From f82d5d731f7c1439a95dda6e1a14d40d58881d69 Mon Sep 17 00:00:00 2001
From: Ariver <shanghai3168@gmail.com>
Date: Tue, 25 Aug 2026 16:18:08 +0800
Subject: [PATCH] checkpoint: 日记改格式前备份
---
X2.Archived/scripts/sync_0inbox.py | 23 ++++++++++++++++-------
1 files changed, 16 insertions(+), 7 deletions(-)
diff --git a/X2.Archived/scripts/sync_0inbox.py b/X2.Archived/scripts/sync_0inbox.py
index 5741037..7b6887d 100644
--- a/X2.Archived/scripts/sync_0inbox.py
+++ b/X2.Archived/scripts/sync_0inbox.py
@@ -27,7 +27,8 @@
--------------------
- 顶层黑名单 EXCLUDE_TOP 整棵子树不扫：X2.Archived / 00index / X1.Knomo / X.Attachment / P3.bobo
- 仅扫描后缀 SCAN_EXT：.md .markdown .excalidraw .canvas；其它后缀一律跳过
-- markdown 同时抽 frontmatter（tags + 关键词字段）与正文 #tag
+- markdown 只抽标准 `tags:` frontmatter 与正文 #tag；topic/category/type 等
+ 元数据字段不再当作标签（避免 copilot 对话、读书笔记的元数据被误建页）
幂等 & 安全
----------
@@ -140,15 +141,18 @@
TAG_RE = re.compile(r"(?<![\w/])#([A-Za-z0-9_\u4e00-\u9fff][\w\u4e00-\u9fff/\-]*)")
SCAN_EXT = {".md", ".markdown", ".excalidraw", ".canvas"}
EXCLUDE_TOP = {"X2.Archived", "00index", "X1.Knomo", "X.Attachment", "P3.bobo"}
+# 嵌套路径排除（只排除指定子树，不影响其同级目录）。
+# 例：02DS/02copilot 是 Obsidian Copilot 的对话日志，其 topic: 字段与聊天正文
+# 不应被当作标签扫描（与 tagfeed 模板的 excludedPaths 保持一致）。
+EXCLUDE_PATHS = {"02DS/02copilot"}
INBOX_DIR = "0inbox"
EXCLUDE_DIRS = {".obsidian", ".trash", ".git", ".agents", ".claude",
".copilot", ".opencode", ".smart-env", ".workbuddy"}
-FM_KEYWORD_KEYS = {
- "aliases", "alias", "keywords", "keyword", "topic", "topics",
- "category", "categories", "subject", "area", "areaOfFocus",
- "people", "person", "project", "projects", "type", "tags",
-}
+# 只把标准 Obsidian 标签字段 `tags:` 当作标签；不再把 topic/category/subject/
+# type/keywords/aliases 等元数据字段误读为标签（否则 copilot 对话的 topic:、
+# 读书笔记的 type: 等会被当成标签疯狂建页）。行内 #标签 仍正常扫描。
+FM_KEYWORD_KEYS = {"tags"}
def split_fm_value(v) -> list:
@@ -358,10 +362,15 @@
all_tags = {}
for root, dirs, files in os.walk(vault):
dirs[:] = [d for d in dirs if d not in EXCLUDE_DIRS and not d.startswith(".")]
- top = os.path.relpath(root, vault).split(os.sep)[0]
+ rel = os.path.relpath(root, vault)
+ top = rel.split(os.sep)[0]
if top in EXCLUDE_TOP:
dirs[:] = []
continue
+ # 嵌套路径排除（如 02DS/02copilot）
+ if any(rel == p or rel.startswith(p + os.sep) for p in EXCLUDE_PATHS):
+ dirs[:] = []
+ continue
for fn in files:
if fn.startswith("."):
continue
--
Gitblit v1.9.3