#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""前端 + README 同步脚本(skillhub-collection)。
职责:
1. 合并 data/ 与 public/data/ 两份已分叉的 local_skills.json(取并集,data/ 为主);
2. 写回两处保持一致(前端 local-skills.html 读 public/data/local_skills.json);
3. 重写 README.md「💾 本地技能库」段(LOCAL-SKILLS 标记内,机器人不会触碰);
4. 供收录流水线调用:新收录技能后跑一次即完成前端+README 同步。
用法: python scripts/sync_frontend.py [--dry-run]
"""
import json
import re
import sys
from datetime import date
from pathlib import Path
ROOT = Path(__file__).resolve().parent.parent
DATA_A = ROOT / "data" / "local_skills.json" # 主数据(机器人 badge 引用)
DATA_B = ROOT / "public" / "data" / "local_skills.json" # 前端数据
README = ROOT / "README.md"
MARK_START = "<!-- LOCAL-SKILLS-START -->"
MARK_END = "<!-- LOCAL-SKILLS-END -->"
CAT_ORDER_LEGACY = [ # 原 README 里的展示顺序,没出现的排后面
"内容创作与发布", "视频创作", "电商与营销", "PPT与演示", "语音与音频",
"数字人与视频配音", "文档与分析", "智能体协作", "产品与项目管理",
"设计与可视化", "AI 创意与可视化", "文档处理", "技能管理", "财务分析",
"文化创作", "邮件与办公自动化", "医疗与行业 Agent", "收录开源技能",
]
def merge(a: dict, b: dict) -> dict:
cats_a = a.get("categories", {})
cats_b = b.get("categories", {})
merged_cats = {}
names = list(cats_a.keys()) + [k for k in cats_b.keys() if k not in cats_a]
for cat in names:
sa = list((cats_a.get(cat) or {}).get("skills", []))
sb = list((cats_b.get(cat) or {}).get("skills", []))
seen, skills = set(), []
for s in sa + sb:
if s not in seen:
seen.add(s)
skills.append(s)
merged_cats[cat] = {"count": len(skills), "skills": skills}
# skill_details:以 A 为准,缺的从 B 补
details = dict(a.get("skill_details", {}))
for k, v in (b.get("skill_details") or {}).items():
details.setdefault(k, v)
# 统计:字段级取 A,数值取两边较大值(保守估计已收录规模)
stats_a = a.get("statistics", {})
stats_b = b.get("statistics", {})
stats = {}
for key in ("by_api_requirement", "by_usage_frequency"):
sa, sb = stats_a.get(key, {}), stats_b.get(key, {})
stats[key] = {k: max(sa.get(k, 0), sb.get(k, 0)) for k in set(sa) | set(sb)}
all_skills = sorted({s for c in merged_cats.values() for s in c["skills"]})
total = len(all_skills)
today = date.today().isoformat()
out = dict(a) # metadata/highlights 等以 A 为骨架
out["categories"] = merged_cats
out["skill_details"] = details
out["statistics"] = stats
out["metadata"] = {**a.get("metadata", {}),
"total_skills": total,
"last_updated": today}
log = list(a.get("update_log", []))
prev_total = a.get("metadata", {}).get("total_skills")
if str(prev_total) != str(total):
log.append({"date": today, "version": "auto-sync",
"changes": [f"前端/README 同步:技能总数 {prev_total} → {total}(合并 data/ 与 public/ 分叉)"]})
out["update_log"] = log[-20:]
return out
def readme_section(data: dict) -> str:
order = [c for c in CAT_ORDER_LEGACY if c in data["categories"]] + \
[c for c in data["categories"] if c not in CAT_ORDER_LEGACY]
lines = ["## 💾 本地技能库", "",
f"共 **{data['metadata']['total_skills']}** 个精选技能,按用途分类"
f"(最后同步 {data['metadata']['last_updated']},由 `scripts/sync_frontend.py` 自动生成本段)。", ""]
for cat in order:
skills = data["categories"][cat]["skills"]
lines.append(f"### {cat}({len(skills)}个)")
for s in skills:
d = data["skill_details"].get(s) or {}
desc = d.get("description_cn") or d.get("description_en") or ""
star = d.get("rating") or ""
suffix = f" - {desc}" if desc else ""
prefix = f"{star} " if star else ""
lines.append(f"- **{s}** {prefix}{suffix}".rstrip())
lines.append("")
return "\n".join(lines)
def apply_readme(text: str, section: str) -> str:
if MARK_START in text and MARK_END in text:
pre = text.split(MARK_START)[0]
post = text.split(MARK_END, 1)[1]
return f"{pre}{MARK_START}\n{section}\n{MARK_END}{post}"
# 无标记:替换整个「## 💾 本地技能库」段(到下一个 ## 标题为止),并补上标记
pat = re.compile(r"## 💾 本地技能库\n.*?(?=\n## )", re.S)
if pat.search(text):
return pat.sub(f"{MARK_START}\n{section}\n{MARK_END}", text, count=1)
# 兜底:追加到 AUTO-SYNC-SKILLS 段之后
return text.replace("<!-- AUTO-SYNC-SKILLS-END -->",
f"<!-- AUTO-SYNC-SKILLS-END -->\n\n{MARK_START}\n{section}\n{MARK_END}", 1)
def main():
dry = "--dry-run" in sys.argv
a = json.loads(DATA_A.read_text(encoding="utf-8"))
b = json.loads(DATA_B.read_text(encoding="utf-8"))
merged = merge(a, b)
payload = json.dumps(merged, ensure_ascii=False, indent=2)
section = readme_section(merged)
readme = README.read_text(encoding="utf-8")
new_readme = apply_readme(readme, section)
n_cats = len(merged["categories"])
total = merged["metadata"]["total_skills"]
print(f"合并结果: {n_cats} 个分类 / {total} 个技能")
for cat, c in merged["categories"].items():
print(f" - {cat}: {c['count']}")
changed_b = set()
for cat in merged["categories"]:
pass
print(f"README: {'将更新' if new_readme != readme else '无变化'}"
f"({len(new_readme.splitlines())} 行)")
if dry:
print("[dry-run] 未写入任何文件")
return 0
DATA_A.write_text(payload + "\n", encoding="utf-8")
DATA_B.write_text(payload + "\n", encoding="utf-8")
if new_readme != readme:
README.write_text(new_readme, encoding="utf-8")
print("✅ 已写入 data/local_skills.json、public/data/local_skills.json、README.md")
return 0
if __name__ == "__main__":
sys.exit(main())