X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=dict%2Flookgrep.c;h=bde37714457bb4346517134ce43492c00a81a2c1;hb=5adb31268b057741850de38eafd42baf8dc947ea;hp=b464e9503cd531e3c85ecdf11ee5e03398baa9b2;hpb=a15e6ea036010b9e0f9be59e81461c8a4e894db4;p=idzebra-moved-to-github.git diff --git a/dict/lookgrep.c b/dict/lookgrep.c index b464e95..bde3771 100644 --- a/dict/lookgrep.c +++ b/dict/lookgrep.c @@ -1,10 +1,85 @@ /* - * Copyright (C) 1994, Index Data I/S + * Copyright (C) 1994-1999, Index Data * All rights reserved. * Sebastian Hammer, Adam Dickmeiss * * $Log: lookgrep.c,v $ - * Revision 1.2 1994-10-04 12:08:07 adam + * Revision 1.24 1999-05-26 07:49:12 adam + * C++ compilation. + * + * Revision 1.23 1999/05/15 14:36:37 adam + * Updated dictionary. Implemented "compression" of dictionary. + * + * Revision 1.22 1999/02/02 14:50:23 adam + * Updated WIN32 code specific sections. Changed header. + * + * Revision 1.21 1998/06/24 12:16:12 adam + * Support for relations on text operands. Open range support in + * DFA module (i.e. [-j], [g-]). + * + * Revision 1.20 1997/10/27 14:33:03 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 1.19 1997/09/18 08:59:18 adam + * Extra generic handle for the character mapping routines. + * + * Revision 1.18 1997/09/05 15:29:58 adam + * Changed prototype for chr_map_input - added const. + * Added support for C++, headers uses extern "C" for public definitions. + * + * Revision 1.17 1996/06/04 10:20:06 adam + * Added support for character mapping. + * + * Revision 1.16 1996/05/24 14:46:04 adam + * Added dict_grep_cmap function to define user-mapping in grep lookups. + * + * Revision 1.15 1996/03/20 09:35:18 adam + * Function dict_lookup_grep got extra parameter, init_pos, which marks + * from which position in pattern approximate pattern matching should occur. + * + * Revision 1.14 1996/02/02 13:43:51 adam + * The public functions simply use char instead of Dict_char to represent + * search strings. Dict_char is used internally only. + * + * Revision 1.13 1996/01/08 09:09:30 adam + * Function dfa_parse got 'const' string argument. + * + * Revision 1.12 1995/12/11 09:04:48 adam + * Bug fix: the lookup/scan/lookgrep didn't handle empty dictionary. + * + * Revision 1.11 1995/12/06 14:43:02 adam + * New function: dict_delete. + * + * Revision 1.10 1995/11/16 17:00:44 adam + * Changed stupid log. + * + * Revision 1.9 1995/10/27 13:58:09 adam + * Makes 'Database unavailable' diagnostic. + * + * Revision 1.8 1995/10/19 14:57:21 adam + * New feature: grep lookup saves length of longest prefix match. + * + * Revision 1.7 1995/10/17 18:01:22 adam + * Userfunc may return non-zero in which case the the grepping stops + * immediately. + * + * Revision 1.6 1995/10/09 16:18:32 adam + * Function dict_lookup_grep got extra client data parameter. + * + * Revision 1.5 1995/09/14 11:52:59 adam + * Grep handle function parameter info is const now. + * + * Revision 1.4 1995/01/24 16:01:02 adam + * Added -ansi to CFLAGS. + * Use new API of dfa module. + * + * Revision 1.3 1994/10/05 12:16:50 adam + * Pagesize is a resource now. + * + * Revision 1.2 1994/10/04 12:08:07 adam * Some bug fixes and some optimizations. * * Revision 1.1 1994/10/03 17:23:04 adam @@ -50,18 +125,18 @@ static INLINE MatchWord get_bit (MatchContext *mc, MatchWord *m, int ch, return m[mc->n * ch + wno] & (1<n = (dfas->no+WORD_BITS) / WORD_BITS; + mc->n = (dfa->no_states+WORD_BITS) / WORD_BITS; mc->range = range; mc->fact = (range+1)*mc->n; - mc->match_mask = xcalloc (mc->n, sizeof(*mc->match_mask)); + mc->match_mask = (MatchWord *) xcalloc (mc->n, sizeof(*mc->match_mask)); - for (s = 0; sno; s++) - if (dfas->sortarray[s]->rule_no) + for (s = 0; sno_states; s++) + if (dfa->states[s]->rule_no) set_bit (mc, mc->match_mask, 0, s); return mc; } @@ -74,7 +149,7 @@ static void rm_MatchContext (MatchContext **mc) } static void mask_shift (MatchContext *mc, MatchWord *Rdst, MatchWord *Rsrc, - DFA_states *dfas, int ch) + struct DFA *dfa, int ch) { int j, s = 0; MatchWord *Rsrc_p = Rsrc, mask; @@ -90,7 +165,7 @@ static void mask_shift (MatchContext *mc, MatchWord *Rdst, MatchWord *Rsrc, { if (mask & 1) { - DFA_state *state = dfas->sortarray[s]; + struct DFA_state *state = dfa->states[s]; int i = state->tran_no; while (--i >= 0) if (ch >= state->trans[i].ch[0] && @@ -99,7 +174,7 @@ static void mask_shift (MatchContext *mc, MatchWord *Rdst, MatchWord *Rsrc, } if (mask & 2) { - DFA_state *state = dfas->sortarray[s+1]; + struct DFA_state *state = dfa->states[s+1]; int i = state->tran_no; while (--i >= 0) if (ch >= state->trans[i].ch[0] && @@ -108,7 +183,7 @@ static void mask_shift (MatchContext *mc, MatchWord *Rdst, MatchWord *Rsrc, } if (mask & 4) { - DFA_state *state = dfas->sortarray[s+2]; + struct DFA_state *state = dfa->states[s+2]; int i = state->tran_no; while (--i >= 0) if (ch >= state->trans[i].ch[0] && @@ -117,7 +192,7 @@ static void mask_shift (MatchContext *mc, MatchWord *Rdst, MatchWord *Rsrc, } if (mask & 8) { - DFA_state *state = dfas->sortarray[s+3]; + struct DFA_state *state = dfa->states[s+3]; int i = state->tran_no; while (--i >= 0) if (ch >= state->trans[i].ch[0] && @@ -126,7 +201,7 @@ static void mask_shift (MatchContext *mc, MatchWord *Rdst, MatchWord *Rsrc, } } s += 4; - if (s >= dfas->no) + if (s >= dfa->no_states) return; mask >>= 4; } @@ -134,7 +209,7 @@ static void mask_shift (MatchContext *mc, MatchWord *Rdst, MatchWord *Rsrc, } static void shift (MatchContext *mc, MatchWord *Rdst, MatchWord *Rsrc, - DFA_states *dfas) + struct DFA *dfa) { int j, s = 0; MatchWord *Rsrc_p = Rsrc, mask; @@ -149,35 +224,35 @@ static void shift (MatchContext *mc, MatchWord *Rdst, MatchWord *Rsrc, { if (mask & 1) { - DFA_state *state = dfas->sortarray[s]; + struct DFA_state *state = dfa->states[s]; int i = state->tran_no; while (--i >= 0) set_bit (mc, Rdst, 0, state->trans[i].to); } if (mask & 2) { - DFA_state *state = dfas->sortarray[s+1]; + struct DFA_state *state = dfa->states[s+1]; int i = state->tran_no; while (--i >= 0) set_bit (mc, Rdst, 0, state->trans[i].to); } if (mask & 4) { - DFA_state *state = dfas->sortarray[s+2]; + struct DFA_state *state = dfa->states[s+2]; int i = state->tran_no; while (--i >= 0) set_bit (mc, Rdst, 0, state->trans[i].to); } if (mask & 8) { - DFA_state *state = dfas->sortarray[s+3]; + struct DFA_state *state = dfa->states[s+3]; int i = state->tran_no; while (--i >= 0) set_bit (mc, Rdst, 0, state->trans[i].to); } } s += 4; - if (s >= dfas->no) + if (s >= dfa->no_states) return; mask >>= 4; } @@ -193,19 +268,20 @@ static void or (MatchContext *mc, MatchWord *Rdst, } static INLINE int move (MatchContext *mc, MatchWord *Rj1, MatchWord *Rj, - Dict_char ch, DFA_states *dfas, MatchWord *Rtmp) + Dict_char ch, struct DFA *dfa, MatchWord *Rtmp, + int range) { int d; MatchWord *Rtmp_2 = Rtmp + mc->n; - mask_shift (mc, Rj1, Rj, dfas, ch); + mask_shift (mc, Rj1, Rj, dfa, ch); for (d = 1; d <= mc->range; d++) { or (mc, Rtmp, Rj, Rj1); /* 2,3 */ - shift (mc, Rtmp_2, Rtmp, dfas); + shift (mc, Rtmp_2, Rtmp, dfa); - mask_shift (mc, Rtmp, Rj+mc->n, dfas, ch); /* 1 */ + mask_shift (mc, Rtmp, Rj+mc->n, dfa, ch); /* 1 */ or (mc, Rtmp, Rtmp_2, Rtmp); /* 1,2,3*/ @@ -221,9 +297,10 @@ static INLINE int move (MatchContext *mc, MatchWord *Rj1, MatchWord *Rj, static int dict_grep (Dict dict, Dict_ptr ptr, MatchContext *mc, - MatchWord *Rj, int pos, - int (*userfunc)(Dict_char *name, char *info), - Dict_char *prefix, DFA_states *dfas) + MatchWord *Rj, int pos, void *client, + int (*userfunc)(char *, const char *, void *), + Dict_char *prefix, struct DFA *dfa, + int *max_pos, int init_pos) { int lo, hi, d; void *p; @@ -233,7 +310,7 @@ static int dict_grep (Dict dict, Dict_ptr ptr, MatchContext *mc, dict_bf_readp (dict->dbf, ptr, &p); lo = 0; hi = DICT_nodir(p)-1; - indxp = (short*) ((char*) p+DICT_PAGESIZE-sizeof(short)); + indxp = (short*) ((char*) p+DICT_bsize(p)-sizeof(short)); while (lo <= hi) { @@ -251,24 +328,33 @@ static int dict_grep (Dict dict, Dict_ptr ptr, MatchContext *mc, MatchWord *Rj0 = Rj + j *mc->fact; MatchWord *Rj1 = Rj + (j+1)*mc->fact; MatchWord *Rj_tmp = Rj + (j+2)*mc->fact; + int range; memcpy (&ch, info+j*sizeof(Dict_char), sizeof(Dict_char)); prefix[pos+j] = ch; + if (pos+j > *max_pos) + *max_pos = pos+j; if (ch == DICT_EOS) { if (was_match) - (*userfunc)(prefix, info+(j+1)*sizeof(Dict_char)); + if ((*userfunc)((char*) prefix, + info+(j+1)*sizeof(Dict_char), client)) + return 1; break; } - move (mc, Rj1, Rj0, ch, dfas, Rj_tmp); + if (pos+j >= init_pos) + range = mc->range; + else + range = 0; + move (mc, Rj1, Rj0, ch, dfa, Rj_tmp, range); for (d = mc->n; --d >= 0; ) - if (Rj1[mc->range*mc->n + d]) + if (Rj1[range*mc->n + d]) break; if (d < 0) break; was_match = 0; for (d = mc->n; --d >= 0; ) - if (Rj1[mc->range*mc->n + d] & mc->match_mask[d]) + if (Rj1[range*mc->n + d] & mc->match_mask[d]) { was_match = 1; break; @@ -280,6 +366,7 @@ static int dict_grep (Dict dict, Dict_ptr ptr, MatchContext *mc, MatchWord *Rj1 = Rj+ mc->fact; MatchWord *Rj_tmp = Rj+2*mc->fact; Dict_char ch; + int range; /* Dict_ptr subptr */ /* Dict_char sub char */ @@ -289,9 +376,15 @@ static int dict_grep (Dict dict, Dict_ptr ptr, MatchContext *mc, memcpy (&ch, info+sizeof(Dict_ptr), sizeof(Dict_char)); prefix[pos] = ch; - move (mc, Rj1, Rj, ch, dfas, Rj_tmp); + if (pos > *max_pos) + *max_pos = pos; + if (pos >= init_pos) + range = mc->range; + else + range = 0; + move (mc, Rj1, Rj, ch, dfa, Rj_tmp, range); for (d = mc->n; --d >= 0; ) - if (Rj1[mc->range*mc->n + d]) + if (Rj1[range*mc->n + d]) break; if (d >= 0) { @@ -299,21 +392,25 @@ static int dict_grep (Dict dict, Dict_ptr ptr, MatchContext *mc, if (info[sizeof(Dict_ptr)+sizeof(Dict_char)]) { for (d = mc->n; --d >= 0; ) - if (Rj1[mc->range*mc->n + d] & mc->match_mask[d]) + if (Rj1[range*mc->n + d] & mc->match_mask[d]) { prefix[pos+1] = DICT_EOS; - (*userfunc)(prefix, info+sizeof(Dict_ptr)+ - sizeof(Dict_char)); + if ((*userfunc)((char*) prefix, + info+sizeof(Dict_ptr)+ + sizeof(Dict_char), client)) + return 1; break; } } memcpy (&subptr, info, sizeof(Dict_ptr)); if (subptr) { - dict_grep (dict, subptr, mc, Rj1, pos+1, - userfunc, prefix, dfas); + if (dict_grep (dict, subptr, mc, Rj1, pos+1, + client, userfunc, prefix, dfa, max_pos, + init_pos)) + return 1; dict_bf_readp (dict->dbf, ptr, &p); - indxp = (short*) ((char*) p+DICT_PAGESIZE-sizeof(short)); + indxp = (short*) ((char*) p+DICT_bsize(p)-sizeof(short)); } } } @@ -322,51 +419,79 @@ static int dict_grep (Dict dict, Dict_ptr ptr, MatchContext *mc, return 0; } -int dict_lookup_grep (Dict dict, Dict_char *pattern, int range, - int (*userfunc)(Dict_char *name, char *info)) +int dict_lookup_grep (Dict dict, const char *pattern, int range, void *client, + int *max_pos, int init_pos, + int (*userfunc)(char *name, const char *info, + void *client)) { MatchWord *Rj; Dict_char prefix[MAX_LENGTH+1]; - char *this_pattern = pattern; - DFA_states *dfas; + const char *this_pattern = pattern; MatchContext *mc; - DFA *dfa = init_dfa(); + struct DFA *dfa = dfa_init(); int i, d; - i = parse_dfa (dfa, &this_pattern, dfa_thompson_chars); +#if 0 + debug_dfa_trav = 1; + debug_dfa_tran = 1; + debug_dfa_followpos = 1; + dfa_verbose = 1; +#endif + + logf (LOG_DEBUG, "dict_lookup_grep range=%d", range); + for (i = 0; pattern[i]; i++) + { + logf (LOG_DEBUG, " %3d %c", pattern[i], + (pattern[i] > ' ' && pattern[i] < 127) ? pattern[i] : '?'); + } + + dfa_set_cmap (dfa, dict->grep_cmap_data, dict->grep_cmap); + + i = dfa_parse (dfa, &this_pattern); if (i || *this_pattern) { - rm_dfa (&dfa); + dfa_delete (&dfa); return -1; } - dfa->root = dfa->top; - dfas = mk_dfas (dfa, 50); - rm_dfa (&dfa); + dfa_mkstate (dfa); - mc = mk_MatchContext (dfas, range); + mc = mk_MatchContext (dfa, range); - Rj = xcalloc ((MAX_LENGTH+1) * mc->n, sizeof(*Rj)); + Rj = (MatchWord *) xcalloc ((MAX_LENGTH+1) * mc->n, sizeof(*Rj)); set_bit (mc, Rj, 0, 0); for (d = 1; d<=mc->range; d++) { int s; memcpy (Rj + mc->n * d, Rj + mc->n * (d-1), mc->n * sizeof(*Rj)); - for (s = 0; sno; s++) + for (s = 0; sno_states; s++) { if (get_bit (mc, Rj, d-1, s)) { - DFA_state *state = dfas->sortarray[s]; + struct DFA_state *state = dfa->states[s]; int i = state->tran_no; while (--i >= 0) set_bit (mc, Rj, d, state->trans[i].to); } } } - i = dict_grep (dict, 1, mc, Rj, 0, userfunc, prefix, dfas); - - rm_dfas (&dfas); + *max_pos = 0; + if (dict->head.root) + i = dict_grep (dict, dict->head.root, mc, Rj, 0, client, + userfunc, prefix, + dfa, max_pos, init_pos); + else + i = 0; + logf (LOG_DEBUG, "max_pos = %d", *max_pos); + dfa_delete (&dfa); xfree (Rj); rm_MatchContext (&mc); return i; } + +void dict_grep_cmap (Dict dict, void *vp, + const char **(*cmap)(void *vp, const char **from, int len)) +{ + dict->grep_cmap = cmap; + dict->grep_cmap_data = vp; +}