feat(native): doc loader for common native (#9941)

This commit is contained in:
darkskygit
2025-02-25 07:50:56 +00:00
parent 26674b0cb8
commit 842c39c3be
42 changed files with 2989 additions and 36 deletions

View File

@@ -0,0 +1,37 @@
use text_splitter::ChunkConfig;
/**
* modified from https://github.com/Abraxas-365/langchain-rust/tree/v4.6.0/src/text_splitter
*/
use super::*;
#[derive(Debug, Clone)]
pub struct TokenSplitter {
splitter_options: SplitterOptions,
}
impl Default for TokenSplitter {
fn default() -> Self {
TokenSplitter::new(SplitterOptions::default())
}
}
impl TokenSplitter {
pub fn new(options: SplitterOptions) -> TokenSplitter {
TokenSplitter {
splitter_options: options,
}
}
}
impl TextSplitter for TokenSplitter {
fn split_text(&self, text: &str) -> Result<Vec<String>, TextSplitterError> {
let chunk_config = ChunkConfig::try_from(&self.splitter_options)?;
Ok(
text_splitter::TextSplitter::new(chunk_config)
.chunks(text)
.map(|x| x.to_string())
.collect(),
)
}
}