# spell-rs
[](https://crates.io/crates/spell)
[](https://travis-ci.org/nbigaouette/spell-rs)
[](https://docs.rs/spell)
[](#license)
_spell-rs_ is a _Spell_ implementation in Rust.
Spell is a _**S**treaming **P**arser for **E**vent **L**ogs using an **L**CS_ (Longest Common Subsequence).
The minimum Rust version is 1.31 due to the `use crate::` syntax.
## Reference
Min Du and Feifei Li. Spell: Streaming Parsing of System Event Logs. 2016 IEEE 16th International
Conference on Data Mining (ICDM) pp. 859-864, Barcelona, Spain, December, 2016.
[DOI: 10.1109/ICDM.2016.0103](https://www.doi.org/10.1109/ICDM.2016.0103).
https://ieeexplore.ieee.org/document/7837916
* Original paper: https://www.cs.utah.edu/~lifeifei/papers/spell.pdf
* Presentation slides: https://www.cs.utah.edu/~mind/papers/spell_slides.pdf
## Details
Spell ingests log lines one by one and build up a list of formatting strings that can be used to
analyze the logs. For example (see [presentation slides](https://www.cs.utah.edu/~mind/papers/spell_slides.pdf),
page 19), given the two log entries:
```text
Temperature (41C) exceeds warning threshold
Temperature (42C, 43C) exceeds warning threshold
```
the LCS is:
```text
Temperature * exceeds warning threshold
```
which can then be used to analyze the logs.
As of cedf57a7a73b1052de937d13150c2b9d8a03237c, the Rust implementation gives the exact same results
as the [Java one](https://github.com/EddiePi/Spell).
## Python Wrapper
Universal Python package (wheel) can be built on all platforms using:
```sh
cd python
# Compile Rust code
pip install --editable .
# Test Python package
python ./test.py
# Build a wheel
python setup.py --verbose bdist_wheel
```
### Docker
This only requires Docker.
```sh
# Build a Docker image containing everything required
./docker/docker_build.sh
# Run the commands to build wheel inside a docker container
./docker/docker_run.sh
```
## Other Implementations
* [Spell](https://github.com/EddiePi/Spell), Java version
* [pyspell](https://github.com/bave/pyspell), a Python version
* [spell](https://github.com/logforensicator/spell), another Python version
## License
This project is licensed under either of
* Apache License, Version 2.0, ([LICENSE-APACHE](LICENSE-APACHE) or
http://www.apache.org/licenses/LICENSE-2.0)
* MIT license ([LICENSE-MIT](LICENSE-MIT) or
http://opensource.org/licenses/MIT)
at your option.