← All modules

A CSV reader that works from an open file handle a line at a time — quoted fields, a header row that becomes the keys, and the separators and quote characters as options — for files too big to slurp.

Version

0.1.4 zef:tony-o

Depends

nothing outside the core

License

see the source

Its own test suite

5 files, green

Checked

2026-09-14 against Raku++ 3.28.0 and Rakudo 2026.08

Where it lives

raku.land · source

Install it #

$ rakupp install CSV::Parser

zef install CSV::Parser writes the same store; either installer leaves the module usable by both engines.

What it is for #

Splitting a line on commas works until the first field that contains one. CSV's quoting rules are simple to state and tedious to get right — quotes around a field that has a comma or a newline in it, doubled quotes for a literal one — and the files that need them are usually the large ones, an export from a database or a spreadsheet, that you want to read row by row rather than into memory at once. This distribution reads from a handle you opened, hands back one row per call, and if the first row is a header it uses those names as the keys.

Rows from a handle #

use CSV::Parser;

my $file = $*TMPDIR.add("towns-{$*PID}.csv");
$file.spurt(qq:to/CSV/);
    name,population,motto
    Ashby,1200,"Small, but proud"
    Brill,950,"Says ""hello"""
    Cowes,10400,
    CSV
my $fh = $file.open;
my $csv = CSV::Parser.new(file_handle => $fh, contains_header_row => True);
until $fh.eof {
    my %row = $csv.get_line;
    next unless %row;
    say %row.keys.sort.map({ "$_=" ~ %row{$_} }).join(' | ');
}
$fh.close;
say $csv.headers.sort.map(*.value).join(',');
$file.unlink;

motto=Small, but proud | name=Ashby | population=1200
motto=Says ""hello"" | name=Brill | population=950
motto= | name=Cowes | population=10400
name,population,motto

get_line reads the next record — a record, not a line, since a quoted field may span several — and answers a hash. With contains_header_row the keys are the header's names; without it they are the column positions as strings, "0", "1" and so on, which headers reports either way. field_separator, line_separator, field_operator (the quote character) and escape_operator turn it into a reader for tab-separated files or a dialect with single quotes.

The one thing to know #

A doubled quote inside a quoted field is left doubled. The standard way to write a literal " in CSV is "", and the second row above shows what this parser does with it: Says ""hello"" comes out with both pairs intact, where every spreadsheet would show Says "hello". The quoting is otherwise honoured — the comma in Small, but proud did not split the field — so this is an omission at one step of the unquoting, and one you can repair after the fact with .subst('""', '"', :g) on the fields that were quoted. Data that contains a lot of quotation marks is the case to check before relying on it.