fastembed 4.4.0

Rust implementation of https://github.com/qdrant/fastembed
Documentation
use std::path::{Path, PathBuf};

use ort::{execution_providers::ExecutionProviderDispatch, session::Session};
use tokenizers::Tokenizer;

use crate::{models::sparse::SparseModel, TokenizerFiles, DEFAULT_CACHE_DIR};

use super::{DEFAULT_EMBEDDING_MODEL, DEFAULT_MAX_LENGTH};

/// Options for initializing the SparseTextEmbedding model
#[derive(Debug, Clone)]
#[non_exhaustive]
pub struct SparseInitOptions {
    pub model_name: SparseModel,
    pub execution_providers: Vec<ExecutionProviderDispatch>,
    pub max_length: usize,
    pub cache_dir: PathBuf,
    pub show_download_progress: bool,
}

impl SparseInitOptions {
    pub fn new(model_name: SparseModel) -> Self {
        Self {
            model_name,
            ..Default::default()
        }
    }

    pub fn with_max_length(mut self, max_length: usize) -> Self {
        self.max_length = max_length;
        self
    }

    pub fn with_cache_dir(mut self, cache_dir: PathBuf) -> Self {
        self.cache_dir = cache_dir;
        self
    }

    pub fn with_execution_providers(
        mut self,
        execution_providers: Vec<ExecutionProviderDispatch>,
    ) -> Self {
        self.execution_providers = execution_providers;
        self
    }

    pub fn with_show_download_progress(mut self, show_download_progress: bool) -> Self {
        self.show_download_progress = show_download_progress;
        self
    }
}

impl Default for SparseInitOptions {
    fn default() -> Self {
        Self {
            model_name: DEFAULT_EMBEDDING_MODEL,
            execution_providers: Default::default(),
            max_length: DEFAULT_MAX_LENGTH,
            cache_dir: Path::new(DEFAULT_CACHE_DIR).to_path_buf(),
            show_download_progress: true,
        }
    }
}

/// Struct for "bring your own" embedding models
///
/// The onnx_file and tokenizer_files are expecting the files' bytes
#[derive(Debug, Clone, PartialEq, Eq)]
#[non_exhaustive]
pub struct UserDefinedSparseModel {
    pub onnx_file: Vec<u8>,
    pub tokenizer_files: TokenizerFiles,
}

impl UserDefinedSparseModel {
    pub fn new(onnx_file: Vec<u8>, tokenizer_files: TokenizerFiles) -> Self {
        Self {
            onnx_file,
            tokenizer_files,
        }
    }
}

/// Rust representation of the SparseTextEmbedding model
pub struct SparseTextEmbedding {
    pub tokenizer: Tokenizer,
    pub(crate) session: Session,
    pub(crate) need_token_type_ids: bool,
    pub(crate) model: SparseModel,
}