php-dataset/src/Readers/CsvReader.php

<?php

namespace NoccyLabs\Dataset\Readers;

use NoccyLabs\Dataset\ReaderInterface;

class CsvReader implements ReaderInterface
{
    private array $files = [];

    private array $options = [];

    private int $currentFile = 0;

    private ?int $loadedFile = null;

    private array $data = [];

    private int $currentIndex = 0;

    private int $counter = 0;

    public function __construct(string $filename, array $options)
    {
        $this->files = glob($filename);
        $this->options = $options;
    }

    private function checkLoadedSlice()
    {
        // If the current file is the loaded file, we're already set
        if ($this->currentFile === $this->loadedFile) return;

        if ($this->currentFile >= count($this->files)) {
            //printf("Reached end of set at slice=%d\n", $this->currentFile);
            return;
        }
        $file = $this->files[$this->currentFile]; 
        $csv = @file($file, FILE_SKIP_EMPTY_LINES|FILE_IGNORE_NEW_LINES);

        $this->loadData($csv);
        $this->loadedFile = $this->currentFile;
        //printf("loaded slice %d: %s\n", $this->currentFile, $file);
    }

    private function loadData(array $data)
    {
        $separator = $this->options['separator']??',';
        $enclosure = $this->options['enclosure']??'"';
        $escape = $this->options['escape']??"\\";


        $head = str_getcsv(array_shift($data));
        $this->data = [];
        foreach ($data as $row) {
            if ($row) {
                $row = str_getcsv($row, $separator, $enclosure, $escape);
                $this->data[] = array_combine($head, $row);
            }
        }
        $this->currentIndex = 0;
    }

    public function rewind(): void
    {
        $this->currentFile = 0;
        $this->currentIndex = 0;
        $this->counter = 0;
        //printf("Rewinding to slice=%d index=%d\n", $this->currentFile, $this->currentIndex);
        $this->checkLoadedSlice();
    }

    public function key()
    {
        //$this->checkLoadedSlice();
        return $this->counter;
    }

    public function current()
    {
        //$this->checkLoadedSlice();   
        return $this->data[$this->currentIndex];
    }

    public function next(): void
    {
        $this->counter++;        
        $this->currentIndex++;
        
        if ($this->currentIndex >= count($this->data)) {
            $this->currentFile++;
            $this->currentIndex = 0;
            //printf("Rolling over to slice=%d index=%d counter=%d\n", $this->currentFile, $this->currentIndex, $this->counter);
        }

        //$this->checkLoadedSlice();   
    }

    public function valid(): bool
    {
        $this->checkLoadedSlice();
        return ($this->currentFile < count($this->files) && ($this->currentIndex < count($this->data)));
        
    }
}