X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=index%2Fzebraapi.c;h=3d9d89f246a9de2427fd4fd41cc5af20b12d4776;hp=b8181a5177ac027d93ab77fea68a70b3589e71ef;hb=13bf0c380e98816595127291b64368dd9b8ca639;hpb=297ba5c5c265a5f869f43a0a211bf9f48f700add diff --git a/index/zebraapi.c b/index/zebraapi.c index b8181a5..3d9d89f 100644 --- a/index/zebraapi.c +++ b/index/zebraapi.c @@ -1,121 +1,28 @@ -/* - * Copyright (C) 1995-2000, Index Data - * All rights reserved. - * - * $Log: zebraapi.c,v $ - * 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 1.128 2004-09-03 14:59:49 heikki Exp $ + Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004 + 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 @@ -125,12 +32,35 @@ #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) + +/* A simple log macro */ +/* don't break with older yazen that lack log_app2 */ +#ifdef LOG_APP2 +#define LOG_API LOG_APP2 +#else +#define LOG_API LOG_DEBUG +#endif -static void zebra_chdir (ZebraService zh) +static Res zebra_open_res (ZebraHandle zh); +static void zebra_close_res (ZebraHandle zh); + + +static void zebra_chdir (ZebraService zs) { - const char *dir = res_get (zh->res, "chdir"); + const char *dir ; + ASSERTZS; + yaz_log(LOG_API,"zebra_chdir"); + dir = res_get (zs->global_res, "chdir"); if (!dir) return; logf (LOG_DEBUG, "chdir %s", dir); @@ -141,304 +71,419 @@ 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_lock (ZebraHandle zh) +static void zebra_flush_reg (ZebraHandle zh) { - zh->errCode = 0; - zh->errString = 0; - if (!zh->service->active) - { - zh->errCode = 1019; - return 1; - } -#if HAVE_SYS_TIMES_H - times (&zh->tms1); -#endif - return 0; + ASSERTZH; + yaz_log(LOG_API,"zebra_flush_reg"); + zh->errCode=0; + zebraExplain_flush (zh->reg->zei, zh); + + extract_flushWriteKeys (zh,1 /* final */); + zebra_index_merge (zh ); } -static void zebra_register_unlock (ZebraHandle zh) -{ -#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; + ASSERTZS; + yaz_log(LOG_API,"zebra_open"); - assert (zs); - if (zs->stop_flag) - return 0; + if (!zs) + return 0; zh = (ZebraHandle) xmalloc (sizeof(*zh)); + yaz_log (LOG_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->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->record_encoding = xstrdup (default_encoding); + + zh->iconv_to_utf8 = + yaz_iconv_open ("UTF-8", default_encoding); + if (zh->iconv_to_utf8 == 0) + yaz_log (LOG_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 (LOG_WARN, "iconv: UTF-8 to %s unsupported", + default_encoding); - zh->key_buf = 0; - zh->admin_databaseName = 0; - zebra_mutex_cond_lock (&zs->session_lock); zh->next = zs->sessions; zs->sessions = zh; zebra_mutex_cond_unlock (&zs->session_lock); + return zh; } -static int zebra_register_activate (ZebraService zh); -static int zebra_register_deactivate (ZebraService 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 = 0; - zebra_mutex_cond_init (&zh->session_lock); - zebra_register_activate (zh); - return zh; + return zebra_start_res(configName, 0, 0); } -static int zebra_register_activate (ZebraService zh) +ZebraService zebra_start_res (const char *configName, Res def_res, Res over_res) { - if (zh->active) - return 0; - yaz_log (LOG_LOG, "zebra_register_activate"); - if (!(zh->res = res_open (zh->configName))) + Res res; + + yaz_log(LOG_API|LOG_LOG,"zebra_start %s",configName); + + if ((res = res_open (configName, def_res, over_res))) { - logf (LOG_WARN, "Failed to read resources `%s'", zh->configName); - return -1; + ZebraService zh = xmalloc (sizeof(*zh)); + + yaz_log (LOG_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); + if (!res_get (zh->global_res, "passwd")) + zh->passwd_db = NULL; + else + { + 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->global_res, "passwd")); + } + zh->path_root = res_get (zh->global_res, "root"); + return zh; } - zebra_chdir (zh); - zh->dh = data1_create (); - if (!zh->dh) - return -1; - zh->bfs = bfs_create (res_get (zh->res, "register")); - if (!zh->bfs) + return 0; +} + + +void zebra_pidfname(ZebraService zs, char *path) +{ + zebra_lock_prefix (zs->global_res, path); + strcat(path, "zebrasrv.pid"); +} + +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 (LOG_LOG|LOG_API, "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) + return 0; + reg->bfs = bfs_create (res_get (res, "register"), reg_path); + if (!reg->bfs) { - data1_destroy(zh->dh); - return -1; + data1_destroy(reg->dh); + return 0; } - bf_lockDir (zh->bfs, res_get (zh->res, "lockDir")); - data1_set_tabpath (zh->dh, res_get(zh->res, "profilePath")); - zh->registerState = -1; /* trigger open of registers! */ - zh->registerChange = 0; - 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; + if (useshadow) + bf_cache (reg->bfs, res_get (res, "shadow")); + + getcwd(cwd, sizeof(cwd)-1); + profilePath = res_get_def(res, "profilePath", DEFAULT_PROFILE_PATH); + yaz_log(LOG_LOG, "profilePath=%s cwd=%s", profilePath, cwd); + + data1_set_tabpath (reg->dh, profilePath); + data1_set_tabroot (reg->dh, reg_path); + reg->recTypes = recTypes_init (reg->dh); + recTypes_default_handlers (reg->recTypes); + + 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; +#if IT_KEY_NEW + reg->keys.codec_handle = iscz1_start(); #endif - zh->zei = 0; + + 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); + zebraRankInstall (reg, rankliv_class); - if (!res_get (zh->res, "passwd")) - zh->passwd_db = NULL; - else - { - 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")); - } + 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 (!(zh->records = rec_open (zh->bfs, 1, 0))) + if (!(reg->records = rec_open (reg->bfs, rw, record_compression))) { logf (LOG_WARN, "rec_open"); - return -1; + return 0; + } + if (rw) + { + reg->matchDict = dict_open (reg->bfs, GMATCH_DICT, 20, 1, 0); } - if (!(zh->dict = dict_open (zh->bfs, FNAME_DICT, 80, 1, 0))) + if (!(reg->dict = dict_open (reg->bfs, FNAME_DICT, 40, rw, 0))) { logf (LOG_WARN, "dict_open"); - return -1; + return 0; } - if (!(zh->sortIdx = sortIdx_open (zh->bfs, 0))) + if (!(reg->sortIdx = sortIdx_open (reg->bfs, rw))) { logf (LOG_WARN, "sortIdx_open"); - return -1; + 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, 1, - 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; + 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, 1, - sizeof (struct it_key), zh->res))) + struct ISAMC_M_s isamc_m; + if (!(reg->isamc = isc_open (reg->bfs, FNAME_ISAMC, + rw, key_isamc_m(res, &isamc_m)))) { - logf (LOG_WARN, "is_open"); - return -1; + logf (LOG_WARN, "isc_open"); + 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, - 1, 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; + logf (LOG_WARN, "isamb_open"); + 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, - 1, 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; + logf (LOG_WARN, "isamb_open"); + return 0; } } -#endif - zh->zei = zebraExplain_open (zh->records, zh->dh, - zh->res, 1, 0 /* rGroup */, - explain_extract); - if (!zh->zei) + if (res_get_match (res, "isam", "null", ISAM_DEFAULT)) + { + struct ISAMC_M_s isamc_m; + + if (!(reg->isamb = isamb_open (reg->bfs, "isamb", + rw, key_isamc_m(res, &isamc_m), -1))) + { + logf (LOG_WARN, "isamb_open"); + return 0; + } + } + reg->zei = zebraExplain_open (reg->records, reg->dh, + res, rw, reg, + explain_extract); + if (!reg->zei) { logf (LOG_WARN, "Cannot obtain EXPLAIN information"); - return -1; + return 0; } - zh->active = 1; - yaz_log (LOG_LOG, "zebra_register_activate ok"); - return 0; + reg->active = 2; + yaz_log (LOG_DEBUG, "zebra_register_open ok p=%p", reg); + return reg; } -void zebra_admin_shutdown (ZebraHandle zh) +int zebra_admin_shutdown (ZebraHandle zh) { - zebraExplain_flush (zh->service->zei, 1, zh); - extract_index (zh); + ASSERTZH; + yaz_log(LOG_API,"zebra_admin_shutdown"); + zh->errCode=0; zebra_mutex_cond_lock (&zh->service->session_lock); zh->service->stop_flag = 1; - if (!zh->service->sessions) - zebra_register_deactivate(zh->service); zebra_mutex_cond_unlock (&zh->service->session_lock); + return 0; } -void zebra_admin_start (ZebraHandle zh) +int zebra_admin_start (ZebraHandle zh) { - ZebraService zs = zh->service; - zh->errCode = 0; + ZebraService zs; + ASSERTZH; + yaz_log(LOG_API,"zebra_admin_start"); + zh->errCode=0; + zs = zh->service; zebra_mutex_cond_lock (&zs->session_lock); - if (!zs->stop_flag) - zebra_register_activate(zs); zebra_mutex_cond_unlock (&zs->session_lock); + return 0; } -static int zebra_register_deactivate (ZebraService zh) +static void zebra_register_close (ZebraService zs, struct zebra_register *reg) { - zh->stop_flag = 0; - if (!zh->active) - return 0; - yaz_log(LOG_LOG, "zebra_register_deactivate"); - zebra_chdir (zh); - if (zh->records) + ASSERTZS; + yaz_log(LOG_LOG|LOG_API, "zebra_register_close p=%p", reg); + reg->stop_flag = 0; + zebra_chdir (zs); + if (reg->records) { - zebraExplain_close (zh->zei, 1); - dict_close (zh->dict); - sortIdx_close (zh->sortIdx); - if (zh->isams) - isams_close (zh->isams); -#if ZMBOL - if (zh->isam) - is_close (zh->isam); - if (zh->isamc) - isc_close (zh->isamc); - if (zh->isamd) - isamd_close (zh->isamd); -#endif - rec_close (&zh->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) + isc_close (reg->isamc); + if (reg->isamb) + isamb_close (reg->isamb); + rec_close (®->records); } - recTypes_destroy (zh->recTypes); - zebra_maps_close (zh->zebra_maps); - zebraRankDestroy (zh); - bfs_destroy (zh->bfs); - data1_destroy (zh->dh); - - if (zh->passwd_db) - passwd_db_close (zh->passwd_db); - res_close (zh->res); - zh->active = 0; - 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 IT_KEY_NEW + if (reg->keys.codec_handle) + iscz1_stop(reg->keys.codec_handle); +#endif + + xfree (reg->key_buf); + xfree (reg->name); + xfree (reg); } -void zebra_stop(ZebraService zh) +int zebra_stop(ZebraService zs) { - if (!zh) - return ; - yaz_log (LOG_LOG, "zebra_stop"); + if (!zs) + return 0; + yaz_log (LOG_LOG|LOG_API, "zebra_stop"); - assert (!zh->sessions); + while (zs->sessions) + { + zebra_close (zs->sessions); + } + + zebra_mutex_cond_destroy (&zs->session_lock); - zebra_mutex_cond_destroy (&zh->session_lock); + if (zs->passwd_db) + passwd_db_close (zs->passwd_db); - zebra_register_deactivate(zh); - xfree (zh->configName); - xfree (zh); + res_close (zs->global_res); + xfree (zs->configName); + xfree (zs); + return 0; } -void zebra_close (ZebraHandle zh) +int zebra_close (ZebraHandle zh) { - ZebraService zs = zh->service; + ZebraService zs; struct zebra_session **sp; + int i; + + yaz_log(LOG_API,"zebra_close"); if (!zh) - return ; + return 0; + ASSERTZH; + zh->errCode=0; + + zs = zh->service; + yaz_log (LOG_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) { @@ -450,10 +495,13 @@ 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->reg_name); + xfree (zh->user_perm); + zh->service=0; /* more likely to trigger an assert */ + xfree (zh->path_reg); xfree (zh); + return 0; } struct map_baseinfo { @@ -465,7 +513,108 @@ 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 (LOG_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 int zebra_select_register (ZebraHandle zh, const char *new_reg) +{ + ASSERTZH; + zh->errCode=0; + if (zh->res && strcmp (zh->reg_name, new_reg) == 0) + return 0; + 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); + + } + return 1; +} + void map_basenames_func (void *vp, const char *name, const char *value) { struct map_baseinfo *p = (struct map_baseinfo *) vp; @@ -480,7 +629,7 @@ 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++) @@ -500,6 +649,9 @@ void map_basenames (ZebraHandle zh, ODR stream, struct map_baseinfo info; struct map_baseinfo *p = &info; int i; + ASSERTZH; + yaz_log(LOG_API,"map_basenames "); + zh->errCode=0; info.zh = zh; info.num_bases = *num_bases; @@ -510,7 +662,7 @@ 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) @@ -524,29 +676,154 @@ void map_basenames (ZebraHandle zh, ODR stream, logf (LOG_LOG, "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) +int zebra_select_database (ZebraHandle zh, const char *basename) +{ + ASSERTZH; + yaz_log(LOG_API,"zebra_select_database %s",basename); + zh->errCode=0; + return zebra_select_databases (zh, 1, &basename); +} + +int zebra_select_databases (ZebraHandle zh, int num_bases, + const char **basenames) +{ + int i; + const char *cp; + int len = 0; + char *new_reg = 0; + ASSERTZH; + yaz_log(LOG_API,"zebra_select_databases n=%d [0]=%s", + num_bases,basenames[0]); + zh->errCode=0; + + if (num_bases < 1) + { + zh->errCode = 23; + return -1; + } + 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]); + + cp = strrchr(basenames[0], '/'); + if (cp) + { + 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 = 23; + return -1; + } + if (len != cp1 - basenames[i] || + memcmp (basenames[i], new_reg, len)) + { + zh->errCode = 23; + return -1; + } + } + else + { + if (cp1) + { + zh->errCode = 23; + return -1; + } + } + } + zebra_select_register (zh, new_reg); + xfree (new_reg); + if (!zh->res) + { + zh->errCode = 109; + return -1; + } + if (!zh->lock_normal || !zh->lock_shadow) + { + zh->errCode = 2; + return -1; + } + return 0; +} + +int zebra_search_RPN (ZebraHandle zh, ODR o, + Z_RPNQuery *query, const char *setname, int *hits) { + const char *max; + zint maxhits; + ASSERTZH; + yaz_log(LOG_API,"zebra_search_rpn"); + zh->errCode=0; zh->hits = 0; - if (zebra_register_lock (zh)) - return; - map_basenames (zh, stream, &num_bases, &basenames); - resultSetAddRPN (zh, stream, decode, query, num_bases, basenames, setname); + *hits = 0; - zebra_register_unlock (zh); + if (zebra_begin_read (zh)) + return 1; + + zebra_livcode_transform(zh, query); + + resultSetAddRPN (zh, odr_extract_mem(o), query, + zh->num_basenames, zh->basenames, setname); + + zebra_end_read (zh); + max = res_get (zh->res, "maxhits"); + if (max) + maxhits=atoi(max); + else { + int i=0; + maxhits=INT_MAX; + while (maxhits>100) { maxhits/=10; i++;} + while (i--) maxhits *= 10; + } + if (zh->hits > maxhits) { /* too large for yaz to handle */ + logf(LOG_LOG,"limiting hits to "ZINT_FORMAT, maxhits); + *hits=maxhits; /* round it down to two digits, to look like rounded */ + } + else + *hits = zh->hits; + return 0; } -void zebra_records_retrieve (ZebraHandle zh, ODR stream, +int zebra_records_retrieve (ZebraHandle zh, ODR stream, const char *setname, Z_RecordComposition *comp, oid_value input_format, int num_recs, ZebraRetrievalRecord *recs) { ZebraPosSet poset; - int i, *pos_array; + int i, *pos_array, ret = 0; + ASSERTZH; + yaz_log(LOG_API,"zebra_records_retrieve n=%d",num_recs); + zh->errCode=0; + + if (!zh->res) + { + zh->errCode = 30; + zh->errString = odr_strdup (stream, setname); + return -1; + } + + zh->errCode = 0; + + if (zebra_begin_read (zh)) + return -1; - if (zebra_register_lock (zh)) - return; pos_array = (int *) xmalloc (num_recs * sizeof(*pos_array)); for (i = 0; ierrCode = 30; zh->errString = nmem_strdup (stream->mem, setname); + ret = -1; } else { for (i = 0; ierrCode = 13; - zh->errString = nmem_strdup (stream->mem, num_str); - break; + recs[i].errCode = 0; + recs[i].format = VAL_SUTRS; + recs[i].len = strlen(poset[i].term); + recs[i].buf = poset[i].term; + recs[i].base = poset[i].db; } - else + else if (poset[i].sysno) { recs[i].errCode = zebra_record_fetch (zh, poset[i].sysno, poset[i].score, @@ -580,42 +857,59 @@ void zebra_records_retrieve (ZebraHandle zh, ODR stream, &recs[i].base); recs[i].errString = NULL; } + else + { + char num_str[20]; + + sprintf (num_str, "%d", pos_array[i]); + zh->errCode = 13; + zh->errString = odr_strdup (stream, num_str); + ret = -1; + break; + } } zebraPosSetDestroy (zh, poset, num_recs); } - zebra_register_unlock (zh); + zebra_end_read (zh); xfree (pos_array); + return ret; } -void zebra_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, +int 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) { - if (zebra_register_lock (zh)) + ASSERTZH; + yaz_log(LOG_API,"zebra_scan"); + zh->errCode=0; + if (zebra_begin_read (zh)) { *entries = 0; *num_entries = 0; - return; + return 1; } - 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); + zh->num_basenames, zh->basenames, position, + num_entries, entries, is_partial, 0, 0); + zebra_end_read (zh); + return 0; } -void zebra_sort (ZebraHandle zh, ODR stream, +int 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)) - return; + ASSERTZH; + yaz_log(LOG_API,"zebra_sort"); + zh->errCode=0; + if (zebra_begin_read (zh)) + return 1; resultSetSort (zh, stream->mem, num_input_setnames, input_setnames, output_setname, sort_sequence, sort_status); - zebra_register_unlock (zh); + zebra_end_read(zh); + return 0; } int zebra_deleleResultSet(ZebraHandle zh, int function, @@ -623,18 +917,21 @@ int zebra_deleleResultSet(ZebraHandle zh, int function, int *statuses) { int i, status; - if (zebra_register_lock (zh)) + ASSERTZH; + yaz_log(LOG_API,"zebra_deleleResultSet n=%d",num_setnames); + zh->errCode=0; + if (zebra_begin_read(zh)) return Z_DeleteStatus_systemProblemAtTarget; switch (function) { - case Z_DeleteRequest_list: + case Z_DeleteResultSetRequest_list: 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_API,"zebra_errCode: %d",zh->errCode); + return zh->errCode; + } + yaz_log(LOG_API,"zebra_errCode: o"); + return 0; } const char *zebra_errString (ZebraHandle zh) { - return diagbib1_str (zh->errCode); + const char *e=""; + if (zh) + e= diagbib1_str (zh->errCode); + yaz_log(LOG_API,"zebra_errString: %s",e); + return e; } char *zebra_errAdd (ZebraHandle zh) { - return zh->errString; + char *a=""; + if (zh) + a= zh->errString; + yaz_log(LOG_API,"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=""; + } } -int zebra_auth (ZebraService zh, const char *user, const char *pass) +int zebra_auth (ZebraHandle zh, const char *user, const char *pass) { - if (!zh->passwd_db || !passwd_db_auth (zh->passwd_db, user, pass)) + const char *p; + char u[40]; + ZebraService zs; + + ASSERTZH; + zh->errCode=0; + 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 0; + + if (!zs->passwd_db || !passwd_db_auth (zs->passwd_db, user, pass)) return 0; return 1; } -void zebra_admin_import_begin (ZebraHandle zh, const char *database) +int zebra_admin_import_begin (ZebraHandle zh, const char *database, + const char *record_type) { - if (zebra_register_lock (zh)) - return; - xfree (zh->admin_databaseName); - zh->admin_databaseName = xstrdup(database); - zebra_register_unlock(zh); + ASSERTZH; + yaz_log(LOG_API,"zebra_admin_import_begin db=%s rt=%s", + database, record_type); + zh->errCode=0; + if (zebra_select_database(zh, database)) + return 1; + if (zebra_begin_trans (zh, 1)) + return 1; + return 0; } -void zebra_admin_import_end (ZebraHandle zh) +int zebra_admin_import_end (ZebraHandle zh) { - zebraExplain_flush (zh->service->zei, 1, zh); - extract_index (zh); + ASSERTZH; + yaz_log(LOG_API,"zebra_admin_import_end"); + zh->errCode=0; + zebra_end_trans (zh); + return 0; } -void zebra_admin_import_segment (ZebraHandle zh, Z_Segment *segment) +int zebra_admin_import_segment (ZebraHandle zh, Z_Segment *segment) { - int sysno; + SYSNO sysno; int i; - if (zebra_register_lock (zh)) - return; + ASSERTZH; + yaz_log(LOG_API,"zebra_admin_import_segment"); + zh->errCode=0; 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) { @@ -708,36 +1046,164 @@ void zebra_admin_import_segment (ZebraHandle zh, Z_Segment *segment) oct->buf); sysno = 0; - extract_rec_in_mem (zh, "grs.sgml", + + zebra_update_record(zh, + 0, /* record Type */ + &sysno, + 0, /* match */ + 0, /* fname */ oct->buf, oct->len, - databaseName, - 0 /* delete_flag */, - 0 /* test_mode */, - &sysno /* sysno */, - 1 /* store_keys */, - 1 /* store_data */, - 0 /* match criteria */); + 0); } } } - zebra_register_unlock(zh); + return 0; } -void zebra_admin_create (ZebraHandle zh, const char *database) +int 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)) + SYSNO sysno = 0; + char *rinfo = 0; + char recid_z[256]; + ASSERTZH; + yaz_log(LOG_API,"zebra_admin_exchange_record ac=%d", action); + zh->errCode=0; + + if (!recid_buf || recid_len <= 0 || recid_len >= sizeof(recid_z)) + return -1; + memcpy (recid_z, recid_buf, recid_len); + recid_z[recid_len] = 0; + + if (zebra_begin_trans(zh, 1)) + return -1; + + 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 -1; + } + + memcpy (&sysno, rinfo+1, sizeof(sysno)); + } + else + { + if (action == 2 || action == 3) /* fail if delete or update */ + { + zebra_end_trans(zh); + return -1; + } + action = 1; /* make it an insert (if it's an update).. */ + } + 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 0; +} + +int delete_w_handle(const char *info, void *handle) +{ + ZebraHandle zh = (ZebraHandle) handle; + ISAMC_P pos; + + 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; +} + +int zebra_drop_database (ZebraHandle zh, const char *database) +{ + int ret = 0; + ASSERTZH; + yaz_log(LOG_API,"zebra_drop_database"); + zh->errCode = 0; + + if (zebra_select_database (zh, database)) + return -1; + if (zebra_begin_trans (zh, 1)) + return -1; + if (zh->reg->isamb) + { + zebraExplain_curDatabase (zh->reg->zei, database); + + zebraExplain_trav_ord(zh->reg->zei, zh, delete_SU_handle); + zebraExplain_removeDatabase(zh->reg->zei, zh); + } + else + { + yaz_log(LOG_WARN, "drop database only supported for isam:b"); + ret = -1; } + zebra_end_trans (zh); + return ret; +} + +int zebra_create_database (ZebraHandle zh, const char *database) +{ + ASSERTZH; + yaz_log(LOG_API,"zebra_create_database"); + zh->errCode=0; + + if (zebra_select_database (zh, database)) + return -1; + if (zebra_begin_trans (zh, 1)) + return -1; + /* announce database */ - if (zebraExplain_newDatabase (zs->zei, database, 0 /* explainDatabase */)) + if (zebraExplain_newDatabase (zh->reg->zei, database, 0 + /* explainDatabase */)) { + zebra_end_trans (zh); zh->errCode = 224; - zh->errString = "Database already exist"; + zh->errString = "database already exist"; + return -1; } - zebra_register_unlock(zh); + zebra_end_trans (zh); + return 0; } int zebra_string_norm (ZebraHandle zh, unsigned reg_id, @@ -745,9 +1211,12 @@ int zebra_string_norm (ZebraHandle zh, unsigned reg_id, char *output_str, int output_len) { WRBUF wrbuf; - if (!zh->service->zebra_maps) + ASSERTZH; + yaz_log(LOG_API,"zebra_string_norm "); + zh->errCode=0; + 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; @@ -759,806 +1228,866 @@ int zebra_string_norm (ZebraHandle zh, unsigned reg_id, return wrbuf_len(wrbuf); } -static void extract_init (struct recExtractCtrl *p, RecWord *w) + +int 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_API,"zebra_set_state v=%d seq=%d", val, seqno); + zh->errCode=0; + + 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 (LOG_DEBUG, "%c %d %ld", val, seqno, p); + fprintf (f, "%c %d %ld\n", val, seqno, p); + fclose (f); + xfree (fname); + return 0; } -static void extract_add_index_string (RecWord *p, const char *string, - int length) +int 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; - 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_API,"zebra_get_state "); + zh->errCode=0; + 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); + return 0; +} - if (!(lead & 1)) - { - memcpy (dst, &attrSet, sizeof(attrSet)); - dst += sizeof(attrSet); - } - if (!(lead & 2)) - { - memcpy (dst, &attrUse, sizeof(attrUse)); - dst += sizeof(attrUse); - } - *dst++ = p->reg_type; - memcpy (dst, string, length); - dst += length; - *dst++ = '\0'; +int 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)++; +int 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; + + 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) +int 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) + if (!zh->res) { - char buf[IT_MAX_WORD+1]; - int i, remain; + zh->errCode = 2; + zh->errString = "zebra_begin_trans: no database selected"; + return -1; + } + ASSERTZHRES; + yaz_log(LOG_API,"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; + zh->errCode = 223; + zh->errString = 0; + return -1; } - 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) + { + zh->errCode = 2; + zh->errString = "zebra_begin_trans: write trans not allowed within read trans"; + return -1; + } + 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; + + zh->errCode=0; + + 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 (LOG_LOG, "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 (LOG_LOG, "previous transaction didn't reach commit"); + bf_commitClean (bfs, rval); + bfs_destroy (bfs); + } + else + { + yaz_log (LOG_WARN, "your previous transaction didn't finish"); + } + } + break; + } + if (pass == 2) + { + yaz_log (LOG_FATAL, "zebra_begin_trans couldn't finish commit"); + abort(); + return -1; + } + 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; + + zh->errCode = 2; + zh->errString = "zebra_begin_trans: cannot open register"; + yaz_log(LOG_FATAL, zh->errString); + return -1; + } } - if (!i) - return; - extract_add_string (p, buf, i); -} - -static void extract_token_add (RecWord *p) -{ - WRBUF wrbuf; - if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0, - p->string, p->length))) + else { - p->string = wrbuf_buf(wrbuf); - p->length = wrbuf_len(wrbuf); + int dirty = 0; + char val; + int seqno; + + (zh->trans_no)++; + + if (zh->trans_no != 1) + { + zebra_flush_reg (zh); + return 0; + } + zh->errCode=0; +#if HAVE_SYS_TIMES_H + times (&zh->tms1); +#endif + if (!zh->res) + { + (zh->trans_no)--; + zh->errCode = 109; + return -1; + } + if (!zh->lock_normal || !zh->lock_shadow) + { + (zh->trans_no)--; + zh->errCode = 2; + return -1; + } + zebra_get_state (zh, &val, &seqno); + if (val == 'd') + val = 'o'; + + if (!zh->reg) + dirty = 1; + else if (seqno != zh->reg->seqno) + { + yaz_log (LOG_LOG, "reopen seqno cur/old %d/%d", + seqno, zh->reg->seqno); + dirty = 1; + } + else if (zh->reg->last_val != val) + { + yaz_log (LOG_LOG, "reopen last cur/old %d/%d", + val, zh->reg->last_val); + dirty = 1; + } + if (!dirty) + return 0; + + 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 = 109; + return -1; + } + zh->reg->last_val = val; + zh->reg->seqno = seqno; } - if (zebra_maps_is_complete (p->zebra_maps, p->reg_type)) - extract_add_complete_field (p); - else - extract_add_incomplete_field(p); + read_res_for_transaction(zh); + return 0; } -static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid) +int zebra_end_trans (ZebraHandle zh) { - ZebraHandle zh = (ZebraHandle) (p->handle); - zebraExplain_addSchema (zh->service->zei, oid); + ZebraTransactionStatus dummy; + yaz_log(LOG_API,"zebra_end_trans"); + return zebra_end_transaction(zh, &dummy); } -static void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno, - int cmd, struct sortKey **skp) +int zebra_end_transaction (ZebraHandle zh, ZebraTransactionStatus *status) { - struct sortKey *sk = *skp; - SortIdx sortIdx = zh->service->sortIdx; + char val; + int seqno; + const char *rval; - 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; -} - -struct encode_info { - int sysno; - int seqno; - int cmd; - char buf[768]; -}; + ASSERTZH; + yaz_log(LOG_API,"zebra_end_transaction"); -void encode_key_init (struct encode_info *i) -{ - i->sysno = 0; - i->seqno = 0; - i->cmd = -1; -} + status->processed = 0; + status->inserted = 0; + status->updated = 0; + status->deleted = 0; + status->utime = 0; + status->stime = 0; -char *encode_key_int (int d, char *bp) -{ - if (d <= 63) - *bp++ = d; - else if (d <= 16383) + if (!zh->res || !zh->reg) { - *bp++ = 64 + (d>>8); - *bp++ = d & 255; + zh->errCode = 2; + zh->errString = "zebra_end_trans: no open transaction"; + return -1; } - else if (d <= 4194303) + if (zh->trans_no != zh->trans_w_no) { - *bp++ = 128 + (d>>16); - *bp++ = (d>>8) & 255; - *bp++ = d & 255; + zh->trans_no--; + if (zh->trans_no != 0) + return 0; + + /* release read lock */ + + zebra_unlock (zh->lock_normal); + zebra_unlock (zh->lock_shadow); } else - { - *bp++ = 192 + (d>>24); - *bp++ = (d>>16) & 255; - *bp++ = (d>>8) & 255; - *bp++ = d & 255; + { /* release write lock */ + zh->trans_no--; + zh->trans_w_no = 0; + + yaz_log (LOG_LOG, "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 (LOG_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 = zh->records_processed; + status->inserted = zh->records_inserted; + status->updated = zh->records_updated; + status->deleted = zh->records_deleted; + + zebra_get_state (zh, &val, &seqno); + if (val != 'd') + { + BFiles bfs = bfs_create (rval, zh->path_reg); + yaz_log (LOG_LOG, "deleting shadow stuff 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); + } - return bp; +#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)); + + status->utime = (long) (zh->tms2.tms_utime - zh->tms1.tms_utime); + status->stime = (long) (zh->tms2.tms_stime - zh->tms1.tms_stime); +#endif + return 0; } -void encode_key_write (char *k, struct encode_info *i, FILE *outf) +int zebra_repository_update (ZebraHandle zh, const char *path) { - struct it_key key; - char *bp = i->buf; - - 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) - { - i->sysno = key.sysno; - i->seqno = 0; - } - 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) - { - logf (LOG_FATAL|LOG_ERRNO, "fwrite"); - exit (1); - } + ASSERTZH; + zh->errCode=0; + logf (LOG_LOG|LOG_API, "updating %s", path); + repositoryUpdate (zh, path); + return zh->errCode; } -static void extract_flushWriteKeys (ZebraHandle zh) +int zebra_repository_delete (ZebraHandle zh, const char *path) { - 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; + zh->errCode=0; + logf (LOG_LOG|LOG_API, "deleting %s", path); + repositoryDelete (zh, path); + return zh->errCode; +} - (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); +int zebra_repository_show (ZebraHandle zh, const char *path) +{ + ASSERTZH; + yaz_log(LOG_API,"zebra_repository_show"); + zh->errCode=0; + repositoryShow (zh, path); + return zh->errCode; +} - if (!(outf = fopen (out_fname, "wb"))) +static int zebra_commit_ex (ZebraHandle zh, int clean_only) +{ + int seqno; + char val; + const char *rval; + BFiles bfs; + ASSERTZH; + zh->errCode=0; + + if (!zh->res) { - logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname); - exit (1); + zh->errCode = 109; + return -1; } - 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) + rval = res_get (zh->res, "shadow"); + if (!rval) { - 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); + logf (LOG_WARN, "Cannot perform commit"); + logf (LOG_WARN, "No shadow area defined"); + return 0; } -#else - qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare); - extract_get_fname_tmp (out_fname, key_file_no); - if (!(outf = fopen (out_fname, "wb"))) + zebra_lock_w (zh->lock_normal); + zebra_lock_r (zh->lock_shadow); + + 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)) { - logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname); - exit (1); - } - 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])) + if (clean_only) + zebra_set_state (zh, 'd', seqno); + else { - 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); + zebra_set_state (zh, 'c', seqno); + + logf (LOG_LOG, "commit start"); + bf_commitExec (bfs); +#ifndef WIN32 + sync (); #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)) + logf (LOG_LOG, "commit clean"); + bf_commitClean (bfs, rval); + seqno++; + zebra_set_state (zh, 'o', seqno); + } + else { - logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname); - exit (1); + logf (LOG_LOG, "nothing to commit"); } - logf (LOG_LOG, "finished section %d", zh->key_file_no); - zh->ptr_i = 0; - zh->key_buf_used = 0; + bfs_destroy (bfs); + + zebra_unlock (zh->lock_shadow); + zebra_unlock (zh->lock_normal); + return 0; +} + +int zebra_clean (ZebraHandle zh) +{ + yaz_log(LOG_API,"zebra_clean"); + return zebra_commit_ex(zh, 1); +} + +int zebra_commit (ZebraHandle zh) +{ + yaz_log(LOG_API,"zebra_commit"); + return zebra_commit_ex(zh, 0); } -static void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno, - int cmd, struct recKeys *reckeys) +int zebra_init (ZebraHandle zh) { - unsigned char attrSet = (unsigned char) -1; - unsigned short attrUse = (unsigned short) -1; - int seqno = 0; - int off = 0; - ZebraExplainInfo zei = zh->service->zei; + const char *rval; + BFiles bfs = 0; + ASSERTZH; + yaz_log(LOG_API,"zebra_init"); + zh->errCode=0; - if (!zh->key_buf) + if (!zh->res) { - 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; + zh->errCode = 109; + return -1; } - zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1); - while (off < reckeys->buf_used) - { - const char *src = reckeys->buf + off; - struct it_key key; - int lead, ch; + rval = res_get (zh->res, "shadow"); + + bfs = bfs_create (res_get (zh->service->global_res, "register"), + zh->path_reg); + if (rval && *rval) + bf_cache (bfs, rval); - lead = *src++; + bf_reset (bfs); + bfs_destroy (bfs); + zebra_set_state (zh, 'o', 0); + return 0; +} - if (!(lead & 1)) - { - memcpy (&attrSet, src, sizeof(attrSet)); - src += sizeof(attrSet); - } - if (!(lead & 2)) - { - memcpy (&attrUse, src, sizeof(attrUse)); - src += sizeof(attrUse); - } - 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; - - ch = zebraExplain_lookupSU (zei, attrSet, attrUse); - if (ch < 0) - ch = zebraExplain_addSU (zei, attrSet, attrUse); - 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; +int zebra_compact (ZebraHandle zh) +{ + BFiles bfs; + ASSERTZH; + yaz_log(LOG_API,"zebra_compact"); + zh->errCode=0; + if (!zh->res) + { + zh->errCode = 109; + return -1; } - assert (off == reckeys->buf_used); + bfs = bfs_create (res_get (zh->res, "register"), zh->path_reg); + inv_compact (bfs); + bfs_destroy (bfs); + return 0; } -static void extract_index (ZebraHandle zh) +int zebra_result (ZebraHandle zh, int *code, char **addinfo) { - extract_flushWriteKeys (zh); - zebra_index_merge (zh); + ASSERTZH; + yaz_log(LOG_API,"zebra_result"); + *code = zh->errCode; + *addinfo = zh->errString; + return 0; } -static int explain_extract (void *handle, Record rec, data1_node *n) +int zebra_shadow_enable (ZebraHandle zh, int value) { - ZebraHandle zh = (ZebraHandle) handle; - struct recExtractCtrl extractCtrl; - int i; - - if (zebraExplain_curDatabase (zh->service->zei, - rec->info[recInfo_databaseName])) - { - abort(); - if (zebraExplain_newDatabase (zh->service->zei, - rec->info[recInfo_databaseName], 0)) - abort (); - } + ASSERTZH; + yaz_log(LOG_API,"zebra_shadow_enable"); + zh->errCode=0; + zh->shadow_enable = value; + return 0; +} - zh->keys.buf_used = 0; - zh->keys.prevAttrUse = -1; - zh->keys.prevAttrSet = -1; - zh->keys.prevSeqNo = 0; - zh->sortKeys = 0; +int zebra_record_encoding (ZebraHandle zh, const char *encoding) +{ + ASSERTZH; + yaz_log(LOG_API,"zebra_record_encoding"); + zh->errCode=0; + xfree (zh->record_encoding); + + /* + * Fixme! + * Something about charset aliases. Oleg??? + */ + + 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); - 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; + zh->record_encoding = xstrdup (encoding); - grs_extract_tree(&extractCtrl, n); + logf(LOG_DEBUG, "Reset record encoding: %s", encoding); + + zh->iconv_to_utf8 = + yaz_iconv_open ("UTF-8", encoding); + if (zh->iconv_to_utf8 == 0) + yaz_log (LOG_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 (LOG_WARN, "iconv: UTF-8 to %s unsupported", encoding); - logf (LOG_LOG, "flush explain record, sysno=%d", rec->sysno); + return 0; +} - if (rec->size[recInfo_delKeys]) - { - struct recKeys delkeys; - struct sortKey *sortKeys = 0; +int zebra_set_resource(ZebraHandle zh, const char *name, const char *value) +{ + ASSERTZH; + yaz_log(LOG_API,"zebra_set_resource %s:%s",name,value); + zh->errCode=0; + res_set(zh->res, name, value); + return 0; +} - 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); - - 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; +const char *zebra_get_resource(ZebraHandle zh, + const char *name, const char *defaultvalue) +{ + const char *v; + ASSERTZH; + v= res_get_def( zh->res, name, (char *)defaultvalue); + zh->errCode=0; + yaz_log(LOG_API,"zebra_get_resource %s:%s",name,v); + return v; +} + +/* moved from zebra_api_ext.c by pop */ +/* FIXME: Should this really be public??? -Heikki */ + +int zebra_trans_no (ZebraHandle zh) +{ + ASSERTZH; + yaz_log(LOG_API,"zebra_trans_no"); + return zh->trans_no; +} + +int zebra_get_shadow_enable (ZebraHandle zh) +{ + yaz_log(LOG_API,"zebra_get_shadow_enable"); + return (zh->shadow_enable); +} + +int zebra_set_shadow_enable (ZebraHandle zh, int value) +{ + yaz_log(LOG_API,"zebra_set_shadow_enable %d",value); + zh->shadow_enable = value; return 0; } -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; +/* almost the same as zebra_records_retrieve ... but how did it work? + I mean for multiple records ??? CHECK ??? */ +void api_records_retrieve (ZebraHandle zh, ODR stream, + const char *setname, Z_RecordComposition *comp, + oid_value input_format, int num_recs, + ZebraRetrievalRecord *recs) +{ + ZebraPosSet poset; + int i, *pos_array; + yaz_log(LOG_API,"api_records_retrieve s=%s n=%d",setname,num_recs); - /* 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))) + if (!zh->res) { - logf (LOG_WARN, "No such record type: %s", recordType); - return 0; + zh->errCode = 30; + zh->errString = odr_strdup (stream, setname); + return; } + + zh->errCode = 0; + + if (zebra_begin_read (zh)) + return; - zh->keys.buf_used = 0; - zh->keys.prevAttrUse = -1; - zh->keys.prevAttrSet = -1; - zh->keys.prevSeqNo = 0; - zh->sortKeys = 0; - - 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++) + pos_array = (int *) xmalloc (num_recs * sizeof(*pos_array)); + for (i = 0; iservice->zebra_maps, i)) - extractCtrl.seqno[i] = 1; - else - extractCtrl.seqno[i] = 0; + logf (LOG_DEBUG, "zebraPosSetCreate error"); + zh->errCode = 30; + zh->errString = nmem_strdup (stream->mem, setname); } - - r = (*recType->extract)(clientData, &extractCtrl); - - if (r == RECCTRL_EXTRACT_EOF) - return 0; - else if (r == RECCTRL_EXTRACT_ERROR) + else { - /* error occured during extraction ... */ -#if 1 - yaz_log (LOG_WARN, "extract error"); -#else - if (rGroup->flagRw && - records_processed < rGroup->fileVerboseLimit) + for (i = 0; irecordType, - fname, (long) recordOffset); + if (poset[i].term) + { + recs[i].errCode = 0; + recs[i].format = VAL_SUTRS; + recs[i].len = strlen(poset[i].term); + recs[i].buf = poset[i].term; + recs[i].base = poset[i].db; + recs[i].sysno = 0; + + } + else if (poset[i].sysno) + { + /* changed here ??? CHECK ??? */ + char *b; + recs[i].errCode = + zebra_record_fetch (zh, poset[i].sysno, poset[i].score, + stream, input_format, comp, + &recs[i].format, + &b, + &recs[i].len, + &recs[i].base); + recs[i].buf = (char *) odr_malloc(stream,recs[i].len); + memcpy(recs[i].buf, b, recs[i].len); + recs[i].errString = 0; /* Hmmm !!! we should get this */ + recs[i].sysno = poset[i].sysno; + recs[i].score = poset[i].score; + } + else + { + char num_str[20]; + + sprintf (num_str, "%d", pos_array[i]); + zh->errCode = 13; + zh->errString = odr_strdup (stream, num_str); + break; + } + } -#endif - return 0; + zebraPosSetDestroy (zh, poset, num_recs); } - 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"); + zebra_end_read (zh); + xfree (pos_array); +} + + +/* --------------------------------------------------------------------------- + Record insert(=update), delete + + 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 +*/ + +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); +} + +int 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) +{ + int res; + yaz_log(LOG_API,"zebra_insert_record sysno=" ZINT_FORMAT, *sysno); + + if (buf_size < 1) buf_size = strlen(buf); + + if (zebra_begin_trans(zh, 1)) return 1; - } - /* match criteria */ + 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); +int 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) +{ + int res; - *sysno = rec->sysno; + yaz_log(LOG_API,"zebra_update_record sysno=" ZINT_FORMAT, *sysno); - recordAttr = rec_init_attr (zh->service->zei, rec); + if (buf_size < 1) buf_size = strlen(buf); -#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); - } - else - { - /* record already exists */ - struct recKeys delkeys; + if (zebra_begin_trans(zh, 1)) + return 1; + 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; +} - rec = rec_get (zh->service->records, *sysno); - assert (rec); - - recordAttr = rec_init_attr (zh->service->zei, rec); +int 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) +{ + int res; + yaz_log(LOG_API,"zebra_delete_record sysno=" ZINT_FORMAT, *sysno); - 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]); - - /* 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; - } + if (buf_size < 1) buf_size = strlen(buf); - /* 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); + if (zebra_begin_trans(zh, 1)) + return 1; + 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; +} - /* set run-number for this record */ - recordAttr->runNumber = - zebraExplain_runNumberIncrement (zh->service->zei, 0); +/* --------------------------------------------------------------------------- + Searching +*/ - /* 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 - } +int zebra_search_PQF (ZebraHandle zh, const char *pqf_query, + const char *setname, int *numhits) +{ + int hits = 0; + int res=-1; + Z_RPNQuery *query; + ODR odr = odr_createmem(ODR_ENCODE); + + yaz_log(LOG_API,"zebra_search_PQF s=%s q=%s",setname, pqf_query); + + query = p_query_rpn (odr, PROTO_Z3950, pqf_query); + + if (!query) + yaz_log (LOG_WARN, "bad query %s\n", pqf_query); else + res=zebra_search_RPN (zh, odr, query, setname, &hits); + + odr_destroy(odr); + + yaz_log(LOG_API,"Hits: %d",hits); + + if (numhits) + *numhits=hits; + + return res; +} + +/* --------------------------------------------------------------------------- + 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 = yaz_sort_spec (stream, sort_spec); + yaz_log(LOG_API,"sort (FIXME) "); + if (!sort_sequence) { - rec->info[recInfo_storeData] = NULL; - rec->size[recInfo_storeData] = 0; + logf(LOG_WARN,"invalid sort specs '%s'", sort_spec); + zh->errCode = 207; + return -1; } - /* update database name */ - xfree (rec->info[recInfo_databaseName]); - rec->info[recInfo_databaseName] = - rec_strdup (databaseName, &rec->size[recInfo_databaseName]); - - /* update offset */ - recordAttr->recordOffset = recordOffset; - /* commit this record */ - rec_put (zh->service->records, &rec); + /* 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++; - return 0; + if (zebra_begin_read (zh)) + return -1; + + resultSetSort (zh, stream->mem, num_input_setnames, input_setnames, + output_setname, sort_sequence, &sort_status); + + zebra_end_read(zh); + return sort_status; } +