feat: all native features on wasm

async-openai-macros/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "async-openai-macros"
-version = "0.2.0"
+version = "0.3.0"
 authors = ["Himanshu Neema"]
 keywords = ["openai", "macros", "ai"]
 description = "Macros for async-openai"
@@ -14,9 +14,6 @@ readme = "README.md"
 [lib]
 proc-macro = true
 
-[features]
-middleware = []
-
 [dependencies]
 syn = { version = "2.0", features = ["full"] }
 quote = "1.0"

async-openai-macros/src/lib.rs

@@ -2,7 +2,7 @@ use proc_macro::TokenStream;
 use quote::{quote, ToTokens};
 use syn::{
     parse::{Parse, ParseStream},
-    parse_macro_input, parse_quote,
+    parse_macro_input,
     punctuated::Punctuated,
     token::Comma,
     FnArg, GenericParam, Generics, ItemFn, Pat, PatType, TypeParam, WhereClause,
@@ -57,7 +57,6 @@ pub fn byot(args: TokenStream, item: TokenStream) -> TokenStream {
     let input = parse_macro_input!(item as ItemFn);
     let mut new_generics = Generics::default();
     let mut param_count = 0;
-    let middleware_enabled = cfg!(feature = "middleware");
 
     // Process function arguments
     let mut new_params = Vec::new();
@@ -83,18 +82,6 @@ pub fn byot(args: TokenStream, item: TokenStream) -> TokenStream {
                         {
                             type_param.bounds.extend(vec![bound.clone()]);
                         }
-                        let needs_middleware_replay_bounds =
-                            bounds_args.bounds.iter().any(|(name, bound)| {
-                                name == &generic_name
-                                    && bound.to_token_stream().to_string().contains("Clone")
-                                    && !bound.to_token_stream().to_string().contains("Display")
-                            });
-                        if middleware_enabled && needs_middleware_replay_bounds {
-                            type_param
-                                .bounds
-                                .push(parse_quote!(crate::middleware::MiddlewareInput));
-                        }
-
                         new_params.push(GenericParam::Type(type_param));
                         param_count += 1;
                         quote! { #pat_ident: #generic_ident }
@@ -136,7 +123,7 @@ pub fn byot(args: TokenStream, item: TokenStream) -> TokenStream {
 
     // Generate return type based on stream flag
     let return_type = if bounds_args.stream {
-        quote! { Result<::std::pin::Pin<Box<dyn ::futures::Stream<Item = Result<R, OpenAIError>> + Send>>, OpenAIError> }
+        quote! { Result<crate::types::stream::StreamResponse<R>, OpenAIError> }
     } else {
         quote! { Result<R, OpenAIError> }
     };

async-openai/Cargo.toml

@@ -49,7 +49,7 @@ chat-completion = ["chat-completion-types", "_api"]
 assistant = ["assistant-types", "_api", ]
 administration = ["administration-types", "_api"]
 completions = ["completion-types", "_api"]
-middleware = ["dep:tower", "_api", "async-openai-macros?/middleware"]
+middleware = ["dep:tower", "_api"]
 
 # Type feature flags - these enable only the types
 response-types = ["dep:derive_builder"]
@@ -162,8 +162,9 @@ bytes = { version = "1.11", optional = true }
 
 # API dependencies - only needed when API features are enabled
 # We use a feature gate to enable these when any API feature is enabled
-async-openai-macros = { path = "../async-openai-macros", version = "0.2.0", optional = true }
+async-openai-macros = { path = "../async-openai-macros", version = "0.3.0", optional = true }
 base64 = { version = "0.22", optional = true }
+futures = { version = "0.3", optional = true }
 rand = { version = "0.9", optional = true }
 reqwest = { version = "0.13", features = [
   "json",
@@ -177,18 +178,17 @@ secrecy = { version = "0.10", features = ["serde"], optional = true }
 serde_urlencoded = { version = "0.7", optional = true }
 url = { version = "2.5", optional = true }
 tower = { version = "0.5", features = ["limit", "retry", "timeout", "util"], optional = true }
+eventsource-stream = { version = "0.2", optional = true }
 ## For Webhook signature verification
 hmac = { version = "0.12", optional = true, default-features = false}
 sha2 = { version = "0.10", optional = true, default-features = false }
 hex = { version = "0.4", optional = true, default-features = false }
 
-## API Non-WASM dependencies (streaming and retry is not implemented for WASM yet)
+## API Non-WASM dependencies
 [target.'cfg(not(target_family = "wasm"))'.dependencies]
-futures = { version = "0.3", optional = true }
 tokio = { version = "1", features = ["fs", "macros"], optional = true }
 tokio-stream = { version = "0.1", optional = true }
 tokio-util = { version = "0.7", features = ["codec", "io-util"], optional = true }
-eventsource-stream = { version = "0.2", optional = true }
 ## For Realtime websocket
 tokio-tungstenite = { version = "0.28", optional = true, default-features = false }
 

async-openai/MIDDLEWARE.md

@@ -91,22 +91,17 @@ Custom tower retry policies can call `middleware::retry::should_retry` to reuse
 
 On native targets retries wait using `tokio::time::sleep`. On WASM retries are immediate.
 
-## Native and WASM bounds
-
-The conceptual middleware boundary stays the same; only the platform thread-safety bounds differ.
+## Error Handling
 
-On native targets, middleware services installed with `Client::with_http_service` must be `Send + Sync + 'static` and return `Send + 'static` futures.
+`OpenAIError::Boxed` is available only when the `middleware` feature is enabled.
 
-On WASM targets, middleware services and futures must be `'static`.
+Custom middleware services installed with `Client::with_http_service` may use any error type that implements `Into<OpenAIError>`. This lets middleware preserve structured errors when it has a dedicated `OpenAIError` conversion.
 
-## Bring Your Own Types Interaction
+Tower's `BoxError` converts into `OpenAIError::Boxed`, which is useful for generic tower layers whose concrete error type is erased. Callers can still downcast the boxed error when they know the original error type.
 
-With the `byot` feature, generated `*_byot` methods keep minimal trait bounds. When `middleware` feature is enabled additional `MiddlewareInput` bounds are added based on native or WASM targets so the input can be stored long enough to rebuild a fresh request for retries.
 
-## Error Handling
+## Bring Your Own Types Interaction
 
-`OpenAIError::Boxed` is available only when the `middleware` feature is enabled.
+With the `byot` feature, generated `*_byot` methods keep the same minimal trait bounds with or without middleware. JSON request bodies are serialized before they enter the replayable middleware request factory; multipart request bodies use the client-level replay bounds required by form handling.
 
-Custom middleware services installed with `Client::with_http_service` may use any error type that implements `Into<OpenAIError>`. This lets middleware preserve structured errors when it has a dedicated `OpenAIError` conversion.
 
-Tower's `BoxError` converts into `OpenAIError::Boxed`, which is useful for generic tower layers whose concrete error type is erased. Callers can still downcast the boxed error when they know the original error type.

async-openai/README.md

@@ -19,7 +19,10 @@
 
 ## Overview
 
-`async-openai` is an unofficial Rust library for OpenAI, based on [OpenAI OpenAPI spec](https://github.com/openai/openai-openapi). It implements all APIs from the spec:
+`async-openai` is an unofficial Rust library for OpenAI, based on [OpenAI OpenAPI spec](https://github.com/openai/openai-openapi). It implements all APIs from the spec.
+
+<details>
+<summary>Feature Flags</summary>
 
 | What | APIs | Crate Feature Flags |
 |---|---|---|
@@ -36,16 +39,18 @@
 | **Administration** | Admin API Keys, Invites, Users, Groups, Roles, Role assignments, Projects, Project users, Project groups, Project service accounts, Project API keys, Project rate limits, Audit logs, Usage, Certificates | `administration` |
 | **Legacy** | Completions | `completions` |
 
-Features that makes `async-openai` unique:
+</details>
+
+
 - Bring your own custom types for Request or Response objects.
-- SSE streaming on available APIs.
-- Customize path, query and headers per request; customize path and headers globally (for all requests).
-- Requests (except SSE streaming) including form submissions are retried with exponential backoff when [rate limited](https://platform.openai.com/docs/guides/rate-limits).
+- Requests are retried with exponential backoff when [rate limited](https://platform.openai.com/docs/guides/rate-limits).
 - Ergonomic builder pattern for all request objects.
-- Granular feature flags to enable any types or apis: good for faster compilation and crate reuse.
-- Microsoft Azure OpenAI Service (only for APIs matching OpenAI spec).
-- WASM (doesn't support streaming yet)
-- Middleware support with [tower](https://crates.io/crates/tower) ecosystem
+- SSE streaming.
+- Customize path, query and headers per request or for all requests.
+- Granular feature flags to enable any types or apis.
+- Microsoft Azure OpenAI Service.
+- WASM.
+- Middleware support with [tower](https://crates.io/crates/tower) ecosystem.
 
 ## Usage
 
@@ -111,18 +116,10 @@ async fn main() -> Result<(), Box<dyn Error>> {
   <sub>Scaled up for README, actual size 256x256</sub>
 </div>
 
-## Webhooks
-
-Support for webhook includes event types, signature verification, and building webhook events from payloads.
-
 ## Bring Your Own Types
 
 Enable methods whose input and outputs are generics with `byot` feature. It creates a new method with same name and `_byot` suffix. 
 
-`byot` requires trait bounds: 
-- a request type (`fn` input parameter) needs to implement `serde::Serialize` or `std::fmt::Display` trait
-- a response type (`fn` ouput parameter) needs to implement `serde::de::DeserializeOwned` trait.
-
 For example, to use `serde_json::Value` as request and response type:
 ```rust
 let response: Value = client
@@ -147,9 +144,11 @@ let response: Value = client
 This can be useful in many scenarios:
 - To use this library with other OpenAI compatible APIs whose types don't exactly match OpenAI. 
 - Extend existing types in this crate with new fields with `serde` (for example with `#[serde(flatten)]`).
-- To avoid verbose types.
+- To avoid typing verbose types.
 - To escape deserialization errors.
 
+`*_byot` methods require same trait bounds as regular methods.
+
 Visit [examples/bring-your-own-type](https://github.com/64bit/async-openai/tree/main/examples/bring-your-own-type)
 directory to learn more.
 
@@ -217,8 +216,7 @@ client
 
 This allows you to use same code (say a `fn`) to call APIs on different OpenAI-compatible providers.
 
-For any struct that implements `Config` trait, wrap it in a smart pointer and cast the pointer to `dyn Config`
-trait object, then create a client with `Box` or `Arc` wrapped configuration.
+Create a client with `Box` or `Arc` wrapped configuration.
 
 For example:
 
@@ -237,6 +235,10 @@ fn chat_completion(client: &Client<Box<dyn Config>>) {
 }
 ```
 
+## Webhooks
+
+Support for webhook includes event types, signature verification, and building webhook events from payloads.
+
 ## Middleware
 
 Middleware is supported via Tower ecosystem, which can be enabled with `middleware` feature. See [middleware](https://github.com/64bit/async-openai/blob/main/async-openai/MIDDLEWARE.md) for more detail.

async-openai/src/assistants/runs.rs

@@ -9,7 +9,6 @@ use crate::{
     Client, RequestOptions,
 };
 
-#[cfg(not(target_family = "wasm"))]
 use crate::types::assistants::AssistantEventStream;
 
 /// Represents an execution run on a thread.
@@ -53,12 +52,11 @@ impl<'c, C: Config> Runs<'c, C> {
     /// Create a run.
     ///
     /// byot: You must ensure "stream: true" in serialized `request`
-    #[cfg(not(target_family = "wasm"))]
     #[crate::byot(
         T0 = serde::Serialize,
         R = serde::de::DeserializeOwned,
         stream = "true",
-        where_clause = "R: std::marker::Send + 'static + TryFrom<eventsource_stream::Event, Error = OpenAIError>"
+        where_clause = "R: crate::traits::MaybeSend + 'static + TryFrom<eventsource_stream::Event, Error = OpenAIError>"
     )]
     #[allow(unused_mut)]
     pub async fn create_stream(
@@ -144,13 +142,12 @@ impl<'c, C: Config> Runs<'c, C> {
     }
 
     /// byot: You must ensure "stream: true" in serialized `request`
-    #[cfg(not(target_family = "wasm"))]
     #[crate::byot(
         T0 = std::fmt::Display,
         T1 = serde::Serialize,
         R = serde::de::DeserializeOwned,
         stream = "true",
-        where_clause = "R: std::marker::Send + 'static + TryFrom<eventsource_stream::Event, Error = OpenAIError>"
+        where_clause = "R: crate::traits::MaybeSend + 'static + TryFrom<eventsource_stream::Event, Error = OpenAIError>"
     )]
     #[allow(unused_mut)]
     pub async fn submit_tool_outputs_stream(

async-openai/src/assistants/threads.rs

@@ -8,7 +8,6 @@ use crate::{
     Client, Messages, RequestOptions, Runs,
 };
 
-#[cfg(not(target_family = "wasm"))]
 use crate::types::assistants::AssistantEventStream;
 
 /// Create threads that assistants can interact with.
@@ -54,12 +53,11 @@ impl<'c, C: Config> Threads<'c, C> {
     /// Create a thread and run it in one request (streaming).
     ///
     /// byot: You must ensure "stream: true" in serialized `request`
-    #[cfg(not(target_family = "wasm"))]
     #[crate::byot(
         T0 = serde::Serialize,
         R = serde::de::DeserializeOwned,
         stream = "true",
-        where_clause = "R: std::marker::Send + 'static + TryFrom<eventsource_stream::Event, Error = OpenAIError>"
+        where_clause = "R: crate::traits::MaybeSend + 'static + TryFrom<eventsource_stream::Event, Error = OpenAIError>"
     )]
     #[allow(unused_mut)]
     pub async fn create_and_run_stream(

async-openai/src/audio/speech.rs

@@ -5,7 +5,6 @@ use crate::{
     Client, RequestOptions,
 };
 
-#[cfg(not(target_family = "wasm"))]
 use crate::types::audio::SpeechResponseStream;
 
 pub struct Speech<'c, C: Config> {
@@ -35,12 +34,11 @@ impl<'c, C: Config> Speech<'c, C> {
     }
 
     /// Generates audio from the input text in SSE stream format.
-    #[cfg(not(target_family = "wasm"))]
     #[crate::byot(
         T0 = serde::Serialize,
         R = serde::de::DeserializeOwned,
         stream = "true",
-        where_clause = "R: std::marker::Send + 'static"
+        where_clause = "R: crate::traits::MaybeSend + 'static"
     )]
     #[allow(unused_mut)]
     pub async fn create_stream(

async-openai/src/audio/transcriptions.rs

@@ -10,7 +10,6 @@ use crate::{
     Client, RequestOptions,
 };
 
-#[cfg(not(target_family = "wasm"))]
 use crate::types::audio::TranscriptionResponseStream;
 
 pub struct Transcriptions<'c, C: Config> {
@@ -30,7 +29,7 @@ impl<'c, C: Config> Transcriptions<'c, C> {
     #[crate::byot(
         T0 = Clone,
         R = serde::de::DeserializeOwned,
-        where_clause =  "reqwest::multipart::Form: crate::traits::AsyncTryFrom<T0, Error = OpenAIError>",
+        where_clause =  "reqwest::multipart::Form: crate::traits::AsyncTryFrom<T0, Error = OpenAIError>, T0: crate::traits::MaybeSend + 'static",
     )]
     pub async fn create(
         &self,
@@ -41,12 +40,11 @@ impl<'c, C: Config> Transcriptions<'c, C> {
             .await
     }
 
-    #[cfg(not(target_family = "wasm"))]
     #[crate::byot(
         T0 = Clone,
         R = serde::de::DeserializeOwned,
         stream = "true",
-        where_clause = "R: std::marker::Send + 'static, reqwest::multipart::Form: crate::traits::AsyncTryFrom<T0, Error = OpenAIError>"
+        where_clause = "R: crate::traits::MaybeSend + 'static, reqwest::multipart::Form: crate::traits::AsyncTryFrom<T0, Error = OpenAIError>, T0: crate::traits::MaybeSend + 'static"
     )]
     #[allow(unused_mut)]
     pub async fn create_stream(
@@ -74,7 +72,7 @@ impl<'c, C: Config> Transcriptions<'c, C> {
     #[crate::byot(
         T0 = Clone,
         R = serde::de::DeserializeOwned,
-        where_clause =  "reqwest::multipart::Form: crate::traits::AsyncTryFrom<T0, Error = OpenAIError>",
+        where_clause =  "reqwest::multipart::Form: crate::traits::AsyncTryFrom<T0, Error = OpenAIError>, T0: crate::traits::MaybeSend + 'static",
     )]
     pub async fn create_verbose_json(
         &self,
@@ -89,7 +87,7 @@ impl<'c, C: Config> Transcriptions<'c, C> {
     #[crate::byot(
         T0 = Clone,
         R = serde::de::DeserializeOwned,
-        where_clause =  "reqwest::multipart::Form: crate::traits::AsyncTryFrom<T0, Error = OpenAIError>",
+        where_clause =  "reqwest::multipart::Form: crate::traits::AsyncTryFrom<T0, Error = OpenAIError>, T0: crate::traits::MaybeSend + 'static",
     )]
     pub async fn create_diarized_json(
         &self,

async-openai/src/audio/translations.rs

@@ -27,7 +27,7 @@ impl<'c, C: Config> Translations<'c, C> {
     #[crate::byot(
         T0 = Clone,
         R = serde::de::DeserializeOwned,
-        where_clause =  "reqwest::multipart::Form: crate::traits::AsyncTryFrom<T0, Error = OpenAIError>",
+        where_clause =  "reqwest::multipart::Form: crate::traits::AsyncTryFrom<T0, Error = OpenAIError>, T0: crate::traits::MaybeSend + 'static",
     )]
     pub async fn create(
         &self,
@@ -42,7 +42,7 @@ impl<'c, C: Config> Translations<'c, C> {
     #[crate::byot(
         T0 = Clone,
         R = serde::de::DeserializeOwned,
-        where_clause =  "reqwest::multipart::Form: crate::traits::AsyncTryFrom<T0, Error = OpenAIError>",
+        where_clause =  "reqwest::multipart::Form: crate::traits::AsyncTryFrom<T0, Error = OpenAIError>, T0: crate::traits::MaybeSend + 'static",
     )]
     pub async fn create_verbose_json(
         &self,

async-openai/src/audio/voice_consents.rs

@@ -26,7 +26,7 @@ impl<'c, C: Config> VoiceConsents<'c, C> {
     #[crate::byot(
         T0 = Clone,
         R = serde::de::DeserializeOwned,
-        where_clause =  "reqwest::multipart::Form: crate::traits::AsyncTryFrom<T0, Error = OpenAIError>",
+        where_clause =  "reqwest::multipart::Form: crate::traits::AsyncTryFrom<T0, Error = OpenAIError>, T0: crate::traits::MaybeSend + 'static",
     )]
     pub async fn create(
         &self,

async-openai/src/audio/voices.rs

@@ -23,7 +23,7 @@ impl<'c, C: Config> Voices<'c, C> {
     #[crate::byot(
         T0 = Clone,
         R = serde::de::DeserializeOwned,
-        where_clause =  "reqwest::multipart::Form: crate::traits::AsyncTryFrom<T0, Error = OpenAIError>",
+        where_clause =  "reqwest::multipart::Form: crate::traits::AsyncTryFrom<T0, Error = OpenAIError>, T0: crate::traits::MaybeSend + 'static",
     )]
     pub async fn create(&self, request: CreateVoiceRequest) -> Result<VoiceResource, OpenAIError> {
         self.client