZMBOL - zh->isam = 0; - zh->isamc = 0; - zh->isamd = 0; -#endif - zh->zei = 0; + reg->bfs = bfs_create (res_get (res, "register"), reg_path); + if (!reg->bfs) + { + data1_destroy(reg->dh); + xfree(reg); + return 0; + } + if (useshadow) + { + if (bf_cache (reg->bfs, res_get (res, "shadow")) == ZEBRA_FAIL) + { + bfs_destroy(reg->bfs); + data1_destroy(reg->dh); + xfree(reg); + return 0; + } + } + + getcwd(cwd, sizeof(cwd)-1); + profilePath = res_get_def(res, "profilePath", DEFAULT_PROFILE_PATH); + yaz_log(YLOG_DEBUG, "profilePath=%s cwd=%s", profilePath, cwd); + + data1_set_tabpath (reg->dh, profilePath); + data1_set_tabroot (reg->dh, reg_path); + reg->recTypes = recTypes_init (zs->record_classes, reg->dh); + + reg->zebra_maps = zebra_maps_open (res, reg_path); + reg->rank_classes = NULL; + + reg->key_buf = 0; + + reg->keys.buf_max = 0; + reg->keys.buf = 0; + reg->keys.codec_handle = iscz1_start(); + + reg->sortKeys.buf = 0; + reg->sortKeys.buf_max = 0; + + reg->records = 0; + reg->dict = 0; + reg->sortIdx = 0; + reg->isams = 0; + reg->matchDict = 0; + reg->isamc = 0; + reg->isamb = 0; + reg->zei = 0; + reg->matchDict = 0; + reg->key_file_no = 0; + reg->ptr_i = 0; - zebraRankInstall (zh, rank1_class); + zebraRankInstall (reg, rank1_class); + zebraRankInstall (reg, rankzv_class); - if (!(zh->records = rec_open (zh->bfs, rw, 0))) + recordCompression = res_get_def (res, "recordCompression", "none"); + if (!strcmp (recordCompression, "none")) + record_compression = REC_COMPRESS_NONE; + if (!strcmp (recordCompression, "bzip2")) + record_compression = REC_COMPRESS_BZIP2; + + if (!(reg->records = rec_open (reg->bfs, rw, record_compression))) { - logf (LOG_WARN, "rec_open"); - return -1; + yaz_log (YLOG_WARN, "rec_open failed"); + return 0; } - if (!(zh->dict = dict_open (zh->bfs, FNAME_DICT, 80, rw, 0))) + if (rw) { - logf (LOG_WARN, "dict_open"); - return -1; + reg->matchDict = dict_open_res (reg->bfs, GMATCH_DICT, 20, 1, 0, res); } - if (!(zh->sortIdx = sortIdx_open (zh->bfs, rw))) + if (!(reg->dict = dict_open_res (reg->bfs, FNAME_DICT, 40, rw, 0, res))) { - logf (LOG_WARN, "sortIdx_open"); - return -1; + yaz_log (YLOG_WARN, "dict_open failed"); + return 0; + } + if (!(reg->sortIdx = sortIdx_open (reg->bfs, rw))) + { + yaz_log (YLOG_WARN, "sortIdx_open failed"); + return 0; } - if (res_get_match (zh->res, "isam", "s", ISAM_DEFAULT)) + if (res_get_match (res, "isam", "s", ISAM_DEFAULT)) { struct ISAMS_M_s isams_m; - if (!(zh->isams = isams_open (zh->bfs, FNAME_ISAMS, rw, - key_isams_m(zh->res, &isams_m)))) + if (!(reg->isams = isams_open (reg->bfs, FNAME_ISAMS, rw, + key_isams_m(res, &isams_m)))) { - logf (LOG_WARN, "isams_open"); - return -1; + yaz_log (YLOG_WARN, "isams_open failed"); + return 0; } } -#if ZMBOL - else if (res_get_match (zh->res, "isam", "i", ISAM_DEFAULT)) + if (res_get_match (res, "isam", "c", ISAM_DEFAULT)) { - if (!(zh->isam = is_open (zh->bfs, FNAME_ISAM, key_compare, rw, - sizeof (struct it_key), zh->res))) + struct ISAMC_M_s isamc_m; + if (!(reg->isamc = isamc_open (reg->bfs, FNAME_ISAMC, + rw, key_isamc_m(res, &isamc_m)))) { - logf (LOG_WARN, "is_open"); - return -1; + yaz_log (YLOG_WARN, "isamc_open failed"); + return 0; } } - else if (res_get_match (zh->res, "isam", "c", ISAM_DEFAULT)) + if (res_get_match (res, "isam", "b", ISAM_DEFAULT)) { struct ISAMC_M_s isamc_m; - if (!(zh->isamc = isc_open (zh->bfs, FNAME_ISAMC, - rw, key_isamc_m(zh->res, &isamc_m)))) + + if (!(reg->isamb = isamb_open (reg->bfs, "isamb", + rw, key_isamc_m(res, &isamc_m), 0))) { - logf (LOG_WARN, "isc_open"); - return -1; + yaz_log (YLOG_WARN, "isamb_open failed"); + return 0; } } - else if (res_get_match (zh->res, "isam", "d", ISAM_DEFAULT)) + if (res_get_match (res, "isam", "bc", ISAM_DEFAULT)) { - struct ISAMD_M_s isamd_m; + struct ISAMC_M_s isamc_m; - if (!(zh->isamd = isamd_open (zh->bfs, FNAME_ISAMD, - rw, key_isamd_m(zh->res, &isamd_m)))) + if (!(reg->isamb = isamb_open (reg->bfs, "isamb", + rw, key_isamc_m(res, &isamc_m), 1))) { - logf (LOG_WARN, "isamd_open"); - return -1; + yaz_log (YLOG_WARN, "isamb_open failed"); + return 0; } } -#endif - zh->zei = zebraExplain_open (zh->records, zh->dh, - zh->res, rw, 0 /* rGroup */, - explain_extract); - if (!zh->zei) + if (res_get_match (res, "isam", "null", ISAM_DEFAULT)) { - logf (LOG_WARN, "Cannot obtain EXPLAIN information"); - return -1; + struct ISAMC_M_s isamc_m; + + if (!(reg->isamb = isamb_open (reg->bfs, "isamb", + rw, key_isamc_m(res, &isamc_m), -1))) + { + yaz_log (YLOG_WARN, "isamb_open failed"); + return 0; + } } - zh->active = 2; - yaz_log (LOG_LOG, "zebra_register_activate ok"); - return 0; + reg->zei = zebraExplain_open (reg->records, reg->dh, + res, rw, reg, + explain_extract); + if (!reg->zei) + { + yaz_log (YLOG_WARN, "Cannot obtain EXPLAIN information"); + return 0; + } + reg->active = 2; + yaz_log (YLOG_DEBUG, "zebra_register_open ok p=%p", reg); + return reg; } -void zebra_admin_shutdown (ZebraHandle zh) +ZEBRA_RES zebra_admin_shutdown (ZebraHandle zh) { + ASSERTZH; + yaz_log(log_level, "zebra_admin_shutdown"); + zebra_clearError(zh); + zebra_mutex_cond_lock (&zh->service->session_lock); zh->service->stop_flag = 1; - if (!zh->service->sessions) - zebra_register_deactivate(zh->service); - zh->service->active = 0; zebra_mutex_cond_unlock (&zh->service->session_lock); + return ZEBRA_OK; } -void zebra_admin_start (ZebraHandle zh) +ZEBRA_RES zebra_admin_start (ZebraHandle zh) { - ZebraService zs = zh->service; - zh->errCode = 0; + ZebraService zs; + ASSERTZH; + yaz_log(log_level, "zebra_admin_start"); + zebra_clearError(zh); + zs = zh->service; zebra_mutex_cond_lock (&zs->session_lock); - if (!zs->stop_flag) - zh->service->active = 1; zebra_mutex_cond_unlock (&zs->session_lock); + return ZEBRA_OK; } -static int zebra_register_deactivate (ZebraService zs) +static void zebra_register_close (ZebraService zs, struct zebra_register *reg) { - zs->stop_flag = 0; - if (zs->active <= 1) - { - yaz_log(LOG_LOG, "zebra_register_deactivate (ignored since active=%d)", - zs->active); - return 0; - } - yaz_log(LOG_LOG, "zebra_register_deactivate"); + ASSERTZS; + assert(reg); + yaz_log(YLOG_DEBUG, "zebra_register_close p=%p", reg); + reg->stop_flag = 0; zebra_chdir (zs); - if (zs->records) - { - zebraExplain_close (zs->zei, 0); - dict_close (zs->dict); - sortIdx_close (zs->sortIdx); - if (zs->isams) - isams_close (zs->isams); -#if ZMBOL - if (zs->isam) - is_close (zs->isam); - if (zs->isamc) - isc_close (zs->isamc); - if (zs->isamd) - isamd_close (zs->isamd); -#endif - rec_close (&zs->records); + if (reg->records) + { + zebraExplain_close (reg->zei); + dict_close (reg->dict); + if (reg->matchDict) + dict_close (reg->matchDict); + sortIdx_close (reg->sortIdx); + if (reg->isams) + isams_close (reg->isams); + if (reg->isamc) + isamc_close (reg->isamc); + if (reg->isamb) + isamb_close (reg->isamb); + rec_close (®->records); } - recTypes_destroy (zs->recTypes); - zebra_maps_close (zs->zebra_maps); - zebraRankDestroy (zs); - bfs_destroy (zs->bfs); - data1_destroy (zs->dh); - if (zs->passwd_db) - passwd_db_close (zs->passwd_db); - zs->active = 1; - return 0; + recTypes_destroy (reg->recTypes); + zebra_maps_close (reg->zebra_maps); + zebraRankDestroy (reg); + bfs_destroy (reg->bfs); + data1_destroy (reg->dh); + + xfree(reg->sortKeys.buf); + xfree(reg->keys.buf); + if (reg->keys.codec_handle) + iscz1_stop(reg->keys.codec_handle); + xfree(reg->key_buf); + xfree(reg->name); + xfree(reg); } -void zebra_stop(ZebraService zs) +ZEBRA_RES zebra_stop(ZebraService zs) { if (!zs) - return ; - yaz_log (LOG_LOG, "zebra_stop"); + return ZEBRA_OK; + yaz_log (log_level, "zebra_stop"); - zebra_mutex_cond_lock (&zs->session_lock); while (zs->sessions) + { zebra_close (zs->sessions); - - zebra_mutex_cond_unlock (&zs->session_lock); - + } + zebra_mutex_cond_destroy (&zs->session_lock); - zebra_register_deactivate(zs); - res_close (zs->res); - xfree (zs->configName); - xfree (zs); + if (zs->passwd_db) + passwd_db_close (zs->passwd_db); + + recTypeClass_destroy(zs->record_classes); + nmem_destroy(zs->nmem); + res_close (zs->global_res); + xfree(zs->configName); + xfree(zs); + return ZEBRA_OK; } -void zebra_close (ZebraHandle zh) +ZEBRA_RES zebra_close (ZebraHandle zh) { - ZebraService zs = zh->service; + ZebraService zs; struct zebra_session **sp; + int i; - yaz_log (LOG_LOG, "zebra_close zh=%p", zh); + yaz_log(log_level, "zebra_close"); if (!zh) - return ; + return ZEBRA_OK; + ASSERTZH; + zh->errCode = 0; + + zs = zh->service; + yaz_log (YLOG_DEBUG, "zebra_close zh=%p", zh); resultSetDestroy (zh, -1, 0, 0); - if (zh->key_buf) - { - xfree (zh->key_buf); - zh->key_buf = 0; - } - xfree (zh->admin_databaseName); + if (zh->reg) + zebra_register_close (zh->service, zh->reg); + zebra_close_res (zh); + + xfree(zh->record_encoding); + + for (i = 0; i < zh->num_basenames; i++) + xfree(zh->basenames[i]); + xfree(zh->basenames); + + if (zh->iconv_to_utf8 != 0) + yaz_iconv_close (zh->iconv_to_utf8); + if (zh->iconv_from_utf8 != 0) + yaz_iconv_close (zh->iconv_from_utf8); + zebra_mutex_cond_lock (&zs->session_lock); + zebra_lock_destroy (zh->lock_normal); + zebra_lock_destroy (zh->lock_shadow); sp = &zs->sessions; while (1) { @@ -548,10 +557,18 @@ void zebra_close (ZebraHandle zh) } sp = &(*sp)->next; } - if (!zs->sessions && zs->stop_flag) - zebra_register_deactivate(zs); zebra_mutex_cond_unlock (&zs->session_lock); - xfree (zh); + xfree(zh->reg_name); + xfree(zh->user_perm); + zh->service = 0; /* more likely to trigger an assert */ + + zebra_limit_destroy(zh->m_limit); + + nmem_destroy(zh->nmem_error); + + xfree(zh->path_reg); + xfree(zh); + return ZEBRA_OK; } struct map_baseinfo { @@ -563,12 +580,130 @@ struct map_baseinfo { char **new_basenames; int new_num_max; }; - + +static Res zebra_open_res (ZebraHandle zh) +{ + Res res = 0; + char fname[512]; + ASSERTZH; + zh->errCode = 0; + + if (zh->path_reg) + { + sprintf (fname, "%.200s/zebra.cfg", zh->path_reg); + res = res_open (fname, zh->service->global_res, 0); + if (!res) + res = zh->service->global_res; + } + else if (*zh->reg_name == 0) + { + res = zh->service->global_res; + } + else + { + yaz_log (YLOG_WARN, "no register root specified"); + return 0; /* no path for register - fail! */ + } + return res; +} + +static void zebra_close_res (ZebraHandle zh) +{ + ASSERTZH; + zh->errCode = 0; + if (zh->res != zh->service->global_res) + res_close (zh->res); + zh->res = 0; +} + +static void zebra_select_register (ZebraHandle zh, const char *new_reg) +{ + ASSERTZH; + zh->errCode = 0; + if (zh->res && strcmp (zh->reg_name, new_reg) == 0) + return; + if (!zh->res) + { + assert (zh->reg == 0); + assert (*zh->reg_name == 0); + } + else + { + if (zh->reg) + { + resultSetInvalidate (zh); + zebra_register_close (zh->service, zh->reg); + zh->reg = 0; + } + zebra_close_res(zh); + } + xfree(zh->reg_name); + zh->reg_name = xstrdup (new_reg); + + xfree(zh->path_reg); + zh->path_reg = 0; + if (zh->service->path_root) + { + zh->path_reg = xmalloc(strlen(zh->service->path_root) + + strlen(zh->reg_name) + 3); + strcpy (zh->path_reg, zh->service->path_root); + if (*zh->reg_name) + { + strcat (zh->path_reg, "/"); + strcat (zh->path_reg, zh->reg_name); + } + } + zh->res = zebra_open_res (zh); + + if (zh->lock_normal) + zebra_lock_destroy (zh->lock_normal); + zh->lock_normal = 0; + + if (zh->lock_shadow) + zebra_lock_destroy (zh->lock_shadow); + zh->lock_shadow = 0; + + if (zh->res) + { + char fname[512]; + const char *lock_area =res_get (zh->res, "lockDir"); + + if (!lock_area && zh->path_reg) + res_set (zh->res, "lockDir", zh->path_reg); + sprintf (fname, "norm.%s.LCK", zh->reg_name); + zh->lock_normal = + zebra_lock_create (res_get(zh->res, "lockDir"), fname, 0); + + sprintf (fname, "shadow.%s.LCK", zh->reg_name); + zh->lock_shadow = + zebra_lock_create (res_get(zh->res, "lockDir"), fname, 0); + + if (!zh->lock_normal || !zh->lock_shadow) + { + if (zh->lock_normal) + { + zebra_lock_destroy(zh->lock_normal); + zh->lock_normal = 0; + } + if (zh->lock_shadow) + { + zebra_lock_destroy(zh->lock_shadow); + zh->lock_shadow = 0; + } + zebra_close_res(zh); + } + } +} + void map_basenames_func (void *vp, const char *name, const char *value) { struct map_baseinfo *p = (struct map_baseinfo *) vp; int i, no; char fromdb[128], todb[8][128]; + + assert(value); + assert(name); + assert(vp); no = sscanf (value, "%127s %127s %127s %127s %127s %127s %127s %127s %127s", @@ -578,7 +713,7 @@ void map_basenames_func (void *vp, const char *name, const char *value) return ; no--; for (i = 0; inum_bases; i++) - if (p->basenames[i] && !strcmp (p->basenames[i], fromdb)) + if (p->basenames[i] && !STRCASECMP (p->basenames[i], fromdb)) { p->basenames[i] = 0; for (i = 0; i < no; i++) @@ -592,14 +727,33 @@ void map_basenames_func (void *vp, const char *name, const char *value) } } +int zebra_select_default_database(ZebraHandle zh) +{ + if (!zh->res) + { + /* no database has been selected - so we select based on + resource setting (including group) + */ + const char *group = res_get(zh->service->global_res, "group"); + const char *v = res_get_prefix(zh->service->global_res, + "database", group, "Default"); + return zebra_select_database(zh, v); + } + return 0; +} + void map_basenames (ZebraHandle zh, ODR stream, int *num_bases, char ***basenames) { struct map_baseinfo info; struct map_baseinfo *p = &info; int i; + ASSERTZH; + yaz_log(log_level, "map_basenames "); + assert(stream); info.zh = zh; + info.num_bases = *num_bases; info.basenames = *basenames; info.new_num_max = 128; @@ -608,7 +762,7 @@ void map_basenames (ZebraHandle zh, ODR stream, odr_malloc (stream, sizeof(*info.new_basenames) * info.new_num_max); info.mem = stream->mem; - res_trav (zh->service->res, "mapdb", &info, map_basenames_func); + res_trav (zh->service->global_res, "mapdb", &info, map_basenames_func); for (i = 0; inum_bases; i++) if (p->basenames[i] && p->new_num_bases < p->new_num_max) @@ -619,122 +773,322 @@ void map_basenames (ZebraHandle zh, ODR stream, *num_bases = info.new_num_bases; *basenames = info.new_basenames; for (i = 0; i<*num_bases; i++) - logf (LOG_LOG, "base %s", (*basenames)[i]); + yaz_log (YLOG_DEBUG, "base %s", (*basenames)[i]); } -void zebra_search_rpn (ZebraHandle zh, ODR stream, ODR decode, - Z_RPNQuery *query, int num_bases, char **basenames, - const char *setname) +ZEBRA_RES zebra_select_database (ZebraHandle zh, const char *basename) { - zh->hits = 0; - if (zebra_register_lock (zh, 0)) - return; - map_basenames (zh, stream, &num_bases, &basenames); - resultSetAddRPN (zh, stream, decode, query, num_bases, basenames, setname); - - zebra_register_unlock (zh); - - logf(LOG_APP,"SEARCH:%d:",zh->hits); + ASSERTZH; + yaz_log(log_level, "zebra_select_database %s",basename); + assert(basename); + return zebra_select_databases (zh, 1, &basename); } -void zebra_records_retrieve (ZebraHandle zh, ODR stream, - const char *setname, Z_RecordComposition *comp, - oid_value input_format, int num_recs, - ZebraRetrievalRecord *recs) +ZEBRA_RES zebra_select_databases (ZebraHandle zh, int num_bases, + const char **basenames) { - ZebraPosSet poset; - int i, *pos_array; + int i; + const char *cp; + int len = 0; + char *new_reg = 0; + ASSERTZH; + assert(basenames); + + yaz_log(log_level, "zebra_select_databases n=%d [0]=%s", + num_bases,basenames[0]); + zh->errCode = 0; + + if (num_bases < 1) + { + zh->errCode = YAZ_BIB1_COMBI_OF_SPECIFIED_DATABASES_UNSUPP; + return ZEBRA_FAIL; + } + for (i = 0; i < zh->num_basenames; i++) + xfree(zh->basenames[i]); + xfree(zh->basenames); + + zh->num_basenames = num_bases; + zh->basenames = xmalloc(zh->num_basenames * sizeof(*zh->basenames)); + for (i = 0; i < zh->num_basenames; i++) + zh->basenames[i] = xstrdup (basenames[i]); - if (zebra_register_lock (zh, 0)) - return; - pos_array = (int *) xmalloc (num_recs * sizeof(*pos_array)); - for (i = 0; ierrCode = 30; - zh->errString = nmem_strdup (stream->mem, setname); + len = cp - basenames[0]; + new_reg = xmalloc(len + 1); + memcpy (new_reg, basenames[0], len); + new_reg[len] = '\0'; } else + new_reg = xstrdup (""); + for (i = 1; ierrCode = 13; - zh->errString = nmem_strdup (stream->mem, num_str); - break; - } - else - { - recs[i].errCode = - zebra_record_fetch (zh, poset[i].sysno, poset[i].score, - stream, input_format, comp, - &recs[i].format, &recs[i].buf, - &recs[i].len, - &recs[i].base); - recs[i].errString = NULL; - } - } - zebraPosSetDestroy (zh, poset, num_recs); + cp1 = strrchr (basenames[i], '/'); + if (cp) + { + if (!cp1) + { + zh->errCode = YAZ_BIB1_COMBI_OF_SPECIFIED_DATABASES_UNSUPP; + return -1; + } + if (len != cp1 - basenames[i] || + memcmp (basenames[i], new_reg, len)) + { + zh->errCode = YAZ_BIB1_COMBI_OF_SPECIFIED_DATABASES_UNSUPP; + return -1; + } + } + else + { + if (cp1) + { + zh->errCode = YAZ_BIB1_COMBI_OF_SPECIFIED_DATABASES_UNSUPP; + return ZEBRA_FAIL; + } + } } - zebra_register_unlock (zh); - xfree (pos_array); + zebra_select_register (zh, new_reg); + xfree(new_reg); + if (!zh->res) + { + zh->errCode = YAZ_BIB1_DATABASE_UNAVAILABLE; + return ZEBRA_FAIL; + } + if (!zh->lock_normal || !zh->lock_shadow) + { + zh->errCode = YAZ_BIB1_TEMPORARY_SYSTEM_ERROR; + return ZEBRA_FAIL; + } + return ZEBRA_OK; +} + +ZEBRA_RES zebra_set_approx_limit(ZebraHandle zh, zint approx_limit) +{ + zh->approx_limit = approx_limit; + return ZEBRA_OK; +} + +ZEBRA_RES zebra_search_RPN(ZebraHandle zh, ODR o, Z_RPNQuery *query, + const char *setname, zint *hits) +{ + ZEBRA_RES r; + ASSERTZH; + assert(o); + assert(query); + assert(hits); + assert(setname); + yaz_log(log_level, "zebra_search_rpn"); + zebra_clearError(zh); + zh->hits = 0; + *hits = 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); + zebra_end_read(zh); + *hits = zh->hits; + return r; +} + +ZEBRA_RES zebra_records_retrieve(ZebraHandle zh, ODR stream, + const char *setname, + Z_RecordComposition *comp, + oid_value input_format, int num_recs, + ZebraRetrievalRecord *recs) +{ + ZebraMetaRecord *poset; + int i; + ZEBRA_RES ret = ZEBRA_OK; + zint *pos_array; + ASSERTZH; + assert(stream); + assert(setname); + assert(recs); + assert(num_recs>0); + + yaz_log(log_level, "zebra_records_retrieve n=%d", num_recs); + + zebra_clearError(zh); + if (!zh->res) + { + zebra_setError(zh, YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST, + setname); + return ZEBRA_FAIL; + } + + if (zebra_begin_read (zh) == ZEBRA_FAIL) + return ZEBRA_FAIL; + + pos_array = (zint *) xmalloc(num_recs * sizeof(*pos_array)); + for (i = 0; i 0) + { + recs[i].buf = (char*) odr_malloc(stream, len); + memcpy(recs[i].buf, buf, len); + } + else + recs[i].buf = buf; + recs[i].score = poset[i].score; + recs[i].sysno = poset[i].sysno; + zebra_snippets_destroy(hit_snippet); + } + else + { + if (ret == ZEBRA_OK) /* only need to set it once */ + zebra_setError_zint(zh, + YAZ_BIB1_PRESENT_REQUEST_OUT_OF_RANGE, + pos_array[i]); + ret = ZEBRA_FAIL; + break; + } + } + zebra_meta_records_destroy(zh, poset, num_recs); + } + zebra_end_read (zh); + xfree(pos_array); + return ret; } -void zebra_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, - oid_value attributeset, - int num_bases, char **basenames, - int *position, int *num_entries, ZebraScanEntry **entries, - int *is_partial) +ZEBRA_RES zebra_scan_PQF(ZebraHandle zh, ODR stream, const char *query, + int *position, + int *num_entries, ZebraScanEntry **entries, + int *is_partial) { - if (zebra_register_lock (zh, 0)) + YAZ_PQF_Parser pqf_parser = yaz_pqf_create (); + Z_AttributesPlusTerm *zapt; + int *attributeSet; + ZEBRA_RES res; + + if (!(zapt = yaz_pqf_scan(pqf_parser, stream, &attributeSet, query))) + { + res = ZEBRA_FAIL; + zh->errCode = YAZ_BIB1_SCAN_MALFORMED_SCAN; + } + else + res = zebra_scan(zh, stream, zapt, VAL_BIB1, + position, num_entries, entries, is_partial); + yaz_pqf_destroy (pqf_parser); + return res; +} + +ZEBRA_RES zebra_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, + oid_value attributeset, + int *position, + int *num_entries, ZebraScanEntry **entries, + int *is_partial) +{ + ZEBRA_RES res; + ASSERTZH; + assert(stream); + assert(zapt); + assert(position); + assert(num_entries); + assert(is_partial); + assert(entries); + yaz_log(log_level, "zebra_scan"); + zebra_clearError(zh); + if (zebra_begin_read (zh) == ZEBRA_FAIL) { *entries = 0; *num_entries = 0; - return; + return ZEBRA_FAIL; } - map_basenames (zh, stream, &num_bases, &basenames); - rpn_scan (zh, stream, zapt, attributeset, - num_bases, basenames, position, - num_entries, entries, is_partial); - zebra_register_unlock (zh); + res = rpn_scan (zh, stream, zapt, attributeset, + zh->num_basenames, zh->basenames, position, + num_entries, entries, is_partial, 0, 0); + zebra_end_read (zh); + return res; } -void zebra_sort (ZebraHandle zh, ODR stream, - int num_input_setnames, const char **input_setnames, - const char *output_setname, Z_SortKeySpecList *sort_sequence, - int *sort_status) +ZEBRA_RES zebra_sort (ZebraHandle zh, ODR stream, + int num_input_setnames, const char **input_setnames, + const char *output_setname, + Z_SortKeySpecList *sort_sequence, + int *sort_status) { - if (zebra_register_lock (zh, 0)) - return; - resultSetSort (zh, stream->mem, num_input_setnames, input_setnames, - output_setname, sort_sequence, sort_status); - zebra_register_unlock (zh); + ZEBRA_RES res; + ASSERTZH; + assert(stream); + assert(num_input_setnames>0); + assert(input_setnames); + assert(sort_sequence); + assert(sort_status); + yaz_log(log_level, "zebra_sort"); + zebra_clearError(zh); + if (zebra_begin_read(zh) == ZEBRA_FAIL) + return ZEBRA_FAIL; + res = resultSetSort(zh, stream->mem, num_input_setnames, input_setnames, + output_setname, sort_sequence, sort_status); + zebra_end_read(zh); + return res; } -int zebra_deleleResultSet(ZebraHandle zh, int function, +int zebra_deleteResultSet(ZebraHandle zh, int function, int num_setnames, char **setnames, int *statuses) { int i, status; - if (zebra_register_lock (zh, 0)) + ASSERTZH; + assert(statuses); + yaz_log(log_level, "zebra_deleteResultSet n=%d",num_setnames); + zebra_clearError(zh);; + if (zebra_begin_read(zh)) return Z_DeleteStatus_systemProblemAtTarget; switch (function) { - case Z_DeleteRequest_list: + case Z_DeleteResultSetRequest_list: + assert(num_setnames>0); + assert(setnames); resultSetDestroy (zh, num_setnames, setnames, statuses); break; - case Z_DeleteRequest_all: + case Z_DeleteResultSetRequest_all: resultSetDestroy (zh, -1, 0, statuses); break; } - zebra_register_unlock (zh); + zebra_end_read (zh); status = Z_DeleteStatus_success; for (i = 0; ierrCode; + if (zh) + { + yaz_log(log_level, "zebra_errCode: %d",zh->errCode); + return zh->errCode; + } + yaz_log(log_level, "zebra_errCode: o"); + return 0; } const char *zebra_errString (ZebraHandle zh) { - return diagbib1_str (zh->errCode); + const char *e = 0; + if (zh) + e= diagbib1_str (zh->errCode); + yaz_log(log_level, "zebra_errString: %s",e); + return e; } char *zebra_errAdd (ZebraHandle zh) { - return zh->errString; + char *a = 0; + if (zh) + a= zh->errString; + yaz_log(log_level, "zebra_errAdd: %s",a); + return a; } -int zebra_hits (ZebraHandle zh) +void zebra_clearError(ZebraHandle zh) { - return zh->hits; + if (zh) + { + zh->errCode = 0; + zh->errString = 0; + } } -int zebra_auth (ZebraService zh, const char *user, const char *pass) +ZEBRA_RES zebra_auth (ZebraHandle zh, const char *user, const char *pass) { - if (!zh->passwd_db || !passwd_db_auth (zh->passwd_db, user, pass)) - { - logf(LOG_APP,"AUTHOK:%s", user?user:"ANONYMOUS"); - return 0; - } + const char *p; + char u[40]; + ZebraService zs; - logf(LOG_APP,"AUTHFAIL:%s", user?user:"ANONYMOUS"); - return 1; + ASSERTZH; + zebra_clearError(zh); + zs= zh->service; + + sprintf(u, "perm.%.30s", user ? user : "anonymous"); + p = res_get(zs->global_res, u); + xfree(zh->user_perm); + zh->user_perm = xstrdup(p ? p : "r"); + + /* users that don't require a password .. */ + if (zh->user_perm && strchr(zh->user_perm, 'a')) + return ZEBRA_OK; + + if (!zs->passwd_db || !passwd_db_auth (zs->passwd_db, user, pass)) + return ZEBRA_OK; + return ZEBRA_FAIL; } -void zebra_admin_import_begin (ZebraHandle zh, const char *database) +ZEBRA_RES zebra_admin_import_begin (ZebraHandle zh, const char *database, + const char *record_type) { - if (zebra_register_lock (zh, 1)) - return; - xfree (zh->admin_databaseName); - zh->admin_databaseName = xstrdup(database); + ASSERTZH; + yaz_log(log_level, "zebra_admin_import_begin db=%s rt=%s", + database, record_type); + zebra_clearError(zh); + if (zebra_select_database(zh, database) == ZEBRA_FAIL) + return ZEBRA_FAIL; + return zebra_begin_trans(zh, 1); } -void zebra_admin_import_end (ZebraHandle zh) +ZEBRA_RES zebra_admin_import_end (ZebraHandle zh) { - zebraExplain_flush (zh->service->zei, 1, zh); - extract_index (zh); - zebra_register_unlock (zh); + ASSERTZH; + yaz_log(log_level, "zebra_admin_import_end"); + zebra_clearError(zh); + return zebra_end_trans(zh); } -void zebra_admin_import_segment (ZebraHandle zh, Z_Segment *segment) +ZEBRA_RES zebra_admin_import_segment (ZebraHandle zh, Z_Segment *segment) { - int sysno; + ZEBRA_RES res = ZEBRA_OK; + SYSNO sysno; int i; - if (zh->service->active < 2) - return; + ASSERTZH; + yaz_log(log_level, "zebra_admin_import_segment"); + zebra_clearError(zh); for (i = 0; inum_segmentRecords; i++) { Z_NamePlusRecord *npr = segment->segmentRecords[i]; - const char *databaseName = npr->databaseName; - if (!databaseName) - databaseName = zh->admin_databaseName; - printf ("--------------%d--------------------\n", i); if (npr->which == Z_NamePlusRecord_intermediateFragment) { Z_FragmentSyntax *fragment = npr->u.intermediateFragment; if (fragment->which == Z_FragmentSyntax_notExternallyTagged) { Odr_oct *oct = fragment->u.notExternallyTagged; - printf ("%.*s", (oct->len > 100 ? 100 : oct->len) , - oct->buf); - sysno = 0; - extract_rec_in_mem (zh, "grs.sgml", - oct->buf, oct->len, - databaseName, - 0 /* delete_flag */, - 0 /* test_mode */, - &sysno /* sysno */, - 1 /* store_keys */, - 1 /* store_data */, - 0 /* match criteria */); + + if ( zebra_update_record(zh, + 0, /* record Type */ + &sysno, + 0, /* match */ + 0, /* fname */ + oct->buf, oct->len, + 0) == ZEBRA_FAIL) + res = ZEBRA_FAIL; } } } + return res; } -void zebra_admin_create (ZebraHandle zh, const char *database) +ZEBRA_RES zebra_admin_exchange_record (ZebraHandle zh, + const char *rec_buf, + size_t rec_len, + const char *recid_buf, size_t recid_len, + int action) + /* 1 = insert. Fail it already exists */ + /* 2 = replace. Fail it does not exist */ + /* 3 = delete. Fail if does not exist */ + /* 4 = update. Insert/replace */ { - ZebraService zs = zh->service; - if (zebra_register_lock(zh, 1)) + ZEBRA_RES res; + SYSNO sysno = 0; + char *rinfo = 0; + char recid_z[256]; + ASSERTZH; + assert(action>0 && action <=4); + assert(rec_buf); + + yaz_log(log_level, "zebra_admin_exchange_record ac=%d", action); + zebra_clearError(zh); + + if (!recid_buf || recid_len <= 0 || recid_len >= sizeof(recid_z)) + return ZEBRA_FAIL; + + memcpy (recid_z, recid_buf, recid_len); + recid_z[recid_len] = 0; + + if (zebra_begin_trans(zh, 1) == ZEBRA_FAIL) + return ZEBRA_FAIL; + + rinfo = dict_lookup (zh->reg->matchDict, recid_z); + if (rinfo) { - zh->errCode = 1019; - return; + if (action == 1) /* fail if insert */ + { + zebra_end_trans(zh); + return ZEBRA_FAIL; + } + + memcpy (&sysno, rinfo+1, sizeof(sysno)); + } + else + { + if (action == 2 || action == 3) /* fail if delete or update */ + { + zebra_end_trans(zh); + return ZEBRA_FAIL; + } + action = 1; /* make it an insert (if it's an update).. */ + } + res = buffer_extract_record (zh, rec_buf, rec_len, + action == 3 ? 1 : 0 /* delete flag */, + 0, /* test mode */ + 0, /* recordType */ + &sysno, + 0, /* match */ + 0, /* fname */ + 0, /* force update */ + 1 /* allow update */ + ); + if (action == 1) + { + dict_insert (zh->reg->matchDict, recid_z, sizeof(sysno), &sysno); + } + else if (action == 3) + { + dict_delete (zh->reg->matchDict, recid_z); + } + zebra_end_trans(zh); + return res; +} + +int delete_w_handle(const char *info, void *handle) +{ + ZebraHandle zh = (ZebraHandle) handle; + ISAM_P pos; + ASSERTZH; + + if (*info == sizeof(pos)) + { + memcpy (&pos, info+1, sizeof(pos)); + isamb_unlink(zh->reg->isamb, pos); } + return 0; +} + +static int delete_SU_handle(void *handle, int ord) +{ + ZebraHandle zh = (ZebraHandle) handle; + char ord_buf[20]; + int ord_len; + + ord_len = key_SU_encode (ord, ord_buf); + ord_buf[ord_len] = '\0'; + + assert (zh->reg->isamb); + dict_delete_subtree(zh->reg->dict, ord_buf, + zh, delete_w_handle); + return 0; +} + +ZEBRA_RES zebra_drop_database(ZebraHandle zh, const char *db) +{ + ZEBRA_RES ret = ZEBRA_OK; + ASSERTZH; + yaz_log(log_level, "zebra_drop_database %s", db); + zebra_clearError(zh); + + if (zebra_select_database (zh, db) == ZEBRA_FAIL) + return ZEBRA_FAIL; + if (zebra_begin_trans (zh, 1) == ZEBRA_FAIL) + return ZEBRA_FAIL; + if (zh->reg->isamb) + { + zebraExplain_curDatabase (zh->reg->zei, db); + + zebraExplain_trav_ord(zh->reg->zei, zh, delete_SU_handle); + zebraExplain_removeDatabase(zh->reg->zei, zh); + } + else + { + yaz_log(YLOG_WARN, "drop database only supported for isam:b"); + ret = ZEBRA_FAIL; + } + zebra_end_trans (zh); + return ret; +} + +ZEBRA_RES zebra_create_database (ZebraHandle zh, const char *db) +{ + ASSERTZH; + yaz_log(log_level, "zebra_create_database %s", db); + assert(db); + zebra_clearError(zh); + + if (zebra_select_database (zh, db) == ZEBRA_FAIL) + return ZEBRA_FAIL; + if (zebra_begin_trans (zh, 1)) + return ZEBRA_FAIL; + /* announce database */ - if (zebraExplain_newDatabase (zs->zei, database, 0 /* explainDatabase */)) + if (zebraExplain_newDatabase (zh->reg->zei, db, 0 + /* explainDatabase */)) { - zh->errCode = 224; - zh->errString = "Database already exist"; + zebra_end_trans (zh); + zebra_setError(zh, YAZ_BIB1_ES_IMMEDIATE_EXECUTION_FAILED, db); + return ZEBRA_FAIL; } - zebraExplain_flush (zh->service->zei, 1, zh); - extract_index (zh); - zebra_register_unlock(zh); + return zebra_end_trans (zh); } int zebra_string_norm (ZebraHandle zh, unsigned reg_id, @@ -851,9 +1368,14 @@ int zebra_string_norm (ZebraHandle zh, unsigned reg_id, char *output_str, int output_len) { WRBUF wrbuf; - if (!zh->service->zebra_maps) + ASSERTZH; + assert(input_str); + assert(output_str); + yaz_log(log_level, "zebra_string_norm "); + zebra_clearError(zh); + if (!zh->reg->zebra_maps) return -1; - wrbuf = zebra_replace(zh->service->zebra_maps, reg_id, "", + wrbuf = zebra_replace(zh->reg->zebra_maps, reg_id, "", input_str, input_len); if (!wrbuf) return -2; @@ -865,837 +1387,882 @@ int zebra_string_norm (ZebraHandle zh, unsigned reg_id, return wrbuf_len(wrbuf); } -static void extract_init (struct recExtractCtrl *p, RecWord *w) +static void zebra_set_state (ZebraHandle zh, int val, int seqno) { - w->zebra_maps = p->zebra_maps; - w->seqnos = p->seqno; - w->attrSet = VAL_BIB1; - w->attrUse = 1016; - w->reg_type = 'w'; - w->extractCtrl = p; + char state_fname[256]; + char *fname; + long p = getpid(); + FILE *f; + ASSERTZH; + yaz_log(log_level, "zebra_set_state v=%d seq=%d", val, seqno); + zebra_clearError(zh); + + sprintf (state_fname, "state.%s.LCK", zh->reg_name); + fname = zebra_mk_fname (res_get(zh->res, "lockDir"), state_fname); + f = fopen (fname, "w"); + + yaz_log (YLOG_DEBUG, "zebra_set_state: %c %d %ld", val, seqno, p); + fprintf (f, "%c %d %ld\n", val, seqno, p); + fclose (f); + xfree(fname); } -static void extract_add_index_string (RecWord *p, const char *string, - int length) +static void zebra_get_state (ZebraHandle zh, char *val, int *seqno) { - char *dst; - unsigned char attrSet; - unsigned short attrUse; - int lead = 0; - int diff = 0; - int *pseqno = &p->seqnos[p->reg_type]; - ZebraHandle zh = p->extractCtrl->handle; - ZebraExplainInfo zei = zh->service->zei; - struct recKeys *keys = &zh->keys; - - if (keys->buf_used+1024 > keys->buf_max) + char state_fname[256]; + char *fname; + FILE *f; + + ASSERTZH; + yaz_log(log_level, "zebra_get_state "); + zebra_clearError(zh); + sprintf (state_fname, "state.%s.LCK", zh->reg_name); + fname = zebra_mk_fname (res_get(zh->res, "lockDir"), state_fname); + f = fopen (fname, "r"); + *val = 'o'; + *seqno = 0; + + if (f) { - char *b; - - b = (char *) xmalloc (keys->buf_max += 128000); - if (keys->buf_used > 0) - memcpy (b, keys->buf, keys->buf_used); - xfree (keys->buf); - keys->buf = b; + fscanf (f, "%c %d", val, seqno); + fclose (f); } - dst = keys->buf + keys->buf_used; - - attrSet = p->attrSet; - if (keys->buf_used > 0 && keys->prevAttrSet == attrSet) - lead |= 1; - else - keys->prevAttrSet = attrSet; - attrUse = p->attrUse; - if (keys->buf_used > 0 && keys->prevAttrUse == attrUse) - lead |= 2; - else - keys->prevAttrUse = attrUse; -#if 1 - diff = 1 + *pseqno - keys->prevSeqNo; - if (diff >= 1 && diff <= 15) - lead |= (diff << 2); - else - diff = 0; -#endif - keys->prevSeqNo = *pseqno; - - *dst++ = lead; + xfree(fname); +} -#if SU_SCHEME - if ((lead & 3) < 3) - { - int ch = zebraExplain_lookupSU (zei, attrSet, attrUse); - if (ch < 0) - { - ch = zebraExplain_addSU (zei, attrSet, attrUse); - yaz_log (LOG_LOG, "addSU set=%d use=%d SU=%d", - attrSet, attrUse, ch); - } - assert (ch > 0); - memcpy (dst, &ch, sizeof(ch)); - dst += sizeof(ch); - } -#else - if (!(lead & 1)) - { - memcpy (dst, &attrSet, sizeof(attrSet)); - dst += sizeof(attrSet); - } - if (!(lead & 2)) - { - memcpy (dst, &attrUse, sizeof(attrUse)); - dst += sizeof(attrUse); - } -#endif - *dst++ = p->reg_type; - memcpy (dst, string, length); - dst += length; - *dst++ = '\0'; +ZEBRA_RES zebra_begin_read (ZebraHandle zh) +{ + return zebra_begin_trans(zh, 0); +} - if (!diff) - { - memcpy (dst, pseqno, sizeof(*pseqno)); - dst += sizeof(*pseqno); - } - keys->buf_used = dst - keys->buf; - if (*pseqno) - (*pseqno)++; +ZEBRA_RES zebra_end_read (ZebraHandle zh) +{ + return zebra_end_trans(zh); } -static void extract_add_sort_string (RecWord *p, const char *string, - int length) +static void read_res_for_transaction(ZebraHandle zh) { - struct sortKey *sk; - ZebraHandle zh = p->extractCtrl->handle; - struct sortKey *sortKeys = zh->sortKeys; + const char *group = res_get(zh->res, "group"); + const char *v; + /* FIXME - do we still use groups ?? */ + + zh->m_group = group; + v = res_get_prefix(zh->res, "followLinks", group, "1"); + zh->m_follow_links = atoi(v); - for (sk = sortKeys; sk; sk = sk->next) - if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse) - return; + zh->m_record_id = res_get_prefix(zh->res, "recordId", group, 0); + zh->m_record_type = res_get_prefix(zh->res, "recordType", group, 0); - sk = (struct sortKey *) xmalloc (sizeof(*sk)); - sk->next = sortKeys; - sortKeys = sk; + v = res_get_prefix(zh->res, "storeKeys", group, "1"); + zh->m_store_keys = atoi(v); - sk->string = (char *) xmalloc (length); - sk->length = length; - memcpy (sk->string, string, length); + v = res_get_prefix(zh->res, "storeData", group, "1"); + zh->m_store_data = atoi(v); - sk->attrSet = p->attrSet; - sk->attrUse = p->attrUse; -} + v = res_get_prefix(zh->res, "explainDatabase", group, "0"); + zh->m_explain_database = atoi(v); -static void extract_add_string (RecWord *p, const char *string, int length) -{ - assert (length > 0); - if (zebra_maps_is_sort (p->zebra_maps, p->reg_type)) - extract_add_sort_string (p, string, length); - else - extract_add_index_string (p, string, length); + v = res_get_prefix(zh->res, "openRW", group, "1"); + zh->m_flag_rw = atoi(v); + + v = res_get_prefix(zh->res, "fileVerboseLimit", group, "100000"); + zh->m_file_verbose_limit = atoi(v); } -static void extract_add_incomplete_field (RecWord *p) +ZEBRA_RES zebra_begin_trans(ZebraHandle zh, int rw) { - const char *b = p->string; - int remain = p->length; - const char **map = 0; - - if (remain > 0) - map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain); - - while (map) + ASSERTZH; + zebra_select_default_database(zh); + if (!zh->res) { - char buf[IT_MAX_WORD+1]; - int i, remain; + zebra_setError(zh, YAZ_BIB1_TEMPORARY_SYSTEM_ERROR, + "zebra_begin_trans: no database selected"); + return ZEBRA_FAIL; + } + ASSERTZHRES; + yaz_log(log_level, "zebra_begin_trans rw=%d",rw); - /* Skip spaces */ - while (map && *map && **map == *CHR_SPACE) - { - remain = p->length - (b - p->string); - if (remain > 0) - map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain); - else - map = 0; - } - if (!map) - break; - i = 0; - while (map && *map && **map != *CHR_SPACE) + if (zh->user_perm) + { + if (rw && !strchr(zh->user_perm, 'w')) { - const char *cp = *map; - - while (i < IT_MAX_WORD && *cp) - buf[i++] = *(cp++); - remain = p->length - (b - p->string); - if (remain > 0) - map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain); - else - map = 0; + zebra_setError( + zh, + YAZ_BIB1_ES_PERMISSION_DENIED_ON_ES_CANNOT_MODIFY_OR_DELETE, + 0); + return ZEBRA_FAIL; } - if (!i) - return; - extract_add_string (p, buf, i); } - (p->seqnos[p->reg_type])++; /* to separate this from next one */ -} - -static void extract_add_complete_field (RecWord *p) -{ - const char *b = p->string; - char buf[IT_MAX_WORD+1]; - const char **map = 0; - int i = 0, remain = p->length; - - if (remain > 0) - map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain); - while (remain > 0 && i < IT_MAX_WORD) + assert (zh->res); + if (rw) { - while (map && *map && **map == *CHR_SPACE) + int pass; + int seqno = 0; + char val = '?'; + const char *rval = 0; + + (zh->trans_no++); + if (zh->trans_w_no) { - remain = p->length - (b - p->string); - if (remain > 0) - map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain); - else - map = 0; + read_res_for_transaction(zh); + return 0; } - if (!map) - break; - - if (i && i < IT_MAX_WORD) - buf[i++] = *CHR_SPACE; - while (map && *map && **map != *CHR_SPACE) + if (zh->trans_no != 1) + { + zebra_setError(zh, YAZ_BIB1_TEMPORARY_SYSTEM_ERROR, + "zebra_begin_trans: no write trans within read"); + return ZEBRA_FAIL; + } + if (zh->reg) { - const char *cp = *map; - - if (i >= IT_MAX_WORD) - break; - while (i < IT_MAX_WORD && *cp) - buf[i++] = *(cp++); - remain = p->length - (b - p->string); - if (remain > 0) - map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, - remain); - else - map = 0; + resultSetInvalidate (zh); + zebra_register_close (zh->service, zh->reg); } + zh->trans_w_no = zh->trans_no; + + zebra_clearError(zh); + + zh->records_inserted = 0; + zh->records_updated = 0; + zh->records_deleted = 0; + zh->records_processed = 0; + +#if HAVE_SYS_TIMES_H + times (&zh->tms1); +#endif + /* lock */ + if (zh->shadow_enable) + rval = res_get (zh->res, "shadow"); + + for (pass = 0; pass < 2; pass++) + { + 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 == 'c') + { + yaz_log (YLOG_WARN, "previous transaction didn't finish commit"); + zebra_unlock (zh->lock_shadow); + zebra_unlock (zh->lock_normal); + zebra_commit (zh); + continue; + } + else if (val == 'd') + { + 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"); + } + } + 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, + 1, rval ? 1 : 0, zh->res, + zh->path_reg); + if (zh->reg) + zh->reg->seqno = seqno; + else + { + zebra_set_state (zh, 'o', seqno); + + zebra_unlock (zh->lock_shadow); + zebra_unlock (zh->lock_normal); + + zh->trans_no--; + zh->trans_w_no = 0; + + zebra_setError(zh, YAZ_BIB1_TEMPORARY_SYSTEM_ERROR, + "zebra_begin_trans: cannot open register"); + yaz_log(YLOG_FATAL, "%s", zh->errString); + return ZEBRA_FAIL; + } } - if (!i) - return; - extract_add_string (p, buf, i); + else + { + int dirty = 0; + char val; + int seqno; + + (zh->trans_no)++; + + if (zh->trans_no != 1) + { + zebra_flush_reg (zh); + return ZEBRA_OK; + } + zebra_clearError(zh); +#if HAVE_SYS_TIMES_H + times (&zh->tms1); +#endif + if (!zh->res) + { + (zh->trans_no)--; + zh->errCode = YAZ_BIB1_DATABASE_UNAVAILABLE; + return ZEBRA_FAIL; + } + if (!zh->lock_normal || !zh->lock_shadow) + { + (zh->trans_no)--; + zh->errCode = YAZ_BIB1_TEMPORARY_SYSTEM_ERROR; + return ZEBRA_FAIL; + } + zebra_get_state (zh, &val, &seqno); + if (val == 'd') + val = 'o'; + + if (!zh->reg) + dirty = 1; + else if (seqno != zh->reg->seqno) + { + yaz_log (YLOG_DEBUG, "reopen seqno cur/old %d/%d", + seqno, zh->reg->seqno); + dirty = 1; + } + else if (zh->reg->last_val != val) + { + yaz_log (YLOG_DEBUG, "reopen last cur/old %d/%d", + val, zh->reg->last_val); + dirty = 1; + } + if (!dirty) + return ZEBRA_OK; + + if (val == 'c') + zebra_lock_r (zh->lock_shadow); + else + zebra_lock_r (zh->lock_normal); + + if (zh->reg) + { + resultSetInvalidate (zh); + zebra_register_close (zh->service, zh->reg); + } + zh->reg = zebra_register_open (zh->service, zh->reg_name, + 0, val == 'c' ? 1 : 0, + zh->res, zh->path_reg); + if (!zh->reg) + { + zebra_unlock (zh->lock_normal); + zebra_unlock (zh->lock_shadow); + zh->trans_no--; + zh->errCode = YAZ_BIB1_DATABASE_UNAVAILABLE; + return ZEBRA_FAIL; + } + zh->reg->last_val = val; + zh->reg->seqno = seqno; + } + read_res_for_transaction(zh); + return ZEBRA_OK; } -static void extract_token_add (RecWord *p) +ZEBRA_RES zebra_end_trans (ZebraHandle zh) { - WRBUF wrbuf; - if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0, - p->string, p->length))) + ZebraTransactionStatus dummy; + ASSERTZH; + yaz_log(log_level, "zebra_end_trans"); + return zebra_end_transaction(zh, &dummy); +} + +ZEBRA_RES zebra_end_transaction (ZebraHandle zh, ZebraTransactionStatus *status) +{ + char val; + int seqno; + const char *rval; + + ASSERTZH; + assert(status); + yaz_log(log_level, "zebra_end_transaction"); + + status->processed = 0; + status->inserted = 0; + status->updated = 0; + status->deleted = 0; + status->utime = 0; + status->stime = 0; + + if (!zh->res || !zh->reg) { - p->string = wrbuf_buf(wrbuf); - p->length = wrbuf_len(wrbuf); + zebra_setError(zh, YAZ_BIB1_TEMPORARY_SYSTEM_ERROR, + "zebra_end_trans: no open transaction"); + return ZEBRA_FAIL; + } + if (zh->trans_no != zh->trans_w_no) + { + zh->trans_no--; + if (zh->trans_no != 0) + return ZEBRA_OK; + + /* release read lock */ + + zebra_unlock (zh->lock_normal); + zebra_unlock (zh->lock_shadow); } - if (zebra_maps_is_complete (p->zebra_maps, p->reg_type)) - extract_add_complete_field (p); else - extract_add_incomplete_field(p); + { /* release write lock */ + zh->trans_no--; + zh->trans_w_no = 0; + + yaz_log (YLOG_DEBUG, "zebra_end_trans"); + rval = res_get (zh->res, "shadow"); + + zebraExplain_runNumberIncrement (zh->reg->zei, 1); + + zebra_flush_reg (zh); + + resultSetInvalidate (zh); + + zebra_register_close (zh->service, zh->reg); + zh->reg = 0; + + yaz_log (YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d " + ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT, + zh->records_processed, zh->records_inserted, + zh->records_updated, zh->records_deleted); + + status->processed = (int) zh->records_processed; + status->inserted = (int) zh->records_inserted; + status->updated = (int) zh->records_updated; + status->deleted = (int) zh->records_deleted; + + zebra_get_state (zh, &val, &seqno); + if (val != 'd') + { + BFiles bfs = bfs_create (rval, zh->path_reg); + yaz_log (YLOG_DEBUG, "deleting shadow val=%c", val); + bf_commitClean (bfs, rval); + bfs_destroy (bfs); + } + if (!rval) + seqno++; + zebra_set_state (zh, 'o', seqno); + + zebra_unlock (zh->lock_shadow); + zebra_unlock (zh->lock_normal); + + } +#if HAVE_SYS_TIMES_H + times (&zh->tms2); + yaz_log (log_level, "user/system: %ld/%ld", + (long) (zh->tms2.tms_utime - zh->tms1.tms_utime), + (long) (zh->tms2.tms_stime - zh->tms1.tms_stime)); + + status->utime = (long) (zh->tms2.tms_utime - zh->tms1.tms_utime); + status->stime = (long) (zh->tms2.tms_stime - zh->tms1.tms_stime); +#endif + return ZEBRA_OK; } -static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid) +int zebra_repository_update (ZebraHandle zh, const char *path) { - ZebraHandle zh = (ZebraHandle) (p->handle); - zebraExplain_addSchema (zh->service->zei, oid); + ASSERTZH; + assert(path); + zebra_clearError(zh); + yaz_log (log_level, "updating %s", path); + repositoryUpdate (zh, path); + return zh->errCode; } -static void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno, - int cmd, struct sortKey **skp) +int zebra_repository_delete (ZebraHandle zh, const char *path) { - struct sortKey *sk = *skp; - SortIdx sortIdx = zh->service->sortIdx; - - sortIdx_sysno (sortIdx, sysno); - while (sk) - { - struct sortKey *sk_next = sk->next; - sortIdx_type (sortIdx, sk->attrUse); - sortIdx_add (sortIdx, sk->string, sk->length); - xfree (sk->string); - xfree (sk); - sk = sk_next; - } - *skp = 0; + ASSERTZH; + assert(path); + zebra_clearError(zh); + yaz_log (log_level, "deleting %s", path); + repositoryDelete (zh, path); + return zh->errCode; } -struct encode_info { - int sysno; - int seqno; - int cmd; - char buf[768]; -}; - -void encode_key_init (struct encode_info *i) +int zebra_repository_show (ZebraHandle zh, const char *path) { - i->sysno = 0; - i->seqno = 0; - i->cmd = -1; + ASSERTZH; + assert(path); + yaz_log(log_level, "zebra_repository_show"); + zebra_clearError(zh); + repositoryShow (zh, path); + return zh->errCode; } -char *encode_key_int (int d, char *bp) +static int zebra_commit_ex(ZebraHandle zh, int clean_only) { - if (d <= 63) - *bp++ = d; - else if (d <= 16383) + int seqno; + char val; + const char *rval; + BFiles bfs; + ASSERTZH; + zebra_clearError(zh); + + zebra_select_default_database(zh); + if (!zh->res) { - *bp++ = 64 + (d>>8); - *bp++ = d & 255; - } - else if (d <= 4194303) - { - *bp++ = 128 + (d>>16); - *bp++ = (d>>8) & 255; - *bp++ = d & 255; + zh->errCode = YAZ_BIB1_DATABASE_UNAVAILABLE; + return -1; } - else + rval = res_get (zh->res, "shadow"); + if (!rval) { - *bp++ = 192 + (d>>24); - *bp++ = (d>>16) & 255; - *bp++ = (d>>8) & 255; - *bp++ = d & 255; + yaz_log (YLOG_WARN, "Cannot perform commit - No shadow area defined"); + return 0; } - return bp; -} -void encode_key_write (char *k, struct encode_info *i, FILE *outf) -{ - struct it_key key; - char *bp = i->buf; + zebra_lock_w (zh->lock_normal); + zebra_lock_r (zh->lock_shadow); - while ((*bp++ = *k++)) - ; - memcpy (&key, k+1, sizeof(struct it_key)); - bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp); - if (i->sysno != key.sysno) + bfs = bfs_create (res_get (zh->res, "register"), zh->path_reg); + + zebra_get_state (zh, &val, &seqno); + + if (rval && *rval) + bf_cache (bfs, rval); + if (bf_commitExists (bfs)) { - i->sysno = key.sysno; - i->seqno = 0; + if (clean_only) + zebra_set_state (zh, 'd', seqno); + else + { + zebra_set_state (zh, 'c', seqno); + + yaz_log (YLOG_DEBUG, "commit start"); + bf_commitExec (bfs); +#ifndef WIN32 + sync (); +#endif + } + yaz_log (YLOG_DEBUG, "commit clean"); + bf_commitClean (bfs, rval); + seqno++; + zebra_set_state (zh, 'o', seqno); } - else if (!i->seqno && !key.seqno && i->cmd == *k) - return; - bp = encode_key_int (key.seqno - i->seqno, bp); - i->seqno = key.seqno; - i->cmd = *k; - if (fwrite (i->buf, bp - i->buf, 1, outf) != 1) + else { - logf (LOG_FATAL|LOG_ERRNO, "fwrite"); - exit (1); + yaz_log (log_level, "nothing to commit"); } + bfs_destroy (bfs); + + zebra_unlock (zh->lock_shadow); + zebra_unlock (zh->lock_normal); + return 0; } -static void extract_flushWriteKeys (ZebraHandle zh) +ZEBRA_RES zebra_clean(ZebraHandle zh) { - FILE *outf; - char out_fname[200]; - char *prevcp, *cp; - struct encode_info encode_info; - int ptr_i = zh->ptr_i; -#if SORT_EXTRA - int i; -#endif - if (!zh->key_buf || ptr_i <= 0) - return; + ASSERTZH; + yaz_log(log_level, "zebra_clean"); + return zebra_commit_ex(zh, 1); +} - (zh->key_file_no)++; - logf (LOG_LOG, "sorting section %d", (zh->key_file_no)); -#if !SORT_EXTRA - qsort (zh->key_buf + zh->ptr_top - ptr_i, ptr_i, sizeof(char*), - key_qsort_compare); - extract_get_fname_tmp (zh, out_fname, zh->key_file_no); +ZEBRA_RES zebra_commit(ZebraHandle zh) +{ + ASSERTZH; + yaz_log(log_level, "zebra_commit"); + return zebra_commit_ex(zh, 0); +} - if (!(outf = fopen (out_fname, "wb"))) +ZEBRA_RES zebra_init(ZebraHandle zh) +{ + const char *rval; + BFiles bfs = 0; + ASSERTZH; + yaz_log(log_level, "zebra_init"); + zebra_clearError(zh); + + zebra_select_default_database(zh); + if (!zh->res) { - logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname); - exit (1); + zebra_setError(zh, YAZ_BIB1_TEMPORARY_SYSTEM_ERROR, + "cannot select default database"); + return ZEBRA_FAIL; } - logf (LOG_LOG, "writing section %d", zh->key_file_no); - prevcp = cp = (zh->key_buf)[zh->ptr_top - ptr_i]; - - encode_key_init (&encode_info); - encode_key_write (cp, &encode_info, outf); - - while (--ptr_i > 0) - { - cp = (zh->key_buf)[zh->ptr_top - ptr_i]; - if (strcmp (cp, prevcp)) - { - encode_key_init (&encode_info); - encode_key_write (cp, &encode_info, outf); - prevcp = cp; - } - else - encode_key_write (cp + strlen(cp), &encode_info, outf); - } -#else - qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare); - extract_get_fname_tmp (out_fname, key_file_no); + rval = res_get (zh->res, "shadow"); - if (!(outf = fopen (out_fname, "wb"))) + bfs = bfs_create (res_get (zh->res, "register"), zh->path_reg); + if (!bfs) { - logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname); - exit (1); + zebra_setError(zh, YAZ_BIB1_TEMPORARY_SYSTEM_ERROR, "bfs_create"); + return ZEBRA_FAIL; } - logf (LOG_LOG, "writing section %d", key_file_no); - i = ptr_i; - prevcp = key_buf[ptr_top-i]; - while (1) - if (!--i || strcmp (prevcp, key_buf[ptr_top-i])) - { - key_y_len = strlen(prevcp)+1; -#if 0 - logf (LOG_LOG, "key_y_len: %2d %02x %02x %s", - key_y_len, prevcp[0], prevcp[1], 2+prevcp); -#endif - qsort (key_buf + ptr_top-ptr_i, ptr_i - i, - sizeof(char*), key_y_compare); - cp = key_buf[ptr_top-ptr_i]; - --key_y_len; - encode_key_init (&encode_info); - encode_key_write (cp, &encode_info, outf); - while (--ptr_i > i) - { - cp = key_buf[ptr_top-ptr_i]; - encode_key_write (cp+key_y_len, &encode_info, outf); - } - if (!i) - break; - prevcp = key_buf[ptr_top-ptr_i]; - } -#endif - if (fclose (outf)) + if (rval && *rval) + bf_cache (bfs, rval); + + bf_reset (bfs); + bfs_destroy (bfs); + zebra_set_state (zh, 'o', 0); + return ZEBRA_OK; +} + +ZEBRA_RES zebra_compact(ZebraHandle zh) +{ + BFiles bfs; + ASSERTZH; + yaz_log(log_level, "zebra_compact"); + zebra_clearError(zh); + if (!zh->res) { - logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname); - exit (1); + zh->errCode = YAZ_BIB1_DATABASE_UNAVAILABLE; + return ZEBRA_FAIL; } - logf (LOG_LOG, "finished section %d", zh->key_file_no); - zh->ptr_i = 0; - zh->key_buf_used = 0; + bfs = bfs_create (res_get (zh->res, "register"), zh->path_reg); + inv_compact (bfs); + bfs_destroy (bfs); + return ZEBRA_OK; } -static void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno, - int cmd, struct recKeys *reckeys) +void zebra_result(ZebraHandle zh, int *code, char **addinfo) { -#if SU_SCHEME -#else - unsigned char attrSet = (unsigned char) -1; - unsigned short attrUse = (unsigned short) -1; -#endif - int seqno = 0; - int off = 0; - int ch = 0; - ZebraExplainInfo zei = zh->service->zei; - - if (!zh->key_buf) - { - int mem = 8*1024*1024; - zh->key_buf = (char**) xmalloc (mem); - zh->ptr_top = mem/sizeof(char*); - zh->ptr_i = 0; - zh->key_buf_used = 0; - zh->key_file_no = 0; - } - zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1); - while (off < reckeys->buf_used) - { - const char *src = reckeys->buf + off; - struct it_key key; - int lead; - - lead = *src++; - -#if SU_SCHEME - if ((lead & 3) < 3) - { - memcpy (&ch, src, sizeof(ch)); - src += sizeof(ch); - } -#else - if (!(lead & 1)) - { - memcpy (&attrSet, src, sizeof(attrSet)); - src += sizeof(attrSet); - } - if (!(lead & 2)) - { - memcpy (&attrUse, src, sizeof(attrUse)); - src += sizeof(attrUse); - } -#endif - if (zh->key_buf_used + 1024 > (zh->ptr_top-zh->ptr_i)*sizeof(char*)) - extract_flushWriteKeys (zh); - ++(zh->ptr_i); - (zh->key_buf)[zh->ptr_top - zh->ptr_i] = - (char*)zh->key_buf + zh->key_buf_used; -#if SU_SCHEME -#else - ch = zebraExplain_lookupSU (zei, attrSet, attrUse); - if (ch < 0) - ch = zebraExplain_addSU (zei, attrSet, attrUse); -#endif - assert (ch > 0); - zh->key_buf_used += - key_SU_code (ch,((char*)zh->key_buf) + zh->key_buf_used); - - while (*src) - ((char*)zh->key_buf) [(zh->key_buf_used)++] = *src++; - src++; - ((char*)(zh->key_buf))[(zh->key_buf_used)++] = '\0'; - ((char*)(zh->key_buf))[(zh->key_buf_used)++] = cmd; - - if (lead & 60) - seqno += ((lead>>2) & 15)-1; - else - { - memcpy (&seqno, src, sizeof(seqno)); - src += sizeof(seqno); - } - key.seqno = seqno; - key.sysno = sysno; - memcpy ((char*)zh->key_buf + zh->key_buf_used, &key, sizeof(key)); - (zh->key_buf_used) += sizeof(key); - off = src - reckeys->buf; - } - assert (off == reckeys->buf_used); + ASSERTZH; + yaz_log(log_level, "zebra_result"); + *code = zh->errCode; + *addinfo = zh->errString; } -static void extract_index (ZebraHandle zh) +void zebra_shadow_enable(ZebraHandle zh, int value) { - extract_flushWriteKeys (zh); - zebra_index_merge (zh); + ASSERTZH; + yaz_log(log_level, "zebra_shadow_enable"); + zebra_clearError(zh); + zh->shadow_enable = value; } -static int explain_extract (void *handle, Record rec, data1_node *n) +ZEBRA_RES zebra_octet_term_encoding(ZebraHandle zh, const char *encoding) { - ZebraHandle zh = (ZebraHandle) handle; - struct recExtractCtrl extractCtrl; - int i; + ASSERTZH; + assert(encoding); + yaz_log(log_level, "zebra_octet_term_encoding"); + zebra_clearError(zh); + + if (zh->iconv_to_utf8 != 0) + yaz_iconv_close(zh->iconv_to_utf8); + if (zh->iconv_from_utf8 != 0) + yaz_iconv_close(zh->iconv_from_utf8); + + zh->iconv_to_utf8 = + yaz_iconv_open ("UTF-8", encoding); + if (zh->iconv_to_utf8 == 0) + yaz_log (YLOG_WARN, "iconv: %s to UTF-8 unsupported", encoding); + zh->iconv_from_utf8 = + yaz_iconv_open (encoding, "UTF-8"); + if (zh->iconv_to_utf8 == 0) + yaz_log (YLOG_WARN, "iconv: UTF-8 to %s unsupported", encoding); + + return ZEBRA_OK; +} - if (zebraExplain_curDatabase (zh->service->zei, - rec->info[recInfo_databaseName])) - { - abort(); - if (zebraExplain_newDatabase (zh->service->zei, - rec->info[recInfo_databaseName], 0)) - abort (); - } +ZEBRA_RES zebra_record_encoding (ZebraHandle zh, const char *encoding) +{ + ASSERTZH; + yaz_log(log_level, "zebra_record_encoding"); + zebra_clearError(zh); + xfree(zh->record_encoding); + zh->record_encoding = 0; + if (encoding) + zh->record_encoding = xstrdup (encoding); + return ZEBRA_OK; +} - zh->keys.buf_used = 0; - zh->keys.prevAttrUse = -1; - zh->keys.prevAttrSet = -1; - zh->keys.prevSeqNo = 0; - zh->sortKeys = 0; - - extractCtrl.init = extract_init; - extractCtrl.tokenAdd = extract_token_add; - extractCtrl.schemaAdd = extract_schema_add; - extractCtrl.dh = zh->service->dh; - for (i = 0; i<256; i++) - extractCtrl.seqno[i] = 0; - extractCtrl.zebra_maps = zh->service->zebra_maps; - extractCtrl.flagShowRecords = 0; - extractCtrl.handle = handle; - - grs_extract_tree(&extractCtrl, n); +void zebra_set_resource(ZebraHandle zh, const char *name, const char *value) +{ + ASSERTZH; + assert(name); + assert(value); + yaz_log(log_level, "zebra_set_resource %s:%s", name, value); + zebra_clearError(zh); + res_set(zh->res, name, value); +} - logf (LOG_LOG, "flush explain record, sysno=%d", rec->sysno); +const char *zebra_get_resource(ZebraHandle zh, + const char *name, const char *defaultvalue) +{ + const char *v; + ASSERTZH; + assert(name); + assert(defaultvalue); + v = res_get_def (zh->res, name, (char *)defaultvalue); + zebra_clearError(zh); + yaz_log(log_level, "zebra_get_resource %s:%s", name, v); + return v; +} - if (rec->size[recInfo_delKeys]) - { - struct recKeys delkeys; - struct sortKey *sortKeys = 0; +/* moved from zebra_api_ext.c by pop */ +/* FIXME: Should this really be public??? -Heikki */ - delkeys.buf_used = rec->size[recInfo_delKeys]; - delkeys.buf = rec->info[recInfo_delKeys]; - extract_flushSortKeys (zh, rec->sysno, 0, &sortKeys); - extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys); - } - extract_flushRecordKeys (zh, rec->sysno, 1, &zh->keys); - extract_flushSortKeys (zh, rec->sysno, 1, &zh->sortKeys); +int zebra_trans_no (ZebraHandle zh) +{ + ASSERTZH; + yaz_log(log_level, "zebra_trans_no"); + return zh->trans_no; +} - xfree (rec->info[recInfo_delKeys]); - rec->size[recInfo_delKeys] = zh->keys.buf_used; - rec->info[recInfo_delKeys] = zh->keys.buf; - zh->keys.buf = NULL; - zh->keys.buf_max = 0; - return 0; +int zebra_get_shadow_enable (ZebraHandle zh) +{ + ASSERTZH; + yaz_log(log_level, "zebra_get_shadow_enable"); + return zh->shadow_enable; } -static int extract_rec_in_mem (ZebraHandle zh, const char *recordType, - const char *buf, size_t buf_size, - const char *databaseName, int delete_flag, - int test_mode, int *sysno, - int store_keys, int store_data, - const char *match_criteria) -{ - RecordAttr *recordAttr; - struct recExtractCtrl extractCtrl; - int i, r; - RecType recType; - char subType[1024]; - void *clientData; - const char *fname = ""; - Record rec; - long recordOffset = 0; - struct zebra_fetch_control fc; - - fc.fd = -1; - fc.record_int_buf = buf; - fc.record_int_len = buf_size; - fc.record_int_pos = 0; - fc.offset_end = 0; - fc.record_offset = 0; - - extractCtrl.offset = 0; - extractCtrl.readf = zebra_record_int_read; - extractCtrl.seekf = zebra_record_int_seek; - extractCtrl.tellf = zebra_record_int_tell; - extractCtrl.endf = zebra_record_int_end; - extractCtrl.fh = &fc; +void zebra_set_shadow_enable (ZebraHandle zh, int value) +{ + ASSERTZH; + yaz_log(log_level, "zebra_set_shadow_enable %d",value); + zh->shadow_enable = value; +} - /* announce database */ - if (zebraExplain_curDatabase (zh->service->zei, databaseName)) - { - if (zebraExplain_newDatabase (zh->service->zei, databaseName, 0)) - return 0; - } - if (!(recType = - recType_byName (zh->service->recTypes, recordType, subType, - &clientData))) - { - logf (LOG_WARN, "No such record type: %s", recordType); - return 0; - } +/* Used by Perl API.. Added the record buffer dup to zebra_records_retrieve + so that it's identicical to the original api_records_retrieve */ +void api_records_retrieve (ZebraHandle zh, ODR stream, + const char *setname, Z_RecordComposition *comp, + oid_value input_format, int num_recs, + ZebraRetrievalRecord *recs) +{ + zebra_records_retrieve(zh, stream, setname, comp, input_format, + num_recs, recs); +} - zh->keys.buf_used = 0; - zh->keys.prevAttrUse = -1; - zh->keys.prevAttrSet = -1; - zh->keys.prevSeqNo = 0; - zh->sortKeys = 0; +/* --------------------------------------------------------------------------- + Record insert(=update), delete - extractCtrl.subType = subType; - extractCtrl.init = extract_init; - extractCtrl.tokenAdd = extract_token_add; - extractCtrl.schemaAdd = extract_schema_add; - extractCtrl.dh = zh->service->dh; - extractCtrl.handle = zh; - extractCtrl.zebra_maps = zh->service->zebra_maps; - extractCtrl.flagShowRecords = 0; - for (i = 0; i<256; i++) - { - if (zebra_maps_is_positioned(zh->service->zebra_maps, i)) - extractCtrl.seqno[i] = 1; - else - extractCtrl.seqno[i] = 0; - } + If sysno is provided, then it's used to identify the record. + If not, and match_criteria is provided, then sysno is guessed + If not, and a record is provided, then sysno is got from there +NOTE: Now returns 0 at success and updates sysno, which is an int* + 20-jun-2003 Heikki +*/ - r = (*recType->extract)(clientData, &extractCtrl); +int zebra_add_record(ZebraHandle zh, + const char *buf, int buf_size) +{ + SYSNO sysno = 0; + return zebra_update_record(zh, 0, &sysno, 0, 0, buf, buf_size, 0); +} - if (r == RECCTRL_EXTRACT_EOF) - return 0; - else if (r == RECCTRL_EXTRACT_ERROR) - { - /* error occured during extraction ... */ -#if 1 - yaz_log (LOG_WARN, "extract error"); -#else - if (rGroup->flagRw && - records_processed < rGroup->fileVerboseLimit) - { - logf (LOG_WARN, "fail %s %s %ld", rGroup->recordType, - fname, (long) recordOffset); - } -#endif - return 0; - } - if (zh->keys.buf_used == 0) - { - /* the extraction process returned no information - the record - is probably empty - unless flagShowRecords is in use */ - if (test_mode) - return 1; - logf (LOG_WARN, "No keys generated for record"); - logf (LOG_WARN, " The file is probably empty"); - return 1; - } - /* match criteria */ +ZEBRA_RES zebra_insert_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 res; + ASSERTZH; + assert(sysno); + assert(buf); + yaz_log(log_level, "zebra_insert_record sysno=" ZINT_FORMAT, *sysno); + + if (buf_size < 1) + buf_size = strlen(buf); + + if (zebra_begin_trans(zh, 1) == ZEBRA_FAIL) + return ZEBRA_FAIL; + res = buffer_extract_record (zh, buf, buf_size, + 0, /* delete_flag */ + 0, /* test_mode */ + recordType, + sysno, + match, fname, + 0, + 0); /* allow_update */ + zebra_end_trans(zh); + return res; +} - if (! *sysno) - { - /* new record */ - if (delete_flag) - { - logf (LOG_LOG, "delete %s %s %ld", recordType, - fname, (long) recordOffset); - logf (LOG_WARN, "cannot delete record above (seems new)"); - return 1; - } - logf (LOG_LOG, "add %s %s %ld", recordType, fname, - (long) recordOffset); - rec = rec_new (zh->service->records); +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 res; + ASSERTZH; + assert(sysno); + assert(buf); + + yaz_log(log_level, "zebra_update_record sysno=" ZINT_FORMAT, *sysno); + + if (buf_size < 1) buf_size = strlen(buf); + + if (zebra_begin_trans(zh, 1) == ZEBRA_FAIL) + return ZEBRA_FAIL; + res = buffer_extract_record (zh, buf, buf_size, + 0, /* delete_flag */ + 0, /* test_mode */ + recordType, + sysno, + match, fname, + force_update, + 1); /* allow_update */ + zebra_end_trans(zh); + return res; +} - *sysno = rec->sysno; +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 res; + ASSERTZH; + assert(sysno); + assert(buf); + yaz_log(log_level, "zebra_delete_record sysno=" ZINT_FORMAT, *sysno); + + if (buf_size < 1) buf_size = strlen(buf); + + if (zebra_begin_trans(zh, 1) == ZEBRA_FAIL) + return ZEBRA_FAIL; + res = buffer_extract_record (zh, buf, buf_size, + 1, /* delete_flag */ + 0, /* test_mode */ + recordType, + sysno, + match,fname, + force_update, + 1); /* allow_update */ + zebra_end_trans(zh); + return res; +} - recordAttr = rec_init_attr (zh->service->zei, rec); +/* --------------------------------------------------------------------------- + Searching +*/ -#if 0 - if (matchStr) - { - dict_insert (matchDict, matchStr, sizeof(*sysno), sysno); - } -#endif - extract_flushRecordKeys (zh, *sysno, 1, &zh->keys); - extract_flushSortKeys (zh, *sysno, 1, &zh->sortKeys); +ZEBRA_RES zebra_search_PQF(ZebraHandle zh, const char *pqf_query, + const char *setname, zint *hits) +{ + zint lhits = 0; + ZEBRA_RES res = ZEBRA_OK; + Z_RPNQuery *query; + ODR odr = odr_createmem(ODR_ENCODE); + ASSERTZH; + assert(pqf_query); + assert(setname); + + yaz_log(log_level, "zebra_search_PQF s=%s q=%s", setname, pqf_query); + + query = p_query_rpn (odr, PROTO_Z3950, pqf_query); + + if (!query) + { + yaz_log (YLOG_WARN, "bad query %s\n", pqf_query); + zh->errCode = YAZ_BIB1_MALFORMED_QUERY; + res = ZEBRA_FAIL; } else - { - /* record already exists */ - struct recKeys delkeys; + res = zebra_search_RPN(zh, odr, query, setname, &lhits); + + odr_destroy(odr); - rec = rec_get (zh->service->records, *sysno); - assert (rec); - - recordAttr = rec_init_attr (zh->service->zei, rec); + yaz_log(log_level, "Hits: " ZINT_FORMAT, lhits); - if (recordAttr->runNumber == - zebraExplain_runNumberIncrement (zh->service->zei, 0)) - { - logf (LOG_LOG, "skipped %s %s %ld", recordType, - fname, (long) recordOffset); - rec_rm (&rec); - return 1; - } - delkeys.buf_used = rec->size[recInfo_delKeys]; - delkeys.buf = rec->info[recInfo_delKeys]; - extract_flushSortKeys (zh, *sysno, 0, &zh->sortKeys); - extract_flushRecordKeys (zh, *sysno, 0, &delkeys); - if (delete_flag) - { - /* record going to be deleted */ - if (!delkeys.buf_used) - { - logf (LOG_LOG, "delete %s %s %ld", recordType, - fname, (long) recordOffset); - logf (LOG_WARN, "cannot delete file above, storeKeys false"); - } - else - { - logf (LOG_LOG, "delete %s %s %ld", recordType, - fname, (long) recordOffset); -#if 0 - if (matchStr) - dict_delete (matchDict, matchStr); -#endif - rec_del (zh->service->records, &rec); - } - rec_rm (&rec); - return 1; - } - else - { - /* record going to be updated */ - if (!delkeys.buf_used) - { - logf (LOG_LOG, "update %s %s %ld", recordType, - fname, (long) recordOffset); - logf (LOG_WARN, "cannot update file above, storeKeys false"); - } - else - { - logf (LOG_LOG, "update %s %s %ld", recordType, - fname, (long) recordOffset); - extract_flushRecordKeys (zh, *sysno, 1, &zh->keys); - } - } - } - /* update file type */ - xfree (rec->info[recInfo_fileType]); - rec->info[recInfo_fileType] = - rec_strdup (recordType, &rec->size[recInfo_fileType]); + if (hits) + *hits = lhits; - /* update filename */ - xfree (rec->info[recInfo_filename]); - rec->info[recInfo_filename] = - rec_strdup (fname, &rec->size[recInfo_filename]); - - /* update delete keys */ - xfree (rec->info[recInfo_delKeys]); - if (zh->keys.buf_used > 0 && store_keys == 1) - { - rec->size[recInfo_delKeys] = zh->keys.buf_used; - rec->info[recInfo_delKeys] = zh->keys.buf; - zh->keys.buf = NULL; - zh->keys.buf_max = 0; - } - else - { - rec->info[recInfo_delKeys] = NULL; - rec->size[recInfo_delKeys] = 0; - } + return res; +} - /* save file size of original record */ - zebraExplain_recordBytesIncrement (zh->service->zei, - - recordAttr->recordSize); -#if 0 - recordAttr->recordSize = fi->file_moffset - recordOffset; - if (!recordAttr->recordSize) - recordAttr->recordSize = fi->file_max - recordOffset; -#else - recordAttr->recordSize = buf_size; -#endif - zebraExplain_recordBytesIncrement (zh->service->zei, - recordAttr->recordSize); - - /* set run-number for this record */ - recordAttr->runNumber = - zebraExplain_runNumberIncrement (zh->service->zei, 0); - - /* update store data */ - xfree (rec->info[recInfo_storeData]); - if (store_data == 1) - { - rec->size[recInfo_storeData] = recordAttr->recordSize; - rec->info[recInfo_storeData] = (char *) - xmalloc (recordAttr->recordSize); -#if 1 - memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize); -#else - if (lseek (fi->fd, recordOffset, SEEK_SET) < 0) - { - logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s", - (long) recordOffset, fname); - exit (1); - } - if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize) - < recordAttr->recordSize) - { - logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s", - recordAttr->recordSize, fname); - exit (1); - } -#endif - } - else +/* --------------------------------------------------------------------------- + Sort - a simplified interface, with optional read locks. +*/ +int zebra_sort_by_specstr (ZebraHandle zh, ODR stream, + const char *sort_spec, + const char *output_setname, + const char **input_setnames) +{ + int num_input_setnames = 0; + int sort_status = 0; + Z_SortKeySpecList *sort_sequence; + ASSERTZH; + assert(stream); + assert(sort_spec); + assert(output_setname); + assert(input_setnames); + sort_sequence = yaz_sort_spec (stream, sort_spec); + yaz_log(log_level, "sort (FIXME) "); + if (!sort_sequence) { - rec->info[recInfo_storeData] = NULL; - rec->size[recInfo_storeData] = 0; + yaz_log(YLOG_WARN, "invalid sort specs '%s'", sort_spec); + zh->errCode = YAZ_BIB1_CANNOT_SORT_ACCORDING_TO_SEQUENCE; + return -1; } - /* update database name */ - xfree (rec->info[recInfo_databaseName]); - rec->info[recInfo_databaseName] = - rec_strdup (databaseName, &rec->size[recInfo_databaseName]); + + /* we can do this, since the perl typemap code for char** will + put a NULL at the end of list */ + while (input_setnames[num_input_setnames]) num_input_setnames++; - /* update offset */ - recordAttr->recordOffset = recordOffset; + if (zebra_begin_read (zh)) + return -1; + + resultSetSort (zh, stream->mem, num_input_setnames, input_setnames, + output_setname, sort_sequence, &sort_status); - /* commit this record */ - rec_put (zh->service->records, &rec); + zebra_end_read(zh); + return sort_status; +} +/* --------------------------------------------------------------------------- + Get BFS for Zebra system (to make alternative storage methods) +*/ +struct BFiles_struct *zebra_get_bfs(ZebraHandle zh) +{ + if (zh && zh->reg) + return zh->reg->bfs; return 0; } + + +/* --------------------------------------------------------------------------- + Set limit for search/scan +*/ +ZEBRA_RES zebra_set_limit(ZebraHandle zh, int complement_flag, zint *ids) +{ + ASSERTZH; + zebra_limit_destroy(zh->m_limit); + zh->m_limit = zebra_limit_create(complement_flag, ids); + return ZEBRA_OK; +} + +/* + Set Error code + addinfo +*/ +void zebra_setError(ZebraHandle zh, int code, const char *addinfo) +{ + zh->errCode = code; + nmem_reset(zh->nmem_error); + zh->errString = addinfo ? nmem_strdup(zh->nmem_error, addinfo) : 0; +} + +void zebra_setError_zint(ZebraHandle zh, int code, zint i) +{ + char vstr[60]; + sprintf(vstr, ZINT_FORMAT, i); + + zh->errCode = code; + nmem_reset(zh->nmem_error); + zh->errString = nmem_strdup(zh->nmem_error, vstr); +} +