Mercurial > hg > Applications > Grep
annotate c/regexParser/regexParser.cc @ 150:a4940a8dd263 pairPro
implement insertCharClass
author | masa |
---|---|
date | Wed, 16 Dec 2015 16:42:18 +0900 |
parents | f1880f25fabf |
children | 6ba059ce9109 |
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 NodePtr regex(RegexInfoPtr); |
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 /** |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
13 * Create a node of regex parse tree. |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
14 * tokenType |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
15 * regexPosition(state) |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
16 * stateTransitionTable |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
17 */ |
112
ec485345daf9
some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
108
diff
changeset
|
18 |
118 | 19 static |
116
66c633575b53
remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
115
diff
changeset
|
20 NodePtr allocateNode() { |
129
b930be74a16e
remove word.h
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
128
diff
changeset
|
21 NodePtr n = NEW(Node); |
125 | 22 n->cc = NULL; |
23 n->left = NULL; | |
24 n->right = NULL; | |
116
66c633575b53
remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
115
diff
changeset
|
25 return n; |
66c633575b53
remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
115
diff
changeset
|
26 } |
66c633575b53
remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
115
diff
changeset
|
27 |
112
ec485345daf9
some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
108
diff
changeset
|
28 static |
134 | 29 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
|
30 NodePtr n = allocateNode(); |
108
70069d4647a0
implement malloc error checking
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
104
diff
changeset
|
31 |
125 | 32 n->tokenType = type; |
134 | 33 n->cc = cc; |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
34 n->left = left; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
35 n->right = right; |
120
5d29b6a1b50f
include Word in Node
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
118
diff
changeset
|
36 n->nodeNumber = ri->nodeNumber; |
5d29b6a1b50f
include Word in Node
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
118
diff
changeset
|
37 ri->nodeNumber++; |
89
50a146c05192
add NodeNumber in Regex Parser tree
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
87
diff
changeset
|
38 |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
39 return n; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
40 } |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
41 |
147 | 42 CharClassPtr createCharClassRange(unsigned long begin, unsigned long end, CharClassPtr left, CharClassPtr right) { |
43 CharClassPtr cc = NEW(CharClass); | |
44 cc->type = 'r'; | |
45 cc->cond.range.begin = begin; | |
46 cc->cond.range.end = end; | |
47 cc->left = left; | |
48 cc->right = right; | |
49 cc->nextState.bitContainer = 0; | |
50 return cc; | |
51 } | |
52 | |
144
d8a4922eceae
remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
143
diff
changeset
|
53 CharClassPtr createCharClassWord(RegexInfoPtr ri) { |
d8a4922eceae
remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
143
diff
changeset
|
54 CharClassPtr cc = NEW(CharClass); |
d8a4922eceae
remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
143
diff
changeset
|
55 cc->type = 'a'; |
d8a4922eceae
remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
143
diff
changeset
|
56 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
|
57 cc->cond.w.length = ri->ptr - ri->tokenValue; |
d8a4922eceae
remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
143
diff
changeset
|
58 cc->nextState.bitContainer = 0; |
d8a4922eceae
remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
143
diff
changeset
|
59 return cc; |
d8a4922eceae
remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
143
diff
changeset
|
60 } |
d8a4922eceae
remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
143
diff
changeset
|
61 |
149
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
62 /* |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
63 cond.range.begin cond.range.end |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
64 |----------------| |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
65 1.b---e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
66 2.b------e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
67 3.b------------e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
68 4.b-----------------------e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
69 5.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 6. b---------e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
73 7. b----------------e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
74 8. b---------------------e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
75 |
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 9. b-----e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
78 10. b--------e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
79 11. b-------------e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
80 |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
81 |----------------| |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
82 12. b-----e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
83 |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
84 |----------------| |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
85 13. b--e |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
86 |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
87 */ |
150 | 88 CharClassPtr insertCharClass(CharClassPtr cc, unsigned long begin, unsigned long end) { |
149
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
89 |
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
90 if (end < cc->cond.range.begin ) { // 1 |
148
d1ebba6e117a
add test routing
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
147
diff
changeset
|
91 CharClassPtr cc1 = createCharClassRange(cc->cond.range.begin,cc->cond.range.end,cc->left,cc->right); |
147 | 92 if (cc->left) { |
93 cc1->left = insertCharClass(cc->left,begin,end); | |
94 return cc1; | |
95 } else { | |
96 CharClassPtr cc2 = createCharClassRange(begin,end,0,0); | |
97 cc1->left = cc2; | |
98 return cc1; | |
99 } | |
149
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
100 } else if (end == cc->cond.range.begin ) { // 2 |
147 | 101 cc->cond.range.begin = begin; |
150 | 102 return cc; |
149
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
103 } 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
|
104 if (begin < cc->cond.range.begin) { // 3,4 |
147 | 105 cc->cond.range.begin = begin; |
106 } | |
150 | 107 return cc; |
149
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
108 } else if (begin > cc->cond.range.end ) { // 13 |
148
d1ebba6e117a
add test routing
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
147
diff
changeset
|
109 CharClassPtr cc1 = createCharClassRange(cc->cond.range.begin,cc->cond.range.end,cc->left,cc->right); |
147 | 110 if (cc->right) { |
148
d1ebba6e117a
add test routing
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
147
diff
changeset
|
111 cc1->right = insertCharClass(cc->right,begin,end); |
147 | 112 return cc1; |
113 } else { | |
114 CharClassPtr cc2 = createCharClassRange(begin,end,0,0); | |
115 cc1->right = cc2; | |
116 return cc1; | |
117 } | |
150 | 118 } |
119 if (cc->right) { | |
120 return insertCharClass(cc->right,cc->cond.range.begin,end); | |
121 } | |
122 if (begin >= cc->cond.range.begin && begin <= cc->cond.range.end) { // 12 | |
123 if (end > cc->cond.range.end) cc->cond.range.end = end; // 11 | |
149
f1880f25fabf
add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
148
diff
changeset
|
124 } else if (begin < cc->cond.range.begin) { // 5 |
147 | 125 cc->cond.range.begin = begin; |
150 | 126 cc->cond.range.end = end; |
127 } else { | |
128 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 | 129 } |
130 return cc; | |
142 | 131 } |
125 | 132 |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
133 // <charClass> ::= '['<literal>'-'<literal>']' |
112
ec485345daf9
some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
108
diff
changeset
|
134 static |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
135 NodePtr charClass(RegexInfoPtr ri) { |
125 | 136 CharClassPtr cc = NEW(CharClass); |
135 | 137 NodePtr n = createNode(ri,'c',cc,0,0); |
126 | 138 cc->type = 'r'; |
142 | 139 cc->nextState.bitContainer = 0; |
148
d1ebba6e117a
add test routing
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
147
diff
changeset
|
140 cc->left = NULL; |
d1ebba6e117a
add test routing
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
147
diff
changeset
|
141 cc->right = NULL; |
142 | 142 RangeListPtr rangeList = &cc->cond.range; |
147 | 143 rangeList->begin = *ri->ptr; |
144 rangeList->end = *ri->ptr; | |
145
50217a0545e8
fix charClass()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
144
diff
changeset
|
145 rangeList->next = NULL; |
130
7925e9abb078
add or flag
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
129
diff
changeset
|
146 |
135 | 147 for (ri->ptr++; *ri->ptr && *ri->ptr != ']'; ri->ptr++) { |
148 if (*ri->ptr == '-') { | |
147 | 149 rangeList->end = *(ri->ptr + 1); |
135 | 150 ri->ptr++; |
151 continue; | |
152 } | |
153 if (ri->ptr[0] == 0 || ri->ptr[0] == ']') break; | |
126 | 154 rangeList->next = NEW(RangeList); |
125 | 155 rangeList = rangeList->next; |
147 | 156 rangeList->begin = *ri->ptr; |
157 rangeList->end = *ri->ptr; | |
125 | 158 rangeList->next = NULL; |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
159 } |
147 | 160 |
150 | 161 RangeListPtr r = cc->cond.range.next; |
162 cc->cond.range.next = 0; | |
163 for (; r; r = r->next) { | |
147 | 164 cc = insertCharClass(cc, r->begin, r->end); |
165 } | |
166 | |
167 n->cc = cc; | |
142 | 168 // TODO literal support |
169 // merge rangeList here | |
135 | 170 if (*ri->ptr) ri->ptr++; |
171 token(ri); | |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
172 return n; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
173 } |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
174 |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
175 // <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
|
176 static |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
177 NodePtr literal(RegexInfoPtr ri) { |
134 | 178 CharClassPtr cc = createCharClassWord(ri); |
144
d8a4922eceae
remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
143
diff
changeset
|
179 token(ri); |
134 | 180 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
|
181 return n; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
182 } |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
183 |
133
ccc673449351
Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
132
diff
changeset
|
184 static |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
185 void token(RegexInfoPtr ri) { |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
186 while (ri->ptr[0] != '\0') { |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
187 if (ri->ptr[0] == '('){ |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
188 ri->ptr++; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
189 ri->tokenType = '('; |
118 | 190 ri->tokenValue = NULL; |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
191 return; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
192 } else if (ri->ptr[0] == ')') { |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
193 ri->ptr++; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
194 ri->tokenType = ')'; |
118 | 195 ri->tokenValue = ri->ptr; |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
196 return; |
135 | 197 } else if (ri->ptr[0] == ']') { |
198 ri->ptr++; | |
199 ri->tokenType = ']'; | |
200 ri->tokenValue = ri->ptr; | |
201 return; | |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
202 } else if (ri->ptr[0] == '|'){ |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
203 ri->ptr++; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
204 ri->tokenType = '|'; |
118 | 205 ri->tokenValue = NULL; |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
206 return; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
207 } else if (ri->ptr[0] == '*'){ |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
208 ri->ptr++; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
209 ri->tokenType = '*'; |
118 | 210 ri->tokenValue = NULL; |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
211 return; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
212 } else if (ri->ptr[0] == '\\'){ |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
213 // need more proccesing |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
214 /* |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
215 \277 |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
216 \0xa5 |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
217 \[ |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
218 \\ |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
219 \utf-8 etc... |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
220 */ |
142 | 221 } else if (ri->ptr[0] == '[') { |
222 ri->ptr++; | |
223 ri->tokenType = 'c'; | |
224 ri->tokenValue = ri->ptr; | |
225 return; | |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
226 } else { |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
227 ri->tokenType = 'a'; |
118 | 228 ri->tokenValue = ri->ptr; |
134 | 229 if (isalnum(ri->ptr[0])) { |
122 | 230 ri->ptr++; |
231 } | |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
232 return; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
233 } |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
234 } |
134 | 235 ri->tokenType = 0; |
236 ri->tokenValue = NULL; | |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
237 return; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
238 } |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
239 |
130
7925e9abb078
add or flag
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
129
diff
changeset
|
240 // <regexAtom> ::= <literal>|<charClass>|<group> |
112
ec485345daf9
some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
108
diff
changeset
|
241 static |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
242 NodePtr regexAtom(RegexInfoPtr ri) { |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
243 |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
244 NodePtr n = NULL; |
124 | 245 if (ri->tokenType == 'c') n = charClass(ri); |
134 | 246 else if (ri->tokenType == 'a') n = literal(ri); |
247 else if (ri->tokenType == '(') { | |
248 n = regex(ri); | |
249 if (ri->tokenType != ')') { | |
250 // error | |
251 } | |
252 token(ri); | |
253 } | |
254 if (ri->tokenType == '*') { | |
255 n = createNode(ri,'*',0,n,0); | |
256 token(ri); | |
257 } | |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
258 |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
259 return n; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
260 } |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
261 |
133
ccc673449351
Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
132
diff
changeset
|
262 // <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
|
263 NodePtr regex(RegexInfoPtr ri) { |
134 | 264 token(ri); |
128 | 265 NodePtr n = regexAtom(ri); |
134 | 266 while (ri->tokenType) { |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
267 if (ri->tokenType == '*') { |
134 | 268 n = createNode(ri,'*',0,n,0); |
269 token(ri); | |
270 return n; | |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
271 } else if (ri->tokenType == '|') { |
134 | 272 n = createNode(ri,'|',0,n,0); |
133
ccc673449351
Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
132
diff
changeset
|
273 NodePtr n1 = regex(ri); |
134 | 274 n->right = n1; |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
275 } else if (ri->tokenType == ')') { |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
276 return n; |
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
277 } else { |
134 | 278 n = createNode(ri,'+',0,n,0); |
279 NodePtr n1 = regexAtom(ri); | |
280 n->right = n1; | |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
281 } |
134 | 282 } |
283 return n; | |
82
1d9bbf922bb6
add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
284 } |