CSV Parser 2.0.1

• Made parsing CSV files without header rows more convenient
• Fixed a compilation error with std::back_inserter on some systems

CSV Parser 2.0.0

• Parser now uses memory-mapped IO for reading from disk thanks to mio
  • CSV files are read in smaller chunks to reduce memory footprint (but parsing is significantly faster)
• CSVReader::read_row() (and CSVReader::iterator) no longer blocks CSVReader::read_csv(), i.e. we can now simultaneously work on CSV data while reading more rows
• Parser internals completely rewritten to use more efficient and easier to maintain/debug data structures
• Fixed bug where single column files could not be parsed
• Fixed errors with parsing empty files
• CSVWriter::write_row() now works with std::array

CSV Parser 2.0 Beta: >300 MBps Edition

• Parser now uses memory-mapped IO for reading from disk
  • On Windows, parser may map entire file into memory or mmap chunks of file iteratively based on available RAM (will extend to all OSes)
• Parser internals completely rewritten to use more efficient and easier to maintain/debug data structures
  • New algorithm involves minimal copying
• Fixed bug where single column files could not be parsed
• Fixed errors with parsing empty files

Fixed memory errors when parsing large files

• Fixed issue with incorrect usage of string_view that led to memory errors when parsing large files such as the 1.4GB Craigslist vehicles dataset #90
• Added ability to have no quote character #83
• Changed VariableColumnPolicy::IGNORE to IGNORE_ROW to avoid clashing with IGNORE macro as defined by WinBase.h #96

Fixed bug with parsing very long rows

• Fixed bug with parsing very long rows (as reported in #92) when the length of the row was greater than 2^16 (the limit of unsigned short)
  • All instances of unsigned short have been replaced by internals::StrBufferPos (size_t) thus giving this parser the theoretical capability of parsing rows that are 2^64 characters long
• Fixed bug recognizing numbers in e-notation when the base did not have a decimal, e.g. 1E-06

Fixed bug with whitespace trimming when a field is entirely whitespace

Fixes incorrect CSV parsing when whitespace trimming is enabled and a field is composed entirely of whitespace characters as reported in #85

First class handling of variable column CSVs

• The behavior for parsing variable-column CSV files can now be simply defined using CSVFormat::variable_columns()
  • Variable-column rows can be kept or silently dropped (default), or result in an error being thrown
  • CSVReader::bad_row_handler() has been removed
• Many annoying clang/gcc warning messages fixed (thanks rpavlik!)
• CSV guessing implementation has been simplified (CSVGuesser is also gone now)

Fixed bug where get<>() threw incorrect overflow errors with unsigned integers

Fixed bug reported in #73

Fixed Issue with Leading Comments

Fixed issue described by #67 where leading comments got concatenated to the first column name

Fixed bug reading rows that begin with empty fields

Fix bug when CSV rows have leading empty fields (#57)

* Fixed bug --> all tests passing :bug:

* Fixed some MSVC warnings