Обнаружил для себя, что распарсить многострочный CSV в php не так просто, как казалось. Встроенные функции, типа str_getcsv и проч, работают только с 1 строкой, но у нас их, как правило, много. Варианты типа разбить строку через explode или прочитать файл через file не учитывают того, что в CSV строке вполне может быть перевод строки, например:

Text

«str1
str2»

Воспользовавшись комментарием соорудил простой класс для парсинга многострочного CSV:

PHP

<?php

 

declare(strict_types=1);

 

// https://www.php.net/manual/ru/function.str-getcsv.php#113220 

final readonly class CsvParser

{

    /**

     * parse a CSV file into a two-dimensional array

     * this seems as simple as splitting a string by lines and commas, but this only works if tricks are performed

     * to ensure that you do NOT split on lines and commas that are inside of double quotes.

     *

     * @return array<int, string[]>

     */

    public function parse(string $str): array

    {

        // match all the non-quoted text and one series of quoted text (or the end of the string)

        // each group of matches will be parsed with the callback, with $matches[1] containing all the non-quoted text,

        // and $matches[3] containing everything inside the quotes

        $str = preg_replace_callback('/([^"]*)("((""|[^"])*)"|$)/', $this->parseQuotes(...), $str);

 

        // remove the very last newline to prevent a 0-field array for the last line

        $str = preg_replace('/n$/', '', $str);

 

        // split on LF and parse each line with a callback

        return array_map($this->parseLine(...), explode("n", $str));

    }

 

    /**

     * replace all the csv-special characters inside double quotes with markers using an escape sequence.

     */

    private function parseQuotes(array $matches): string

    {

        // anything inside the quotes that might be used to split the string into lines and fields later,

        // needs to be quoted. The only character we can guarantee as safe to use, because it will never appear in the unquoted text, is a CR

        // So we're going to use CR as a marker to make escape sequences for CR, LF, Quotes, and Commas.

        $str = $matches[3] ?? '';

        $str = str_replace(["r", "n", '""', ','], ["rR", "rN", "rQ", "rC"], $str);

 

        // The unquoted text is where commas and newlines are allowed, and where the splits will happen

        // We're going to remove all CRs from the unquoted text, by normalizing all line endings to just LF

        // This ensures us that the only place CR is used, is as the escape sequences for quoted text

        return preg_replace('/rn?/', "n", $matches[1]).$str;

    }

 

    /**

     * split on comma and parse each field with a callback.

     */

    private function parseLine(string $line): array

    {

        return array_map($this->parseField(...), explode(',', $line));

    }

 

    /**

     * restore any csv-special characters that are part of the data.

     */

    private function parseField(string $field): string

    {

        return str_replace(["rC", "rQ", "rN", "rR"], [',', '"', "n", "r"], $field);

    }

}

Источник: