@iducky/media-agent 1.1.0

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

@@ -0,0 +1,128 @@
+"""Legacy operation, executed only through main()."""
+
+def main():
+    #!/usr/bin/env python3
+    """Enterprise login flow. Usage: python3 enterprise_login.py <profile_id> [--site creator|leads]"""
+    from media_agent.runtime.paths import runtime_home
+    import json, time, sys, os
+    from pathlib import Path
+    from datetime import datetime, timezone, timedelta
+    from media_agent.runtime.browser import launch_persistent_context
+
+    BASE = runtime_home()
+    ALIAS = sys.argv[1] if len(sys.argv) > 1 else sys.exit(1)
+    SITE = 'leads'  # default
+    for i, a in enumerate(sys.argv):
+        if a == '--site' and i + 1 < len(sys.argv):
+            SITE = sys.argv[i + 1]; break
+
+    PROFILE_DIR = BASE / 'profiles' / ALIAS
+    BROWSER_DATA = str(PROFILE_DIR / 'browser_data')
+    LOCK_FILE = BASE / 'locks' / f'{ALIAS}.lock'
+    SCREENSHOTS_DIR = BASE / 'screenshots'
+    SCREENSHOTS_DIR.mkdir(parents=True, exist_ok=True)
+
+    # ===== Lock =====
+    LOCK_FILE.parent.mkdir(parents=True, exist_ok=True)
+    if LOCK_FILE.exists():
+        try:
+            d = json.loads(LOCK_FILE.read_text()); pid = d.get('pid')
+            if pid:
+                try: os.kill(pid, 0); print(f'STATUS|PROFILE_LOCKED|pid={pid}', flush=True); sys.exit(3)
+                except OSError: LOCK_FILE.unlink()
+        except: LOCK_FILE.unlink()
+
+    TASK_ID = f'login_{SITE}_{ALIAS}_{datetime.now().strftime("%Y%m%d_%H%M%S")}'
+    LOCK_FILE.write_text(json.dumps({'task_id': TASK_ID, 'locked_at': datetime.now(timezone.utc).isoformat(), 'pid': os.getpid(), 'site': SITE}, indent=2))
+    print(f'LOCK|acquired|{TASK_ID}', flush=True)
+
+    # ===== Launch =====
+    cfg = json.loads((PROFILE_DIR / 'config.json').read_text())
+    sites = cfg.get('sites', {})
+    site_cfg = sites.get(SITE, {})
+    target_url = site_cfg.get('target_url', cfg['account'].get('target_url', 'https://leads.cluerich.com/pc/auth/login'))
+
+    ctx = launch_persistent_context(
+        user_data_dir=BROWSER_DATA, headless=False, stealth_args=True,
+        viewport=cfg['fingerprint']['viewport'],
+        locale=cfg['fingerprint']['locale'], timezone=cfg['fingerprint']['timezone'],
+        humanize=cfg.get('humanize', True), geoip=False,
+    )
+    print('LAUNCH|browser started', flush=True)
+    page = ctx.pages[0] if ctx.pages else ctx.new_page()
+
+    # ===== Navigate =====
+    page.goto(target_url, wait_until='load', timeout=60000)
+    time.sleep(5)
+    print(f'URL|{page.url}', flush=True)
+    print(f'SITE|{SITE}', flush=True)
+    print(f'TITLE|{page.title()}', flush=True)
+
+    # ===== Screenshot =====
+    ss = str(SCREENSHOTS_DIR / f'{ALIAS}_{SITE}_login_page.png')
+    page.screenshot(path=ss, full_page=True)
+    print(f'SCREENSHOT|{ss}', flush=True)
+
+    # ===== Check for QR code =====
+    qr_path = str(SCREENSHOTS_DIR / f'{ALIAS}_{SITE}_qr.png')
+    qr_found = False
+    for i in range(10):
+        try:
+            for div in page.query_selector_all('div'):
+                box = div.bounding_box()
+                if box and 150 < box['width'] < 350 and 150 < box['height'] < 350:
+                    div.screenshot(path=qr_path)
+                    if Path(qr_path).exists() and Path(qr_path).stat().st_size > 2000:
+                        print(f'QR_OK|{qr_path}|size={Path(qr_path).stat().st_size}', flush=True)
+                        qr_found = True; break
+        except: pass
+        if qr_found: break
+        time.sleep(1)
+
+    if not qr_found:
+        page.screenshot(path=qr_path, full_page=True)
+        print(f'QR_FALLBACK|{qr_path}', flush=True)
+
+    # ===== Output state =====
+    print(f'STATE|WAIT_QR_SCAN', flush=True)
+    print(f'SUMMARY|task_id={TASK_ID}|profile_id={ALIAS}|site={SITE}|state=WAIT_QR_SCAN|url={page.url}|browser=open|lock=locked|next_action=wait_user_scan', flush=True)
+
+    # ===== Wait for user scan =====
+    SIG = BASE / 'locks' / f'{ALIAS}.continue'
+    print(f'WAIT|等待用户扫码...', flush=True)
+    for i in range(600):
+        time.sleep(2)
+        if SIG.exists():
+            SIG.unlink(missing_ok=True); print(f'SCAN_DONE', flush=True); break
+        if i % 15 == 0: print(f'WAITING|{i*2}s', flush=True)
+
+    # ===== After scan =====
+    time.sleep(5)
+    print(f'URL_AFTER_SCAN|{page.url}', flush=True)
+    print(f'TITLE|{page.title()}', flush=True)
+
+    body = page.content().lower()
+    if '运营工作台' in body or '数据分析' in body or '线索管理' in body or 'creator-micro' in page.url:
+        print('LOGIN_SUCCESS|进入后台', flush=True)
+    else:
+        print('LOGIN_CHECK|需进一步验证', flush=True)
+
+    ss2 = str(SCREENSHOTS_DIR / f'{ALIAS}_{SITE}_post_scan.png')
+    page.screenshot(path=ss2, full_page=True)
+    print(f'SCREENSHOT|{ss2}', flush=True)
+
+    # Keep open for manual inspection
+    print('KA|保持打开，等待.close信号', flush=True)
+    for i in range(600):
+        time.sleep(2)
+        if (BASE / 'locks' / f'{ALIAS}.close').exists(): break
+
+    # Close
+    ctx.close()
+    print('CLOSED', flush=True)
+    LOCK_FILE.unlink(missing_ok=True)
+    print('LOCK_RELEASED', flush=True)
+    print('DONE', flush=True)
+
+if __name__ == '__main__':
+    main()

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

@@ -0,0 +1,70 @@
+"""Legacy operation, executed only through main()."""
+
+def main():
+    #!/usr/bin/env python3
+    """Export short video detail data. Usage: python3 export_short_video.py <profile_id> <date>"""
+    from media_agent.runtime.paths import runtime_home
+    import json, time, os, sys, re, shutil
+    from pathlib import Path
+    from datetime import datetime, timezone, timedelta
+    from media_agent.runtime.browser import launch_persistent_context
+
+    BASE = runtime_home()
+    ALIAS = sys.argv[1] if len(sys.argv) > 1 else sys.exit(1)
+    TARGET_DATE = sys.argv[2] if len(sys.argv) > 2 else (datetime.now() - timedelta(1)).strftime('%Y-%m-%d')
+
+    PROFILE_DIR = BASE / 'profiles' / ALIAS
+    BROWSER_DATA = str(PROFILE_DIR / 'browser_data')
+    LOCK_FILE = BASE / 'locks' / f'{ALIAS}.lock'
+
+    LOCK_FILE.parent.mkdir(parents=True, exist_ok=True)
+    if LOCK_FILE.exists():
+        try:
+            os.kill(json.loads(LOCK_FILE.read_text()).get('pid', 0), 0)
+            print('PROFILE_LOCKED'); sys.exit(3)
+        except:
+            LOCK_FILE.unlink()
+    LOCK_FILE.write_text(json.dumps({'task_id': 'export-short-video', 'pid': os.getpid()}, indent=2))
+
+    cfg = json.loads((PROFILE_DIR / 'config.json').read_text())
+    ctx = launch_persistent_context(
+        user_data_dir=BROWSER_DATA, headless=False, stealth_args=True,
+        viewport=cfg['fingerprint']['viewport'],
+        locale=cfg['fingerprint']['locale'],
+        timezone=cfg['fingerprint']['timezone'],
+        humanize=False, geoip=False
+    )
+    page = ctx.pages[0] if ctx.pages else ctx.new_page()
+    page.goto('https://leads.cluerich.com/pc/analysis/short-video/list', wait_until='load', timeout=60000)
+    time.sleep(8)
+
+    body = page.content().lower()
+    if not any(k in body for k in ['运营工作台', '数据分析']):
+        print('LOGIN_REQUIRED'); ctx.close(); LOCK_FILE.unlink(); sys.exit(2)
+    print('LOGIN_OK', flush=True)
+
+    ts = datetime.now().strftime('%Y%m%d_%H%M%S')
+    OUT_DIR = BASE / 'tasks' / 'douyin_enterprise_short_video' / ts
+    OUT_DIR.mkdir(parents=True, exist_ok=True)
+
+    page.evaluate('window.scrollTo(0, 1600)')
+    time.sleep(2)
+
+    export_btn = page.locator('button:has-text("导出数据")').first
+    if export_btn.count() > 0:
+        with page.expect_download(timeout=60000) as dl_info:
+            export_btn.click()
+        download = dl_info.value
+        fname = download.suggested_filename
+        fpath = OUT_DIR / fname
+        download.save_as(str(fpath))
+        print(f'DOWNLOADED|{fpath}|{fpath.stat().st_size} bytes', flush=True)
+    else:
+        print('EXPORT_BUTTON_NOT_FOUND', flush=True)
+
+    ctx.close()
+    LOCK_FILE.unlink(missing_ok=True)
+    print('DONE', flush=True)
+
+if __name__ == '__main__':
+    main()