X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fzrpn.c;h=0f05b2dfe9894cf5d9495d34214f7d92be8d90c7;hb=ac66787b71dd244e895a262a82b2a4944f8fea6b;hp=812d78fb9e008957659fc21ee6b1a065019a4dd1;hpb=69da23537c6bb71ab948e079708bf8ea090de73f;p=idzebra-moved-to-github.git diff --git a/index/zrpn.c b/index/zrpn.c index 812d78f..0f05b2d 100644 --- a/index/zrpn.c +++ b/index/zrpn.c @@ -1,361 +1,9 @@ /* - * Copyright (C) 1995-2000, Index Data + * Copyright (C) 1995-2002, Index Data * All rights reserved. * Sebastian Hammer, Adam Dickmeiss * - * $Log: zrpn.c,v $ - * Revision 1.104 2000-04-05 09:49:35 adam - * On Unix, zebra/z'mbol uses automake. - * - * Revision 1.103 2000/03/20 19:08:36 adam - * Added remote record import using Z39.50 extended services and Segment - * Requests. - * - * Revision 1.102 2000/03/15 15:00:31 adam - * First work on threaded version. - * - * Revision 1.101 2000/03/02 14:35:03 adam - * Fixed proximity handling. - * - * Revision 1.100 1999/12/28 15:48:12 adam - * Minor Fix. - * - * Revision 1.99 1999/12/23 09:03:32 adam - * Changed behaviour of trunc=105 so that * is regular .* and ! is regular . - * - * Revision 1.98 1999/11/30 13:48:04 adam - * Improved installation. Updated for inclusion of YAZ header files. - * - * Revision 1.97 1999/10/14 14:33:50 adam - * Added truncation 5=106. - * - * Revision 1.96 1999/09/23 10:05:05 adam - * Implemented structure=105 searching. - * - * Revision 1.95 1999/09/07 07:19:21 adam - * Work on character mapping. Implemented replace rules. - * - * Revision 1.94 1999/07/20 13:59:18 adam - * Fixed bug that occurred when phrases had 0 hits. - * - * Revision 1.93 1999/06/17 14:38:40 adam - * Bug fix: Scan SEGV'ed when getting unknown use attribute. - * - * Revision 1.92 1999/05/26 07:49:13 adam - * C++ compilation. - * - * Revision 1.91 1999/02/02 14:51:13 adam - * Updated WIN32 code specific sections. Changed header. - * - * Revision 1.90 1998/11/16 16:03:43 adam - * Moved loggin utilities to Yaz. Was implemented in file zlogs.c. - * - * Revision 1.89 1998/11/16 10:11:55 adam - * Added addtional info for error 114 - unsupported use attribute. - * - * Revision 1.88 1998/10/18 07:54:52 adam - * Additional info added for diagnostics 114 (Unsupported use attribute) and - * 121 (Unsupported attribute set). - * - * Revision 1.87 1998/09/28 11:19:12 adam - * Fix for Compiled ASN.1. - * - * Revision 1.86 1998/09/22 10:48:20 adam - * Minor changes in search API. - * - * Revision 1.85 1998/09/22 10:03:43 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 1.84 1998/09/18 12:41:00 adam - * Fixed bug with numerical relations. - * - * Revision 1.83 1998/09/02 13:53:19 adam - * Extra parameter decode added to search routines to implement - * persistent queries. - * - * Revision 1.82 1998/06/26 11:16:40 quinn - * Added support (un-optimised) for left and left/right truncation - * - * Revision 1.81 1998/06/24 12:16:14 adam - * Support for relations on text operands. Open range support in - * DFA module (i.e. [-j], [g-]). - * - * Revision 1.80 1998/06/23 15:33:34 adam - * Added feature to specify sort criteria in query (type 7 specifies - * sort flags). - * - * Revision 1.79 1998/06/22 11:35:09 adam - * Minor changes. - * - * Revision 1.78 1998/06/08 14:43:17 adam - * Added suport for EXPLAIN Proxy servers - added settings databasePath - * and explainDatabase to facilitate this. Increased maximum number - * of databases and attributes in one register. - * - * Revision 1.77 1998/05/20 10:12:22 adam - * Implemented automatic EXPLAIN database maintenance. - * Modified Zebra to work with ASN.1 compiled version of YAZ. - * - * Revision 1.76 1998/04/02 14:35:29 adam - * First version of Zebra that works with compiled ASN.1. - * - * Revision 1.75 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. - * - * Revision 1.74 1998/02/10 12:03:06 adam - * Implemented Sort. - * - * Revision 1.73 1998/01/29 13:40:11 adam - * Better logging for scan service. - * - * Revision 1.72 1998/01/07 13:53:41 adam - * Queries using simple ranked operands returns right number of hits. - * - * Revision 1.71 1997/12/18 10:54:24 adam - * New method result set method rs_hits that returns the number of - * hits in result-set (if known). The ranked result set returns real - * number of hits but only when not combined with other operands. - * - * Revision 1.70 1997/10/31 12:34:43 adam - * Changed a few log statements. - * - * Revision 1.69 1997/10/29 12:05:02 adam - * Server produces diagnostic "Unsupported Attribute Set" when appropriate. - * - * Revision 1.68 1997/10/27 14:33:06 adam - * Moved towards generic character mapping depending on "structure" - * field in abstract syntax file. Fixed a few memory leaks. Fixed - * bug with negative integers when doing searches with relational - * operators. - * - * Revision 1.67 1997/09/29 09:06:10 adam - * Removed one static var in order to make this module thread safe. - * - * Revision 1.66 1997/09/25 14:58:03 adam - * Windows NT port. - * - * Revision 1.65 1997/09/22 12:39:06 adam - * Added get_pos method for the ranked result sets. - * - * Revision 1.64 1997/09/18 08:59:20 adam - * Extra generic handle for the character mapping routines. - * - * Revision 1.63 1997/09/17 12:19:18 adam - * Zebra version corresponds to YAZ version 1.4. - * Changed Zebra server so that it doesn't depend on global common_resource. - * - * Revision 1.62 1997/09/05 15:30:09 adam - * Changed prototype for chr_map_input - added const. - * Added support for C++, headers uses extern "C" for public definitions. - * - * Revision 1.61 1997/02/10 10:21:14 adam - * Bug fix: in search terms character (^) wasn't observed. - * - * Revision 1.60 1997/01/31 11:10:34 adam - * Bug fix: Leading and trailing white space weren't removed in scan tokens. - * - * Revision 1.59 1997/01/17 11:31:46 adam - * Bug fix: complete phrase search didn't work. - * - * Revision 1.58 1996/12/23 15:30:45 adam - * Work on truncation. - * Bug fix: result sets weren't deleted after server shut down. - * - * Revision 1.57 1996/11/11 13:38:02 adam - * Added proximity support in search. - * - * Revision 1.56 1996/11/08 11:10:32 adam - * Buffers used during file match got bigger. - * Compressed ISAM support everywhere. - * Bug fixes regarding masking characters in queries. - * Redesigned Regexp-2 queries. - * - * Revision 1.55 1996/11/04 14:07:44 adam - * Moved truncation code to trunc.c. - * - * Revision 1.54 1996/10/29 14:09:52 adam - * Use of cisam system - enabled if setting isamc is 1. - * - * Revision 1.53 1996/06/26 09:21:43 adam - * Bug fix: local attribute set wasn't obeyed in scan. - * - * Revision 1.52 1996/06/17 14:26:20 adam - * Function gen_regular_rel changed to handle negative numbers. - * - * Revision 1.51 1996/06/11 10:54:15 quinn - * Relevance work - * - * Revision 1.50 1996/06/07 08:51:53 adam - * Bug fix: Character mapping was broken (introducued by last revision). - * - * Revision 1.49 1996/06/04 10:18:11 adam - * Search/scan uses character mapping module. - * - * Revision 1.48 1996/05/28 15:15:01 adam - * Bug fix: Didn't handle unknown database correctly. - * - * Revision 1.47 1996/05/15 18:36:28 adam - * Function trans_term transforms unsearchable characters to blanks. - * - * Revision 1.46 1996/05/15 11:57:56 adam - * Fixed bug introduced by set/field mapping in search operations. - * - * Revision 1.45 1996/05/14 11:34:00 adam - * Scan support in multiple registers/databases. - * - * Revision 1.44 1996/05/14 06:16:44 adam - * Compact use/set bytes used in search service. - * - * Revision 1.43 1996/05/09 09:54:43 adam - * Server supports maps from one logical attributes to a list of physical - * attributes. - * The extraction process doesn't make space consuming 'any' keys. - * - * Revision 1.42 1996/05/09 07:28:56 quinn - * Work towards phrases and multiple registers - * - * Revision 1.41 1996/03/20 09:36:43 adam - * Function dict_lookup_grep got extra parameter, init_pos, which marks - * from which position in pattern approximate pattern matching should occur. - * Approximate pattern matching is used in relevance=re-2. - * - * Revision 1.40 1996/02/02 13:44:44 adam - * The public dictionary functions simply use char instead of Dict_char - * to represent search strings. Dict_char is used internally only. - * - * Revision 1.39 1996/01/03 16:22:13 quinn - * operator->roperator - * - * Revision 1.38 1995/12/11 09:12:55 adam - * The rec_get function returns NULL if record doesn't exist - will - * happen in the server if the result set records have been deleted since - * the creation of the set (i.e. the search). - * The server saves a result temporarily if it is 'volatile', i.e. the - * set is register dependent. - * - * Revision 1.37 1995/12/06 15:05:28 adam - * More verbose in count_set. - * - * Revision 1.36 1995/12/06 12:41:27 adam - * New command 'stat' for the index program. - * Filenames can be read from stdin by specifying '-'. - * Bug fix/enhancement of the transformation from terms to regular - * expressons in the search engine. - * - * Revision 1.35 1995/11/27 09:29:00 adam - * Bug fixes regarding conversion to regular expressions. - * - * Revision 1.34 1995/11/16 17:00:56 adam - * Better logging of rpn query. - * - * Revision 1.33 1995/11/01 13:58:28 quinn - * Moving data1 to yaz/retrieval - * - * Revision 1.32 1995/10/27 14:00:11 adam - * Implemented detection of database availability. - * - * Revision 1.31 1995/10/17 18:02:10 adam - * New feature: databases. Implemented as prefix to words in dictionary. - * - * Revision 1.30 1995/10/16 09:32:38 adam - * More work on relational op. - * - * Revision 1.29 1995/10/13 16:01:49 adam - * Work on relations. - * - * Revision 1.28 1995/10/13 12:26:43 adam - * Optimization of truncation. - * - * Revision 1.27 1995/10/12 17:07:22 adam - * Truncation works. - * - * Revision 1.26 1995/10/12 12:40:54 adam - * Bug fixes in rpn_prox. - * - * Revision 1.25 1995/10/10 13:59:24 adam - * Function rset_open changed its wflag parameter to general flags. - * - * Revision 1.24 1995/10/09 16:18:37 adam - * Function dict_lookup_grep got extra client data parameter. - * - * Revision 1.23 1995/10/06 16:33:37 adam - * Use attribute mappings. - * - * Revision 1.22 1995/10/06 15:07:39 adam - * Structure 'local-number' handled. - * - * Revision 1.21 1995/10/06 13:52:06 adam - * Bug fixes. Handler may abort further scanning. - * - * Revision 1.20 1995/10/06 11:06:33 adam - * Scan entries include 'occurrences' now. - * - * Revision 1.19 1995/10/06 10:43:56 adam - * Scan added. 'occurrences' in scan entries not set yet. - * - * Revision 1.18 1995/10/04 16:57:20 adam - * Key input and merge sort in one pass. - * - * Revision 1.17 1995/10/04 12:55:17 adam - * Bug fix in ranked search. Use=Any keys inserted. - * - * Revision 1.16 1995/10/02 16:24:40 adam - * Use attribute actually used in search requests. - * - * Revision 1.15 1995/10/02 15:18:52 adam - * New member in recRetrieveCtrl: diagnostic. - * - * Revision 1.14 1995/09/28 12:10:32 adam - * Bug fixes. Field prefix used in queries. - * - * Revision 1.13 1995/09/18 14:17:50 adam - * Minor changes. - * - * Revision 1.12 1995/09/15 14:45:21 adam - * Retrieve control. - * Work on truncation. - * - * Revision 1.11 1995/09/14 11:53:27 adam - * First work on regular expressions/truncations. - * - * Revision 1.10 1995/09/11 15:23:26 adam - * More work on relevance search. - * - * Revision 1.9 1995/09/11 13:09:35 adam - * More work on relevance feedback. - * - * Revision 1.8 1995/09/08 14:52:27 adam - * Minor changes. Dictionary is lower case now. - * - * Revision 1.7 1995/09/07 13:58:36 adam - * New parameter: result-set file descriptor (RSFD) to support multiple - * positions within the same result-set. - * Boolean operators: and, or, not implemented. - * Result-set references. - * - * Revision 1.6 1995/09/06 16:11:18 adam - * Option: only one word key per file. - * - * Revision 1.5 1995/09/06 10:33:04 adam - * More work on present. Some log messages removed. - * - * Revision 1.4 1995/09/05 15:28:40 adam - * More work on search engine. - * - * Revision 1.3 1995/09/04 15:20:22 adam - * Minor changes. - * - * Revision 1.2 1995/09/04 12:33:43 adam - * Various cleanup. YAZ util used instead. - * - * Revision 1.1 1995/09/04 09:10:40 adam - * More work on index add/del/update. - * Merge sort implemented. - * Initial work on z39 server. - * + * $Id: zrpn.c,v 1.109 2002-02-18 11:46:58 adam Exp $ */ #include #include @@ -399,7 +47,8 @@ typedef struct { Z_AttributesPlusTerm *zapt; } AttrType; -static int attr_find (AttrType *src, oid_value *attributeSetP) +static int attr_find_ex (AttrType *src, oid_value *attributeSetP, + const char **string_value) { int num_attributes; @@ -433,19 +82,34 @@ static int attr_find (AttrType *src, oid_value *attributeSetP) return *element->value.numeric; break; case Z_AttributeValue_complex: - if (src->minor >= element->value.complex->num_list || - element->value.complex->list[src->minor]->which != + if (src->minor >= element->value.complex->num_list) + break; + if (element->value.complex->list[src->minor]->which == Z_StringOrNumeric_numeric) - break; - ++(src->minor); - if (element->attributeSet && attributeSetP) - { - oident *attrset; - - attrset = oid_getentbyoid (element->attributeSet); - *attributeSetP = attrset->value; - } - return *element->value.complex->list[src->minor-1]->u.numeric; + { + ++(src->minor); + if (element->attributeSet && attributeSetP) + { + oident *attrset; + + attrset = oid_getentbyoid (element->attributeSet); + *attributeSetP = attrset->value; + } + return + *element->value.complex->list[src->minor-1]->u.numeric; + } + else if (element->value.complex->list[src->minor]->which == + Z_StringOrNumeric_string) + { + if (!string_value) + break; + ++(src->minor); + *string_value = + element->value.complex->list[src->minor-1]->u.string; + return -2; + } + else + break; default: assert (0); } @@ -455,6 +119,11 @@ static int attr_find (AttrType *src, oid_value *attributeSetP) return -1; } +static int attr_find (AttrType *src, oid_value *attributeSetP) +{ + return attr_find_ex (src, attributeSetP, 0); +} + static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt, int type) { @@ -475,6 +144,7 @@ struct grep_info { int isam_p_indx; ZebraHandle zh; int reg_type; + ZebraSet termset; }; static void term_untrans (ZebraHandle zh, int reg_type, @@ -528,9 +198,24 @@ static void add_isam_p (const char *name, const char *info, assert (*info == sizeof(*p->isam_p_buf)); memcpy (p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf)); -#if 0 - term_untrans (p->zh, p->reg_type, term_tmp, name+2); - logf (LOG_DEBUG, "grep: %s", term_tmp); +#if 1 + if (p->termset) + { + const char *db; + int set, use; + char term_tmp[512]; + int su_code = 0; + int len = key_SU_decode (&su_code, name); + + term_untrans (p->zh, p->reg_type, term_tmp, name+len+1); + logf (LOG_LOG, "grep: %d %c %s", su_code, name[len], term_tmp); + zebraExplain_lookup_ord (p->zh->service->zei, + su_code, &db, &set, &use); + logf (LOG_LOG, "grep: set=%d use=%d db=%s", set, use, db); + + resultSetAddTerm (p->zh, p->termset, name[len], db, + set, use, term_tmp); + } #endif (p->isam_p_indx)++; } @@ -574,6 +259,9 @@ static int term_100 (ZebraMaps zebra_maps, int reg_type, int i = 0; int j = 0; + const char *space_start = 0; + const char *space_end = 0; + if (!term_pre (zebra_maps, reg_type, src, NULL, NULL)) return 0; s0 = *src; @@ -581,8 +269,33 @@ static int term_100 (ZebraMaps zebra_maps, int reg_type, { s1 = s0; map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0)); - if (space_split && **map == *CHR_SPACE) - break; + if (space_split) + { + if (**map == *CHR_SPACE) + break; + } + else /* complete subfield only. */ + { + if (**map == *CHR_SPACE) + { /* save space mapping for later .. */ + space_start = s1; + space_end = s0; + continue; + } + else if (space_start) + { /* reload last space */ + while (space_start < space_end) + { + if (!isalnum (*space_start) && *space_start != '-') + dst[i++] = '\\'; + dst_term[j++] = *space_start; + dst[i++] = *space_start++; + } + /* and reset */ + space_start = space_end = 0; + } + } + /* add non-space char */ while (s1 < s0) { if (!isalnum (*s1) && *s1 != '-') @@ -1168,7 +881,7 @@ static int string_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, else term_dict[prefix_len++] = '('; - ord_len = key_SU_code (ord, ord_buf); + ord_len = key_SU_encode (ord, ord_buf); for (i = 0; iterm_no = 0; +#endif + grep_info->isam_p_size = 0; + grep_info->isam_p_buf = NULL; + grep_info->zh = zh; + grep_info->reg_type = reg_type; + grep_info->termset = 0; + + attr_init (&termset, zapt, 8); + termset_value_numeric = + attr_find_ex (&termset, NULL, &termset_value_string); + if (termset_value_numeric != -1) + { + char resname[32]; + const char *termset_name = 0; + if (termset_value_numeric != -2) + { + + sprintf (resname, "%d", termset_value_numeric); + termset_name = resname; + } + else + termset_name = termset_value_string; + logf (LOG_LOG, "creating termset set %s", termset_name); + grep_info->termset = resultSetAdd (zh, termset_name, 1); + if (!grep_info->termset) + { + zh->errCode = 128; + zh->errString = nmem_strdup (stream, termset_name); + return -1; + } + } + return 0; +} + + static RSET rpn_search_APT_phrase (ZebraHandle zh, Z_AttributesPlusTerm *zapt, const char *termz_org, @@ -1635,14 +1395,8 @@ static RSET rpn_search_APT_phrase (ZebraHandle zh, char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type); const char *termp = termz; -#ifdef TERM_COUNT - grep_info.term_no = 0; -#endif - grep_info.isam_p_size = 0; - grep_info.isam_p_buf = NULL; - grep_info.zh = zh; - grep_info.reg_type = reg_type; - + if (grep_info_prepare (zh, zapt, &grep_info, reg_type, stream)) + return 0; while (1) { logf (LOG_DEBUG, "APT_phrase termp=%s", termp); @@ -1694,14 +1448,9 @@ static RSET rpn_search_APT_or_list (ZebraHandle zh, struct grep_info grep_info; char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type); const char *termp = termz; -#ifdef TERM_COUNT - grep_info.term_no = 0; -#endif - grep_info.isam_p_size = 0; - grep_info.isam_p_buf = NULL; - grep_info.zh = zh; - grep_info.reg_type = reg_type; + if (grep_info_prepare (zh, zapt, &grep_info, reg_type, stream)) + return 0; while (1) { logf (LOG_DEBUG, "APT_or_list termp=%s", termp); @@ -1760,14 +1509,8 @@ static RSET rpn_search_APT_and_list (ZebraHandle zh, char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type); const char *termp = termz; -#ifdef TERM_COUNT - grep_info.term_no = 0; -#endif - grep_info.isam_p_size = 0; - grep_info.isam_p_buf = NULL; - grep_info.zh = zh; - grep_info.reg_type = reg_type; - + if (grep_info_prepare (zh, zapt, &grep_info, reg_type, stream)) + return 0; while (1) { logf (LOG_DEBUG, "APT_and_list termp=%s", termp); @@ -1928,7 +1671,7 @@ static int numeric_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt, else term_dict[prefix_len++] = '('; - ord_len = key_SU_code (ord, ord_buf); + ord_len = key_SU_encode (ord, ord_buf); for (i = 0; iservice->zebra_maps, zapt, ®_id, &search_type, - &rank_type, &complete_flag, &sort_flag); + rank_type, &complete_flag, &sort_flag); logf (LOG_DEBUG, "reg_id=%c", reg_id); logf (LOG_DEBUG, "complete_flag=%d", complete_flag); @@ -2320,18 +2057,24 @@ static RSET rpn_search_structure (ZebraHandle zh, Z_RPNStructure *zs, logf (LOG_DEBUG, "rpn_search_ref"); r = resultSetRef (zh, zs->u.simple->u.resultSetId); if (!r) + { r = rset_create (rset_kind_null, NULL); + zh->errCode = 30; + zh->errString = + nmem_strdup (stream, zs->u.simple->u.resultSetId); + return 0; + } } else { zh->errCode = 3; - return NULL; + return 0; } } else { zh->errCode = 3; - return NULL; + return 0; } return r; } @@ -2478,7 +2221,7 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, unsigned reg_id; char *search_type = NULL; - char *rank_type = NULL; + char rank_type[128]; int complete_flag; int sort_flag; *list = 0; @@ -2486,17 +2229,20 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, if (attributeset == VAL_NONE) attributeset = VAL_BIB1; - logf (LOG_DEBUG, "position = %d, num = %d", pos, num); + yaz_log (LOG_DEBUG, "position = %d, num = %d set=%d", + pos, num, attributeset); attr_init (&use, zapt, 1); use_value = attr_find (&use, &attributeset); if (zebra_maps_attr (zh->service->zebra_maps, zapt, ®_id, &search_type, - &rank_type, &complete_flag, &sort_flag)) + rank_type, &complete_flag, &sort_flag)) { + *num_entries = 0; zh->errCode = 113; return ; } + yaz_log (LOG_DEBUG, "use_value = %d", use_value); if (use_value == -1) use_value = 1016; @@ -2521,6 +2267,7 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, { zh->errString = basenames[base_no]; zh->errCode = 109; /* Database unavailable */ + *num_entries = 0; return; } for (local_attr = attp.local_attributes; local_attr && ord_no < 32; @@ -2536,6 +2283,7 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, } if (ord_no == 0) { + *num_entries = 0; zh->errCode = 113; return; } @@ -2562,7 +2310,7 @@ void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt, for (j = 0; jlist[j].term = NULL; - prefix_len += key_SU_code (ords[i], termz + prefix_len); + prefix_len += key_SU_encode (ords[i], termz + prefix_len); termz[prefix_len++] = reg_id; termz[prefix_len] = 0; strcpy (scan_info->prefix, termz);