如何捕获ANTLR语法中的任何内容？

Question

I have a grammar that looks like a markup/markdown language. We use it to produce our textbooks.

It is something like:

[chapter Introduction]

    [section First program]

    Java is pretty cool, **we love it**, let's learn.

    Use the ::javacc:: to compile stuff.

    [title C# is also cool]

    bla bla 

    [code]

    some java code in here

    [/code]

We have this sort of [tag xxx]content[/tag] markup language. I wrote the grammar for that, but it doesn't work for all cases. My main doubt is how to capture the content inside of [code] or even [title] , which can be anything.

To capture [section blabla] , I tried the following:

secao      : '[section ' secao_nome ']';
    secao_nome : (~']'+?);

I tried (~']'+?) to capture everything but the closing tag. That was my main idea: write many regexes like that, one for each tag that I have, and make them ignore the "close tag". For example, I tried to do (~'::'+?) to capture the content of the italic (which ends with :: ).

I also tried to have a generic token for the inside content. However, I need to ignore :: , ** , and all the symbols that actually mean something depending on the context. So, my expression RAW : (~[\\n\\[\\]'**''::''__''%%'' '0-9\\"] | ':')+; doesn't work.

You can see my full grammar here. Sorry that the names are in portuguese:

grammar Tubaina;

    @header {
        package br.com.caelum.tubaina.antlr;
    }

    afc                 : capitulo conteudos+;

    capitulo            : '[chapter ' capitulo_nome ']';
    capitulo_nome       : (~']'+?)*;

    conteudos           : enter* conteudo+ enter*;
    conteudo            : (secao | texto | subsecao | label | box | codigo | lista | imagem | exercicios | index | tabela | quote | todo | note);

    secao               : '[section ' secao_nome ']';
    secao_nome          : (~'['+?);

    quote               : '[quote ' quote_texto '--' quote_autor ']';
    quote_texto         : (~'--'+?);
    quote_autor         : (~']'+?);

    tabela              : '[table "' tabela_nome '"]' tabela_linhas+;
    tabela_nome         : (~'"'+?);
    tabela_linhas       : '[row]' tabela_colunas+ '[/row]';
    tabela_colunas      : '[col]' tabela_conteudo '[/col]';
    tabela_conteudo     : conteudo;

    index               : '[index ' index_nome ']';
    index_nome          : (~']'+?);

    exercicios          : '[exercise]' questoes '[/exercise]';
    questoes            : (enter* questao_def enter*)+;
    questao_def         : '[question]' enter* questao resposta_def? enter* '[/question]';
    questao             : (conteudo | enter)+; 
    resposta_def        : enter* '[answer]' resposta '[/answer]';
    resposta            : (texto | enter)+; 

    imagem                  : '[img ' espaco* imagem_path espaco* imagem_tamanho_def? espaco* (imagem_comentario_def? | ']');
    imagem_path             : (~' '+?);
    imagem_tamanho_def      : 'w=' imagem_tamanho '%';
    imagem_tamanho          : NUMERO;
    imagem_comentario_def   : '"' imagem_comentario '"]';
    imagem_comentario       : (~'"'+?);

    lista               : lista_numerada | lista_nao_numerada;
    lista_numerada      : '[list ' lista_tipo ']' item* '[/list]';
    lista_tipo          : 'number' | 'roman' | 'letter';
    lista_nao_numerada  : '[list]' item* '[/list]';
    item                : enter* '*' texto* enter*;

    todo                : todo_comando todo_comentario ']';
    todo_comando        : '[todo ' | '[TODO ';
    todo_comentario     : (~']'+?);

    note                : '[note]' note_conteudo+ '[/note]';
    note_conteudo       : (enter* texto enter*);

    box                 : '[box ' box_titulo ']' box_conteudo+ '[/box]';
    box_conteudo        : (enter* conteudos+ enter*);
    box_titulo          :  (~']'+?);

    subsecao            : '[title ' subsecao_nome ']';
    subsecao_nome       : (~']'+?);

    label               : '[label ' label_nome ']';
    label_nome          : (~']'+?);

    codigo                  : codigo_com_linguagem | codigo_sem_linguagem | codigo_do_arquivo;
    codigo_do_arquivo       : '[code ' linguagem 'filename=' codigo_arquivo_path '[/code]';
    codigo_arquivo_path     : (~' '+?);
    codigo_raw              : (~'[/code]'+?);
    linguagem               : (~' '+?);
    codigo_sem_linguagem    : '[code]' codigo_raw '[/code]';
    codigo_com_linguagem    : '[code ' linguagem codigo_fechado codigo_raw '[/code]';
    codigo_fechado          : ' #]' | ']';

    texto               : paragrafo | negrito | italico | underline | inline;
    paragrafo           : linha enter?;
    linha               : (~'\n'+?);
    negrito             : '**' linha '**';
    italico             : '::' linha '::';
    underline           : '__' linha '__';
    inline              : '%%' linha '%%';

    enter                       : N | TAB;
    espaco                      : ESPACO;

    N                   : ['\n'];
    TAB                 : '\t';
    ESPACO : ' ';
    NUMERO : [0-9]+;

    WS                  : (' ' | '\t') -> skip;

Also, my attempt with the generic regex is here: https://github.com/mauricioaniche/tubaina-antlr-grammar/blob/f381ad0e3d1bd458922165c7166c7f1c55cea6c2/src/br/com/caelum/tubaina/antlr/Tubaina.g4

My question is: how can I write a grammar to a language like that, in which I have tags and any content inside them? Any ideas?

Thanks in advance!

Answer 1

I'm not sure about antlr, so I'm posting this answer that might help you with the regex idea.

You could use a regex like this:

\[code\]([\s\S]+)\[/code\]|\[title (.+)\]

Working demo

Match information

MATCH 1
2.  [165-180]   `C# is also cool`
MATCH 2
1.  [207-241]   `

    some java code in here

    `

I've put both regex in a compound one using OR to show you the idea. If you are able to use 2 regex then you can use the following:

\[code\]([\s\S]+)\[/code\]   <-- to capture the [code]XX[/code] content
\[title (.+)\]               <-- to capture the [title XX] content