cii-text-analysis/save/Lib/StringClassifier.php.save

<?php

declare(strict_types=1);

namespace Classifier\Lib;

use Symfony\Component\Console\Style\SymfonyStyle;
use Symfony\Component\Console\Input\ArgvInput;
use Symfony\Component\Console\Output\ConsoleOutput;

class StringClassifier
{
    const OUTPUT_DIR = __DIR__ . '/../output';
    const GOOD_FILE = self::OUTPUT_DIR . '/positive.php';
    const BAD_FILE = self::OUTPUT_DIR . '/negative.php';
    const IGNORE_FILE = self::OUTPUT_DIR . '/ignored.php'; // Nouveau fichier

    /** @var string[] */
    private $strings = [];

    /** @var SymfonyStyle */
    private $io;

    /** @var array{good: string[], bad: string[], ignored: string[]} */
    private $results = ['good' => [], 'bad' => [], 'ignored' => []];

    public function __construct(array $strings)
    {
        $this->io = new SymfonyStyle(new ArgvInput(), new ConsoleOutput());

        $this->loadExistingData();

        $cleaned = array_unique(array_map(function (string $value): string {
            return preg_replace('/\s+/', ' ', trim(mb_strtolower($value)));
        }, $strings));

        $alreadyDone = array_merge(
            $this->results['good'],
            $this->results['bad'],
            $this->results['ignored']
        );

        $this->strings = array_diff($cleaned, $alreadyDone);

        if (!is_dir(self::OUTPUT_DIR)) {
            mkdir(self::OUTPUT_DIR, 0777, true);
        }
    }

    private function loadExistingData(): void
    {
        $files = [
            'good' => self::GOOD_FILE,
            'bad' => self::BAD_FILE,
            'ignored' => self::IGNORE_FILE
        ];

        foreach ($files as $key => $path) {
            if (file_exists($path)) {
                $this->results[$key] = include $path;
            }
        }
    }

    public function runQuestions(): void
    {
        $count = count($this->strings);
        if ($count === 0) {
            $this->io->success("Tous les messages fournis sont déjà labellisés ou ignorés !");
            return;
        }

        $this->io->title('Labellisation des données (Mode Résilient)');
        $this->io->note([
            "Appuyez sur 'i' pour écarter un message définitivement.",
            "Appuyez sur 'q' pour sauvegarder et quitter."
        ]);

        $this->io->progressStart($count);

        foreach ($this->strings as $string) {
            $this->io->newLine(2);
            $this->io->text("<info>Message :</info> $string");

            $answer = $this->io->choice(
                'Action pour ce message ?',
                [
                    'y' => 'Oui (Alerte)',
                    'n' => 'Non (Autre)',
                    'i' => 'Ignorer (Passer)',
                    'q' => 'Quitter et Sauvegarder'
                ],
                'n'
            );

            if ($answer === 'q') {
                $this->io->note("Interruption demandée...");
                break;
            }

            if ($answer === 'y') {
                $this->results['good'][] = $string;
            } elseif ($answer === 'i') {
                $this->results['ignored'][] = $string;
            } else {
                $this->results['bad'][] = $string;
            }

            $this->io->progressAdvance();
        }

        $this->io->progressFinish();
        $this->saveResults();

        $this->io->success("Sauvegarde effectuée. Ignorés : " . count($this->results['ignored']));
    }

    private function saveResults(): void
    {
        $this->exportToFile(self::GOOD_FILE, $this->results['good']);
        $this->exportToFile(self::BAD_FILE, $this->results['bad']);
        $this->exportToFile(self::IGNORE_FILE, $this->results['ignored']);
    }

    private function exportToFile(string $path, array $data): void
    {
        sort($data);
        $content = "<?php\n\nreturn " . var_export($data, true) . ";\n";
        file_put_contents($path, $content);
    }
}