The post_install initializes var/archivebox as the data directory,
but users need to know where it is for subsequent commands. The
caveats block is shown after brew install/upgrade to guide users.
https://claude.ai/code/session_01Vx1EsNrNySgsc8Y67dGzCn
preremove.sh stops the service during upgrades (to avoid stale venv
binaries), but postinstall.sh wasn't restarting it. Now postinstall
checks if the service was enabled and restarts it after the venv is
rebuilt, so upgrades don't leave a previously running service down.
https://claude.ai/code/session_01Vx1EsNrNySgsc8Y67dGzCn
- preremove.sh: Stop the systemd service during upgrades (not just
remove/purge) so the running process doesn't use stale venv binaries
while postinstall replaces them. Only disable + remove venv on full
removal.
- debian.yml: Fail loudly when release lookup fails during a release
event (exit 1), but still skip gracefully for workflow_dispatch
manual testing (exit 0). Prevents silently broken .deb publication.
- archivebox.rb + build_brew.sh + homebrew.yml: Add post_install that
initializes ArchiveBox in var/archivebox with DATA_DIR set, using
correct Homebrew system() syntax (no env hash as first arg).
https://claude.ai/code/session_01Vx1EsNrNySgsc8Y67dGzCn
docker.yml uses github.event_name to decide between full release tags
(semver, latest) vs CI-only tags (branch, sha). When release.yml calls
it via `uses:`, the child sees event_name='workflow_call' which was
hitting the non-release path. Now workflow_call is treated the same as
workflow_dispatch so published releases get proper Docker tags.
https://claude.ai/code/session_01Vx1EsNrNySgsc8Y67dGzCn
- install.sh: Replace awk float comparison with integer major/minor check
so Python 3.9 is correctly rejected as < 3.13
- nfpm.yaml: Add git/curl/wget as recommends so users have basic archiving
tools out of the box without needing `archivebox install`
- debian.yml: Restore git/curl/wget in CI smoke test to match what the
package recommends, instead of relying on runner preinstalls
- debian.yml: Guard release upload to skip gracefully when no GitHub
Release exists (fixes workflow_dispatch failures)
https://claude.ai/code/session_01Vx1EsNrNySgsc8Y67dGzCn
The release job was duplicating ~70 lines of PyPI fetching + formula
generation that build_brew.sh already handles. Now it just calls
./bin/build_brew.sh. Also merged the two Linux-only brew setup steps.
https://claude.ai/code/session_01Vx1EsNrNySgsc8Y67dGzCn
Remove all non-essential dependencies from both package formats.
The .deb now only depends on python3, pip, and venv. The brew
formula only depends on python@3.13. All other runtime deps
(node, chrome, yt-dlp, wget, ripgrep, etc.) are installed
on-demand by `archivebox install` at runtime.
Removed from brew formula:
- 6 depends_on entries (node, git, wget, curl, ripgrep, yt-dlp)
- on_linux block (pkg-config, openssl, libffi)
- post_install hook (was running archivebox install)
- service block (users run archivebox server directly)
Removed from .deb:
- 6 depends entries (nodejs, npm, git, wget, curl, ripgrep)
- recommends section (yt-dlp, ffmpeg, chromium)
Net: -126 lines across packaging files.
https://claude.ai/code/session_01Vx1EsNrNySgsc8Y67dGzCn
- systemd service: use /usr/bin/archivebox wrapper (exports venv PATH
for bundled tools like yt-dlp) instead of direct venv binary
- install.sh: prefer python3.13, fail early with clear error if < 3.13,
add comment clarifying the manual (unpinned) fallback behavior
- debian.yml release job: fall back to pyproject.toml version when
github.event.release.tag_name is empty (workflow_dispatch path)
- nfpm.yaml: clarify that install.sh enforces the real >= 3.13 constraint
- CI pre-seed step: expanded comment explaining why we pre-seed with
python3.13 and how it relates to real installs
https://claude.ai/code/session_01Vx1EsNrNySgsc8Y67dGzCn
- Pin cache-apt-pkgs-action to commit SHA for supply-chain safety
- Fix Homebrew post_install to use with_env block instead of env hash
in system() call (idiomatic Homebrew pattern)
- Add clarifying comments to service file, preremove.sh, and nfpm.yaml
explaining user/group creation, directory ownership, and upgrade handling
https://claude.ai/code/session_01Vx1EsNrNySgsc8Y67dGzCn
- Move .deb upload to GitHub Release into a separate job that runs after tests pass
- Fix workflow_call event propagation so release jobs run when called from release.yml
- Fix setup.sh post-install verification to check `which archivebox` first (works for brew/deb)
- Fix README.md: detect architecture with dpkg instead of hardcoding amd64
- Fix README.md: remove --setup flag from apt install instructions
https://claude.ai/code/session_01Vx1EsNrNySgsc8Y67dGzCn
- Remove --setup flag from systemd service and CI (not valid in 0.9.x)
- Remove release triggers from debian/homebrew workflows (handled by release.yml)
- Fix brew post_install to set DATA_DIR so it initializes in var/archivebox
- Add PATH export to deb wrapper script for bundled console scripts
- Remove pip install fallback in install.sh (strict version pinning)
- Guard preremove.sh cleanup to only run on remove/purge, not upgrade
- Initialize SDIST_URL/SDIST_SHA256 in build_brew.sh (nounset safety)
- Pin awalsh128/cache-apt-pkgs-action to v1.6.0 (supply chain safety)
https://claude.ai/code/session_01Vx1EsNrNySgsc8Y67dGzCn
- test-parallel.yml: mkdir -p tests/out before pytest --basetemp
(fixes FileNotFoundError in chrome test fixture)
- debian.yml: fix archivebox add command (--parser url_list removed
in 0.9.x), remove || true so failures are caught
- setup.sh: revert apt section to always use pip install, not .deb
https://claude.ai/code/session_01Vx1EsNrNySgsc8Y67dGzCn
- homebrew.yml: Build local sdist, generate formula with file:// URL and
real resource stanzas via homebrew-pypi-poet, run full
`brew install --build-from-source` on both macOS and Linux (Linuxbrew)
- debian.yml: Pre-seed venv with local wheel before dpkg install so
postinstall succeeds even for unreleased versions; test init/status/add
- Both workflows trigger on push (path-filtered) and release
- Release job generates formula with PyPI URL and pushes to tap
https://claude.ai/code/session_01Vx1EsNrNySgsc8Y67dGzCn
- Fix debian.yml: pin nfpm version, add permissions, improve test job
with user creation, init test, and status check
- Fix homebrew.yml: use PyPI JSON API (macOS-compatible, no grep -oP),
wait for PyPI availability on release, use generated formula not template,
add Linux (Linuxbrew) test job alongside macOS
- Add release.yml orchestrator: pip → deb + brew + docker in order
- Add workflow_call triggers to pip.yml and docker.yml
- Fix build_brew.sh: replace grep -oP with Python-based PyPI API,
add on_linux deps (pkg-config, openssl, libffi)
- Fix setup.sh: use GitHub API to find correct .deb download URL
(filename includes version number)
- Fix postinstall.sh: create archivebox system user, pin version from
package, check for systemd before daemon-reload
- Fix preremove.sh: stop service before removal, check for systemd
- Fix install.sh: fallback to latest if pinned version not on PyPI
- Add on_linux deps to brew formula for Linuxbrew support
- Tested: .deb builds, installs, creates user, runs archivebox init
https://claude.ai/code/session_01Vx1EsNrNySgsc8Y67dGzCn
- Add Homebrew formula (brew_dist/archivebox.rb) using virtualenv pattern
with auto-generation via homebrew-pypi-poet in bin/build_brew.sh
- Add Debian packaging via nFPM (pkg/debian/) with thin .deb that pip-installs
archivebox into /opt/archivebox/venv on postinstall
- Add build/release scripts: bin/{build,release}_{brew,deb}.sh
- Update CI workflows to build packages on release and test them
- Update README apt/brew install instructions with working commands
- Update bin/setup.sh to use .deb download instead of old Launchpad PPA
https://claude.ai/code/session_01Vx1EsNrNySgsc8Y67dGzCn
Fixes#1139
## Summary
This PR fixes: Feature Request: Add AI-assisted summarization, tagging,
search, and more using LLMs / RAG
## Changes
```
archivebox/core/models.py | 12 +++++++-----
1 file changed, 7 insertions(+), 5 deletions(-)
```
## Testing
Please review the changes carefully. The fix was verified against the
existing test suite.
---
*This PR was created with the assistance of Claude Sonnet 4.6 by
Anthropic | effort: low. Happy to make any adjustments!*
<!-- This is an auto-generated description by cubic. -->
---
## Summary by cubic
Returns tags as a JSON array in Snapshot.to_dict() and accepts both list
and comma-separated tags in from_json(), making search exports and
RAG/LLM integrations easier. Fixes#1139.
- **New Features**
- Tags export is now a sorted JSON list for deterministic output.
- Imports accept list or string formats; trims whitespace and
deduplicates tags for compatibility.
<sup>Written for commit 08b0dfaf12.
Summary will update on new commits.</sup>
<!-- End of auto-generated description by cubic. -->
Previously, `archivebox search --json` exported tags as a comma-separated
string (e.g. "tag1,tag2"), which required manual parsing by consumers like
LlamaIndex, LangChain, and other RAG frameworks.
Now `to_dict()` returns tags as a proper JSON array (e.g. ["tag1", "tag2"]),
making the export directly usable as structured metadata in LLM/RAG pipelines
without additional preprocessing.
`from_json()` is updated to accept both list and string formats for backward
compatibility with existing JSON imports.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
<!-- IMPORTANT: Do not submit PRs with only formatting / PEP8 / line
length changes. -->
# Summary
Add the maintainer info of the ArchiveBox AUR package for
accountability. Much of the packaging has changed since the time of its
initial contribution and I as the current maintainer will make sure
these changes will work smoothly moving forward. I will also make sure
this AUR package will be up to date once the 0.9.x branch is released.
# Related issues
<!-- e.g. #123 or Roadmap goal #
https://github.com/pirate/ArchiveBox/wiki/Roadmap -->
# Changes these areas
- [ ] Bugfixes
- [ ] Feature behavior
- [ ] Command line interface
- [ ] Configuration options
- [ ] Internal architecture
- [ ] Snapshot data layout on disk
<!-- This is an auto-generated description by cubic. -->
---
## Summary by cubic
Update README to tag the current maintainer of the Arch AUR package.
Adds “maintained by @jasongodev” next to the original contributor to
improve accountability and clarify support.
<sup>Written for commit 0d05fd8c53.
Summary will update on new commits.</sup>
<!-- End of auto-generated description by cubic. -->
<!-- IMPORTANT: Do not submit PRs with only formatting / PEP8 / line
length changes. -->
# Summary
This PR fixes the docker image build. Also fixes the uuid7 not found
error on the first run of `archivebox init`.
<!-- This is an auto-generated description by cubic. -->
---
## Summary by cubic
Fixes the Docker image build and the uuid7 error on first init. We now
use uv-managed Python 3.13 and patch uuid.uuid7 before Django
migrations.
- **Bug Fixes**
- Docker: switch to uv-managed Python, create venv with uv --python,
skip version check at build, and start with --init.
- UUID7: add uuid_compat, import it early, and monkey-patch uuid.uuid7
on <3.14 to keep migrations working.
- **Dependencies**
- Bump Python to 3.13.
- Require uuid_extensions on Python <3.14.
<sup>Written for commit 9aa4f0de58.
Summary will update on new commits.</sup>
<!-- End of auto-generated description by cubic. -->
## Summary
Implements native LDAP authentication support for ArchiveBox.
## Changes
- Create `archivebox/config/ldap.py` with LDAPConfig class
- Create `archivebox/ldap/` Django app with custom auth backend
- Update `core/settings.py` to conditionally load LDAP when enabled
- Add LDAP_CREATE_SUPERUSER support to auto-grant superuser privileges
- Add comprehensive tests in test_auth_ldap.py (no mocks, no skips)
- LDAP only activates if django-auth-ldap is installed and
LDAP_ENABLED=True
- Helpful error messages when LDAP libraries are missing or config is
incomplete
## Implementation Approach
- ✅ Native integration (not a plugin)
- ✅ Conditional loading based on libraries + config
- ✅ Separate Django app for LDAP logic
- ✅ Clean if statements in settings.py
- ✅ No mixing LDAP code with rest of codebase
Fixes#1664🤖 Generated with [Claude Code](https://claude.ai/code)
- Create archivebox/config/ldap.py with LDAPConfig class
- Create archivebox/ldap/ Django app with custom auth backend
- Update core/settings.py to conditionally load LDAP when enabled
- Add LDAP_CREATE_SUPERUSER support to auto-grant superuser privileges
- Add comprehensive tests in test_auth_ldap.py (no mocks, no skips)
- LDAP only activates if django-auth-ldap is installed and LDAP_ENABLED=True
- Helpful error messages when LDAP libraries are missing or config is incomplete
Fixes#1664
Co-authored-by: Nick Sweeting <pirate@users.noreply.github.com>
