状态机学习（四）解析四则运算式 词法分析

#include <string>
#include <iostream>
using namespace std;

char* testStr = "12+345-6*8/9";

typedef enum {
BAD_TOKEN,
NUMBER_TOKEN,
ADD_OPERATOR_TOKEN,
SUB_OPERATOR_TOKEN,
MUL_OPERATOR_TOKEN,
DIV_OPERATOR_TOKEN,
END_OF_LINE_TOKEN
} TokenKind;

typedef struct {
TokenKind kind;
unsigned value;
string s;
} Token;

typedef enum {
INITIAL_STATUS,
IN_INT_PART_STATUS
} LexerStatus;
//==========================================

static size_t gParseIndex = 0;

bool GetToken(Token& token,const string& strContent) {
LexerStatus status = INITIAL_STATUS;
token.kind = BAD_TOKEN;
char currentChar = '\0';
size_t numBegIndex = 0;
size_t numEndIndex = 0;
while (gParseIndex < strContent.size())
{
//数字状态跳出
if(status == IN_INT_PART_STATUS &&
!isdigit(strContent[gParseIndex]) ) {
numEndIndex = gParseIndex;
token.kind = NUMBER_TOKEN;
token.s = strContent.substr(numBegIndex, numEndIndex- numBegIndex);
return true;
}
// 空格略过 换行结束函数
if (isspace(strContent[gParseIndex])) {
if (strContent[gParseIndex] == '\r' ||
strContent[gParseIndex] == '\n') {
token.kind = END_OF_LINE_TOKEN;
return true;
}
else {
gParseIndex++;
continue;
}
}

if (isdigit(strContent[gParseIndex])) {
if (status != IN_INT_PART_STATUS) {
status = IN_INT_PART_STATUS;
numBegIndex = gParseIndex;
}
gParseIndex++;
continue;
}

if (strContent[gParseIndex] == '+') {
token.kind = ADD_OPERATOR_TOKEN;
token.s = strContent[gParseIndex];
gParseIndex++;
return true;
}
else if (strContent[gParseIndex] == '-') {
token.kind = SUB_OPERATOR_TOKEN;
token.s = strContent[gParseIndex];
gParseIndex++;
return true;
}
else if (strContent[gParseIndex] == '*') {
token.kind = MUL_OPERATOR_TOKEN;
token.s = strContent[gParseIndex];
gParseIndex++;
return true;
}
else if (strContent[gParseIndex] == '/') {
token.kind = DIV_OPERATOR_TOKEN;
token.s = strContent[gParseIndex];
gParseIndex++;
return true;
}

cerr << "Parse Error " << endl;
token.kind = BAD_TOKEN;
return false;
}
// 最后结尾 检查是否数字字符串状态需要跳出
if (status == IN_INT_PART_STATUS &&
gParseIndex == strContent.size()) {
numEndIndex = gParseIndex;
token.kind = NUMBER_TOKEN;
token.s = strContent.substr(numBegIndex, numEndIndex - numBegIndex);
return true;
}

token.kind = END_OF_LINE_TOKEN;
return true;
}

#define TEST_MACRO
#ifdef TEST_MACRO

void ParseLinetest(string s) {
Token token;
gParseIndex = 0;
while (1) {
if (GetToken(token, s)) {
if (token.kind == END_OF_LINE_TOKEN) {
break;
}
else {
cout << "kind = " << token.kind
<< ", str = " << token.s << endl;
}
}

}
}

int main()
{
ParseLinetest(testStr);
return 0;
}