跳到內容

螢幕錄製測試

此測試透過將有頭瀏覽器工作階段錄製為 MP4 檔案,驗證完整的 VNC/瀏覽器堆疊。它證明了:

  • Xvfb 正在運行並接受 :99 上的連線
  • Chromium 可以在虛擬顯示器上以有頭模式啟動
  • ffmpeg 可以透過 x11grab 擷取顯示畫面
  • 產生的 MP4 是有效且可播放的影片
  1. 輪詢 xdpyinfo -display :99 直到 X 顯示器準備就緒(與 entrypoint 相同的模式)
  2. 啟動 ffmpeg 以 10 fps 錄製顯示畫面
  3. 透過 Playwright 的 sync_api 啟動有頭模式 Chromium
  4. 導航至 https://example.com(暫停 3 秒),然後導航至 https://httpbin.org(暫停 3 秒)
  5. 向 ffmpeg 發送 SIGTERM 以完成 MP4 的最終化處理
  6. 回報輸出路徑和檔案大小

將以下內容儲存為容器內的 vnc-browser-test.py,或從此儲存庫的 docs/tests/ 目錄複製:

#!/usr/bin/env python3
"""VNC/browser stack integration test.
Proves that Xvfb, headed Chromium (via Playwright), and ffmpeg screen
recording all work together inside the devcontainer by recording a
browser navigating to two websites and producing an MP4 video.
Usage: python3 vnc-browser-test.py [output_path]
Default output: /tmp/recording.mp4
"""
import os
import signal
import subprocess
import sys
import time
DISPLAY = os.environ.get("DISPLAY", ":99")
DEFAULT_OUTPUT = "/tmp/recording.mp4"
RESOLUTION = "1280x1024"
FRAMERATE = "10"
XVFB_TIMEOUT = 50 # iterations (~5 s at 0.1 s each)
SITES = [
("https://example.com", 3),
("https://httpbin.org", 3),
]
def wait_for_xvfb():
"""Poll xdpyinfo until the X display is ready (mirrors entrypoint.sh)."""
print(f"Waiting for Xvfb on {DISPLAY} ...")
for i in range(XVFB_TIMEOUT):
result = subprocess.run(
["xdpyinfo", "-display", DISPLAY],
stdout=subprocess.DEVNULL,
stderr=subprocess.DEVNULL,
)
if result.returncode == 0:
print(f" Display ready after {i * 0.1:.1f}s")
return
time.sleep(0.1)
sys.exit(f"ERROR: Xvfb display {DISPLAY} not ready after {XVFB_TIMEOUT * 0.1:.1f}s")
def start_recording(output_path):
"""Start ffmpeg x11grab recording and return the process."""
cmd = [
"ffmpeg",
"-y",
"-f", "x11grab",
"-video_size", RESOLUTION,
"-framerate", FRAMERATE,
"-i", f"{DISPLAY}.0",
"-c:v", "libx264",
"-preset", "ultrafast",
"-pix_fmt", "yuv420p",
output_path,
]
print(f"Starting ffmpeg recording -> {output_path}")
return subprocess.Popen(
cmd,
stdout=subprocess.DEVNULL,
stderr=subprocess.PIPE,
)
def stop_recording(proc):
"""Send SIGTERM and wait for ffmpeg to finalize the MP4."""
print("Stopping ffmpeg ...")
proc.send_signal(signal.SIGTERM)
try:
proc.wait(timeout=10)
except subprocess.TimeoutExpired:
proc.kill()
proc.wait()
def browse_sites():
"""Launch headed Chromium via Playwright and visit each site."""
from playwright.sync_api import sync_playwright # noqa: PLC0415
with sync_playwright() as pw:
browser = pw.chromium.launch(
headless=False,
args=[
"--no-sandbox",
"--disable-gpu",
f"--display={DISPLAY}",
],
)
page = browser.new_page(viewport={"width": 1280, "height": 1024})
for url, pause in SITES:
print(f" Navigating to {url} ...")
try:
page.goto(url, timeout=15000)
except Exception as exc: # noqa: BLE001
print(f" Warning: navigation error ({exc}), continuing")
time.sleep(pause)
browser.close()
def main():
output_path = sys.argv[1] if len(sys.argv) > 1 else DEFAULT_OUTPUT
wait_for_xvfb()
ffmpeg = start_recording(output_path)
# Give ffmpeg a moment to initialize before browser activity
time.sleep(1)
try:
browse_sites()
finally:
stop_recording(ffmpeg)
if os.path.isfile(output_path):
size = os.path.getsize(output_path)
print(f"\nSUCCESS: {output_path} ({size:,} bytes)")
else:
sys.exit(f"ERROR: output file not created: {output_path}")
if __name__ == "__main__":
main()
Terminal window
python3 vnc-browser-test.py /tmp/recording.mp4
Waiting for Xvfb on :99 ...
Display ready after 0.3s
Starting ffmpeg recording -> /tmp/recording.mp4
Navigating to https://example.com ...
Navigating to https://httpbin.org ...
Stopping ffmpeg ...
SUCCESS: /tmp/recording.mp4 (1,234,567 bytes)

MP4 應顯示 Chromium 視窗在 fluxbox 桌面上渲染兩個網站的畫面。使用 ffplayvlc 播放,或從容器下載檔案進行驗證。

症狀可能原因
Xvfb display :99 not readyXvfb 未啟動 — 請檢查 entrypoint 是否執行了 VNC 堆疊
ffmpeg: x11grab error顯示器不匹配或 ffmpeg 缺少 x11grab 支援
Playwright launch errorChromium 未安裝或缺少 --no-sandbox 參數
MP4 為 0 位元組ffmpeg 在寫入前被終止 — 請檢查 SIGTERM 處理