tech-book-extractor-skills 1.0.8 → 1.0.10

package/README.md

@@ -2,6 +2,8 @@
 
 Claude Code 技能：技术书深度萃取——两阶段流水线。
 
+[![npm version](https://img.shields.io/npm/v/tech-book-extractor-skills)](https://www.npmjs.com/package/tech-book-extractor-skills)
+
 ## 安装
 
 ```bash

package/bin/install.js

@@ -6,7 +6,6 @@ const os = require("os");
 
 const skillsSource = path.join(__dirname, "..", "skills");
 const skillsDest = path.join(os.homedir(), ".claude", "skills");
-const scriptsSrc = path.join(__dirname, "..", "stage1", "complexity_scanner.py");
 const scriptsDest = path.join(os.homedir(), ".claude", "scripts");
 
 fs.mkdirSync(skillsDest, { recursive: true });
@@ -34,14 +33,9 @@ for (const skill of skills) {
   console.log(`✓ skill: ${skill} → ${dest}`);
 }
 
-// 复制预处理脚本
-const scannerDest = path.join(scriptsDest, "complexity_scanner.py");
-fs.copyFileSync(scriptsSrc, scannerDest);
-console.log(`✓ script: complexity_scanner.py → ${scannerDest}`);
-
-// 复制 PDF 文本提取脚本
+// 复制 Python 脚本
 const scriptsDir = path.join(__dirname, "..", "scripts");
-for (const pyScript of ["pdf_extract_utils.py", "extract_book.py", "extract_chapter.py"]) {
+for (const pyScript of ["complexity_scanner.py", "pdf_extract_utils.py", "extract_book.py", "extract_chapter.py"]) {
   const dest = path.join(scriptsDest, pyScript);
   fs.copyFileSync(path.join(scriptsDir, pyScript), dest);
   console.log(`✓ script: ${pyScript} → ${dest}`);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "tech-book-extractor-skills",
-  "version": "1.0.8",
+  "version": "1.0.10",
   "description": "Claude Code skills for deep technical book reading — structure parsing (Stage 1) and chapter extraction (Stage 2).",
   "bin": {
     "tech-book-extractor-skills": "bin/install.js"
@@ -8,16 +8,13 @@
   "files": [
     "skills/",
     "bin/",
-    "stage1/complexity_scanner.py",
+    "scripts/complexity_scanner.py",
     "scripts/pdf_extract_utils.py",
     "scripts/extract_book.py",
     "scripts/extract_chapter.py"
   ],
   "scripts": {
-    "install-skills": "node bin/install.js",
-    "sync": "node scripts/sync.js",
-    "release:patch": "npm run sync && npm version patch && npm publish",
-    "release:minor": "npm run sync && npm version minor && npm publish"
+    "install-skills": "node bin/install.js"
   },
   "repository": {
     "type": "git",

package/scripts/extract_book.py

@@ -16,7 +16,7 @@ extract_book.py — 整本书文本提取脚本
 
 输出：
   {output_dir}/{书名}/{书名}-fulltext.md
-  {output_dir}/{书名}/images/          # 仅 --export-images 时
+  中间结果（图片等）：/tmp/tech-book-extractor/{书名}/
 """
 
 from __future__ import annotations
@@ -32,6 +32,7 @@ from pdf_extract_utils import (
     extract_pages,
     save_results,
     extract_book_name,
+    get_tmp_dir,
 )
 
 
@@ -67,7 +68,9 @@ def main():
     book_name = extract_book_name(pdf_path.name)
     output_dir = Path(args.output) / book_name
     output_path = output_dir / f"{book_name}-fulltext.md"
-    image_dir = str(output_dir / "images") if args.export_images else ""
+    # 中间结果（图片等）固定放 /tmp，不污染用户输出目录
+    tmp_dir = get_tmp_dir(book_name)
+    image_dir = str(tmp_dir / "images") if args.export_images else ""
 
     # 页码范围
     total_pages = pdf_page_count(pdf_path)

package/scripts/extract_chapter.py

@@ -17,6 +17,7 @@ extract_chapter.py — 单章文本提取脚本
 
 输出：
   {output_dir}/{书名}/chapters/{chapter_id}-raw.md
+  中间结果（图片等）：/tmp/tech-book-extractor/{书名}/
 """
 
 from __future__ import annotations
@@ -32,6 +33,7 @@ from pdf_extract_utils import (
     extract_pages,
     save_results,
     extract_book_name,
+    get_tmp_dir,
     parse_page_range,
 )
 
@@ -80,7 +82,9 @@ def main():
     book_name = extract_book_name(pdf_path.name)
     output_dir = Path(args.output) / book_name / "chapters"
     output_path = output_dir / f"{chapter_id}-raw.md"
-    image_dir = str(Path(args.output) / book_name / "images") if args.export_images else ""
+    # 中间结果（图片等）固定放 /tmp，不污染用户输出目录
+    tmp_dir = get_tmp_dir(book_name)
+    image_dir = str(tmp_dir / "images") if args.export_images else ""
 
     page_count = page_end - page_start + 1
     print(f"📖 {book_name}")

package/scripts/pdf_extract_utils.py

@@ -518,6 +518,13 @@ def save_results(
 # 辅助函数
 # ═══════════════════════════════════════════════════════════════════
 
+def get_tmp_dir(book_name: str) -> Path:
+    """中间结果目录：/tmp/tech-book-extractor/<书名>/"""
+    path = Path("/tmp/tech-book-extractor") / book_name
+    path.mkdir(parents=True, exist_ok=True)
+    return path
+
+
 def extract_book_name(pdf_path: str | Path) -> str:
     """从 PDF 文件名提取书名（去后缀、去特殊字符）"""
     name = Path(pdf_path).stem

package/skills/chapter-drill/SKILL.md

@@ -8,6 +8,21 @@ description: "技术书章节深度萃取——把一章钻透。自动生成骨
 
 ---
 
+## ⛔ 硬约束：骨架必须先于萃取
+
+**本章 skill 的执行前提是 Stage 1 骨架（`stage1-skeleton.json`）已存在。**
+
+执行前必须检查：`{output_dir}/{书名}/stage1-skeleton.json` 是否存在。
+
+| 骨架状态 | 行为 |
+|---------|------|
+| 存在 | ✅ 读取骨架，进入萃取流程 |
+| 不存在 | 🚫 **中止本章萃取**，提示用户先运行 `/book-map` 生成骨架 |
+
+> **绝对禁止**：在骨架缺失时跳过骨架直接萃取。骨架是萃取的导航系统，没有骨架的萃取等于盲飞。
+
+---
+
 ## 前置步骤：从 PDF 提取单章文本
 
 如果章节原文在 PDF 里，先用提取脚本拿到文字：