Voting

The Note You're Voting On

1 year ago
old MacOS (up to ~2001) and old Office For MacOS (up to 2007? I think) use carriage-return for newlines, 
Microsoft Windows use carriage-return+line-feed for newlines,
Unix (Linux and modern MacOS) use line-feeds, 
Some systems use BOM/byte-order-masks just to say they use UTF-8, i've even encountered one-BOM-per-CSV-row!

For a csv-file parser handling all the above cases, I wrote:

<?php
function parse_csv(string $csv, string $separator = ","): array
{
    $csv = strtr(
        $csv,
        [
            "\xEF\xBB\xBF" => "", // remove UTF-8 byte order masks, if present
            "\r\n" => "\n", // Windows CrLf=> Unix Lf
            "\r" => "\n" // old-MacOS Cr => Unix Lf
            // (both modern MacOS and Linux use Lf .. Windows is the only outlier)
        ]
    );
    $lines = explode("\n", $csv);
    $keys = str_getcsv(array_shift($lines), $separator);
    $ret = array();
    foreach ($lines as $lineno => $line) {
        if (strlen($line) < 1) {
            // ... probably malformed csv, but we'll allow it
            continue;
        }
        $parsed = str_getcsv($line, $separator);
        if (count($parsed) !== count($keys)) {
            throw new \RuntimeException("error on csv line #{$lineno}: count mismatch:" . count($parsed) . ' !== ' . count($keys) . ": " . var_export([
                'error' => 'count mismatch',
                'keys' => $keys,
                'parsed' => $parsed,
                'line' => $line
            ], true));
        }
        $ret[] = array_combine($keys, $parsed);
    }
    return $ret;
}
?>
<< Back to user notes page