Mercurial > hg > Members > masakoha > testcode
annotate regexParser/regexParser.cc @ 278:99d635926ef3
ceriumCreateAnyState implemented
author | Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp> |
---|---|
date | Sat, 30 Jan 2016 21:19:05 +0900 |
parents | 3f3c9902bb6d |
children | 2f3e7bba038e |
rev | line source |
---|---|
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
1 #include <stdlib.h> |
89
50a146c05192
add NodeNumber in Regex Parser tree
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
87
diff
changeset
|
2 #include <stdio.h> |
122 | 3 #include <string.h> |
4 #include <ctype.h> | |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
5 #include "regexParser.h" |
115
ca30f8334741
rename createRegexParser.cc to regexParser.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
112
diff
changeset
|
6 |
112
ec485345daf9
some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
108
diff
changeset
|
7 static NodePtr charClass(RegexInfoPtr); |
ec485345daf9
some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
108
diff
changeset
|
8 static void token(RegexInfoPtr); |
ec485345daf9
some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
108
diff
changeset
|
9 static NodePtr regexAtom(RegexInfoPtr); |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
10 |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
11 /** |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
12 * Create a node of regex parse tree. |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
13 * tokenType |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
14 * regexPosition(state) |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
15 * stateTransitionTable |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
16 */ |
118 | 17 static |
116
66c633575b53
remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
115
diff
changeset
|
18 NodePtr allocateNode() { |
129
b930be74a16e
remove word.h
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
128
diff
changeset
|
19 NodePtr n = NEW(Node); |
125 | 20 n->cc = NULL; |
260
3f3c9902bb6d
initialize node in allocateNode()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
255
diff
changeset
|
21 n->stateNum = 0; |
3f3c9902bb6d
initialize node in allocateNode()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
255
diff
changeset
|
22 n->nextStateNum = 0; |
3f3c9902bb6d
initialize node in allocateNode()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
255
diff
changeset
|
23 n->state = NULL; |
3f3c9902bb6d
initialize node in allocateNode()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
255
diff
changeset
|
24 n->nextState = NULL; |
125 | 25 n->left = NULL; |
26 n->right = NULL; | |
116
66c633575b53
remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
115
diff
changeset
|
27 return n; |
66c633575b53
remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
115
diff
changeset
|
28 } |
66c633575b53
remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
115
diff
changeset
|
29 |
187
ef798db705e9
remove some warnings and errors(not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
185
diff
changeset
|
30 NodePtr createNode(RegexInfoPtr ri,unsigned char type,CharClassPtr cc, NodePtr left, NodePtr right) { |
116
66c633575b53
remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
115
diff
changeset
|
31 NodePtr n = allocateNode(); |
125 | 32 n->tokenType = type; |
134 | 33 n->cc = cc; |
183
7ae0a3070647
implement generateTransitionList
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
180
diff
changeset
|
34 n->state = NULL; |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
35 n->left = left; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
36 n->right = right; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
37 return n; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
38 } |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
39 |
144
d8a4922eceae
remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
143
diff
changeset
|
40 CharClassPtr createCharClassWord(RegexInfoPtr ri) { |
d8a4922eceae
remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
143
diff
changeset
|
41 CharClassPtr cc = NEW(CharClass); |
d8a4922eceae
remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
143
diff
changeset
|
42 cc->type = 'a'; |
192
ecf70fb215a5
print charclass
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
190
diff
changeset
|
43 cc->left = NULL; |
ecf70fb215a5
print charclass
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
190
diff
changeset
|
44 cc->right = NULL; |
144
d8a4922eceae
remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
143
diff
changeset
|
45 cc->cond.w.word = ri->tokenValue; |
d8a4922eceae
remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
143
diff
changeset
|
46 cc->cond.w.length = ri->ptr - ri->tokenValue; |
178
5e8c6857934c
implement charClassMerge
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
175
diff
changeset
|
47 cc->cond.range.begin = cc->cond.range.end = *ri->tokenValue; |
192
ecf70fb215a5
print charclass
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
190
diff
changeset
|
48 cc->cond.range.next = NULL; |
144
d8a4922eceae
remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
143
diff
changeset
|
49 return cc; |
d8a4922eceae
remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
143
diff
changeset
|
50 } |
d8a4922eceae
remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
143
diff
changeset
|
51 |
149
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
52 /* |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
53 cond.range.begin cond.range.end |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
54 |----------------| |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
55 1.b---e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
56 2.b------e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
57 3.b------------e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
58 4.b-----------------------e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
59 5.b----------------------------e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
60 |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
61 |----------------| |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
62 6. b---------e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
63 7. b----------------e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
64 8. b---------------------e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
65 |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
66 |----------------| |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
67 9. b-----e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
68 10. b--------e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
69 11. b-------------e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
70 |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
71 |----------------| |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
72 12. b-----e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
73 |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
74 |----------------| |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
75 13. b--e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
76 |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
77 */ |
180
d97bcab546e8
implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
179
diff
changeset
|
78 CharClassPtr insertCharClass(CharClassPtr cc, unsigned long begin, unsigned long end) { |
178
5e8c6857934c
implement charClassMerge
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
175
diff
changeset
|
79 if (cc == NULL) { |
188
109d22faf7b5
remove errors and warnings
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
187
diff
changeset
|
80 createCharClassRange(begin,end,0,0,0); |
178
5e8c6857934c
implement charClassMerge
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
175
diff
changeset
|
81 } |
149
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
82 if (end < cc->cond.range.begin ) { // 1 |
147 | 83 if (cc->left) { |
180
d97bcab546e8
implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
179
diff
changeset
|
84 cc->left = insertCharClass(cc->left,begin,end); |
147 | 85 } else { |
188
109d22faf7b5
remove errors and warnings
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
187
diff
changeset
|
86 cc->left = createCharClassRange(begin,end,0,0,0); |
147 | 87 } |
152 | 88 return cc; |
149
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
89 } else if (end == cc->cond.range.begin ) { // 2 |
147 | 90 cc->cond.range.begin = begin; |
150 | 91 return cc; |
149
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
92 } else if (end <= cc->cond.range.end) { // 3,4,6,7,9,10 |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
93 if (begin < cc->cond.range.begin) { // 3,4 |
147 | 94 cc->cond.range.begin = begin; |
95 } | |
150 | 96 return cc; |
149
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
97 } else if (begin > cc->cond.range.end ) { // 13 |
147 | 98 if (cc->right) { |
180
d97bcab546e8
implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
179
diff
changeset
|
99 cc->right = insertCharClass(cc->right,begin,end); |
147 | 100 } else { |
188
109d22faf7b5
remove errors and warnings
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
187
diff
changeset
|
101 cc->right = createCharClassRange(begin,end,0,0,0); |
147 | 102 } |
151 | 103 return cc; |
150 | 104 } |
105 if (cc->right) { | |
151 | 106 CharClassPtr right = cc->right; |
107 begin = cc->cond.range.begin; | |
108 free(cc); | |
180
d97bcab546e8
implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
179
diff
changeset
|
109 return insertCharClass(right,begin,end); |
150 | 110 } |
111 if (begin >= cc->cond.range.begin && begin <= cc->cond.range.end) { // 12 | |
151 | 112 if (end > cc->cond.range.end) cc->cond.range.end = end; // 11,8 |
149
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
113 } else if (begin < cc->cond.range.begin) { // 5 |
147 | 114 cc->cond.range.begin = begin; |
150 | 115 cc->cond.range.end = end; |
116 } else { | |
117 printf("insertCharClass Error : begin %lu end %lu cc->begin %lu cc->end %lu\n", begin,end,cc->cond.range.begin,cc->cond.range.end); | |
147 | 118 } |
119 return cc; | |
142 | 120 } |
125 | 121 |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
122 // <charClass> ::= '['<literal>'-'<literal>']' |
112
ec485345daf9
some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
108
diff
changeset
|
123 static |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
124 NodePtr charClass(RegexInfoPtr ri) { |
125 | 125 CharClassPtr cc = NEW(CharClass); |
135 | 126 NodePtr n = createNode(ri,'c',cc,0,0); |
126 | 127 cc->type = 'r'; |
142 | 128 cc->nextState.bitContainer = 0; |
148
d1ebba6e117a
add test routing
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
147
diff
changeset
|
129 cc->left = NULL; |
d1ebba6e117a
add test routing
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
147
diff
changeset
|
130 cc->right = NULL; |
212
b0ae5273925c
implement allocateCCState()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
192
diff
changeset
|
131 cc->stateNum = 0; |
142 | 132 RangeListPtr rangeList = &cc->cond.range; |
147 | 133 rangeList->begin = *ri->ptr; |
134 rangeList->end = *ri->ptr; | |
145
50217a0545e8
fix charClass()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
144
diff
changeset
|
135 rangeList->next = NULL; |
130
7925e9abb078
add or flag
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
129
diff
changeset
|
136 |
135 | 137 for (ri->ptr++; *ri->ptr && *ri->ptr != ']'; ri->ptr++) { |
138 if (*ri->ptr == '-') { | |
147 | 139 rangeList->end = *(ri->ptr + 1); |
135 | 140 ri->ptr++; |
141 continue; | |
142 } | |
143 if (ri->ptr[0] == 0 || ri->ptr[0] == ']') break; | |
126 | 144 rangeList->next = NEW(RangeList); |
125 | 145 rangeList = rangeList->next; |
147 | 146 rangeList->begin = *ri->ptr; |
147 rangeList->end = *ri->ptr; | |
125 | 148 rangeList->next = NULL; |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
149 } |
147 | 150 |
150 | 151 RangeListPtr r = cc->cond.range.next; |
152 cc->cond.range.next = 0; | |
153 for (; r; r = r->next) { | |
180
d97bcab546e8
implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
179
diff
changeset
|
154 cc = insertCharClass(cc, r->begin, r->end); |
147 | 155 } |
192
ecf70fb215a5
print charclass
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
190
diff
changeset
|
156 cc->cond.range.next = 0; |
142 | 157 // TODO literal support |
158 // merge rangeList here | |
135 | 159 if (*ri->ptr) ri->ptr++; |
160 token(ri); | |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
161 return n; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
162 } |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
163 |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
164 // <literal> ::= [a-z][A-Z][0-9] |
112
ec485345daf9
some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
108
diff
changeset
|
165 static |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
166 NodePtr literal(RegexInfoPtr ri) { |
134 | 167 CharClassPtr cc = createCharClassWord(ri); |
144
d8a4922eceae
remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
143
diff
changeset
|
168 token(ri); |
134 | 169 NodePtr n = createNode(ri,'a',cc,0,0); |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
170 return n; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
171 } |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
172 |
133
ccc673449351
Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
132
diff
changeset
|
173 static |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
174 void token(RegexInfoPtr ri) { |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
175 while (ri->ptr[0] != '\0') { |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
176 if (ri->ptr[0] == '('){ |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
177 ri->ptr++; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
178 ri->tokenType = '('; |
118 | 179 ri->tokenValue = NULL; |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
180 return; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
181 } else if (ri->ptr[0] == ')') { |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
182 ri->ptr++; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
183 ri->tokenType = ')'; |
118 | 184 ri->tokenValue = ri->ptr; |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
185 return; |
135 | 186 } else if (ri->ptr[0] == ']') { |
187 ri->ptr++; | |
188 ri->tokenType = ']'; | |
189 ri->tokenValue = ri->ptr; | |
190 return; | |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
191 } else if (ri->ptr[0] == '|'){ |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
192 ri->ptr++; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
193 ri->tokenType = '|'; |
118 | 194 ri->tokenValue = NULL; |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
195 return; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
196 } else if (ri->ptr[0] == '*'){ |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
197 ri->ptr++; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
198 ri->tokenType = '*'; |
118 | 199 ri->tokenValue = NULL; |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
200 return; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
201 } else if (ri->ptr[0] == '\\'){ |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
202 // need more proccesing |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
203 /* |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
204 \277 |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
205 \0xa5 |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
206 \[ |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
207 \\ |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
208 \utf-8 etc... |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
209 */ |
142 | 210 } else if (ri->ptr[0] == '[') { |
211 ri->ptr++; | |
212 ri->tokenType = 'c'; | |
213 ri->tokenValue = ri->ptr; | |
214 return; | |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
215 } else { |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
216 ri->tokenType = 'a'; |
118 | 217 ri->tokenValue = ri->ptr; |
134 | 218 if (isalnum(ri->ptr[0])) { |
122 | 219 ri->ptr++; |
220 } | |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
221 return; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
222 } |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
223 } |
134 | 224 ri->tokenType = 0; |
225 ri->tokenValue = NULL; | |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
226 return; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
227 } |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
228 |
130
7925e9abb078
add or flag
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
129
diff
changeset
|
229 // <regexAtom> ::= <literal>|<charClass>|<group> |
112
ec485345daf9
some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
108
diff
changeset
|
230 static |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
231 NodePtr regexAtom(RegexInfoPtr ri) { |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
232 |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
233 NodePtr n = NULL; |
124 | 234 if (ri->tokenType == 'c') n = charClass(ri); |
134 | 235 else if (ri->tokenType == 'a') n = literal(ri); |
236 else if (ri->tokenType == '(') { | |
237 n = regex(ri); | |
238 if (ri->tokenType != ')') { | |
239 // error | |
240 } | |
241 token(ri); | |
242 } | |
243 if (ri->tokenType == '*') { | |
244 n = createNode(ri,'*',0,n,0); | |
245 token(ri); | |
246 } | |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
247 return n; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
248 } |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
249 |
133
ccc673449351
Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
132
diff
changeset
|
250 // <regex> ::= <regexAtom> | <regexAtom>'*'<regex> | <regexAtom>'|'<regex> | <regexAtom><regexAtom>'*' | <regexAtom><regex> |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
251 NodePtr regex(RegexInfoPtr ri) { |
134 | 252 token(ri); |
128 | 253 NodePtr n = regexAtom(ri); |
134 | 254 while (ri->tokenType) { |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
255 if (ri->tokenType == '*') { |
134 | 256 n = createNode(ri,'*',0,n,0); |
257 token(ri); | |
258 return n; | |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
259 } else if (ri->tokenType == '|') { |
134 | 260 n = createNode(ri,'|',0,n,0); |
133
ccc673449351
Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
132
diff
changeset
|
261 NodePtr n1 = regex(ri); |
134 | 262 n->right = n1; |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
263 } else if (ri->tokenType == ')') { |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
264 return n; |
255
61d4d466e64c
fix Makefile
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
214
diff
changeset
|
265 } else if (ri->tokenType == ']') { |
61d4d466e64c
fix Makefile
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
214
diff
changeset
|
266 // error |
61d4d466e64c
fix Makefile
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
214
diff
changeset
|
267 return n; |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
268 } else { |
134 | 269 n = createNode(ri,'+',0,n,0); |
180
d97bcab546e8
implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
179
diff
changeset
|
270 NodePtr n1 = regexAtom(ri); |
d97bcab546e8
implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
179
diff
changeset
|
271 n->right = n1; |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
272 } |
134 | 273 } |
274 return n; | |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
275 } |