📄 re.cpp
字号:
#include <iostream>
#include "RE.h"
//////////////////////////////////////////////////////////////////////////
// class Scanner
//
void Scanner::get()
{
int c;
do
{
c = re[off ++];
}while(c == ' ' || c == '\t');
if(('a' <= c && c <= 'z') || ('0' <= c && c <= '9')) // charset
{
token = CHARS;
value = rawMap[c - '0'];
}
else if(c == 'E' || c == 'O' || c == '+' || c == '-' ||
c == '~' || c == '(' || c == ')' || c == '*')
{
token = SYMBOL;
value = c;
}
else if(c == 0)
{
token = ENDSTRING;
}
else
{
token = SCAN_ERR;
value = c;
}
}
// count how many chacters are used, and build the map
void Scanner::analyze()
{
int c, i = 0, counter;
// Map character into rawMap
while( (c = re[i++]) != NULL)
{
if((('a' <= c && c <= 'z') || ('0' <= c && c <= '9')) && rawMap[c - '0'] == 0)
{
rawMap[c - '0'] = 1;
mapSize++;
}
}
if(map != NULL) delete []map;
map = new int[mapSize];
// Build map, map[num] can map num to character
counter = 0;
for(i = 0;i < MAP_SIZE;i ++)
{
if(rawMap[i])
{
map[counter] = i + '0';
rawMap[i] = counter ++;
}
else
{
rawMap[i] = -1;
}
}
}
//////////////////////////////////////////////////////////////////////////
// class RegularExp
//
void RegularExp::parse(void)
{
scanner.get();
nfa = unionSet();
if(scanner.token != scanner.ENDSTRING)
{
error("Error end of the Expression");
}
nfa->makeDeadState();
}
NFA *RegularExp::unionSet()
{
NFA *n1, *n2;
n1 = intersectionSet();
while(scanner.token == scanner.SYMBOL && scanner.value == '+')
{
scanner.get();
n2 = intersectionSet();
n1->unionSet(n2);
delete n2;
}
return n1;
}
NFA *RegularExp::intersectionSet()
{
NFA *n1, *n2;
n1 = join();
while(scanner.token == scanner.SYMBOL && scanner.value == '-')
{
scanner.get();
n2 = join();
n1->intersectionSet(n2);
delete n2;
}
return n1;
}
NFA *RegularExp::join()
{
NFA *n1, *n2;
n1 = complementSet();
while( (scanner.token == scanner.SYMBOL && scanner.value == '(') ||
(scanner.token == scanner.SYMBOL && scanner.value == '~') ||
(scanner.token == scanner.SYMBOL && scanner.value == 'E') ||
(scanner.token == scanner.CHARS) )
{
n2 = complementSet();
n1->join(n2);
delete n2;
}
return n1;
}
NFA *RegularExp::complementSet()
{
NFA *n1;
int counter = 0;
while(scanner.token == scanner.SYMBOL && scanner.value == '~')
{
scanner.get();
++counter;
}
n1 = closure();
if(counter % 2 == 1)
{
n1->complementSet();
}
return n1;
}
NFA *RegularExp::closure()
{
NFA *n1;
int once = 0;
n1 = element();
while(scanner.token == scanner.SYMBOL && scanner.value == '*')
{
scanner.get();
once = 1;
}
if(once)
{
n1->closure();
}
return n1;
}
NFA *RegularExp::element()
{
NFA *n1;
if(scanner.token == scanner.SYMBOL && scanner.value == '(')
{
scanner.get();
n1 = unionSet();
if(scanner.token == scanner.SYMBOL && scanner.value == ')')
{
scanner.get();
}
else
{
error("() must match");
}
}
else if(scanner.token == scanner.SYMBOL && scanner.value == 'E')
{
n1 = new NFA(scanner.mapSize);
n1->emptyStr();
scanner.get();
}
else if(scanner.token == scanner.SYMBOL && scanner.value == 'O')
{
n1 = new NFA(scanner.mapSize);
n1->emptySet();
scanner.get();
}
else if(scanner.token == scanner.CHARS)
{
n1 = new NFA(scanner.mapSize);
n1->alphabet(scanner.value);
scanner.get();
}
else
{
error("Unknow string");
}
return n1;
}
// Deal with the errors occurd in parsing
int RegularExp::error(const char *msg)
{
cerr << msg << endl;
return 0;
}
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -