ANTLR#1：描述一个简单计算器

ANTLR是什么鬼？引用官网的说明，

What is ANTLR?

ANTLR (ANother Tool for Language Recognition) is a powerful parser generator for reading, processing, executing, or translating structured text or binary files. It’s widely used to build languages, tools, and frameworks. From a grammar, ANTLR generates a parser that can build and walk parse trees.

从编译器的角度来看，ANTLR可以用来帮助我们完成编译器前端所要完成的一些工作，词法分析（Lexical Analysis），语法分析（Syntax Analysis），生成抽象语法树（Abstract Syntax Tree，AST）等。语义分析（Semantic Analysis)，例如类型检查，需要我们自己完成。有一点要啰嗦一下，在ANTLR4中已经不再提供生成AST的功能（在ANTLR3中可以通过Options指定output=AST来生成AST），而是生成了分析树（Parse Tree）。关于AST和PT的优劣，可以看看SO上面的这个回答。

要使用ANTLR生成语言的词法分析器（Lexer）跟语法分析器（Parser），我们需要告诉ANTLR我们的语言的文法（Grammar）。ANTLR采用的是上下文无关文法（Context Free Grammar），使用类似BNF的符号集来描述。使用上下文无关文法的语言比较常用的Parser有两种，LL Parser和LR Parser，而ANTLR帮我们生成的是前者。

alright，上面啰嗦了这么多，接下来我们通过一个简单的栗子来小试牛刀，看看ANTLR是怎么玩的。

一个简单计算器的grammar，

grammar Cal;

options
{
  // antlr will generate java lexer and parser
  language = Java;
}

@lexer::header {
package me.kisimple.just4fun.antlr;
}

@parser::header {
package me.kisimple.just4fun.antlr;
}

 lexer rules:
PLUS : '+'
;
MINUS : '-'
;
MULTIPLY : '*'
;
DIVIDE : '/'
;
LPAREN : '('
;
RPAREN : ')'
;
ZERO : '0'
;
NUMBER : ZERO | [1-9][0-9]*
;
WS : [ \t\r\n]+ -> skip
;

 parser rules:
program : calExpr
;
calExpr : multExpr ((PLUS|MINUS) multExpr)*
; 
multExpr : atom ((MULTIPLY|DIVIDE) atom)*
; 
atom:  LPAREN calExpr RPAREN | NUMBER
;

使用 ANTLR提供的工具 来帮我们生成Lexer和Parser，

> java org.antlr.v4.Tool -visitor Cal.g4

-visitor用来告诉ANTLR帮我们生成PT的Visitor，我们可以通过Visitor来访问PT，另外也可以通过Listener来遍历PT，二者的区别如下，

The biggest difference between the listener and visitor mechanisms is that listener methods are called independently by an ANTLR-provided walker object, whereas visitor methods must walk their children with explicit visit calls. Forgetting to invoke visitor methods on a node’s children, means those subtrees don’t get visited.

ANTLR还帮我们生成了token文件，Cal.tokens，内容如下

PLUS=1
MINUS=2
MULTIPLY=3
DIVIDE=4
LPAREN=5
RPAREN=6
ZERO=7
NUMBER=8
WS=9
'+'=1
'-'=2
'*'=3
'/'=4
'('=5
')'=6
'0'=7

ANTLR给在我们的grammar文件中出现的token赋予了一个整数类型，这些类型在Parser中定义了常量，语法分析语义分析时不可避免的要用到它们。

接下来我们实现一个Visitor来访问PT，我们直接边访问边做计算，访问结束后我们的计算结果也就出来了（所以这里做的是一个解释器：）

package me.kisimple.just4fun.antlr;

import org.antlr.v4.runtime.tree.ParseTree;
import org.antlr.v4.runtime.tree.TerminalNode;

import java.util.List;
import java.util.Stack;

public class CalVisitorImpl extends CalBaseVisitor<Long> {

    @Override public Long visitCalExpr(CalParser.CalExprContext ctx) {
        List<ParseTree> children = ctx.children;
        Stack<Long> atomStack = new Stack<Long>();
        Stack<TerminalNode> opStack = new Stack<TerminalNode>();

        for (ParseTree child : children) {
            if(child instanceof CalParser.MultExprContext) {
                Long current = visitMultExpr((CalParser.MultExprContext) child);
                if(!opStack.isEmpty()) {
                    TerminalNode op = opStack.pop();
                    Long last = atomStack.pop();
                    switch (op.getSymbol().getType()) {
                        case CalParser.PLUS:
                            atomStack.push(last + current);
                            break;
                        case CalParser.MINUS:
                            atomStack.push(last - current);
                            break;
                    }
                } else {
                    atomStack.push(current);
                }

            } else {
                opStack.push((TerminalNode)child);
            }
        }
        return atomStack.pop();
    }

    @Override public Long visitMultExpr(CalParser.MultExprContext ctx) {
        List<ParseTree> children = ctx.children;
        Stack<Long> atomStack = new Stack<Long>();
        Stack<TerminalNode> opStack = new Stack<TerminalNode>();

        for (ParseTree child : children) {
            if(child instanceof CalParser.AtomContext) {
                Long current = visitAtom((CalParser.AtomContext) child);
                if(!opStack.isEmpty()) {
                    TerminalNode op = opStack.pop();
                    Long last = atomStack.pop();
                    switch (op.getSymbol().getType()) {
                        case CalParser.MULTIPLY:
                            atomStack.push(last * current);
                            break;
                        case CalParser.DIVIDE:
                            atomStack.push(last / current);
                            break;
                    }
                } else {
                    atomStack.push(current);
                }

            } else {
                opStack.push((TerminalNode)child);
            }
        }
        return atomStack.pop();
    }

    @Override public Long visitAtom(CalParser.AtomContext ctx) {
        TerminalNode node = ctx.NUMBER();
        if(node != null) {
            String text = ctx.getText();
            return Long.valueOf(text);
        }
        // ignore '(' and ')'
        return visitCalExpr(ctx.calExpr());
    }

}

使用方式如下，

 public static void main(String[] args) throws Throwable {

        System.out.println(((7 + 7) * (7 + 7) - 7) * 7 - 77 / 7);
        String expr = "((7 + 7) * (7 + 7) - 7) * 7 - 77 / 7";

        ANTLRInputStream input = new ANTLRInputStream(expr);
        CalLexer lexer = new CalLexer(input);

        CommonTokenStream tokens = new CommonTokenStream(lexer);
        tokens.fill();

        CalParser parser = new CalParser(tokens);

        ParserRuleContext tree = parser.program();

        CalVisitorImpl visitor = new CalVisitorImpl();
        System.out.println(visitor.visit(tree));

    }

可以通过ANTLR提供的IDEA插件来看看这颗分析树，

Error Handling

还有一点要说明下，默认情况下，ANTLR生成的Parser在做语法分析时，如果遇到错误会直接略过出错的token，例如String expr = "7 77 + 7";分析树如下图，

输出是这样子的，

line 1:2 extraneous input '77' expecting {<EOF>, '+', '-', '*', '/'}
14

77 直接被忽略，执行了 7 + 7 得到上面的结果。但是可以看到，其实Parser是知道有错误了才会报 extraneous input 。因此我们可以参考默认的 ANTLRErrorStrategy ，也就是 DefaultErrorStrategy 来重新实现错误处理的逻辑，

public class IErrorStrategy extends BailErrorStrategy {

    @Override
    public Token recoverInline(Parser recognizer)
            throws RecognitionException {
        // SINGLE TOKEN DELETION
        Token matchedSymbol = singleTokenDeletion(recognizer);
        if (matchedSymbol != null)
            return super.recoverInline(recognizer);

        // SINGLE TOKEN INSERTION
        singleTokenInsertion(recognizer);
        return super.recoverInline(recognizer);
    }

    @Override
    public void sync(Parser recognizer) throws RecognitionException {
        ATNState s = recognizer.getInterpreter().atn.states.get(recognizer.getState());
        if (inErrorRecoveryMode(recognizer))
            return;

        TokenStream tokens = recognizer.getInputStream();
        int la = tokens.LA(1);

        // try cheaper subset first; might get lucky. seems to shave a wee bit off
        if ( recognizer.getATN().nextTokens(s).contains(la) || la==Token.EOF )
            return;

        // Return but don't end recovery. only do that upon valid token match
        if (recognizer.isExpectedToken(la))
            return;

        singleTokenDeletion(recognizer);
        throw new ParseCancellationException(
                new InputMismatchException(recognizer));
    }

}

使用Parser的时候需要设置一下， parser.setErrorHandler(new IErrorStrategy()); 然后就可以看到成功报错并且提示了：）

line 1:2 extraneous input '77' expecting {<EOF>, '+', '-', '*', '/'}
Exception in thread "main" org.antlr.v4.runtime.misc.ParseCancellationException
    at me.kisimple.just4fun.antlr.IErrorStrategy.sync(IErrorStrategy.java:44)
    at me.kisimple.just4fun.antlr.CalParser.multExpr(CalParser.java:251)
    at me.kisimple.just4fun.antlr.CalParser.calExpr(CalParser.java:172)
    at me.kisimple.just4fun.antlr.CalParser.program(CalParser.java:116)
    at me.kisimple.just4fun.Main.main(Main.java:33)
    at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
    at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:57)
    at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
    at java.lang.reflect.Method.invoke(Method.java:606)
    at com.intellij.rt.execution.application.AppMain.main(AppMain.java:134)
Caused by: org.antlr.v4.runtime.InputMismatchException
    ... 10 more

参考资料

• ANTLR 4 Documentation
• What is the difference between ANTLR 3 and 4?
• https://theantlrguy.atlassian.net/wiki/display/ANTLR4/Lexer+Rules
• https://theantlrguy.atlassian.net/wiki/display/ANTLR4/Parser+Rules
• https://theantlrguy.atlassian.net/wiki/display/ANTLR4/Parse+Tree+Listeners
• http://www.ibm.com/developerworks/cn/java/j-lo-antlr/index.html
• http://meri-stuff.blogspot.com/2011/08/antlr-tutorial-hello-word.html