项目初始化

daily/cursor_editor.py

@@ -0,0 +1,144 @@
+"""Cursor 编辑层：JSON 数据 → 主题 / 速览 / 中文描述。"""
+
+from __future__ import annotations
+
+import json
+import logging
+from pathlib import Path
+from typing import Any
+
+from daily.config import ROOT, env
+from daily.llm_client import extract_json_object, has_llm_configured, llm_chat
+from daily.text_utils import clip_text
+from daily.report_data import editorial_json_path, save_json, skill_id
+
+logger = logging.getLogger(__name__)
+
+_SKILL_DIR = ROOT / "skills" / "daily-editor"
+
+
+def is_enabled() -> bool:
+    raw = (env("DAILY_CURSOR_EDITOR") or "").strip().lower()
+    if raw in {"1", "true", "yes", "on"}:
+        return has_llm_configured()
+    if raw in {"0", "false", "no", "off"}:
+        return False
+    return False
+
+
+def _load_skill_prompt() -> str:
+    skill_path = _SKILL_DIR / "SKILL.md"
+    if skill_path.exists():
+        return skill_path.read_text(encoding="utf-8").strip()
+    return "你是技术早报编辑。根据输入 JSON 输出编辑结果 JSON。"
+
+
+def _build_system_prompt() -> str:
+    skill = _load_skill_prompt()
+    return (
+        f"{skill}\n\n"
+        "再次强调：只输出 JSON 对象，包含 theme_line、highlights（3条）、descriptions。"
+    )
+
+
+def run_editorial(llm_input: dict[str, Any], *, date_str: str) -> dict[str, Any] | None:
+    """调用 LLM 生成 editorial；失败返回 None。"""
+    if not is_enabled():
+        return None
+    system = _build_system_prompt()
+    user = json.dumps(llm_input, ensure_ascii=False, indent=2)
+    try:
+        raw = llm_chat(system, user)
+    except Exception as exc:
+        logger.warning("Cursor 编辑失败，回退规则模式：%s", exc)
+        return None
+    if not raw:
+        logger.warning("Cursor 编辑无响应，回退规则模式")
+        return None
+    parsed = extract_json_object(raw)
+    if not parsed.get("theme_line") and not parsed.get("descriptions"):
+        logger.warning("Cursor 编辑 JSON 无效，回退规则模式")
+        return None
+    editorial = _normalize_editorial(parsed)
+    save_json(editorial_json_path(date_str), editorial)
+    return editorial
+
+
+def _normalize_editorial(raw: dict[str, Any]) -> dict[str, Any]:
+    theme = str(raw.get("theme_line") or "").strip()
+    highlights_raw = raw.get("highlights") or []
+    highlights: list[str] = []
+    if isinstance(highlights_raw, list):
+        for item in highlights_raw:
+            if isinstance(item, str) and item.strip():
+                highlights.append(item.strip())
+    descriptions_raw = raw.get("descriptions") or {}
+    descriptions: dict[str, str] = {}
+    if isinstance(descriptions_raw, dict):
+        for key, value in descriptions_raw.items():
+            if isinstance(value, str) and value.strip():
+                limit = 40 if str(key).startswith("github:") else 36
+                descriptions[str(key)] = clip_text(value, limit)
+    return {
+        "theme_line": theme,
+        "highlights": highlights[:3],
+        "descriptions": descriptions,
+    }
+
+
+def theme_line_from_editorial(editorial: dict[str, Any]) -> str:
+    theme = editorial.get("theme_line", "")
+    if not theme:
+        return ""
+    if "今日主题" in theme:
+        return theme if theme.startswith("**") else f"**{theme}**"
+    return f"**今日主题**：{theme}"
+
+
+def apply_descriptions(
+    *,
+    trending: list[dict[str, Any]],
+    hot: list[dict[str, Any]],
+    github_trending: list[dict[str, Any]],
+    github_emerging: list[dict[str, Any]],
+    github_topic: list[dict[str, Any]],
+    ai_news: dict[str, Any],
+    descriptions: dict[str, str],
+) -> None:
+    if not descriptions:
+        return
+
+    for item in trending + hot:
+        key = f"skill:{skill_id(item)}"
+        if key in descriptions:
+            item["description"] = descriptions[key]
+
+    for repo_list in (github_trending, github_emerging, github_topic):
+        for item in repo_list:
+            key = f"github:{item.get('repo', '')}"
+            if key in descriptions:
+                item["description"] = descriptions[key]
+
+    if not ai_news.get("enabled"):
+        return
+
+    def _apply_news_item(item: dict[str, Any]) -> None:
+        key = f"news:{item.get('link', '')}"
+        if key in descriptions:
+            item["summary"] = descriptions[key]
+
+    for cat in ai_news.get("categories") or []:
+        for item in cat.get("items") or []:
+            _apply_news_item(item)
+    for item in ai_news.get("flat") or []:
+        _apply_news_item(item)
+
+
+def load_cached_editorial(date_str: str) -> dict[str, Any] | None:
+    path = editorial_json_path(date_str)
+    if not path.exists():
+        return None
+    try:
+        return _normalize_editorial(json.loads(path.read_text(encoding="utf-8")))
+    except (OSError, json.JSONDecodeError):
+        return None