Reading CSV files

phit includes a native CSV reader, Phit\read_csv(), that tokenizes a file, guesses each column's type, and parses the values in Rust. Its rules follow readr 2 and match phibble's PHP reader. When the extension is loaded, phibble's DataFrame::fromCsv() uses it.

Edit the text below and the page reads it again. The sample shows several of the rules. A ZIP code with a leading zero stays text, quoted numbers with grouping marks read as doubles, and true and false spellings read as booleans. The invalid date 2026-02-30 fails to parse, and the short last row is padded with NA and reported as a problem.

Code

use function Phit\read_csv;

$result = read_csv($path, ['delimiter' => ',']);

$result['names'];     // column names
$result['columns'];   // per column: type, values, and the count that failed to parse
$result['problems'];  // ragged rows and unterminated quotes

$df = DataFrame::fromCsv($path);  // phibble, through the same reader

The phit documentation lists every option and guessing rule. In phibble's 1,000,000-row benchmark on an Apple Silicon Mac, DataFrame::fromCsv() takes 0.12 seconds with the extension and 1.0 seconds with the PHP reader.