annotate regexParser/grepWalk.cc @ 315:66012db6a717

add wordMode in grepWalk
author mir3636
date Sat, 07 May 2016 19:18:43 +0900
parents 1188debbef10
children c4d33b7c3ccd
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
231
d67649929e96 add grepWalk
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
1 #include <stdio.h>
299
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
2 #include <stdlib.h>
231
d67649929e96 add grepWalk
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
3
d67649929e96 add grepWalk
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
4 #include "grepWalk.h"
234
df4d04b3c34a implement grepWalk (not confirm correct working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 232
diff changeset
5 #include "subsetConstruction.h"
308
1188debbef10 separate CharClass
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 299
diff changeset
6 #include "CharClass.h"
299
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
7 #include "threadedSearch.h"
234
df4d04b3c34a implement grepWalk (not confirm correct working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 232
diff changeset
8
299
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
9 StatePtr nextState(BitVector bi,TransitionGeneratorPtr tg) {
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
10 // create tSearch in next state.
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
11 StatePtr state = tg->stateArray[bi.bitContainer];
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
12 if (state == NULL) {
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
13 // on the fly subset construction.
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
14 state = createState(tg,bi);
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
15 determinize(state,tg);
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
16 tg->stateArray[bi.bitContainer] = state;
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
17 }
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
18 return state;
293
948428caf616 NFA maximum match worked
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 266
diff changeset
19 }
948428caf616 NFA maximum match worked
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 266
diff changeset
20
299
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
21 void grepWalk(TransitionGeneratorPtr tg,Buffer buff) {
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
22 TSValue tsv = createTSValue(tg,buff);
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
23 tg->stateStart = NEW(State);
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
24 *tg->stateStart = *tg->stateList;
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
25 tg->stateStart->accept = false; // Start state never accept
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
26 StatePtr state = tg->stateStart;
234
df4d04b3c34a implement grepWalk (not confirm correct working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 232
diff changeset
27
299
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
28 #if DEBUG
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
29 TSValuePtr tsvp = &tsv; // make tsv visible in lldb
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
30 #endif
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
31 next: while (tsv.buff.buffptr < tsv.buff.buffend) {
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
32 if (state->accept) {
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
33 tsv = tg->stateMatch(tsv);
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
34 }
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
35 CharClassWalkerPtr ccw = createCharClassWalker(state->cc);
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
36 if (!hasNext(ccw)) {
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
37 // matched start again
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
38 state = tg->stateStart;
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
39 ccw = createCharClassWalker(state->cc);
235
4aab1e93a971 fix condition grepWalk.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 234
diff changeset
40 }
299
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
41 unsigned char c = *tsv.buff.buffptr++;
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
42 // printState(tsv.current->state);
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
43 while (hasNext(ccw)) {
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
44 CharClassPtr cc = getNext(ccw);
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
45 if (c<cc->cond.range.begin) {
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
46 state = tg->stateStart;
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
47 tsv = tg->stateSkip(tsv);
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
48 goto next;
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
49 } else if (c<=cc->cond.range.end) {
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
50 // range matched.
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
51 if (cc->cond.w.word) {
315
66012db6a717 add wordMode in grepWalk
mir3636
parents: 308
diff changeset
52 WordPtr w;
66012db6a717 add wordMode in grepWalk
mir3636
parents: 308
diff changeset
53 for (w = &cc->cond.w.word;w;w = w->next) {
66012db6a717 add wordMode in grepWalk
mir3636
parents: 308
diff changeset
54 // match the word.
66012db6a717 add wordMode in grepWalk
mir3636
parents: 308
diff changeset
55 // if (not match) continue;
66012db6a717 add wordMode in grepWalk
mir3636
parents: 308
diff changeset
56 if (strncmp(w.word,tsv.buff.buffptr-1,w.length)) break;
66012db6a717 add wordMode in grepWalk
mir3636
parents: 308
diff changeset
57 }
66012db6a717 add wordMode in grepWalk
mir3636
parents: 308
diff changeset
58 if (!w) continue;
66012db6a717 add wordMode in grepWalk
mir3636
parents: 308
diff changeset
59 tsv.buff.buffptr += w.length - 1;
299
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
60 }
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
61 state = nextState(cc->nextState,tg);
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
62 goto next;
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
63 }
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
64 }
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
65 state = tg->stateStart;
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
66 tsv = tg->stateSkip(tsv);
234
df4d04b3c34a implement grepWalk (not confirm correct working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 232
diff changeset
67 }
299
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
68 #if DEBUG
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
69 *tsvp = tsv;
bdfe0a32c48f grepWalk
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 293
diff changeset
70 #endif
234
df4d04b3c34a implement grepWalk (not confirm correct working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 232
diff changeset
71 }
df4d04b3c34a implement grepWalk (not confirm correct working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 232
diff changeset
72