annotate regexParser/regexParser.h @ 186:3e8aae8beba9 pairPro

fix createTransitionGenerator()
author Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
date Thu, 24 Dec 2015 20:37:29 +0900
parents 1da1b2eacb84
children ef798db705e9
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
1 #include "error.h"
126
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
2
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
3 #define NEW(type) (type*)malloc(sizeof(type))
175
3be0fbcd4b52 implement createTransitionGenerator
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 167
diff changeset
4 #define SYNTAX_NODENUMBER 100
126
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
5
144
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 142
diff changeset
6 typedef struct bitVector {
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 142
diff changeset
7 unsigned long bitContainer;
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 142
diff changeset
8 }BitVector,*BitVectorPtr;
d8a4922eceae remove some errors (not working)
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 142
diff changeset
9
184
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
10 typedef struct stateStack {
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
11 BitVector state;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
12 struct stateStack *next;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
13 } StateStack, *StateStackPtr;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
14
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
15 typedef struct transitionGenerator {
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
16 long stateMax;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
17 StateStack stack;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
18 StatePtr state;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
19 StatePtr stateArray;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
20 StatePtr currentState;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
21 } TransitionGenerator, *TransitionGeneratorPtr;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
22
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
23 typedef struct tgValue {
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
24 bool asterisk;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
25 int stateBegin;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
26 int stateEnd;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
27 TransitionGeneratorPtr tg;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
28 } TGValue, *TGValuePtr;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
29
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
30 enum charClassStackState {
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
31 LEFT,
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
32 SELF,
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
33 RIGHT
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
34 };
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
35
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
36 typedef struct charClassStack {
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
37 charClassStackState turn;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
38 CharClassPtr cc;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
39 struct charClassStack *next;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
40 } CharClassStack, *CharClassStackPtr;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
41
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
42 typedef struct charClassWalker {
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
43 CharClassStackPtr stack;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
44 CharClassPtr next;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
45 } CharClassWalker, *CharClassWalkerPtr;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
46
129
b930be74a16e remove word.h
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 126
diff changeset
47 typedef struct word {
b930be74a16e remove word.h
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 126
diff changeset
48 unsigned char *word;
b930be74a16e remove word.h
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 126
diff changeset
49 int length;
b930be74a16e remove word.h
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 126
diff changeset
50 } Word, *WordPtr;
b930be74a16e remove word.h
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 126
diff changeset
51
126
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
52 typedef struct utf8Range {
142
de0f332d560c insert charClassMerge function
masa
parents: 133
diff changeset
53 unsigned long begin;
de0f332d560c insert charClassMerge function
masa
parents: 133
diff changeset
54 unsigned long end;
de0f332d560c insert charClassMerge function
masa
parents: 133
diff changeset
55 struct utf8Range *next; // only used in the parser.
126
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
56 } RangeList , *RangeListPtr;
639b0b437ebf remove error (do not work)
masa
parents: 125
diff changeset
57
178
5e8c6857934c implement charClassMerge
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 175
diff changeset
58 typedef struct condition {
142
de0f332d560c insert charClassMerge function
masa
parents: 133
diff changeset
59 RangeList range;
de0f332d560c insert charClassMerge function
masa
parents: 133
diff changeset
60 Word w;
116
66c633575b53 remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 115
diff changeset
61 } Condition, *ConditionList;
66c633575b53 remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 115
diff changeset
62
77
7f53a587bf97 add regexParser.h
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
63 typedef struct charClass {
111
1d30f70702df add determinize.cc and transition.cc
masa
parents: 110
diff changeset
64 unsigned char type;
115
ca30f8334741 rename createRegexParser.cc to regexParser.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 111
diff changeset
65 struct charClass *left;
ca30f8334741 rename createRegexParser.cc to regexParser.cc
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 111
diff changeset
66 struct charClass *right;
142
de0f332d560c insert charClassMerge function
masa
parents: 133
diff changeset
67 Condition cond;
de0f332d560c insert charClassMerge function
masa
parents: 133
diff changeset
68 BitVector nextState;
77
7f53a587bf97 add regexParser.h
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
69 } CharClass, *CharClassPtr;
7f53a587bf97 add regexParser.h
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
70
7f53a587bf97 add regexParser.h
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
71 typedef struct node {
89
50a146c05192 add NodeNumber in Regex Parser tree
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 85
diff changeset
72 unsigned char tokenType;
111
1d30f70702df add determinize.cc and transition.cc
masa
parents: 110
diff changeset
73 unsigned long nodeNumber;
1d30f70702df add determinize.cc and transition.cc
masa
parents: 110
diff changeset
74 CharClassPtr cc;
183
7ae0a3070647 implement generateTransitionList
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 180
diff changeset
75 int stateNum;
7ae0a3070647 implement generateTransitionList
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 180
diff changeset
76 int nextStateNum;
7ae0a3070647 implement generateTransitionList
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 180
diff changeset
77 StatePtr state;
77
7f53a587bf97 add regexParser.h
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
78 struct node *left;
7f53a587bf97 add regexParser.h
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
79 struct node *right;
7f53a587bf97 add regexParser.h
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents:
diff changeset
80 } Node, *NodePtr;
116
66c633575b53 remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 115
diff changeset
81
184
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
82 typedef struct state {
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
83 BitVector bitState;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
84 CharClassPtr cc;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
85 NodePtr node;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
86 struct state *next;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
87 } State, *StatePtr;
1da1b2eacb84 gather struct
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 183
diff changeset
88
116
66c633575b53 remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 115
diff changeset
89 typedef struct regexInfo {
66c633575b53 remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 115
diff changeset
90 unsigned char *ptr;
66c633575b53 remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 115
diff changeset
91 unsigned char tokenType;
118
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 116
diff changeset
92 unsigned char *tokenValue;
178
5e8c6857934c implement charClassMerge
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 175
diff changeset
93 int stateNumber;
116
66c633575b53 remove error and warning
Masataka Kohagura <kohagura@cr.ie.u-ryukyu.ac.jp>
parents: 115
diff changeset
94 } RegexInfo, *RegexInfoPtr;
155
6cd0141bed6c impl charClassMerge
masa
parents: 145
diff changeset
95
6cd0141bed6c impl charClassMerge
masa
parents: 145
diff changeset
96 CharClassPtr createCharClassRange(unsigned long begin, unsigned long end, CharClassPtr left, CharClassPtr right);