[antlr-interest] Tree walking

[Jeff Dahl]

Jeff Dahl wrote:

> Why does SAMPLE.TREE.G not walk the entire tree?  Instead of
>
>    ROOT
>    |-- A
>    |   `-- B
>    |       |-- C
>    |       |   |-- attribute1
>    |       |   |   `-- value 1
>    |       |   |-- attribute2
>    |       |   |   `-- value 2
>    |       |   `-- attribute3
>    |       |       `-- value 3
>    |       `-- D
>    |           `-- attribute4
>    |               `-- value 4
>    |-- E
>    |   `-- attribute5
>    |       `-- value 5
>    `-- F
>        `-- attribute6
>            `-- value 6
>
> SAMPLE.TREE.G outputs
>
>    ROOT
>    |-- A
>    |   `-- B
>    |       |-- C
>    |       `-- D
>    |-- E
>    `-- F

I figured out while SAMPLE.TREE.G was not walking the entire tree-- I 
had SAMPLE.TREE.G defined in a different file and did not use the 
exportVocab/importVocab options.  As such, a new ...TokenTypes.java file 
was created that did not correspond to the TokenTypes.java SAMPLE.G 
created.  For example, the TokenTypes created by SAMPLE.G were:

    int EOF = 1;
    int NULL_TREE_LOOKAHEAD = 3;
    int ROOT = 4;
    int O_SECTION = 5;
    int C_SECTION = 6;
    int KEYWORD = 7;
    int VALUE = 8;
    int SECTION = 9;
    int CHARS = 10;
    int WS = 11;
    int NEWLINE = 12;

while the TokenTypes created by SAMPLE.TREE.G were:

    int EOF = 1;
    int NULL_TREE_LOOKAHEAD = 3;
    int ROOT = 4;
    int O_SECTION = 5;
    int KEYWORD = 6;
    int VALUE = 7;

Clearly, when the tree walker was looking for a KEYWORD(6) and the token 
came up as a KEYWORD(7) the tree parser would fail.

The solution is to use the exportVocab/importVocab 
<http://www.antlr.org/doc/options.html#importVocab> options.