Files
attrsum/README.md
T
clawbot 2d996df335
check / check (push) Successful in 2m33s
Run lint and tests as phases of the Dockerfile (closes #5)
The Dockerfile gets a lint phase on the pinned golangci-lint v2.12.2
image and a test phase on the Debian Go image. The tests run as an
unprivileged user, because root reads a file with mode 0000 and the
permission test then fails. The build stage copies a file from each
phase, so no build finishes unless both pass, and it no longer runs
make check. script/lint and script/test each build their phase,
uncached and tagged; script/cibuild bootstraps, runs script/check, then
builds the image. script/bootstrap no longer installs golangci-lint.
README.md and TODO.md describe the new setup.

Model: opus-5-5
2026-10-06 04:32:11 +00:00

117 lines
3.4 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
[**attrsum**](https://git.eeqj.de/sneak/attrsum/) is a **Go
1.22** command-line utility that **adds, updates, verifies, and clears per-file
file content checksums stored in extended attributes (xattrs) on macOS (APFS) and
Linux**, released under the [WTFPL v2](http://www.wtfpl.net/).
Original release 2025-05-08.
Current version 1.0 (2025-05-08).
---
## Getting Started — Quick Build
```bash
# prerequisites: Go 1.22+
git clone https://git.eeqj.de/sneak/attrsum.git
cd attrsum
go build -o attrsum .
```
### Install
```bash
go install git.eeqj.de/sneak/attrsum@latest # into GOPATH/bin or $(go env GOBIN)
```
Semantic Versioning 2.0.0 is used for tags.
---
## Usage
```bash
# add checksum & timestamp xattrs to every regular file under one or more paths
attrsum sum add DIR1 DIR2 file.txt
# update checksum only when file mtime is newer than stored sumtime
attrsum sum update DIR1 DIR2
# verify checksums, stop on first error
attrsum check DIR1 DIR2
# verify every file, reporting each result, keep going after errors
attrsum -v check --continue DIR1 DIR2
# remove checksum & timestamp xattrs
attrsum clear DIR1 DIR2
# read paths from stdin (use - as argument)
find /data -name "*.jpg" | attrsum sum add -
# quiet mode (suppress progress bar and summary)
attrsum -q sum add DIR
```
| xattr key | meaning |
|---------------------------------------------|--------------------------------|
| `user.berlin.sneak.app.attrsum.checksum` | base-58 multihash (sha2-256) |
| `user.berlin.sneak.app.attrsum.sumtime` | RFC 3339 timestamp of checksum |
Flags:
* `-v, --verbose` — per-file log output
* `-q, --quiet` — suppress all output except errors (no progress bar or summary)
* `--exclude PATTERN` — skip paths matching rsync/Doublestar glob
* `--exclude-dotfiles` — skip any path component that starts with `.`
All commands display a progress bar with ETA and print a summary report to stderr on completion (unless `--quiet` is specified).
`attrsum` **never follows symlinks** and skips non-regular files (sockets, devices, …).
---
## Why?
Apple APFS and Linux ext3/ext4 **store no per-file content checksums**, so
silent data corruption can pass unnoticed. `attrsum` keeps a portable checksum **inside each file’s xattrs**, providing integrity
verification that travels with the file itself—no external database
required. Now you can trust a USB stick didn't eat your data.
---
## TODO
Future improvements under consideration:
- **Dry-run mode (`--dry-run`, `-n`)** — show what would be done without making changes
- **JSON output (`--json`)** — machine-readable output for scripting and integration
- **Parallel processing (`-j N`)** — use multiple goroutines for faster checksumming on large trees
- **Exit code documentation** — formalize and document exit codes for scripting
---
## Contributing
* Author & maintainer: **sneak** – <sneak@sneak.berlin>
* Issues / PRs: <https://git.eeqj.de/sneak/attrsum/>
* Code must pass `make check`, which runs the tests and golangci-lint as
phases of the `Dockerfile` (Docker is required) and checks formatting
with `gofmt`.
* No CLA; contributions are under WTFPL v2.
---
## Community & Support
Bug tracker and wiki are in the Gitea repo linked above.
No formal Code of Conduct; be excellent to each other.
---
## License
*Everything is permitted.*
See [WTFPL v2](http://www.wtfpl.net/txt/copying/).