annotate c/regexParser/main.cc @ 66:f8fb3b463f70

fix when '|' come procces
author Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
date Tue, 04 Aug 2015 16:41:41 +0900
parents 20b7d4e958bb
children 4842ca2cf8ee
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
55
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
1 /*
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
2 * <literal> ::= [a-z][A-Z][0-9]
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
3 * <charClass> ::= '['<literal>'-'<literal>']'
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
4 * <string> ::= <literal><literal>*
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
5 * <group> ::= '('<regex>')'
66
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
6 * <regexAtom> ::= <literal>|<charClass>|<group>
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
7 * <regex> ::= <regexAtom>|<regexAtom>'*'|<regexAtom>'|'<regex>|<regexAtom><regex>
55
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
8 */
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
9
45
89a198fa6b23 add dfrTobin
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
10 #include <stdio.h>
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
11 #include <stdlib.h>
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
12 #include <string.h>
57
71b497d25273 fix literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 56
diff changeset
13
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
14 typedef struct charClass {
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
15 unsigned char table[256];
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
16 struct utf8Range {
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
17 unsigned char *begin;
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
18 unsigned char *end;
64
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
19 struct utf8Range *next;
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
20 } *rangeList;
64
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
21 } CharClass, *CharClassPtr;
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
22
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
23 typedef struct node {
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
24 unsigned char type;
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
25 union value {
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
26 charClass *cc;
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
27 unsigned char *string;
64
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
28 };
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
29 struct node *self;
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
30 struct node *left;
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
31 struct node *right;
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
32 } Node, *NodePtr;
52
a2826bf4e80a remove magic number
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 51
diff changeset
33
64
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
34 unsigned char *ptr;
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
35 unsigned char tokenType;
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
36 int tokenValue;
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
37 NodePtr regexHeadNode;
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
38
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
39 NodePtr charClass();
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
40 NodePtr string();
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
41 NodePtr group();
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
42 NodePtr orexp();
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
43 NodePtr asterisk();
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
44 NodePtr regex();
57
71b497d25273 fix literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 56
diff changeset
45 NodePtr createNode(char,NodePtr,NodePtr);
64
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
46 extern void token();
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
47 extern NodePtr regexAtom();
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
48
55
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
49
62
a49b4a8b8c14 implement isLiteral
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 61
diff changeset
50 bool isLiteral(char c) {
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
51 if (*ptr > 0x7f) return true;
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
52 else if (*ptr == '(') return false;
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
53 else if (*ptr == '[') return false;
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
54 else if (*ptr == '|') return false;
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
55 else if (*ptr == '*') return false;
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
56 return true;
62
a49b4a8b8c14 implement isLiteral
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 61
diff changeset
57 }
a49b4a8b8c14 implement isLiteral
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 61
diff changeset
58
60
8616a045a7f4 impl asterisk
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 59
diff changeset
59 void printNodeDate(NodePtr n) {
8616a045a7f4 impl asterisk
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 59
diff changeset
60 puts("---------------------");
64
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
61 // printf("Self Node char : %c\n", n->character);
60
8616a045a7f4 impl asterisk
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 59
diff changeset
62 printf("Self Node addr : %p\n", n->self);
8616a045a7f4 impl asterisk
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 59
diff changeset
63 printf("left Node addr : %p\n", n->left);
8616a045a7f4 impl asterisk
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 59
diff changeset
64 printf("right Node addr : %p\n", n->right);
8616a045a7f4 impl asterisk
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 59
diff changeset
65 puts("---------------------");
8616a045a7f4 impl asterisk
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 59
diff changeset
66 puts("");
8616a045a7f4 impl asterisk
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 59
diff changeset
67 }
8616a045a7f4 impl asterisk
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 59
diff changeset
68
57
71b497d25273 fix literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 56
diff changeset
69 NodePtr createNode(char character, NodePtr left, NodePtr right) {
71b497d25273 fix literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 56
diff changeset
70 NodePtr n = (NodePtr)malloc(sizeof(Node));
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
71 n->self = n;
64
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
72 // n->character = character;
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
73 n->left = left;
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
74 n->right = right;
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
75
60
8616a045a7f4 impl asterisk
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 59
diff changeset
76 printNodeDate(n);
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
77 return n;
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
78 }
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
79
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
80 // <charClass> ::= '['<literal>'-'<literal>']'
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
81 NodePtr charClass() {
62
a49b4a8b8c14 implement isLiteral
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 61
diff changeset
82 ptr++;
a49b4a8b8c14 implement isLiteral
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 61
diff changeset
83 NodePtr n = (NodePtr)malloc(sizeof(Node));
a49b4a8b8c14 implement isLiteral
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 61
diff changeset
84 return n;
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
85 }
55
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
86
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
87 // <literal> ::= [a-z][A-Z][0-9]
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
88 NodePtr literal() {
65
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 64
diff changeset
89 NodePtr n = createNode(*ptr,0,0);
57
71b497d25273 fix literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 56
diff changeset
90 ptr++;
71b497d25273 fix literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 56
diff changeset
91 return n;
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
92 }
55
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
93
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
94 // <string> ::= <literal><literal>*
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
95 NodePtr string() {
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
96 char c = *ptr;
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
97 NodePtr n = NULL;
61
67cade0e35b0 impl isLiteral
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 60
diff changeset
98 if (isLiteral(c)) {
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
99 n = createNode(0,literal(),string());
55
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
100 } else {
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
101 n = createNode(0,0,0);
55
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
102 }
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
103 return n;
55
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
104 }
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
105
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
106 // <group> ::= '('<regex>')'
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
107 NodePtr group() {
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
108 token();
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
109 NodePtr n = regex();
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
110 token();
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
111 if (*ptr == ')') {
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
112 n = createNode('(',n,0);
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
113 } else {
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
114 // ) reqiured
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
115 }
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
116 return n;
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
117 }
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
118
55
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
119
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
120 void token() {
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
121 while (*ptr != '\0') {
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
122 if ((*ptr == '(') || (*ptr == ')')) {
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
123 tokenType = *ptr++;
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
124 tokenValue = 0;
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
125 return ;
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
126 } else if (*ptr == '[') {
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
127 tokenType = '[';
64
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
128 tokenValue = *ptr;
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
129 if (ptr[1] == ']') {
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
130 ptr++;
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
131 }
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
132 while (*ptr != ']') ptr++;
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
133 ptr++;
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
134 return;
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
135 } else if (*ptr == '|'){
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
136 tokenType = '|';
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
137 tokenValue = 0;
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
138 return;
60
8616a045a7f4 impl asterisk
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 59
diff changeset
139 } else if (*ptr == '*'){
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
140 tokenType = '*';
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
141 tokenValue = 0;
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
142 return;
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
143 }
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
144
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
145 tokenType = 'a';
64
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
146 tokenValue = *ptr;
65
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 64
diff changeset
147 return;
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
148
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
149 if (*ptr == '\\') ptr++; // need more proccesing
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
150 /*
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
151 \277
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
152 \0xa5
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
153 \[
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
154 \\
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
155 \utf-8 etc...
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
156 */
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
157
55
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
158 }
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
159 }
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
160
64
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
161 // <regexAtom> ::= <literal>|<charClass>|<group>
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
162 NodePtr regexAtom() {
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
163
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
164 token();
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
165 NodePtr n = NULL;
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
166 if (tokenType == 'a') n = literal();
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
167 else if (tokenType == '[') n = charClass();
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
168 else if (tokenType == '(') n = group();
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
169
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
170 return n;
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
171 }
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
172
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
173 // <regex> ::= <regexAtom>|<regexAtom>'*'|<regexAtom>'|'<regex>|<regexAtom><regex>
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
174 NodePtr regex() {
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
175 NodePtr n = regexAtom();
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
176 while (*ptr) {
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
177 token();
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
178 if (tokenType == '*') {
66
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
179 n = createNode('*',n,0); ptr++;
64
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
180 } else if (tokenType == '|') {
66
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
181 ptr++;
64
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
182 NodePtr n1 = regex();
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
183 n = createNode('|',n,n1);
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
184 } else {
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
185 NodePtr n1 = regex();
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
186 n = createNode('+',n,n1);
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
187 }
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
188 }
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
189 return n;
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
190 }
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
191
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
192
66
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
193 /*
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
194 * e.g.
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
195 *
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
196 * % ./regexParser -regex abc
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
197 *
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
198 * #-c
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
199 * #-+
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
200 * # #-b
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
201 * +
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
202 * #-a
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
203 *
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
204 * % ./regexParser -regex (a*|bc)d
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
205 *
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
206 *
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
207 * #-d
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
208 * +
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
209 * # #-c
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
210 * # #-+
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
211 * # # #-b
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
212 * #-|
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
213 * #
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
214 * #-*
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
215 * #-a
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
216 *
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
217 */
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
218 void printTree(NodePtr n) {
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
219
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
220 }
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
221
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
222
62
a49b4a8b8c14 implement isLiteral
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 61
diff changeset
223 int main(int argc, char **argv)
55
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
224 {
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
225 for (int i = 1; i < argc; i++) {
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
226 if (strcmp(argv[i],"-regex") == 0) {
64
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
227 ptr = (unsigned char*)argv[i+1]; i++;
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
228 }
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
229 }
55
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
230
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
231 printf("regex : %s\n",ptr);
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
232 NodePtr n = regex();
66
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
233 printTree(n);
45
89a198fa6b23 add dfrTobin
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
234 return 0;
89a198fa6b23 add dfrTobin
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
235 }