void end_parse()
{
switch (state_)
{
case states::unquoted_string:
before_record();
end_unquoted_string_value();
after_field();
break;
case states::escaped_value:
if (parameters_.quote_escape_char() == parameters_.quote_char())
{
before_record();
end_quoted_string_value();
after_field();
}
break;
}
if (column_index_ > 0)
{
after_record();
}
switch (stack_[top_])
{
case modes::array:
if (!pop(modes::array))
{
err_handler_->error(std::error_code(csv_parser_errc::unexpected_eof, csv_error_category()), *this);
}
break;
case modes::object:
if (!pop(modes::object))
{
err_handler_->error(std::error_code(csv_parser_errc::unexpected_eof, csv_error_category()), *this);
}
break;
case modes::header:
if (!pop(modes::header))
{
err_handler_->error(std::error_code(csv_parser_errc::unexpected_eof, csv_error_category()), *this);
}
break;
}
handler_->end_array(*this);
if (!pop(modes::done))
{
err_handler_->error(std::error_code(csv_parser_errc::unexpected_eof, csv_error_category()), *this);
}
handler_->end_json();
}
void end_parse(std::error_code& ec)
{
switch (state_)
{
case csv_state_type::unquoted_string:
if (parameters_.trim_leading() || parameters_.trim_trailing())
{
trim_string_buffer(parameters_.trim_leading(),parameters_.trim_trailing());
}
if (!parameters_.ignore_empty_lines() || (column_index_ > 0 || value_buffer_.length() > 0))
{
if (column_index_ == 0)
{
before_record();
}
if (stack_[top_] != csv_mode_type::subfields)
{
before_field();
}
end_unquoted_string_value();
after_field();
}
break;
case csv_state_type::escaped_value:
if (parameters_.quote_escape_char() == parameters_.quote_char())
{
if (column_index_ == 0)
{
before_record();
}
if (stack_[top_] != csv_mode_type::subfields)
{
before_field();
}
end_quoted_string_value(ec);
if (ec) return;
after_field();
}
break;
default:
break;
}
if (column_index_ > 0)
{
after_record();
}
switch (stack_[top_])
{
case csv_mode_type::header:
pop_mode(csv_mode_type::header);
break;
case csv_mode_type::data:
pop_mode(csv_mode_type::data);
break;
default:
break;
}
if (parameters_.mapping() == mapping_type::m_columns)
{
basic_json_fragment_filter<CharT> fragment_filter(handler_);
handler_.begin_object(*this);
for (size_t i = 0; i < column_names_.size(); ++i)
{
handler_.name(column_names_[i],*this);
decoders_[i].end_array(*this);
decoders_[i].end_json();
decoders_[i].get_result().dump_fragment(fragment_filter);
}
handler_.end_object(*this);
}
else
{
handler_.end_array(*this);
}
if (!pop_mode(csv_mode_type::initial))
{
err_handler_.fatal_error(csv_parser_errc::unexpected_eof, *this);
ec = csv_parser_errc::unexpected_eof;
return;
}
handler_.end_json();
}
void parse(Char const* p, size_t start, size_t length)
{
index_ = start;
for (; index_ < length && state_ != states::done; ++index_)
{
curr_char_ = p[index_];
all_states:
switch (state_)
{
case states::comment:
if (curr_char_ == '\n')
{
state_ = states::expect_value;
}
else if (prev_char_ == '\r')
{
state_ = states::expect_value;
goto all_states;
}
break;
case states::expect_value:
if (column_ == 1 && curr_char_ == parameters_.comment_starter())
{
state_ = states::comment;
}
else
{
state_ = states::unquoted_string;
goto all_states;
}
break;
case states::between_fields:
if (curr_char_ == '\r' || (prev_char_ != '\r' && curr_char_ == '\n'))
{
after_record();
state_ = states::expect_value;
}
else if (curr_char_ == parameters_.field_delimiter())
{
state_ = states::expect_value;
}
break;
case states::escaped_value:
{
if (curr_char_ == parameters_.quote_char())
{
string_buffer_.push_back(curr_char_);
state_ = states::quoted_string;
}
else if (parameters_.quote_escape_char() == parameters_.quote_char())
{
before_record();
end_quoted_string_value();
after_field();
state_ = states::between_fields;
goto all_states;
}
}
break;
case states::quoted_string:
{
if (curr_char_ == parameters_.quote_escape_char())
{
state_ = states::escaped_value;
}
else if (curr_char_ == parameters_.quote_char())
{
before_record();
end_quoted_string_value();
after_field();
state_ = states::between_fields;
}
else
{
string_buffer_.push_back(curr_char_);
}
}
break;
case states::unquoted_string:
{
if (curr_char_ == '\r' || (prev_char_ != '\r' && curr_char_ == '\n'))
{
before_record();
end_unquoted_string_value();
after_field();
after_record();
state_ = states::expect_value;
}
else if (curr_char_ == '\n')
{
if (prev_char_ != '\r')
{
before_record();
end_unquoted_string_value();
after_field();
after_record();
state_ = states::expect_value;
}
}
else if (curr_char_ == parameters_.field_delimiter())
{
before_record();
end_unquoted_string_value();
after_field();
state_ = states::expect_value;
}
else if (curr_char_ == parameters_.quote_char())
{
string_buffer_.clear();
state_ = states::quoted_string;
}
else
{
string_buffer_.push_back(curr_char_);
}
}
break;
default:
err_handler_->error(std::error_code(csv_parser_errc::invalid_state, csv_error_category()), *this);
break;
}
if (line_ > parameters_.max_lines())
{
state_ = states::done;
}
switch (curr_char_)
{
case '\r':
++line_;
column_ = 1;
break;
case '\n':
if (prev_char_ != '\r')
{
++line_;
}
column_ = 1;
break;
default:
++column_;
break;
}
prev_char_ = curr_char_;
}
}
void parse(const CharT* p, size_t start, size_t length, std::error_code& ec)
{
index_ = start;
for (; index_ < length && state_ != csv_state_type::done; ++index_)
{
curr_char_ = p[index_];
all_csv_states:
switch (state_)
{
case csv_state_type::comment:
if (curr_char_ == '\n')
{
state_ = csv_state_type::expect_value;
}
else if (prev_char_ == '\r')
{
state_ = csv_state_type::expect_value;
goto all_csv_states;
}
break;
case csv_state_type::expect_value:
if (column_ == 1 && curr_char_ == parameters_.comment_starter())
{
state_ = csv_state_type::comment;
}
else
{
state_ = csv_state_type::unquoted_string;
goto all_csv_states;
}
break;
case csv_state_type::escaped_value:
{
if (curr_char_ == parameters_.quote_char())
{
value_buffer_.push_back(static_cast<CharT>(curr_char_));
state_ = csv_state_type::quoted_string;
}
else if (parameters_.quote_escape_char() == parameters_.quote_char())
{
state_ = csv_state_type::between_fields;
goto all_csv_states;
}
}
break;
case csv_state_type::quoted_string:
{
if (curr_char_ == parameters_.quote_escape_char())
{
state_ = csv_state_type::escaped_value;
}
else if (curr_char_ == parameters_.quote_char())
{
state_ = csv_state_type::between_fields;
}
else
{
value_buffer_.push_back(static_cast<CharT>(curr_char_));
}
}
break;
case csv_state_type::between_fields:
if (prev_char_ == '\r' && curr_char_ == '\n')
{
}
else if (curr_char_ == '\r' || curr_char_ == '\n')
{
if (parameters_.trim_leading() || parameters_.trim_trailing())
{
trim_string_buffer(parameters_.trim_leading(),parameters_.trim_trailing());
}
if (!parameters_.ignore_empty_lines() || (column_index_ > 0 || value_buffer_.length() > 0))
{
if (column_index_ == 0)
{
before_record();
}
if (stack_[top_] != csv_mode_type::subfields)
{
before_field();
}
end_quoted_string_value(ec);
if (ec) return;
after_field();
after_record();
}
state_ = csv_state_type::expect_value;
}
else if (curr_char_ == parameters_.field_delimiter() || (parameters_.subfield_delimiter().second && curr_char_ == parameters_.subfield_delimiter().first))
{
if (column_index_ == 0 && stack_[top_] != csv_mode_type::subfields)
{
before_record();
}
if (parameters_.trim_leading() || parameters_.trim_trailing())
{
trim_string_buffer(parameters_.trim_leading(),parameters_.trim_trailing());
}
if (stack_[top_] != csv_mode_type::subfields)
{
before_field();
if (parameters_.subfield_delimiter().second && curr_char_ == parameters_.subfield_delimiter().first)
{
before_multi_valued_field();
}
}
end_quoted_string_value(ec);
if (ec) return;
if (curr_char_ == parameters_.field_delimiter())
{
after_field();
}
state_ = csv_state_type::unquoted_string;
}
break;
case csv_state_type::unquoted_string:
{
if (prev_char_ == '\r' && curr_char_ == '\n')
{
}
else if (curr_char_ == '\r' || curr_char_ == '\n')
{
if (parameters_.trim_leading() || parameters_.trim_trailing())
{
trim_string_buffer(parameters_.trim_leading(),parameters_.trim_trailing());
}
if (!parameters_.ignore_empty_lines() || (column_index_ > 0 || value_buffer_.length() > 0))
{
if (column_index_ == 0)
{
before_record();
}
if (stack_[top_] != csv_mode_type::subfields)
{
before_field();
}
end_unquoted_string_value();
after_field();
after_record();
}
state_ = csv_state_type::expect_value;
}
else if (curr_char_ == parameters_.field_delimiter() || (parameters_.subfield_delimiter().second && curr_char_ == parameters_.subfield_delimiter().first))
{
if (column_index_ == 0 && stack_[top_] != csv_mode_type::subfields)
{
before_record();
}
if (parameters_.trim_leading() || parameters_.trim_trailing())
{
trim_string_buffer(parameters_.trim_leading(),parameters_.trim_trailing());
}
if (stack_[top_] != csv_mode_type::subfields)
{
before_field();
if (parameters_.subfield_delimiter().second && curr_char_ == parameters_.subfield_delimiter().first)
{
before_multi_valued_field();
}
}
end_unquoted_string_value();
if (curr_char_ == parameters_.field_delimiter())
{
after_field();
}
state_ = csv_state_type::unquoted_string;
}
else if (curr_char_ == parameters_.quote_char())
{
value_buffer_.clear();
state_ = csv_state_type::quoted_string;
}
else
{
value_buffer_.push_back(static_cast<CharT>(curr_char_));
}
}
break;
default:
err_handler_.fatal_error(csv_parser_errc::invalid_state, *this);
ec = csv_parser_errc::invalid_state;
return;
}
if (line_ > parameters_.max_lines())
{
state_ = csv_state_type::done;
}
switch (curr_char_)
{
case '\r':
++line_;
column_ = 1;
break;
case '\n':
if (prev_char_ != '\r')
{
++line_;
}
column_ = 1;
break;
default:
++column_;
break;
}
prev_char_ = curr_char_;
}
}