annotate regexParser/regexParser.cc @ 180:d97bcab546e8 pairPro

implement getNext
author Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
date Thu, 24 Dec 2015 17:56:28 +0900
parents 6cf8252f3912
children 7ae0a3070647
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 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
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
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
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
22 n->cc = NULL;
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
23 n->left = NULL;
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
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
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
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
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;
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;
180
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
36 n->nodeNumber = ri->stateNumber;
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
37 ri->stateNumber++;
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
144
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
42 CharClassPtr createCharClassWord(RegexInfoPtr ri) {
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
43 CharClassPtr cc = NEW(CharClass);
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
44 cc->type = 'a';
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) {
178
5e8c6857934c implement charClassMerge
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 175
diff changeset
78 if (cc == NULL) {
180
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
79 createCharClassRange(begin,end,0,0);
178
5e8c6857934c implement charClassMerge
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 175
diff changeset
80 }
149
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
81 if (end < cc->cond.range.begin ) { // 1
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
82 if (cc->left) {
180
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
83 cc->left = insertCharClass(cc->left,begin,end);
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
84 } else {
180
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
85 cc->left = createCharClassRange(begin,end,0,0);
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
86 }
152
1c9e8ba64f6a add implement charclassMerge (not working)
masa
parents: 151
diff changeset
87 return cc;
149
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
88 } else if (end == cc->cond.range.begin ) { // 2
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
89 cc->cond.range.begin = begin;
150
a4940a8dd263 implement insertCharClass
masa
parents: 149
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.end) { // 3,4,6,7,9,10
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
92 if (begin < cc->cond.range.begin) { // 3,4
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
93 cc->cond.range.begin = begin;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
94 }
150
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
95 return cc;
149
f1880f25fabf add insertCharClass images
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 148
diff changeset
96 } else if (begin > cc->cond.range.end ) { // 13
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
97 if (cc->right) {
180
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
98 cc->right = insertCharClass(cc->right,begin,end);
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
99 } else {
180
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
100 cc->right = createCharClassRange(begin,end,0,0);
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
101 }
151
6ba059ce9109 test file
Nozomi
parents: 150
diff changeset
102 return cc;
150
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
103 }
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
104 if (cc->right) {
151
6ba059ce9109 test file
Nozomi
parents: 150
diff changeset
105 CharClassPtr right = cc->right;
6ba059ce9109 test file
Nozomi
parents: 150
diff changeset
106 begin = cc->cond.range.begin;
6ba059ce9109 test file
Nozomi
parents: 150
diff changeset
107 free(cc);
180
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
108 return insertCharClass(right,begin,end);
150
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
109 }
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
110 if (begin >= cc->cond.range.begin && begin <= cc->cond.range.end) { // 12
151
6ba059ce9109 test file
Nozomi
parents: 150
diff changeset
111 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
112 } else if (begin < cc->cond.range.begin) { // 5
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
113 cc->cond.range.begin = begin;
150
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
114 cc->cond.range.end = end;
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
115 } else {
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
116 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
117 }
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
118 return cc;
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
119 }
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
120
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
121 // <charClass> ::= '['<literal>'-'<literal>']'
112
ec485345daf9 some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 108
diff changeset
122 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
123 NodePtr charClass(RegexInfoPtr ri) {
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
124 CharClassPtr cc = NEW(CharClass);
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
125 NodePtr n = createNode(ri,'c',cc,0,0);
126
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
126 cc->type = 'r';
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
127 cc->nextState.bitContainer = 0;
148
d1ebba6e117a add test routing
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 147
diff changeset
128 cc->left = NULL;
d1ebba6e117a add test routing
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 147
diff changeset
129 cc->right = NULL;
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
130 RangeListPtr rangeList = &cc->cond.range;
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
131 rangeList->begin = *ri->ptr;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
132 rangeList->end = *ri->ptr;
145
50217a0545e8 fix charClass()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 144
diff changeset
133 rangeList->next = NULL;
130
7925e9abb078 add or flag
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 129
diff changeset
134
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
135 for (ri->ptr++; *ri->ptr && *ri->ptr != ']'; ri->ptr++) {
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
136 if (*ri->ptr == '-') {
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
137 rangeList->end = *(ri->ptr + 1);
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
138 ri->ptr++;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
139 continue;
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;
126
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
142 rangeList->next = NEW(RangeList);
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
143 rangeList = rangeList->next;
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
144 rangeList->begin = *ri->ptr;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
145 rangeList->end = *ri->ptr;
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
146 rangeList->next = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
147 }
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
148
150
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
149 RangeListPtr r = cc->cond.range.next;
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
150 cc->cond.range.next = 0;
a4940a8dd263 implement insertCharClass
masa
parents: 149
diff changeset
151 for (; r; r = r->next) {
180
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
152 cc = insertCharClass(cc, r->begin, r->end);
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
153 }
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
154
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
155 // TODO literal support
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
156 // merge rangeList here
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
157 if (*ri->ptr) ri->ptr++;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
158 token(ri);
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
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
162 // <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
163 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
164 NodePtr literal(RegexInfoPtr ri) {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
165 CharClassPtr cc = createCharClassWord(ri);
144
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
166 token(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
167 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
168 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
169 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
170
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
171 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
172 void token(RegexInfoPtr ri) {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
173 while (ri->ptr[0] != '\0') {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
174 if (ri->ptr[0] == '('){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
175 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
176 ri->tokenType = '(';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
177 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
178 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
179 } else if (ri->ptr[0] == ')') {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
180 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
181 ri->tokenType = ')';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
182 ri->tokenValue = ri->ptr;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
183 return;
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
184 } else if (ri->ptr[0] == ']') {
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
185 ri->ptr++;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
186 ri->tokenType = ']';
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
187 ri->tokenValue = ri->ptr;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
188 return;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
189 } else if (ri->ptr[0] == '|'){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
190 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
191 ri->tokenType = '|';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
192 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
193 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
194 } else if (ri->ptr[0] == '*'){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
195 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
196 ri->tokenType = '*';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
197 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
198 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
199 } else if (ri->ptr[0] == '\\'){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
200 // need more proccesing
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
201 /*
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
202 \277
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
203 \0xa5
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
204 \[
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
205 \\
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
206 \utf-8 etc...
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
207 */
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
208 } else if (ri->ptr[0] == '[') {
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
209 ri->ptr++;
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
210 ri->tokenType = 'c';
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
211 ri->tokenValue = ri->ptr;
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
212 return;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
213 } else {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
214 ri->tokenType = 'a';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
215 ri->tokenValue = ri->ptr;
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
216 if (isalnum(ri->ptr[0])) {
122
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 121
diff changeset
217 ri->ptr++;
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 121
diff changeset
218 }
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
219 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
220 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
221 }
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
222 ri->tokenType = 0;
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
223 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
224 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
225 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
226
130
7925e9abb078 add or flag
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 129
diff changeset
227 // <regexAtom> ::= <literal>|<charClass>|<group>
112
ec485345daf9 some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 108
diff changeset
228 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
229 NodePtr regexAtom(RegexInfoPtr ri) {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
230
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
231 NodePtr n = NULL;
124
masa
parents: 122
diff changeset
232 if (ri->tokenType == 'c') n = charClass(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
233 else if (ri->tokenType == 'a') n = literal(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
234 else if (ri->tokenType == '(') {
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
235 n = regex(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
236 if (ri->tokenType != ')') {
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
237 // error
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
238 }
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
239 token(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
240 }
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
241 if (ri->tokenType == '*') {
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
242 n = createNode(ri,'*',0,n,0);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
243 token(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
244 }
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
245 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
246 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
247
178
5e8c6857934c implement charClassMerge
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 175
diff changeset
248 RegexInfoPtr createRegexInfo (RegexInfoPtr ri) {
5e8c6857934c implement charClassMerge
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 175
diff changeset
249 ri->stateNumber++;
5e8c6857934c implement charClassMerge
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 175
diff changeset
250 return ri;
5e8c6857934c implement charClassMerge
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 175
diff changeset
251 }
5e8c6857934c implement charClassMerge
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 175
diff changeset
252
5e8c6857934c implement charClassMerge
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 175
diff changeset
253
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
254 // <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
255 NodePtr regex(RegexInfoPtr ri) {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
256 token(ri);
128
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 127
diff changeset
257 NodePtr n = regexAtom(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
258 while (ri->tokenType) {
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
259 if (ri->tokenType == '*') {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
260 n = createNode(ri,'*',0,n,0);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
261 token(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
262 return n;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
263 } else if (ri->tokenType == '|') {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
264 n = createNode(ri,'|',0,n,0);
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
265 NodePtr n1 = regex(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
266 n->right = n1;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
267 } else if (ri->tokenType == ')') {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
268 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
269 } else {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
270 n = createNode(ri,'+',0,n,0);
180
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
271 NodePtr n1 = regexAtom(ri);
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
272 n->right = n1;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
273 }
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
274 }
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
275 return n;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
276 }