Many, many times it is written in the documentation differently than it actually is.

[gander]

Bug Report

Many, many times it is written in the documentation differently than it actually is.

Information	Description
Version	9.21.0
PHP version	8.x
OS Platform	Linux

Summary

e.g. here https://csv.thephpleague.com/9.0/reader/tabular-data-reader/ getRecords

The getRecords enables iterating over all records from the current object. If the optional $header argument is given, it will be used as a mapper on the record and will update the record header and the value position.

Currently, each call to getRecords() returns a records mapped from headers. Each method returns mapped if I have set setHeaderOffset.

[nyamsprod]

@gander thanks for using the package could you provide an example of what you are describing because I fail to understand the actual issue you are referring to 🤔

[gander]

CSV with header row

$reader = Reader::createFromStream(fopen(__DIR__.'/input.csv', 'r'))
    ->setHeaderOffset(0)
    ->getRecords()
;

Result:

[[
 "x" => "A",
 "y" => "B",
 "z" => "C",
]]

Should be:

[[
 0 => "A",
 1 => "B",
 2 => "C",
]]

[gander]

Following documentation, $header argument should be TRULY optional, and to map, should be used code:

$records = $reader->getRecords($reader->getHeader());

[gander]

For now, my workaround:

$reader = Reader::createFromStream(fopen(__DIR__.'/input.csv', 'r'));
$header = $reader->slice(0, 1)->first();
$records = iterator_to_array($reader->slice(1));

[nyamsprod]

CSV with header row

$reader = Reader::createFromStream(fopen(__DIR__.'/input.csv', 'r'))
    ->setHeaderOffset(0)
    ->getRecords()
;

Result:

[[
 "x" => "A",
 "y" => "B",
 "z" => "C",
]]

Should be:

[[
 0 => "A",
 1 => "B",
 2 => "C",
]]

You are incorrect ... it is doing exactly what it is said in the documentation.

By setting the header offset you are instructing the Reader class to use the content of the row at index 0 as the header and whenever you call the getRecords method the header is calculated and used to create the associative array that you have. Hence why I fail to understand your issue. Also the library remove the selected header record from the result as it is the header and not the actual content of the CSV.

If you wish not to use the header offset and have it included in your CSV object just do not use setHeaderOffset or use setHeaderOffset(null).

So this is definitely not a bug but expected behaviour.

[gander]

So how do you use setHeaderOffset, get the header, and have the records not mapped with the header?

[nyamsprod]

if you do not care about the header being set just iterate over the value and use array_values

$reader = Reader::createFromPath(__DIR__.'/input.csv', 'r');
$reader->setHeaderOffset(0);
iterator_to_array($reader->map(fn (array $row) => array_values($row)))

[gander]

Still need header, some data mapped, and not mapped, so my workaround

$header = $reader->slice(0, 1)->first();
$reader->slice(666, 1)->mapHeader($header)->first();

[gander]

I still think that there is room for improvement, but in the documentation. In one place it says that setHeaderOffset will result in mapping, but for every single function that has a different result, depending on the use of setHeaderOffset, there should be such an annotation.

[nyamsprod]

I really do not understand what you are trying to achieve but I guess it serves your business logic and that's fine. The only thing I know it's that what you are doing is not a workaround because there is no bug to work around 🤷🏾‍♂️