asterocr 1.0.0__py3-none-any.whl

asterocr/kaggle_bridge/colab_auto_runner.py

@@ -0,0 +1,380 @@
+import os
+import sys
+import time
+import json
+from playwright.sync_api import sync_playwright
+
+PROFILE_DIR = os.path.join(os.environ.get("LOCALAPPDATA", ""), "Google", "Chrome", "User Data")
+
+import re
+
+def check_quota_exceeded(page):
+    try:
+        import re
+        # Broadly search for quota limits anywhere in the DOM (including shadow DOMs)
+        loc = page.locator("text=/Cannot connect to (TPU|GPU) backend|usage limits in Colab|Usage limit exceeded/i")
+        if loc.count() > 0 and loc.first.is_visible():
+            print("[!] TAI KHOAN NAY DA HET HAN MUC GPU/TPU (Quota Exceeded)!")
+            print("[*] Chuan bi dong trinh duyet va chuyen sang Account tiep theo...")
+            return True
+    except Exception as e:
+        pass
+    return False
+
+def check_and_clear_sessions(page):
+    try:
+        # Bơm Script JS trực tiếp vào trình duyệt để tìm và click nút xuyên qua mọi tầng mây Shadow DOM!
+        js_clicker = """
+        () => {
+            function findAndClick(text) {
+                let els = document.querySelectorAll('*');
+                for (let el of els) {
+                    if (el.shadowRoot) {
+                        let btn = el.shadowRoot.querySelector('mwc-button, paper-button, button');
+                        if (btn && btn.textContent && btn.textContent.toLowerCase().includes(text.toLowerCase())) {
+                            btn.click();
+                            return true;
+                        }
+                    }
+                    if (el.textContent && el.textContent.toLowerCase() === text.toLowerCase()) {
+                        let btn = el.closest('mwc-button, paper-button, button') || el;
+                        btn.click();
+                        return true;
+                    }
+                }
+                return false;
+            }
+            return findAndClick('Manage sessions');
+        }
+        """
+        
+        # Thử click bằng JS trước
+        clicked_by_js = page.evaluate(js_clicker)
+        
+        # Nếu JS thất bại thì dùng Playwright
+        manage_btn = page.locator("text=Manage sessions")
+        if clicked_by_js or (manage_btn.count() > 0 and manage_btn.first.is_visible()):
+            if not clicked_by_js:
+                manage_btn.first.click(force=True)
+                
+            print("[*] Phat hien loi 'Too many sessions'! Tu dong dọn dẹp...")
+            time.sleep(4)
+            
+            # Nhấn nút "Terminate other sessions"
+            term_other = page.locator("text=Terminate other sessions")
+            if term_other.count() > 0 and term_other.first.is_visible():
+                term_other.first.click(force=True)
+                time.sleep(2)
+            
+            # Click thêm vào tất cả các biểu tượng thùng rác (nếu có)
+            trash_icons = page.locator("[title*='Terminate'], [aria-label*='Terminate'], [icon='delete']")
+            for i in range(trash_icons.count()):
+                try:
+                    if trash_icons.nth(i).is_visible():
+                        trash_icons.nth(i).click(force=True)
+                        time.sleep(1)
+                except:
+                    pass
+            
+            # Đóng hộp thoại
+            close_btn = page.locator("text=Close")
+            if close_btn.count() > 0 and close_btn.first.is_visible():
+                close_btn.first.click(force=True)
+            
+            try:
+                page.keyboard.press("Escape")
+            except:
+                pass
+            time.sleep(2)
+            print("[+] Da don dep xong session cu! Tiep tuc chay code tren account hien tai...")
+            return True
+    except Exception as e:
+        print(f"[DEBUG] Loi khi xoa session: {e}")
+    return False
+
+def main():
+    print("=" * 65)
+    print("  AsterOCR Google Colab Zero-Touch Engine")
+    print("=" * 65)
+
+def run_with_profile(profile_dir, is_tpu, COLAB_NOTEBOOK_CODE):
+    print(f"[*] Su dung Bot Profile tai: {profile_dir}")
+
+    with sync_playwright() as p:
+        browser = p.chromium.launch_persistent_context(
+            user_data_dir=profile_dir,
+            channel="chrome",
+            headless=False,
+            ignore_default_args=["--enable-automation"],
+            args=[
+                '--disable-blink-features=AutomationControlled',
+                '--disable-infobars',
+                '--no-sandbox',
+                '--disable-dev-shm-usage',
+                '--window-size=1920,1080'
+            ],
+            user_agent="Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/121.0.0.0 Safari/537.36",
+            no_viewport=True
+        )
+        
+        page = browser.pages[0] if browser.pages else browser.new_page()
+        
+        # Bơm Script tàng hình (Stealth Evasion) vào mỗi khi load trang
+        page.add_init_script("""
+            Object.defineProperty(navigator, 'webdriver', {
+                get: () => undefined
+            });
+            window.navigator.chrome = {
+                runtime: {}
+            };
+            Object.defineProperty(navigator, 'languages', {
+                get: () => ['en-US', 'en']
+            });
+            Object.defineProperty(navigator, 'plugins', {
+                get: () => [1, 2, 3]
+            });
+        """)
+        
+        print("[*] Truc tiep truy cap vao Google Colab...")
+        page.goto("https://colab.research.google.com/#create=true")
+        
+        # Check login
+        try:
+            page.wait_for_selector('.monaco-editor', timeout=15000)
+            print("[+] Da dang nhap Google Colab (Co Monaco Editor).")
+        except:
+            print("[!] Colab yeu cau dang nhap hoac load qua cham! Vui long tu dang nhap tren cua so trinh duyet.")
+            try:
+                page.wait_for_selector('.monaco-editor', timeout=60000)
+                print("[+] Da dang nhap thanh cong!")
+            except:
+                print("[!] Qua gio dang nhap hoac loi mang. Chuyen sang Profile tiep theo...")
+                return "ERROR"
+
+        
+        print("[*] Cau hinh phan cung (TPU/GPU)...")
+        time.sleep(2)
+        try:
+            # Click Runtime -> Change runtime type (force=True de xuyen qua cac bang popup huong dan cua Google)
+            try:
+                page.locator('#runtime-menu-button').click(force=True, timeout=5000)
+            except:
+                page.get_by_text("Runtime", exact=True).click(force=True, timeout=5000)
+                
+            time.sleep(1)
+            try:
+                page.locator('#ok').click(force=True, timeout=1000) # clear any dialogs
+            except:
+                pass
+                
+            page.get_by_text("Change runtime type").click(force=True, timeout=5000)
+            time.sleep(2)
+            
+            # Select hardware based on new Colab UI
+            if is_tpu:
+                # Try new TPU names first, fallback to old generic TPU
+                try:
+                    page.get_by_text("v5e-1 TPU").first.click(force=True, timeout=3000)
+                except:
+                    page.get_by_text("TPU").first.click(force=True, timeout=3000)
+            else:
+                try:
+                    page.get_by_text("T4 GPU").first.click(force=True, timeout=3000)
+                except:
+                    page.get_by_text("GPU").first.click(force=True, timeout=3000)
+            time.sleep(1)
+            
+            # Click Save button
+            try:
+                page.get_by_role("button", name="Save").click(timeout=3000)
+            except:
+                try:
+                    page.locator('button:has-text("Save")').last.click(force=True, timeout=3000)
+                except:
+                    pass
+            
+            # Ultimate fallback: Press Enter to submit the dialog
+            time.sleep(1)
+            page.keyboard.press("Enter")
+            time.sleep(3)
+            check_and_clear_sessions(page)
+            if check_quota_exceeded(page):
+                return "QUOTA_EXCEEDED"
+            print("[+] Cau hinh phan cung thanh cong!")
+        except Exception as e:
+            print(f"[!] Khong the cau hinh tu dong phan cung: {e}")
+
+        print("[*] Tiem code va bat dau chay...")
+        try:
+            # Click into the first cell
+            page.click('.monaco-editor')
+            time.sleep(1)
+            page.keyboard.press("Control+A")
+            page.keyboard.press("Backspace")
+            time.sleep(1)
+            # Type the code using a synthetic paste event to bypass auto-indent and clipboard locks!
+            page.evaluate('''
+                (code) => {
+                    const dataTransfer = new DataTransfer();
+                    dataTransfer.setData('text/plain', code);
+                    const event = new ClipboardEvent('paste', {
+                        clipboardData: dataTransfer,
+                        bubbles: true,
+                        cancelable: true
+                    });
+                    document.activeElement.dispatchEvent(event);
+                }
+            ''', COLAB_NOTEBOOK_CODE)
+            print("[*] Dang cho Colab cap phat GPU (Co the mat 10-15s)...")
+            time.sleep(2)
+            page.keyboard.press("Control+Enter")
+            
+            # Vòng lặp chờ 15s để bắt lỗi "Too many sessions" hoặc "Quota Exceeded"
+            # Vì Colab mất một lúc để Allocating, nếu check quá nhanh sẽ bị hụt!
+            is_quota_exceeded = False
+            for _ in range(8):
+                time.sleep(2)
+                if check_and_clear_sessions(page):
+                    time.sleep(2)
+                    page.keyboard.press("Control+Enter")
+                if check_quota_exceeded(page):
+                    is_quota_exceeded = True
+                    break
+            
+            if is_quota_exceeded:
+                return "QUOTA_EXCEEDED"
+                
+            print("[+] Da tiem code va vuot qua buoc kiem tra GPU!")
+        except Exception as e:
+            print(f"[!] Loi khi paste code: {e}")
+
+        print("[*] Bat che do Auto-Watcher vinh vien...")
+        while True:
+            try:
+                time.sleep(10)
+                # Kiem tra nut Reconnect
+                reconnect_btn = page.query_selector('colab-connect-button[connected="false"], #connect, colab-connect-button:has-text("Reconnect"), colab-connect-button:has-text("Connect")')
+                if reconnect_btn:
+                    print("[*] Mat ket noi! Tu dong Reconnect...")
+                    reconnect_btn.click()
+                    time.sleep(5)
+                    page.keyboard.press("Control+F9")
+
+                # Dialog "Are you still there" hoac error
+                dialog_ok = page.query_selector('md-dialog mwc-button[dialogaction="ok"], paper-button:has-text("OK")')
+                if dialog_ok:
+                    print("[*] Dong hop thoai Canh bao...")
+                    dialog_ok.click()
+                    time.sleep(3)
+                    page.keyboard.press("Control+F9")
+
+                # Kiem tra trang thai chay cua Notebook (Idle Detection)
+                # Kiem tra xem co cell nao dang "Running" khong (Bang cach tim nut Stop)
+                is_running = page.evaluate('''
+                    () => {
+                        let btns = document.querySelectorAll('colab-run-button');
+                        if (btns.length === 0) return true; // Chua the xac dinh, gia su dang chay de an toan
+                        for (let b of btns) {
+                            if (b.title && b.title.toLowerCase().includes('stop')) return true;
+                            if (b.getAttribute('aria-label') && b.getAttribute('aria-label').toLowerCase().includes('stop')) return true;
+                            let icon = b.querySelector('iron-icon');
+                            if (icon && icon.getAttribute('icon') === 'colab:stop') return true;
+                        }
+                        return false; // Khong co cell nao dang chay
+                    }
+                ''')
+                
+                if not is_running:
+                    print("[!] Phat hien Notebook bi DUNG chay (Idle/Crashed). Tu dong kich hoat chay lai...")
+                    page.keyboard.press("Control+Enter")
+                    time.sleep(5)
+
+                # Dialog "Too many sessions"
+                if check_and_clear_sessions(page):
+                    page.keyboard.press("Control+Enter")
+
+                # Kiem tra Quota Limit Exceeded
+                if check_quota_exceeded(page):
+                    return "QUOTA_EXCEEDED"
+
+                # Keep-alive
+                page.evaluate('() => { if (window.colab && window.colab.global) { window.colab.global.notebook.kernel.keepAlive(); } }')
+            except Exception:
+                time.sleep(10)
+
+def main():
+    print("=" * 65)
+    print("  AsterOCR Google Colab Zero-Touch Engine")
+    print("=" * 65)
+    tunnel_url = sys.argv[1] if len(sys.argv) > 1 else "https://asterai.sglocalhub.com"
+    print(f"[*] Engine URL: {tunnel_url}")
+    
+    # Read config to know if TPU or GPU
+    is_tpu = False
+    config_path = os.path.join(os.path.dirname(__file__), "..", "..", "scratch", "colab_run_config.json")
+    if os.path.exists(config_path):
+        import json
+        with open(config_path, "r", encoding="utf-8") as f:
+            cfg = json.load(f)
+            if cfg.get("device") == "colab_tpu":
+                is_tpu = True
+
+    COLAB_NOTEBOOK_CODE = f'''# --- ASTEROCR COLAB WORKER (ZERO-TOUCH) ---
+import requests, os, time, subprocess, sys
+server_url = "{tunnel_url}"
+
+while True:
+    try:
+        print(f"[+] Xin code tu {{server_url}}...")
+        res = requests.get(f"{{server_url}}/api/colab/worker.py", timeout=20)
+        if res.status_code == 200 and not res.text.startswith("<!DOCTYPE"):
+            with open("worker.py", "w", encoding="utf-8") as f:
+                f.write(res.text)
+            print("[🚀] Bat dau Huan luyen...")
+            p = subprocess.Popen([sys.executable, "-u", "worker.py"], stdout=subprocess.PIPE, stderr=subprocess.STDOUT, text=True, bufsize=1)
+            for line in p.stdout:
+                print(line, end='', flush=True)
+            p.wait()
+            print(f"[!] Tien trinh ket thuc voi ma loi: {{p.returncode}}")
+        else:
+            print(f"[ERROR] Loi tai code. Status: {{res.status_code}}")
+            if res.text.startswith("<!DOCTYPE"):
+                print("[ERROR] Nhan duoc HTML thay vi Code Python! Cloudflare Tunnel dang bi chan hoac chua san sang.")
+        print("[!] Phien ket thuc. Tu dong khoi dong lai sau 5s...")
+        time.sleep(5)
+    except Exception as e:
+        print(f"[!] Ket noi gian doan: {{e}}. Thu lai sau 10s...")
+        time.sleep(10)
+'''
+
+    profiles = [
+        os.path.join(os.environ.get("LOCALAPPDATA", ""), "Google", "Chrome", "AsterOCR_Profile"),
+        os.path.join(os.environ.get("LOCALAPPDATA", ""), "Google", "Chrome", "AsterOCR_Profile_2"),
+        os.path.join(os.environ.get("LOCALAPPDATA", ""), "Google", "Chrome", "AsterOCR_Profile_3")
+    ]
+    
+    print("[*] Bat dau chu trinh Multi-Profile Fallback...")
+    
+    for i, profile_dir in enumerate(profiles, start=1):
+        print(f"\\n{'='*50}")
+        print(f"[*] DANG SU DUNG TAI KHOAN {i}/3")
+        print(f"{'='*50}")
+        
+        status = run_with_profile(profile_dir, is_tpu, COLAB_NOTEBOOK_CODE)
+        
+        if status == "QUOTA_EXCEEDED":
+            print(f"[!] Tai khoan {i} da het Quota. Tu dong chuyen sang tai khoan tiep theo...")
+            time.sleep(5)
+            continue
+        elif status == "ERROR":
+            print(f"[!] Tai khoan {i} gap loi ket noi. Thu tai khoan tiep theo...")
+            time.sleep(5)
+            continue
+        else:
+            print("[+] Tien trinh dung dot ngot hoac bi huy.")
+            break
+            
+    print("[!] TAT CA TAI KHOAN DEU DA HET QUOTA HOAC GAP LOI!")
+
+if __name__ == "__main__":
+    main()