feat(db): 0003 migration + LlmSetting/Source 模型加 blocklist_tags

- 新增 alembic 0003: sources.blocklist_tags + llm_settings.blocklist_tags(JSONB)
- 两层配置:全局(llm_settings) + per-source(sources),合并去重后注入 classify prompt
- 默认空数组,不影响存量数据;admin API 在下个 commit 暴露编辑入口

backend/alembic/versions/0003_blocklist_tags.py

@@ -0,0 +1,56 @@
+"""Sources / llm_settings 加 blocklist_tags(屏蔽分类标签)
+
+设计:
+- sources.blocklist_tags:      当前源级屏蔽词(per-source)
+- llm_settings.blocklist_tags: 全局屏蔽词(对所有源生效)
+- 合并去重后注入到 classify prompt,LLM 判断文章类别是否命中 → drop
+
+注意:
+- 这两个字段在 classify 阶段之前是空数组(默认 '[]'::jsonb),
+  不会影响存量数据;admin_llm/admin.py 会暴露编辑入口。
+
+Revision ID: 0003
+Revises: 0002
+Create Date: 2026-06-09
+"""
+from __future__ import annotations
+
+from typing import Sequence, Union
+
+import sqlalchemy as sa
+from alembic import op
+from sqlalchemy.dialects.postgresql import JSONB
+
+revision: str = "0003"
+down_revision: Union[str, None] = "0002"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    # === sources 加 per-source 屏蔽词 ===
+    op.add_column(
+        "sources",
+        sa.Column(
+            "blocklist_tags",
+            JSONB,
+            nullable=False,
+            server_default=sa.text("'[]'::jsonb"),
+        ),
+    )
+
+    # === llm_settings 加全局屏蔽词 ===
+    op.add_column(
+        "llm_settings",
+        sa.Column(
+            "blocklist_tags",
+            JSONB,
+            nullable=False,
+            server_default=sa.text("'[]'::jsonb"),
+        ),
+    )
+
+
+def downgrade() -> None:
+    op.drop_column("llm_settings", "blocklist_tags")
+    op.drop_column("sources", "blocklist_tags")

backend/app/models/llm_setting.py

@@ -11,6 +11,7 @@ from __future__ import annotations
 from datetime import datetime
 
 from sqlalchemy import Boolean, DateTime, Integer, String, Text, func
+from sqlalchemy.dialects.postgresql import JSONB
 from sqlalchemy.orm import Mapped, mapped_column
 
 from app.database import Base
@@ -28,6 +29,13 @@ class LlmSetting(Base):
     commentary_prompt: Mapped[str | None] = mapped_column(Text)
     image_prompt_template: Mapped[str | None] = mapped_column(Text)
 
+    # === 全局屏蔽分类标签(如 ["体育", "娱乐"])===
+    # 与 sources.blocklist_tags 合并去重后注入 classify prompt;
+    # 命中则删文章(drop)
+    blocklist_tags: Mapped[list[str]] = mapped_column(
+        JSONB, nullable=False, default=list, server_default="[]"
+    )
+
     # === 插图参数 ===
     image_size: Mapped[str] = mapped_column(String(16), default="768x512", nullable=False)
 

backend/app/models/source.py

@@ -14,6 +14,7 @@ from sqlalchemy import (
     Text,
     func,
 )
+from sqlalchemy.dialects.postgresql import JSONB
 from sqlalchemy.orm import Mapped, mapped_column, relationship
 
 from app.database import Base
@@ -50,6 +51,10 @@ class Source(Base):
     language_src: Mapped[str | None] = mapped_column(String(8))
     priority: Mapped[int] = mapped_column(Integer, default=50, nullable=False, index=True)
     headers_json: Mapped[dict | None] = mapped_column(JSON)
+    # 源级屏蔽分类标签(如 ["体育", "娱乐"]);与 llm_settings.blocklist_tags 合并后使用
+    blocklist_tags: Mapped[list[str]] = mapped_column(
+        JSONB, nullable=False, default=list, server_default="[]"
+    )
     last_fetched_at: Mapped[datetime | None] = mapped_column(DateTime(timezone=True))
     last_status: Mapped[str | None] = mapped_column(String(64))
     consecutive_failures: Mapped[int] = mapped_column(Integer, default=0, nullable=False)