@harmonyos-arkts/d2h 0.1.11 → 0.1.12

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

@@ -1,249 +1,249 @@
-#!/usr/bin/env python3
-"""
-merge_appgraph.py — AppGraph(analyze android) facts × forge scan_features 语义合并。
-
-定位：scan_features 只见鸿蒙侧静态树（.id/行号/if 分支），缺行为语义；
-AppGraph facts 是安卓权威行为账本（控件/入口/事件/handler/callTrace/导航边/trigger 分类），
-但无鸿蒙侧定位事实。本脚本按 token 集合 join（answerFeature ↔ feature_answer_switch）
-把 Android 语义注入 features.json，产出同名 schema 的增强版，供 evaluate_coverage /
-用例设计（handler 名直接揭示联动语义）/ agent 融合（ref 体系对齐迁移报告）消费。
-
-用法:
-  merge_appgraph.py --features features.json --facts <android>/.migration/spec/facts \
-                    [--out features.merged.json]
-无 AppGraph facts 时退出码 2（调用方回退纯 scan_features）。
-"""
-import argparse
-import glob
-import json
-import os
-import re
-
-
-def tokens(s):
-    """camelCase + snake_case 统一切词：answerFeature 与 feature_answer → {answer, feature}。"""
-    parts = [p.lower() for p in re.findall(r'[A-Z]?[a-z0-9]+', str(s or ''))]
-    return set(parts) - {'menu', 'click', 'button', 'entry', 'the', 'id', 'app'}
-
-
-def norm_name(s):
-    return re.sub(r'[^a-z0-9]', '', str(s or '').lower())
-
-
-def join_key_candidates(name):
-    t = tokens(name)
-    return t, norm_name(name)
-
-
-def match(names, pool, event=None):
-    """names: 候选语义名（carrier/label/event 提取）；规则：norm 相等 > token 全等 >
-    android ⊂ forge id（容许 _switch 类小后缀）；checked 类事件优先 *_switch 目标。"""
-    for name in names:
-        if not name:
-            continue
-        ct, cn = tokens(name), norm_name(name)
-        if not ct and not cn:
-            continue
-        for ft in sorted(pool):
-            if cn and norm_name(ft) == cn:
-                return ft
-        cands = []
-        for ft in sorted(pool):
-            ftt = tokens(ft)
-            if not ftt or not ct:
-                continue
-            if ct == ftt:
-                cands.append((0, ft))
-            elif ct < ftt and len(ftt) <= len(ct) + 2:
-                cands.append((1, ft))
-        if not cands:
-            continue
-        cands.sort()
-        if event and str(event).startswith('checked'):
-            for _, ft in cands:
-                if 'switch' in ft or 'toggle' in ft:
-                    return ft
-        return cands[0][1]
-    return None
-
-
-def event_tokens(event):
-    """menu-click:privacy_policy → ['privacy_policy']；checked-change → []。"""
-    m = re.match(r'^menu-click:(\w+)$', str(event or ''))
-    return [m.group(1)] if m else []
-
-
-def harvest_ids(project_dir):
-    """收割工程内可静态定位的 ID：.id('x') 字面量 + 组件参数 *Id: 'x'（如 rowId）。
-    模板拼接 .id(`${rowId}_switch`) 无法静态解析，由 prop 值 + 已知后缀兜底。"""
-    pool = {}
-    root = os.path.join(project_dir or '.', '')
-    for path in glob.glob(os.path.join(root, '**', '*.ets'), recursive=True):
-        if os.sep + 'ohosTest' in path or os.sep + '.hypium_forge' in path:
-            continue
-        try:
-            src = open(path, encoding='utf-8').read()
-        except Exception:
-            continue
-        for m in re.finditer(r"\.id\(\s*['\"`]([\w-]+)['\"`]\s*\)", src):
-            line = src[:m.start()].count('\n') + 1
-            pool.setdefault(m.group(1), {'file': os.path.relpath(path, root), 'line': line})
-        for m in re.finditer(r"\b\w*[Ii]d\s*:\s*['\"]([\w-]+)['\"]", src):
-            for cand in (m.group(1), f"{m.group(1)}_switch"):
-                pool.setdefault(cand, {'file': os.path.relpath(path, root), 'line': src[:m.start()].count('\n') + 1})
-    return pool
-
-
-def load_appgraph(facts_dir):
-    controls, entries, nav = [], [], []
-    inter_path = os.path.join(facts_dir, 'interactions.json')
-    if os.path.isdir(facts_dir):
-        pages = sorted(glob.glob(os.path.join(facts_dir, 'pages', '*.json')))
-    else:
-        pages = []
-    for p in pages:
-        if os.path.basename(p) == 'index.json' or os.path.basename(p) == 'unassigned.json':
-            continue
-        try:
-            d = json.load(open(p, encoding='utf-8'))
-        except Exception:
-            continue
-        controls += d.get('controls', []) or []
-        entries += d.get('entries', []) or []
-    if os.path.exists(inter_path):
-        try:
-            nav = json.load(open(inter_path, encoding='utf-8')).get('interactions', []) or []
-        except Exception:
-            nav = []
-    return controls, entries, nav
-
-
-def main():
-    ap = argparse.ArgumentParser()
-    ap.add_argument('--features', required=True)
-    ap.add_argument('--facts', required=True)
-    ap.add_argument('--out', default=None)
-    args = ap.parse_args()
-
-    if not os.path.isdir(args.facts):
-        print(f'[x] AppGraph facts 不存在: {args.facts}')
-        raise SystemExit(2)
-    feats = json.load(open(args.features, encoding='utf-8'))
-    controls, entries, nav = load_appgraph(args.facts)
-    if not entries and not controls:
-        print('[x] facts 为空，回退 scan_features')
-        raise SystemExit(2)
-
-    forge_targets = []
-    for pg in feats.get('pages', []):
-        for it in pg.get('interactions', []):
-            forge_targets.append(it.get('target'))
-    id_pool = harvest_ids(feats.get('project_dir', '.'))
-    pool = set(forge_targets) | set(id_pool.keys())
-
-    android_entries = []
-    for e in entries:
-        rec = {
-            'name': e.get('carrierName') or e.get('displayLabel'),
-            'event': e.get('event'),
-            'handlers': e.get('handlerNames', []) or [],
-            'trigger': e.get('triggerCategory'),
-            'ref': e.get('entryRef'),
-            'label': e.get('displayLabel'),
-            'android_evidence': (e.get('sourceEvidence') or [{}])[0].get('file', ''),
-        }
-        android_entries.append(rec)
-
-    android_controls = []
-    for c in controls:
-        android_controls.append({
-            'name': c.get('name'),
-            'ref': c.get('controlRef'),
-            'evidence': (c.get('sourceEvidence') or [{}])[0].get('file', ''),
-        })
-
-    matched, unmatched = {}, []
-    for ae in android_entries:
-        if ae['trigger'] != 'user':
-            continue
-        cand = [ae['name'], ae['label']] + event_tokens(ae['event'])
-        ft = match(cand, pool, ae['event'])
-        if ft:
-            ae['forge_target'] = ft
-            matched.setdefault(ft, []).append(ae)
-        else:
-            unmatched.append(ae)
-
-    nav_edges = [{'from': n.get('fromRef'), 'to': n.get('toRef'), 'relation': n.get('relation')}
-                 for n in nav]
-
-    enriched_pages = []
-    for pg in feats.get('pages', []):
-        for it in pg.get('interactions', []):
-            hits = matched.get(it.get('target'), [])
-            if hits:
-                h = hits[0]
-                it['android_event'] = h['event']
-                it['android_handlers'] = h['handlers']
-                it['android_ref'] = h['ref']
-                it['android_evidence'] = h['android_evidence']
-        enriched_pages.append(pg)
-
-    entry_path = next((p['path'] for p in enriched_pages if enriched_pages and p['path'] == feats.get('entry', {}).get('page')),
-                      enriched_pages[0]['path'] if enriched_pages else None)
-    entry_idx = next((i for i, p in enumerate(enriched_pages) if p['path'] == entry_path), 0)
-
-    def owner_unit(h, loc):
-        """归属：checked-change → 拥有 lineN:Toggle 的组件单元（并移除被具体化的 lineN）；
-        其余 → 入口页。"""
-        if str(h['event']).startswith('checked'):
-            for i, p in enumerate(enriched_pages):
-                for it in p.get('interactions', []):
-                    if str(it.get('target', '')).startswith('line') and it.get('type') == 'Toggle':
-                        p['interactions'] = [x for x in p['interactions'] if x is not it]
-                        return i
-        return entry_idx
-
-    for ft, hits in matched.items():
-        if any(ft == it.get('target') for pg in enriched_pages for it in pg.get('interactions', [])):
-            continue
-        loc = id_pool.get(ft, {})
-        h = hits[0]
-        enriched_pages[owner_unit(h, loc)]['interactions'].append({
-            'kind': 'click' if 'click' in str(h['event']) else 'toggle',
-            'target': ft,
-            'desc': f"android:{h['name']}({h['event']})",
-            'has_id': ft in id_pool,
-            'line': loc.get('line', 0),
-            'type': 'IdRef',
-            'android_event': h['event'],
-            'android_handlers': h['handlers'],
-            'android_ref': h['ref'],
-            'android_evidence': h['android_evidence'],
-            'android_file': loc.get('file', ''),
-        })
-
-    feats['pages'] = enriched_pages
-    feats['android'] = {
-        'controls': android_controls,
-        'entries_user': [e for e in android_entries if e['trigger'] == 'user'],
-        'entries_system': [e for e in android_entries if e['trigger'] != 'user'],
-        'nav_edges': nav_edges,
-        'unmatched_user_entries': unmatched,
-        'source': args.facts,
-    }
-
-    out = args.out or args.features.replace('.json', '.merged.json')
-    json.dump(feats, open(out, 'w', encoding='utf-8'), ensure_ascii=False, indent=1)
-    print(f'合并完成: controls={len(android_controls)} user入口={len(feats["android"]["entries_user"])} '
-          f'系统入口={len(feats["android"]["entries_system"])} join命中={len(matched)}/{len(matched)+len(unmatched)} '
-          f'导航边={len(nav_edges)} → {out}')
-    for ft, hits in sorted(matched.items()):
-        for h in hits:
-            print(f'  [+] {h["name"]}({h["event"]}) → {ft}  handlers={h["handlers"]}')
-    for u in unmatched:
-        print(f'  [?] 未join: {u["name"]} ({u["event"]}) handlers={u["handlers"]}')
-
-
-if __name__ == '__main__':
-    main()
+#!/usr/bin/env python3
+"""
+merge_appgraph.py — AppGraph(analyze android) facts × forge scan_features 语义合并。
+
+定位：scan_features 只见鸿蒙侧静态树（.id/行号/if 分支），缺行为语义；
+AppGraph facts 是安卓权威行为账本（控件/入口/事件/handler/callTrace/导航边/trigger 分类），
+但无鸿蒙侧定位事实。本脚本按 token 集合 join（answerFeature ↔ feature_answer_switch）
+把 Android 语义注入 features.json，产出同名 schema 的增强版，供 evaluate_coverage /
+用例设计（handler 名直接揭示联动语义）/ agent 融合（ref 体系对齐迁移报告）消费。
+
+用法:
+  merge_appgraph.py --features features.json --facts <android>/.migration/spec/facts \
+                    [--out features.merged.json]
+无 AppGraph facts 时退出码 2（调用方回退纯 scan_features）。
+"""
+import argparse
+import glob
+import json
+import os
+import re
+
+
+def tokens(s):
+    """camelCase + snake_case 统一切词：answerFeature 与 feature_answer → {answer, feature}。"""
+    parts = [p.lower() for p in re.findall(r'[A-Z]?[a-z0-9]+', str(s or ''))]
+    return set(parts) - {'menu', 'click', 'button', 'entry', 'the', 'id', 'app'}
+
+
+def norm_name(s):
+    return re.sub(r'[^a-z0-9]', '', str(s or '').lower())
+
+
+def join_key_candidates(name):
+    t = tokens(name)
+    return t, norm_name(name)
+
+
+def match(names, pool, event=None):
+    """names: 候选语义名（carrier/label/event 提取）；规则：norm 相等 > token 全等 >
+    android ⊂ forge id（容许 _switch 类小后缀）；checked 类事件优先 *_switch 目标。"""
+    for name in names:
+        if not name:
+            continue
+        ct, cn = tokens(name), norm_name(name)
+        if not ct and not cn:
+            continue
+        for ft in sorted(pool):
+            if cn and norm_name(ft) == cn:
+                return ft
+        cands = []
+        for ft in sorted(pool):
+            ftt = tokens(ft)
+            if not ftt or not ct:
+                continue
+            if ct == ftt:
+                cands.append((0, ft))
+            elif ct < ftt and len(ftt) <= len(ct) + 2:
+                cands.append((1, ft))
+        if not cands:
+            continue
+        cands.sort()
+        if event and str(event).startswith('checked'):
+            for _, ft in cands:
+                if 'switch' in ft or 'toggle' in ft:
+                    return ft
+        return cands[0][1]
+    return None
+
+
+def event_tokens(event):
+    """menu-click:privacy_policy → ['privacy_policy']；checked-change → []。"""
+    m = re.match(r'^menu-click:(\w+)$', str(event or ''))
+    return [m.group(1)] if m else []
+
+
+def harvest_ids(project_dir):
+    """收割工程内可静态定位的 ID：.id('x') 字面量 + 组件参数 *Id: 'x'（如 rowId）。
+    模板拼接 .id(`${rowId}_switch`) 无法静态解析，由 prop 值 + 已知后缀兜底。"""
+    pool = {}
+    root = os.path.join(project_dir or '.', '')
+    for path in glob.glob(os.path.join(root, '**', '*.ets'), recursive=True):
+        if os.sep + 'ohosTest' in path or os.sep + '.hypium_forge' in path:
+            continue
+        try:
+            src = open(path, encoding='utf-8').read()
+        except Exception:
+            continue
+        for m in re.finditer(r"\.id\(\s*['\"`]([\w-]+)['\"`]\s*\)", src):
+            line = src[:m.start()].count('\n') + 1
+            pool.setdefault(m.group(1), {'file': os.path.relpath(path, root), 'line': line})
+        for m in re.finditer(r"\b\w*[Ii]d\s*:\s*['\"]([\w-]+)['\"]", src):
+            for cand in (m.group(1), f"{m.group(1)}_switch"):
+                pool.setdefault(cand, {'file': os.path.relpath(path, root), 'line': src[:m.start()].count('\n') + 1})
+    return pool
+
+
+def load_appgraph(facts_dir):
+    controls, entries, nav = [], [], []
+    inter_path = os.path.join(facts_dir, 'interactions.json')
+    if os.path.isdir(facts_dir):
+        pages = sorted(glob.glob(os.path.join(facts_dir, 'pages', '*.json')))
+    else:
+        pages = []
+    for p in pages:
+        if os.path.basename(p) == 'index.json' or os.path.basename(p) == 'unassigned.json':
+            continue
+        try:
+            d = json.load(open(p, encoding='utf-8'))
+        except Exception:
+            continue
+        controls += d.get('controls', []) or []
+        entries += d.get('entries', []) or []
+    if os.path.exists(inter_path):
+        try:
+            nav = json.load(open(inter_path, encoding='utf-8')).get('interactions', []) or []
+        except Exception:
+            nav = []
+    return controls, entries, nav
+
+
+def main():
+    ap = argparse.ArgumentParser()
+    ap.add_argument('--features', required=True)
+    ap.add_argument('--facts', required=True)
+    ap.add_argument('--out', default=None)
+    args = ap.parse_args()
+
+    if not os.path.isdir(args.facts):
+        print(f'[x] AppGraph facts 不存在: {args.facts}')
+        raise SystemExit(2)
+    feats = json.load(open(args.features, encoding='utf-8'))
+    controls, entries, nav = load_appgraph(args.facts)
+    if not entries and not controls:
+        print('[x] facts 为空，回退 scan_features')
+        raise SystemExit(2)
+
+    forge_targets = []
+    for pg in feats.get('pages', []):
+        for it in pg.get('interactions', []):
+            forge_targets.append(it.get('target'))
+    id_pool = harvest_ids(feats.get('project_dir', '.'))
+    pool = set(forge_targets) | set(id_pool.keys())
+
+    android_entries = []
+    for e in entries:
+        rec = {
+            'name': e.get('carrierName') or e.get('displayLabel'),
+            'event': e.get('event'),
+            'handlers': e.get('handlerNames', []) or [],
+            'trigger': e.get('triggerCategory'),
+            'ref': e.get('entryRef'),
+            'label': e.get('displayLabel'),
+            'android_evidence': (e.get('sourceEvidence') or [{}])[0].get('file', ''),
+        }
+        android_entries.append(rec)
+
+    android_controls = []
+    for c in controls:
+        android_controls.append({
+            'name': c.get('name'),
+            'ref': c.get('controlRef'),
+            'evidence': (c.get('sourceEvidence') or [{}])[0].get('file', ''),
+        })
+
+    matched, unmatched = {}, []
+    for ae in android_entries:
+        if ae['trigger'] != 'user':
+            continue
+        cand = [ae['name'], ae['label']] + event_tokens(ae['event'])
+        ft = match(cand, pool, ae['event'])
+        if ft:
+            ae['forge_target'] = ft
+            matched.setdefault(ft, []).append(ae)
+        else:
+            unmatched.append(ae)
+
+    nav_edges = [{'from': n.get('fromRef'), 'to': n.get('toRef'), 'relation': n.get('relation')}
+                 for n in nav]
+
+    enriched_pages = []
+    for pg in feats.get('pages', []):
+        for it in pg.get('interactions', []):
+            hits = matched.get(it.get('target'), [])
+            if hits:
+                h = hits[0]
+                it['android_event'] = h['event']
+                it['android_handlers'] = h['handlers']
+                it['android_ref'] = h['ref']
+                it['android_evidence'] = h['android_evidence']
+        enriched_pages.append(pg)
+
+    entry_path = next((p['path'] for p in enriched_pages if enriched_pages and p['path'] == feats.get('entry', {}).get('page')),
+                      enriched_pages[0]['path'] if enriched_pages else None)
+    entry_idx = next((i for i, p in enumerate(enriched_pages) if p['path'] == entry_path), 0)
+
+    def owner_unit(h, loc):
+        """归属：checked-change → 拥有 lineN:Toggle 的组件单元（并移除被具体化的 lineN）；
+        其余 → 入口页。"""
+        if str(h['event']).startswith('checked'):
+            for i, p in enumerate(enriched_pages):
+                for it in p.get('interactions', []):
+                    if str(it.get('target', '')).startswith('line') and it.get('type') == 'Toggle':
+                        p['interactions'] = [x for x in p['interactions'] if x is not it]
+                        return i
+        return entry_idx
+
+    for ft, hits in matched.items():
+        if any(ft == it.get('target') for pg in enriched_pages for it in pg.get('interactions', [])):
+            continue
+        loc = id_pool.get(ft, {})
+        h = hits[0]
+        enriched_pages[owner_unit(h, loc)]['interactions'].append({
+            'kind': 'click' if 'click' in str(h['event']) else 'toggle',
+            'target': ft,
+            'desc': f"android:{h['name']}({h['event']})",
+            'has_id': ft in id_pool,
+            'line': loc.get('line', 0),
+            'type': 'IdRef',
+            'android_event': h['event'],
+            'android_handlers': h['handlers'],
+            'android_ref': h['ref'],
+            'android_evidence': h['android_evidence'],
+            'android_file': loc.get('file', ''),
+        })
+
+    feats['pages'] = enriched_pages
+    feats['android'] = {
+        'controls': android_controls,
+        'entries_user': [e for e in android_entries if e['trigger'] == 'user'],
+        'entries_system': [e for e in android_entries if e['trigger'] != 'user'],
+        'nav_edges': nav_edges,
+        'unmatched_user_entries': unmatched,
+        'source': args.facts,
+    }
+
+    out = args.out or args.features.replace('.json', '.merged.json')
+    json.dump(feats, open(out, 'w', encoding='utf-8'), ensure_ascii=False, indent=1)
+    print(f'合并完成: controls={len(android_controls)} user入口={len(feats["android"]["entries_user"])} '
+          f'系统入口={len(feats["android"]["entries_system"])} join命中={len(matched)}/{len(matched)+len(unmatched)} '
+          f'导航边={len(nav_edges)} → {out}')
+    for ft, hits in sorted(matched.items()):
+        for h in hits:
+            print(f'  [+] {h["name"]}({h["event"]}) → {ft}  handlers={h["handlers"]}')
+    for u in unmatched:
+        print(f'  [?] 未join: {u["name"]} ({u["event"]}) handlers={u["handlers"]}')
+
+
+if __name__ == '__main__':
+    main()