session-to-obsidian 0.1.0__tar.gz

session_to_obsidian-0.1.0/.github/workflows/ci.yml

@@ -0,0 +1,18 @@
+name: ci
+on:
+  push:
+    branches: [main]
+  pull_request:
+jobs:
+  test:
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        python-version: ["3.10", "3.12"]
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: ${{ matrix.python-version }}
+      - run: pip install pytest -e .
+      - run: pytest -q

session_to_obsidian-0.1.0/.github/workflows/release.yml

@@ -0,0 +1,19 @@
+name: release
+# 推一個版本 tag(例如 v0.1.0)就自動 build + 發佈到 PyPI。
+# 用 PyPI Trusted Publishing(OIDC),不需要任何 token/secret。
+on:
+  push:
+    tags: ["v*"]
+jobs:
+  publish:
+    runs-on: ubuntu-latest
+    permissions:
+      id-token: write   # Trusted Publishing 需要;PyPI 用它驗證身分,免 token
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+      - run: pip install build
+      - run: python -m build
+      - uses: pypa/gh-action-pypi-publish@release/v1

session_to_obsidian-0.1.0/.gitignore

@@ -0,0 +1,9 @@
+__pycache__/
+*.py[cod]
+*.egg-info/
+build/
+dist/
+.venv/
+venv/
+.pytest_cache/
+.DS_Store

session_to_obsidian-0.1.0/CHANGELOG.md

@@ -0,0 +1,25 @@
+# Changelog
+
+All notable changes to this project are documented here. Format follows
+[Keep a Changelog](https://keepachangelog.com/), and this project adheres to
+[Semantic Versioning](https://semver.org/).
+
+## [Unreleased]
+
+## [0.1.0] - 2026-06-19
+
+Initial release.
+
+### Added
+- `s2o import <file> --vault <dir>` — import a Claude Code or GitHub Copilot
+  Chat session into Obsidian as a topic-organized recap note + topic-grouped
+  Canvas, and update a cross-session `INDEX.md`.
+- Auto source detection: Claude native jsonl, Copilot Chat append-log (replayed).
+- Semantic topic clustering via the local `claude` CLI (headless, no API key);
+  `--cluster time` fallback with no LLM.
+- Bilingual output (`--lang en|zh-TW`, default `en`; `S2O_LANG` env).
+- Frontmatter `originSessionId` for dedupe — re-importing a session replaces it.
+- `s2o index` to rebuild the index.
+
+[Unreleased]: https://github.com/timo9378/session-to-obsidian/compare/v0.1.0...HEAD
+[0.1.0]: https://github.com/timo9378/session-to-obsidian/releases/tag/v0.1.0

session_to_obsidian-0.1.0/CONTRIBUTING.md

@@ -0,0 +1,40 @@
+# Contributing
+
+Thanks for your interest! This is a small, dependency-free tool — contributions that keep it simple are very welcome.
+
+## Setup
+
+```bash
+git clone https://github.com/timo9378/session-to-obsidian
+cd session-to-obsidian
+pip install pytest -e .   # or use a venv
+pytest
+```
+
+## Project layout
+
+```
+s2o/
+  parse.py        jsonl → steps (no LLM)
+  adapters/       source detection + restore (copilot append-log replay, claude native)
+  cluster.py      ask-list → {title, topics}; backends: claude_cli (host Claude), time
+  render.py       steps + topics → .md (TOC + topic sections) + .canvas
+  index.py        sessions/ → INDEX.md
+  slugs.py        title → cross-platform filename slug
+  i18n.py         output labels + clustering prompt (en / zh-TW)
+  cli.py          orchestration (import / index)
+```
+
+Design rationale lives in [SPEC.md](SPEC.md) — read it before changing clustering or rendering.
+
+## Guidelines
+
+- **No runtime dependencies.** Standard library only. The LLM step shells out to the `claude` CLI.
+- Add a test for any bug you fix or behavior you add (`tests/test_s2o.py`).
+- The hard-won bits (Copilot append-log replay, heading-demote + fence-balancing, topic clustering contract) are covered by tests — keep them green.
+- Adding a new source (e.g. another AI IDE)? Write an adapter in `s2o/adapters/` returning `(records, meta)` and wire it into `detect()`.
+- Adding a language? Extend `LABELS` and `_RULES_NAME` in `s2o/i18n.py`.
+
+## Reporting issues
+
+Include the source type (Claude / Copilot), the command you ran, and what the output looked like vs what you expected.

session_to_obsidian-0.1.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 timo9378
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

session_to_obsidian-0.1.0/PKG-INFO

@@ -0,0 +1,93 @@
+Metadata-Version: 2.4
+Name: session-to-obsidian
+Version: 0.1.0
+Summary: 把 Claude Code / GitHub Copilot 對話 session 變成 Obsidian 裡依語意主題分區、可翻找、可給 Claude compact 後回想的回顧筆記
+Project-URL: Homepage, https://github.com/timo9378/session-to-obsidian
+Project-URL: Repository, https://github.com/timo9378/session-to-obsidian
+Project-URL: Issues, https://github.com/timo9378/session-to-obsidian/issues
+Project-URL: Changelog, https://github.com/timo9378/session-to-obsidian/blob/main/CHANGELOG.md
+Author: timo9378
+License-Expression: MIT
+License-File: LICENSE
+Keywords: claude-code,copilot,knowledge-management,llm,obsidian
+Classifier: Environment :: Console
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Topic :: Utilities
+Requires-Python: >=3.10
+Description-Content-Type: text/markdown
+
+# session-to-obsidian
+
+[![CI](https://github.com/timo9378/session-to-obsidian/actions/workflows/ci.yml/badge.svg)](https://github.com/timo9378/session-to-obsidian/actions/workflows/ci.yml)
+[![PyPI](https://img.shields.io/pypi/v/session-to-obsidian.svg)](https://pypi.org/project/session-to-obsidian/)
+[![Python](https://img.shields.io/badge/python-3.10%2B-blue.svg)](https://www.python.org/)
+[![License: MIT](https://img.shields.io/badge/license-MIT-blue.svg)](LICENSE)
+
+Turn a **Claude Code** or **GitHub Copilot Chat** session into an Obsidian note that is organized **by topic**, easy to browse, good for writing from, and useful as a memory layer your AI can recall after a context compaction.
+
+繁體中文說明見 **[README.zh-TW.md](README.zh-TW.md)**.
+
+Not a verbatim log dump. It splits one session into **coherent topic threads** (not time slices). Under each topic: *what you asked + the assistant's full narrative + the tools/files/screenshots involved*, with a table of contents on top — plus a cross-session `INDEX.md`.
+
+## Why
+
+One AI coding session often does several unrelated things (fix a bug, tweak some UI, detour into something else). Later you want to find "how did I solve that thing" — and a timeline is useless, because what you remember is the **topic**, not the timestamp. This tool uses an LLM to cluster steps into topics so the note is organized the way you remember it.
+
+The first audience is the author's own future self (a tool you keep using); open source is a bonus.
+
+## Install
+
+```bash
+pipx install session-to-obsidian
+```
+
+Topic clustering calls your **local `claude` CLI** (Claude Code) in headless mode — no API key, uses your existing subscription. No Claude Code? It still works, clustering just degrades to time-slicing (see below).
+
+## Usage
+
+```bash
+# A Claude Code session (jsonl lives in ~/.claude/projects/<proj>/<uuid>.jsonl)
+s2o import ~/.claude/projects/myproj/abc123.jsonl --vault ~/Obsidian
+
+# A Copilot Chat session (VS Code's chatSessions/<uuid>.jsonl append-log)
+s2o import "~/…/workspaceStorage/<hash>/chatSessions/<uuid>.jsonl" --vault ~/Obsidian
+
+# Output language (labels + generated titles). Default: en
+s2o import <file> --vault ~/Obsidian --lang zh-TW
+
+# No LLM — pure time-segmentation (offline / no Claude Code)
+s2o import <file> --vault ~/Obsidian --cluster time
+
+# Rebuild the index only
+s2o index --vault ~/Obsidian
+```
+
+Source (Claude native / Copilot append-log) is auto-detected. Output goes to `<vault>/90-Meta/sessions/<date>-<source>-<title>/` with a `.md` (recap) and `.canvas` (topic-grouped node graph). Set `S2O_VAULT` to skip `--vault`; `S2O_LANG` to set a default language. Re-importing the same session replaces the previous note (deduped by an `originSessionId` stored in frontmatter).
+
+## How it works
+
+```
+detect source → adapter restores steps → extract "ask + files + first-line gist"
+   → cluster (LLM) → render .md/.canvas → update INDEX.md
+```
+
+Clustering only feeds the LLM **the ask list + each step's touched filenames + the assistant's first line** (never the full transcript), so token use is bounded. Compound asks go to their dominant topic; topics may be non-contiguous in time; each step keeps its original number.
+
+## Limitations
+
+- **Topic clustering needs the local `claude` CLI** (Claude Code). Without it, use `--cluster time`.
+- Images are not extracted by default (`--images` to enable). By Obsidian convention images live under `_attachments/` (gitignore them; sync via Syncthing to avoid bloating git).
+- Topic names / titles are LLM-generated; quality depends on the model. Re-run a single session to regenerate.
+
+## Development
+
+```bash
+pip install pytest -e .
+pytest
+```
+
+## License
+
+MIT — see [LICENSE](LICENSE).

session_to_obsidian-0.1.0/README.md

@@ -0,0 +1,73 @@
+# session-to-obsidian
+
+[![CI](https://github.com/timo9378/session-to-obsidian/actions/workflows/ci.yml/badge.svg)](https://github.com/timo9378/session-to-obsidian/actions/workflows/ci.yml)
+[![PyPI](https://img.shields.io/pypi/v/session-to-obsidian.svg)](https://pypi.org/project/session-to-obsidian/)
+[![Python](https://img.shields.io/badge/python-3.10%2B-blue.svg)](https://www.python.org/)
+[![License: MIT](https://img.shields.io/badge/license-MIT-blue.svg)](LICENSE)
+
+Turn a **Claude Code** or **GitHub Copilot Chat** session into an Obsidian note that is organized **by topic**, easy to browse, good for writing from, and useful as a memory layer your AI can recall after a context compaction.
+
+繁體中文說明見 **[README.zh-TW.md](README.zh-TW.md)**.
+
+Not a verbatim log dump. It splits one session into **coherent topic threads** (not time slices). Under each topic: *what you asked + the assistant's full narrative + the tools/files/screenshots involved*, with a table of contents on top — plus a cross-session `INDEX.md`.
+
+## Why
+
+One AI coding session often does several unrelated things (fix a bug, tweak some UI, detour into something else). Later you want to find "how did I solve that thing" — and a timeline is useless, because what you remember is the **topic**, not the timestamp. This tool uses an LLM to cluster steps into topics so the note is organized the way you remember it.
+
+The first audience is the author's own future self (a tool you keep using); open source is a bonus.
+
+## Install
+
+```bash
+pipx install session-to-obsidian
+```
+
+Topic clustering calls your **local `claude` CLI** (Claude Code) in headless mode — no API key, uses your existing subscription. No Claude Code? It still works, clustering just degrades to time-slicing (see below).
+
+## Usage
+
+```bash
+# A Claude Code session (jsonl lives in ~/.claude/projects/<proj>/<uuid>.jsonl)
+s2o import ~/.claude/projects/myproj/abc123.jsonl --vault ~/Obsidian
+
+# A Copilot Chat session (VS Code's chatSessions/<uuid>.jsonl append-log)
+s2o import "~/…/workspaceStorage/<hash>/chatSessions/<uuid>.jsonl" --vault ~/Obsidian
+
+# Output language (labels + generated titles). Default: en
+s2o import <file> --vault ~/Obsidian --lang zh-TW
+
+# No LLM — pure time-segmentation (offline / no Claude Code)
+s2o import <file> --vault ~/Obsidian --cluster time
+
+# Rebuild the index only
+s2o index --vault ~/Obsidian
+```
+
+Source (Claude native / Copilot append-log) is auto-detected. Output goes to `<vault>/90-Meta/sessions/<date>-<source>-<title>/` with a `.md` (recap) and `.canvas` (topic-grouped node graph). Set `S2O_VAULT` to skip `--vault`; `S2O_LANG` to set a default language. Re-importing the same session replaces the previous note (deduped by an `originSessionId` stored in frontmatter).
+
+## How it works
+
+```
+detect source → adapter restores steps → extract "ask + files + first-line gist"
+   → cluster (LLM) → render .md/.canvas → update INDEX.md
+```
+
+Clustering only feeds the LLM **the ask list + each step's touched filenames + the assistant's first line** (never the full transcript), so token use is bounded. Compound asks go to their dominant topic; topics may be non-contiguous in time; each step keeps its original number.
+
+## Limitations
+
+- **Topic clustering needs the local `claude` CLI** (Claude Code). Without it, use `--cluster time`.
+- Images are not extracted by default (`--images` to enable). By Obsidian convention images live under `_attachments/` (gitignore them; sync via Syncthing to avoid bloating git).
+- Topic names / titles are LLM-generated; quality depends on the model. Re-run a single session to regenerate.
+
+## Development
+
+```bash
+pip install pytest -e .
+pytest
+```
+
+## License
+
+MIT — see [LICENSE](LICENSE).

session_to_obsidian-0.1.0/README.zh-TW.md

@@ -0,0 +1,69 @@
+# session-to-obsidian
+
+[![CI](https://github.com/timo9378/session-to-obsidian/actions/workflows/ci.yml/badge.svg)](https://github.com/timo9378/session-to-obsidian/actions/workflows/ci.yml)
+[![PyPI](https://img.shields.io/pypi/v/session-to-obsidian.svg)](https://pypi.org/project/session-to-obsidian/)
+[![License: MIT](https://img.shields.io/badge/license-MIT-blue.svg)](LICENSE)
+
+把一場 **Claude Code** 或 **GitHub Copilot Chat** 的對話 session,變成 Obsidian 裡**依語意主題分區、可快速翻找、可給 Claude compact 後回想、可拿來寫文章**的回顧筆記。
+
+> English: **[README.md](README.md)**
+
+不是逐字 log dump。它把一場 session 拆成**連貫的主題串**(不是時間段),每條主題底下是「我問 + 助手完整敘述 + 用到的工具/檔案/截圖」,頂部附目錄,並維護一份跨 session 的 `INDEX.md` 索引。
+
+## 為什麼
+
+AI coding 一場 session 常常一次解好幾件事(修個 bug、順手改 UI、中間插一段別的)。事後想翻「那個問題當時怎麼解的」,時間軸沒用——你記得的是**主題**,不是幾點做的。這工具用 LLM 把步驟分到主題,讓筆記照你記憶的方式組織。第一受眾是作者本人;開源是順手。
+
+## 安裝
+
+```bash
+pipx install session-to-obsidian
+```
+
+語意分群預設呼叫你**本機的 `claude` CLI**(Claude Code)——零 API 金鑰、吃你現有訂閱。沒裝 Claude Code 也能用,分群退化成時間分段。
+
+## 用法
+
+```bash
+# Claude Code session(jsonl 在 ~/.claude/projects/<專案>/<uuid>.jsonl)
+s2o import ~/.claude/projects/myproj/abc123.jsonl --vault ~/Obsidian
+
+# Copilot Chat session(VS Code 的 chatSessions/<uuid>.jsonl append-log)
+s2o import "~/…/workspaceStorage/<hash>/chatSessions/<uuid>.jsonl" --vault ~/Obsidian
+
+# 輸出語言(標籤 + 產生的標題),預設 en
+s2o import <file> --vault ~/Obsidian --lang zh-TW
+
+# 不靠 LLM,純時間分段(離線 / 無 Claude Code)
+s2o import <file> --vault ~/Obsidian --cluster time
+
+# 只重建索引
+s2o index --vault ~/Obsidian
+```
+
+來源自動偵測。輸出到 `<vault>/90-Meta/sessions/<日期>-<來源>-<標題>/`(`.md` + `.canvas`)。`S2O_VAULT` 可省 `--vault`,`S2O_LANG` 設預設語言。重匯同一場會自動取代舊版(靠 frontmatter 的 `originSessionId` 去重)。
+
+## 它怎麼運作
+
+```
+偵測來源 → adapter 還原步驟 → 抽「提問+檔名+助手首句」→ 分群(LLM)→ 渲染 .md/.canvas → 更新 INDEX.md
+```
+
+分群只餵**提問清單 + 該步檔名 + 助手首句**(不餵全文),token 有界。複合提問歸主導主題、主題可跨時間非連續、每步保留原始步號。
+
+## 限制
+
+- **語意分群需要本機 `claude` CLI**(Claude Code)。沒有就 `--cluster time`。
+- 圖片預設不抽(`--images` 開啟);圖走 `_attachments/`(自行 gitignore + Syncthing,避免 git 變肥)。
+- 主題命名/標題由 LLM 產,品質視模型而定;不滿意可重跑單場。
+
+## 開發
+
+```bash
+pip install pytest -e .
+pytest
+```
+
+## License
+
+MIT — 見 [LICENSE](LICENSE)。

session_to_obsidian-0.1.0/SPEC.md

@@ -0,0 +1,58 @@
+# session-to-obsidian — 專案規格(SPEC)
+
+把 Claude Code session(jsonl)轉成 Obsidian 裡**可快速翻找、可給 Claude compact 後回想、可拿來寫文章**的結構化紀錄。
+
+## 1. 目的 / 受眾
+- **單場回顧 + 寫文章素材**:翻找「某問題當時怎麼解的」,加速回憶;`.md` 是主力。
+- **Claude 跨 compact 的記憶層**:context 被壓縮後,Claude 能回去查前面做了什麼/怎麼解 → 這份結構化紀錄是持久記憶。
+- **受眾**:AI coding agent 使用者 ∩ Obsidian 使用者(交集窄)。**第一受眾是作者本人**(永久使用者容器);開源是順手,不以採用率為成敗。
+
+## 2. 形態 / 發佈
+- **一個自有專案**,發佈成 **Claude Code skill/plugin(SKILL.md + 腳本)**。
+- **語意那步用 host 的 Claude(零外部 LLM 依賴)**;想跑 standalone/cron 再選配外部/地端 LLM(hybrid)。
+- 「skill」只是呼叫格式,骨子是專案。skill 不穩的疑慮 → 把 agent 的工作限到**只有「主題分群」**,其餘全是死板腳本。
+
+## 3. 分工(誰做、要不要 LLM)
+| 步驟 | 誰 | LLM |
+|---|---|---|
+| 解析 jsonl → 步驟/提問/工具/檔案/截圖/時間 | 腳本 | ❌ |
+| **把步驟分到主題串 + 命名** | host agent(隔離 sub-agent) | ✅ 唯一一處 |
+| 組 `.md`(目錄 + 主題分區 + 提問當小標 + 完整作法 + 截圖) | 腳本 | ❌ |
+| 建圖 (b) 單場主題級 / (c) 跨 session | 腳本(有標籤就純機械) | ❌ |
+| 儲存(git 文字 / Syncthing 圖) | 腳本 | ❌ |
+
+## 4. Token 控制(主題分群)
+- 原則 = **不餵全 transcript**,但給「便宜且高鑑別度」的結構訊號。實測(2026-06-19,dcfce9f1):
+  **只餵提問清單會壞** —— context-dependent 的提問(「這是連結」「放到 X 後要 run…」)光看字面無法歸線,
+  agent 會把它併進前一條主題、還生出空主題。**修法 = 提問 + 該步碰的檔名(basename,上限 6)+ 助手首句 gist(上限 160 字)**,
+  量仍有界(每步 ≈ 一兩百字),救回 context-dependent 步驟。`--dump-intents` 已內建這三項。
+- 跑在**隔離 sub-agent(Task)**:不吃主 session context、不害它提早 compact。各場 sub-agent 讀自己的 dump、寫自己的 topics.json。
+- 分群契約:每步恰好一主題(不重不漏,render 端驗證);**主題可非連續**(隔時收尾步歸回原線);複合提問不拆、歸主導線;
+  主題按最早步號排;步號保留原始編號(主題視圖仍可還原時序)。tiny 場(≤5步同一事)→ 1 主題,別硬切。
+
+## 5. 輸出
+- **`<slug>.md`**:每步「我問 + 助手完整敘述 + 工具/檔案 + 截圖」,依主題分區 + 頂部目錄。**核心價值**。導覽靠 Obsidian 原生 Outline/搜尋。
+- **(b) 單場主題級節點圖**:主題串當節點(非每步、非檔案 hub —— 那是毛球教訓)。對「一場多主題」的 session 有用。
+- **(c) 跨 session 圖**:每場一節點。關聯訊號(由強到弱):**共用檔案(權重,設門檻)+ 同專案分簇 + 共用主題標籤(白拿)**;**embedding 語意相似 = opt-in**(新依賴);模糊字串搜尋最弱、不用。避免毛球 → 只用強邊。
+
+## 6. 儲存架構(已定案)
+- **git 只放文字**(.md/.canvas)→ 永不肥。
+- **圖走 Syncthing 的 vault `_attachments/`**(gitignore)→ server↔手機,免 git bloat。
+- 全圖原檔私有封存;文章要用的圖才逐張走 blog pipeline 公開(session 截圖含敏感,別全公開)。
+- 手機:termux SSH(免 PAT)或 Obsidian-Git(HTTPS+PAT);**手機 pull-only**(多寫手會打架);Android 共用儲存 `git config core.filemode false`。
+
+## 7. 未解 / 待做(下次開工)
+- ✅ **主題分群 step**(2026-06-19 完成):`--dump-intents`(提問+files+gist)→ 隔離 sub-agent 分群+命名+總標題 → `--topics` render。20 場已重匯。
+- ✅ **.md 導覽**:頂部 TOC(wikilink heading 連結)+ 主題分區 + 原始步號;助手敘述的 ATX 標題 demote 成粗體、fence 強制配對 → Obsidian Outline 只剩 主題/步驟。
+- ✅ **手刻 mesh 砍掉**(候選3降範圍);單場 `.canvas` 改主題分組彩色框。
+- **(b)(c) 圖的呈現形式**:仍未做。傾向原生 graph(linked notes,scoped local graph)而非手刻布局。跨 session(c)未動。
+- ✅ **檔名帶標題**(2026-06-19):session 資料夾/檔名 = `<date>-<source>-<標題slug>`(跨平台 sanitize + 截長),側欄一眼看出主題,不再是 uuid。
+- ✅ **recall 入口**(2026-06-19):`build_index.py` 掃 `sessions/*/*.md` 產 `INDEX.md`(全 session + 主題地圖,按月分組)。
+  雙受眾:人一頁掃完、Claude compact 後讀這張當 recall 地圖(找到 → 開 .md → Outline 跳主題)。慣例已寫進 vault `CLAUDE.md`。
+  進階(未做):`/recall <query>` skill 直接回傳命中主題(屬候選5打包範圍)。
+- **打包成 skill/plugin**(SKILL.md 把 extract→cluster→render 串起來)+ 選配 standalone CLI。骨架已成形(三段拆分),待寫 SKILL.md。
+
+## 現況(2026-06-19)
+`session_to_obsidian.py`:解析 → `--dump-intents`(提問+files+gist 給分群 agent)→ `--topics`(吃 {title,topics} render 主題分區 .md + TOC + 主題分組 canvas);沒給 `--topics` 退回時間分段。`NOIMG=1` 純文字。
+**20 場全部以主題分群重匯完成**(9 copilot + 11 Claude),coverage/headings/fence parity 全驗證通過。dump/topics 暫存 `/tmp/s2o/`。mesh 已移除。
+待:SKILL.md 打包、recall 入口、(b)(c) 圖、選配重抽圖(目前 text-only)。

session_to_obsidian-0.1.0/pyproject.toml

@@ -0,0 +1,34 @@
+[build-system]
+requires = ["hatchling"]
+build-backend = "hatchling.build"
+
+[project]
+name = "session-to-obsidian"
+version = "0.1.0"
+description = "把 Claude Code / GitHub Copilot 對話 session 變成 Obsidian 裡依語意主題分區、可翻找、可給 Claude compact 後回想的回顧筆記"
+readme = "README.md"
+requires-python = ">=3.10"
+license = "MIT"
+license-files = ["LICENSE"]
+authors = [{ name = "timo9378" }]
+keywords = ["claude-code", "copilot", "obsidian", "llm", "knowledge-management"]
+classifiers = [
+    "Programming Language :: Python :: 3",
+    "Programming Language :: Python :: 3.10",
+    "Programming Language :: Python :: 3.12",
+    "Environment :: Console",
+    "Topic :: Utilities",
+]
+dependencies = []  # 純標準函式庫;語意分群靠本機 `claude` CLI(選用)
+
+[project.scripts]
+s2o = "s2o.cli:main"
+
+[project.urls]
+Homepage = "https://github.com/timo9378/session-to-obsidian"
+Repository = "https://github.com/timo9378/session-to-obsidian"
+Issues = "https://github.com/timo9378/session-to-obsidian/issues"
+Changelog = "https://github.com/timo9378/session-to-obsidian/blob/main/CHANGELOG.md"
+
+[tool.hatch.build.targets.wheel]
+packages = ["s2o"]

session_to_obsidian-0.1.0/s2o/__init__.py

@@ -0,0 +1,2 @@
+"""session-to-obsidian:把 Claude Code / Copilot session 變成 Obsidian 主題回顧。"""
+__version__ = "0.1.0"

session_to_obsidian-0.1.0/s2o/adapters/__init__.py

@@ -0,0 +1,29 @@
+"""來源偵測 + 載入。回傳 (records[Claude格式], meta{date,title})。"""
+from __future__ import annotations
+import json
+from . import claude, copilot
+
+
+def detect(src_path) -> str:
+    """嗅第一行:Copilot append-log 首行有 kind 鍵;Claude 原生是 message/type 訊息。"""
+    with open(src_path, encoding="utf-8") as f:
+        for line in f:
+            if not line.strip():
+                continue
+            try:
+                o = json.loads(line)
+            except Exception:
+                return "claude"
+            return "copilot" if "kind" in o else "claude"
+    return "claude"
+
+
+def load(src_path, source: str = "auto"):
+    if source == "auto":
+        source = detect(src_path)
+    mod = {"claude": claude, "copilot": copilot}.get(source)
+    if mod is None:
+        raise ValueError(f"未知來源:{source}(支援 claude / copilot / auto)")
+    records, meta = mod.load(src_path)
+    meta["source"] = source
+    return records, meta

session_to_obsidian-0.1.0/s2o/adapters/claude.py

@@ -0,0 +1,15 @@
+"""Claude Code 原生 session(jsonl)— 本來就是 Claude 格式,讀進來即可。"""
+from __future__ import annotations
+from ..parse import read_jsonl, session_id
+
+
+def load(src_path):
+    """→ (records, meta)。meta.date = 第一筆 timestamp;session_id 供去重;title 留空。"""
+    records = read_jsonl(src_path)
+    date = ""
+    for r in records:
+        ts = r.get("timestamp")
+        if ts:
+            date = ts
+            break
+    return records, {"date": date, "title": "", "session_id": session_id(records)}