KO | EN

GitLite

Minimalist Open Source Repository Search Engine

gitlite — search

#typescript #ai-agents #cli #open-source #ai #claude-code #codex #deepseek-harness #dsh-plugin #react #windows #developer-tools

shy

★ 43 Open GitHub ↗

Shunting Yard Rule Engine in Rust

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Shy - A Shunting Yard Rules Engine

This is the beginnings of an IOT Rules processing and alerting system. The goals are:

- **Simple syntax**. Engineers can write rule expressions using a simple infix notation.
  - **Simple file format**. Engineers store the rules in simple text files (such as CSV or YAML)
  - **External rules files**. Application loads the rules from these external files, so recompilation is not necessary and formulas are not embedded in the code where they are inaccessible to non-programmers.
  - **Low memory**. Rules engine will require little memory...
  - **Real-time** ... and will be fast enough to use for real-time applications (thus must not use a garbage collector, a feature of **Rust**).
  - **Cached rules**. Compiled rules can be cached and reused.
  - **Business Context**. The rules can read and write from variables stored in business objects (the execution context) that conform to a simple interface.
  - **Sensor statistics**. Streams of measurement data can be processed efficiently, with small amounts of memory required to gather basic statistics like the mean, median, min, max, various quantiles, standard deviation, etc. This requires the use of **frugal streaming** algorithms. 
  - **Off-line mode**. Since sensors may only be connected to the Internet sporadically and lack enough memory to store data for several days, low memory usage is critical!
  - **Sensor rules**. The rules can access the sensor statistics to make judgments.
  - **Actions**. Passing rules can trigger the execution of alerting actions.
  - **Remote update**. If the application is running on an edge device, new rules can be sent to it remotely and hot-swapped with the previous rules.
  - **RESTful Micro service**. The Rules engine can be executed as a REST service, caching sets of pre-compiled rules and executing them against a different context with each request.

The following parts are in working order:

- Expression lexical analysis, parsing, and compilation
  - Expression execution
  - Expression caching
  - Creation of Context objects
  - Execution of expressions against a context in a command line REPL
  - Converting to and from **serde-json** **Value** objects (as preparation for the serialization and deserialization of rules which will be needed to stand up a REST API server.)
  - Starting a Web Service and executing _some_ of the necessary RESTful API routes

For an overview of the rule syntax and how to use the REPL to test expressions interactively (without a Web Service), see this file:

[Shy Rule Syntax and Usage](./src/README.md)

To operate the Web Server and learn the REST API to its commands, read this:

[REST API for Shy Rules Service](./src/service/README.md)

(Note: The above link lists which REST routes have been implemented so far.)

shy

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README