Skip to content

Latest commit

 

History

History
109 lines (90 loc) · 4.57 KB

Schema.md

File metadata and controls

109 lines (90 loc) · 4.57 KB
Raw

Syntax Tree Structure

How It Works

For example, CSharpProcessorUsingAntlr3 class generates the two following xml files corresponding to class K {} and class K { void m() {} }.

<compilation_unit id="0">
  .. snip ..
          <class_declaration id="50">
  .. snip ..
            <class_body id="621">
              <TOKENS id="char_literal667">
                <WS hidden=" " startline="1" startpos="7" endline="1" endpos="8" />
                <TOKEN startline="1" startpos="8" endline="1" endpos="9">{</TOKEN>
              </TOKENS>
              <TOKENS id="char_literal669">
                <WS hidden=" " startline="1" startpos="9" endline="1" endpos="10" />
                <TOKEN startline="1" startpos="10" endline="1" endpos="11">}</TOKEN>
              </TOKENS>
            </class_body>
          </class_declaration>
        </type_declaration>
  .. snip ..
</compilation_unit>

and

<compilation_unit id="0">
  .. snip ..
          <class_declaration id="50">
  .. snip ..
            <class_body id="621">
              <TOKENS id="char_literal667">
                <WS hidden=" " startline="1" startpos="7" endline="1" endpos="8" />
                <TOKEN startline="1" startpos="8" endline="1" endpos="9">{</TOKEN>
              </TOKENS>
              <class_member_declarations id="668">
                <class_member_declaration id="670">
                  <method_declaration id="673">
                    <method_header id="707">
  .. snip ..
                    </method_header>
                    <method_body id="708">
  .. snip ..
                    </method_body>
                  </method_declaration>
                </class_member_declaration>
              </class_member_declarations>
              <TOKENS id="char_literal669">
                <WS hidden=" " startline="1" startpos="21" endline="1" endpos="22" />
                <TOKEN startline="1" startpos="22" endline="1" endpos="23">}</TOKEN>
              </TOKENS>
            </class_body>
          </class_declaration>
        </type_declaration>
  .. snip ..
</compilation_unit>

The class_body element in the first xml contains only two tokens of { and }, in contrast, one in the second xml contains also the class_member_declarations element corresponding to void m() {}. This difference is caused by the ANTLR EBNF grammar (cs.g) that has the following parsing rule.

class_body: '{' class_member_declarations? '}' ;

This rule indicates that the class_body elements can have one or no class_member_declarations element.

What Is ID Attribute

As you can see, each non-terminal node, whose name starts a lower letter, has an id attribute. The id value identifies the location where the corresponding node appears in reft sides of the EBNF rules. For example, the id value indicates that a multiplicative_expression element corresponds to the first or second multiplicative_expression rules in the additive_expression rule.

additive_expression: multiplicative_expression ( ('+'|'-') multiplicative_expression )* ;

Assume that unique integers as id values are annotated in the rule as follows.

additive_expression:
  multiplicative_expression<id: 1> ( ('+'<id: 2> | '-'<id: 3>) multiplicative_expression<id: 4> )* ;

Then, one possible xml may be as follows.

<additive_expression>
  <multiplicative_expression id="1">
    .. snip ..
  </multiplicative_expression>
  <TOKENS>
    <TOKEN id="2">+</TOKEN>
  </TOKENS>
  <multiplicative_expression id="4">
    .. snip ..
  </multiplicative_expression>
  <TOKENS>
    <TOKEN id="3">-</TOKEN>
  </TOKENS>
  <multiplicative_expression id="4">
    .. snip ..
  </multiplicative_expression>
</additive_expression>

As another example, the id values can also identify the locatoin of the corresponding non-terminal node.

bracket_expression: '[' expression_list? ']' ;
element_initializer: non_assignment_expression  | '{' expression_list '}' ;

Assume that unique integers as id values are annotated in the rule as follows.

bracket_expression: '['<id: 5> expression_list<id: 6>? ']'<id: 7> ;
element_initializer:
  non_assignment_expression<id: 8> | '{'<id: 9> expression_list<id: 10> '}'<id: 11> ;