From 1d055261b4144dbf86b2658437015b15d4dd9bff Mon Sep 17 00:00:00 2001 From: Richard Date: Sun, 4 Sep 2022 00:32:56 +0100 Subject: initial --- include/jsoncons_ext/csv/csv_cursor.hpp | 358 ++++++++++++++++++++++++++++++++ 1 file changed, 358 insertions(+) create mode 100644 include/jsoncons_ext/csv/csv_cursor.hpp (limited to 'include/jsoncons_ext/csv/csv_cursor.hpp') diff --git a/include/jsoncons_ext/csv/csv_cursor.hpp b/include/jsoncons_ext/csv/csv_cursor.hpp new file mode 100644 index 0000000..67d55a6 --- /dev/null +++ b/include/jsoncons_ext/csv/csv_cursor.hpp @@ -0,0 +1,358 @@ +// Copyright 2018 Daniel Parker +// Distributed under the Boost license, Version 1.0. +// (See accompanying file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) + +// See https://github.com/danielaparker/jsoncons for latest version + +#ifndef JSONCONS_CSV_CSV_CURSOR_HPP +#define JSONCONS_CSV_CSV_CURSOR_HPP + +#include // std::allocator +#include +#include +#include +#include +#include +#include // std::basic_istream +#include +#include +#include +#include +#include +#include +#include +#include + +namespace jsoncons { namespace csv { + +template,class Allocator=std::allocator> +class basic_csv_cursor : public basic_staj_cursor, private virtual ser_context +{ +public: + using source_type = Source; + using char_type = CharT; + using allocator_type = Allocator; +private: + static constexpr size_t default_max_buffer_size = 16384; + + typedef typename std::allocator_traits:: template rebind_alloc char_allocator_type; + + text_source_adaptor source_; + basic_csv_parser parser_; + basic_staj_visitor cursor_visitor_; + + // Noncopyable and nonmoveable + basic_csv_cursor(const basic_csv_cursor&) = delete; + basic_csv_cursor& operator=(const basic_csv_cursor&) = delete; + +public: + using string_view_type = jsoncons::basic_string_view; + + // Constructors that throw parse exceptions + + template + basic_csv_cursor(Sourceable&& source, + const basic_csv_decode_options& options = basic_csv_decode_options(), + std::function err_handler = default_csv_parsing(), + const Allocator& alloc = Allocator(), + typename std::enable_if,Sourceable>::value>::type* = 0) + : source_(std::forward(source)), + parser_(options,err_handler,alloc), + cursor_visitor_(accept_all) + { + if (!done()) + { + next(); + } + } + + template + basic_csv_cursor(Sourceable&& source, + const basic_csv_decode_options& options = basic_csv_decode_options(), + std::function err_handler = default_csv_parsing(), + const Allocator& alloc = Allocator(), + typename std::enable_if,Sourceable>::value>::type* = 0) + : source_(), + parser_(options,err_handler,alloc), + cursor_visitor_(accept_all) + { + jsoncons::basic_string_view sv(std::forward(source)); + initialize_with_string_view(sv); + } + + + // Constructors that set parse error codes + template + basic_csv_cursor(Sourceable&& source, + std::error_code& ec) + : basic_csv_cursor(std::allocator_arg, Allocator(), + std::forward(source), + basic_csv_decode_options(), + default_csv_parsing(), + ec) + { + } + + template + basic_csv_cursor(Sourceable&& source, + const basic_csv_decode_options& options, + std::error_code& ec) + : basic_csv_cursor(std::allocator_arg, Allocator(), + std::forward(source), + options, + default_csv_parsing(), + ec) + { + } + + template + basic_csv_cursor(Sourceable&& source, + const basic_csv_decode_options& options, + std::function err_handler, + std::error_code& ec) + : basic_csv_cursor(std::allocator_arg, Allocator(), + std::forward(source), + options, + err_handler, + ec) + { + } + + template + basic_csv_cursor(std::allocator_arg_t, const Allocator& alloc, + Sourceable&& source, + const basic_csv_decode_options& options, + std::function err_handler, + std::error_code& ec, + typename std::enable_if,Sourceable>::value>::type* = 0) + : source_(std::forward(source)), + parser_(options,err_handler,alloc), + cursor_visitor_(accept_all) + { + if (!done()) + { + next(ec); + } + } + + template + basic_csv_cursor(std::allocator_arg_t, const Allocator& alloc, + Sourceable&& source, + const basic_csv_decode_options& options, + std::function err_handler, + std::error_code& ec, + typename std::enable_if,Sourceable>::value>::type* = 0) + : source_(), + parser_(options,err_handler,alloc), + cursor_visitor_(accept_all) + { + jsoncons::basic_string_view sv(std::forward(source)); + initialize_with_string_view(sv, ec); + } + + template + typename std::enable_if,Sourceable>::value>::type + reset(Sourceable&& source) + { + source_ = std::forward(source); + parser_.reinitialize(); + cursor_visitor_.reset(); + if (!done()) + { + next(); + } + } + + template + typename std::enable_if,Sourceable>::value>::type + reset(Sourceable&& source) + { + source_ = {}; + parser_.reinitialize(); + cursor_visitor_.reset(); + initialize_with_string_view(std::forward(source)); + } + + template + typename std::enable_if,Sourceable>::value>::type + reset(Sourceable&& source, std::error_code& ec) + { + source_ = std::forward(source); + parser_.reinitialize(); + cursor_visitor_.reset(); + if (!done()) + { + next(ec); + } + } + + template + typename std::enable_if,Sourceable>::value>::type + reset(Sourceable&& source, std::error_code& ec) + { + source_ = {}; + parser_.reinitialize(); + initialize_with_string_view(std::forward(source), ec); + } + + bool done() const override + { + return parser_.done(); + } + + const basic_staj_event& current() const override + { + return cursor_visitor_.event(); + } + + void read_to(basic_json_visitor& visitor) override + { + std::error_code ec; + read_to(visitor, ec); + if (ec) + { + JSONCONS_THROW(ser_error(ec,parser_.line(),parser_.column())); + } + } + + void read_to(basic_json_visitor& visitor, + std::error_code& ec) override + { + if (staj_to_saj_event(cursor_visitor_.event(), visitor, *this, ec)) + { + read_next(visitor, ec); + } + } + + void next() override + { + std::error_code ec; + next(ec); + if (ec) + { + JSONCONS_THROW(ser_error(ec,parser_.line(),parser_.column())); + } + } + + void next(std::error_code& ec) override + { + read_next(ec); + } + + static bool accept_all(const basic_staj_event&, const ser_context&) + { + return true; + } + + const ser_context& context() const override + { + return *this; + } + + bool eof() const + { + return parser_.source_exhausted() && source_.eof(); + } + + std::size_t line() const override + { + return parser_.line(); + } + + std::size_t column() const override + { + return parser_.column(); + } + + friend + basic_staj_filter_view operator|(basic_csv_cursor& cursor, + std::function&, const ser_context&)> pred) + { + return basic_staj_filter_view(cursor, pred); + } + +#if !defined(JSONCONS_NO_DEPRECATED) + + JSONCONS_DEPRECATED_MSG("Instead, use read_to(basic_json_visitor&)") + void read(basic_json_visitor& visitor) + { + read_to(visitor); + } + + JSONCONS_DEPRECATED_MSG("Instead, use read_to(basic_json_visitor&, std::error_code&)") + void read(basic_json_visitor& visitor, + std::error_code& ec) + { + read_to(visitor, ec); + } +#endif +private: + + void initialize_with_string_view(string_view_type sv) + { + auto r = unicode_traits::detect_json_encoding(sv.data(), sv.size()); + if (!(r.encoding == unicode_traits::encoding_kind::utf8 || r.encoding == unicode_traits::encoding_kind::undetected)) + { + JSONCONS_THROW(ser_error(json_errc::illegal_unicode_character,parser_.line(),parser_.column())); + } + std::size_t offset = (r.ptr - sv.data()); + parser_.update(sv.data()+offset,sv.size()-offset); + if (!done()) + { + next(); + } + } + + void initialize_with_string_view(string_view_type sv, std::error_code& ec) + { + auto r = unicode_traits::detect_encoding_from_bom(sv.data(), sv.size()); + if (!(r.encoding == unicode_traits::encoding_kind::utf8 || r.encoding == unicode_traits::encoding_kind::undetected)) + { + ec = json_errc::illegal_unicode_character; + return; + } + std::size_t offset = (r.ptr - sv.data()); + parser_.update(sv.data()+offset,sv.size()-offset); + if (!done()) + { + next(ec); + } + } + + void read_next(std::error_code& ec) + { + read_next(cursor_visitor_, ec); + } + + void read_next(basic_json_visitor& visitor, std::error_code& ec) + { + parser_.restart(); + while (!parser_.stopped()) + { + if (parser_.source_exhausted()) + { + auto s = source_.read_buffer(ec); + if (ec) return; + if (s.size() > 0) + { + parser_.update(s.data(),s.size()); + } + } + parser_.parse_some(visitor, ec); + if (ec) return; + } + } +}; + +using csv_stream_cursor = basic_csv_cursor>; +using csv_string_cursor = basic_csv_cursor>; +using wcsv_stream_cursor = basic_csv_cursor>; +using wcsv_string_cursor = basic_csv_cursor>; + +using csv_cursor = basic_csv_cursor; +using wcsv_cursor = basic_csv_cursor; + +}} + +#endif + -- cgit v1.2.3