Method embed

python/src/embeddings/client.rs:33–55

Generate embedding for a single text CRITICAL: This method releases the GIL during execution, enabling true parallelism from Python threads. Multiple threads can call this method simultaneously.

(&self, py: Python<'_>, text: String)

    /// CRITICAL: This method releases the GIL during execution, enabling true parallelism
    /// from Python threads. Multiple threads can call this method simultaneously.
    fn embed(&self, py: Python<'_>, text: String) -> PyResult<Vec<f32>> {
        // Validate input
        if text.is_empty() {
            return Err(PyErr::new::<pyo3::exceptions::PyValueError, _>(
                "Text input cannot be empty",
            ));
        }

        let service = Arc::clone(&self.service);
        let rt = get_runtime();

        // CRITICAL FIX: Release GIL during async execution
        // This enables multiple Python threads to execute embed() in parallel
        py.allow_threads(|| {
            rt.block_on(async move {
                let response = service
                    .embed_text(&text)
                    .await
                    .map_err(to_py_runtime_error)?;
                Ok(response)
            })
        })
    }

    /// Generate embeddings for multiple texts
    ///