@kernel-sig/console 0.1.0

package/backend/app/domain/sig_info.py

@@ -0,0 +1,201 @@
+"""``sig-info.yaml`` 解析。
+
+这是 openEuler 各 SIG 的**权威元数据文件**，机器可读，位于
+``openeuler/community`` 的 ``sig/Kernel/sig-info.yaml``。成员名单、
+仓库清单、分支负责人都在这里，由社区工具链维护。
+
+此前用的是 ``README.md`` 的 Maintainer 列表加 ``committers.md`` 的模块表，
+两者都是给人看的文档：README 只有 4 位维护者的展示名与 gitee handle，
+committers.md 是按内核路径划分的分工表。合起来只有 101 人，且**没有邮箱、
+没有组织、没有 atomgit 账号**（实测 14% 的人只有 gitee 账号）。
+``sig-info.yaml`` 有 133 人且字段完整。
+
+两份材料不冲突，但用途不同，因此都保留：
+- ``sig-info.yaml`` → 成员是谁、管哪些仓库、负责哪些分支
+- ``committers.md`` → 改动了哪个内核路径该找谁（yaml 里没有路径信息）
+"""
+
+from __future__ import annotations
+
+from dataclasses import dataclass, field
+
+import yaml
+
+# 角色优先级。一个人可以同时是维护者与某仓库的 committer，
+# 取最高的那个作为主角色，全部角色另存。
+ROLE_RANK = {"maintainer": 3, "committer": 2, "contributor": 1}
+
+
+@dataclass
+class RosterEntry:
+    """名单里的一个人。"""
+
+    name: str
+    atomgit_id: str | None = None
+    gitee_id: str | None = None
+    email: str | None = None
+    organization: str | None = None
+
+    role: str = "contributor"
+    """最高角色。"""
+
+    roles: list[str] = field(default_factory=list)
+    repositories: list[str] = field(default_factory=list)
+    """作为 committer / contributor 参与的仓库。"""
+
+    admin_repositories: list[str] = field(default_factory=list)
+    kept_branches: list[str] = field(default_factory=list)
+    """作为 keeper 负责的分支。"""
+
+    @property
+    def key(self) -> str:
+        """去重键。优先用 atomgit 账号 —— 那是在本平台能匹配到 PR 作者的字段。"""
+        return self.atomgit_id or self.gitee_id or self.name
+
+
+@dataclass
+class BranchKeeper:
+    repo: str
+    branch: str
+    keepers: list[str] = field(default_factory=list)
+    """keeper 的 atomgit/gitee 账号。yaml 里只有 id，姓名要回名单里查。"""
+
+
+@dataclass
+class SIGInfo:
+    name: str
+    description: str | None = None
+    mailing_list: str | None = None
+    roster: list[RosterEntry] = field(default_factory=list)
+    repositories: list[str] = field(default_factory=list)
+    branch_keepers: list[BranchKeeper] = field(default_factory=list)
+
+
+def _person(raw: object) -> RosterEntry | None:
+    if not isinstance(raw, dict):
+        return None
+    name = str(raw.get("name") or "").strip()
+    atomgit = str(raw.get("atomgit_id") or "").strip() or None
+    gitee = str(raw.get("gitee_id") or "").strip() or None
+    if not name and not atomgit and not gitee:
+        return None
+    return RosterEntry(
+        name=name or (atomgit or gitee or ""),
+        atomgit_id=atomgit,
+        gitee_id=gitee,
+        email=str(raw.get("email") or "").strip() or None,
+        organization=str(raw.get("organization") or "").strip() or None,
+    )
+
+
+def _as_list(value: object) -> list:
+    if value is None:
+        return []
+    return value if isinstance(value, list) else [value]
+
+
+def parse_sig_info(text: str) -> SIGInfo | None:
+    """解析 sig-info.yaml。结构不合预期时返回 None，不抛异常。
+
+    yaml 是上游维护的，字段增删由他们决定；解析失败只该让这轮同步跳过，
+    不该让整个名单同步任务报红 —— 名单是低频数据，下轮再试即可。
+    """
+    try:
+        payload = yaml.safe_load(text)
+    except yaml.YAMLError:
+        return None
+    if not isinstance(payload, dict):
+        return None
+
+    info = SIGInfo(
+        name=str(payload.get("name") or "Kernel"),
+        description=str(payload.get("description") or "") or None,
+        mailing_list=str(payload.get("mailing_list") or "") or None,
+    )
+
+    entries: dict[str, RosterEntry] = {}
+
+    def merge(entry: RosterEntry | None, role: str) -> RosterEntry | None:
+        if entry is None:
+            return None
+        existing = entries.get(entry.key)
+        if existing is None:
+            entries[entry.key] = entry
+            existing = entry
+        # 补全字段：同一个人在不同段落出现时可能只写了 id
+        for attr in ("name", "atomgit_id", "gitee_id", "email", "organization"):
+            if not getattr(existing, attr) and getattr(entry, attr):
+                setattr(existing, attr, getattr(entry, attr))
+        if role not in existing.roles:
+            existing.roles.append(role)
+        if ROLE_RANK.get(role, 0) > ROLE_RANK.get(existing.role, 0):
+            existing.role = role
+        return existing
+
+    for raw in _as_list(payload.get("maintainers")):
+        merge(_person(raw), "maintainer")
+
+    for repo_entry in _as_list(payload.get("repositories")):
+        if not isinstance(repo_entry, dict):
+            continue
+        repo = repo_entry.get("repo")
+        # repo 是仓库名列表（同一份配置同时管源码仓与构建仓）
+        repos = [str(item) for item in _as_list(repo) if isinstance(item, str)]
+        info.repositories.extend(repos)
+
+        for raw in _as_list(repo_entry.get("repo_admin")):
+            entry = merge(_person(raw), "committer")
+            if entry is not None:
+                entry.admin_repositories.extend(repos)
+        for raw in _as_list(repo_entry.get("committers")):
+            entry = merge(_person(raw), "committer")
+            if entry is not None:
+                entry.repositories.extend(repos)
+        for raw in _as_list(repo_entry.get("contributors")):
+            entry = merge(_person(raw), "contributor")
+            if entry is not None:
+                entry.repositories.extend(repos)
+
+    for raw_branch in _as_list(payload.get("branches")):
+        if not isinstance(raw_branch, dict):
+            continue
+        # keeper 只有 id，没有姓名与邮箱 —— 姓名要回名单里查，
+        # 因此这里先存账号，解析结束后统一补名
+        keeper_logins = []
+        for raw in _as_list(raw_branch.get("keeper")):
+            person = _person(raw)
+            if person is not None:
+                keeper_logins.append(person.key)
+        for target in _as_list(raw_branch.get("repo_branch")):
+            if not isinstance(target, dict):
+                continue
+            branch = str(target.get("branch") or "").strip()
+            repo = str(target.get("repo") or "").strip()
+            if branch:
+                info.branch_keepers.append(
+                    BranchKeeper(repo=repo, branch=branch, keepers=list(keeper_logins))
+                )
+
+    # keeper 若不在 committer 名单里，也要作为成员出现 —— 能管分支的人
+    # 不列在成员页上会让人以为名单漏了
+    for branch in info.branch_keepers:
+        for login in branch.keepers:
+            if login not in entries:
+                entries[login] = RosterEntry(name=login, role="committer", roles=["committer"])
+                if login.startswith("http") or "." in login:
+                    entries[login].gitee_id = login
+                else:
+                    entries[login].atomgit_id = login
+
+    for entry in entries.values():
+        for branch in info.branch_keepers:
+            if entry.key in branch.keepers and branch.branch not in entry.kept_branches:
+                entry.kept_branches.append(branch.branch)
+        entry.repositories = sorted(set(entry.repositories))
+        entry.admin_repositories = sorted(set(entry.admin_repositories))
+
+    info.roster = sorted(
+        entries.values(), key=lambda item: (-ROLE_RANK.get(item.role, 0), item.name)
+    )
+    info.repositories = sorted(set(info.repositories))
+    return info