在第7,8章我们分别学习了Pascal控制语句的解析,解释执行,还涉及到了一点点优化,虽然好像花的时间不多,但是用Antlr之后就会发现还是很啰嗦。Antlr对于编译器作者来说,或许不是性能最好的,但是对于语言前期的论证,运行时原型的快速构建都有非常重要意义,尤其语法较多(也就是EBNF Rule条数很多)。更进一步说,即使有很多优化,Antlr的源代码和代码生成(Codegen)这一块完全可以定制。我们可以在语言,运行时稳定之后来个手写版:)。好了,废话不多说了,先看控制语句的语法。
==>> 本章中文版源代码下载:svn co http://wci.googlecode.com/svn/branches/ch8_antlr/ 源代码使用了UTF-8编码,下载到本地请修改!
控制语句的语法
这几个控制语句,FOR最复杂,CASE其次,下面的语法描述严格遵循第7章的分析树显示。
1: statement:
2: compound_statement | assignment_statement | control_statement;
3: control_statement:
4: while_statement | repeat_statement | for_statement | if_statement | case_statement;
5: while_statement:
>^(LOOP ^(NOT $t1) $s1);
7: repeat_statement:
>^(LOOP $s1 $t1);
9: for_statement
10: @init{
11: PascalNode id_node = null;
12: PascalNode id_cp1=null;
13: PascalNode id_cp2=null;
14: PascalNode id_cp3=null;
15: PascalNode op_node=null;
16: boolean increment = true;
18: ((PascalAntlrToken)one_constant.getToken()).setValue(new Integer(1));
19: }:
20: (FOR a1=assignment_statement {
21: id_node =(PascalNode) ((PascalNode)a1.getTree()).getChild(0);
22: id_cp1 = (PascalNode)id_node.dupNode();
23: id_cp2= (PascalNode)id_node.dupNode();
24: id_cp3= (PascalNode)id_node.dupNode();
25: }
>^(LT {id_cp1} $e11))
>^(COMPOUND $a1 ^(LOOP ^(TEST $q) $s2 ^(ASSIGN {id_cp2} ^({op_node} {id_cp3} {one_constant}))));
29: if_statement:
> ^(IF $t1 $s1 $s2?);
31: case_statement:
>^(SELECT $e1 $bhs);
33: case_branches:
> case_branch+;
35: case_branch:
>^(SELECT_BRANCH $c1 $s1);
37: constant_list:
> ^(SELECT_CONSTANTS constant+);
39: constant:
40: NUMBER | NUMBER_REAL | STRING;