Add response_format option

src/client.rs

@@ -157,6 +157,7 @@ impl ChatGPT {
                 temperature: self.config.temperature,
                 top_p: self.config.top_p,
                 max_tokens: self.config.max_tokens,
+                response_format: self.config.response_format,
                 frequency_penalty: self.config.frequency_penalty,
                 presence_penalty: self.config.presence_penalty,
                 reply_count: self.config.reply_count,
@@ -198,6 +199,7 @@ impl ChatGPT {
                 temperature: self.config.temperature,
                 top_p: self.config.top_p,
                 max_tokens: self.config.max_tokens,
+                response_format: self.config.response_format,
                 frequency_penalty: self.config.frequency_penalty,
                 presence_penalty: self.config.presence_penalty,
                 reply_count: self.config.reply_count,
@@ -230,6 +232,7 @@ impl ChatGPT {
                 temperature: self.config.temperature,
                 top_p: self.config.top_p,
                 max_tokens: self.config.max_tokens,
+                response_format: self.config.response_format,
                 frequency_penalty: self.config.frequency_penalty,
                 presence_penalty: self.config.presence_penalty,
                 reply_count: self.config.reply_count,
@@ -273,6 +276,7 @@ impl ChatGPT {
                 temperature: self.config.temperature,
                 top_p: self.config.top_p,
                 max_tokens: self.config.max_tokens,
+                response_format: self.config.response_format,
                 frequency_penalty: self.config.frequency_penalty,
                 presence_penalty: self.config.presence_penalty,
                 reply_count: self.config.reply_count,
@@ -370,6 +374,7 @@ impl ChatGPT {
                 stream: false,
                 temperature: self.config.temperature,
                 top_p: self.config.top_p,
+                response_format: self.config.response_format,
                 frequency_penalty: self.config.frequency_penalty,
                 presence_penalty: self.config.presence_penalty,
                 reply_count: self.config.reply_count,
@@ -407,6 +412,7 @@ impl ChatGPT {
                 stream: false,
                 temperature: self.config.temperature,
                 top_p: self.config.top_p,
+                response_format: self.config.response_format,
                 frequency_penalty: self.config.frequency_penalty,
                 presence_penalty: self.config.presence_penalty,
                 reply_count: self.config.reply_count,

src/config.rs

@@ -1,5 +1,5 @@
-use std::{fmt::Display, str::FromStr};
 use std::time::Duration;
+use std::{fmt::Display, str::FromStr};
 
 #[cfg(feature = "functions")]
 use crate::functions::FunctionValidationStrategy;
@@ -18,6 +18,8 @@ pub struct ModelConfiguration {
     pub top_p: f32,
     /// Controls the maximum number of tokens to generate in the completion
     pub max_tokens: Option<u32>,
+    /// Format of the response from the model, e.g., text or json_object.
+    pub response_format: ResponseFormat,
     /// Determines how much to penalize new tokens passed on their existing presence so far
     pub presence_penalty: f32,
     /// Determines how much to penalize new tokens based on their existing frequency so far
@@ -40,6 +42,7 @@ impl Default for ModelConfiguration {
             temperature: 0.5,
             top_p: 1.0,
             max_tokens: None,
+            response_format: ResponseFormat::default(),
             presence_penalty: 0.0,
             frequency_penalty: 0.0,
             reply_count: 1,
@@ -51,6 +54,38 @@ impl Default for ModelConfiguration {
     }
 }
 
+/// Specifies the format of the response.
+#[derive(Serialize, Debug, Clone, Copy, PartialEq, PartialOrd, Builder)]
+#[builder(default, setter(into))]
+pub struct ResponseFormat {
+    /// The type of format for the response.
+    pub format_type: FormatType,
+}
+
+impl Default for ResponseFormat {
+    fn default() -> Self {
+        Self {
+            format_type: FormatType::default(),
+        }
+    }
+}
+
+/// Specifies the type of the format.
+#[derive(Serialize, Debug, Clone, Copy, PartialEq, PartialOrd)]
+#[serde(rename_all = "snake_case")]
+pub enum FormatType {
+    /// Standard text format.
+    Text,
+    /// JSON object format.
+    JsonObject,
+}
+
+impl Default for FormatType {
+    fn default() -> Self {
+        FormatType::Text
+    }
+}
+
 /// The engine version for ChatGPT
 #[derive(Serialize, Debug, Default, Copy, Clone, PartialEq, PartialOrd)]
 #[allow(non_camel_case_types)]

src/types.rs

@@ -1,3 +1,4 @@
+use crate::config::ResponseFormat;
 #[cfg(feature = "functions")]
 use crate::functions::FunctionCall;
 use serde::{Deserialize, Deserializer, Serialize};
@@ -85,6 +86,8 @@ pub struct CompletionRequest<'a> {
     pub stream: bool,
     /// The extra randomness of response
     pub temperature: f32,
+    /// Format of the response from the model, e.g., text or json_object.
+    pub response_format: ResponseFormat,
     /// Controls diversity via nucleus sampling, not recommended to use with temperature
     pub top_p: f32,
     /// Controls the maximum number of tokens to generate in the completion