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_reader.hpp | 348 ++++++++++++++++++++++++++++++++ 1 file changed, 348 insertions(+) create mode 100644 include/jsoncons_ext/csv/csv_reader.hpp (limited to 'include/jsoncons_ext/csv/csv_reader.hpp') diff --git a/include/jsoncons_ext/csv/csv_reader.hpp b/include/jsoncons_ext/csv/csv_reader.hpp new file mode 100644 index 0000000..f10211a --- /dev/null +++ b/include/jsoncons_ext/csv/csv_reader.hpp @@ -0,0 +1,348 @@ +// Copyright 2013 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_READER_HPP +#define JSONCONS_CSV_CSV_READER_HPP + +#include +#include +#include +#include // std::allocator +#include // std::move +#include // std::basic_istream +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +namespace jsoncons { namespace csv { + + template,class Allocator=std::allocator> + class basic_csv_reader + { + struct stack_item + { + stack_item() noexcept + : array_begun_(false) + { + } + + bool array_begun_; + }; + using char_type = CharT; + using temp_allocator_type = Allocator; + typedef typename std::allocator_traits:: template rebind_alloc char_allocator_type; + + basic_csv_reader(const basic_csv_reader&) = delete; + basic_csv_reader& operator = (const basic_csv_reader&) = delete; + + basic_default_json_visitor default_visitor_; + text_source_adaptor source_; + basic_json_visitor& visitor_; + basic_csv_parser parser_; + + public: + // Structural characters + static constexpr size_t default_max_buffer_size = 16384; + //! Parse an input stream of CSV text into a json object + /*! + \param is The input stream to read from + */ + + template + basic_csv_reader(Sourceable&& source, + basic_json_visitor& visitor, + const Allocator& alloc = Allocator()) + + : basic_csv_reader(std::forward(source), + visitor, + basic_csv_decode_options(), + default_csv_parsing(), + alloc) + { + } + + template + basic_csv_reader(Sourceable&& source, + basic_json_visitor& visitor, + const basic_csv_decode_options& options, + const Allocator& alloc = Allocator()) + + : basic_csv_reader(std::forward(source), + visitor, + options, + default_csv_parsing(), + alloc) + { + } + + template + basic_csv_reader(Sourceable&& source, + basic_json_visitor& visitor, + std::function err_handler, + const Allocator& alloc = Allocator()) + : basic_csv_reader(std::forward(source), + visitor, + basic_csv_decode_options(), + err_handler, + alloc) + { + } + + template + basic_csv_reader(Sourceable&& source, + basic_json_visitor& visitor, + const basic_csv_decode_options& options, + std::function err_handler, + const Allocator& alloc = Allocator()) + : source_(std::forward(source)), + visitor_(visitor), + parser_(options, err_handler, alloc) + + { + } + + ~basic_csv_reader() noexcept = default; + + void read() + { + std::error_code ec; + read(ec); + if (ec) + { + JSONCONS_THROW(ser_error(ec,parser_.line(),parser_.column())); + } + } + + void read(std::error_code& ec) + { + read_internal(ec); + } + + std::size_t line() const + { + return parser_.line(); + } + + std::size_t column() const + { + return parser_.column(); + } + + bool eof() const + { + return parser_.source_exhausted() && source_.eof(); + } + + private: + + void read_internal(std::error_code& ec) + { + if (source_.is_error()) + { + ec = csv_errc::source_error; + return; + } + 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; + } + } + }; + + template,class Allocator=std::allocator> + class legacy_basic_csv_reader + { + struct stack_item + { + stack_item() noexcept + : array_begun_(false) + { + } + + bool array_begun_; + }; + using char_type = CharT; + using temp_allocator_type = Allocator; + typedef typename std::allocator_traits:: template rebind_alloc char_allocator_type; + + legacy_basic_csv_reader(const legacy_basic_csv_reader&) = delete; + legacy_basic_csv_reader& operator = (const legacy_basic_csv_reader&) = delete; + + basic_default_json_visitor default_visitor_; + text_source_adaptor source_; + basic_json_visitor& visitor_; + basic_csv_parser parser_; + + public: + // Structural characters + static constexpr size_t default_max_buffer_size = 16384; + //! Parse an input stream of CSV text into a json object + /*! + \param is The input stream to read from + */ + + template + legacy_basic_csv_reader(Sourceable&& source, + basic_json_visitor& visitor, + const Allocator& alloc = Allocator()) + + : legacy_basic_csv_reader(std::forward(source), + visitor, + basic_csv_decode_options(), + default_csv_parsing(), + alloc) + { + } + + template + legacy_basic_csv_reader(Sourceable&& source, + basic_json_visitor& visitor, + const basic_csv_decode_options& options, + const Allocator& alloc = Allocator()) + + : legacy_basic_csv_reader(std::forward(source), + visitor, + options, + default_csv_parsing(), + alloc) + { + } + + template + legacy_basic_csv_reader(Sourceable&& source, + basic_json_visitor& visitor, + std::function err_handler, + const Allocator& alloc = Allocator()) + : legacy_basic_csv_reader(std::forward(source), + visitor, + basic_csv_decode_options(), + err_handler, + alloc) + { + } + + template + legacy_basic_csv_reader(Sourceable&& source, + basic_json_visitor& visitor, + const basic_csv_decode_options& options, + std::function err_handler, + const Allocator& alloc = Allocator(), + typename std::enable_if,Sourceable>::value>::type* = 0) + : source_(std::forward(source)), + visitor_(visitor), + parser_(options, err_handler, alloc) + { + } + + template + legacy_basic_csv_reader(Sourceable&& source, + basic_json_visitor& visitor, + const basic_csv_decode_options& options, + std::function err_handler, + const Allocator& alloc = Allocator(), + typename std::enable_if,Sourceable>::value>::type* = 0) + : source_(), + visitor_(visitor), + parser_(options, err_handler, alloc) + { + jsoncons::basic_string_view sv(std::forward(source)); + 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)) + { + 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); + } + + ~legacy_basic_csv_reader() noexcept = default; + + void read() + { + std::error_code ec; + read(ec); + if (ec) + { + JSONCONS_THROW(ser_error(ec,parser_.line(),parser_.column())); + } + } + + void read(std::error_code& ec) + { + read_internal(ec); + } + + std::size_t line() const + { + return parser_.line(); + } + + std::size_t column() const + { + return parser_.column(); + } + + bool eof() const + { + return parser_.source_exhausted() && source_.eof(); + } + + private: + + void read_internal(std::error_code& ec) + { + if (source_.is_error()) + { + ec = csv_errc::source_error; + return; + } + 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; + } + } + }; + +#if !defined(JSONCONS_NO_DEPRECATED) + using csv_reader = legacy_basic_csv_reader; + using wcsv_reader = legacy_basic_csv_reader; +#endif + + using csv_string_reader = basic_csv_reader>; + using wcsv_string_reader = basic_csv_reader>; + using csv_stream_reader = basic_csv_reader>; + using wcsv_stream_reader = basic_csv_reader>; + +}} + +#endif -- cgit v1.2.3