项目初始化

bot/skills_service.py

@@ -0,0 +1,317 @@
+"""skills.sh 数据查询与命令解析。"""
+
+from __future__ import annotations
+
+import json
+import logging
+import re
+import time
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Any, Literal
+
+import certifi
+import httpx
+
+logger = logging.getLogger(__name__)
+
+FEED_URLS = [
+    # jsDelivr 在国内通常比 raw.githubusercontent.com 更稳定
+    "https://cdn.jsdelivr.net/gh/NeverSight/skills.sh_feed@main/data/feed.json",
+    "https://raw.githubusercontent.com/NeverSight/skills.sh_feed/main/data/feed.json",
+]
+CACHE_TTL_SECONDS = 600
+CACHE_DIR = Path(__file__).resolve().parent / ".cache"
+CACHE_FILE = CACHE_DIR / "feed.json"
+
+_cache: dict[str, Any] = {"data": None, "fetched_at": 0.0}
+
+Board = Literal["trending", "hot", "all"]
+
+
+@dataclass
+class Command:
+    kind: Literal["help", "list", "search", "detail"]
+    board: Board = "trending"
+    limit: int = 10
+    query: str = ""
+
+
+def _fetch_json(url: str) -> dict[str, Any]:
+    headers = {
+        "User-Agent": "skills-hot-bot/1.0",
+        "Accept": "application/json",
+    }
+    with httpx.Client(
+        timeout=httpx.Timeout(20.0, connect=10.0),
+        verify=certifi.where(),
+        follow_redirects=True,
+    ) as client:
+        resp = client.get(url, headers=headers)
+        resp.raise_for_status()
+        return resp.json()
+
+
+def _load_disk_cache() -> dict[str, Any] | None:
+    if not CACHE_FILE.exists():
+        return None
+    try:
+        return json.loads(CACHE_FILE.read_text(encoding="utf-8"))
+    except (OSError, json.JSONDecodeError) as exc:
+        logger.warning("读取本地缓存失败: %s", exc)
+        return None
+
+
+def _save_disk_cache(data: dict[str, Any]) -> None:
+    CACHE_DIR.mkdir(parents=True, exist_ok=True)
+    CACHE_FILE.write_text(json.dumps(data, ensure_ascii=False), encoding="utf-8")
+
+
+def load_feed(force: bool = False) -> dict[str, Any]:
+    now = time.time()
+    if not force and _cache["data"] and now - _cache["fetched_at"] < CACHE_TTL_SECONDS:
+        return _cache["data"]
+
+    errors: list[str] = []
+    for url in FEED_URLS:
+        for attempt in range(3):
+            try:
+                data = _fetch_json(url)
+                _cache["data"] = data
+                _cache["fetched_at"] = now
+                _save_disk_cache(data)
+                logger.info("skills 数据已更新: %s", url)
+                return data
+            except Exception as exc:
+                msg = f"{url} (#{attempt + 1}): {exc}"
+                errors.append(msg)
+                logger.debug("拉取失败 %s", msg)
+                time.sleep(0.5 * (attempt + 1))
+
+    stale = _load_disk_cache()
+    if stale:
+        logger.warning("网络不可用，回退到本地缓存")
+        _cache["data"] = stale
+        _cache["fetched_at"] = now
+        return stale
+
+    raise RuntimeError(f"无法获取 skills 数据。最近错误: {errors[-1]}")
+
+
+def warm_feed_cache() -> None:
+    """启动时预加载，避免首条消息才触发网络请求。"""
+    load_feed(force=True)
+
+
+def _normalize_text(text: str) -> str:
+    text = re.sub(r"@\S+\s*", "", text)
+    return text.strip().lower()
+
+
+def _parse_limit(raw: str | None, default: int = 10) -> int:
+    if not raw:
+        return default
+    try:
+        n = int(raw)
+    except ValueError:
+        return default
+    return max(1, min(n, 30))
+
+
+def _match_list(raw: str, board: Board, aliases: str) -> Command | None:
+    m = re.match(rf"^({aliases})(?:\s+top)?\s*(\d+)?$", raw)
+    if m:
+        return Command(kind="list", board=board, limit=_parse_limit(m.group(2)))
+    m = re.match(rf"^(查|查询)\s+({aliases})(?:\s+top)?\s*(\d+)?$", raw)
+    if m:
+        return Command(kind="list", board=board, limit=_parse_limit(m.group(3)))
+    return None
+
+
+def parse_command(text: str) -> Command:
+    raw = _normalize_text(text)
+    if not raw or raw in {"help", "帮助", "?", "h"}:
+        return Command(kind="help")
+
+    for board, aliases in (
+        ("trending", "trending|趋势|top"),
+        ("hot", "hot|实时|热门"),
+        ("all", "all|总榜|alltime|all-time"),
+    ):
+        cmd = _match_list(raw, board, aliases)
+        if cmd:
+            return cmd
+
+    m = re.match(r"^(search|搜索|find|查)\s+(.+)$", raw)
+    if m:
+        return Command(kind="search", query=m.group(2).strip(), limit=5)
+
+    m = re.match(r"^(detail|详情|skill|info)\s+(.+)$", raw)
+    if m:
+        return Command(kind="detail", query=m.group(2).strip())
+
+    if raw.startswith("trending") or raw.startswith("趋势"):
+        parts = raw.split(maxsplit=1)
+        return Command(kind="list", board="trending", limit=_parse_limit(parts[1] if len(parts) > 1 else None))
+
+    return Command(kind="search", query=raw, limit=5)
+
+
+def _format_installs(n: int | float) -> str:
+    if n >= 1_000_000:
+        return f"{n / 1_000_000:.1f}M"
+    if n >= 1_000:
+        return f"{n / 1_000:.1f}K"
+    return str(int(n))
+
+
+def _board_items(feed: dict[str, Any], board: Board) -> list[dict[str, Any]]:
+    key = {"trending": "topTrending", "hot": "topHot", "all": "topAllTime"}[board]
+    return feed.get(key, [])
+
+
+def _board_title(board: Board) -> str:
+    return {
+        "trending": "Trending（近期增长）",
+        "hot": "Hot（实时热度）",
+        "all": "All Time（总安装榜）",
+    }[board]
+
+
+def format_list(board: Board, limit: int) -> str:
+    feed = load_feed()
+    items = _board_items(feed, board)[:limit]
+    updated = feed.get("updatedAt", "未知")[:10]
+
+    lines = [
+        f"**skills.sh {_board_title(board)} Top {limit}**",
+        f"> 数据更新：{updated}",
+        "",
+    ]
+
+    for i, item in enumerate(items, 1):
+        title = item.get("title", "?")
+        source = item.get("source", "?")
+        installs = _format_installs(item.get("installs", 0))
+        desc = item.get("description", "")
+        if len(desc) > 80:
+            desc = desc[:77] + "..."
+        link = item.get("link", "")
+        lines.append(f"{i}. **{title}** · {installs}")
+        lines.append(f"   `{source}`")
+        if desc:
+            lines.append(f"   {desc}")
+        if link:
+            lines.append(f"   [查看]({link})")
+        lines.append("")
+
+    return "\n".join(lines).strip()
+
+
+def format_search(query: str, limit: int) -> str:
+    feed = load_feed()
+    q = query.lower()
+    seen: set[str] = set()
+    matches: list[dict[str, Any]] = []
+
+    for board in ("topTrending", "topHot", "topAllTime"):
+        for item in feed.get(board, []):
+            item_id = item.get("id") or item.get("title", "")
+            if item_id in seen:
+                continue
+            haystack = " ".join(
+                [
+                    item.get("title", ""),
+                    item.get("source", ""),
+                    item.get("description", ""),
+                ]
+            ).lower()
+            if q in haystack:
+                seen.add(item_id)
+                matches.append(item)
+            if len(matches) >= limit:
+                break
+        if len(matches) >= limit:
+            break
+
+    if not matches:
+        return f"未找到与 **{query}** 相关的 skill。\n\n试试：`trending 10` / `hot 10` / `搜索 react`"
+
+    lines = [f"**搜索「{query}」** 共 {len(matches)} 条", ""]
+    for i, item in enumerate(matches, 1):
+        title = item.get("title", "?")
+        source = item.get("source", "?")
+        installs = _format_installs(item.get("installs", 0))
+        link = item.get("link", "")
+        lines.append(f"{i}. **{title}** · {installs} · `{source}`")
+        if link:
+            lines.append(f"   [查看]({link})")
+    return "\n".join(lines)
+
+
+def format_detail(name: str) -> str:
+    feed = load_feed()
+    q = name.lower().strip()
+    best: dict[str, Any] | None = None
+
+    for board in ("topTrending", "topHot", "topAllTime"):
+        for item in feed.get(board, []):
+            title = (item.get("title") or "").lower()
+            item_id = (item.get("id") or "").lower()
+            if title == q or q in title or q in item_id:
+                if best is None or item.get("installs", 0) > best.get("installs", 0):
+                    best = item
+
+    if not best:
+        return f"未找到 skill：**{name}**\n\n试试：`搜索 {name}`"
+
+    desc = best.get("description", "无描述")
+    return "\n".join(
+        [
+            f"**{best.get('title', '?')}**",
+            f"`{best.get('source', '?')}`",
+            f"安装量：**{_format_installs(best.get('installs', 0))}**",
+            "",
+            desc,
+            "",
+            f"[skills.sh 详情]({best.get('link', 'https://skills.sh')})",
+            "",
+            f"安装：`npx skills add {best.get('source', '')}/{best.get('title', '')}`",
+        ]
+    )
+
+
+def format_help() -> str:
+    return "\n".join(
+        [
+            "**Skills 助手 · 命令帮助**",
+            "",
+            "`trending 10` / `趋势 10` — 近期增长榜",
+            "`hot 10` / `实时 10` — 实时热度榜",
+            "`all 10` / `总榜 10` — 历史总安装榜",
+            "`搜索 react` / `search tdd` — 关键词搜索",
+            "`详情 find-skills` — 查看单个 skill",
+            "`preview` / `截图` / `预览` — 单页截图",
+            "`browser 场景名` — 执行 YAML 场景（见 bot/scenarios/）",
+            "自然语言 — 如：访问登录页，输入账号密码，点击登录，点击智能体管理，截图",
+            "`preview /about 5173` — 指定路径和端口",
+            "",
+            "示例：",
+            "• trending top10",
+            "• 查 grill",
+            "• 详情 remotion-render",
+        ]
+    )
+
+
+def handle_command(text: str) -> str:
+    cmd = parse_command(text)
+    if cmd.kind == "help":
+        return format_help()
+    if cmd.kind == "list":
+        return format_list(cmd.board, cmd.limit)
+    if cmd.kind == "search":
+        return format_search(cmd.query, cmd.limit)
+    if cmd.kind == "detail":
+        return format_detail(cmd.query)
+    return format_help()