cpp_iterator.hpp

C++的一个好库。。。现在很流行

/*=============================================================================
    Boost.Wave: A Standard compliant C++ preprocessor library

    Definition of the preprocessor iterator
    
    http://www.boost.org/

    Copyright (c) 2001-2005 Hartmut Kaiser. Distributed under the Boost
    Software License, Version 1.0. (See accompanying file
    LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
=============================================================================*/

#if !defined(CPP_ITERATOR_HPP_175CA88F_7273_43FA_9039_BCF7459E1F29_INCLUDED)
#define CPP_ITERATOR_HPP_175CA88F_7273_43FA_9039_BCF7459E1F29_INCLUDED

#include <string>
#include <vector>
#include <list>
#include <cstdlib>
#include <cctype>

#include <boost/assert.hpp>
#include <boost/shared_ptr.hpp>
#include <boost/filesystem/path.hpp>
#include <boost/filesystem/operations.hpp>
#include <boost/spirit/iterator/multi_pass.hpp>
#include <boost/spirit/tree/parse_tree_utils.hpp>
#include <boost/pool/pool_alloc.hpp>

#include <boost/wave/wave_config.hpp>

#include <boost/wave/util/insert_whitespace_detection.hpp>
#include <boost/wave/util/eat_whitespace.hpp>
#include <boost/wave/util/macro_helpers.hpp>
#include <boost/wave/util/cpp_macromap_utils.hpp>
#include <boost/wave/util/interpret_pragma.hpp>
#include <boost/wave/util/transform_iterator.hpp>
#include <boost/wave/util/functor_input.hpp>

#include <boost/wave/grammars/cpp_grammar_gen.hpp>
#include <boost/wave/grammars/cpp_expression_grammar_gen.hpp>
#if BOOST_WAVE_ENABLE_COMMANDLINE_MACROS != 0
#include <boost/wave/grammars/cpp_predef_macros_gen.hpp>
#endif

#include <boost/wave/cpp_iteration_context.hpp>
#include <boost/wave/cpp_exceptions.hpp>
#include <boost/wave/language_support.hpp>

///////////////////////////////////////////////////////////////////////////////
namespace boost {
namespace wave {
namespace util {

///////////////////////////////////////////////////////////////////////////////
// retrieve the macro name from the parse tree
template <typename ParseNodeT, typename TokenT, typename PositionT>
inline void  
retrieve_macroname(ParseNodeT const &node, boost::spirit::parser_id id, 
    TokenT &macroname, PositionT const &act_pos)
{
ParseNodeT const *name_node = 0;

    using boost::spirit::find_node;
    if (!find_node(node, id, &name_node)) 
    {
        // ill formed define statement (unexpected, should not happen)
        BOOST_WAVE_THROW(preprocess_exception, bad_define_statement, 
            "bad parse tree (unexpected)", act_pos);
    }
    
typename ParseNodeT::children_t const &children = name_node->children;

    if (0 == children.size() || 
        children[0].value.begin() == children[0].value.end()) 
    {
        // ill formed define statement (unexpected, should not happen)
        BOOST_WAVE_THROW(preprocess_exception, bad_define_statement, 
            "bad parse tree (unexpected)", act_pos);
    }

// retrieve the macro name
    macroname = *children[0].value.begin();
}

///////////////////////////////////////////////////////////////////////////////
// retrieve the macro parameters or the macro definition from the parse tree
template <typename ParseNodeT, typename TokenT, typename ContainerT>
inline bool  
retrieve_macrodefinition(
    ParseNodeT const &node, boost::spirit::parser_id id, 
    ContainerT &macrodefinition, TokenT const &/*t*/)
{
    using namespace boost::wave;
    typedef typename ParseNodeT::const_tree_iterator const_tree_iterator;

// find macro parameters/macro definition inside the parse tree
std::pair<const_tree_iterator, const_tree_iterator> nodes;

    using boost::spirit::get_node_range;
    if (get_node_range(node, id, nodes)) {
    // copy all parameters to the supplied container
        typename ContainerT::iterator last_nonwhite = macrodefinition.end();
        const_tree_iterator end = nodes.second;
        
        for (const_tree_iterator cit = nodes.first; cit != end; ++cit) {
            if ((*cit).value.begin() != (*cit).value.end()) {
            typename ContainerT::iterator inserted = macrodefinition.insert(
                macrodefinition.end(), *(*cit).value.begin());
                
                if (!IS_CATEGORY(macrodefinition.back(), WhiteSpaceTokenType) &&
                    T_NEWLINE != token_id(macrodefinition.back()) &&
                    T_EOF != token_id(macrodefinition.back()))
                {
                    last_nonwhite = inserted;
                }
            }
        }
        
    // trim trailing whitespace (leading whitespace is trimmed by the grammar)
        if (last_nonwhite != macrodefinition.end()) {
            macrodefinition.erase(++last_nonwhite, macrodefinition.end());
        }
        return true;
    }
    return false;
}

#if BOOST_WAVE_ENABLE_COMMANDLINE_MACROS != 0
///////////////////////////////////////////////////////////////////////////////
//  add an additional predefined macro given by a string (MACRO(x)=definition)
template <typename ContextT>
bool add_macro_definition(ContextT &ctx, std::string macrostring,
    bool is_predefined, boost::wave::language_support language)
{
    typedef typename ContextT::token_type token_type;
    typedef typename ContextT::lexer_type lexer_type;
    typedef typename token_type::position_type position_type;
    typedef boost::wave::grammars::predefined_macros_grammar_gen<lexer_type> 
        predef_macros_type;

    using namespace boost::wave;
    using namespace std;    // isspace is in std namespace for some systems
    
// skip leading whitespace
std::string::iterator begin = macrostring.begin();
std::string::iterator end = macrostring.end();

    while(begin != end && isspace(*begin))
        ++begin;
        
// parse the macro definition
position_type act_pos("command line", 0);
boost::spirit::tree_parse_info<lexer_type> hit = 
    predef_macros_type::parse_predefined_macro(
        lexer_type(begin, end, position_type(), language), lexer_type());

    if (!hit.match || (!hit.full && T_EOF != token_id(*hit.stop))) {
        BOOST_WAVE_THROW(preprocess_exception, bad_macro_definition, 
            macrostring.c_str(), act_pos);
    }
    
// retrieve the macro definition from the parse tree
token_type macroname;
std::vector<token_type> macroparameters;
typename ContextT::token_sequence_type macrodefinition;
bool has_parameters = false;

    boost::wave::util::retrieve_macroname(*hit.trees.begin(), 
        predef_macros_type::rule_ids.plain_define_id, macroname, act_pos);
    has_parameters = boost::wave::util::retrieve_macrodefinition(*hit.trees.begin(), 
        predef_macros_type::rule_ids.macro_parameters_id, macroparameters, 
        token_type());
    boost::wave::util::retrieve_macrodefinition(*hit.trees.begin(), 
        predef_macros_type::rule_ids.macro_definition_id, macrodefinition,
        token_type());

//  If no macrodefinition is given, and the macro string does not end with a 
//  '=', then the macro should be defined with the value '1'
    if (0 == macrodefinition.size() && 
        '=' != macrostring[macrostring.size()-1])
    {
        macrodefinition.push_back(token_type(T_INTLIT, "1", act_pos));
    }
    
// add the new macro to the macromap
    return ctx.add_macro_definition(macroname, has_parameters, macroparameters, 
        macrodefinition, is_predefined);
}
#endif // BOOST_WAVE_ENABLE_COMMANDLINE_MACROS != 0

///////////////////////////////////////////////////////////////////////////////
}   // namespace util

///////////////////////////////////////////////////////////////////////////////
//  forward declaration
template <typename ContextT> class pp_iterator;

namespace impl {

///////////////////////////////////////////////////////////////////////////////
//  
//  pp_iterator_functor
//
///////////////////////////////////////////////////////////////////////////////
template <typename ContextT> 
class pp_iterator_functor {

public:
// interface to the boost::spirit::multi_pass_policies::functor_input policy
    typedef typename ContextT::token_type               result_type;

//  eof token
    static result_type const eof;

private:
    typedef typename ContextT::lexer_type               lexer_type;
    typedef typename result_type::string_type           string_type;
    typedef boost::wave::grammars::cpp_grammar_gen<lexer_type> 
        cpp_grammar_type;

//  iteration context related types (an iteration context represents a current
//  position in an included file)
    typedef base_iteration_context<lexer_type>    base_iteration_context_type;
    typedef 
        iteration_context<lexer_type, typename ContextT::input_policy_type>
        iteration_context_type;

// parse tree related types
    typedef 
        boost::spirit::node_val_data_factory<boost::spirit::nil_t> 
        node_factory_t;
    typedef 
        boost::spirit::tree_match<lexer_type, node_factory_t> 
        parse_tree_match_t;
    typedef typename parse_tree_match_t::node_t         parse_node_type;       // tree_node<node_val_data<> >
    typedef typename parse_tree_match_t::parse_node_t   parse_node_value_type; // node_val_data<>
    typedef typename parse_tree_match_t::container_t    parse_tree_type;       // parse_node_type::children_t

// type of a token sequence
    typedef typename ContextT::token_sequence_type      token_sequence_type;
    
public:
    template <typename IteratorT>
    pp_iterator_functor(ContextT &ctx_, IteratorT const &first_, 
            IteratorT const &last_, typename ContextT::position_type const &pos_,
            boost::wave::language_support language)
    :   ctx(ctx_), 
        iter_ctx(new base_iteration_context_type(
                lexer_type(first_, last_, pos_, language), lexer_type(), 
                pos_.get_file().c_str()
            )), 
        seen_newline(true), must_emit_line_directive(false),
        act_pos(ctx_.get_main_pos()), //last_line(0),
        eater(need_preserve_comments(ctx_.get_language()))
    {
        act_pos.set_file(pos_.get_file());
#if BOOST_WAVE_SUPPORT_PRAGMA_ONCE != 0
        ctx_.set_current_filename(pos_.get_file().c_str());
#endif
    }
    
// get the next preprocessed token
    result_type const &operator()();

// get the last recognized token (for error processing etc.)
    result_type const &current_token() const { return act_token; }

protected:
    friend class pp_iterator<ContextT>;
    void on_include_helper(char const *t, char const *s, bool is_system, 
        bool include_next);
    
protected:
    result_type const &get_next_token();
    result_type const &pp_token(bool consider_emitting_line_directive = false);

    bool pp_directive();
    bool dispatch_directive(boost::spirit::tree_parse_info<lexer_type> const &hit);

    void on_include(string_type const &s, bool is_system, bool include_next);
    void on_include(typename parse_tree_type::const_iterator const &begin,
        typename parse_tree_type::const_iterator const &end, bool include_next);

    void on_define(parse_node_type const &node);
    void on_undefine(result_type const &t);
    
    void on_ifdef(typename parse_tree_type::const_iterator const &begin,
        typename parse_tree_type::const_iterator const &end);
    void on_ifndef(typename parse_tree_type::const_iterator const &begin,
        typename parse_tree_type::const_iterator const &end);
    void on_else();
    void on_endif();
    void on_illformed(typename result_type::string_type const &s);
        
    void on_line(typename parse_tree_type::const_iterator const &begin,
        typename parse_tree_type::const_iterator const &end);
    void on_if(typename parse_tree_type::const_iterator const &begin,
        typename parse_tree_type::const_iterator const &end);
    void on_elif(typename parse_tree_type::const_iterator const &begin,
        typename parse_tree_type::const_iterator const &end);
    void on_error(typename parse_tree_type::const_iterator const &begin,
        typename parse_tree_type::const_iterator const &end);
#if BOOST_WAVE_SUPPORT_WARNING_DIRECTIVE != 0
    void on_warning(typename parse_tree_type::const_iterator const &begin,
        typename parse_tree_type::const_iterator const &end);
#endif
    bool on_pragma(typename parse_tree_type::const_iterator const &begin,
        typename parse_tree_type::const_iterator const &end);

    result_type const &emit_line_directive();
    bool returned_from_include();

    bool interpret_pragma(token_sequence_type const &pragma_body,
        token_sequence_type &result);

private:
    ContextT &ctx;              // context, this iterator is associated with
    boost::shared_ptr<base_iteration_context_type> iter_ctx;
    
    bool seen_newline;              // needed for recognizing begin of line
    bool must_emit_line_directive;  // must emit a line directive
    result_type act_token;          // current token
    typename result_type::position_type &act_pos;   // current fileposition (references the macromap)
//    unsigned int last_line;         // line number of the previous token
        
    token_sequence_type unput_queue;     // tokens to be preprocessed again
    token_sequence_type pending_queue;   // tokens already preprocessed
    
    // detect whether to insert additional whitespace in between two adjacent 
    // tokens, which otherwise would form a different token type, if 
    // retokenized
    boost::wave::util::insert_whitespace_detection whitespace; 
    
    // remove not needed whitespace from the output stream
    boost::wave::util::eat_whitespace<result_type> eater;
};

///////////////////////////////////////////////////////////////////////////////
//  eof token
template <typename ContextT>
typename pp_iterator_functor<ContextT>::result_type const
    pp_iterator_functor<ContextT>::eof;

///////////////////////////////////////////////////////////////////////////////
//
//  returned_from_include()
//