Kata

We need valid CSV parser!

A CSV (Comma-Separated Values) file is a file that represents tabular data in a simple, text-only manner. At its simplest, a CSV file contains rows of data separated by newlines, and each field is separated by commas.

This parser needs to not only handle CSVs using commas to delimit fields, but it also needs to handle complex field values (which may be wrapped in quotes, and may span multiple lines), and also different delimiters and quote characters.

Examples

A simple CSV would look like this:

a,b,c
d,e,f

which should parse to:

[['a', 'b', 'c'], ['d', 'e', 'f']

While a more complex one might be:

one,"two wraps
onto ""two"" lines",three
4,,6

that becomes

[['one', 'two wraps\nonto "two" lines', 'three'], ['4', '', '6']]

Specification:

Basics

• The parser should return an array of arrays, one array for each row of the CSV file (not necessarily each line of text!).
• Rows are delimited by the newline character ("\n").
• Each row is divided by a separator character, by default the comma (,). All characters between separators are part of the value — do not trim the field value.
• Fields are assumed to be strings — don't convert them to numbers or other types in this kata.
• Empty fields are valid — don't discard empty values at the beginning, middle or end of a row. These should be included as an empty string.
• Likewise, an empty row is still valid, and effectively contains a single empty field.
• For this kata, expect uneven rows. Just include the actual fields in each row, even if the rows have a different number of fields.

Quoted Fields

• The parser should handle quoted fields.
• A quoted field starts and ends with the same character, and every character in between makes up the field value, including separator characters. The default quote character is a double quote (").
• Quoted fields may span multiple lines — don't assume a newline means a new row!
• Quoted fields only start immediately following a separator character, newline, or start of the file. If a quote character occurs anywhere else, it should be treated as a normal field value.
• They should be immediately followed by a separator, newline, or the end of the file.
• If a quote character occurs within a quoted field, it is simply doubled. For example, the value foo "bar" baz will be encoded as "foo ""bar"" baz". The parser should identify and unescape these values.
• You should check for unclosed quoted fields, and throw an error, but this is not tested here.

Alternate Characters

• The parser should handle alternate characters for the separator and quote.
• You may safely assume that the values provided are a single character, but bonus points for error checking.