Skip to content

release-validation.md

Latest commit

 

History

History
138 lines (98 loc) · 4.48 KB

release-validation.md

File metadata and controls

138 lines (98 loc) · 4.48 KB
Raw

Release Validation

Last updated: 2026-04-22

This document keeps two truths separate:

  • the stable release baseline (v2.4.0)
  • the current main branch snapshot, which already contains newer AI + Obsidian improvements

Stable release baseline: v2.4.0

The checks used to cut v2.4.0 remain the minimum baseline for future stable tags.

Installation validation

bash scripts/bootstrap.sh --install-python
bash scripts/bootstrap.sh
.venv/bin/python -m py_compile scripts/process_share_links.py scripts/run_regression.py
.venv/bin/python -m unittest discover -s tests -v

Latest stable-baseline outcome on 2026-04-19:

  • bootstrap.sh --install-python: passed
  • bootstrap.sh: passed
  • py_compile: passed
  • unittest: passed (33 tests at the time)

Automated live regression

Public-link regression entrypoint:

.venv/bin/python scripts/run_regression.py --preset extended --analysis-mode heuristic

extended covers:

  • GitHub
  • Zhihu
  • CSDN
  • Toutiao
  • Bilibili

Stable-baseline outcome on 2026-04-19: passed (5/5 success)

Platform Result Extraction path
GitHub success github api + readme
Zhihu success scrapling fetch [.Post-RichText]
CSDN success scrapling fetch [#article_content]
Toutiao success scrapling stealthy-fetch [.article-content]
Bilibili success yt-dlp download + whisper-cli

Current main branch snapshot

These checks describe the repository as it exists on main, not the last stable tag.

Local verification on 2026-04-22

.venv/bin/python -m py_compile scripts/process_share_links.py scripts/douyin_auth.py scripts/run_regression.py
.venv/bin/python -m unittest discover -s tests -v

Outcome:

  • py_compile: passed
  • unittest: passed (67 tests)

Representative real outputs currently present

These are not abstract claims; they correspond to real notes / reports already generated in a working vault.

Source type Result Extraction path
GitHub repo success deepwiki overview
Douyin short video success playwright douyin download + whisper-cli

Representative notes already written by the current pipeline:

  • NousResearch/hermes-agent
  • OpenCode 保姆级配置与实战指南

Manual real-link regression notes

In addition to the public preset above, the project has manually verified representative public share links during the hardening cycles around 2026-03-26, 2026-04-19, and 2026-04-22.

Platform Result Typical extraction path
GitHub success deepwiki overview or github api + readme
Zhihu success article extraction
CSDN success article extraction
Toutiao success scrapling stealthy-fetch [.article-content]
Bilibili success yt-dlp download + whisper-cli
WeChat success scrapling stealthy-fetch [#js_content]
Xiaohongshu success yt-dlp download + whisper-cli
X/Twitter success yt-dlp download + whisper-cli
Weibo partial yt-dlp metadata only on very short clips
Douyin success playwright douyin download + whisper-cli

Notes:

  • partial is an honest outcome, not a hidden failure.
  • social-platform success rates can change over time because anti-bot behavior changes
  • browser cookies, saved auth, and referer support exist for users who need stronger access stability
  • Feishu / Feishu Wiki upload is intentionally outside the current release scope

Public CI vs self-hosted runner

Recommended split:

  • public CI:
    • runtime bootstrap
    • compile checks
    • unit tests
    • lightweight public-link regression
    • mocked Douyin auth gating and fallback logic
  • self-hosted runner / desktop smoke test:
    • real Douyin QR login
    • saved-cookie reuse
    • Playwright media fallback on a real share link

Important note:

  • public CI should not attempt real Douyin QR login
  • real QR login only makes sense in a self-hosted runner, VNC session, or remote desktop where a human can actually scan the code
  • if you intentionally use such an environment, open the gate explicitly with CONTENT_PROCESSOR_ALLOW_NON_TTY_DOUYIN_LOGIN=1

Stable release gate

Before publishing a new stable tag:

  • the stable-baseline installation checks should pass on a clean runtime
  • unit tests should remain green
  • at least one GitHub link, one article page, one dynamic page, and one media link should succeed
  • mixed-outcome platforms such as Weibo should be documented honestly instead of being hidden