X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=dict%2Flookgrep.c;h=fba076f0a464864fa5764e9aaf9abcd3d89f77e8;hp=b71a484dadab4ffa08e4c3d207ad2f27a478fd7e;hb=a030c87bc444608639905eca95e29f84a4f1d991;hpb=e9b13c0966913e073e98f9c5d7c5f4deed7c4c46 diff --git a/dict/lookgrep.c b/dict/lookgrep.c index b71a484..fba076f 100644 --- a/dict/lookgrep.c +++ b/dict/lookgrep.c @@ -1,13 +1,23 @@ -/* - * Copyright (C) 1994, Index Data I/S - * All rights reserved. - * Sebastian Hammer, Adam Dickmeiss - * - * $Log: lookgrep.c,v $ - * Revision 1.1 1994-10-03 17:23:04 adam - * First version of dictionary lookup with regular expressions and errors. - * - */ +/* This file is part of the Zebra server. + Copyright (C) 1994-2011 Index Data + + 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 this program; if not, write to the Free Software + Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA + +*/ + + #include #include @@ -15,38 +25,41 @@ #include #include -#include +#include "dict-p.h" typedef unsigned MatchWord; #define WORD_BITS 32 +#define MAX_LENGTH 1024 + +/* This code is based + * Sun Wu and Udi Manber: Fast Text Searching Allowing Errors. + * Communications of the ACM, pp. 83-91, Vol. 35, No. 10, Oct. 1992, USA. + * PostScript version of the paper in its submitted form: agrep1.ps) + * recommended reading to understand AGREP ! + * + * http://www.tgries.de/agrep/#AGREP1PS + * http://www.tgries.de/agrep/doc/agrep1ps.zip + */ typedef struct { - int n; /* no of MatchWord needed */ - int range; /* max no. of errors */ - MatchWord *Sc; /* Mask Sc */ + int n; /* no of MatchWord needed */ + int range; /* max no. of errors */ + int fact; /* (range+1)*n */ + MatchWord *match_mask; /* match_mask */ } MatchContext; #define INLINE -static INLINE void set_bit (MatchContext *mc, MatchWord *m, int ch, int state) +static INLINE void set_bit(MatchContext *mc, MatchWord *m, int ch, int state) { int off = state & (WORD_BITS-1); int wno = state / WORD_BITS; - + m[mc->n * ch + wno] |= 1<n * ch + wno] &= ~(1<n * ch + wno] & (1<n = (dfas->no+WORD_BITS) / WORD_BITS; + mc->n = (dfa->no_states+WORD_BITS) / WORD_BITS; mc->range = range; - mc->Sc = xcalloc (sizeof(*mc->Sc) * 256 * mc->n, 1); - - for (i=0; ino; i++) - { - int j; - DFA_state *state = dfas->sortarray[i]; + mc->fact = (range+1)*mc->n; + mc->match_mask = (MatchWord *) xcalloc(mc->n, sizeof(*mc->match_mask)); - for (j=0; jtran_no; j++) - { - int ch; - int ch0 = state->trans[j].ch[0]; - int ch1 = state->trans[j].ch[1]; - assert (ch0 >= 0 && ch1 >= 0); - - for (ch = ch0; ch <= ch1; ch++) - set_bit (mc, mc->Sc, ch, i); - } - } + for (s = 0; sno_states; s++) + if (dfa->states[s]->rule_no) + set_bit(mc, mc->match_mask, 0, s); return mc; } +static void rm_MatchContext(MatchContext **mc) +{ + xfree((*mc)->match_mask); + xfree(*mc); + *mc = NULL; +} -static void mask_shift (MatchContext *mc, MatchWord *Rdst, MatchWord *Rsrc, - DFA_states *dfas, int ch) +static void mask_shift(MatchContext *mc, MatchWord *Rdst, MatchWord *Rsrc, + struct DFA *dfa, int ch) { int j, s = 0; MatchWord *Rsrc_p = Rsrc, mask; -#if 1 for (j = 0; jn; j++) Rdst[j] = 0; -#else - Rdst[0] = 1; - for (j = 1; jn; j++) - Rdst[j] = 0; -#endif while (1) { mask = *Rsrc_p++; @@ -106,51 +107,51 @@ 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] && ch <= state->trans[i].ch[1]) - set_bit (mc, Rdst, 0, state->trans[i].to); + 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) if (ch >= state->trans[i].ch[0] && ch <= state->trans[i].ch[1]) - set_bit (mc, Rdst, 0, state->trans[i].to); + 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) if (ch >= state->trans[i].ch[0] && ch <= state->trans[i].ch[1]) - set_bit (mc, Rdst, 0, state->trans[i].to); + 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) if (ch >= state->trans[i].ch[0] && ch <= state->trans[i].ch[1]) - set_bit (mc, Rdst, 0, state->trans[i].to); + set_bit(mc, Rdst, 0, state->trans[i].to); } } s += 4; - if (s >= dfas->no) + if (s >= dfa->no_states) return; mask >>= 4; } } } -static void shift (MatchContext *mc, MatchWord *Rdst, MatchWord *Rsrc, - DFA_states *dfas) +static void shift(MatchContext *mc, MatchWord *Rdst, MatchWord *Rsrc, + struct DFA *dfa) { int j, s = 0; MatchWord *Rsrc_p = Rsrc, mask; @@ -165,89 +166,94 @@ 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); + 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); + 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); + 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); + set_bit(mc, Rdst, 0, state->trans[i].to); } } s += 4; - if (s >= dfas->no) + if (s >= dfa->no_states) return; mask >>= 4; } } } -static void or (MatchContext *mc, MatchWord *Rdst, - MatchWord *Rsrc1, MatchWord *Rsrc2) +static void or(MatchContext *mc, MatchWord *Rdst, + MatchWord *Rsrc1, MatchWord *Rsrc2) { int i; for (i = 0; in; i++) Rdst[i] = Rsrc1[i] | Rsrc2[i]; } -static int move (MatchContext *mc, MatchWord *Rj1, MatchWord *Rj, - Dict_char ch, DFA_states *dfas, MatchWord *Rtmp) +static INLINE int move(MatchContext *mc, MatchWord *Rj1, MatchWord *Rj, + Dict_char ch, struct DFA *dfa, MatchWord *Rtmp, + int range) { int d; - MatchWord *Rj_a = Rtmp; - MatchWord *Rj_b = Rtmp + mc->n; - MatchWord *Rj_c = Rtmp + 2*mc->n; + 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++) { - mask_shift (mc, Rj_b, Rj+d*mc->n, dfas, ch); /* 1 */ - - or (mc, Rj_a, Rj+(d-1)*mc->n, Rj1+(d-1)*mc->n); /* 2,3 */ + or(mc, Rtmp, Rj, Rj1); /* 2,3 */ - shift (mc, Rj_c, Rj_a, dfas); - - or (mc, Rj_a, Rj_b, Rj_c); /* 1,2,3*/ + shift(mc, Rtmp_2, Rtmp, dfa); + + mask_shift(mc, Rtmp, Rj+mc->n, dfa, ch); /* 1 */ + + or(mc, Rtmp, Rtmp_2, Rtmp); /* 1,2,3*/ + + Rj1 += mc->n; - or (mc, Rj1+d*mc->n, Rj_a, Rj+(d-1)*mc->n); /* 1,2,3,4 */ + or(mc, Rj1, Rtmp, Rj); /* 1,2,3,4 */ + + Rj += mc->n; } return 1; } -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) +static int grep(Dict dict, Dict_ptr ptr, MatchContext *mc, + 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; + int lo, hi, d; void *p; short *indxp; char *info; - dict_bf_readp (dict->dbf, ptr, &p); + 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) { if (indxp[-lo] > 0) @@ -261,79 +267,99 @@ static int dict_grep (Dict dict, Dict_ptr ptr, MatchContext *mc, for (j=0; ; j++) { Dict_char ch; - int s, d; - MatchWord *Rj0 = Rj + j *(1+mc->range)*mc->n; - MatchWord *Rj1 = Rj + (j+1)*(1+mc->range)*mc->n; - MatchWord *Rj_tmp = Rj + (j+2)*(1+mc->range)*mc->n; + 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)); + 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)); + { + int ret = userfunc((char*) prefix, + info+(j+1)*sizeof(Dict_char), client); + if (ret) + return ret; + } break; } - - was_match = 0; - 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; - for (s = 0; sno; s++) - { - if (dfas->sortarray[s]->rule_no) - if (get_bit (mc, Rj1+mc->range*mc->n, 0, s)) - was_match = 1; - } - for (d = 0; d <= mc->range; d++) - reset_bit (mc, Rj1+d*mc->n, 0, dfas->no); + was_match = 0; + for (d = mc->n; --d >= 0; ) + if (Rj1[range*mc->n + d] & mc->match_mask[d]) + { + was_match = 1; + break; + } } } else { - MatchWord *Rj1 = Rj+ (1+mc->range)*mc->n; - MatchWord *Rj_tmp = Rj+2*(1+mc->range)*mc->n; + MatchWord *Rj1 = Rj+ mc->fact; + MatchWord *Rj_tmp = Rj+2*mc->fact; Dict_char ch; - int d; + int range; /* Dict_ptr subptr */ /* Dict_char sub char */ /* unsigned char length of information */ /* char * information */ info = (char*)p - indxp[-lo]; - memcpy (&ch, info+sizeof(Dict_ptr), sizeof(Dict_char)); + 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) { Dict_ptr subptr; if (info[sizeof(Dict_ptr)+sizeof(Dict_char)]) { - int s; - for (s = 0; sno; s++) - if (dfas->sortarray[s]->rule_no) - if (get_bit (mc, Rj1+mc->range*mc->n, 0, s)) - { - prefix[pos+1] = DICT_EOS; - (*userfunc)(prefix, info+sizeof(Dict_ptr)+ - sizeof(Dict_char)); - break; - } + for (d = mc->n; --d >= 0; ) + if (Rj1[range*mc->n + d] & mc->match_mask[d]) + { + int ret; + prefix[pos+1] = DICT_EOS; + ret = userfunc((char*) prefix, + info+sizeof(Dict_ptr)+ + sizeof(Dict_char), client); + if (ret) + return ret; + break; + } } - memcpy (&subptr, info, sizeof(Dict_ptr)); + memcpy(&subptr, info, sizeof(Dict_ptr)); if (subptr) { - dict_grep (dict, subptr, mc, Rj1, pos+1, - userfunc, prefix, dfas); - dict_bf_readp (dict->dbf, ptr, &p); - indxp = (short*) ((char*) p+DICT_PAGESIZE-sizeof(short)); + int ret = grep(dict, subptr, mc, Rj1, pos+1, + client, userfunc, prefix, dfa, max_pos, + init_pos); + if (ret) + return ret; + + dict_bf_readp(dict->dbf, ptr, &p); + indxp = (short*) ((char*) p+DICT_bsize(p)-sizeof(short)); } } } @@ -342,54 +368,89 @@ 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[2048]; - char *this_pattern = pattern; - DFA_states *dfas; + Dict_char prefix[MAX_LENGTH+1]; + const char *this_pattern = pattern; MatchContext *mc; - DFA *dfa = init_dfa(); - int i, d; + struct DFA *dfa = dfa_init(); + int i, d, ret = 0; + +#if 0 + debug_dfa_trav = 1; + debug_dfa_tran = 1; + debug_dfa_followpos = 1; + dfa_verbose = 1; +#endif + + dfa_anyset_includes_nl(dfa); + + yaz_log(YLOG_DEBUG, "dict_lookup_grep range=%d", range); + for (i = 0; pattern[i]; i++) + { + yaz_log(YLOG_DEBUG, " %2d %3d %c", i, pattern[i], + (pattern[i] > ' ' && pattern[i] < 127) ? pattern[i] : '?'); + } + + dfa_set_cmap(dfa, dict->grep_cmap_data, dict->grep_cmap); - i = parse_dfa (dfa, &this_pattern, dfa_thompson_chars); + i = dfa_parse(dfa, &this_pattern); if (i || *this_pattern) { - rm_dfa (&dfa); + yaz_log(YLOG_WARN, "dfa_parse fail=%d", i); + dfa_delete(&dfa); return -1; } - dfa->root = dfa->top; - dfas = mk_dfas (dfa, 200); - rm_dfa (&dfa); + dfa_mkstate(dfa); - mc = mk_MatchContext (dfas, range); + mc = mk_MatchContext(dfa, range); - Rj = xcalloc ((1000+dict_strlen(pattern)+range+2)*(range+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++) + memcpy(Rj + mc->n * d, Rj + mc->n * (d-1), mc->n * sizeof(*Rj)); + for (s = 0; s < dfa->no_states; s++) { - if (get_bit (mc, Rj, d-1, 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); + set_bit(mc, Rj, d, state->trans[i].to); } } } - i = dict_grep (dict, 1, mc, Rj, 0, userfunc, prefix, dfas); - - rm_dfas (&dfas); - xfree (Rj); - return i; + *max_pos = 0; + if (dict->head.root) + ret = grep(dict, dict->head.root, mc, Rj, 0, client, + userfunc, prefix, + dfa, max_pos, init_pos); + yaz_log(YLOG_DEBUG, "max_pos = %d", *max_pos); + dfa_delete(&dfa); + xfree(Rj); + rm_MatchContext(&mc); + return ret; } - +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; +} +/* + * Local variables: + * c-basic-offset: 4 + * c-file-style: "Stroustrup" + * indent-tabs-mode: nil + * End: + * vim: shiftwidth=4 tabstop=8 expandtab + */