runtime/C/doxygen/atsections.dox

/// \page atsections Using Sections Within Grammar Files
///
/// \section intro Introduction
///
/// A C targeted grammar can make use of special annotations within a grammar
/// file, which are prefixed with the <b>\@</b> character. These sections cause the 
/// the placement of their contents within the generated code at defined points
/// such as within the generated C header file.
///
/// The general form of these annotations is:
///
/// \code
/// section
///   : '@' (( 'parser' | 'lexer' ) '::')? SECTIONNAME '{' yourcode '}'
///   ;
/// \endcode
///
/// If the 'parser' or lexer keywords are left out of the specification, then the
/// ANTLR tool assumes a lexer target for a lexer grammar, a parser target for a parser
/// or tree parser grammar, and a parser target for a combined lexer/parser grammar. You
/// are advised as a matter of course to include the parser or lexer target keyword.
///
/// Documentation regarding the \@sections available for a grammar targeted at C now
/// follows.
///
/// \subsection psrinit Sections \@init and \@declarations
///
/// Java targeted grammars allow the special section <code>\@init</code> to be placed after the declaration
/// of a rule (lexer, parser and tree parser rules). This allows you to both declare and initialize
/// variables that are local to the code generated for that rule. You can then reference them within
/// your rule action code.
///
/// With the C target, the generated code is subject to the restrictions of C semantics and this
/// means that you must declare any local variables, then assign to them afterwards. As well as the
/// <code>\@init</code> section, which C programmers should use to initialize their local variables, the C
/// target provides the <code>\@declarations</code> section, which is also a rule based section. This section
/// is where the C programmer should declare the local variables, thus separating their declaration
/// from their initialization. Here is an example:
///
/// \code
/// translation_unit
/// @declarations
/// {
/// 	pANTLR3_BOOLEAN	hasUsing;
/// }
/// @init
/// {
/// 
/// 	// Assume no Using directives
/// 	//
/// 	hasUsing = ANTLR3_FALSE;
/// 
/// }
///     : rulea ruleb ...
///
/// \endcode
///
/// Using the <code>\@declarations</code> and <code>\@init</code> sections guarantees that your generated code will
/// compile correctly on any standard C compiler (assuming, of course, that you type in valid C code.)
///
/// \subsection psrheader \@header section.
///
/// The <code>\@parser::header</code> or <code>\@lexer::header</code> annotations cause the code they encapsulate
/// to be placed at the start of each generated file, regardless of whether it is a .c or .h file. This can
/// be useful for inserting copyright information and so on in all your generated files.
///
/// \bNOTE: Be careful not to confuse this concept with placing code in the generated .h header file. The name choice is
/// unfortunate, but was already used in the Java target to allow the placement of \c imports statements
/// in generated java classes. We have therefore kept the intent of this section the same.
///
/// Here is an example:
////
/// \code
/// @lexer::header
/// {
///   // Copyright (c) Jim Idle 2007 - All your grammar are belong to us.
/// }
///
/// @parser::header
/// {
///   // Copyright (c) Jim Idle 2007 - All your grammar are belong to us.
/// }
/// \endcode
///
///
/// \subsection hdrinclude \@includes section
///
/// The <code>\@parser::includes</code> or <code>\@lexer::includes</code> annotations cause
/// the code they encapsulate to be placed in the generated .h file, \b after the standard
/// includes required by the ANTLR generated code.
///
/// Here you could for instance place a <code>\#include</code>
/// statement to cause your grammar code to include some standard definitions. Because you
/// may use multiple parsers and lexers in your solution, you should probably not place
/// <code>#define</code> statements here, but in the <code>\@postinclude</code> section. Then you
/// may create different <code>\#defines</code> for different recognizers. 
///
/// Here is an example:
////
/// \code
/// @lexer::includes
/// {
///   #include "myprojectcommondefs.h"
/// }
///
/// @parser::includes
/// {
///   #include "myprojectcommondefs.h"
/// }
/// \endcode
///
/// 
/// \subsection hdrpreinclude \@preincludes section
///
/// The <code>\@parser::preincludes</code> or <code>\@lexer::preincludes</code> annotations cause
/// the code they encapsulate to be placed in the generated .h file, \b before the standard
/// includes required by the ANTLR generated code.
///
/// You should use this section when you wish to place #defines and other definitions
/// in the code before the standard ANTLR runtime includes defined them. This allows you
/// to override any predefined symbols and options that the includes otherwise take
/// defaults for. For instance, if you have built a version of the runtime with a 
/// special version of malloc, you can <code>\#define</code> #ANTLR3_MALLOC to match the definition
/// you used for the ANTLR runtime library.
///
/// \subsection hdrpostinclude \@postinclude section
///
/// The <code>\@parser::postinclude</code> or <code>\@lexer::postinclude</code> annotations cause
/// the code they encapsulate to be placed in the generated <b>.C</b> file, after the generated include
/// file (which includes the standard ANTLR3C library includes.
///
/// Code you place here then will be subject to any macros defined by your own includes, by the 
/// generated include and by the standard ANTLR3 includes. This is a good place to <code>\#undef</code> 
/// anything that you don;t like the default values of, but cannot override before the includes
/// define them.
///
/// This is also a good place to <code>#define</code> any macros you may wish to use in the generated
/// .c file. As you can include multiple parsers in your projects, you will need to include the
/// generated .h file of each of them, possibly globally, but almost certainly in a context where you
/// are including more than one .h file simultaneously. Hence if you commonly use the same macro
/// names for accessing structures and so on, and they change from grammar to grammar, you should 
/// define them here to avoid creating conflicting definitions in the header files.
///