阶段3（下）：RAG、向量数据库专题；接入 BOSS直聘调研（research/、岗位对照和延伸技术栈）；新增 cs 领域和 12 个技术栈骨架；面试题全部补齐追问并加入校验

tools/build_catalog.py

@@ -17,11 +17,11 @@ import sys
 from pathlib import Path
 
 ROOT = Path(__file__).resolve().parents[1]
-CONTENT_DIRS = ["plan", "topics", "scenarios"]
+CONTENT_DIRS = ["plan", "topics", "scenarios", "research"]
 SCHEMA_VERSION = 1
 
-TYPES = {"index", "lesson", "scenario", "interview", "project", "week"}
-DOMAINS = {"backend", "devops", "python", "ai", "frontend", "cross", "plan"}
+TYPES = {"index", "lesson", "scenario", "interview", "project", "week", "research"}
+DOMAINS = {"backend", "devops", "python", "ai", "frontend", "cs", "cross", "plan"}
 LEVELS = {"basic", "intermediate", "advanced"}
 QUESTION_LEVELS = ["基础", "进阶", "场景题"]
 ID_RE = re.compile(r"^[a-z0-9]+(?:\.[a-z0-9-]+)+$")
@@ -79,11 +79,12 @@ def parse_frontmatter(text: str) -> tuple[dict | None, str, list[str]]:
     return meta, text[end + 5:], errs
 
 
-def scan_body(body: str) -> tuple[bool, list[str], list[str], dict[str, int]]:
-    """返回：围栏是否闭合、二级标题、链接目标、按级别统计的面试题数。"""
+def scan_body(body: str) -> tuple[bool, list[str], list[str], dict[str, int], list[str]]:
+    """返回：围栏是否闭合、二级标题、链接目标、按级别统计的面试题数、缺少参考答案或追问的题目。"""
     in_fence, fence_mark = False, ""
     headings, links = [], []
     counts = {k: 0 for k in QUESTION_LEVELS}
+    questions: list[list] = []          # [标题, 有参考答案, 有追问]
     for line in body.splitlines():
         m = FENCE_RE.match(line)
         if m:
@@ -99,8 +100,17 @@ def scan_body(body: str) -> tuple[bool, list[str], list[str], dict[str, int]]:
         q = QUESTION_RE.match(line)
         if q:
             counts[q.group(1)] += 1
+            questions.append([line[4:].strip(), False, False])
+        elif line.startswith("#"):
+            if questions and line.startswith(("# ", "## ", "### ")):
+                questions.append([None, True, True])   # 题目之后的其他标题，结束上一题
+        elif questions:
+            questions[-1][1] |= line.startswith("**参考答案**")
+            questions[-1][2] |= line.startswith("**追问**")
         links += LINK_RE.findall(line)
-    return not in_fence, headings, links, counts
+    incomplete = [f"{t}（缺少{'“**参考答案**”' if not a else ''}{'“**追问**”' if not f else ''}）"
+                  for t, a, f in questions if t and not (a and f)]
+    return not in_fence, headings, links, counts, incomplete
 
 
 def main() -> int:
@@ -136,7 +146,9 @@ def main() -> int:
         for w in meta.get("week", []) if isinstance(meta.get("week"), list) else []:
             if not isinstance(w, int) or not 1 <= w <= 16:
                 errors.append(f"{rel}: week 取值 {w!r} 必须是 1~16 的整数")
-        closed, headings, links, counts = scan_body(body)
+        closed, headings, links, counts, incomplete = scan_body(body)
+        if t == "interview":
+            errors += [f"{rel}: 面试题 {q}" for q in incomplete]
         if not closed:
             errors.append(f"{rel}: 代码块围栏没有闭合")
         for sec in REQUIRED_SECTIONS.get(t, []):