annotate regexParser/regexParser.cc @ 314:a4484c02cba5

add wordMode in regexParser
author mir3636
date Sat, 07 May 2016 18:38:54 +0900
parents 1188debbef10
children fa590a7272ae
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"
308
1188debbef10 separate CharClass
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 300
diff changeset
6 #include "CharClass.h"
115
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
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 */
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
18 static
116
66c633575b53 remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 115
diff changeset
19 NodePtr allocateNode() {
129
b930be74a16e remove word.h
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 128
diff changeset
20 NodePtr n = NEW(Node);
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
21 n->cc = NULL;
260
3f3c9902bb6d initialize node in allocateNode()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 255
diff changeset
22 n->stateNum = 0;
3f3c9902bb6d initialize node in allocateNode()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 255
diff changeset
23 n->nextStateNum = 0;
3f3c9902bb6d initialize node in allocateNode()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 255
diff changeset
24 n->state = NULL;
3f3c9902bb6d initialize node in allocateNode()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 255
diff changeset
25 n->nextState = NULL;
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
26 n->left = NULL;
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
27 n->right = NULL;
116
66c633575b53 remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 115
diff changeset
28 return n;
66c633575b53 remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 115
diff changeset
29 }
66c633575b53 remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 115
diff changeset
30
187
ef798db705e9 remove some warnings and errors(not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 185
diff changeset
31 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
32 NodePtr n = allocateNode();
125
4d6ac69801ad fix createNode
Nozomi
parents: 122
diff changeset
33 n->tokenType = type;
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
34 n->cc = cc;
183
7ae0a3070647 implement generateTransitionList
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 180
diff changeset
35 n->state = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
36 n->left = left;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
37 n->right = right;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
38 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
39 }
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 // <charClass> ::= '['<literal>'-'<literal>']'
112
ec485345daf9 some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 108
diff changeset
42 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
43 NodePtr charClass(RegexInfoPtr ri) {
300
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
44 CharClassPtr cc = NULL;
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
45 NodePtr n = createNode(ri,'c',cc,0,0);
300
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
46 unsigned char begin = *ri->ptr;
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
47 unsigned char end = *ri->ptr;
130
7925e9abb078 add or flag
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 129
diff changeset
48
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
49 for (ri->ptr++; *ri->ptr && *ri->ptr != ']'; ri->ptr++) {
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
50 if (*ri->ptr == '-') {
300
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
51 end = *(ri->ptr + 1);
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
52 cc = insertCharClass(cc, begin, end);
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
53 ri->ptr++;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
54 continue;
300
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
55 } else {
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
56 cc = insertCharClass(cc, begin, end);
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
57 }
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
58 if (ri->ptr[0] == 0 || ri->ptr[0] == ']') break;
300
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
59 begin = *ri->ptr;
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
60 end = *ri->ptr;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
61 }
300
3e78631a6222 remove ragne list
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 287
diff changeset
62 n->cc = insertCharClass(cc, begin, end);
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
63 // TODO literal support
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
64 // merge rangeList here
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
65 if (*ri->ptr) ri->ptr++;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
66 token(ri);
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
67 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
68 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
69
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
70 // <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
71 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
72 NodePtr literal(RegexInfoPtr ri) {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
73 CharClassPtr cc = createCharClassWord(ri);
314
a4484c02cba5 add wordMode in regexParser
mir3636
parents: 308
diff changeset
74 if (ri->wordMode) {
a4484c02cba5 add wordMode in regexParser
mir3636
parents: 308
diff changeset
75 unsigned char *word = ri->ptr - 1;
a4484c02cba5 add wordMode in regexParser
mir3636
parents: 308
diff changeset
76 while (ri->tokenType == 'a') {
a4484c02cba5 add wordMode in regexParser
mir3636
parents: 308
diff changeset
77 token(ri);
a4484c02cba5 add wordMode in regexParser
mir3636
parents: 308
diff changeset
78 }
a4484c02cba5 add wordMode in regexParser
mir3636
parents: 308
diff changeset
79 cc->cond.w.length = ri->ptr - word;
a4484c02cba5 add wordMode in regexParser
mir3636
parents: 308
diff changeset
80 if (cc->cond.w.length > 1)
a4484c02cba5 add wordMode in regexParser
mir3636
parents: 308
diff changeset
81 cc->cond.w.word = word;
a4484c02cba5 add wordMode in regexParser
mir3636
parents: 308
diff changeset
82 } else
a4484c02cba5 add wordMode in regexParser
mir3636
parents: 308
diff changeset
83 token(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
84 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
85 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
86 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
87
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
88 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
89 void token(RegexInfoPtr ri) {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
90 while (ri->ptr[0] != '\0') {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
91 if (ri->ptr[0] == '('){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
92 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
93 ri->tokenType = '(';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
94 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
95 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
96 } else if (ri->ptr[0] == ')') {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
97 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
98 ri->tokenType = ')';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
99 ri->tokenValue = ri->ptr;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
100 return;
135
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
101 } else if (ri->ptr[0] == ']') {
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
102 ri->ptr++;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
103 ri->tokenType = ']';
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
104 ri->tokenValue = ri->ptr;
e1a262ec75f0 impl charclass
masa
parents: 134
diff changeset
105 return;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
106 } else if (ri->ptr[0] == '|'){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
107 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
108 ri->tokenType = '|';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
109 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
110 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
111 } else if (ri->ptr[0] == '*'){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
112 ri->ptr++;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
113 ri->tokenType = '*';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
114 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
115 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
116 } else if (ri->ptr[0] == '\\'){
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
117 // need more proccesing
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
118 /*
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
119 \277
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
120 \0xa5
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
121 \[
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
122 \\
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
123 \utf-8 etc...
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
124 */
287
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 260
diff changeset
125 ri->ptr++;
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 260
diff changeset
126 ri->tokenType = 'a';
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 260
diff changeset
127 ri->tokenValue = ri->ptr;
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 260
diff changeset
128 ri->ptr++;
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 260
diff changeset
129 return;
142
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
130 } else if (ri->ptr[0] == '[') {
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
131 ri->ptr++;
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
132 ri->tokenType = 'c';
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
133 ri->tokenValue = ri->ptr;
de0f332d560c insert charClassMerge function
masa
parents: 136
diff changeset
134 return;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
135 } else {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
136 ri->tokenType = 'a';
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
137 ri->tokenValue = ri->ptr;
287
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 260
diff changeset
138 // if (isalnum(ri->ptr[0])) {
122
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 121
diff changeset
139 ri->ptr++;
287
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 260
diff changeset
140 // }
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
141 return;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
142 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
143 }
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
144 ri->tokenType = 0;
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
145 ri->tokenValue = NULL;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
146 return;
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
130
7925e9abb078 add or flag
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 129
diff changeset
149 // <regexAtom> ::= <literal>|<charClass>|<group>
112
ec485345daf9 some function use static
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 108
diff changeset
150 static
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
151 NodePtr regexAtom(RegexInfoPtr ri) {
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 NodePtr n = NULL;
124
masa
parents: 122
diff changeset
154 if (ri->tokenType == 'c') n = charClass(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
155 else if (ri->tokenType == 'a') n = literal(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
156 else if (ri->tokenType == '(') {
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
157 n = regex(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
158 if (ri->tokenType != ')') {
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
159 // error
287
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 260
diff changeset
160 fprintf(stderr,"unclosed ')' before %s \n", ri->ptr);
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents: 260
diff changeset
161 return createNode(ri,0,0,0,0);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
162 }
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
163 token(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
164 }
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
165 if (ri->tokenType == '*') {
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
166 n = createNode(ri,'*',0,n,0);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
167 token(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
168 }
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
169 return n;
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
170 }
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
171
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
172 // <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
173 NodePtr regex(RegexInfoPtr ri) {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
174 token(ri);
128
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 127
diff changeset
175 NodePtr n = regexAtom(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
176 while (ri->tokenType) {
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
177 if (ri->tokenType == '*') {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
178 n = createNode(ri,'*',0,n,0);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
179 token(ri);
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
180 return n;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
181 } else if (ri->tokenType == '|') {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
182 n = createNode(ri,'|',0,n,0);
133
ccc673449351 Look ahead '*'
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 132
diff changeset
183 NodePtr n1 = regex(ri);
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
184 n->right = n1;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
185 } else if (ri->tokenType == ')') {
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
186 return n;
255
61d4d466e64c fix Makefile
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 214
diff changeset
187 } else if (ri->tokenType == ']') {
61d4d466e64c fix Makefile
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 214
diff changeset
188 // error
61d4d466e64c fix Makefile
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 214
diff changeset
189 return n;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
190 } else {
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
191 n = createNode(ri,'+',0,n,0);
180
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
192 NodePtr n1 = regexAtom(ri);
d97bcab546e8 implement getNext
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 179
diff changeset
193 n->right = n1;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
194 }
134
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
195 }
dbafc753078e fix concatination & selection & grouping
masa
parents: 133
diff changeset
196 return n;
82
1d9bbf922bb6 add createRegexTree.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
197 }