<?php
/*************************************************************
# 076 License

Copyright (c) テクニカル諏訪子

Permission is hereby granted to any person obtaining a copy of the software
Little Beast (the "Software") to use, modify, merge, copy, publish, distribute,
sublicense, and/or sell copies of the Software, subject to the following conditions:

    1. **Origin Attribution**:
       - You must not misrepresent the origin of the Software; you must not claim
         you created the original Software.
       - If the Software is used in a product, you must either:
         a. Provide clear attribution in the product's documentation, user interface,
            or other visible areas, **OR**
         b. Pay the original developers a fee they specify in writing.
    2. **Usage Restriction**:
       - The Software, or any derivative works, dependencies, or libraries
         incorporating it, must not be used for censorship or to suppress freedom of
         speech, expression, or creativity. Prohibited uses include, but are not
         limited to:
         - Censorship of so-called "hate speech", visuals, non-mainstream opinions,
           ideas, or objective reality.
         - Tools or systems designed to restrict access to information or
           artistic works.
    3. **Notice Preservation**:
       - This license and the above copyright notice must remain intact in all copies
         of the source code.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED,
INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A
PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF
CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE
OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 */
namespace Std\Lib\Db\Convert;

use Std\Lib\Db\DbException;
use Std\Lib\Db\Sqlite\Blob;

/**
 * LittleBeastのJSONのデータ（data/）から読む（Storesを参照）
 *
 * データのディレクトリが無い・空なら、テーブルも無い。SQLiteのファイル（data/site.sqlite等）は飛ばす。
 */
final class JsonSource implements Source {
  private array $warnings = [];
  private ?array $files = null;
  private ?array $data = null;

  /** @param string $dir  データのディレクトリ（普通はROOT/data） */
  public function __construct(private string $dir) {
    $this->dir = rtrim($dir, '/\\');
  }

  public function warnings(): array {
    return array_values(array_unique($this->warnings));
  }

  /** データのファイル（相対パス => 絶対パス） */
  private function files(): array {
    if ($this->files !== null) return $this->files;
    $out = [];
    if (is_dir($this->dir)) {
      $it = new \RecursiveIteratorIterator(new \RecursiveDirectoryIterator($this->dir, \FilesystemIterator::SKIP_DOTS));
      foreach ($it as $f) {
        if (!$f->isFile()) continue;
        $rel = str_replace('\\', '/', substr($f->getPathname(), strlen($this->dir) + 1));
        $out[$rel] = $f->getPathname();
      }
    }
    ksort($out, SORT_STRING);
    return $this->files = $out;
  }

  /** 全部の行を作る（データは小さいので、メモリーに） */
  private function load(): array {
    if ($this->data !== null) return $this->data;
    $d = array_fill_keys(array_keys(Stores::tables()), []);
    $generic = [];
    foreach ($this->files() as $rel => $abs) {
      foreach (Stores::SKIP as $re) {
        if (preg_match($re, $rel)) {
          if (str_starts_with($rel, 'db/') && str_ends_with($rel, '.json')) $generic[$rel] = $abs;
          continue 2;
        }
      }
      $text = (string)@file_get_contents($abs);
      // SQLiteのデータベース（例：data/site.sqlite）とジャーナル
      if (str_starts_with($text, "SQLite format 3\0") || preg_match('/-(journal|wal|shm)$/', $rel)) {
        $this->warnings[] = "SQLiteのファイルは変換しません：{$rel}";
        continue;
      }
      if (!$this->store($rel, $text, $d)) $d['lb_files'][] = [$rel, new Blob($text)];
    }
    $this->data = ['stores' => $d, 'generic' => $generic];
    return $this->data;
  }

  /**
   * ファイルを決まった形のテーブルにする
   *
   * @return bool  falseならlb_files
   */
  private function store(string $rel, string $text, array &$d): bool {
    $isJson = str_ends_with($rel, '.json');
    $json = $isJson ? json_decode($text, true) : null;
    if ($isJson && json_last_error() !== JSON_ERROR_NONE) return false;
    if (preg_match('#^user/(\d+)\.([A-Za-z0-9]+)\.json$#', $rel, $m)) {
      if (!is_array($json)) return false;
      $role = $json['role'] ?? null;
      $d['lb_users'][] = [(int)$m[1], $m[2], isset($json['email']) && is_string($json['email']) ? $json['email'] : null, is_int($role) ? $role : null, $text];
      return true;
    }
    if ($rel === 'comments/ap.json') {
      if (!is_array($json)) return false;
      foreach ($json as $k => $v) $d['lb_comment_ap'][] = [(string)$k, self::encode($v)];
      return true;
    }
    if (preg_match('#^comments/(.+)\.json$#', $rel, $m) && Stores::safeThread($m[1])) {
      if (!is_array($json) || (isset($json['comments']) && !is_array($json['comments']))) return false;
      $comments = $json['comments'] ?? [];
      unset($json['comments']);
      $d['lb_comment_threads'][] = [$m[1], self::encode((object)$json)];
      $seq = 0;
      foreach ($comments as $id => $c) $d['lb_comments'][] = [$m[1], (string)$id, $seq++, self::encode($c)];
      return true;
    }
    if (preg_match('#^notifications/([^/]+)\.json$#', $rel, $m) && Stores::safeName($m[1])) {
      if (!is_array($json) || !array_is_list($json)) return false;
      foreach ($json as $i => $n) $d['lb_notifications'][] = [$m[1], $i, self::encode($n)];
      return true;
    }
    if ($rel === 'cron/cron.json') {
      if (!is_array($json) || !array_is_list($json)) return false;
      foreach ($json as $i => $j) $d['lb_cron'][] = [$i, self::encode($j)];
      return true;
    }
    if ($rel === 'followers.txt' || $rel === 'following.txt') {
      $t = $rel === 'followers.txt' ? 'lb_followers' : 'lb_following';
      $lines = array_values(array_filter(array_map('trim', explode("\n", $text)), fn($l) => $l !== ''));
      foreach ($lines as $i => $l) $d[$t][] = [$i, $l];
      return true;
    }
    if (preg_match('#^ap/notes/([^/]+)\.json$#', $rel, $m) && Stores::safeName($m[1])) {
      if (!is_array($json)) return false;
      $d['lb_ap_notes'][] = [$m[1], $text];
      return true;
    }
    return false;
  }

  public static function encode(mixed $v): string {
    return (string)json_encode($v, Stores::JSON_FLAGS);
  }

  public function tables(): array {
    $data = $this->load();
    $out = [];
    foreach (Stores::tables() as $name => $t) {
      if (!empty($data['stores'][$name])) $out[] = $t;
    }
    foreach ($data['generic'] as $rel => $abs) {
      $g = $this->generic($abs);
      if ($g === null) {
        $this->warnings[] = "読めないテーブルのファイル：{$rel}";
        continue;
      }
      $out[] = $g['table'];
    }
    return $out;
  }

  /** data/db/{テーブル}.json */
  private function generic(string $abs): ?array {
    static $cache = [];
    if (isset($cache[$abs])) return $cache[$abs];
    $j = json_decode((string)@file_get_contents($abs), true);
    if (!is_array($j) || !isset($j['table'], $j['columns']) || !is_array($j['columns'])) return null;
    $t = ['name' => (string)$j['table'], 'columns' => $j['columns'], 'pk' => $j['pk'] ?? [], 'indexes' => $j['indexes'] ?? []];
    foreach ($t['columns'] as $i => $c) {
      if (isset($c['default']['value']) && is_array($c['default']['value'])) $t['columns'][$i]['default']['value'] = self::decodeValue($c['default']['value']);
    }
    return $cache[$abs] = ['table' => $t, 'rows' => $j['rows'] ?? []];
  }

  /** JSONの値（{"$blob": base64}、{"$float": "NaN"}） */
  public static function decodeValue(mixed $v): mixed {
    if (is_array($v)) {
      if (isset($v['$blob'])) return new Blob((string)base64_decode((string)$v['$blob']));
      if (isset($v['$float'])) return match ($v['$float']) { 'NaN' => NAN, 'Infinity' => INF, default => -INF };
      return json_encode($v, JSON_UNESCAPED_UNICODE | JSON_UNESCAPED_SLASHES);
    }
    return $v;
  }

  public function maxLengths(array $t): array {
    $out = [];
    $this->rows([strtolower($t['name'])], function(string $name, array $rows) use ($t, &$out): void {
      foreach ($t['columns'] as $i => $c) {
        foreach ($rows as $r) {
          $v = $r[$i] ?? null;
          $len = $v instanceof Blob ? strlen($v->data) : (is_string($v) ? mb_strlen($v, 'UTF-8') : 0);
          $out[$c['name']] = max($out[$c['name']] ?? 0, $len);
        }
      }
    });
    return $out;
  }

  public function rows(array $tables, callable $sink): void {
    $data = $this->load();
    foreach ($data['stores'] as $name => $rows) {
      if (!in_array($name, $tables, true) || empty($rows)) continue;
      foreach (array_chunk($rows, 500) as $chunk) $sink($name, $chunk);
    }
    foreach ($data['generic'] as $abs) {
      $g = $this->generic($abs);
      if ($g === null || !in_array(strtolower($g['table']['name']), $tables, true)) continue;
      foreach (array_chunk($g['rows'], 500) as $chunk) {
        $sink($g['table']['name'], array_map(fn($r) => array_map([self::class, 'decodeValue'], (array)$r), $chunk));
      }
    }
  }
}