X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fkcompare.c;h=ec85430cd1bec09a6797e1b0ed6e7bb2c80ce51c;hb=3726bf6622da6a8b983bb4cbb7d654e84c3216d7;hp=b5b3b631bc8fbd824b20e7a2a703cc2fbacb2217;hpb=774c155f12ee6473f38ece849a401dd8426178d3;p=idzebra-moved-to-github.git diff --git a/index/kcompare.c b/index/kcompare.c index b5b3b63..ec85430 100644 --- a/index/kcompare.c +++ b/index/kcompare.c @@ -1,10 +1,73 @@ /* - * Copyright (C) 1994-1995, Index Data I/S + * Copyright (C) 1994-1999, Index Data * All rights reserved. * Sebastian Hammer, Adam Dickmeiss * * $Log: kcompare.c,v $ - * Revision 1.14 1995-10-30 15:08:08 adam + * Revision 1.33 1999-07-14 10:59:26 adam + * Changed functions isc_getmethod, isams_getmethod. + * Improved fatal error handling (such as missing EXPLAIN schema). + * + * Revision 1.32 1999/07/13 13:21:15 heikki + * Managing negative deltas + * + * Revision 1.31 1999/07/06 09:37:04 heikki + * Working on isamh - not ready yet. + * + * Revision 1.30 1999/06/30 15:07:23 heikki + * Adding isamh stuff + * + * Revision 1.29 1999/06/30 09:08:23 adam + * Added coder to reset. + * + * Revision 1.28 1999/05/26 07:49:13 adam + * C++ compilation. + * + * Revision 1.27 1999/05/12 13:08:06 adam + * First version of ISAMS. + * + * Revision 1.26 1999/02/02 14:50:54 adam + * Updated WIN32 code specific sections. Changed header. + * + * Revision 1.25 1998/06/08 15:26:06 adam + * Minor changes. + * + * Revision 1.24 1998/06/08 14:43:12 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 1.23 1998/03/05 08:45:12 adam + * New result set model and modular ranking system. Moved towards + * descent server API. System information stored as "SGML" records. + * + * Revision 1.22 1997/09/22 12:39:06 adam + * Added get_pos method for the ranked result sets. + * + * Revision 1.21 1997/09/17 12:19:13 adam + * Zebra version corresponds to YAZ version 1.4. + * Changed Zebra server so that it doesn't depend on global common_resource. + * + * Revision 1.20 1996/12/23 15:30:44 adam + * Work on truncation. + * Bug fix: result sets weren't deleted after server shut down. + * + * Revision 1.19 1996/12/11 12:08:00 adam + * Added better compression. + * + * Revision 1.18 1996/10/29 14:09:44 adam + * Use of cisam system - enabled if setting isamc is 1. + * + * Revision 1.17 1996/06/04 10:18:58 adam + * Minor changes - removed include of ctype.h. + * + * Revision 1.16 1996/05/13 14:23:05 adam + * Work on compaction of set/use bytes in dictionary. + * + * Revision 1.15 1995/11/20 16:59:46 adam + * New update method: the 'old' keys are saved for each records. + * + * Revision 1.14 1995/10/30 15:08:08 adam * Bug fixes. * * Revision 1.13 1995/10/27 14:00:11 adam @@ -58,7 +121,6 @@ #include #include #include -#include #include #include "index.h" @@ -71,6 +133,25 @@ void key_logdump (int logmask, const void *p) logf (logmask, "%7d s=%-4d", key.sysno, key.seqno); } +int key_compare_it (const void *p1, const void *p2) +{ + if (((struct it_key *) p1)->sysno != ((struct it_key *) p2)->sysno) + { + if (((struct it_key *) p1)->sysno > ((struct it_key *) p2)->sysno) + return 2; + else + return -2; + } + if (((struct it_key *) p1)->seqno != ((struct it_key *) p2)->seqno) + { + if (((struct it_key *) p1)->seqno > ((struct it_key *) p2)->seqno) + return 1; + else + return -1; + } + return 0; +} + int key_compare (const void *p1, const void *p2) { struct it_key i1, i2; @@ -83,7 +164,6 @@ int key_compare (const void *p1, const void *p2) else return -2; } -#if IT_KEY_HAVE_SEQNO if (i1.seqno != i2.seqno) { if (i1.seqno > i2.seqno) @@ -91,15 +171,6 @@ int key_compare (const void *p1, const void *p2) else return -1; } -#else - if (i1.freq != i2.freq) - { - if (i1.freq > i2.freq) - return 1; - else - return -1; - } -#endif return 0; } @@ -118,20 +189,191 @@ int key_qsort_compare (const void *p1, const void *p2) return cp1[l] - cp2[l]; } -int index_char_cvt (int c) +int key_get_pos (const void *p) +{ + struct it_key key; + memcpy (&key, p, sizeof(key)); + return key.seqno; +} + +struct iscz1_code_info { + struct it_key key; +}; + +static void *iscz1_code_start (int mode) { - return tolower (c); + struct iscz1_code_info *p = (struct iscz1_code_info *) + xmalloc (sizeof(*p)); + p->key.sysno = 0; + p->key.seqno = 0; + return p; } -int index_word_prefix (char *string, int attset_ordinal, - int local_attribute, - char *databaseName) +static void iscz1_code_reset (void *vp) +{ + struct iscz1_code_info *p = (struct iscz1_code_info *) vp; + p->key.sysno = 0; + p->key.seqno = 0; +} + +static void iscz1_code_stop (int mode, void *p) +{ + xfree (p); +} + +void iscz1_encode_int (unsigned d, char **dst) +{ + unsigned char *bp = (unsigned char*) *dst; + + if (d <= 63) + *bp++ = d; + else if (d <= 16383) + { + *bp++ = 64 | (d>>8); + *bp++ = d & 255; + } + else if (d <= 4194303) + { + *bp++ = 128 | (d>>16); + *bp++ = (d>>8) & 255; + *bp++ = d & 255; + } + else + { + *bp++ = 192 | (d>>24); + *bp++ = (d>>16) & 255; + *bp++ = (d>>8) & 255; + *bp++ = d & 255; + } + *dst = (char *) bp; +} + +int iscz1_decode_int (unsigned char **src) +{ + unsigned c = *(*src)++; + switch (c & 192) + { + case 0: + return c; + case 64: + return ((c & 63) << 8) + *(*src)++; + case 128: + c = ((c & 63) << 8) + *(*src)++; + c = (c << 8) + *(*src)++; + return c; + } + if (c&32) /* expand sign bit to high bits */ + c = ((c | 63) << 8) + *(*src)++; + else + c = ((c & 63) << 8) + *(*src)++; + c = (c << 8) + *(*src)++; + c = (c << 8) + *(*src)++; + + return c; +} + +static void iscz1_code_item (int mode, void *vp, char **dst, char **src) +{ + struct iscz1_code_info *p = (struct iscz1_code_info *) vp; + struct it_key tkey; + int d; + + if (mode == ISAMC_ENCODE) + { + memcpy (&tkey, *src, sizeof(struct it_key)); + d = tkey.sysno - p->key.sysno; + if (d) + { + iscz1_encode_int (2*tkey.seqno + 1, dst); + iscz1_encode_int (d, dst); + p->key.sysno += d; + p->key.seqno = tkey.seqno; + } + else + { + iscz1_encode_int (2*(tkey.seqno - p->key.seqno), dst); + p->key.seqno = tkey.seqno; + } + (*src) += sizeof(struct it_key); + } + else + { + d = iscz1_decode_int ((unsigned char **) src); + if (d & 1) + { + p->key.seqno = d>>1; + p->key.sysno += iscz1_decode_int ((unsigned char **) src); + } + else + p->key.seqno += d>>1; + memcpy (*dst, &p->key, sizeof(struct it_key)); + (*dst) += sizeof(struct it_key); + } +} + +ISAMC_M key_isamc_m (Res res, ISAMC_M me) +{ + isc_getmethod (me); + + me->compare_item = key_compare; + + me->code_start = iscz1_code_start; + me->code_item = iscz1_code_item; + me->code_stop = iscz1_code_stop; + me->code_reset = iscz1_code_reset; + + me->debug = atoi(res_get_def (res, "isamcDebug", "0")); + + return me; +} + +ISAMS_M key_isams_m (Res res, ISAMS_M me) +{ + isams_getmethod (me); + + me->compare_item = key_compare; + + me->code_start = iscz1_code_start; + me->code_item = iscz1_code_item; + me->code_stop = iscz1_code_stop; + + me->debug = atoi(res_get_def (res, "isamsDebug", "0")); + + return me; +} + +ISAMH_M key_isamh_m (Res res) +{ + static ISAMH_M me = NULL; + + if (me) + return me; + + me = isamh_getmethod (); + + me->compare_item = key_compare; + + me->code_start = iscz1_code_start; + me->code_item = iscz1_code_item; + me->code_stop = iscz1_code_stop; + me->code_reset = iscz1_code_reset; + + me->debug = atoi(res_get_def (res, "isamhDebug", "9")); + + return me; +} + + +int key_SU_code (int ch, char *out) { int i; - sprintf (string, "%s@%c%04d", databaseName, - attset_ordinal + '0', local_attribute); - for (i = 0; string[i]; i++) - string[i] = index_char_cvt (string[i]); + for (i = 0; ch; i++) + { + if (ch > 63) + out[i] = 128 + (ch & 63); + else + out[i] = 1 + ch; + ch = ch >> 6; + } return i; } -