Splitters Crate

Recoco Splitters

Intelligent text splitting and parsing for Recoco.

This crate implements sophisticated text splitting strategies, primarily leveraging Tree-sitter to perform syntax-aware chunking of source code and structured documents.

🚀 Why Tree-sitter?

Standard text splitters often break code in the middle of functions or classes, destroying context. recoco-splitters understands the syntax of the language it is processing, ensuring that chunks respect logical boundaries (e.g., keeping a whole function together).

📦 Supported Languages

To minimize binary size, Recoco feature-gates every language parser. Enable only what you need in your Cargo.toml.

[dependencies]
recoco-splitters = { version = "...", features = ["python", "rust"] }

Feature	Language
all	all languages
c	C
c-sharp	C#
cpp	C++
css	CSS
fortran	Fortran
go	Go
html	HTML
java	Java
javascript	JavaScript
json	JSON
kotlin	Kotlin
markdown	Markdown
php	PHP
python	Python
r	R
ruby	Ruby
rust	Rust
scala	Scala
solidity	Solidity
sql	SQL
swift	Swift
toml	TOML
typescript	TypeScript
xml	XML
yaml	YAML

🧩 Splitter Strategies

• Recursive Character Splitter: Standard splitting by separators (paragraphs, newlines, etc.).
• Recursive Syntax Splitter: Tree-sitter based splitting that respects code blocks and syntax nodes.

📄 License

Apache-2.0. See main repository for details.