@kernel-sig/console 0.1.0

package/backend/app/models/classification.py

@@ -0,0 +1,144 @@
+"""分类结果。
+
+分类走两级：规则引擎优先（确定性、零成本、可解释），规则未命中或置信度低
+时才调用 AI。人工覆盖过的结果不会被自动任务覆盖回去 ——
+这是对使用者预期的基本尊重。
+"""
+
+import uuid
+from datetime import datetime
+from enum import StrEnum
+
+from sqlalchemy import (
+    BigInteger,
+    Boolean,
+    DateTime,
+    Enum,
+    Float,
+    ForeignKey,
+    Index,
+    String,
+    Text,
+    UniqueConstraint,
+    func,
+)
+from sqlalchemy.dialects.postgresql import ARRAY, UUID
+from sqlalchemy.orm import Mapped, mapped_column
+
+from app.models.base import Base, TimestampMixin
+
+
+class SubjectType(StrEnum):
+    PULL_REQUEST = "pull_request"
+    ISSUE = "issue"
+
+
+class PRKind(StrEnum):
+    """PR/Issue 的内容类别。
+
+    取值与 ``app.domain.classification.Kind`` 一一对应，两侧必须一致 ——
+    规则算出的值写不进数据库的表现是"分类静默失败"。
+    """
+
+    CVE = "cve"
+    RELEASE = "release"
+    BACKPORT = "backport"
+    DRIVER_NEW = "driver_new"
+    DRIVER_UPDATE = "driver_update"
+    SOC_SUPPORT = "soc_support"
+    OUT_OF_TREE = "out_of_tree"
+    BUGFIX = "bugfix"
+    FEATURE = "feature"
+    PERF = "perf"
+    REFACTOR = "refactor"
+    DOC = "docs"
+    CLEANUP = "cleanup"
+    UNKNOWN = "unknown"
+
+
+class ClassificationSource(StrEnum):
+    RULE = "rule"
+    AI = "ai"
+    MANUAL = "manual"
+
+
+class Classification(Base, TimestampMixin):
+    __tablename__ = "ksp_classification"
+    __table_args__ = (
+        UniqueConstraint("subject_type", "subject_id", name="uq_ksp_classification_subject"),
+        Index("ix_ksp_classification_kind", "kind"),
+    )
+
+    id: Mapped[uuid.UUID] = mapped_column(UUID(as_uuid=True), primary_key=True, default=uuid.uuid4)
+    subject_type: Mapped[SubjectType] = mapped_column(
+        Enum(
+            SubjectType,
+            name="classification_subject_type",
+            values_callable=lambda e: [m.value for m in e],
+        ),
+        nullable=False,
+    )
+    subject_id: Mapped[uuid.UUID] = mapped_column(UUID(as_uuid=True), nullable=False, index=True)
+    repository_id: Mapped[uuid.UUID] = mapped_column(
+        UUID(as_uuid=True),
+        ForeignKey("ksp_repository.id", ondelete="CASCADE"),
+        nullable=False,
+        index=True,
+    )
+
+    kind: Mapped[PRKind] = mapped_column(
+        Enum(PRKind, name="pr_kind", values_callable=lambda e: [m.value for m in e]),
+        nullable=False,
+        default=PRKind.UNKNOWN,
+    )
+    # 从变更文件路径推导的子系统，如 net / mm / arm64
+    subsystem: Mapped[str | None] = mapped_column(String(64), nullable=True, index=True)
+    related_subsystems: Mapped[list[str] | None] = mapped_column(ARRAY(Text), nullable=True)
+
+    cve_ids: Mapped[list[str] | None] = mapped_column(ARRAY(Text), nullable=True, index=True)
+    # 从正文 fixes #N / resolves #N 解析出的关联 Issue 编号
+    linked_issues: Mapped[list[int] | None] = mapped_column(ARRAY(BigInteger), nullable=True)
+
+    confidence: Mapped[float] = mapped_column(Float, nullable=False, default=1.0)
+    source: Mapped[ClassificationSource] = mapped_column(
+        Enum(
+            ClassificationSource,
+            name="classification_source",
+            values_callable=lambda e: [m.value for m in e],
+        ),
+        nullable=False,
+        default=ClassificationSource.RULE,
+    )
+    # 命中的规则名，便于排查"为什么被分到这一类"
+    rule_name: Mapped[str | None] = mapped_column(String(64), nullable=True)
+    reason: Mapped[str | None] = mapped_column(Text, nullable=True)
+
+    # 人工覆盖后置位，自动任务不再改写
+    is_override: Mapped[bool] = mapped_column(Boolean, nullable=False, default=False)
+    overridden_by: Mapped[uuid.UUID | None] = mapped_column(
+        UUID(as_uuid=True), ForeignKey("ksp_user.id", ondelete="SET NULL"), nullable=True
+    )
+    overridden_at: Mapped[datetime | None] = mapped_column(DateTime(timezone=True), nullable=True)
+
+    def __repr__(self) -> str:
+        return f"<Classification {self.subject_type.value}:{self.kind.value}>"
+
+
+class LabelSyncState(Base):
+    """记录已建到 AtomGit 的标签，避免重复创建与重复写入。"""
+
+    __tablename__ = "ksp_label_sync_state"
+    __table_args__ = (
+        UniqueConstraint("repository_id", "label_name", name="uq_ksp_label_sync_repo_name"),
+    )
+
+    id: Mapped[int] = mapped_column(BigInteger, primary_key=True, autoincrement=True)
+    repository_id: Mapped[uuid.UUID] = mapped_column(
+        UUID(as_uuid=True),
+        ForeignKey("ksp_repository.id", ondelete="CASCADE"),
+        nullable=False,
+    )
+    label_name: Mapped[str] = mapped_column(String(128), nullable=False)
+    created_at: Mapped[datetime] = mapped_column(
+        DateTime(timezone=True), server_default=func.now(), nullable=False
+    )

package/backend/app/models/credential.py

@@ -0,0 +1,56 @@
+"""外部凭据（AtomGit token、LLM API Key 等）。
+
+安全约束：明文只以 AES-256-GCM 密文形态落库，密钥由 KSC_SECRET_KEY
+经 HKDF 派生。界面永不回显明文，只展示指纹供人工比对。
+"""
+
+import uuid
+from datetime import datetime
+from enum import StrEnum
+
+from sqlalchemy import DateTime, Enum, ForeignKey, LargeBinary, String
+from sqlalchemy.dialects.postgresql import JSONB, UUID
+from sqlalchemy.orm import Mapped, mapped_column
+
+from app.models.base import Base, TimestampMixin
+
+
+class CredentialKind(StrEnum):
+    ATOMGIT_TOKEN = "atomgit_token"
+    LLM_API_KEY = "llm_api_key"
+
+
+class Credential(Base, TimestampMixin):
+    __tablename__ = "ksp_credential"
+
+    id: Mapped[uuid.UUID] = mapped_column(UUID(as_uuid=True), primary_key=True, default=uuid.uuid4)
+    kind: Mapped[CredentialKind] = mapped_column(
+        Enum(
+            CredentialKind,
+            name="credential_kind",
+            values_callable=lambda e: [m.value for m in e],
+        ),
+        nullable=False,
+        index=True,
+    )
+    name: Mapped[str] = mapped_column(String(128), nullable=False, unique=True)
+
+    # 密文（AES-256-GCM，格式 v1:<nonce>:<ciphertext>）
+    encrypted_value: Mapped[bytes] = mapped_column(LargeBinary, nullable=False)
+    # SHA-256 前 16 字符，用于界面展示"已配置哪一份"而无需回显明文
+    fingerprint: Mapped[str] = mapped_column(String(32), nullable=False)
+
+    # 非敏感附加信息，如 token 对应的 AtomGit 账号名
+    meta: Mapped[dict | None] = mapped_column(JSONB, nullable=True)
+
+    last_verified_at: Mapped[datetime | None] = mapped_column(
+        DateTime(timezone=True), nullable=True
+    )
+    last_verify_error: Mapped[str | None] = mapped_column(String(500), nullable=True)
+
+    created_by: Mapped[uuid.UUID | None] = mapped_column(
+        UUID(as_uuid=True), ForeignKey("ksp_user.id", ondelete="SET NULL"), nullable=True
+    )
+
+    def __repr__(self) -> str:
+        return f"<Credential {self.name} kind={self.kind.value} fp={self.fingerprint}>"

package/backend/app/models/issue.py

@@ -0,0 +1,132 @@
+"""Issue 及其关联数据。"""
+
+import uuid
+from datetime import datetime
+from enum import StrEnum
+
+from sqlalchemy import (
+    BigInteger,
+    DateTime,
+    Enum,
+    ForeignKey,
+    Index,
+    Integer,
+    String,
+    Text,
+    UniqueConstraint,
+    func,
+)
+from sqlalchemy.dialects.postgresql import ARRAY, JSONB, UUID
+from sqlalchemy.orm import Mapped, mapped_column
+
+from app.models.base import Base, TimestampMixin
+
+
+class IssueType(StrEnum):
+    """AtomGit 原生提供 issue_type，无需从标签推断。"""
+
+    BUG = "内核缺陷"
+    FEATURE = "内核需求"
+    TASK = "任务"
+    UNKNOWN = "未知"
+
+
+class IssueLinkSource(StrEnum):
+    PR_BODY = "pr_body"
+    ISSUE_BODY = "issue_body"
+    ATOMGIT_LINK = "atomgit_link"
+    MANUAL = "manual"
+
+
+class Issue(Base, TimestampMixin):
+    __tablename__ = "ksp_issue"
+    __table_args__ = (
+        UniqueConstraint("repository_id", "number", name="uq_ksp_issue_repo_number"),
+        Index("ix_ksp_issue_label_names", "label_names", postgresql_using="gin"),
+    )
+
+    id: Mapped[uuid.UUID] = mapped_column(UUID(as_uuid=True), primary_key=True, default=uuid.uuid4)
+    repository_id: Mapped[uuid.UUID] = mapped_column(
+        UUID(as_uuid=True),
+        ForeignKey("ksp_repository.id", ondelete="CASCADE"),
+        nullable=False,
+        index=True,
+    )
+    number: Mapped[int] = mapped_column(Integer, nullable=False)
+
+    title: Mapped[str] = mapped_column(Text, nullable=False)
+    body: Mapped[str | None] = mapped_column(Text, nullable=True)
+    state: Mapped[str] = mapped_column(String(32), nullable=False, index=True)
+
+    # AtomGit 的结构化字段，比用标签猜测更可靠
+    issue_type: Mapped[str | None] = mapped_column(String(64), nullable=True, index=True)
+    issue_state: Mapped[str | None] = mapped_column(String(64), nullable=True)
+    priority: Mapped[int | None] = mapped_column(Integer, nullable=True)
+    priority_label: Mapped[str | None] = mapped_column(String(64), nullable=True)
+
+    author_login: Mapped[str | None] = mapped_column(String(128), nullable=True, index=True)
+    assignee_login: Mapped[str | None] = mapped_column(String(128), nullable=True, index=True)
+
+    labels: Mapped[list | None] = mapped_column(JSONB, nullable=True)
+    label_names: Mapped[list[str] | None] = mapped_column(ARRAY(Text), nullable=True)
+
+    comments_count: Mapped[int] = mapped_column(Integer, nullable=False, default=0)
+
+    atomgit_created_at: Mapped[datetime | None] = mapped_column(
+        DateTime(timezone=True), nullable=True
+    )
+    atomgit_updated_at: Mapped[datetime | None] = mapped_column(
+        DateTime(timezone=True), nullable=True
+    )
+    finished_at: Mapped[datetime | None] = mapped_column(DateTime(timezone=True), nullable=True)
+    html_url: Mapped[str | None] = mapped_column(String(512), nullable=True)
+
+    first_seen_at: Mapped[datetime] = mapped_column(
+        DateTime(timezone=True), server_default=func.now(), nullable=False
+    )
+    last_synced_at: Mapped[datetime | None] = mapped_column(DateTime(timezone=True), nullable=True)
+    content_hash: Mapped[str | None] = mapped_column(String(64), nullable=True)
+
+    def __repr__(self) -> str:
+        return f"<Issue #{self.number} {self.issue_type}>"
+
+
+class PRIssueLink(Base):
+    """PR ↔ Issue 关联。
+
+    优先使用 AtomGit 的显式关联接口；正文正则匹配作为补充，
+    因此用 confidence 区分可信度。
+    """
+
+    __tablename__ = "ksp_pr_issue_link"
+    __table_args__ = (
+        UniqueConstraint("pull_request_id", "issue_number", name="uq_ksp_pr_issue_link"),
+    )
+
+    id: Mapped[int] = mapped_column(BigInteger, primary_key=True, autoincrement=True)
+    pull_request_id: Mapped[uuid.UUID] = mapped_column(
+        UUID(as_uuid=True),
+        ForeignKey("ksp_pull_request.id", ondelete="CASCADE"),
+        nullable=False,
+        index=True,
+    )
+    repository_id: Mapped[uuid.UUID] = mapped_column(
+        UUID(as_uuid=True),
+        ForeignKey("ksp_repository.id", ondelete="CASCADE"),
+        nullable=False,
+    )
+    issue_number: Mapped[int] = mapped_column(Integer, nullable=False, index=True)
+
+    source: Mapped[IssueLinkSource] = mapped_column(
+        Enum(
+            IssueLinkSource,
+            name="issue_link_source",
+            values_callable=lambda e: [m.value for m in e],
+        ),
+        nullable=False,
+    )
+    # 显式关联为 1.0，正文正则匹配略低，便于按可信度筛选
+    confidence: Mapped[float] = mapped_column(nullable=False, default=1.0)
+    created_at: Mapped[datetime] = mapped_column(
+        DateTime(timezone=True), server_default=func.now(), nullable=False
+    )

package/backend/app/models/meeting.py

@@ -0,0 +1,189 @@
+"""SIG 例会及其版本进展报告。
+
+数据源是 Etherpad 上的活文档，按时间倒序累积。这不是一个"同步一次就完"
+的数据源 —— 每次例会前主持人在同一页追加新一节，因此同步是按页整体解析后
+按日期 upsert。
+
+会议里最有价值的是「进展update」议题下的版本进展报告：tag 区间、合入补丁数、
+CVE 清单、按类别归好的补丁列表。它同时充当双周合入总结与各版本的 release notes，
+所以单独建表而不是塞进一个 JSON 字段 —— 补丁条目要能按 PR 号反查
+（"这个 PR 是在哪次例会上宣布合入的"）。
+"""
+
+import uuid
+from datetime import date, datetime
+from enum import StrEnum
+
+from sqlalchemy import (
+    Date,
+    DateTime,
+    ForeignKey,
+    Index,
+    Integer,
+    String,
+    Text,
+    UniqueConstraint,
+)
+from sqlalchemy.dialects.postgresql import ARRAY, JSONB, UUID
+from sqlalchemy.orm import Mapped, mapped_column, relationship
+
+from app.models.base import Base, TimestampMixin
+
+
+class MeetingStatus(StrEnum):
+    SCHEDULED = "scheduled"
+    DONE = "done"
+    CANCELLED = "cancelled"
+
+
+class Meeting(Base, TimestampMixin):
+    __tablename__ = "ksp_meeting"
+    __table_args__ = (
+        # 一场会议一个日期。唯一约束本身建了索引，不再重复加。
+        UniqueConstraint("held_on", name="uq_ksp_meeting_held_on"),
+    )
+
+    id: Mapped[uuid.UUID] = mapped_column(UUID(as_uuid=True), primary_key=True, default=uuid.uuid4)
+
+    held_on: Mapped[date] = mapped_column(Date, nullable=False)
+    start_time: Mapped[str | None] = mapped_column(String(8), nullable=True)
+    title: Mapped[str] = mapped_column(String(128), nullable=False)
+
+    host: Mapped[str | None] = mapped_column(String(64), nullable=True)
+    next_host: Mapped[str | None] = mapped_column(String(64), nullable=True)
+    host_order: Mapped[list[str] | None] = mapped_column(ARRAY(String(64)), nullable=True)
+
+    meeting_url: Mapped[str | None] = mapped_column(String(512), nullable=True)
+    minutes_url: Mapped[str | None] = mapped_column(String(512), nullable=True)
+
+    # 与 PullRequest.gate_stage 一致：存字符串而非枚举，便于以后加取值
+    status: Mapped[str] = mapped_column(
+        String(16), nullable=False, default=MeetingStatus.SCHEDULED.value, index=True
+    )
+    legacy: Mapped[str | None] = mapped_column(Text, nullable=True)
+
+    source_url: Mapped[str] = mapped_column(String(512), nullable=False)
+    source_hash: Mapped[str | None] = mapped_column(String(64), nullable=True)
+    """本节纪要内容指纹。Etherpad 整页同步，未变化的会议不必重写。"""
+
+    last_synced_at: Mapped[datetime | None] = mapped_column(DateTime(timezone=True), nullable=True)
+
+    agendas: Mapped[list["MeetingAgenda"]] = relationship(
+        back_populates="meeting",
+        cascade="all, delete-orphan",
+        order_by="MeetingAgenda.sequence",
+    )
+    reports: Mapped[list["ReleaseReport"]] = relationship(
+        back_populates="meeting",
+        cascade="all, delete-orphan",
+    )
+
+    def __repr__(self) -> str:
+        return f"<Meeting {self.held_on} {self.title}>"
+
+
+class MeetingAgenda(Base, TimestampMixin):
+    __tablename__ = "ksp_meeting_agenda"
+    __table_args__ = (
+        # 唯一约束的首列就是 meeting_id，按会议取议题走这个索引，
+        # 再给 meeting_id 单加一个索引是重复的
+        UniqueConstraint("meeting_id", "sequence", name="uq_ksp_meeting_agenda_seq"),
+        Index("ix_ksp_meeting_agenda_owner", "owner"),
+    )
+
+    id: Mapped[uuid.UUID] = mapped_column(UUID(as_uuid=True), primary_key=True, default=uuid.uuid4)
+    meeting_id: Mapped[uuid.UUID] = mapped_column(
+        UUID(as_uuid=True),
+        ForeignKey("ksp_meeting.id", ondelete="CASCADE"),
+        nullable=False,
+    )
+
+    sequence: Mapped[int] = mapped_column(Integer, nullable=False)
+    """出现顺序。议题编号在纪要里会重复（实测两处「议题一」），不能当键。"""
+
+    num: Mapped[str | None] = mapped_column(String(8), nullable=True)
+    title: Mapped[str] = mapped_column(Text, nullable=False)
+    owner: Mapped[str | None] = mapped_column(String(64), nullable=True)
+    body: Mapped[str | None] = mapped_column(Text, nullable=True)
+
+    pull_numbers: Mapped[list[int] | None] = mapped_column(ARRAY(Integer), nullable=True)
+    issue_numbers: Mapped[list[int] | None] = mapped_column(ARRAY(Integer), nullable=True)
+
+    meeting: Mapped[Meeting] = relationship(back_populates="agendas")
+
+    def __repr__(self) -> str:
+        return f"<Agenda {self.sequence} {self.title[:24]}>"
+
+
+class ReleaseReport(Base, TimestampMixin):
+    """一次例会上一个分支的版本进展报告。
+
+    对应纪要里的一段 ``【OLK-6.6】...``。既是双周合入总结，也是该版本的
+    release notes —— 它给出的正是"这半个月合入了什么"。
+    """
+
+    __tablename__ = "ksp_release_report"
+    __table_args__ = (
+        # 同上：唯一约束首列覆盖了按会议查询
+        UniqueConstraint("meeting_id", "branch", name="uq_ksp_release_report_branch"),
+        Index("ix_ksp_release_report_branch", "branch"),
+    )
+
+    id: Mapped[uuid.UUID] = mapped_column(UUID(as_uuid=True), primary_key=True, default=uuid.uuid4)
+    meeting_id: Mapped[uuid.UUID] = mapped_column(
+        UUID(as_uuid=True),
+        ForeignKey("ksp_meeting.id", ondelete="CASCADE"),
+        nullable=False,
+    )
+
+    branch: Mapped[str] = mapped_column(String(64), nullable=False)
+    """规范化后的分支名，与 ``domain.release`` 一致。"""
+
+    from_tag: Mapped[str | None] = mapped_column(String(32), nullable=True)
+    to_tag: Mapped[str | None] = mapped_column(String(32), nullable=True)
+
+    patch_count: Mapped[int | None] = mapped_column(Integer, nullable=True)
+    """作者自报的合入补丁数。与 ``listed_count`` 不同，后者是正文里真正列出的条数。"""
+
+    listed_count: Mapped[int] = mapped_column(Integer, nullable=False, default=0)
+
+    iso_urls: Mapped[list[str] | None] = mapped_column(ARRAY(String(512)), nullable=True)
+    cve_ids: Mapped[list[str] | None] = mapped_column(ARRAY(String(32)), nullable=True)
+
+    groups: Mapped[list | None] = mapped_column(JSONB, nullable=True)
+    """分组元信息（名称与自报条数）。条目本身在 ``ReleaseReportPatch``。"""
+
+    meeting: Mapped[Meeting] = relationship(back_populates="reports")
+
+    def __repr__(self) -> str:
+        return f"<ReleaseReport {self.branch} {self.from_tag}..{self.to_tag}>"
+
+
+class ReleaseReportPatch(Base, TimestampMixin):
+    """进展报告里列出的一条补丁。
+
+    单独建表是为了能按 ``pull_number`` 反查：PR 详情页可以显示
+    "该 PR 在 2026-09-04 的例会上宣布合入"，这是把会议与代码连起来的那根线。
+    """
+
+    __tablename__ = "ksp_release_report_patch"
+    __table_args__ = (
+        Index("ix_ksp_release_report_patch_pull", "pull_number"),
+        Index("ix_ksp_release_report_patch_report", "report_id", "sequence"),
+    )
+
+    id: Mapped[uuid.UUID] = mapped_column(UUID(as_uuid=True), primary_key=True, default=uuid.uuid4)
+    report_id: Mapped[uuid.UUID] = mapped_column(
+        UUID(as_uuid=True),
+        ForeignKey("ksp_release_report.id", ondelete="CASCADE"),
+        nullable=False,
+    )
+
+    sequence: Mapped[int] = mapped_column(Integer, nullable=False)
+    group_name: Mapped[str | None] = mapped_column(String(64), nullable=True)
+    title: Mapped[str] = mapped_column(Text, nullable=False)
+    pull_number: Mapped[int | None] = mapped_column(Integer, nullable=True)
+    url: Mapped[str | None] = mapped_column(String(512), nullable=True)
+
+    def __repr__(self) -> str:
+        return f"<ReportPatch {self.pull_number} {self.title[:24]}>"