📄 dfa.c
字号:
/* Create the first states. */ num_start_states = lastsc * 2; for ( i = 1; i <= num_start_states; ++i ) { numstates = 1; /* For each start condition, make one state for the case when * we're at the beginning of the line (the '^' operator) and * one for the case when we're not. */ if ( i % 2 == 1 ) nset[numstates] = scset[(i / 2) + 1]; else nset[numstates] = mkbranch( scbol[i / 2], scset[i / 2] ); nset = epsclosure( nset, &numstates, accset, &nacc, &hashval ); if ( snstods( nset, numstates, accset, nacc, hashval, &ds ) ) { numas += nacc; totnst += numstates; ++todo_next; if ( variable_trailing_context_rules && nacc > 0 ) check_trailing_context( nset, numstates, accset, nacc ); } } if ( ! fullspd ) { if ( ! snstods( nset, 0, accset, 0, 0, &end_of_buffer_state ) ) flexfatal( _( "could not create unique end-of-buffer state" ) ); ++numas; ++num_start_states; ++todo_next; } while ( todo_head < todo_next ) { targptr = 0; totaltrans = 0; for ( i = 1; i <= numecs; ++i ) state[i] = 0; ds = ++todo_head; dset = dss[ds]; dsize = dfasiz[ds]; if ( trace ) fprintf( stderr, _( "state # %d:\n" ), ds ); sympartition( dset, dsize, symlist, duplist ); for ( sym = 1; sym <= numecs; ++sym ) { if ( symlist[sym] ) { symlist[sym] = 0; if ( duplist[sym] == NIL ) { /* Symbol has unique out-transitions. */ numstates = symfollowset( dset, dsize, sym, nset ); nset = epsclosure( nset, &numstates, accset, &nacc, &hashval ); if ( snstods( nset, numstates, accset, nacc, hashval, &newds ) ) { totnst = totnst + numstates; ++todo_next; numas += nacc; if ( variable_trailing_context_rules && nacc > 0 ) check_trailing_context( nset, numstates, accset, nacc ); } state[sym] = newds; if ( trace ) fprintf( stderr, "\t%d\t%d\n", sym, newds ); targfreq[++targptr] = 1; targstate[targptr] = newds; ++numuniq; } else { /* sym's equivalence class has the same * transitions as duplist(sym)'s * equivalence class. */ targ = state[duplist[sym]]; state[sym] = targ; if ( trace ) fprintf( stderr, "\t%d\t%d\n", sym, targ ); /* Update frequency count for * destination state. */ i = 0; while ( targstate[++i] != targ ) ; ++targfreq[i]; ++numdup; } ++totaltrans; duplist[sym] = NIL; } } if ( caseins && ! useecs ) { register int j; for ( i = 'A', j = 'a'; i <= 'Z'; ++i, ++j ) { if ( state[i] == 0 && state[j] != 0 ) /* We're adding a transition. */ ++totaltrans; else if ( state[i] != 0 && state[j] == 0 ) /* We're taking away a transition. */ --totaltrans; state[i] = state[j]; } } numsnpairs += totaltrans; if ( ds > num_start_states ) check_for_backing_up( ds, state ); if ( nultrans ) { nultrans[ds] = state[NUL_ec]; state[NUL_ec] = 0; /* remove transition */ } if ( fulltbl ) { outn( " {" ); /* Supply array's 0-element. */ if ( ds == end_of_buffer_state ) mk2data( -end_of_buffer_state ); else mk2data( end_of_buffer_state ); for ( i = 1; i < num_full_table_rows; ++i ) /* Jams are marked by negative of state * number. */ mk2data( state[i] ? state[i] : -ds ); dataflush(); outn( " },\n" ); } else if ( fullspd ) place_state( state, ds, totaltrans ); else if ( ds == end_of_buffer_state ) /* Special case this state to make sure it does what * it's supposed to, i.e., jam on end-of-buffer. */ stack1( ds, 0, 0, JAMSTATE ); else /* normal, compressed state */ { /* Determine which destination state is the most * common, and how many transitions to it there are. */ comfreq = 0; comstate = 0; for ( i = 1; i <= targptr; ++i ) if ( targfreq[i] > comfreq ) { comfreq = targfreq[i]; comstate = targstate[i]; } bldtbl( state, ds, totaltrans, comstate, comfreq ); } } if ( fulltbl ) dataend(); else if ( ! fullspd ) { cmptmps(); /* create compressed template entries */ /* Create tables for all the states with only one * out-transition. */ while ( onesp > 0 ) { mk1tbl( onestate[onesp], onesym[onesp], onenext[onesp], onedef[onesp] ); --onesp; } mkdeftbl(); } flex_free( (void *) accset ); flex_free( (void *) nset ); }/* snstods - converts a set of ndfa states into a dfa state * * synopsis * is_new_state = snstods( int sns[numstates], int numstates, * int accset[num_rules+1], int nacc, * int hashval, int *newds_addr ); * * On return, the dfa state number is in newds. */int snstods( sns, numstates, accset, nacc, hashval, newds_addr )int sns[], numstates, accset[], nacc, hashval, *newds_addr; { int didsort = 0; register int i, j; int newds, *oldsns; for ( i = 1; i <= lastdfa; ++i ) if ( hashval == dhash[i] ) { if ( numstates == dfasiz[i] ) { oldsns = dss[i]; if ( ! didsort ) { /* We sort the states in sns so we * can compare it to oldsns quickly. * We use bubble because there probably * aren't very many states. */ bubble( sns, numstates ); didsort = 1; } for ( j = 1; j <= numstates; ++j ) if ( sns[j] != oldsns[j] ) break; if ( j > numstates ) { ++dfaeql; *newds_addr = i; return 0; } ++hshcol; } else ++hshsave; } /* Make a new dfa. */ if ( ++lastdfa >= current_max_dfas ) increase_max_dfas(); newds = lastdfa; dss[newds] = allocate_integer_array( numstates + 1 ); /* If we haven't already sorted the states in sns, we do so now, * so that future comparisons with it can be made quickly. */ if ( ! didsort ) bubble( sns, numstates ); for ( i = 1; i <= numstates; ++i ) dss[newds][i] = sns[i]; dfasiz[newds] = numstates; dhash[newds] = hashval; if ( nacc == 0 ) { if ( reject ) dfaacc[newds].dfaacc_set = (int *) 0; else dfaacc[newds].dfaacc_state = 0; accsiz[newds] = 0; } else if ( reject ) { /* We sort the accepting set in increasing order so the * disambiguating rule that the first rule listed is considered * match in the event of ties will work. We use a bubble * sort since the list is probably quite small. */ bubble( accset, nacc ); dfaacc[newds].dfaacc_set = allocate_integer_array( nacc + 1 ); /* Save the accepting set for later */ for ( i = 1; i <= nacc; ++i ) { dfaacc[newds].dfaacc_set[i] = accset[i]; if ( accset[i] <= num_rules ) /* Who knows, perhaps a REJECT can yield * this rule. */ rule_useful[accset[i]] = true; } accsiz[newds] = nacc; } else { /* Find lowest numbered rule so the disambiguating rule * will work. */ j = num_rules + 1; for ( i = 1; i <= nacc; ++i ) if ( accset[i] < j ) j = accset[i]; dfaacc[newds].dfaacc_state = j; if ( j <= num_rules ) rule_useful[j] = true; } *newds_addr = newds; return 1; }/* symfollowset - follow the symbol transitions one step * * synopsis * numstates = symfollowset( int ds[current_max_dfa_size], int dsize, * int transsym, int nset[current_max_dfa_size] ); */int symfollowset( ds, dsize, transsym, nset )int ds[], dsize, transsym, nset[]; { int ns, tsp, sym, i, j, lenccl, ch, numstates, ccllist; numstates = 0; for ( i = 1; i <= dsize; ++i ) { /* for each nfa state ns in the state set of ds */ ns = ds[i]; sym = transchar[ns]; tsp = trans1[ns]; if ( sym < 0 ) { /* it's a character class */ sym = -sym; ccllist = cclmap[sym]; lenccl = ccllen[sym]; if ( cclng[sym] ) { for ( j = 0; j < lenccl; ++j ) { /* Loop through negated character * class. */ ch = ccltbl[ccllist + j]; if ( ch == 0 ) ch = NUL_ec; if ( ch > transsym ) /* Transsym isn't in negated * ccl. */ break; else if ( ch == transsym ) /* next 2 */ goto bottom; } /* Didn't find transsym in ccl. */ nset[++numstates] = tsp; } else for ( j = 0; j < lenccl; ++j ) { ch = ccltbl[ccllist + j]; if ( ch == 0 ) ch = NUL_ec; if ( ch > transsym ) break; else if ( ch == transsym ) { nset[++numstates] = tsp; break; } } } else if ( sym >= 'A' && sym <= 'Z' && caseins ) flexfatal( _( "consistency check failed in symfollowset" ) ); else if ( sym == SYM_EPSILON ) { /* do nothing */ } else if ( ABS( ecgroup[sym] ) == transsym ) nset[++numstates] = tsp; bottom: ; } return numstates; }/* sympartition - partition characters with same out-transitions * * synopsis * sympartition( int ds[current_max_dfa_size], int numstates, * int symlist[numecs], int duplist[numecs] ); */void sympartition( ds, numstates, symlist, duplist )int ds[], numstates;int symlist[], duplist[]; { int tch, i, j, k, ns, dupfwd[CSIZE + 1], lenccl, cclp, ich; /* Partitioning is done by creating equivalence classes for those * characters which have out-transitions from the given state. Thus * we are really creating equivalence classes of equivalence classes. */ for ( i = 1; i <= numecs; ++i ) { /* initialize equivalence class list */ duplist[i] = i - 1; dupfwd[i] = i + 1; } duplist[1] = NIL; dupfwd[numecs] = NIL; for ( i = 1; i <= numstates; ++i ) { ns = ds[i]; tch = transchar[ns]; if ( tch != SYM_EPSILON ) { if ( tch < -lastccl || tch >= csize ) { flexfatal( _( "bad transition character detected in sympartition()" ) ); } if ( tch >= 0 ) { /* character transition */ int ec = ecgroup[tch]; mkechar( ec, dupfwd, duplist ); symlist[ec] = 1; } else { /* character class */ tch = -tch; lenccl = ccllen[tch]; cclp = cclmap[tch]; mkeccl( ccltbl + cclp, lenccl, dupfwd, duplist, numecs, NUL_ec ); if ( cclng[tch] ) { j = 0; for ( k = 0; k < lenccl; ++k ) { ich = ccltbl[cclp + k]; if ( ich == 0 ) ich = NUL_ec; for ( ++j; j < ich; ++j ) symlist[j] = 1; } for ( ++j; j <= numecs; ++j ) symlist[j] = 1; } else for ( k = 0; k < lenccl; ++k ) { ich = ccltbl[cclp + k]; if ( ich == 0 ) ich = NUL_ec; symlist[ich] = 1; } } } } }
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -