Mercurial > hg > Applications > Grep

comparison regexParser/CharClass.cc @ 309:058c87665213

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
small fix
author Shinji KONO <kono@ie.u-ryukyu.ac.jp>
date Mon, 08 Feb 2016 18:04:28 +0900
parents 1188debbef10
children df27e6cab846
comparison
equal deleted inserted replaced
308:1188debbef10 309:058c87665213
8 #include "BitVector.h" 8 #include "BitVector.h"
9 #include "error.h" 9 #include "error.h"
10 10
11 #include "CharClass.h" 11 #include "CharClass.h"
12 12
13
14 CharClassPtr createCharClassWord(RegexInfoPtr ri) {
15 CharClassPtr cc = NEW(CharClass);
16 cc->type = 'a';
17 cc->left = NULL;
18 cc->right = NULL;
19 cc->cond.w.word = ri->tokenValue;
20 cc->cond.w.length = ri->ptr - ri->tokenValue;
21 cc->cond.range.begin = cc->cond.range.end = *ri->tokenValue;
22 return cc;
23 }
24
25 /*
26 cond.range.begin cond.range.end
27 |----------------|
28 1.b---e
29 2.b------e
30 3.b------------e
31 4.b-----------------------e
32 5.b----------------------------e
33
34 |----------------|
35 6. b---------e
36 7. b----------------e
37 8. b---------------------e
38
39 |----------------|
40 9. b-----e
41 10. b--------e
42 11. b-------------e
43
44 |----------------|
45 12. b-----e
46
47 |----------------|
48 13. b--e
49
50 */
51 CharClassPtr insertCharClass(CharClassPtr cc, unsigned long begin, unsigned long end) {
52 if (begin>end) {
53 unsigned long tmp = begin; begin = end; end = tmp;
54 }
55 if (cc == NULL) {
56 return createCharClassRange(begin,end,0,0,0);
57 }
58 if (end < cc->cond.range.begin ) { // 1
59 if (cc->left) {
60 cc->left = insertCharClass(cc->left,begin,end);
61 } else {
62 cc->left = createCharClassRange(begin,end,0,0,0);
63 }
64 return cc;
65 } else if (end == cc->cond.range.begin ) { // 2
66 cc->cond.range.begin = begin;
67 return cc;
68 } else if (end <= cc->cond.range.end) { // 3,4,6,7,9,10
69 if (begin < cc->cond.range.begin) { // 3,4
70 cc->cond.range.begin = begin;
71 }
72 return cc;
73 } else if (begin > cc->cond.range.end ) { // 13
74 if (cc->right) {
75 cc->right = insertCharClass(cc->right,begin,end);
76 } else {
77 cc->right = createCharClassRange(begin,end,0,0,0);
78 }
79 return cc;
80 }
81 if (cc->right) {
82 CharClassPtr right = cc->right;
83 begin = cc->cond.range.begin;
84 free(cc);
85 return insertCharClass(right,begin,end);
86 }
87 if (begin >= cc->cond.range.begin && begin <= cc->cond.range.end) { // 12
88 if (end > cc->cond.range.end) cc->cond.range.end = end; // 11,8
89 } else if (begin < cc->cond.range.begin) { // 5
90 cc->cond.range.begin = begin;
91 cc->cond.range.end = end;
92 } else {
93 printf("insertCharClass Error : begin %lu end %lu cc->begin %lu cc->end %lu\n", begin,end,cc->cond.range.begin,cc->cond.range.end);
94 }
95 return cc;
96 }
97
98
99 CharClassPtr createCharClassRange(unsigned long begin, unsigned long end,unsigned long state, CharClassPtr left, CharClassPtr right) { 13 CharClassPtr createCharClassRange(unsigned long begin, unsigned long end,unsigned long state, CharClassPtr left, CharClassPtr right) {
100 CharClassPtr cc = NEW(CharClass); 14 CharClassPtr cc = NEW(CharClass);
101 cc->type = 'r';
102 cc->cond.range.begin = begin; 15 cc->cond.range.begin = begin;
103 cc->cond.range.end = end; 16 cc->cond.range.end = end;
104 cc->cond.w.word = NULL; 17 cc->cond.w.word = NULL;
105 cc->cond.w.length = 0; 18 cc->cond.w.length = 0;
106 cc->left = left; 19 cc->left = left;
107 cc->right = right; 20 cc->right = right;
108 cc->nextState.bitContainer = state; 21 cc->nextState.bitContainer = state;
109 return cc; 22 return cc;
110 } 23 }
24
25 CharClassPtr createCharClassWord(RegexInfoPtr ri) {
26 CharClassPtr cc = NEW(CharClass);
27 cc->left = NULL;
28 cc->right = NULL;
29 cc->cond.w.word = ri->tokenValue;
30 cc->cond.w.length = ri->ptr - ri->tokenValue;
31 cc->cond.range.begin = cc->cond.range.end = *ri->tokenValue;
32 return cc;
33 }
34
35 /*
36 cond.range.begin cond.range.end
37 |----------------|
38 1.b---e
39 2.b------e
40 3.b------------e
41 4.b-----------------------e
42 5.b----------------------------e
43
44 |----------------|
45 6. b---------e
46 7. b----------------e
47 8. b---------------------e
48
49 |----------------|
50 9. b-----e
51 10. b--------e
52 11. b-------------e
53
54 |----------------|
55 12. b-----e
56
57 |----------------|
58 13. b--e
59
60 */
61 CharClassPtr insertCharClass(CharClassPtr cc, unsigned long begin, unsigned long end) {
62 if (begin>end) {
63 unsigned long tmp = begin; begin = end; end = tmp;
64 }
65 if (cc == NULL) {
66 return createCharClassRange(begin,end,0,0,0);
67 }
68 if (end < cc->cond.range.begin ) { // 1
69 if (cc->left) {
70 cc->left = insertCharClass(cc->left,begin,end);
71 } else {
72 cc->left = createCharClassRange(begin,end,0,0,0);
73 }
74 return cc;
75 } else if (end == cc->cond.range.begin ) { // 2
76 cc->cond.range.begin = begin;
77 return cc;
78 } else if (end <= cc->cond.range.end) { // 3,4,6,7,9,10
79 if (begin < cc->cond.range.begin) { // 3,4
80 cc->cond.range.begin = begin;
81 }
82 return cc;
83 } else if (begin > cc->cond.range.end ) { // 13
84 if (cc->right) {
85 cc->right = insertCharClass(cc->right,begin,end);
86 } else {
87 cc->right = createCharClassRange(begin,end,0,0,0);
88 }
89 return cc;
90 }
91 if (cc->right) {
92 CharClassPtr right = cc->right;
93 begin = cc->cond.range.begin;
94 free(cc);
95 return insertCharClass(right,begin,end);
96 }
97 if (begin >= cc->cond.range.begin && begin <= cc->cond.range.end) { // 12
98 if (end > cc->cond.range.end) cc->cond.range.end = end; // 11,8
99 } else if (begin < cc->cond.range.begin) { // 5
100 cc->cond.range.begin = begin;
101 cc->cond.range.end = end;
102 } else {
103 printf("insertCharClass Error : begin %lu end %lu cc->begin %lu cc->end %lu\n", begin,end,cc->cond.range.begin,cc->cond.range.end);
104 }
105 return cc;
106 }
107
108
111 109
112 CharClassPtr charClassMerge(CharClassPtr cc,unsigned long begin, unsigned long end, BitVector nextState) ; 110 CharClassPtr charClassMerge(CharClassPtr cc,unsigned long begin, unsigned long end, BitVector nextState) ;
113 111
114 CharClassPtr mergeCCTree(CharClassPtr cc,unsigned char mBegin,unsigned char mEnd,BitVector nextState) { 112 CharClassPtr mergeCCTree(CharClassPtr cc,unsigned char mBegin,unsigned char mEnd,BitVector nextState) {
115 CharClassPtr cc1; 113 CharClassPtr cc1;