#!/usr/bin/env python3 # -*- coding: utf-8 -*- """一键出片 端到端:真实跑完向导 4 步,然后断言时间轴里真的有转场/运镜关键帧/字幕效果。 依赖:系统 chromium + playwright(与 scripts/e2e_template.py 相同)。 用法:python3 scripts/e2e_quickvideo_fx.py """ import sys, time, json, re from playwright.sync_api import sync_playwright BASE = "https://127.0.0.1:3005" APPID = "qv_fx_%d" % int(time.time()) STAMP = time.strftime("%H:%M:%S") def log(msg): print("[%s] %s" % (STAMP, msg), flush=True) GUARD = """async () => { try { const root = await navigator.storage.getDirectory(); const db = await new Promise((res, rej) => { const rq = indexedDB.open('H5VideoLocalDB'); rq.onupgradeneeded = () => rq.result.createObjectStore('folderHandles'); rq.onsuccess = () => res(rq.result); rq.onerror = () => rej(rq.error); }); const done = await new Promise((res, rej) => { const tx = db.transaction('folderHandles', 'readwrite'); const st = tx.objectStore('folderHandles'); const rec = { id: 'root', handle: root }; // 应用自己建的 store 用内联键(keyPath='id'),我们脚本建的是外置键,两种都兼容 if (st.keyPath) st.put(rec); else st.put(rec, 'root'); tx.oncomplete = () => res(true); tx.onerror = () => rej(tx.error); }); db.close(); return done; } catch (e) { return String(e); } }""" def js_click(page, expr, desc): ok = page.evaluate("() => { const e = %s; if (e) { e.click(); return true; } return false; }" % expr) log("点 %s: %s" % (desc, ok)) return ok def wait_text(page, text, seconds=120, label=""): """等页面出现某个文案 —— 用它当『到第几步了』的唯一判据,别用会撞车的按钮片段。""" for _ in range(seconds): page.wait_for_timeout(1000) if page.evaluate("() => document.body.innerText.includes(%s)" % json.dumps(text)): log("到位[%s]: %s" % (label, text)) return True log("等待超时[%s]: %s" % (label, text)) return False def pick(page, shows, option, desc): """打开 semi Select 并选一项:shows=当前可见文案(定位哪个下拉),option=要选的文案""" opened = page.evaluate( "() => { const s=[...document.querySelectorAll('.semi-select')].find(x=>x.innerText.includes('%s'));" " if(s){s.click();return true;} return false; }" % shows ) if not opened: log("下拉没找到:%s" % desc) return False page.wait_for_timeout(600) ok = page.evaluate( "() => { const o=[...document.querySelectorAll('.semi-select-option')].find(x=>x.innerText.trim()==='%s');" " if(o){o.click();return true;} return false; }" % option ) log("选 %s -> %s: %s" % (desc, option, ok)) return ok def make_images(): from PIL import Image, ImageDraw out = [] for i, color in enumerate([(200, 60, 60), (60, 140, 200), (80, 170, 90)], 1): p = "/tmp/qv_img%d.png" % i im = Image.new("RGB", (960, 720), color) d = ImageDraw.Draw(im) d.text((40, 40), "QV FX TEST %d" % i, fill=(255, 255, 255)) d.rectangle([20, 20, 940, 700], outline=(255, 255, 255), width=6) im.save(p) out.append(p) return out def main(): imgs = make_images() log("测试图片: %s" % imgs) with sync_playwright() as p: browser = p.chromium.launch( executable_path="/usr/bin/chromium", headless=True, args=["--no-sandbox", "--disable-gpu", "--disable-dev-shm-usage", "--mute-audio", "--autoplay-policy=no-user-gesture-required"], ) ctx = browser.new_context(ignore_https_errors=True) page = ctx.new_page() errors = [] bad = [] page.on("pageerror", lambda e: errors.append(str(e))) page.on("console", lambda m: errors.append("console:" + m.text[:200]) if m.type == "error" else None) # 把向导自己的收尾日志打出来(里面有统一轨道/转场计数) page.on("console", lambda m: log("页面日志 %s" % m.text[:220]) if "[qv]" in m.text else None) page.on("response", lambda r: bad.append("%d %s" % (r.status, r.url[:120])) if r.status >= 400 else None) log("打开工程 %s" % APPID) page.goto("%s/editor/%s" % (BASE, APPID), wait_until="domcontentloaded", timeout=120000) page.wait_for_timeout(4000) log("守卫: %s" % page.evaluate(GUARD)) page.reload(wait_until="domcontentloaded") # 等编辑器真正起来(侧栏是纯文本,没有 aria-label,只能等文案出现) page.wait_for_selector('#h5dsCanvas', timeout=90000) try: page.wait_for_function("() => document.body.innerText.includes('AI创作')", timeout=90000) log("侧栏已就绪") except Exception as e: log("侧栏等待超时: %s" % e) page.wait_for_timeout(3000) # 打开 AI 面板 -> 一键出片(入口是文本叶子,aria-label 不可用) js_click(page, """(() => { const w = document.createTreeWalker(document.body, NodeFilter.SHOW_TEXT); let n; while (n = w.nextNode()) { const t = (n.textContent || '').trim(); if (t === 'AI创作' && n.parentElement) return n.parentElement; } return null; })()""", "AI创作") page.wait_for_timeout(2500) js_click(page, "[...document.querySelectorAll('button')].find(x => (x.innerText||'').trim() === '开始一键出片') " "|| [...document.querySelectorAll('button')].find(x => (x.innerText||'').includes('一键出片'))", "开始一键出片") page.wait_for_timeout(3000) # 步骤1:上传 3 张图 + 标题 fi = page.query_selector("div.semi-modal input[type=file]") if not fi: fi = page.query_selector("input[type=file][accept='image/*']") if not fi: log("ERROR: 找不到向导的图片输入框") print(json.dumps({"ok": False, "why": "no-file-input", "errors": errors[-5:]}, ensure_ascii=False)) return 1 fi.set_input_files(imgs) page.wait_for_timeout(2500) page.fill("input[placeholder*='三步搞定']", "测试标题效果验证") page.wait_for_timeout(500) js_click(page, "[...document.querySelectorAll('button')].find(x=>(x.innerText||'').includes('① 识别图片'))", "生成文案") # 到第 2 步的唯一标志:第 1 步的「下一步:配置声音」按钮出现(最长 180s) got = wait_text(page, "下一步:配置声音", 180, "第2步") log("文案生成完成: %s" % got) if not got: # 兜底:手动加 3 段 log("AI 文案没出来,改用手动加段") for i in range(3): js_click(page, "[...document.querySelectorAll('button')].find(x=>(x.innerText||'').trim()==='加一段')", "加一段") page.wait_for_timeout(500) tas = page.query_selector_all(".semi-modal textarea") for i, ta in enumerate(tas[:3]): ta.fill("第%d段测试口播内容" % (i + 1)) page.wait_for_timeout(500) js_click(page, "[...document.querySelectorAll('button')].find(x=>(x.innerText||'').includes('下一步:配置声音'))", "下一步") page.wait_for_timeout(2000) # 步骤3:合成全部配音 js_click(page, "[...document.querySelectorAll('button')].find(x=>(x.innerText||'').includes('合成全部配音'))", "合成全部配音") # 合成完会自动跳到第 4 步;只认效果面板的特征文案「图片转场」, # 不能用「确认生成」——第 3 步的「下一步:确认生成」也含这几个字,会误判。 done = wait_text(page, "图片转场", 300, "第4步效果面板") log("配音合成完成(自动进入确认页): %s" % done) page.wait_for_timeout(1500) # 步骤4:设置画面/字幕效果(全部用非默认值,证明控件真的生效) body_before = page.inner_text(".semi-modal-body") log("确认页含效果面板: %s" % ("图片转场" in body_before and "字幕效果" in body_before)) pick(page, "自动轮换", "左侧擦除", "转场") pick(page, "缓慢推近", "左右平移", "图片动效") pick(page, "底部", "居中", "字幕位置") pick(page, "黄字黑边", "白字黑边", "字幕配色") pick(page, "放大弹入", "放大 + 淡入", "字幕入场") # 大小滑块:聚焦后按右键 5 次(0.6 -> 1.1) moved = page.evaluate("""() => { const h = document.querySelector('.semi-modal .semi-slider-handle'); if (!h) return 'no-handle'; h.focus(); return 'focused'; }""") if moved == "focused": for _ in range(5): page.keyboard.press("ArrowRight") page.wait_for_timeout(400) log("字幕大小滑块: %s" % moved) page.wait_for_timeout(800) log("确认页摘要: %s" % page.inner_text(".semi-modal .semi-modal-body").split("\n")[0:14]) # 生成 js_click(page, "[...document.querySelectorAll('button')].find(x=>(x.innerText||'').includes('③ 确认生成视频'))", "确认生成视频") page.wait_for_timeout(12000) # 断言:读回工程数据 res = page.evaluate("""() => { const ed = window.__editor && window.__editor.movie; if (!ed) return {err:'no-editor'}; const d = ed.data || {}; const els = d.elements || []; const imgs = els.filter(e => e.type === 'image'); const trs = d.transitions || []; const caps = d.captions || []; // 转场有效性:startElementId 必须存在,且同轨道相邻 const byId = {}; els.forEach(e => byId[e.id] = e); const check = trs.map(t => { const a = byId[t.startElementId]; if (!a) return {name: t.name, ok: false, why: 'startElementId 不存在'}; // 注意:不要用 e !== a 排除自身 —— indexOf(a) 会永远返回 -1,next 就会取错 const same = els.filter(e => e.trackIndex === a.trackIndex) .sort((m, n) => m.startTime - n.startTime); const i = same.indexOf(a); const next = same[i + 1]; const gap = next ? Number((a.startTime + a.duration - next.startTime).toFixed(3)) : null; return { name: t.name, tname: t.transitionName, dur: t.duration, gap, ok: !!next && Math.abs(gap) < 0.01 }; }); return { imgCount: imgs.length, sameTrack: imgs.length ? new Set(imgs.map(e => e.trackIndex)).size : 0, elementsFull: els.map(e => ({ id: e.id, type: e.type, t: e.startTime, d: e.duration, track: e.trackIndex })), transitionsFull: trs.map(t => ({ id: t.id, startElementId: t.startElementId, tname: t.transitionName, dur: t.duration })), imgs: imgs.map(e => ({ t: e.startTime, d: e.duration, track: e.trackIndex, frames: (e.frames || []).length, f0: (e.frames || [])[0] || null, f1: (e.frames || [])[(e.frames || []).length - 1] || null, style: { x: e.style && e.style.x, y: e.style && e.style.y, w: e.style && e.style.width, h: e.style && e.style.height } })), transitionCount: trs.length, transitionChecks: check, captionCount: caps.length, caption0: caps[0] ? { x: caps[0].style.x, y: caps[0].style.y, textScale: caps[0].textScale, fill: caps[0].textStyle && caps[0].textStyle.fill, stroke: caps[0].textStyle && caps[0].textStyle.stroke && caps[0].textStyle.stroke.color, frames: (caps[0].frames || []).length, f0: (caps[0].frames || [])[0] || null, f1: (caps[0].frames || [])[(caps[0].frames || []).length - 1] || null, } : null, canvas: { w: d.width, h: d.height }, }; }""") log("元素明细: " + json.dumps(res.get('elementsFull'), ensure_ascii=False)) log("转场明细: " + json.dumps(res.get('transitionsFull'), ensure_ascii=False)) log("工程数据: " + json.dumps(res, ensure_ascii=False)[:2400]) # 播一下确认渲染 play = page.evaluate("""() => { const m = window.__editor && window.__editor.movie; if (!m) return 'no-movie'; try { m.currentTime = 1.5; if (typeof m.play === 'function') m.play(); return 'playing'; } catch (e) { return 'err:' + e.message; } }""") page.wait_for_timeout(4000) after = page.evaluate("""() => { const m = window.__editor && window.__editor.movie; const st = document.querySelector('#h5dsCanvas') && document.querySelector('#h5dsCanvas').innerText || ''; return { time: m ? m.currentTime : null, loading: /loading\\s+\\d/.test(st) }; }""") log("播放: %s -> %s" % (play, json.dumps(after, ensure_ascii=False))) page.screenshot(path="/tmp/qv_fx.png", full_page=False) log("截图: /tmp/qv_fx.png") # 转场区间内截一帧,确认转场真的在渲染(而不是只有数据) mid = page.evaluate("""() => { const st = window.__editor; const m = st && st.movie; if (!m) return 'no-movie'; const trs = (m.data && m.data.transitions) || []; if (!trs.length) return 'no-transitions'; const tr = trs[0]; const el = (m.data.elements || []).find(e => e.id === tr.startElementId); if (!el) return 'no-start-element'; const t = el.startTime + el.duration - 0.5; m.pause(); if (typeof st.updateMovie === 'function') st.updateMovie(t); else m.currentTime = t; return 'seek=' + Math.round(t * 100) / 100; }""") page.wait_for_timeout(2500) log("转场区间定位: %s" % mid) page.screenshot(path="/tmp/qv_fx_transition.png", full_page=False) log("转场截图: /tmp/qv_fx_transition.png") # 转场窗口内多点采样,看是不是渐变过渡(红->蓝中间应出现混合色)而不是硬切 samples = page.evaluate("""() => { const m = window.__editor && window.__editor.movie; const trs = (m && m.data && m.data.transitions) || []; if (!trs.length) return null; const el = m.data.elements.find(e => e.id === trs[0].startElementId); const end = el.startTime + el.duration; const win = trs[0].duration || 1; const pts = []; for (let k = 1; k <= 4; k++) pts.push(Number((end - win + (win * k) / 5).toFixed(2))); return pts; }""") if samples: log("转场窗口采样点: %s" % samples) for idx, tt in enumerate(samples): # 必须先暂停再跳帧;跳帧要用 Store 的 updateMovie(t)(内部是 movie.step), # 直接赋 movie.currentTime 不会驱动渲染,四个采样点会停在同一帧。 page.evaluate( "(t) => { const st = window.__editor;" " if (st && st.movie && st.movie.pause) st.movie.pause();" " if (st && st.updateMovie) st.updateMovie(t);" " else if (st && st.movie) st.movie.currentTime = t; }", tt, ) page.wait_for_timeout(1200) # 确认跳帧真的生效(currentTime 赋值未必驱动渲染) cur = page.evaluate("""() => { const m = window.__editor.movie; const tc = (document.body.innerText.split('\\n').join(' ').match(/\\d{2}:\\d{2}:\\d{2}/g) || []).slice(0, 3); return { movieTime: Math.round(m.currentTime * 100) / 100, timecodes: tc }; }""") page.locator("#h5dsCanvas").screenshot(path="/tmp/qv_tr_%d.png" % idx) log("采样 t=%.2f -> %s" % (tt, json.dumps(cur, ensure_ascii=False))) # 暂停态 step() 可能不渲染转场滤镜,所以再做一次「边播边截帧」的验证 capinfo = page.evaluate("""() => { const m = window.__editor.movie; const trs = (m.data && m.data.transitions) || []; if (!trs.length) return null; const el = m.data.elements.find(e => e.id === trs[0].startElementId); const end = el.startTime + el.duration; return { start: Number((end - (trs[0].duration || 1) - 0.5).toFixed(2)), boundary: end }; }""") if capinfo: page.evaluate("(t) => { const s=window.__editor; if (s.movie.pause) s.movie.pause(); s.updateMovie(t); }", capinfo["start"]) page.wait_for_timeout(900) page.evaluate("() => window.__editor.movie.play()") pts = [] for i in range(18): page.locator("#h5dsCanvas").screenshot(path="/tmp/qvp_%02d.png" % i) pts.append(page.evaluate("() => Math.round(window.__editor.movie.currentTime * 100) / 100")) log("播放采样起点=%.2f 边界=%.2f 时刻表=%s" % (capinfo["start"], capinfo["boundary"], pts)) # 保存 + 刷新,确认一键出片的结果不会丢 saved = page.evaluate("""() => { const s = window.__editor; if (typeof s.saveApp === 'function') { s.saveApp(() => {}); return 'saveApp 已调用'; } return '没有 saveApp'; }""") log("手动保存: %s" % saved) page.wait_for_timeout(8000) page.reload(wait_until="domcontentloaded") page.wait_for_selector("#h5dsCanvas", timeout=90000) page.wait_for_timeout(12000) after_reload = page.evaluate("""() => { const d = (window.__editor && window.__editor.movie && window.__editor.movie.data) || {}; return { els: (d.elements || []).length, caps: (d.captions || []).length, trans: (d.transitions || []).length }; }""") log("刷新后工程数据: %s" % json.dumps(after_reload, ensure_ascii=False)) log("JS 错误: %s" % (errors[-4:] if errors else "无")) log("非2xx请求: %s" % (bad[:8] if bad else "无")) browser.close() return 0 if __name__ == "__main__": sys.exit(main())