Method stream

python/src/llm/client.rs:653–710

(
        &self,
        prompt: String,
        max_tokens: Option<i64>,
        temperature: Option<f64>,
    )

    #[instrument(skip(self), fields(prompt_len = prompt.len()))]
    #[pyo3(signature = (prompt, max_tokens=None, temperature=None))]
    fn stream(
        &self,
        prompt: String,
        max_tokens: Option<i64>,
        temperature: Option<f64>,
    ) -> PyResult<StreamIterator> {
        // Validate input
        if prompt.is_empty() {
            return Err(validation_error(
                "prompt",
                Some(&prompt),
                "Prompt cannot be empty",
            ));
        }

        // Validate max_tokens
        let validated_max_tokens = if let Some(tokens) = max_tokens {
            if tokens <= 0 {
                return Err(validation_error(
                    "max_tokens",
                    Some(&tokens.to_string()),
                    "max_tokens must be greater than 0",
                ));
            }
            if tokens > u32::MAX as i64 {
                return Err(validation_error(
                    "max_tokens",
                    Some(&tokens.to_string()),
                    "max_tokens is too large",
                ));
            }
            Some(tokens as u32)
        } else {
            None
        };

        // Validate temperature
        let validated_temperature = if let Some(temp) = temperature {
            if !(0.0..=2.0).contains(&temp) {
                return Err(validation_error(
                    "temperature",
                    Some(&temp.to_string()),
                    "temperature must be between 0.0 and 2.0",
                ));
            }
            Some(temp as f32)
        } else {
            None
        };

        Ok(StreamIterator::new(
            Arc::clone(&self.provider),
            self.config.clone(),
            prompt,
            validated_max_tokens,
            validated_temperature,
        ))
    }