@harmonyos-arkts/d2h 0.1.11 → 0.1.13

package/skills/d2h-hypium-forge/scripts/collect_runtime_pages.py

@@ -1,169 +1,169 @@
-#!/usr/bin/env python3
-"""
-collect_runtime_pages.py — 运行时页面文本采集器（设计前置，消灭上下文假设错位）。
-
-流程: 启动 App → dump 首页 → 逐个点击底部 Tab → dump →
-      对每个已 dump 页面提取「可点击文本」（clickable 或短文本按钮类）逐个浅层点击 → dump 新页面 → 返回。
-产出 runtime_pages.json: [{"key": "home|tab:我的|click:设置", "texts": [...], "hints": [...], "from": 父页key}]
-
-用法: collect_runtime_pages.py --device 127.0.0.1:5559 --bundle com.x [--max-pages 20] [--out runtime_pages.json]
-"""
-import argparse
-import json
-import re
-import subprocess
-import time
-
-from hypium import BY, MatchPattern
-
-HDC = '/Applications/DevEco-Studio.app/Contents/sdk/default/openharmony/toolchains/hdc'
-
-
-def dump_texts(sn, tag):
-    """hdc dumpLayout → (texts, clickable_texts, hints)。"""
-    dev = f'/data/local/tmp/rp_{tag}.json'
-    subprocess.run(['bash', '-c', f'{HDC} -t {sn} shell uitest dumpLayout -p {dev}'],
-                   capture_output=True, timeout=30)
-    local = f'/tmp/rp_{tag}.json'
-    subprocess.run(['bash', '-c', f'{HDC} -t {sn} file recv {dev} {local}'],
-                   capture_output=True, timeout=30)
-    try:
-        d = json.load(open(local, encoding='utf-8'))
-    except Exception:
-        return [], [], []
-    texts, clickables, hints = [], [], []
-
-    def walk(n):
-        a = n.get('attributes', {})
-        t = a.get('text')
-        if t:
-            ts = str(t).strip()
-            if ts and len(ts) < 20 and not re.match(r'^[\d:,.\s]+$', ts):
-                texts.append(ts)
-                if str(a.get('clickable')) == 'true':
-                    clickables.append(ts)
-        if a.get('hint'):
-            hints.append(str(a['hint'])[:16])
-        for c in n.get('children', []) or []:
-            walk(c)
-
-    walk(d)
-    seen_t, seen_c = [], []
-    for t in texts:
-        if t not in seen_t:
-            seen_t.append(t)
-    for t in clickables:
-        if t not in seen_c:
-            seen_c.append(t)
-    return seen_t, seen_c, hints
-
-
-def main():
-    ap = argparse.ArgumentParser()
-    ap.add_argument('--device', required=True)
-    ap.add_argument('--bundle', required=True)
-    ap.add_argument('--ability', default='EntryAbility')
-    ap.add_argument('--max-pages', type=int, default=18)
-    ap.add_argument('--out', default='runtime_pages.json')
-    args = ap.parse_args()
-
-    from hypium import UiDriver
-    d = UiDriver.connect(device_sn=args.device)
-    # 热启动（冷启动后 UI agent 短暂不可用，实测文本点击会静默失败）
-    d.start_app(args.bundle, args.ability)
-    for _ in range(20):
-        try:
-            if d.find_component(BY.text('首页')):
-                break
-        except Exception:
-            pass
-        time.sleep(1)
-    time.sleep(2)
-
-    pages = []
-    queue = []  # (key, clickable_texts) 待浅探
-
-    def cap(key):
-        texts, clicks, hints = dump_texts(args.device, abs(hash(key)) % 10000)
-        pages.append({'key': key, 'texts': texts[:40], 'clickables': clicks[:15], 'hints': hints[:5]})
-        print(f'[+] {key}: {len(texts)} 文本 / {len(clicks)} 可点', flush=True)
-        return clicks
-
-    clicks = cap('home')
-    queue.append(('home', clicks))
-
-    # 底部 Tab（取 home 可点文本中位于常见 Tab 词表的）
-    TABS = ['首页', '主页', '分类', '购物车', '我的', '预约', '课程', '发现']
-    if not clicks:
-        clicks = [t for t in pages[0]['texts'] if 2 <= len(t) <= 6]
-        pages[0]['clickables'] = clicks
-    tab_hits = [t for t in clicks if any(tw in t for tw in TABS)]
-    for t in tab_hits[:5]:
-        try:
-            click_text(d, t)
-            time.sleep(1.8)
-            key = f'tab:{t}'
-            c2 = cap(key)
-            queue.append((key, c2))
-            # 回 home 供下一个 tab 点击
-            try:
-                click_text(d, '首页')
-            except Exception:
-                d.press_back()
-            time.sleep(1.5)
-        except Exception as e:
-            print(f'  [x] tab {t}: {str(e)[:60]}', flush=True)
-
-    # BFS 浅层：每页最多取 2 个入口点击（限页数）
-    visited_clicks = set()
-    while len(pages) < args.max_pages and queue:
-        pkey, clicks = queue.pop(0)
-        if not clicks:
-            pg = next((p for p in pages if p['key'] == pkey), None)
-            clicks = [t for t in (pg['texts'] if pg else []) if 2 <= len(t) <= 6]
-        n = 0
-        for t in clicks:
-            if len(pages) >= args.max_pages or n >= 2:
-                break
-            if t in visited_clicks or len(t) < 2 or t in ('返回', '确定', '取消'):
-                continue
-            visited_clicks.add(t)
-            try:
-                click_text(d, t)
-                time.sleep(1.8)
-                key = f'{pkey}>{t}'
-                c3 = cap(key)
-                queue.append((key, c3))
-                n += 1
-            except Exception:
-                continue
-            finally:
-                try:
-                    d.press_back(); time.sleep(1.2)
-                except Exception:
-                    pass
-
-    json.dump(pages, open(args.out, 'w', encoding='utf-8'), ensure_ascii=False, indent=1)
-    print(f'采集完成: {len(pages)} 页 → {args.out}')
-
-
-def click_text(d, txt, wait=6):
-    import time as _t
-    deadline = _t.time() + wait
-    while _t.time() < deadline:
-        try:
-            comps = d.find_all_components(BY.text(txt, MatchPattern.CONTAINS))
-            if comps:
-                c = comps[-1]
-                b = c.getBounds()
-                d.touch(((b.left + b.right) // 2, (b.top + b.bottom) // 2))
-                _t.sleep(1.2)
-                return
-        except Exception:
-            pass
-        _t.sleep(0.4)
-    raise AssertionError('not clickable: ' + txt)
-
-
-if __name__ == '__main__':
-    main()
+#!/usr/bin/env python3
+"""
+collect_runtime_pages.py — 运行时页面文本采集器（设计前置，消灭上下文假设错位）。
+
+流程: 启动 App → dump 首页 → 逐个点击底部 Tab → dump →
+      对每个已 dump 页面提取「可点击文本」（clickable 或短文本按钮类）逐个浅层点击 → dump 新页面 → 返回。
+产出 runtime_pages.json: [{"key": "home|tab:我的|click:设置", "texts": [...], "hints": [...], "from": 父页key}]
+
+用法: collect_runtime_pages.py --device 127.0.0.1:5559 --bundle com.x [--max-pages 20] [--out runtime_pages.json]
+"""
+import argparse
+import json
+import re
+import subprocess
+import time
+
+from hypium import BY, MatchPattern
+
+HDC = '/Applications/DevEco-Studio.app/Contents/sdk/default/openharmony/toolchains/hdc'
+
+
+def dump_texts(sn, tag):
+    """hdc dumpLayout → (texts, clickable_texts, hints)。"""
+    dev = f'/data/local/tmp/rp_{tag}.json'
+    subprocess.run(['bash', '-c', f'{HDC} -t {sn} shell uitest dumpLayout -p {dev}'],
+                   capture_output=True, timeout=30)
+    local = f'/tmp/rp_{tag}.json'
+    subprocess.run(['bash', '-c', f'{HDC} -t {sn} file recv {dev} {local}'],
+                   capture_output=True, timeout=30)
+    try:
+        d = json.load(open(local, encoding='utf-8'))
+    except Exception:
+        return [], [], []
+    texts, clickables, hints = [], [], []
+
+    def walk(n):
+        a = n.get('attributes', {})
+        t = a.get('text')
+        if t:
+            ts = str(t).strip()
+            if ts and len(ts) < 20 and not re.match(r'^[\d:,.\s]+$', ts):
+                texts.append(ts)
+                if str(a.get('clickable')) == 'true':
+                    clickables.append(ts)
+        if a.get('hint'):
+            hints.append(str(a['hint'])[:16])
+        for c in n.get('children', []) or []:
+            walk(c)
+
+    walk(d)
+    seen_t, seen_c = [], []
+    for t in texts:
+        if t not in seen_t:
+            seen_t.append(t)
+    for t in clickables:
+        if t not in seen_c:
+            seen_c.append(t)
+    return seen_t, seen_c, hints
+
+
+def main():
+    ap = argparse.ArgumentParser()
+    ap.add_argument('--device', required=True)
+    ap.add_argument('--bundle', required=True)
+    ap.add_argument('--ability', default='EntryAbility')
+    ap.add_argument('--max-pages', type=int, default=18)
+    ap.add_argument('--out', default='runtime_pages.json')
+    args = ap.parse_args()
+
+    from hypium import UiDriver
+    d = UiDriver.connect(device_sn=args.device)
+    # 热启动（冷启动后 UI agent 短暂不可用，实测文本点击会静默失败）
+    d.start_app(args.bundle, args.ability)
+    for _ in range(20):
+        try:
+            if d.find_component(BY.text('首页')):
+                break
+        except Exception:
+            pass
+        time.sleep(1)
+    time.sleep(2)
+
+    pages = []
+    queue = []  # (key, clickable_texts) 待浅探
+
+    def cap(key):
+        texts, clicks, hints = dump_texts(args.device, abs(hash(key)) % 10000)
+        pages.append({'key': key, 'texts': texts[:40], 'clickables': clicks[:15], 'hints': hints[:5]})
+        print(f'[+] {key}: {len(texts)} 文本 / {len(clicks)} 可点', flush=True)
+        return clicks
+
+    clicks = cap('home')
+    queue.append(('home', clicks))
+
+    # 底部 Tab（取 home 可点文本中位于常见 Tab 词表的）
+    TABS = ['首页', '主页', '分类', '购物车', '我的', '预约', '课程', '发现']
+    if not clicks:
+        clicks = [t for t in pages[0]['texts'] if 2 <= len(t) <= 6]
+        pages[0]['clickables'] = clicks
+    tab_hits = [t for t in clicks if any(tw in t for tw in TABS)]
+    for t in tab_hits[:5]:
+        try:
+            click_text(d, t)
+            time.sleep(1.8)
+            key = f'tab:{t}'
+            c2 = cap(key)
+            queue.append((key, c2))
+            # 回 home 供下一个 tab 点击
+            try:
+                click_text(d, '首页')
+            except Exception:
+                d.press_back()
+            time.sleep(1.5)
+        except Exception as e:
+            print(f'  [x] tab {t}: {str(e)[:60]}', flush=True)
+
+    # BFS 浅层：每页最多取 2 个入口点击（限页数）
+    visited_clicks = set()
+    while len(pages) < args.max_pages and queue:
+        pkey, clicks = queue.pop(0)
+        if not clicks:
+            pg = next((p for p in pages if p['key'] == pkey), None)
+            clicks = [t for t in (pg['texts'] if pg else []) if 2 <= len(t) <= 6]
+        n = 0
+        for t in clicks:
+            if len(pages) >= args.max_pages or n >= 2:
+                break
+            if t in visited_clicks or len(t) < 2 or t in ('返回', '确定', '取消'):
+                continue
+            visited_clicks.add(t)
+            try:
+                click_text(d, t)
+                time.sleep(1.8)
+                key = f'{pkey}>{t}'
+                c3 = cap(key)
+                queue.append((key, c3))
+                n += 1
+            except Exception:
+                continue
+            finally:
+                try:
+                    d.press_back(); time.sleep(1.2)
+                except Exception:
+                    pass
+
+    json.dump(pages, open(args.out, 'w', encoding='utf-8'), ensure_ascii=False, indent=1)
+    print(f'采集完成: {len(pages)} 页 → {args.out}')
+
+
+def click_text(d, txt, wait=6):
+    import time as _t
+    deadline = _t.time() + wait
+    while _t.time() < deadline:
+        try:
+            comps = d.find_all_components(BY.text(txt, MatchPattern.CONTAINS))
+            if comps:
+                c = comps[-1]
+                b = c.getBounds()
+                d.touch(((b.left + b.right) // 2, (b.top + b.bottom) // 2))
+                _t.sleep(1.2)
+                return
+        except Exception:
+            pass
+        _t.sleep(0.4)
+    raise AssertionError('not clickable: ' + txt)
+
+
+if __name__ == '__main__':
+    main()