molenda.net/src/Application/Services/BlogProvider.php

<?php

declare(strict_types=1);

namespace App\Application\Services;

class BlogProvider
{
    private array $tableCache = [];
    private string $cacheDir;

    public function __construct()
    {
        $this->cacheDir = dirname(__DIR__, 3) . '/var/cache/markov';
        if (!is_dir($this->cacheDir)) {
            @mkdir($this->cacheDir, 0775, true);
        }
    }

    private function generate_markov_table($text, $look_forward = 4)
    {
        // build counts in a single pass and compute cumulative weights for fast selection
        $len = strlen($text);
        $table = [];

        // single pass to build counts
        $max = $len - $look_forward;
        for ($i = 0; $i < $max; $i++) {
            $key = substr($text, $i, $look_forward);
            $next = substr($text, $i + $look_forward, $look_forward);
            if (!isset($table[$key])) $table[$key] = [];
            if (isset($table[$key][$next])) {
                $table[$key][$next]++;
            } else {
                $table[$key][$next] = 1;
            }
        }

        // convert counts into cumulative arrays with totals for fast selection
        // to save memory we map next-token strings to integer ids and keep a shared token list
        $tokenIndex = [];
        $tokenList = [];
        $nextId = 0;

        foreach ($table as $key => $counts) {
            $cum = [];
            $sum = 0;
            foreach ($counts as $item => $weight) {
                if (!isset($tokenIndex[$item])) {
                    $tokenIndex[$item] = $nextId;
                    $tokenList[$nextId] = $item;
                    $nextId++;
                }
                $id = $tokenIndex[$item];
                $sum += $weight;
                $cum[] = ['id' => $id, 'cum' => $sum];
            }
            $table[$key] = ['total' => $sum, 'cum' => $cum];
        }

        // attach token list for this table so picker can map ids back to strings
        $table['_tokens'] = $tokenList;

        return $table;
    }

    private function generate_markov_text($length, $table, $look_forward = 4)
    {
        // pick a random starting state
        $states = array_keys($table);
        if (!$states) return '';
        $char = $states[array_rand($states)];
        $o = $char;

        $iterations = (int)ceil($length / max(1, $look_forward));
        for ($i = 0; $i < $iterations; $i++) {
            $entry = $table[$char] ?? null;
            if ($entry && !empty($entry['cum'])) {
                $newId = $this->pick_weighted($entry);
                if ($newId !== false) {
                    $tokenList = $table['_tokens'] ?? [];
                    $newchar = $tokenList[$newId] ?? false;
                    if ($newchar !== false) {
                        $char = $newchar;
                        $o .= $newchar;
                        continue;
                    }
                }
            }
            // fallback: pick another random state
            $char = $states[array_rand($states)];
            $o .= $char;
        }

        return $o;
    }

    private function pick_weighted(array $entry)
    {
        // entry contains 'total' and 'cum' keys
        if (empty($entry['cum']) || empty($entry['total'])) return false;
        $rand = mt_rand(1, $entry['total']);
        foreach ($entry['cum'] as $pair) {
            if ($rand <= $pair['cum']) return $pair['id'];
        }
        return false;
    }

    private function generate($length)
    {
        $path = dirname(__DIR__, 3) . '/lib/php_text.txt';
        if (!is_readable($path)) {
            throw new \RuntimeException("Missing data file: {$path}");
        }

        $table = $this->getTableForFile($path);
        return $this->generate_markov_text($length, $table);
    }

    private function generate_code($length)
    {
        $path = dirname(__DIR__, 3) . '/lib/php_code.txt';
        if (!is_readable($path)) {
            throw new \RuntimeException("Missing data file: {$path}");
        }

        $table = $this->getTableForFile($path);
        return $this->generate_markov_text($length, $table, 4);
    }

    private function getTableForFile(string $path, int $look_forward = 4): array
    {
        $cacheKey = md5($path . '|' . $look_forward);
        if (isset($this->tableCache[$cacheKey])) return $this->tableCache[$cacheKey];

        $cacheFile = $this->cacheDir . '/markov_' . $cacheKey . '.ser';
        $srcMTime = filemtime($path) ?: 0;

        if (is_readable($cacheFile)) {
            $data = @unserialize(@file_get_contents($cacheFile));
            if (is_array($data) && isset($data['mtime']) && $data['mtime'] === $srcMTime && isset($data['table'])) {
                $this->tableCache[$cacheKey] = $data['table'];
                return $data['table'];
            }
        }

        $text = file_get_contents($path);
        $table = $this->generate_markov_table($text, $look_forward);

        // write cache best-effort
        try {
            @file_put_contents($cacheFile, serialize(['mtime' => $srcMTime, 'table' => $table]), LOCK_EX);
        } catch (\Throwable $e) {
            // ignore
        }

        $this->tableCache[$cacheKey] = $table;
        return $table;
    }

    private function stubify(string $title): string
    {
        return strtolower(str_replace(' ', '-', $title));
    }

    private function get_title(): string
    {
        $length = random_int(4, 30);
        $title = $this->generate($length * 10);
        $title = str_replace(['.', ',', '!', '?', '"', '—'], '', $title);
        $title = substr($title, 0, strpos($title, ' ', $length));
        return ucwords($title);
    }

        private function get_headline(string $stub): string
        {
            return $this->generate(100) . '...';
    }

    private function getContent(string $stub): string
    {
        $paragraphCount = random_int(5, 16);
        $paragraphs = [];
        for ($i = 0; $i < $paragraphCount; $i++) {
            if ($i === 0) {
                $paragraphs[] = '<p><strong>' . htmlspecialchars($this->generate(100)) . '</strong></p>';
            } elseif ($i % 3 === 2) {
                $paragraphs[] = '<div class="skill-item"><pre><code>' . htmlspecialchars($this->generate_code(500)) . '</code></pre></div>';
            } else {
                $paragraphs[] = '<p>' . htmlspecialchars($this->generate(500)) . '</p>';
            }
        }

        return implode("\n\n", $paragraphs);
    }

    public function getRandomArticles(int $count): array
    {
        $articles = [];
        for ($i = 0; $i < $count; $i++) {
            $title = $this->get_title();
            $articles[] = [
                'title' => $title,
                'stub' => $this->stubify($title),
                'headline' => $this->get_headline($this->stubify($title))
            ];
        }
        return $articles;
    }

    public function getByStub(string $stub): ?array
    {
        return [
            'title' => ucwords(str_replace('-', ' ', $stub)),
            'stub' => $stub,
            'headline' => ucfirst($this->get_headline($stub)),
            'content' => $this->getContent($stub)
        ];
    }
}