@iducky/media-agent 1.1.0

package/src/media_agent/platforms/douyin/douyin_full_login.py

@@ -0,0 +1,391 @@
+"""Legacy operation, executed only through main()."""
+
+def main():
+    #!/usr/bin/env python3
+    """
+    douyin_creator_personal_test: 完整登录与登录态持久化流程
+    """
+    from media_agent.runtime.paths import runtime_home
+    import json, time, sys, os, signal
+    from pathlib import Path
+    from datetime import datetime, timezone
+    from media_agent.runtime.browser import launch_persistent_context
+
+    BASE = runtime_home()
+    ALIAS = 'douyin_creator_personal_test'
+    PROFILE_DIR = BASE / 'profiles' / ALIAS
+    BROWSER_DATA = str(PROFILE_DIR / 'browser_data')
+    LOCK_FILE = BASE / 'locks' / f'{ALIAS}.lock'
+    QR_SCREENSHOT = str(BASE / 'screenshots' / f'{ALIAS}_qr.png')
+    FINAL_SCREENSHOT = str(BASE / 'screenshots' / f'{ALIAS}_login_result.png')
+    PERSISTENCE_SCREENSHOT = str(BASE / 'screenshots' / f'{ALIAS}_persistence.png')
+    TASK_ID = f'full_login_{ALIAS}_{datetime.now().strftime("%Y%m%d_%H%M%S")}'
+
+    # ========== Report accumulator ==========
+    R = {'stage': 'init', 'errors': []}
+    def report(k, v):
+        R[k] = v
+
+    # ========== Stage 1: Lock & launch ==========
+    report('A_profile_path', str(PROFILE_DIR))
+    report('A_browser_data', BROWSER_DATA)
+
+    LOCK_FILE.parent.mkdir(parents=True, exist_ok=True)
+    if LOCK_FILE.exists():
+        # Check if it's a zombie lock (no matching process)
+        try:
+            lock_data = json.loads(LOCK_FILE.read_text())
+            pid = lock_data.get('pid')
+            if pid:
+                try:
+                    os.kill(pid, 0)  # Signal 0 = existence check
+                    print(f'ERROR|Lock active: pid={pid}', flush=True)
+                    sys.exit(1)
+                except OSError:
+                    print(f'WARN|Zombie lock pid={pid}, cleaning', flush=True)
+                    LOCK_FILE.unlink()
+        except:
+            LOCK_FILE.unlink()
+
+    LOCK_FILE.write_text(json.dumps(
+        {'task_id': TASK_ID, 'locked_at': datetime.now(timezone.utc).isoformat(),
+         'pid': os.getpid(), 'manual_control': True}, indent=2))
+    print(f'LOCK|{TASK_ID}', flush=True)
+    report('I_lock_acquired', True)
+
+    config = json.loads((PROFILE_DIR / 'config.json').read_text())
+    fp = config['fingerprint']
+    context = launch_persistent_context(
+        user_data_dir=BROWSER_DATA,
+        headless=False, stealth_args=True, viewport=fp['viewport'],
+        locale=fp['locale'], timezone=fp['timezone'], humanize=False, geoip=False,
+    )
+    print('LAUNCH|Browser started', flush=True)
+    page = context.pages[0] if context.pages else context.new_page()
+    page.goto('https://creator.douyin.com/', wait_until='domcontentloaded', timeout=30000)
+    time.sleep(5)
+    report('B_launch_url', page.url)
+
+    # ========== Check if already logged in ==========
+    ch = page.content().lower()
+    already_in = '扫码登录' not in ch and '登录/注册' not in ch
+    if already_in:
+        print('ALREADY_LOGGED_IN|直接进入登录成功检查', flush=True)
+        report('B_first_scan', 'skipped_already_logged_in')
+        # Jump to login check
+        logged_in = True
+        # (goto login check section)
+    else:
+        report('B_first_scan', 'qr_sent')
+
+        # ========== Stage 2: QR code ==========
+        for i in range(30):
+            try:
+                for div in page.query_selector_all('div'):
+                    box = div.bounding_box()
+                    if box and abs(box['width'] - 250) < 10 and abs(box['height'] - 250) < 10:
+                        div.screenshot(path=QR_SCREENSHOT)
+                        if Path(QR_SCREENSHOT).exists() and Path(QR_SCREENSHOT).stat().st_size > 2000:
+                            print(f'QR_OK|{QR_SCREENSHOT}', flush=True); break
+            except: pass
+            if Path(QR_SCREENSHOT).exists() and Path(QR_SCREENSHOT).stat().st_size > 2000: break
+            time.sleep(1)
+        print('QR_READY|请扫码，完成后回复：首次扫码完成', flush=True)
+        sys.stdout.flush()
+
+        # ========== Stage 3: Wait for scan ==========
+        SIG = BASE / 'locks' / f'{ALIAS}.continue'
+        print('WAIT_SCAN', flush=True)
+        for i in range(900):
+            time.sleep(1)
+            if SIG.exists():
+                SIG.unlink(missing_ok=True)
+                print('SCAN_DONE', flush=True); break
+        report('C_verify_triggered', False)
+
+        # ========== Stage 4: Identity verification popup ==========
+        print('WAIT_VERIFY', flush=True)
+        verify_found = False
+        for i in range(30):
+            time.sleep(1)
+            try:
+                ch = page.content().lower()
+                if '身份验证' in ch:
+                    print('VERIFY_POPUP|身份验证弹窗已出现', flush=True)
+                    report('C_verify_triggered', True)
+                    verify_found = True
+                    break
+            except: pass
+
+        if verify_found:
+            # Click "接收短信验证码" using text locator
+            try:
+                opt = page.locator('text=接收短信验证码').first
+                if opt and opt.is_visible():
+                    opt.click()
+                    time.sleep(2)
+                    print('SMS_CLICKED|点击接收短信验证码', flush=True)
+                    report('D_verify_method', 'sms_code')
+                else:
+                    # Fallback: JS click
+                    page.evaluate('''() => {
+                        const els = document.querySelectorAll('div, span, button');
+                        for (const el of els) {
+                            if (el.textContent.includes('接收短信验证码')) {
+                                el.click(); return;
+                            }
+                        }
+                    }''')
+                    time.sleep(2)
+                    print('SMS_CLICKED_JS', flush=True)
+                    report('D_verify_method', 'sms_code_js')
+            except Exception as e:
+                print(f'SMS_CLICK_ERR|{e}', flush=True)
+                report('D_verify_method', f'failed: {e}')
+
+            # Wait for SMS input field
+            print('WAIT_SMS_INPUT', flush=True)
+            for i in range(30):
+                time.sleep(1)
+                try:
+                    ch = page.content().lower()
+                    if '验证码' in ch and '请输入' in ch:
+                        print('SMS_INPUT_READY', flush=True)
+                        break
+                except: pass
+
+            # Confirm SMS sent to phone number
+            print('SMS_PROMPT|验证码输入框已准备好，请发送本次短信验证码。', flush=True)
+            sys.stdout.flush()
+
+            # ========== Stage 5: Wait for SMS code ==========
+            SIG_CODE = BASE / 'locks' / f'{ALIAS}.code_done'
+            print('WAIT_CODE', flush=True)
+            code_received = False
+            for i in range(900):
+                time.sleep(1)
+                if SIG_CODE.exists():
+                    SIG_CODE.unlink(missing_ok=True)
+                    code_received = True
+                    print('CODE_DONE', flush=True)
+                    break
+                # Also check if page changed (auto-login)
+                try:
+                    ch = page.content().lower()
+                    if '扫码登录' not in ch and '登录/注册' not in ch:
+                        print('AUTO_LOGIN|页面已跳转', flush=True)
+                        break
+                except: pass
+
+            if code_received:
+                # The code is stored in a temp file (not in logs/state)
+                code_file = BASE / 'locks' / f'{ALIAS}.code_value'
+                if code_file.exists():
+                    code = code_file.read_text().strip()
+                    code_file.unlink()
+                    print(f'CODE_READ|长度={len(code)}', flush=True)
+
+                    # Fill code using the correct modal locator
+                    try:
+                        # Locate the second_verify_mask modal
+                        modal = page.locator('div[class*="second_verify_mask"]')
+                        modal_count = modal.count()
+                        code_input = modal.locator('input#button-input[name="button-input"][maxlength="6"]')
+                        ci_count = code_input.count()
+                        ci_visible = code_input.is_visible()
+                        ci_enabled = code_input.is_enabled()
+                        print(f'MODAL|count={modal_count}|input_count={ci_count}|visible={ci_visible}|enabled={ci_enabled}', flush=True)
+
+                        if ci_count > 0 and ci_visible:
+                            code_input.scroll_into_view_if_needed()
+                            time.sleep(0.3)
+                            code_input.click()
+                            time.sleep(0.3)
+                            code_input.fill(code)
+                            time.sleep(0.5)
+                            val = code_input.input_value()
+                            if len(val) == 6:
+                                print(f'CODE_FILLED|ok', flush=True)
+                                report('E_code_submit', 'filled_ok')
+                            else:
+                                print(f'CODE_FILLED|len={len(val)}', flush=True)
+                                report('E_code_submit', f'filled_short_{len(val)}')
+                        else:
+                            # Fallback: use direct selector
+                            print(f'MODAL_LOOKUP_FALLBACK', flush=True)
+                            code_input2 = page.locator('input#button-input[type="tel"][maxlength="6"]')
+                            ci2_count = code_input2.count()
+                            print(f'FALLBACK|count={ci2_count}', flush=True)
+                            if ci2_count > 0:
+                                code_input2.first.scroll_into_view_if_needed()
+                                time.sleep(0.3)
+                                code_input2.first.click()
+                                time.sleep(0.3)
+                                code_input2.first.fill(code)
+                                time.sleep(0.5)
+                                val = code_input2.first.input_value()
+                                if len(val) == 6:
+                                    print(f'CODE_FILLED|ok', flush=True)
+                                    report('E_code_submit', 'filled_ok')
+                                else:
+                                    print(f'CODE_FILLED|len={len(val)}', flush=True)
+                                    report('E_code_submit', f'filled_short_{len(val)}')
+                            else:
+                                print('CODE_INPUT_NOT_FOUND', flush=True)
+                                report('E_code_submit', 'input_not_found')
+                    except Exception as e:
+                        print(f'CODE_FILL_ERR|{e}', flush=True)
+                        report('E_code_submit', f'error: {e}')
+
+                    # Click 验证 button using JS within the same modal
+                    try:
+                        result = page.evaluate('''() => {
+                            const modal = document.querySelector('div[class*="second_verify_mask"]');
+                            if (!modal) return 'modal_not_found';
+                            const els = modal.querySelectorAll('div, span, button, a');
+                            for (const el of els) {
+                                const t = el.textContent.trim();
+                                if (t === '验证' && el.offsetParent !== null) {
+                                    el.click(); return 'clicked';
+                                }
+                            }
+                            return 'not_found';
+                        }''')
+                        time.sleep(1)
+                        print(f'VERIFY_CLICKED|{result}', flush=True)
+                        report('E_verify_clicked', result)
+                    except Exception as e:
+                        print(f'VERIFY_ERR|{e}', flush=True)
+                        report('E_verify_clicked', f'error: {e}')
+
+        # ========== Stage 6: Login check (120s) ==========
+        print('POST_VERIFY|等待登录...', flush=True)
+        logged_in = False
+        for i in range(60):
+            time.sleep(2)
+            try:
+                cookies = context.cookies()
+                has_sid = any(c['name'] == 'sessionid' for c in cookies)
+                has_sid_ss = any(c['name'] == 'sessionid_ss' for c in cookies)
+                has_sid_g = any(c['name'] == 'sid_guard' for c in cookies)
+                ch = page.content().lower()
+                still_login = '扫码登录' in ch or '登录/注册' in ch
+                verify_open = '身份验证' in ch
+                print(f'POST|{i*2}s|url={page.url}|login_form={still_login}|verify={verify_open}|sid={has_sid}|sid_ss={has_sid_ss}|sid_g={has_sid_g}', flush=True)
+                if has_sid and has_sid_ss and not still_login:
+                    logged_in = True
+                    print(f'LOGIN_SUCCESS|sessionid出现在{i*2}s', flush=True)
+                    break
+                # Check for console elements
+                if not still_login and not verify_open:
+                    logged_in = True
+                    print(f'LOGIN_SUCCESS|页面已进入控制台 at {i*2}s', flush=True)
+                    break
+            except: pass
+
+        report('F_login_url', page.url)
+        try:
+            ch = page.content().lower()
+            report('F_login_form_still', '扫码登录' in ch or '登录/注册' in ch)
+            report('F_verify_still', '身份验证' in ch)
+        except: pass
+
+        cookies = context.cookies()
+        report('G_cookie_sessionid', any(c['name'] == 'sessionid' for c in cookies))
+        report('G_cookie_sessionid_ss', any(c['name'] == 'sessionid_ss' for c in cookies))
+        report('G_cookie_sid_guard', any(c['name'] == 'sid_guard' for c in cookies))
+        report('G_cookie_uid_tt', any(c['name'] == 'uid_tt' for c in cookies))
+        for ck in cookies:
+            if ck['name'] in ['sessionid', 'sessionid_ss', 'sid_guard', 'uid_tt', 'has_biz_token', 'passport_csrf_token']:
+                print(f'CK|name={ck["name"]}|domain={ck["domain"]}|httponly={ck.get("httpOnly")}|expires={ck.get("expires","session")}', flush=True)
+
+        # Save state
+        state = json.loads((PROFILE_DIR / 'state.json').read_text())
+        state['is_logged_in'] = logged_in
+        state['session_start_at'] = datetime.now(timezone.utc).isoformat() if logged_in else None
+        state['last_login_check'] = datetime.now(timezone.utc).isoformat()
+        state['last_activity'] = datetime.now(timezone.utc).isoformat()
+        (PROFILE_DIR / 'state.json').write_text(json.dumps(state, indent=2, ensure_ascii=False))
+
+        # Screenshot
+        try:
+            page.screenshot(path=FINAL_SCREENSHOT, full_page=True)
+            print(f'SCREENSHOT|{FINAL_SCREENSHOT}', flush=True)
+            report('K_screenshot', FINAL_SCREENSHOT)
+        except: pass
+
+        report('F_logged_in', logged_in)
+
+        # ========== Stage 7: Graceful close ==========
+        if logged_in:
+            print('LOGIN_OK|正常关闭浏览器保存数据...', flush=True)
+            try:
+                context.close()
+                print('CLOSED|浏览器已正常关闭', flush=True)
+                report('H_browser_closed', True)
+            except Exception as e:
+                print(f'CLOSE_ERR|{e}', flush=True)
+                report('H_browser_closed', f'error: {e}')
+            time.sleep(10)
+            print('SLEEP_DONE|等待10秒落盘完成', flush=True)
+
+            # ========== Stage 8: Re-launch for persistence ==========
+            print('VERIFY_PERSISTENCE|重新打开验证免登录...', flush=True)
+            LOCK_FILE.unlink(missing_ok=True)
+            # Re-acquire lock
+            TASK_ID2 = f'persist_{ALIAS}_{datetime.now().strftime("%Y%m%d_%H%M%S")}'
+            LOCK_FILE.write_text(json.dumps(
+                {'task_id': TASK_ID2, 'locked_at': datetime.now(timezone.utc).isoformat(),
+                 'pid': os.getpid(), 'manual_control': True}, indent=2))
+            print(f'LOCK2|{TASK_ID2}', flush=True)
+
+            context2 = launch_persistent_context(
+                user_data_dir=BROWSER_DATA,
+                headless=False, stealth_args=True, viewport=fp['viewport'],
+                locale=fp['locale'], timezone=fp['timezone'], humanize=False, geoip=False,
+            )
+            page2 = context2.pages[0] if context2.pages else context2.new_page()
+            page2.goto('https://creator.douyin.com/', wait_until='domcontentloaded', timeout=30000)
+            time.sleep(5)
+            print(f'VERIFY_URL|{page2.url}', flush=True)
+            report('J_reopen_url', page2.url)
+            ch2 = page2.content().lower()
+            persisted = '扫码登录' not in ch2 and '登录/注册' not in ch2
+            print(f'PERSISTED|{persisted}', flush=True)
+            report('J_persisted', persisted)
+            cookies2 = context2.cookies()
+            report('J_reopen_has_sessionid', any(c['name'] == 'sessionid' for c in cookies2))
+            try:
+                page2.screenshot(path=PERSISTENCE_SCREENSHOT, full_page=True)
+                print(f'VERIFY_SS|{PERSISTENCE_SCREENSHOT}', flush=True)
+                report('K_persistence_screenshot', PERSISTENCE_SCREENSHOT)
+            except: pass
+            context2.close()
+            print('VERIFY_DONE', flush=True)
+            LOCK_FILE.unlink(missing_ok=True)
+            report('I_lock_released', True)
+        else:
+            print('LOGIN_FAILED|保持浏览器打开30秒后关闭...', flush=True)
+            report('H_browser_closed', False)
+            for i in range(30):
+                time.sleep(1)
+                if (BASE / 'locks' / f'{ALIAS}.close').exists():
+                    break
+            try:
+                context.close()
+                print('CLOSED', flush=True)
+            except: pass
+            LOCK_FILE.unlink(missing_ok=True)
+            report('I_lock_released', True)
+
+    # ========== Final report ==========
+    print('='*60, flush=True)
+    print('FINAL_REPORT', flush=True)
+    print('='*60, flush=True)
+    for k, v in sorted(R.items()):
+        print(f'  {k}: {v}', flush=True)
+    print('='*60, flush=True)
+    print('DONE', flush=True)
+
+if __name__ == '__main__':
+    main()