refactor: 统一所有渠道后端，对齐 research 技能

GitHub: REST API → gh CLI（官方工具，认证后完整能力） Bilibili: 自写 API → yt-dlp（和 YouTube 统一后端，支持搜索 bilisearch） YouTube: 新增搜索功能（ytsearch via yt-dlp） README 中英文同步更新： - 平台表格：小红书/Exa/GitHub/YouTube/B站描述全部更新 - 选型表格：新增 gh CLI、xiaohongshu-mcp，更新 yt-dlp/Exa 描述 - 按需解锁：去掉 Exa Key 注册步骤（已自动配置） - 配置难度说明：新增「自动配置」「mcporter」级别
2026-02-24 13:25:57 +01:00
parent bb9ce6f7f1
commit 88eaebe263
5 changed files with 298 additions and 230 deletions
@@ -1,11 +1,13 @@
 # -*- coding: utf-8 -*-
-"""GitHub — via GitHub REST API (free, no config needed).
+"""GitHub — via gh CLI.

-Backend: GitHub API v3
-Swap to: gh CLI, or any GitHub API wrapper
+Backend: gh CLI (https://cli.github.com)
+Swap to: GitHub REST API
 """

-import requests
+import json
+import shutil
+import subprocess
 from urllib.parse import urlparse
 from .base import Channel, ReadResult, SearchResult
 from typing import List
@@ -14,115 +16,124 @@ from typing import List
 class GitHubChannel(Channel):
    name = "github"
    description = "GitHub 仓库和代码"
-    backends = ["GitHub API"]
+    backends = ["gh CLI"]
    tier = 0

-    API = "https://api.github.com"
+    def _gh(self, args: list, timeout: int = 15) -> str:
+        r = subprocess.run(
+            ["gh"] + args,
+            capture_output=True, text=True, timeout=timeout,
+        )
+        if r.returncode != 0:
+            raise RuntimeError(r.stderr or r.stdout)
+        return r.stdout

-    def _headers(self, config=None):
-        h = {"Accept": "application/vnd.github+json"}
-        token = config.get("github_token") if config else None
-        if token:
-            h["Authorization"] = f"Bearer {token}"
-        return h
-
-    def check(self, config=None):
-        import shutil
-        token = config.get("github_token") if config else None
-        has_gh = shutil.which("gh")
-        if token or has_gh:
-            return "ok", "完整可用（读取、搜索、Fork、Issue、PR 等）"
-        return "ok", "公开仓库可读可搜。配置 gh CLI 或 github_token 可解锁 Fork、Issue、PR 等操作"
+    def _gh_json(self, args: list, timeout: int = 15) -> dict:
+        return json.loads(self._gh(args + ["--json"], timeout))

    def can_handle(self, url: str) -> bool:
-        domain = urlparse(url).netloc.lower()
-        return "github.com" in domain
+        return "github.com" in urlparse(url).netloc.lower()
+
+    def check(self, config=None):
+        if not shutil.which("gh"):
+            return "warn", "gh CLI 未安装。安装：https://cli.github.com 。公开仓库仍可通过 Jina Reader 读取"
+        try:
+            self._gh(["auth", "status"], timeout=5)
+            return "ok", "完整可用（读取、搜索、Fork、Issue、PR 等）"
+        except Exception:
+            return "ok", "gh CLI 已装但未认证。运行 gh auth login 可解锁完整功能"

    async def read(self, url: str, config=None) -> ReadResult:
-        path = urlparse(url).path.strip("/").split("/")
+        if not shutil.which("gh"):
+            # Fallback to Jina Reader for public repos
+            from agent_reach.channels.web import WebChannel
+            return await WebChannel().read(url, config)

+        path = urlparse(url).path.strip("/").split("/")
        if len(path) < 2:
-            raise ValueError(f"Invalid GitHub URL: {url}")
+            from agent_reach.channels.web import WebChannel
+            return await WebChannel().read(url, config)

        owner, repo = path[0], path[1]
-        headers = self._headers(config)

-        # Issues/PRs
+        # Issues / PRs
        if len(path) >= 4 and path[2] in ("issues", "pull"):
-            num = path[3]
-            resp = requests.get(f"{self.API}/repos/{owner}/{repo}/issues/{num}", headers=headers, timeout=15)
-            resp.raise_for_status()
-            data = resp.json()
-
-            # Get comments
-            comments_text = ""
-            if data.get("comments", 0) > 0:
-                cr = requests.get(f"{self.API}/repos/{owner}/{repo}/issues/{num}/comments",
-                                  headers=headers, params={"per_page": 20}, timeout=15)
-                if cr.ok:
-                    for c in cr.json():
-                        comments_text += f"\n\n---\n**{c.get('user', {}).get('login', '')}** ({c.get('created_at', '')}):\n{c.get('body', '')}"
-
-            return ReadResult(
-                title=data.get("title", ""),
-                content=(data.get("body", "") or "") + comments_text,
-                url=url,
-                author=data.get("user", {}).get("login", ""),
-                date=data.get("created_at", ""),
-                platform="github",
-                extra={"state": data.get("state"), "comments": data.get("comments", 0),
-                       "reactions": data.get("reactions", {}).get("total_count", 0)},
-            )
+            return await self._read_issue(owner, repo, path[3], url)

        # Repo
-        resp = requests.get(f"{self.API}/repos/{owner}/{repo}", headers=headers, timeout=15)
-        resp.raise_for_status()
-        data = resp.json()
+        return await self._read_repo(owner, repo, url)

-        # Get README
-        readme_text = ""
-        rr = requests.get(f"{self.API}/repos/{owner}/{repo}/readme", headers=headers, timeout=15)
-        if rr.ok:
-            import base64
-            readme_data = rr.json()
-            if readme_data.get("encoding") == "base64":
-                readme_text = base64.b64decode(readme_data["content"]).decode("utf-8", errors="replace")
+    async def _read_repo(self, owner: str, repo: str, url: str) -> ReadResult:
+        slug = f"{owner}/{repo}"
+        try:
+            # Get repo info
+            info = self._gh(["repo", "view", slug])
+            # Get README
+            try:
+                readme = self._gh(
+                    ["api", f"repos/{slug}/readme", "--jq", ".content"],
+                    timeout=10,
+                )
+                import base64
+                readme_text = base64.b64decode(readme).decode("utf-8", errors="replace")
+            except Exception:
+                readme_text = ""

-        return ReadResult(
-            title=f"{owner}/{repo}",
-            content=readme_text or data.get("description", ""),
-            url=url,
-            author=owner,
-            platform="github",
-            extra={"stars": data.get("stargazers_count", 0), "forks": data.get("forks_count", 0),
-                   "language": data.get("language", ""), "description": data.get("description", "")},
-        )
+            content = readme_text or info
+            return ReadResult(
+                title=slug, content=content, url=url,
+                author=owner, platform="github",
+            )
+        except Exception:
+            from agent_reach.channels.web import WebChannel
+            return await WebChannel().read(url)
+
+    async def _read_issue(self, owner: str, repo: str, num: str, url: str) -> ReadResult:
+        slug = f"{owner}/{repo}"
+        try:
+            out = self._gh(["issue", "view", num, "-R", slug])
+            return ReadResult(
+                title=f"{slug}#{num}", content=out, url=url,
+                platform="github",
+            )
+        except Exception:
+            # Might be a PR
+            try:
+                out = self._gh(["pr", "view", num, "-R", slug])
+                return ReadResult(
+                    title=f"{slug}#{num}", content=out, url=url,
+                    platform="github",
+                )
+            except Exception:
+                from agent_reach.channels.web import WebChannel
+                return await WebChannel().read(url)

    async def search(self, query: str, config=None, **kwargs) -> List[SearchResult]:
+        if not shutil.which("gh"):
+            raise ValueError("GitHub search requires gh CLI. Install: https://cli.github.com")
+
        language = kwargs.get("language")
        limit = kwargs.get("limit", 5)

-        q = query
+        args = ["search", "repos", query, "--sort", "stars", f"--limit={limit}"]
        if language:
-            q += f" language:{language}"
-
-        resp = requests.get(
-            f"{self.API}/search/repositories",
-            headers=self._headers(config),
-            params={"q": q, "sort": "stars", "per_page": min(limit, 30)},
-            timeout=15,
-        )
-        resp.raise_for_status()
+            args += [f"--language={language}"]

+        out = self._gh(args, timeout=15)
        results = []
-        for repo in resp.json().get("items", []):
-            results.append(SearchResult(
-                title=repo.get("full_name", ""),
-                url=repo.get("html_url", ""),
-                snippet=repo.get("description", ""),
-                date=repo.get("updated_at", ""),
-                extra={"stars": repo.get("stargazers_count", 0),
-                       "forks": repo.get("forks_count", 0),
-                       "language": repo.get("language", "")},
-            ))
+        for line in out.strip().split("\n"):
+            if not line.strip():
+                continue
+            parts = line.split("\t")
+            if len(parts) >= 1:
+                slug = parts[0].strip()
+                desc = parts[1].strip() if len(parts) > 1 else ""
+                stars = parts[3].strip() if len(parts) > 3 else ""
+                lang = parts[5].strip() if len(parts) > 5 else ""
+                results.append(SearchResult(
+                    title=slug,
+                    url=f"https://github.com/{slug}",
+                    snippet=desc,
+                    extra={"stars": stars, "language": lang},
+                ))
        return results