Atitit 发帖机系列（8）词法分析器v5 版本新特性说明）

v5 增加对sql单引号的内部支持。可以作为string

结构调整，使用递归法重构循环发。。放弃循环发。

V4 java dsl词法分析使用循环

V3 sql的词法分析

atitit..sql update语法的词法分析,与语法ast构建 -

V1版 anno 解析器基于fsm设计模式

Atitit.antlr实现词法分析 - attilax

/AtiPlatf_ee/src/com/attilax/fsm/JavaTokenScannerV2.java

package com.attilax.fsm;

import java.util.Collection;

import java.util.List;

import com.attilax.io.filex;

import com.attilax.io.pathx;

import com.attilax.json.AtiJson;

import com.google.common.collect.Lists;

public class JavaTokenScannerV2 {

public static void main(String[] args) {

String f = pathx.prjPath_webrootMode() + "/tokenT/a.txt";

String s = "meth(\\\"select from table where a='bb' \",'str2',\'s3\') ";

s = "meth(\"select from tab where a='abc'\",'str2','s3')";

// s = filex.read(f);

List<Token> li = new JavaTokenScannerV2().getTokens(s);

System.out.println(AtiJson.toJson(li));

}

List<Token> tokens = Lists.newArrayList();

String curTokenTxt = "";

String splitors = "(),";

String curStat = "ini";

String curDbquoStat = "ini";

private String code;

public char[] code_char_arr;

public char cur_char;

int gColumn = -1;

@SuppressWarnings("unchecked")

public List<Token> getTokens(String codeStr) {

List<Token> li = Lists.newArrayList();

code_char_arr = codeStr.toCharArray();

while (true) {

Object tk;

try {

tk = nextTokens();

} catch (TokenEndEx e) {

break;

}

if (tk instanceof Token)

li.add((Token) tk);

else if (tk instanceof List)

li.addAll((Collection<? extends Token>) tk);

else

throw new RuntimeException("token type err,curchar:" + this.cur_char + ",colidx:" + this.gColumn);

}

return li;

}

public Object nextTokens() throws TokenEndEx {

// code_char_arr = code.toCharArray();

gColumn++;

if (gColumn > code_char_arr.length - 1)

throw new TokenEndEx(new String(code_char_arr));

cur_char = code_char_arr[gColumn];

// cur_char=cur_char;

if (this.curTokenTxt.equals("1598"))

System.out.println("dbg");

if (this.gColumn == 30)

System.out.println("dbg");

// get next char,,then change stat

// jude cur char and cur stat...then if or not chage stat

switch (cur_char) {

case '(':

return BrkStartEvt();

// break;

case ')':

return brkEndEvt();

case '\'':

return sQuoEvt();

case '\"':

return dbQuoEvt();

case ':':

return colonEvt();

case ',':

return commaEvt();

default:

return normalCharEvt();

// break;

}

private Object normalCharEvt() throws TokenEndEx {

// normal char

curTokenTxt = curTokenTxt + String.valueOf(cur_char);

// gColumn += 1;

return nextTokens();

}

private Object commaEvt() throws TokenEndEx {

List<Token> tokens = Lists.newArrayList();

if (cur_char == ',' && !this.curStat.equals("squoStart") && !this.curStat.equals("dbquoStart")) {

if (this.curTokenTxt.trim().length() > 0) {

String curTokenTye = "con";

if (this.curTokenTxt.startsWith(":"))

curTokenTye = "op";

else if (this.curStat.equals("colon"))

curTokenTye = "op";

Token tk4 = new Token(this.curTokenTxt).setType(curTokenTye);

tokens.add(tk4);

return tokens;

}

Token tk3 = new Token(String.valueOf(cur_char)).setType("spltr");

tokens.add(tk3);

this.curTokenTxt = "";

this.curStat = "commaStat";

return tk3;

}

return normalCharEvt();

}

private Object colonEvt() {

List<Token> tokens = Lists.newArrayList();

if (cur_char == ':' && !this.curStat.equals("strStart")) {

if (this.curTokenTxt.trim().length() > 0) {

String curTokenTye = "con";

Token tk4 = new Token(this.curTokenTxt).setType(curTokenTye);

tokens.add(tk4);

}

Token tk3 = new Token(String.valueOf(cur_char)).setType("op");

tokens.add(tk3);

this.curTokenTxt = "";

this.curStat = "colon";

return tokens;

}

return tokens;

}

private Object dbQuoEvt() throws TokenEndEx {

// ---------------str type

// first dbquo

if ((cur_char == '\"') && (this.curDbquoStat != "dbquoStart")) //

{

this.curStat = "dbquoStart";

this.curDbquoStat = "dbquoStart";

this.curTokenTxt = "";

return nextTokens();

}

// for close dbquo

if ((cur_char == '\"') && this.curDbquoStat.equals("dbquoStart")) {

Token tk3 = new Token(this.curTokenTxt).setType("str");

tokens.add(tk3);

this.curTokenTxt = "";

this.curStat = "dbquoEnd";

this.curDbquoStat = "dbquoEnd";

return tk3;

}

return normalCharEvt();

}

private Object sQuoEvt() throws TokenEndEx {

char c = this.cur_char;

// for in dbquo ..single quo ..none process as normal char

// first single quoe

if (c == '\'' && (this.curStat != "dbquoStart") && this.curStat != "squoStart") //

{

this.curStat = "squoStart";

this.curTokenTxt = "";

return nextTokens();

}

// for close single quoue

if ((c == '\'') && this.curStat.equals("squoStart")) {

Token tk3 = new Token(this.curTokenTxt).setType("str");

tokens.add(tk3);

this.curTokenTxt = "";

this.curStat = "squoEnd";

return tk3;

}

return normalCharEvt();

}

private Object brkEndEvt() {

char c = this.cur_char;

if (c == ')' && !this.curStat.equals("strStart")) { // && cur stat

// =brk start

if (this.curTokenTxt.length() > 0) // jeig cant smp... last end brk

// is impt..if smp continue

// ,then cant add end brk to

// token

{

String type = gettype_4curCharIsBrkend(this.curTokenTxt, this.curStat);

Token tk3 = new Token(this.curTokenTxt).setType(type);

tokens.add(tk3);

}

Token tk2 = new Token(")").setType("op");

tokens.add(tk2);

this.curTokenTxt = "";

this.curStat = "brkEnd";

return tk2;

}

return c;

}

作者:: 绰号:老哇的爪子（全名：：Attilax Akbar Al Rapanui 阿提拉克斯阿克巴阿尔拉帕努伊）

汉字名：艾提拉（艾龙）， EMAIL:1466519819@qq.com

转载请注明来源： http://www.cnblogs.com/attilax/

--Atiend