annotate c/regexParser/regexParser.cc @ 142:de0f332d560c pairPro

insert charClassMerge function
author masa
date Fri, 11 Dec 2015 14:54:00 +0900
parents 15815fcb6c2f
children 32977f5a2ed0
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';
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
33 cc->cond.w.word = ri->tokenValue;
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
34 cc->cond.w.length = ri->ptr - ri->tokenValue;
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
35 cc->nextState.bitContainer = 0;
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
36 token(ri);
126
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
37
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
38 return cc;
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
39 }
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
40
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
41 static
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
42 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
43 NodePtr n = allocateNode();
108
70069d4647a0 implement malloc error checking
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 104
diff changeset
44
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
45 n->tokenType = type;
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
46 n->cc = cc;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
47 n->left = left;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
48 n->right = right;
120
5d29b6a1b50f include Word in Node
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 118
diff changeset
49 n->nodeNumber = ri->nodeNumber;
5d29b6a1b50f include Word in Node
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 118
diff changeset
50 ri->nodeNumber++;
89
50a146c05192 add NodeNumber in Regex Parser tree
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 87
diff changeset
51
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
52 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
53 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
54
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
55 CharClassPtr charClassMerge(CharClassPtr src, CharClassPtr add) {
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
56 // 重なっているccの領域を分割する
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
57 // 必要ならばnextStateを重ねあわせる
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
58 // 変更があった場合は新しくリストを作って返す
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
59 if (src->type == 'a') {
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
60 if (add->type == 'a') {
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
61 if (src->cond.w.word[0] > add->cond.w.word[0]) {
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
62 // add のほうが小さいので小さい順のccをつくる
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
63 CharClassPtr left = charClassMerge(add->left.src);
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
64 return createCharClassWord(add->cond->w.word, left, add->right);
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
65 } else {
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
66
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
67 }
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
68 } else if (add->type == 'c') {
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
69 //
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
70 if (src->cond.w.word[0] < add->cond.range.begin) {
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
71
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
72 } else (src->cond->w.word[0] < add->end) {
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
73 } else if (src->type == 'c') {
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
74
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
75 }
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
76 }
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
77
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
78 // <charClass> ::= '['<literal>'-'<literal>']'
112
ec485345daf9 some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 108
diff changeset
79 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
80 NodePtr charClass(RegexInfoPtr ri) {
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
81 CharClassPtr cc = NEW(CharClass);
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
82 NodePtr n = createNode(ri,'c',cc,0,0);
126
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
83 cc->type = 'r';
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
84 cc->nextState.bitContainer = 0;
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
85 RangeListPtr rangeList = &cc->cond.range;
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
86 rangeList->begin = ri->ptr;
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
87 rangeList->end = ri->ptr;
130
7925e9abb078 add or flag
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 129
diff changeset
88
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
89 for (ri->ptr++; *ri->ptr && *ri->ptr != ']'; ri->ptr++) {
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
90 if (*ri->ptr == '-') {
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
91 rangeList->end = ri->ptr + 1;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
92 ri->ptr++;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
93 continue;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
94 }
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
95 if (ri->ptr[0] == 0 || ri->ptr[0] == ']') break;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
96 if (ri->ptr[0] == rangeList->end[0] + 1) {
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
97 rangeList->end = ri->ptr;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
98 continue;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
99 }
126
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
100 rangeList->next = NEW(RangeList);
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
101 rangeList = rangeList->next;
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
102 rangeList->begin = ri->ptr;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
103 rangeList->end = ri->ptr;
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
104 rangeList->next = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
105 }
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
106 // TODO literal support
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
107 // merge rangeList here
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
108 if (*ri->ptr) ri->ptr++;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
109 token(ri);
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
110 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
111 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
112
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
113 // <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
114 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
115 NodePtr literal(RegexInfoPtr ri) {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
116 CharClassPtr cc = createCharClassWord(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
117 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
118 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
119 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
120
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
121 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
122 void token(RegexInfoPtr ri) {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
123 while (ri->ptr[0] != '\0') {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
124 if (ri->ptr[0] == '('){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
125 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
126 ri->tokenType = '(';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
127 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
128 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
129 } else if (ri->ptr[0] == ')') {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
130 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
131 ri->tokenType = ')';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
132 ri->tokenValue = ri->ptr;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
133 return;
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
134 } else if (ri->ptr[0] == ']') {
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
135 ri->ptr++;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
136 ri->tokenType = ']';
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
137 ri->tokenValue = ri->ptr;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
138 return;
82
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 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
141 ri->tokenType = '|';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
142 ri->tokenValue = NULL;
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
143 ri->orNum++;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
144 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
145 } else if (ri->ptr[0] == '*'){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
146 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
147 ri->tokenType = '*';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
148 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
149 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
150 } else if (ri->ptr[0] == '\\'){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
151 // need more proccesing
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 \277
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
154 \0xa5
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 \\
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
157 \utf-8 etc...
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
158 */
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
159 } else if (ri->ptr[0] == '[') {
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
160 ri->ptr++;
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
161 ri->tokenType = 'c';
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
162 ri->tokenValue = ri->ptr;
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
163 return;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
164 } else {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
165 ri->tokenType = 'a';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
166 ri->tokenValue = ri->ptr;
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
167 if (isalnum(ri->ptr[0])) {
122
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 121
diff changeset
168 ri->ptr++;
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 121
diff changeset
169 }
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
170 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
171 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
172 }
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
173 ri->tokenType = 0;
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
174 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
175 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
176 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
177
130
7925e9abb078 add or flag
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 129
diff changeset
178 // <regexAtom> ::= <literal>|<charClass>|<group>
112
ec485345daf9 some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 108
diff changeset
179 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
180 NodePtr regexAtom(RegexInfoPtr ri) {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
181
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
182 NodePtr n = NULL;
124
masa
parents: 122
diff changeset
183 if (ri->tokenType == 'c') n = charClass(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
184 else if (ri->tokenType == 'a') n = literal(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
185 else if (ri->tokenType == '(') {
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
186 n = regex(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
187 if (ri->tokenType != ')') {
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
188 // error
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
189 }
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
190 token(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
191 }
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
192 if (ri->tokenType == '*') {
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
193 n = createNode(ri,'*',0,n,0);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
194 token(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
195 }
82
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 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
198 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
199
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
200 // <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
201 NodePtr regex(RegexInfoPtr ri) {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
202 token(ri);
128
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 127
diff changeset
203 NodePtr n = regexAtom(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
204 while (ri->tokenType) {
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
205 if (ri->tokenType == '*') {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
206 n = createNode(ri,'*',0,n,0);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
207 token(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
208 return n;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
209 } else if (ri->tokenType == '|') {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
210 n = createNode(ri,'|',0,n,0);
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
211 NodePtr n1 = regex(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
212 n->right = n1;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
213 } else if (ri->tokenType == ')') {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
214 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
215 } else {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
216 n = createNode(ri,'+',0,n,0);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
217 NodePtr n1 = regexAtom(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
218 n->right = n1;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
219 }
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
220 }
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
221 return n;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
222 }