annotate c/regexParser/regexParser.cc @ 135:e1a262ec75f0 pairPro

impl charclass
author masa
date Fri, 04 Dec 2015 19:07:10 +0900
parents dbafc753078e
children 15815fcb6c2f
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 #include "error.h"
ca30f8334741 rename createRegexParser.cc to regexParser.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 112
diff changeset
7
112
ec485345daf9 some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 108
diff changeset
8 static NodePtr charClass(RegexInfoPtr);
ec485345daf9 some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 108
diff changeset
9 static void token(RegexInfoPtr);
ec485345daf9 some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 108
diff changeset
10 static NodePtr regexAtom(RegexInfoPtr);
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
11 NodePtr regex(RegexInfoPtr);
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 /**
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
14 * Create a node of regex parse tree.
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
15 * tokenType
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
16 * regexPosition(state)
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
17 * stateTransitionTable
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
18 */
112
ec485345daf9 some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 108
diff changeset
19
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
20 static
116
66c633575b53 remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 115
diff changeset
21 NodePtr allocateNode() {
129
b930be74a16e remove word.h
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 128
diff changeset
22 NodePtr n = NEW(Node);
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
23 n->cc = NULL;
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
24 n->left = NULL;
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
25 n->right = NULL;
116
66c633575b53 remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 115
diff changeset
26 return n;
66c633575b53 remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 115
diff changeset
27 }
66c633575b53 remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 115
diff changeset
28
112
ec485345daf9 some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 108
diff changeset
29 static
126
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
30 CharClassPtr createCharClassWord(RegexInfoPtr ri) {
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
31 CharClassPtr cc = NEW(CharClass);
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
32 cc->type = 'a';
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
33 cc->cond = NEW(Condition);
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
34 cc->cond->w = NEW(Word);
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
35 cc->cond->w->word = ri->tokenValue;
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
36 cc->cond->w->length = ri->ptr - ri->tokenValue;
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
37 token(ri);
126
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
38
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
39 return cc;
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
40 }
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
41
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
42 static
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
43 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
44 NodePtr n = allocateNode();
108
70069d4647a0 implement malloc error checking
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 104
diff changeset
45
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
46 n->tokenType = type;
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
47 n->cc = cc;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
48 n->left = left;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
49 n->right = right;
120
5d29b6a1b50f include Word in Node
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 118
diff changeset
50 n->nodeNumber = ri->nodeNumber;
5d29b6a1b50f include Word in Node
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 118
diff changeset
51 ri->nodeNumber++;
89
50a146c05192 add NodeNumber in Regex Parser tree
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 87
diff changeset
52
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
53 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
54 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
55
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
56
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
57 // <charClass> ::= '['<literal>'-'<literal>']'
112
ec485345daf9 some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 108
diff changeset
58 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
59 NodePtr charClass(RegexInfoPtr ri) {
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
60 CharClassPtr cc = NEW(CharClass);
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
61 NodePtr n = createNode(ri,'c',cc,0,0);
126
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
62 cc->type = 'r';
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
63 cc->cond = NEW(Condition);
126
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
64 cc->cond->range = NEW(RangeList);
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
65 cc->cond->range->begin = ri->ptr;
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
66 cc->cond->range->end = ri->ptr;
126
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
67 cc->cond->range->next = NULL;
122
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 121
diff changeset
68
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 121
diff changeset
69
126
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
70 RangeListPtr rangeList = cc->cond->range;
130
7925e9abb078 add or flag
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 129
diff changeset
71
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
72 for (ri->ptr++; *ri->ptr && *ri->ptr != ']'; ri->ptr++) {
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
73 if (*ri->ptr == '-') {
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
74 rangeList->end = ri->ptr + 1;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
75 ri->ptr++;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
76 continue;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
77 }
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
78 if (ri->ptr[0] == 0 || ri->ptr[0] == ']') break;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
79 if (ri->ptr[0] == rangeList->end[0] + 1) {
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
80 rangeList->end = ri->ptr;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
81 continue;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
82 }
126
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
83 rangeList->next = NEW(RangeList);
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
84 rangeList = rangeList->next;
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
85 rangeList->begin = ri->ptr;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
86 rangeList->end = ri->ptr;
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
87 rangeList->next = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
88 }
124
masa
parents: 122
diff changeset
89 // TODO literal support
122
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 121
diff changeset
90
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
91 if (*ri->ptr) ri->ptr++;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
92 token(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
93 token(ri);
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
94 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
95 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
96
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
97 // <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
98 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
99 NodePtr literal(RegexInfoPtr ri) {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
100 CharClassPtr cc = createCharClassWord(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
101 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
102 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
103 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
104
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
105 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
106 void token(RegexInfoPtr ri) {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
107 while (ri->ptr[0] != '\0') {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
108 if (ri->ptr[0] == '('){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
109 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
110 ri->tokenType = '(';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
111 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
112 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
113 } else if (ri->ptr[0] == ')') {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
114 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
115 ri->tokenType = ')';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
116 ri->tokenValue = ri->ptr;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
117 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
118 } else if (ri->ptr[0] == '[') {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
119 ri->ptr++;
122
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 121
diff changeset
120 ri->tokenType = 'c';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
121 ri->tokenValue = ri->ptr;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
122 return;
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
123 } else if (ri->ptr[0] == ']') {
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
124 ri->ptr++;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
125 ri->tokenType = ']';
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
126 ri->tokenValue = ri->ptr;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
127 return;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
128 } else if (ri->ptr[0] == '|'){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
129 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
130 ri->tokenType = '|';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
131 ri->tokenValue = NULL;
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
132 ri->orNum++;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
133 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
134 } else if (ri->ptr[0] == '*'){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
135 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
136 ri->tokenType = '*';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
137 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
138 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
139 } else if (ri->ptr[0] == '\\'){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
140 // need more proccesing
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
141 /*
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
142 \277
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
143 \0xa5
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
144 \[
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
145 \\
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
146 \utf-8 etc...
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
147 */
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
148 } else {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
149 ri->tokenType = 'a';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
150 ri->tokenValue = ri->ptr;
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
151 if (isalnum(ri->ptr[0])) {
122
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 121
diff changeset
152 ri->ptr++;
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 121
diff changeset
153 }
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
154 return;
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 }
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
157 ri->tokenType = 0;
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
158 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
159 return;
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
130
7925e9abb078 add or flag
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 129
diff changeset
162 // <regexAtom> ::= <literal>|<charClass>|<group>
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 regexAtom(RegexInfoPtr ri) {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
165
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
166 NodePtr n = NULL;
124
masa
parents: 122
diff changeset
167 if (ri->tokenType == 'c') n = charClass(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
168 else if (ri->tokenType == 'a') n = literal(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
169 else if (ri->tokenType == '(') {
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
170 n = regex(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
171 if (ri->tokenType != ')') {
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
172 // error
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
173 }
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
174 token(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
175 }
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
176 if (ri->tokenType == '*') {
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
177 n = createNode(ri,'*',0,n,0);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
178 token(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
179 }
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
180
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
181 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
182 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
183
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
184 // <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
185 NodePtr regex(RegexInfoPtr ri) {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
186 token(ri);
128
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 127
diff changeset
187 NodePtr n = regexAtom(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
188 while (ri->tokenType) {
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
189 if (ri->tokenType == '*') {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
190 n = createNode(ri,'*',0,n,0);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
191 token(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
192 return n;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
193 } else if (ri->tokenType == '|') {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
194 n = createNode(ri,'|',0,n,0);
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
195 NodePtr n1 = regex(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
196 n->right = n1;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
197 } else if (ri->tokenType == ')') {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
198 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
199 } else {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
200 n = createNode(ri,'+',0,n,0);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
201 NodePtr n1 = regexAtom(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
202 n->right = n1;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
203 }
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
204 }
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
205 return n;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
206 }