Files
rank24.pl/.paul/codebase/concerns.md
Jacek Pyziak 0776c4531e Dodano mapę kodu w .paul/codebase/ (7 dokumentów)
Wygenerowano przez równoległą analizę czterech agentów: stack, architektura,
konwencje, integracje, testy, baza danych oraz wykryte problemy i dług techniczny.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-05 19:57:07 +02:00

98 lines
5.0 KiB
Markdown

# Concerns & Technical Debt — rank24.pl
## Security — CRITICAL
### Hardcoded Credentials (must fix before any public exposure)
| Secret | Location | Risk |
|--------|---------|------|
| MySQL password | `config.php` lines 2-5 | Full DB access if repo leaked |
| FTP password | `.vscode/ftp-kr.json`, `.vscode/sftp.json` | Full server access |
| DataForSEO API key | `autoload/class.Cron.php` ~lines 160, 262, 354 | API abuse / billing fraud |
| SMTP password | `autoload/class.S.php` ~lines 293-300 | Email spoofing |
**Remediation**: move all secrets to environment variables or a `.env` file excluded from VCS.
### SQL Injection
- `autoload/class.Cron.php` ~line 200: raw string concatenation in DELETE query
- `autoload/class.GoogleRank.php` lines 74, 96, 100, 136, 158, 162: raw string concat in UPDATE queries
- `autoload/class.DataBase.php` lines 15, 47, 82: mixed OPD with string building
**Remediation**: use Medoo's parameterized methods or PDO `bindValue()` for all dynamic values.
### Other Security Issues (MEDIUM)
- **Weak password hashing**: `md5($pass1)` in `autoload/class.DataBase.php` line 31 — use `password_hash()`
- **No CSRF protection**: state-changing AJAX operations in `ajax.php` lack CSRF tokens
- **Path traversal**: `autoload/class.DataBase.php` ~line 57 — user-supplied `image_folder` concatenated into file path without validation
- **Client-supplied MIME type**: file type validation in `class.DataBase.php` checks `$file['type']` (attacker-controlled)
- **Insecure deserialization**: `@unserialize()` used in `autoload/class.FileCache.php` line 43 and `autoload/opd.statement.php`
- **XSS**: `\S::get()` reads raw `$_POST`/`$_GET` without sanitization; values reach HTML output in multiple templates
## Technical Debt
### God Classes
- `autoload/class.S.php` — 700+ lines; handles sessions, email, DNS, CSV, URL, string utils, DB helpers. Should be split.
- `autoload/class.GoogleRank.php` — 300+ lines; proxy selection logic repeated 4+ times with no extraction.
- `autoload/class.Cron.php` — 400+ lines; hardcoded credentials, multiple large functions.
### Code Duplication
- Proxy selection + backoff UPDATE query repeated verbatim ~4 times in `class.GoogleRank.php`
- Google block-detection strings (`"Our systems have detected unusual traffic"`) duplicated in multiple methods
### Global State Anti-Pattern
Every class does `global $db, $mdb, $user, $config, $cache;` — no DI, no service container. Makes refactoring and testing very difficult.
### Two ORM Layers
Both `$db` (OPD) and `$mdb` (Medoo) are initialized and used. Older code paths use OPD raw queries; newer paths use Medoo. Inconsistent access patterns throughout.
### Deprecated PHP Patterns
- Old-style constructor: `function DataEdit()` in `autoload/class.DataEdit.php` line 32 (should be `__construct()`)
- `global` variable injection instead of constructor parameters
- Short open tags `<?` in templates — requires `short_open_tag = On` in php.ini
### Missing Abstractions
- No HTTP client wrapper — cURL used directly in `class.GoogleScraper.php`, no retry/logging
- No proxy manager class — proxy logic spread across `GoogleRank`, `GoogleSite`, `class.S`
- No logger — debugging done via `file_put_contents('google-rank.txt', ...)` and commented-out `\S::pre()`
- No centralized input validation layer — all `\S::get()` calls are point-of-use
## Reliability Concerns
### Cron Jobs
- No retry logic — single failure = job skipped until next run
- No max execution time enforcement — long foreach loops can time out silently
- `cron.php` outputs JSON directly — no structured logging for cron daemon
- API responses partially unchecked: `class.Cron.php` ~line 162 returns `'ok'` without verifying data integrity
### Scraping / Proxy
- Proxy rotation is deterministic (`ORDER BY used ASC LIMIT 1`) — predictable, easier for Google to detect
- No rate limiting or delays between requests in batch operations
- No exponential backoff — same fixed 10s timeout regardless of failure history (`CURLOPT_TIMEOUT = 10`)
- Block detection only covers known string patterns — new CAPTCHA formats would be missed
- `file_put_contents('google-rank.txt', $result)` in `class.GoogleRank.php` line ~185 — debug log left in production
### Error Handling
- Global `error_reporting` suppresses notices, warnings, deprecations — real errors can be masked
- cURL operations in `class.GoogleScraper.php` have no try/catch — silent failures return `-1`
- File operations in `class.DataBase.php` use `@` suppression — orphaned temp files possible
- No circuit breaker — scraping continues even after repeated proxy failures
## Minor Issues
- Debug variable `$debbbb` left in `autoload/class.S.php` line ~357
- Commented-out dead code in `autoload/class.Cron.php` lines 56-61
- Duplicate key in array: `autoload/class.Cron.php` lines ~364, ~376 — `'se_id'` set twice with different values
- Session-based state won't scale to multi-server deployment (no session locking)
- All proxy requests use `http://` (line ~120 `class.GoogleRank.php`) — credentials sniffable on network