daw_json_parse_policy_hash_comments.h

Go to the documentation of this file.

// Copyright (c) Darrell Wright
//
// Distributed under the Boost Software License, Version 1.0. (See accompanying
// file LICENSE or copy at http://www.boost.org/LICENSE_1_0.txt)
//
// Official repository: https://github.com/beached/daw_json_link
//
 
#pragma once
 
#include "daw/json/impl/version.h"
 
#include "daw/json/impl/daw_json_assert.h"
#include "daw/json/impl/daw_json_parse_common.h"
#include "daw/json/impl/daw_json_parse_policy_policy_details.h"
#include "daw/json/impl/daw_not_const_ex_functions.h"
 
#include <daw/daw_attributes.h>
#include <daw/daw_constant.h>
#include <daw/daw_likely.h>
#include <daw/daw_not_null.h>
#include <daw/daw_traits.h>
 
#include <cstddef>
#include <cstdint>
#include <type_traits>
 
namespace daw::json {
  inline namespace DAW_JSON_VER {
    class HashCommentSkippingPolicy {
      template<typename ParseState>
      DAW_ATTRIB_FLATINLINE static constexpr void
      skip_comments_unchecked( ParseState &parse_state ) {
        while( parse_state.front( ) == '#' ) {
          parse_state.remove_prefix( );
          parse_state.template move_to_next_of_unchecked<'\n'>( );
          parse_state.remove_prefix( );
        }
      }
 
      template<typename ParseState>
      DAW_ATTRIB_FLATINLINE static constexpr void
      skip_comments_checked( ParseState &parse_state ) {
        while( parse_state.has_more( ) and parse_state.front( ) == '#' ) {
          parse_state.remove_prefix( );
          parse_state.template move_to_next_of_checked<'\n'>( );
          // The last line may not end with a newline
          if( ( ParseState::is_zero_terminated_string or
                parse_state.has_more( ) ) and
              parse_state.front( ) == '\n' ) {
            parse_state.remove_prefix( );
          }
        }
      }
 
      template<typename ParseState>
      DAW_ATTRIB_FLATINLINE static constexpr void
      skip_comments( ParseState &parse_state ) {
        if constexpr( ParseState::is_unchecked_input ) {
          skip_comments_unchecked( parse_state );
        } else {
          skip_comments_checked( parse_state );
        }
      }
 
    public:
      template<typename ParseState>
      DAW_ATTRIB_FLATINLINE static constexpr void
      trim_left_checked( ParseState &parse_state ) {
        skip_comments_checked( parse_state );
        while( parse_state.has_more( ) and parse_state.is_space_unchecked( ) ) {
          parse_state.remove_prefix( );
          skip_comments_checked( parse_state );
        }
      }
 
      template<typename ParseState>
      DAW_ATTRIB_FLATINLINE static constexpr void
      trim_left_unchecked( ParseState &parse_state ) {
        skip_comments_unchecked( parse_state );
        while( parse_state.is_space_unchecked( ) ) {
          parse_state.remove_prefix( );
          // Comments can follow whitespace
          skip_comments_unchecked( parse_state );
        }
      }
 
      template<typename ParseState>
      DAW_ATTRIB_FLATINLINE static constexpr void
      move_next_member_unchecked( ParseState &parse_state ) {
        parse_state.move_next_member_or_end_unchecked( );
      }
 
      template<char... keys, typename ParseState>
      DAW_ATTRIB_FLATINLINE static constexpr void
      move_to_next_of( ParseState &parse_state ) {
        skip_comments( parse_state );
        while( true ) {
          // Check before reading front( ), skip_comments can reach the end
          daw_json_assert_weak( parse_state.has_more( ),
                                ErrorReason::UnexpectedEndOfData,
                                parse_state );
          if( parse_policy_details::in<keys...>( parse_state.front( ) ) ) {
            break;
          }
          parse_state.remove_prefix( );
          skip_comments( parse_state );
        }
      }
 
      DAW_ATTRIB_FLATINLINE static constexpr bool is_literal_end( char c ) {
        return c == '\0' or c == ',' or c == ']' or c == '}' or c == '#';
      }
 
      template<json_details::SkipBracketedType BracketedType,
               typename ParseState>
      DAW_ATTRIB_FLATINLINE static constexpr ParseState
      skip_bracketed_item_checked( ParseState &parse_state ) {
        DAW_CPP23_STATIC_LOCAL constexpr char PrimLeft =
          BracketedType == json_details::SkipBracketedType::Class ? '{' : '[';
        using PrimRight = daw::constant<PrimLeft == '{' ? '}' : ']'>;
        using SecLeft = daw::constant<PrimLeft == '{' ? '[' : '{'>;
        using SecRight = daw::constant<SecLeft::value == '{' ? '}' : ']'>;
 
        // Not checking for Left as it is required to be skipped already
        auto result = parse_state;
        std::size_t cnt = 0;
        std::uint32_t prime_bracket_count = 1;
        std::uint32_t second_bracket_count = 0;
        auto ptr_first = parse_state.first;
        auto const ptr_last = parse_state.last;
        // A null range is a logic error, not bad input
        daw_json_dbg_ensure( ptr_first != nullptr and ptr_last != nullptr,
                             ErrorReason::InvalidNull );
        if( DAW_UNLIKELY( ptr_first >= ptr_last ) ) {
          return result;
        }
        if( *ptr_first == PrimLeft ) {
          ++ptr_first;
        }
        while( DAW_LIKELY( ptr_first < ptr_last ) ) {
          // TODO: use if/else if or put switch into IILE
          switch( *ptr_first ) {
          case '\\':
            ++ptr_first;
            break;
          case '"':
            ++ptr_first;
            // ptr_first is past an opening quote, so it is not null
            ptr_first += json_details::mem_skip_until_end_of_string<
                           ParseState::is_unchecked_input,
                           typename ParseState::exec_tag_t>(
                           daw::not_null<char const *>( daw::never_null, ptr_first ),
                           daw::not_null<char const *>( daw::never_null, ptr_last ) )
                           .get( ) -
                         ptr_first;
            daw_json_ensure( ptr_first < ptr_last,
                             ErrorReason::UnexpectedEndOfData,
                             parse_state );
            break;
          case ',':
            if( prime_bracket_count == 1 and second_bracket_count == 0 ) {
              ++cnt;
            }
            break;
          case PrimLeft:
            ++prime_bracket_count;
            break;
          case PrimRight::value:
            --prime_bracket_count;
            if( prime_bracket_count == 0 ) {
              daw_json_ensure( second_bracket_count == 0,
                               ErrorReason::InvalidBracketing,
                               parse_state );
              ++ptr_first;
              // We include the close primary bracket in the range so that
              // subsequent parsers have a terminator inside their range
              result.last = ptr_first;
              result.counter = cnt;
              parse_state.first = ptr_first;
              return result;
            }
            break;
          case SecLeft::value:
            ++second_bracket_count;
            break;
          case SecRight::value:
            --second_bracket_count;
            break;
          case '#':
            ++ptr_first;
            while( ptr_first < ptr_last and *ptr_first != '\n' ) {
              ++ptr_first;
            }
            if( ptr_first < ptr_last ) {
              continue;
            }
            break;
          }
          ++ptr_first;
        }
        daw_json_ensure( ( prime_bracket_count == 0 ) &
                           ( second_bracket_count == 0 ),
                         ErrorReason::InvalidBracketing,
                         parse_state );
        // We include the close primary bracket in the range so that subsequent
        // parsers have a terminator inside their range
        result.last = ptr_first;
        result.counter = cnt;
        parse_state.first = ptr_first;
        return result;
      }
 
      template<json_details::SkipBracketedType BracketedType,
               typename ParseState>
      DAW_ATTRIB_FLATINLINE static constexpr ParseState
      skip_bracketed_item_unchecked( ParseState &parse_state ) {
        DAW_CPP23_STATIC_LOCAL constexpr char PrimLeft =
          BracketedType == json_details::SkipBracketedType::Class ? '{' : '[';
        // Not checking for Left as it is required to be skipped already
        using PrimRight = daw::constant<PrimLeft == '{' ? '}' : ']'>;
        using SecLeft = daw::constant<PrimLeft == '{' ? '[' : '{'>;
        using SecRight = daw::constant<SecLeft::value == '{' ? '}' : ']'>;
 
        auto result = parse_state;
        std::size_t cnt = 0;
        std::uint32_t prime_bracket_count = 1;
        std::uint32_t second_bracket_count = 0;
        auto ptr_first = parse_state.first;
        auto const ptr_last = parse_state.last;
        // A null range is a logic error, not bad input
        daw_json_dbg_ensure( ptr_first != nullptr and ptr_last != nullptr,
                             ErrorReason::InvalidNull );
        if( *ptr_first == PrimLeft ) {
          ++ptr_first;
        }
        while( true ) {
          switch( *ptr_first ) {
          case '\\':
            ++ptr_first;
            break;
          case '"':
            ++ptr_first;
 
            // ptr_first is past an opening quote, so it is not null
            ptr_first += json_details::mem_skip_until_end_of_string<
                           ParseState::is_unchecked_input,
                           typename ParseState::exec_tag_t>(
                           daw::not_null<char const *>( daw::never_null, ptr_first ),
                           daw::not_null<char const *>( daw::never_null, ptr_last ) )
                           .get( ) -
                         ptr_first;
            break;
          case ',':
            if( prime_bracket_count == 1 and second_bracket_count == 0 ) {
              ++cnt;
            }
            break;
          case PrimLeft:
            ++prime_bracket_count;
            break;
          case PrimRight::value:
            --prime_bracket_count;
            if( prime_bracket_count == 0 ) {
              ++ptr_first;
              // We include the close primary bracket in the range so that
              // subsequent parsers have a terminator inside their range
              result.last = ptr_first;
              result.counter = cnt;
              parse_state.first = ptr_first;
              return result;
            }
            break;
          case SecLeft::value:
            ++second_bracket_count;
            break;
          case SecRight::value:
            --second_bracket_count;
            break;
          case '#':
            ++ptr_first;
            while( *ptr_first != '\n' ) {
              ++ptr_first;
            }
            break;
          }
          ++ptr_first;
        }
        DAW_UNREACHABLE( );
      }
    };
  } // namespace DAW_JSON_VER
} // namespace daw::json