A class for parsing CSV data from any std::istream, including non-seekable sources such as pipes and decompression filters. More...

Public Member Functions
	StreamParser (TStream &source, const CSVFormat &format, const ColNamesPtr &col_names=nullptr)

	StreamParser (TStream &source, internals::ParseFlagMap parse_flags, internals::WhitespaceMap ws_flags)

std::string &	get_csv_head () override

void	next (size_t bytes=CSV_CHUNK_SIZE_DEFAULT) override
	Parse the next block of data.

Public Member Functions inherited from csv::internals::IBasicCSVParser
	IBasicCSVParser (const CSVFormat &, const ColNamesPtr &)

	IBasicCSVParser (const ParseFlagMap &parse_flags, const WhitespaceMap &ws_flags)

bool	eof ()
	Whether or not we have reached the end of source.

ResolvedFormat	get_resolved_format ()

void	end_feed ()
	Indicate the last block of data has been parsed.

CONSTEXPR_17 ParseFlags	parse_flag (const char ch) const noexcept

CONSTEXPR_17 ParseFlags	compound_parse_flag (const char ch) const noexcept

CONSTEXPR bool	utf8_bom () const
	Whether or not this CSV has a UTF-8 byte order mark.

void	set_output (RowCollection &rows)

Additional Inherited Members
Protected Member Functions inherited from csv::internals::IBasicCSVParser
CONSTEXPR bool	no_chunk () const
	Whether or not source needs to be read in chunks.

size_t	parse ()
	Parse the current chunk of data and return the completed-row prefix length.

void	reset_data_ptr ()
	Create a new RawCSVDataPtr for a new chunk of data.

void	resolve_format_from_head (const CSVFormat &format)

Protected Attributes inherited from csv::internals::IBasicCSVParser
CSVRow	current_row_

RawCSVDataPtr	data_ptr_ = nullptr

ColNamesPtr	col_names_ = nullptr

RawCSVFieldList *	fields_ = nullptr

int	field_start_ = UNINITIALIZED_FIELD

size_t	field_length_ = 0

SentinelVecs	simd_sentinels_
	Precomputed SIMD broadcast vectors for find_next_non_special.

ParseFlagMap	parse_flags_
	An array where the (i + 128)th slot gives the ParseFlags for ASCII character i.

bool	eof_ = false

ResolvedFormat	format

size_t	source_size_ = 0
	The size of the incoming CSV.

Detailed Description

template<typename TStream>
class csv::internals::StreamParser< TStream >

A class for parsing CSV data from any std::istream, including non-seekable sources such as pipes and decompression filters.

Chunk boundary handling: parse() returns the byte offset of the start of the last incomplete row in the current chunk (the "remainder"). Rather than seeking back to re-read those bytes (which requires a seekable stream), they are saved in leftover_ and prepended to the next chunk. This is semantically identical to the old seek-back approach but works on any istream and avoids the syscall overhead of seekg().

Format resolution: The constructor reads a head buffer from the stream via get_csv_head() and passes it to resolve_format_from_head(), which infers the delimiter and header row when not explicitly set. The head bytes are stored in leftover_ so the first next() call re-parses them without re-reading.

Definition at line 316 of file basic_csv_parser.hpp.