Antlr语法Java代码

Question

I have defined a grammar language using Antlr3. ANTLRWorks generates the lexer and parser code.

Below is a sample grammar:

grammar i;

@header {
package com.data;
}

elements    : (value (',' value)*)?;
insert      : 'INSERT INTO table' 'VALUES' '('elements')'';';
WS      : (' '|'\t'|'\f'|'\n'|'\r')+ {skip();}; // handle white space between keywords

The above grammar is only a example. I am actually developing a big grammar for my company.

I have imported the grammar into com.data package of my project and can use the Lexer and Parser generated. What I would like to find out:

• Does the Java code have to be written within the grammar?
• I would prefer to use the Lexer and Parser, and write the Java code within another class instead within the grammar.

According to the following post: ANTLR: Is there a simple example? it shows a simple example. However, in my case the grammar is big and would prefer the grammar to be imported and used. I have many rules:

For example, the above grammar generates this code:

public void insert() throws RecognitionException;

What is the best way to handle this? I have lots of rules, which are return type of void.

Each rule has a specific meaning in Java terms. I don't think the best way would be to get the syntax and then check call my code. For example:

String input = "INSERT INTO table VALUES (null);";
if(input.equals("INSERT INTO table VALUES (null);") {
     insertData();
}

public void insertData() {
   // insert data into database
}

It is a good idea to check against every syntax and call my code?

Answer 1

The answer to your two questions:

1. No, the Java code does NOT have to be embedded within the grammar. You can generate an AST instead. My grammar is simple enough, so I don't do that.
2. The generated AST can be processed using the Observer/Vistor pattern. Google should turn up something there for you.