Spatie\PdfToText\Pdf

Complexity

Total Complexity

10

Size/Duplication

Total Lines	79
Duplicated Lines	0 %

Coupling/Cohesion

Components	1
Dependencies	3

Importance

Changes

0

7 Methods

Rating	Name	Duplication	Size	Complexity
A	__construct()	0	4	1
A	setPdf()	0	10	2
A	setOptions()	0	6	1
A	addOptions()	0	9	1
A	parseOptions()	0	17	2
A	text()	0	10	2
A	getText()	0	8	1

<?php

namespace Spatie\PdfToText;

use Spatie\PdfToText\Exceptions\CouldNotExtractText;
use Spatie\PdfToText\Exceptions\PdfNotFound;
use Symfony\Component\Process\Process;

class Pdf
{
    protected $pdf;

    protected $binPath;

    protected $options = [];

    public function __construct(string $binPath = null)
    {
        $this->binPath = $binPath ?? '/usr/bin/pdftotext';
    }

    public function setPdf(string $pdf): self
    {
        if (!is_readable($pdf)) {
            throw new PdfNotFound("Could not read `{$pdf}`");
        }

        $this->pdf = $pdf;

        return $this;
    }

    public function setOptions(array $options): self
    {
        $this->options = $this->parseOptions($options);

        return $this;
    }

    public function addOptions(array $options): self
    {
        $this->options = array_merge(
            $this->options,
            $this->parseOptions($options)
        );

        return $this;
    }

    protected function parseOptions(array $options): array
    {
        $mapper = function (string $content) : array {
            $content = trim($content);
            if ('-' !== ($content[0] ?? '')) {
                $content = '-'.$content;
            }

            return explode(' ', $content, 2);
        };

        $reducer = function (array $carry, array $option) : array {
            return array_merge($carry, $option);
        };

        return array_reduce(array_map($mapper, $options), $reducer, []);
    }

    public function text() : string
    {
        $process = new Process(array_merge([$this->binPath], $this->options, [$this->pdf, '-']));
        $process->run();
        if (!$process->isSuccessful()) {
            throw new CouldNotExtractText($process);
        }

        return trim($process->getOutput(), " \t\n\r\0\x0B\x0C");
    }

    public static function getText(string $pdf, string $binPath = null, array $options = []) : string
    {
        return (new static($binPath))
            ->setOptions($options)
            ->setPdf($pdf)
            ->text()
        ;
    }
}