annotate c/regexParser/regexParser.cc @ 147:84d32375383a pairPro

implement insertCharClass
author masa
date Tue, 15 Dec 2015 17:14:35 +0900
parents 50217a0545e8
children d1ebba6e117a
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;
120
5d29b6a1b50f include Word in Node
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 118
diff changeset
36 n->nodeNumber = ri->nodeNumber;
5d29b6a1b50f include Word in Node
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 118
diff changeset
37 ri->nodeNumber++;
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
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
42 CharClassPtr createCharClassRange(unsigned long begin, unsigned long end, CharClassPtr left, CharClassPtr right) {
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
43 CharClassPtr cc = NEW(CharClass);
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
44 cc->type = 'r';
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
45 cc->cond.range.begin = begin;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
46 cc->cond.range.end = end;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
47 cc->left = left;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
48 cc->right = right;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
49 cc->nextState.bitContainer = 0;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
50 return cc;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
51 }
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
52
144
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
53 CharClassPtr createCharClassWord(RegexInfoPtr ri) {
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
54 CharClassPtr cc = NEW(CharClass);
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
55 cc->type = 'a';
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
56 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
57 cc->cond.w.length = ri->ptr - ri->tokenValue;
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
58 cc->nextState.bitContainer = 0;
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
59 return cc;
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
60 }
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
61
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
62 CharClassPtr insertCharClass(CharClassPtr cc, unsigned char begin, unsigned char end) {
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
63 if (end < cc->cond.range.begin ) {
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
64 CharClassPtr cc1 = createCharClassRange(cc->begin,cc->end,cc->left,cc->right);
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
65 if (cc->left) {
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
66 cc1->left = insertCharClass(cc->left,begin,end);
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
67 return cc1;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
68 } else {
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
69 CharClassPtr cc2 = createCharClassRange(begin,end,0,0);
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
70 cc1->left = cc2;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
71 return cc1;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
72 }
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
73 } else if (end == cc->cond.range.begin ) {
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
74 cc->cond.range.begin = begin;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
75 } else if (end <= cc->cond.range.end) {
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
76 if (begin < cc->cond.range.begin) {
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
77 cc->cond.range.begin = begin;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
78 }
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
79 } else if (begin > cc->cond.range.end ) {
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
80 CharClassPtr cc1 = createCharClassRange(cc->begin,cc->end,cc->left,cc->right);
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
81 if (cc->right) {
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
82 cc1->rigt = insertCharClass(cc->right,begin,end);
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
83 return cc1;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
84 } else {
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
85 CharClassPtr cc2 = createCharClassRange(begin,end,0,0);
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
86 cc1->right = cc2;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
87 return cc1;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
88 }
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
89 } else if (begin == cc->cond.range.end ) {
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
90 cc->cond.range.end = end;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
91 } else if (begin < cc->cond.range.begin) {
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
92 cc->cond.range.begin = begin;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
93 }
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
94 return cc;
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
95 }
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
96
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
97 // <charClass> ::= '['<literal>'-'<literal>']'
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 charClass(RegexInfoPtr ri) {
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
100 CharClassPtr cc = NEW(CharClass);
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
101 NodePtr n = createNode(ri,'c',cc,0,0);
126
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
102 cc->type = 'r';
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
103 cc->nextState.bitContainer = 0;
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
104 RangeListPtr rangeList = &cc->cond.range;
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
105 rangeList->begin = *ri->ptr;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
106 rangeList->end = *ri->ptr;
145
50217a0545e8 fix charClass()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 144
diff changeset
107 rangeList->next = NULL;
130
7925e9abb078 add or flag
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 129
diff changeset
108
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
109 for (ri->ptr++; *ri->ptr && *ri->ptr != ']'; ri->ptr++) {
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
110 if (*ri->ptr == '-') {
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
111 rangeList->end = *(ri->ptr + 1);
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
112 ri->ptr++;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
113 continue;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
114 }
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
115 if (ri->ptr[0] == 0 || ri->ptr[0] == ']') break;
144
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
116 if (ri->ptr[0] == rangeList->end + 1) {
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
117 rangeList->end = *ri->ptr;
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
118 continue;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
119 }
126
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
120 rangeList->next = NEW(RangeList);
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
121 rangeList = rangeList->next;
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
122 rangeList->begin = *ri->ptr;
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
123 rangeList->end = *ri->ptr;
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
124 rangeList->next = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
125 }
147
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
126
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
127 for (RangeListPtr r = &cc->cond.range; r; r = r->next) {
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
128 cc = insertCharClass(cc, r->begin, r->end);
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
129 }
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
130
84d32375383a implement insertCharClass
masa
parents: 145
diff changeset
131 n->cc = cc;
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
132 // TODO literal support
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
133 // merge rangeList here
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
134 if (*ri->ptr) ri->ptr++;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
135 token(ri);
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
136 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
137 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
138
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
139 // <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
140 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
141 NodePtr literal(RegexInfoPtr ri) {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
142 CharClassPtr cc = createCharClassWord(ri);
144
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 143
diff changeset
143 token(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
144 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
145 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
146 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
147
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
148 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
149 void token(RegexInfoPtr ri) {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
150 while (ri->ptr[0] != '\0') {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
151 if (ri->ptr[0] == '('){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
152 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
153 ri->tokenType = '(';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
154 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
155 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
156 } else if (ri->ptr[0] == ')') {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
157 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
158 ri->tokenType = ')';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
159 ri->tokenValue = ri->ptr;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
160 return;
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
161 } else if (ri->ptr[0] == ']') {
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
162 ri->ptr++;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
163 ri->tokenType = ']';
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
164 ri->tokenValue = ri->ptr;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
165 return;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
166 } else if (ri->ptr[0] == '|'){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
167 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
168 ri->tokenType = '|';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
169 ri->tokenValue = NULL;
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 } else if (ri->ptr[0] == '*'){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
172 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
173 ri->tokenType = '*';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
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 } else if (ri->ptr[0] == '\\'){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
177 // need more proccesing
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
178 /*
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
179 \277
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
180 \0xa5
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 \\
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
183 \utf-8 etc...
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
184 */
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
185 } else if (ri->ptr[0] == '[') {
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
186 ri->ptr++;
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
187 ri->tokenType = 'c';
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
188 ri->tokenValue = ri->ptr;
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
189 return;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
190 } else {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
191 ri->tokenType = 'a';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
192 ri->tokenValue = ri->ptr;
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
193 if (isalnum(ri->ptr[0])) {
122
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 121
diff changeset
194 ri->ptr++;
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 121
diff changeset
195 }
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
196 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
197 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
198 }
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
199 ri->tokenType = 0;
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
200 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
201 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
202 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
203
130
7925e9abb078 add or flag
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 129
diff changeset
204 // <regexAtom> ::= <literal>|<charClass>|<group>
112
ec485345daf9 some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 108
diff changeset
205 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
206 NodePtr regexAtom(RegexInfoPtr ri) {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
207
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
208 NodePtr n = NULL;
124
masa
parents: 122
diff changeset
209 if (ri->tokenType == 'c') n = charClass(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
210 else if (ri->tokenType == 'a') n = literal(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
211 else if (ri->tokenType == '(') {
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
212 n = regex(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
213 if (ri->tokenType != ')') {
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
214 // error
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
215 }
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
216 token(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
217 }
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
218 if (ri->tokenType == '*') {
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
219 n = createNode(ri,'*',0,n,0);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
220 token(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
221 }
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
222
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
223 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
224 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
225
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
226 // <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
227 NodePtr regex(RegexInfoPtr ri) {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
228 token(ri);
128
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 127
diff changeset
229 NodePtr n = regexAtom(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
230 while (ri->tokenType) {
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
231 if (ri->tokenType == '*') {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
232 n = createNode(ri,'*',0,n,0);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
233 token(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
234 return n;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
235 } else if (ri->tokenType == '|') {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
236 n = createNode(ri,'|',0,n,0);
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
237 NodePtr n1 = regex(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
238 n->right = n1;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
239 } else if (ri->tokenType == ')') {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
240 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
241 } else {
134
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 NodePtr n1 = regexAtom(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
244 n->right = n1;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
245 }
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
246 }
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
247 return n;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
248 }