X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=dict%2Flookgrep.c;h=2a42cb63592bc1233b0f69f5fa221b9e9e447df9;hb=1dcfe532fcc0447de623e717d70c46ab553329b9;hp=b71a484dadab4ffa08e4c3d207ad2f27a478fd7e;hpb=e9b13c0966913e073e98f9c5d7c5f4deed7c4c46;p=idzebra-moved-to-github.git diff --git a/dict/lookgrep.c b/dict/lookgrep.c index b71a484..2a42cb6 100644 --- a/dict/lookgrep.c +++ b/dict/lookgrep.c @@ -4,7 +4,20 @@ * Sebastian Hammer, Adam Dickmeiss * * $Log: lookgrep.c,v $ - * Revision 1.1 1994-10-03 17:23:04 adam + * 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 * First version of dictionary lookup with regular expressions and errors. * */ @@ -19,11 +32,13 @@ typedef unsigned MatchWord; #define WORD_BITS 32 +#define MAX_LENGTH 1024 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 @@ -32,19 +47,10 @@ 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 = 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) + 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++; @@ -106,7 +100,7 @@ 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] && @@ -115,7 +109,7 @@ 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] && @@ -124,7 +118,7 @@ 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] && @@ -133,7 +127,7 @@ 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] && @@ -142,7 +136,7 @@ static void mask_shift (MatchContext *mc, MatchWord *Rdst, MatchWord *Rsrc, } } s += 4; - if (s >= dfas->no) + if (s >= dfa->no_states) return; mask >>= 4; } @@ -150,7 +144,7 @@ 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; @@ -165,35 +159,35 @@ 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; } @@ -208,26 +202,28 @@ static void or (MatchContext *mc, MatchWord *Rdst, 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 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; @@ -236,10 +232,10 @@ static 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) + int (*userfunc)(Dict_char *name, const char *info), + Dict_char *prefix, struct DFA *dfa) { - int lo, hi; + int lo, hi, d; void *p; short *indxp; char *info; @@ -247,7 +243,8 @@ 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_pagesize(dict)-sizeof(short)); + while (lo <= hi) { if (indxp[-lo] > 0) @@ -261,10 +258,9 @@ 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; memcpy (&ch, info+j*sizeof(Dict_char), sizeof(Dict_char)); prefix[pos+j] = ch; @@ -274,30 +270,26 @@ static int dict_grep (Dict dict, Dict_ptr ptr, MatchContext *mc, (*userfunc)(prefix, info+(j+1)*sizeof(Dict_char)); break; } - - was_match = 0; - move (mc, Rj1, Rj0, ch, dfas, Rj_tmp); + move (mc, Rj1, Rj0, ch, dfa, Rj_tmp); for (d = mc->n; --d >= 0; ) if (Rj1[mc->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[mc->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; /* Dict_ptr subptr */ /* Dict_char sub char */ @@ -307,7 +299,7 @@ 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); + move (mc, Rj1, Rj, ch, dfa, Rj_tmp); for (d = mc->n; --d >= 0; ) if (Rj1[mc->range*mc->n + d]) break; @@ -316,24 +308,23 @@ static int dict_grep (Dict dict, Dict_ptr ptr, MatchContext *mc, 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[mc->range*mc->n + d] & mc->match_mask[d]) + { + prefix[pos+1] = DICT_EOS; + (*userfunc)(prefix, info+sizeof(Dict_ptr)+ + sizeof(Dict_char)); + break; + } } memcpy (&subptr, info, sizeof(Dict_ptr)); if (subptr) { dict_grep (dict, subptr, mc, Rj1, pos+1, - userfunc, prefix, dfas); + userfunc, prefix, dfa); dict_bf_readp (dict->dbf, ptr, &p); - indxp = (short*) ((char*) p+DICT_PAGESIZE-sizeof(short)); + indxp = (short*) ((char*) p+DICT_pagesize(dict) + -sizeof(short)); } } } @@ -343,53 +334,47 @@ static int dict_grep (Dict dict, Dict_ptr ptr, MatchContext *mc, } int dict_lookup_grep (Dict dict, Dict_char *pattern, int range, - int (*userfunc)(Dict_char *name, char *info)) + int (*userfunc)(Dict_char *name, const char *info)) { MatchWord *Rj; - Dict_char prefix[2048]; + Dict_char prefix[MAX_LENGTH+1]; char *this_pattern = pattern; - DFA_states *dfas; MatchContext *mc; - DFA *dfa = init_dfa(); + struct DFA *dfa = dfa_init(); int i, d; - i = parse_dfa (dfa, &this_pattern, dfa_thompson_chars); + 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, 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 = 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); + i = dict_grep (dict, 1, mc, Rj, 0, userfunc, prefix, dfa); - rm_dfas (&dfas); + dfa_delete (&dfa); xfree (Rj); + rm_MatchContext (&mc); return i; } - - -