X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fzebraapi.c;h=74fa36c048f717e3472efd10c583e05695bf0779;hb=eb8857cfa2ceaa2635f14888b95d6f71b8def045;hp=02893e9bfc30689247ee5345b3d6150fcae15e81;hpb=fbbce5a85247803288df5d97e5061889f118c826;p=idzebra-moved-to-github.git diff --git a/index/zebraapi.c b/index/zebraapi.c index 02893e9..74fa36c 100644 --- a/index/zebraapi.c +++ b/index/zebraapi.c @@ -1,5 +1,5 @@ -/* $Id: zebraapi.c,v 1.83 2003-02-12 15:45:59 heikki Exp $ - Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002 +/* $Id: zebraapi.c,v 1.90 2003-03-05 00:08:04 adam Exp $ + Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003 Index Data Aps This file is part of the Zebra server. @@ -33,8 +33,11 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA #endif #include +#include +#include #include "index.h" #include +#include "zebraapi.h" /* simple asserts to validate the most essential input args */ #define ASSERTZH assert(zh && zh->service) @@ -103,6 +106,7 @@ ZebraHandle zebra_open (ZebraService zs) zh->basenames = 0; zh->trans_no = 0; + zh->trans_w_no = 0; zh->lock_normal = 0; zh->lock_shadow = 0; @@ -219,6 +223,8 @@ struct zebra_register *zebra_register_open (ZebraService zs, const char *name, reg->keys.buf_max = 0; reg->keys.buf = 0; + reg->sortKeys.buf = 0; + reg->sortKeys.buf_max = 0; reg->records = 0; reg->dict = 0; @@ -234,6 +240,7 @@ struct zebra_register *zebra_register_open (ZebraService zs, const char *name, reg->key_file_no = 0; zebraRankInstall (reg, rank1_class); + zebraRankInstall (reg, rankzv_class); recordCompression = res_get_def (res, "recordCompression", "none"); if (!strcmp (recordCompression, "none")) @@ -398,10 +405,12 @@ static void zebra_register_close (ZebraService zs, struct zebra_register *reg) bfs_destroy (reg->bfs); data1_destroy (reg->dh); + xfree (reg->sortKeys.buf); + xfree (reg->keys.buf); + xfree (reg->key_buf); xfree (reg->name); xfree (reg); - yaz_log(LOG_DEBUG, "zebra_register_close 2"); } void zebra_stop(ZebraService zs) @@ -430,6 +439,7 @@ void zebra_close (ZebraHandle zh) { ZebraService zs; struct zebra_session **sp; + int i; if (!zh) return; @@ -442,13 +452,16 @@ void zebra_close (ZebraHandle zh) return ; resultSetDestroy (zh, -1, 0, 0); - 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) @@ -921,7 +934,7 @@ void zebra_admin_import_begin (ZebraHandle zh, const char *database, zh->errCode=0; if (zebra_select_database(zh, database)) return; - zebra_begin_trans (zh); + zebra_begin_trans (zh, 1); xfree (zh->admin_databaseName); zh->admin_databaseName = xstrdup(database); } @@ -1023,7 +1036,7 @@ void zebra_admin_create (ZebraHandle zh, const char *database) if (zebra_select_database (zh, database)) return; - zebra_begin_trans (zh); + zebra_begin_trans (zh, 1); zs = zh->service; /* announce database */ @@ -1101,232 +1114,255 @@ void zebra_get_state (ZebraHandle zh, char *val, int *seqno) int zebra_begin_read (ZebraHandle zh) { - int dirty = 0; - char val; - int seqno; - ASSERTZH; - - assert (zh->res); - - (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) - 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) - { - zh->errCode = 109; - return -1; - } - zh->reg->last_val = val; - zh->reg->seqno = seqno; - - return 0; + return zebra_begin_trans(zh, 0); } void zebra_end_read (ZebraHandle zh) { - ASSERTZH; - (zh->trans_no)--; - - if (zh->trans_no != 0) - return; - zh->errCode=0; - -#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 - - zebra_unlock (zh->lock_normal); - zebra_unlock (zh->lock_shadow); + zebra_end_trans(zh); } -void zebra_begin_trans (ZebraHandle zh) +int zebra_begin_trans (ZebraHandle zh, int rw) { - int pass; - int seqno = 0; - char val = '?'; - const char *rval = 0; ASSERTZHRES; - assert (zh->res); - - (zh->trans_no++); - if (zh->trans_no != 1) + if (rw) { - return; - } - zh->errCode=0; - - yaz_log (LOG_LOG, "zebra_begin_trans"); - - zh->records_inserted = 0; - zh->records_updated = 0; - zh->records_deleted = 0; - zh->records_processed = 0; + int pass; + int seqno = 0; + char val = '?'; + const char *rval = 0; + + (zh->trans_no++); + if (zh->trans_w_no) + return 0; + if (zh->trans_no != 1) + { + zh->errCode = 2; + zh->errString = "write trans not allowed within read trans"; + return -1; + } + zh->trans_w_no = zh->trans_no; + zh->errCode=0; + + yaz_log (LOG_LOG, "zebra_begin_trans"); + + zh->records_inserted = 0; + zh->records_updated = 0; + zh->records_deleted = 0; + zh->records_processed = 0; + #if HAVE_SYS_TIMES_H - times (&zh->tms1); + 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); - } + /* lock */ + if (zh->shadow_enable) + rval = res_get (zh->res, "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') + for (pass = 0; pass < 2; pass++) { 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); + zebra_lock_r (zh->lock_normal); + zebra_lock_w (zh->lock_shadow); } else { - yaz_log (LOG_WARN, "your previous transaction didn't finish"); + 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; } - break; + 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); + + zh->reg->seqno = seqno; } - if (pass == 2) + else { - yaz_log (LOG_FATAL, "zebra_begin_trans couldn't finish commit"); - abort(); - return; + 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) + 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) + { + zh->errCode = 109; + return -1; + } + zh->reg->last_val = val; + zh->reg->seqno = seqno; } - 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); + return 0; +} - zh->reg->seqno = seqno; +void zebra_end_trans (ZebraHandle zh) { + ZebraTransactionStatus dummy; + zebra_end_transaction(zh, &dummy); } -void zebra_end_trans (ZebraHandle zh) +void zebra_end_transaction (ZebraHandle zh, ZebraTransactionStatus *status) { char val; int seqno; const char *rval; - ASSERTZH; - - zh->trans_no--; - if (zh->trans_no != 0) - return; - zh->errCode=0; - - yaz_log (LOG_LOG, "zebra_end_trans"); - rval = res_get (zh->res, "shadow"); - zebraExplain_runNumberIncrement (zh->reg->zei, 1); - - zebra_flush_reg (zh); - - zebra_register_close (zh->service, zh->reg); - zh->reg = 0; + ASSERTZH; - yaz_log (LOG_LOG, "Records: %7d i/u/d %d/%d/%d", - zh->records_processed, zh->records_inserted, - zh->records_updated, zh->records_deleted); + status->processed = 0; + status->inserted = 0; + status->updated = 0; + status->deleted = 0; + status->utime = 0; + status->stime = 0; - zebra_get_state (zh, &val, &seqno); - if (val != 'd') + if (zh->trans_no != zh->trans_w_no) { - 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); + zh->trans_no--; + if (zh->trans_no != 0) + return; - zebra_unlock (zh->lock_shadow); - zebra_unlock (zh->lock_normal); + /* release read lock */ + zebra_unlock (zh->lock_normal); + zebra_unlock (zh->lock_shadow); + } + else + { /* 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); + + zebra_register_close (zh->service, zh->reg); + zh->reg = 0; + + yaz_log (LOG_LOG, "Records: %7d i/u/d %d/%d/%d", + 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); + + } #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)); - + (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 } @@ -1456,7 +1492,7 @@ int zebra_record_insert (ZebraHandle zh, const char *buf, int len) int olderr; ASSERTZH; zh->errCode=0; - zebra_begin_trans (zh); + zebra_begin_trans (zh, 1); if (zh->errCode) return 0; /* bad sysno */ extract_rec_in_mem (zh, "grs.sgml", @@ -1541,3 +1577,323 @@ const char *zebra_get_resource(ZebraHandle zh, zh->errCode=0; return res_get_def( zh->res, name, (char *)defaultvalue); } + +/* moved from zebra_api_ext.c by pop */ + +int zebra_trans_no (ZebraHandle zh) { + return (zh->trans_no); +} + + + +int zebra_get_shadow_enable (ZebraHandle zh) { + return (zh->shadow_enable); +} + +void zebra_set_shadow_enable (ZebraHandle zh, int value) { + zh->shadow_enable = value; +} + +void init_recordGroup (struct recordGroup *rg) { + rg->groupName = NULL; + rg->databaseName = NULL; + rg->path = NULL; + rg->recordId = NULL; + rg->recordType = NULL; + rg->flagStoreData = -1; + rg->flagStoreKeys = -1; + rg->flagRw = 1; + rg->databaseNamePath = 0; + rg->explainDatabase = 0; + rg->fileVerboseLimit = 100000; + rg->followLinks = -1; +} + + +/* This is from extract.c... it seems useful, when extract_rec_in mem is + called... and in general... Should be moved to somewhere else */ +void res_get_recordGroup (ZebraHandle zh, + struct recordGroup *rGroup, + const char *ext) { + char gprefix[128]; + char ext_res[128]; + + if (!rGroup->groupName || !*rGroup->groupName) + *gprefix = '\0'; + else + sprintf (gprefix, "%s.", rGroup->groupName); + + /* determine file type - depending on extension */ + if (!rGroup->recordType) { + sprintf (ext_res, "%srecordType.%s", gprefix, ext); + if (!(rGroup->recordType = res_get (zh->res, ext_res))) { + sprintf (ext_res, "%srecordType", gprefix); + rGroup->recordType = res_get (zh->res, ext_res); + } + } + /* determine match criteria */ + if (!rGroup->recordId) { + sprintf (ext_res, "%srecordId.%s", gprefix, ext); + if (!(rGroup->recordId = res_get (zh->res, ext_res))) { + sprintf (ext_res, "%srecordId", gprefix); + rGroup->recordId = res_get (zh->res, ext_res); + } + } + + /* determine database name */ + if (!rGroup->databaseName) { + sprintf (ext_res, "%sdatabase.%s", gprefix, ext); + if (!(rGroup->databaseName = res_get (zh->res, ext_res))) { + sprintf (ext_res, "%sdatabase", gprefix); + rGroup->databaseName = res_get (zh->res, ext_res); + } + } + if (!rGroup->databaseName) + rGroup->databaseName = "Default"; + + /* determine if explain database */ + sprintf (ext_res, "%sexplainDatabase", gprefix); + rGroup->explainDatabase = + atoi (res_get_def (zh->res, ext_res, "0")); + + /* storeData */ + if (rGroup->flagStoreData == -1) { + const char *sval; + sprintf (ext_res, "%sstoreData.%s", gprefix, ext); + if (!(sval = res_get (zh->res, ext_res))) { + sprintf (ext_res, "%sstoreData", gprefix); + sval = res_get (zh->res, ext_res); + } + if (sval) + rGroup->flagStoreData = atoi (sval); + } + if (rGroup->flagStoreData == -1) rGroup->flagStoreData = 0; + + /* storeKeys */ + if (rGroup->flagStoreKeys == -1) { + const char *sval; + + sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext); + sval = res_get (zh->res, ext_res); + if (!sval) { + sprintf (ext_res, "%sstoreKeys", gprefix); + sval = res_get (zh->res, ext_res); + } + if (!sval) sval = res_get (zh->res, "storeKeys"); + if (sval) rGroup->flagStoreKeys = atoi (sval); + } + if (rGroup->flagStoreKeys == -1) rGroup->flagStoreKeys = 0; + +} + + +/* 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; + + if (!zh->res) + { + zh->errCode = 30; + zh->errString = odr_strdup (stream, setname); + return; + } + + zh->errCode = 0; + + if (zebra_begin_read (zh)) + return; + + pos_array = (int *) xmalloc (num_recs * sizeof(*pos_array)); + for (i = 0; ierrCode = 30; + zh->errString = nmem_strdup (stream->mem, setname); + } + else + { + for (i = 0; ierrCode = 13; + zh->errString = odr_strdup (stream, num_str); + break; + } + + } + zebraPosSetDestroy (zh, poset, num_recs); + } + zebra_end_read (zh); + xfree (pos_array); +} + + +/* --------------------------------------------------------------------------- + Record insert(=update), delete + + If sysno is provided, then it's used to identify the reocord. + If not, and match_criteria is provided, then sysno is guessed + If not, and a record is provided, then sysno is got from there +*/ + +int zebra_update_record (ZebraHandle zh, + struct recordGroup *rGroup, + const char *recordType, + int sysno, const char *match, const char *fname, + const char *buf, int buf_size) + +{ + int res; + + if (buf_size < 1) buf_size = strlen(buf); + + zebra_begin_trans(zh, 1); + res=bufferExtractRecord (zh, buf, buf_size, rGroup, + 0, // delete_flag + 0, // test_mode, + recordType, + &sysno, + match, fname); + zebra_end_trans(zh); + return sysno; +} + +int zebra_delete_record (ZebraHandle zh, + struct recordGroup *rGroup, + const char *recordType, + int sysno, const char *match, const char *fname, + const char *buf, int buf_size) +{ + int res; + + if (buf_size < 1) buf_size = strlen(buf); + + zebra_begin_trans(zh, 1); + res=bufferExtractRecord (zh, buf, buf_size, rGroup, + 1, // delete_flag + 0, // test_mode, + recordType, + &sysno, + match,fname); + zebra_end_trans(zh); + return sysno; +} + +/* --------------------------------------------------------------------------- + Searching + + zebra_search_RPN is the same as zebra_search_rpn, except that read locking + is not mandatory. (it's repeatable now, also in zebraapi.c) +*/ + +void zebra_search_RPN (ZebraHandle zh, ODR decode, ODR stream, + Z_RPNQuery *query, const char *setname, int *hits) +{ + zh->hits = 0; + *hits = 0; + + if (zebra_begin_read (zh)) + return; + resultSetAddRPN (zh, decode, stream, query, + zh->num_basenames, zh->basenames, setname); + + zebra_end_read (zh); + + *hits = zh->hits; +} + +int zebra_search_PQF (ZebraHandle zh, + ODR odr_input, ODR odr_output, + const char *pqf_query, + const char *setname) + +{ + int hits; + Z_RPNQuery *query; + query = p_query_rpn (odr_input, PROTO_Z3950, pqf_query); + + if (!query) { + logf (LOG_WARN, "bad query %s\n", pqf_query); + odr_reset (odr_input); + return(0); + } + zebra_search_RPN (zh, odr_input, odr_output, query, setname, &hits); + + odr_reset (odr_input); + odr_reset (odr_output); + + return(hits); +} + +/* --------------------------------------------------------------------------- + Sort - a simplified interface, with optional read locks. +*/ +int sort (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); + if (!sort_sequence) { + logf(LOG_WARN,"invalid sort specs '%s'", sort_spec); + zh->errCode = 207; + return (-1); + } + + /* 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++; + + 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); +}