Merge branch 'dongri:main' into main

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "openai-api-rs"
-version = "5.2.3"
+version = "5.2.4"
 edition = "2021"
 authors = ["Dongri Jin <dongrium@gmail.com>"]
 license = "MIT"

README.md

@@ -7,7 +7,7 @@ Check out the [docs.rs](https://docs.rs/openai-api-rs/).
 Cargo.toml
 ```toml
 [dependencies]
-openai-api-rs = "5.2.3"
+openai-api-rs = "5.2.4"
 ```
 
 ## Usage

examples/audio_transcriptions.rs

@@ -12,7 +12,14 @@ async fn main() -> Result<(), Box<dyn std::error::Error>> {
         WHISPER_1.to_string(),
     );
 
-    let result = client.audio_transcription(req).await?;
+    let req_json = req.clone().response_format("json".to_string());
+
+    let result = client.audio_transcription(req_json).await?;
+    println!("{:?}", result);
+
+    let req_raw = req.clone().response_format("text".to_string());
+
+    let result = client.audio_transcription_raw(req_raw).await?;
     println!("{:?}", result);
 
     Ok(())

src/realtime/server_event.rs

@@ -31,7 +31,7 @@ pub struct ConversationCreated {
 #[derive(Debug, Serialize, Deserialize, Clone)]
 pub struct InputAudioBufferCommited {
     pub event_id: String,
-    pub previous_item_id: String,
+    pub previous_item_id: Option<String>,
     pub item_id: String,
 }
 

src/realtime/types.rs

@@ -218,6 +218,8 @@ pub enum ResponseStatus {
 #[derive(Debug, Serialize, Deserialize, Clone)]
 #[serde(tag = "type")]
 pub enum ResponseStatusDetail {
+    #[serde(rename = "cancelled")]
+    Cancelled { reason: CancelledReason },
     #[serde(rename = "incomplete")]
     Incomplete { reason: IncompleteReason },
     #[serde(rename = "failed")]
@@ -230,6 +232,13 @@ pub struct FailedError {
     pub message: String,
 }
 
+#[derive(Debug, Serialize, Deserialize, Clone)]
+#[serde(rename_all = "snake_case")]
+pub enum CancelledReason {
+    TurnDetected,
+    ClientCancelled,
+}
+
 #[derive(Debug, Serialize, Deserialize, Clone)]
 #[serde(rename_all = "snake_case")]
 pub enum IncompleteReason {

src/v1/api.rs

@@ -211,6 +211,13 @@ impl OpenAIClient {
         self.handle_response(response).await
     }
 
+    async fn post_form_raw(&self, path: &str, form: Form) -> Result<Bytes, APIError> {
+        let request = self.build_request(Method::POST, path).await;
+        let request = request.multipart(form);
+        let response = request.send().await?;
+        Ok(response.bytes().await?)
+    }
+
     async fn handle_response<T: serde::de::DeserializeOwned>(
         &self,
         response: Response,
@@ -303,10 +310,34 @@ impl OpenAIClient {
         &self,
         req: AudioTranscriptionRequest,
     ) -> Result<AudioTranscriptionResponse, APIError> {
+        // https://platform.openai.com/docs/api-reference/audio/createTranslation#audio-createtranslation-response_format
+        if let Some(response_format) = &req.response_format {
+            if response_format != "json" && response_format != "verbose_json" {
+                return Err(APIError::CustomError {
+                    message: "response_format must be either 'json' or 'verbose_json' please use audio_transcription_raw".to_string(),
+                });
+            }
+        }
         let form = Self::create_form(&req, "file")?;
         self.post_form("audio/transcriptions", form).await
     }
 
+    pub async fn audio_transcription_raw(
+        &self,
+        req: AudioTranscriptionRequest,
+    ) -> Result<Bytes, APIError> {
+        // https://platform.openai.com/docs/api-reference/audio/createTranslation#audio-createtranslation-response_format
+        if let Some(response_format) = &req.response_format {
+            if response_format != "text" && response_format != "srt" && response_format != "vtt" {
+                return Err(APIError::CustomError {
+                    message: "response_format must be either 'text', 'srt' or 'vtt', please use audio_transcription".to_string(),
+                });
+            }
+        }
+        let form = Self::create_form(&req, "file")?;
+        self.post_form_raw("audio/transcriptions", form).await
+    }
+
     pub async fn audio_translation(
         &self,
         req: AudioTranslationRequest,