Mercurial > hg > Applications > Grep
annotate c/regexParser/main.cc @ 58:4053c3e0fa7f
implement group()
author | Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp> |
---|---|
date | Fri, 12 Jun 2015 19:02:00 +0900 |
parents | 71b497d25273 |
children | af189c727733 |
rev | line source |
---|---|
55 | 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>')' |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
6 * <or> ::= <regex>'|'<regex> |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
7 * <*> ::= <regex>'*' |
58
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
8 * <regex> ::= <string>|<or>|<charClass>|<group>|<*> |
55 | 9 */ |
10 | |
45 | 11 #include <stdio.h> |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
12 #include <stdlib.h> |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
13 #include <string.h> |
57
71b497d25273
fix literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
56
diff
changeset
|
14 |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
15 typedef struct node { |
58
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
16 struct node *self; |
57
71b497d25273
fix literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
56
diff
changeset
|
17 char character; |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
18 struct node *left; |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
19 struct node *right; |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
20 } Node, *NodePtr; |
52
a2826bf4e80a
remove magic number
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
51
diff
changeset
|
21 |
58
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
22 char *ptr; |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
23 NodePtr regexHeadNode; |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
24 |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
25 NodePtr charClass(); |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
26 NodePtr string(); |
58
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
27 NodePtr group(); |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
28 NodePtr _or(); |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
29 NodePtr asterisk(); |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
30 NodePtr regex(); |
57
71b497d25273
fix literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
56
diff
changeset
|
31 NodePtr createNode(char,NodePtr,NodePtr); |
55 | 32 |
57
71b497d25273
fix literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
56
diff
changeset
|
33 NodePtr createNode(char character, NodePtr left, NodePtr right) { |
71b497d25273
fix literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
56
diff
changeset
|
34 NodePtr n = (NodePtr)malloc(sizeof(Node)); |
58
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
35 n->self = n; |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
36 n->character = character; |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
37 n->left = left; |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
38 n->right = right; |
58
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
39 |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
40 return n; |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
41 } |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
42 |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
43 // <charClass> ::= '['<literal>'-'<literal>']' |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
44 NodePtr charClass() { |
58
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
45 ptr++; |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
46 NodePtr n = createNode(0,0,0); |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
47 return n; |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
48 } |
55 | 49 |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
50 // <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
|
51 NodePtr literal() { |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
52 char c = *ptr; |
57
71b497d25273
fix literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
56
diff
changeset
|
53 NodePtr n = createNode(c,0,0); |
71b497d25273
fix literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
56
diff
changeset
|
54 ptr++; |
71b497d25273
fix literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
56
diff
changeset
|
55 return n; |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
56 } |
55 | 57 |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
58 // <string> ::= <literal><literal>* |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
59 NodePtr string() { |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
60 char c = *ptr; |
58
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
61 NodePtr n = NULL; |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
62 printf("%c\n",c); |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
63 if (('a'<=c && c<='z')||('A'<=c && c<='Z')||('0'<=c && c<='9')) { |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
64 n = createNode(0,literal(),string()); |
55 | 65 } else { |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
66 n = createNode(0,0,0); |
55 | 67 } |
58
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
68 return n; |
55 | 69 } |
70 | |
58
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
71 // <group> ::= '('<regex>')' | '('<regex>'|'<regex>')' |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
72 NodePtr group() { |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
73 NodePtr n; |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
74 if (*ptr == ')') { |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
75 n = createNode(0,0,0); |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
76 ptr++; |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
77 } else { |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
78 ptr++; |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
79 n = regex(); |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
80 } |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
81 |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
82 return n; |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
83 } |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
84 |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
85 |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
86 // <or> ::= <regex>'|'<regex> |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
87 NodePtr _or() { |
58
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
88 NodePtr n = createNode('|',regexHeadNode,regex()); |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
89 return n; |
55 | 90 } |
91 | |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
92 // <*> ::= <regex>'*' |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
93 NodePtr asterisk() { |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
94 |
55 | 95 } |
96 | |
58
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
97 // <regex> ::= <string>|<or>|<charClass>|<group>|<*> |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
98 NodePtr regex() { |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
99 |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
100 NodePtr n; |
49
f76fe618d5a7
implement bitset
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
48
diff
changeset
|
101 |
58
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
102 while (*ptr != '\0') { |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
103 if ((*ptr == '(') || (*ptr == ')')) { |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
104 n = group(); |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
105 } else if (*ptr == '[') { |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
106 n = charClass(); |
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
107 } else if (*ptr == '|'){ |
57
71b497d25273
fix literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
56
diff
changeset
|
108 n = _or(); |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
109 } else { |
57
71b497d25273
fix literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
56
diff
changeset
|
110 n = string(); |
58
4053c3e0fa7f
implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
57
diff
changeset
|
111 regexHeadNode = n; |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
112 } |
55 | 113 } |
50
bb0e88fbbe02
impl bitGet and bitSet
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
49
diff
changeset
|
114 |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
115 return n; |
55 | 116 } |
117 | |
118 int | |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
119 main(int argc, char **argv) |
55 | 120 { |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
121 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
|
122 if (strcmp(argv[i],"-regex") == 0) { |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
123 ptr = argv[i+1]; i++; |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
124 } |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
125 } |
55 | 126 |
56
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
127 printf("regex : %s\n",ptr); |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
128 NodePtr n = regex(); |
8901bc071d33
implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
129 |
45 | 130 return 0; |
131 } |