StreamCSV

A CSV-Parsing library with an emphasis on performance.

StreamCSV can read CSV data from streams of data. This means you can parse arbitrarily large amounts of data without having to fit it all into memory at once.

The high-level streamcsv module

A simple example, this will iterate over all records ("lines") of a CSV file:

for idx, record in streamcsv.records(io.open("test.csv")) do
	print(table.unpack(record))
end

This file will be treated as having no initial header line. To indicate that the first line should be read as a header, set the header option to true:

for idx, record in streamcsv.records(io.open("test2.csv"), {header=true}) do
	print(table.unpack(record))
end

When a header is not part of the CSV file, but is otherwise known, it can be passed explicitly:

streamcsv.records("1,2,3", {header="a,b,c"})
-- or
streamcsv.records("1,2,3", {header={"a", "b", "c"}})

The write module

This module takes care of writing CSV data to strings. It has one function each for writing out fields, records and entire files.

The low-level streamcsv.read module

Generally, all the functions of this module are structured as follows:

Arguments

1. An initial string (can be empty)
2. At what position in the initial string to start parsing
3. A string containing the record and field separators (default "\n,")
4. A generator function that returns the next string from the stream or nil

Note that if the generator function should return an empty string, the parser functions will immediately attempt to call it again, potentially leading to an endless loop.

Return values

1. The main result of the function
2. The new current string (same as argument, or result of generator)
3. The first position in the current string that has not yet been consumed