annotate c/regexParser/main.cc @ 75:6541eae41a73

fix group function
author masa
date Fri, 28 Aug 2015 18:40:33 +0900
parents b11dfc48f698
children d98a036441e2
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>']'
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
4 * <group> ::= '('<regex>')'
66
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
5 * <regexAtom> ::= <literal>|<charClass>|<group>
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
6 * <regex> ::= <regexAtom>|<regexAtom>'*'|<regexAtom>'|'<regex>|<regexAtom><regex>
55
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
7 */
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
8
45
89a198fa6b23 add dfrTobin
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
9 #include <stdio.h>
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
10 #include <stdlib.h>
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
11 #include <string.h>
57
71b497d25273 fix literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 56
diff changeset
12
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
13 typedef struct charClass {
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
14 unsigned char table[256];
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
15 struct utf8Range {
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
16 unsigned char *begin;
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
17 unsigned char *end;
64
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
18 struct utf8Range *next;
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
19 } *rangeList;
64
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
20 } CharClass, *CharClassPtr;
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
21
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
22 typedef struct node {
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
23 unsigned char type;
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
24 union value {
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
25 charClass *cc;
67
4842ca2cf8ee print character
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 66
diff changeset
26 unsigned char character;
4842ca2cf8ee print character
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 66
diff changeset
27 } Value, *ValuePtr;
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
28 struct node *self;
68
d27b3af1fe75 remove string()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 67
diff changeset
29 struct node *parent;
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();
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
40 NodePtr group();
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
41 NodePtr orexp();
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
42 NodePtr asterisk();
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
43 NodePtr regex();
67
4842ca2cf8ee print character
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 66
diff changeset
44 NodePtr createNode(unsigned char,NodePtr,NodePtr);
64
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
45 extern void token();
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
46 extern NodePtr regexAtom();
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
47
55
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
48
62
a49b4a8b8c14 implement isLiteral
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 61
diff changeset
49 bool isLiteral(char c) {
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
50 if (*ptr > 0x7f) return true;
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
51 else if (*ptr == '(') return false;
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 return true;
62
a49b4a8b8c14 implement isLiteral
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 61
diff changeset
56 }
a49b4a8b8c14 implement isLiteral
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 61
diff changeset
57
67
4842ca2cf8ee print character
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 66
diff changeset
58 NodePtr createNode(unsigned char character, NodePtr left, NodePtr right) {
57
71b497d25273 fix literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 56
diff changeset
59 NodePtr n = (NodePtr)malloc(sizeof(Node));
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
60 n->self = n;
67
4842ca2cf8ee print character
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 66
diff changeset
61 n->Value.character = character;
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
62 n->left = left;
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
63 n->right = right;
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
64 return n;
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
65 }
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
66
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
67 // <charClass> ::= '['<literal>'-'<literal>']'
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
68 NodePtr charClass() {
62
a49b4a8b8c14 implement isLiteral
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 61
diff changeset
69 NodePtr n = (NodePtr)malloc(sizeof(Node));
67
4842ca2cf8ee print character
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 66
diff changeset
70 unsigned char startChar = *ptr;
4842ca2cf8ee print character
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 66
diff changeset
71 while (*ptr == '-') {
4842ca2cf8ee print character
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 66
diff changeset
72 ptr++;
4842ca2cf8ee print character
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 66
diff changeset
73 }
4842ca2cf8ee print character
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 66
diff changeset
74 unsigned char endChar = *ptr;
4842ca2cf8ee print character
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 66
diff changeset
75 unsigned char *charTable = (unsigned char*)malloc(sizeof(char)*256);
4842ca2cf8ee print character
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 66
diff changeset
76
62
a49b4a8b8c14 implement isLiteral
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 61
diff changeset
77 return n;
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
78 }
55
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
79
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
80 // <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
81 NodePtr literal() {
65
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 64
diff changeset
82 NodePtr n = createNode(*ptr,0,0);
57
71b497d25273 fix literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 56
diff changeset
83 ptr++;
71b497d25273 fix literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 56
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
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
87 // <group> ::= '('<regex>')'
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
88 NodePtr group() {
75
6541eae41a73 fix group function
masa
parents: 74
diff changeset
89 return regex();
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
90 }
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
91
55
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
92
71
58d2b10988c9 move ptr++ in token()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 70
diff changeset
93
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
94 void token() {
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
95 while (*ptr != '\0') {
74
b11dfc48f698 fix grouping function
masa
parents: 73
diff changeset
96 if (*ptr == '('){
71
58d2b10988c9 move ptr++ in token()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 70
diff changeset
97 ptr++;
70
87dff3a124ab resolve segmentation fault(But not correct performance)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 69
diff changeset
98 tokenType = '(';
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
99 tokenValue = 0;
67
4842ca2cf8ee print character
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 66
diff changeset
100 if (ptr[1] == ')') {
4842ca2cf8ee print character
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 66
diff changeset
101 ptr++;
4842ca2cf8ee print character
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 66
diff changeset
102 }
4842ca2cf8ee print character
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 66
diff changeset
103 return;
74
b11dfc48f698 fix grouping function
masa
parents: 73
diff changeset
104 } else if (*ptr == ')') {
b11dfc48f698 fix grouping function
masa
parents: 73
diff changeset
105 ptr++;
b11dfc48f698 fix grouping function
masa
parents: 73
diff changeset
106 tokenType = ')';
b11dfc48f698 fix grouping function
masa
parents: 73
diff changeset
107 tokenValue = *ptr;
b11dfc48f698 fix grouping function
masa
parents: 73
diff changeset
108 return;
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
109 } else if (*ptr == '[') {
71
58d2b10988c9 move ptr++ in token()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 70
diff changeset
110 ptr++;
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
111 tokenType = '[';
64
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
112 tokenValue = *ptr;
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
113 if (ptr[1] == ']') {
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
114 ptr++;
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
115 }
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
116 return;
58
4053c3e0fa7f implement group()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 57
diff changeset
117 } else if (*ptr == '|'){
71
58d2b10988c9 move ptr++ in token()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 70
diff changeset
118 ptr++;
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
119 tokenType = '|';
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
120 tokenValue = 0;
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
121 return;
60
8616a045a7f4 impl asterisk
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 59
diff changeset
122 } else if (*ptr == '*'){
71
58d2b10988c9 move ptr++ in token()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 70
diff changeset
123 ptr++;
63
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
124 tokenType = '*';
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
125 tokenValue = 0;
8fd3d35e9861 add token function
masa
parents: 62
diff changeset
126 return;
73
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 72
diff changeset
127 } else if (*ptr == '\\'){
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 72
diff changeset
128 // need more proccesing
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 72
diff changeset
129 /*
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 72
diff changeset
130 \277
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 72
diff changeset
131 \0xa5
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 72
diff changeset
132 \[
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 72
diff changeset
133 \\
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 72
diff changeset
134 \utf-8 etc...
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 72
diff changeset
135 */
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 72
diff changeset
136 } else {
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 72
diff changeset
137 tokenType = 'a';
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 72
diff changeset
138 tokenValue = *ptr;
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 72
diff changeset
139 return;
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
140 }
55
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
141 }
71
58d2b10988c9 move ptr++ in token()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 70
diff changeset
142
58d2b10988c9 move ptr++ in token()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 70
diff changeset
143 tokenType = 0;
58d2b10988c9 move ptr++ in token()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 70
diff changeset
144 tokenValue = 0;
58d2b10988c9 move ptr++ in token()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 70
diff changeset
145 return;
55
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
146 }
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
147
64
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
148 // <regexAtom> ::= <literal>|<charClass>|<group>
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
149 NodePtr regexAtom() {
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
150
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
151 token();
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
152 NodePtr n = NULL;
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
153 if (tokenType == 'a') n = literal();
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
154 else if (tokenType == '[') n = charClass();
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
155 else if (tokenType == '(') n = group();
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
156
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
157 return n;
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
158 }
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
159
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
160 // <regex> ::= <regexAtom>|<regexAtom>'*'|<regexAtom>'|'<regex>|<regexAtom><regex>
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
161 NodePtr regex() {
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
162 NodePtr n = regexAtom();
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
163 while (*ptr) {
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 if (tokenType == '*') {
71
58d2b10988c9 move ptr++ in token()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 70
diff changeset
166 n = createNode('*',n,0);
64
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
167 } else if (tokenType == '|') {
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
168 NodePtr n1 = regex();
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
169 n = createNode('|',n,n1);
74
b11dfc48f698 fix grouping function
masa
parents: 73
diff changeset
170 } else if (tokenType == ')') {
b11dfc48f698 fix grouping function
masa
parents: 73
diff changeset
171 return n;
64
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
172 } else {
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
173 NodePtr n1 = regex();
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
174 n = createNode('+',n,n1);
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
175 }
67
4842ca2cf8ee print character
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 66
diff changeset
176 } return n;
68
d27b3af1fe75 remove string()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 67
diff changeset
177 }
d27b3af1fe75 remove string()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 67
diff changeset
178
69
eecddded9b91 implement printTree
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 68
diff changeset
179 void descendTree(NodePtr n) {
eecddded9b91 implement printTree
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 68
diff changeset
180 static int d = 0;
68
d27b3af1fe75 remove string()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 67
diff changeset
181 if (n->right != NULL) {
d27b3af1fe75 remove string()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 67
diff changeset
182 d++;
69
eecddded9b91 implement printTree
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 68
diff changeset
183 descendTree(n->right);
eecddded9b91 implement printTree
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 68
diff changeset
184 d--;
eecddded9b91 implement printTree
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 68
diff changeset
185 }
eecddded9b91 implement printTree
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 68
diff changeset
186 printf("%*c%c\n",d*4, ' ',n->Value.character);
eecddded9b91 implement printTree
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 68
diff changeset
187 if (n->left != NULL) {
68
d27b3af1fe75 remove string()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 67
diff changeset
188 d++;
69
eecddded9b91 implement printTree
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 68
diff changeset
189 descendTree(n->left);
eecddded9b91 implement printTree
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 68
diff changeset
190 d--;
68
d27b3af1fe75 remove string()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 67
diff changeset
191 }
d27b3af1fe75 remove string()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 67
diff changeset
192 }
d27b3af1fe75 remove string()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 67
diff changeset
193
66
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
194 void printTree(NodePtr n) {
72
c4b934048e2a remove printNodeDate
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 71
diff changeset
195 puts("---Print Node----");
69
eecddded9b91 implement printTree
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 68
diff changeset
196 descendTree(n);
72
c4b934048e2a remove printNodeDate
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 71
diff changeset
197 puts("-----------------");
66
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
198 }
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
62
a49b4a8b8c14 implement isLiteral
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 61
diff changeset
201 int main(int argc, char **argv)
55
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
202 {
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
203 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
204 if (strcmp(argv[i],"-regex") == 0) {
64
e0ad6c145f89 remove some errors
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 63
diff changeset
205 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
206 }
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
207 }
55
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 54
diff changeset
208
56
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
209 printf("regex : %s\n",ptr);
8901bc071d33 implement string() and literal()
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 55
diff changeset
210 NodePtr n = regex();
66
f8fb3b463f70 fix when '|' come procces
Masataka Kohagura <e085726@ie.u-ryukyu.ac.jp>
parents: 65
diff changeset
211 printTree(n);
45
89a198fa6b23 add dfrTobin
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
212 return 0;
89a198fa6b23 add dfrTobin
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
213 }