annotate regexParser/regexParser.cc @ 190:3e8e5780ad4a pairPro

change node::State to State
author Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
date Fri, 25 Dec 2015 15:30:52 +0900
parents 109d22faf7b5
children ecf70fb215a5
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
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
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 121
diff changeset
3 #include <string.h>
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 121
diff changeset
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
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
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
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
20 n->cc = NULL;
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
21 n->left = NULL;
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
22 n->right = NULL;
116
66c633575b53 remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 115
diff changeset
23 return n;
66c633575b53 remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 115
diff changeset
24 }
66c633575b53 remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 115
diff changeset
25
187
ef798db705e9 remove some warnings and errors(not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 185
diff changeset
26 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
27 NodePtr n = allocateNode();
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
28 n->tokenType = type;
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
29 n->cc = cc;
183
7ae0a3070647 implement generateTransitionList
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 180
diff changeset
30 n->state = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
31 n->left = left;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
32 n->right = right;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
33 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
34 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
35
144
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
36 CharClassPtr createCharClassWord(RegexInfoPtr ri) {
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
37 CharClassPtr cc = NEW(CharClass);
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
38 cc->type = 'a';
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
39 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
40 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
41 cc->cond.range.begin = cc->cond.range.end = *ri->tokenValue;
144
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
42 return cc;
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
43 }
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
44
149
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
45 /*
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
46 cond.range.begin cond.range.end
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
47 |----------------|
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
48 1.b---e
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
49 2.b------e
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
50 3.b------------e
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
51 4.b-----------------------e
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
52 5.b----------------------------e
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
53
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 6. b---------e
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
56 7. b----------------e
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
57 8. b---------------------e
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
58
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
59 |----------------|
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
60 9. b-----e
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
61 10. b--------e
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
62 11. b-------------e
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
63
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 12. b-----e
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 |----------------|
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
68 13. b--e
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
69
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
70 */
180
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
71 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
72 if (cc == NULL) {
188
109d22faf7b5 remove errors and warnings
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 187
diff changeset
73 createCharClassRange(begin,end,0,0,0);
178
5e8c6857934c implement charClassMerge
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 175
diff changeset
74 }
149
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
75 if (end < cc->cond.range.begin ) { // 1
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
76 if (cc->left) {
180
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
77 cc->left = insertCharClass(cc->left,begin,end);
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
78 } else {
188
109d22faf7b5 remove errors and warnings
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 187
diff changeset
79 cc->left = createCharClassRange(begin,end,0,0,0);
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
80 }
152
1c9e8ba64f6a add implement charclassMerge (not working)
masa
parents: 151
diff changeset
81 return cc;
149
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
82 } else if (end == cc->cond.range.begin ) { // 2
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
83 cc->cond.range.begin = begin;
150
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
84 return cc;
149
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
85 } 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
86 if (begin < cc->cond.range.begin) { // 3,4
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
87 cc->cond.range.begin = begin;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
88 }
150
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
89 return cc;
149
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
90 } else if (begin > cc->cond.range.end ) { // 13
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
91 if (cc->right) {
180
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
92 cc->right = insertCharClass(cc->right,begin,end);
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
93 } else {
188
109d22faf7b5 remove errors and warnings
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 187
diff changeset
94 cc->right = createCharClassRange(begin,end,0,0,0);
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
95 }
151
6ba059ce9109 test file
Nozomi
parents: 150
diff changeset
96 return cc;
150
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
97 }
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
98 if (cc->right) {
151
6ba059ce9109 test file
Nozomi
parents: 150
diff changeset
99 CharClassPtr right = cc->right;
6ba059ce9109 test file
Nozomi
parents: 150
diff changeset
100 begin = cc->cond.range.begin;
6ba059ce9109 test file
Nozomi
parents: 150
diff changeset
101 free(cc);
180
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
102 return insertCharClass(right,begin,end);
150
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
103 }
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
104 if (begin >= cc->cond.range.begin && begin <= cc->cond.range.end) { // 12
151
6ba059ce9109 test file
Nozomi
parents: 150
diff changeset
105 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
106 } else if (begin < cc->cond.range.begin) { // 5
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
107 cc->cond.range.begin = begin;
150
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
108 cc->cond.range.end = end;
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
109 } else {
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
110 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
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
111 }
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
112 return cc;
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
113 }
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
114
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
115 // <charClass> ::= '['<literal>'-'<literal>']'
112
ec485345daf9 some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 108
diff changeset
116 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
117 NodePtr charClass(RegexInfoPtr ri) {
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
118 CharClassPtr cc = NEW(CharClass);
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
119 NodePtr n = createNode(ri,'c',cc,0,0);
126
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
120 cc->type = 'r';
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
121 cc->nextState.bitContainer = 0;
148
d1ebba6e117a add test routing
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 147
diff changeset
122 cc->left = NULL;
d1ebba6e117a add test routing
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 147
diff changeset
123 cc->right = NULL;
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
124 RangeListPtr rangeList = &cc->cond.range;
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
125 rangeList->begin = *ri->ptr;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
126 rangeList->end = *ri->ptr;
145
50217a0545e8 fix charClass()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 144
diff changeset
127 rangeList->next = NULL;
130
7925e9abb078 add or flag
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 129
diff changeset
128
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
129 for (ri->ptr++; *ri->ptr && *ri->ptr != ']'; ri->ptr++) {
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
130 if (*ri->ptr == '-') {
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
131 rangeList->end = *(ri->ptr + 1);
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
132 ri->ptr++;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
133 continue;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
134 }
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
135 if (ri->ptr[0] == 0 || ri->ptr[0] == ']') break;
126
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
136 rangeList->next = NEW(RangeList);
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
137 rangeList = rangeList->next;
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
138 rangeList->begin = *ri->ptr;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
139 rangeList->end = *ri->ptr;
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
140 rangeList->next = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
141 }
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
142
150
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
143 RangeListPtr r = cc->cond.range.next;
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
144 cc->cond.range.next = 0;
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
145 for (; r; r = r->next) {
180
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
146 cc = insertCharClass(cc, r->begin, r->end);
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
147 }
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
148
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
149 // TODO literal support
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
150 // merge rangeList here
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
151 if (*ri->ptr) ri->ptr++;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
152 token(ri);
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
153 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
154 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
155
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
156 // <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
157 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
158 NodePtr literal(RegexInfoPtr ri) {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
159 CharClassPtr cc = createCharClassWord(ri);
144
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
160 token(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
161 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
162 return n;
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
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
165 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
166 void token(RegexInfoPtr ri) {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
167 while (ri->ptr[0] != '\0') {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
168 if (ri->ptr[0] == '('){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
169 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
170 ri->tokenType = '(';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
171 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
172 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
173 } else if (ri->ptr[0] == ')') {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
174 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
175 ri->tokenType = ')';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
176 ri->tokenValue = ri->ptr;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
177 return;
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
178 } else if (ri->ptr[0] == ']') {
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
179 ri->ptr++;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
180 ri->tokenType = ']';
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
181 ri->tokenValue = ri->ptr;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
182 return;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
183 } else if (ri->ptr[0] == '|'){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
184 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
185 ri->tokenType = '|';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
186 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
187 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
188 } else if (ri->ptr[0] == '*'){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
189 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
190 ri->tokenType = '*';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
191 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
192 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
193 } else if (ri->ptr[0] == '\\'){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
194 // need more proccesing
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
195 /*
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
196 \277
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
197 \0xa5
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
198 \[
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
199 \\
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
200 \utf-8 etc...
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
201 */
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
202 } else if (ri->ptr[0] == '[') {
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
203 ri->ptr++;
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
204 ri->tokenType = 'c';
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
205 ri->tokenValue = ri->ptr;
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
206 return;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
207 } else {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
208 ri->tokenType = 'a';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
209 ri->tokenValue = ri->ptr;
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
210 if (isalnum(ri->ptr[0])) {
122
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 121
diff changeset
211 ri->ptr++;
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 121
diff changeset
212 }
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
213 return;
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 }
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
216 ri->tokenType = 0;
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
217 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
218 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
219 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
220
130
7925e9abb078 add or flag
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 129
diff changeset
221 // <regexAtom> ::= <literal>|<charClass>|<group>
112
ec485345daf9 some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 108
diff changeset
222 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
223 NodePtr regexAtom(RegexInfoPtr ri) {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
224
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
225 NodePtr n = NULL;
124
masa
parents: 122
diff changeset
226 if (ri->tokenType == 'c') n = charClass(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
227 else if (ri->tokenType == 'a') n = literal(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
228 else if (ri->tokenType == '(') {
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
229 n = regex(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
230 if (ri->tokenType != ')') {
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
231 // error
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
232 }
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
233 token(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
234 }
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
235 if (ri->tokenType == '*') {
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
236 n = createNode(ri,'*',0,n,0);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
237 token(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
238 }
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
239 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
240 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
241
178
5e8c6857934c implement charClassMerge
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 175
diff changeset
242 RegexInfoPtr createRegexInfo (RegexInfoPtr ri) {
5e8c6857934c implement charClassMerge
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 175
diff changeset
243 ri->stateNumber++;
5e8c6857934c implement charClassMerge
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 175
diff changeset
244 return ri;
5e8c6857934c implement charClassMerge
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 175
diff changeset
245 }
5e8c6857934c implement charClassMerge
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 175
diff changeset
246
5e8c6857934c implement charClassMerge
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 175
diff changeset
247
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
248 // <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
249 NodePtr regex(RegexInfoPtr ri) {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
250 token(ri);
128
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 127
diff changeset
251 NodePtr n = regexAtom(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
252 while (ri->tokenType) {
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
253 if (ri->tokenType == '*') {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
254 n = createNode(ri,'*',0,n,0);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
255 token(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
256 return n;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
257 } else if (ri->tokenType == '|') {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
258 n = createNode(ri,'|',0,n,0);
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
259 NodePtr n1 = regex(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
260 n->right = n1;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
261 } else if (ri->tokenType == ')') {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
262 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
263 } else {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
264 n = createNode(ri,'+',0,n,0);
180
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
265 NodePtr n1 = regexAtom(ri);
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
266 n->right = n1;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
267 }
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
268 }
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
269 return n;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
270 }