feat: support RAG (#560)

* feat: support RAG

* support more embeddings models and implement concurrent embedding api

* show the progress of addings paths

* ignore embedding context when saving message

* embedding model max_chunk_size => default_chunk_size

* support pdf and pandoc formats (docx, epub, ipynb)

1 files changed, 3 insertions, 6 deletions

diff --git a/src/client/replicate.rs b/src/client/replicate.rs
index 92c7e18..e96ed64 100644
--- a/src/client/replicate.rs
+++ b/src/client/replicate.rs
@@ -1,8 +1,5 @@
-use super::{
-    catch_error, prompt_format::*, sse_stream, ChatCompletionsData, ChatCompletionsOutput, Client,
-    ExtraConfig, Model, ModelData, ModelPatches, PromptAction, PromptKind, ReplicateClient,
-    SseHandler, SseMmessage,
-};
+use super::*;
+use super::prompt_format::*;
 
 use anyhow::{anyhow, Result};
 use async_trait::async_trait;
@@ -36,7 +33,7 @@ impl ReplicateClient {
         api_key: &str,
     ) -> Result<RequestBuilder> {
         let mut body = build_chat_completions_body(data, &self.model)?;
-        self.patch_request_body(&mut body);
+        self.patch_chat_completions_body(&mut body);
 
         let url = format!("{API_BASE}/models/{}/predictions", self.model.name());