annotate regexParser/regexParser.cc @ 300:3e78631a6222

remove ragne list
author Shinji KONO <kono@ie.u-ryukyu.ac.jp>
date Wed, 03 Feb 2016 13:56:01 +0900
parents 2f3e7bba038e
children 1188debbef10
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;
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
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
25 n->left = NULL;
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
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
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
32 n->tokenType = type;
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
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;
144
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
48 return cc;
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
49 }
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
50
149
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
51 /*
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
52 cond.range.begin cond.range.end
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 1.b---e
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
55 2.b------e
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
56 3.b------------e
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
57 4.b-----------------------e
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
58 5.b----------------------------e
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 |----------------|
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
61 6. b---------e
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
62 7. b----------------e
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
63 8. b---------------------e
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 |----------------|
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
66 9. b-----e
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
67 10. b--------e
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
68 11. 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 |----------------|
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
71 12. b-----e
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
72
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 13. 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 */
180
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
77 CharClassPtr insertCharClass(CharClassPtr cc, unsigned long begin, unsigned long end) {
300
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
78 if (begin>end) {
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
79 unsigned long tmp = begin; begin = end; end = tmp;
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
80 }
178
5e8c6857934c implement charClassMerge
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 175
diff changeset
81 if (cc == NULL) {
300
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
82 return createCharClassRange(begin,end,0,0,0);
178
5e8c6857934c implement charClassMerge
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 175
diff changeset
83 }
149
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
84 if (end < cc->cond.range.begin ) { // 1
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
85 if (cc->left) {
180
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
86 cc->left = insertCharClass(cc->left,begin,end);
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
87 } else {
188
109d22faf7b5 remove errors and warnings
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 187
diff changeset
88 cc->left = createCharClassRange(begin,end,0,0,0);
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
89 }
152
1c9e8ba64f6a add implement charclassMerge (not working)
masa
parents: 151
diff changeset
90 return cc;
149
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
91 } else if (end == cc->cond.range.begin ) { // 2
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
92 cc->cond.range.begin = begin;
150
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
93 return cc;
149
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
94 } 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
95 if (begin < cc->cond.range.begin) { // 3,4
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
96 cc->cond.range.begin = begin;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
97 }
150
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
98 return cc;
149
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
99 } else if (begin > cc->cond.range.end ) { // 13
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
100 if (cc->right) {
180
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
101 cc->right = insertCharClass(cc->right,begin,end);
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
102 } else {
188
109d22faf7b5 remove errors and warnings
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 187
diff changeset
103 cc->right = createCharClassRange(begin,end,0,0,0);
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
104 }
151
6ba059ce9109 test file
Nozomi
parents: 150
diff changeset
105 return cc;
150
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
106 }
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
107 if (cc->right) {
151
6ba059ce9109 test file
Nozomi
parents: 150
diff changeset
108 CharClassPtr right = cc->right;
6ba059ce9109 test file
Nozomi
parents: 150
diff changeset
109 begin = cc->cond.range.begin;
6ba059ce9109 test file
Nozomi
parents: 150
diff changeset
110 free(cc);
180
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
111 return insertCharClass(right,begin,end);
150
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
112 }
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
113 if (begin >= cc->cond.range.begin && begin <= cc->cond.range.end) { // 12
151
6ba059ce9109 test file
Nozomi
parents: 150
diff changeset
114 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
115 } else if (begin < cc->cond.range.begin) { // 5
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
116 cc->cond.range.begin = begin;
150
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
117 cc->cond.range.end = end;
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
118 } else {
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
119 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
120 }
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
121 return cc;
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
122 }
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
123
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
124 // <charClass> ::= '['<literal>'-'<literal>']'
112
ec485345daf9 some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 108
diff changeset
125 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
126 NodePtr charClass(RegexInfoPtr ri) {
300
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
127 CharClassPtr cc = NULL;
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
128 NodePtr n = createNode(ri,'c',cc,0,0);
300
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
129 unsigned char begin = *ri->ptr;
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
130 unsigned char end = *ri->ptr;
130
7925e9abb078 add or flag
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 129
diff changeset
131
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
132 for (ri->ptr++; *ri->ptr && *ri->ptr != ']'; ri->ptr++) {
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
133 if (*ri->ptr == '-') {
300
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
134 end = *(ri->ptr + 1);
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
135 cc = insertCharClass(cc, begin, end);
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
136 ri->ptr++;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
137 continue;
300
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
138 } else {
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
139 cc = insertCharClass(cc, begin, end);
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
140 }
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
141 if (ri->ptr[0] == 0 || ri->ptr[0] == ']') break;
300
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
142 begin = *ri->ptr;
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
143 end = *ri->ptr;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
144 }
300
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
145 n->cc = insertCharClass(cc, begin, end);
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
146 // TODO literal support
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
147 // merge rangeList here
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
148 if (*ri->ptr) ri->ptr++;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
149 token(ri);
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
150 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
151 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
152
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
153 // <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
154 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
155 NodePtr literal(RegexInfoPtr ri) {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
156 CharClassPtr cc = createCharClassWord(ri);
144
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
157 token(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
158 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
159 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
160 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
161
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
162 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
163 void token(RegexInfoPtr ri) {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
164 while (ri->ptr[0] != '\0') {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
165 if (ri->ptr[0] == '('){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
166 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
167 ri->tokenType = '(';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
168 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
169 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
170 } else if (ri->ptr[0] == ')') {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
171 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
172 ri->tokenType = ')';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
173 ri->tokenValue = ri->ptr;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
174 return;
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
175 } else if (ri->ptr[0] == ']') {
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
176 ri->ptr++;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
177 ri->tokenType = ']';
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
178 ri->tokenValue = ri->ptr;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
179 return;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
180 } else if (ri->ptr[0] == '|'){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
181 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
182 ri->tokenType = '|';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
183 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
184 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
185 } else if (ri->ptr[0] == '*'){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
186 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
187 ri->tokenType = '*';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
188 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
189 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
190 } else if (ri->ptr[0] == '\\'){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
191 // need more proccesing
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
192 /*
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
193 \277
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
194 \0xa5
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 \\
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
197 \utf-8 etc...
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
198 */
287
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 260
diff changeset
199 ri->ptr++;
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 260
diff changeset
200 ri->tokenType = 'a';
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 260
diff changeset
201 ri->tokenValue = ri->ptr;
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 260
diff changeset
202 ri->ptr++;
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 260
diff changeset
203 return;
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
204 } else if (ri->ptr[0] == '[') {
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
205 ri->ptr++;
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
206 ri->tokenType = 'c';
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
207 ri->tokenValue = ri->ptr;
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
208 return;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
209 } else {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
210 ri->tokenType = 'a';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
211 ri->tokenValue = ri->ptr;
287
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 260
diff changeset
212 // if (isalnum(ri->ptr[0])) {
122
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 121
diff changeset
213 ri->ptr++;
287
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 260
diff changeset
214 // }
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
215 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
216 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
217 }
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
218 ri->tokenType = 0;
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
219 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
220 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
221 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
222
130
7925e9abb078 add or flag
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 129
diff changeset
223 // <regexAtom> ::= <literal>|<charClass>|<group>
112
ec485345daf9 some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 108
diff changeset
224 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
225 NodePtr regexAtom(RegexInfoPtr ri) {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
226
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
227 NodePtr n = NULL;
124
masa
parents: 122
diff changeset
228 if (ri->tokenType == 'c') n = charClass(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
229 else if (ri->tokenType == 'a') n = literal(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
230 else if (ri->tokenType == '(') {
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
231 n = regex(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
232 if (ri->tokenType != ')') {
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
233 // error
287
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 260
diff changeset
234 fprintf(stderr,"unclosed ')' before %s \n", ri->ptr);
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 260
diff changeset
235 return createNode(ri,0,0,0,0);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
236 }
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
237 token(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
238 }
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
239 if (ri->tokenType == '*') {
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
240 n = createNode(ri,'*',0,n,0);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
241 token(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
242 }
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
243 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
244 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
245
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
246 // <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
247 NodePtr regex(RegexInfoPtr ri) {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
248 token(ri);
128
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 127
diff changeset
249 NodePtr n = regexAtom(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
250 while (ri->tokenType) {
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
251 if (ri->tokenType == '*') {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
252 n = createNode(ri,'*',0,n,0);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
253 token(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
254 return n;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
255 } else if (ri->tokenType == '|') {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
256 n = createNode(ri,'|',0,n,0);
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
257 NodePtr n1 = regex(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
258 n->right = n1;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
259 } else if (ri->tokenType == ')') {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
260 return n;
255
61d4d466e64c fix Makefile
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 214
diff changeset
261 } else if (ri->tokenType == ']') {
61d4d466e64c fix Makefile
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 214
diff changeset
262 // error
61d4d466e64c fix Makefile
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 214
diff changeset
263 return n;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
264 } else {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
265 n = createNode(ri,'+',0,n,0);
180
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
266 NodePtr n1 = regexAtom(ri);
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
267 n->right = n1;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
268 }
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
269 }
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
270 return n;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
271 }