feat: ernie migrates to v2 api (#1130)

7 files changed, 16 insertions, 378 deletions

diff --git a/Argcfile.sh b/Argcfile.sh
index cf1e8ff..ea77211 100755
--- a/Argcfile.sh
+++ b/Argcfile.sh
@@ -294,21 +294,6 @@ chat-vertexai() {
 -d "$(_build_body vertexai "$@")" 
 }
 
-# @cmd Chat with ernie api
-# @meta require-tools jq
-# @env ERNIE_API_KEY!
-# @option -m --model=ernie-tiny-8k $ERNIE_MODEL
-# @flag -S --no-stream
-# @arg text~
-chat-ernie() {
-    auth_url="https://aip.baidubce.com/oauth/2.0/token?grant_type=client_credentials&client_id=$ERNIE_API_KEY&client_secret=$ERNIE_SECRET_KEY"
-    ACCESS_TOKEN="$(curl -fsSL "$auth_url" | jq -r '.access_token')"
-    url="https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/$argc_model?access_token=$ACCESS_TOKEN"
-    _wrapper curl -i "$url" \
--X POST \
--d "$(_build_body ernie "$@")"
-}
-
 _argc_before() {
     OPENAI_COMPATIBLE_PROVIDERS=( \
         openai,gpt-4o-mini,https://api.openai.com/v1 \
@@ -316,6 +301,7 @@ _argc_before() {
         cloudflare,@cf/meta/llama-3.1-8b-instruct,https://api.cloudflare.com/client/v4/accounts/${CLOUDFLARE_ACCOUNT_ID}/ai/v1 \
         deepinfra,meta-llama/Meta-Llama-3.1-8B-Instruct,https://api.deepinfra.com/v1/openai \
         deepseek,deepseek-chat,https://api.deepseek.com \
+        ernie,ernie-4.0-turbo-8k-latest,https://qianfan.baidubce.com/v2 \
         fireworks,accounts/fireworks/models/llama-v3p1-8b-instruct,https://api.fireworks.ai/inference/v1 \
         github,gpt-4o-mini,https://models.inference.ai.azure.com \
         groq,llama-3.1-8b-instant,https://api.groq.com/openai/v1 \
@@ -380,7 +366,7 @@ _choice_provider() {
 }
 
 _choice_client() {
-    printf "%s\n" gemini claude cohere azure-openai vertexai bedrock ernie
+    printf "%s\n" gemini claude cohere azure-openai vertexai bedrock
 }
 
 _choice_openai_compatible_provider() {
@@ -442,17 +428,6 @@ _build_body() {
     "safetySettings":[{"category":"HARM_CATEGORY_HARASSMENT","threshold":"BLOCK_ONLY_HIGH"},{"category":"HARM_CATEGORY_HATE_SPEECH","threshold":"BLOCK_ONLY_HIGH"},{"category":"HARM_CATEGORY_SEXUALLY_EXPLICIT","threshold":"BLOCK_ONLY_HIGH"},{"category":"HARM_CATEGORY_DANGEROUS_CONTENT","threshold":"BLOCK_ONLY_HIGH"}]
 }'
             ;;
-        ernie)
-            echo '{
-    "messages": [
-        {
-            "role": "user",
-            "content": "'"$*"'"
-        }
-    ],
-    "stream": '$stream'
-}'
-            ;;
         *)
             _die "error: unsupported build body for $kind"
             ;;
diff --git a/config.example.yaml b/config.example.yaml
index fe41a61..8e81f8a 100644
--- a/config.example.yaml
+++ b/config.example.yaml
@@ -246,9 +246,10 @@ clients:
     api_key: xxx
 
   # See https://cloud.baidu.com/doc/WENXINWORKSHOP/index.html
-  - type: ernie
+  - type: openai-compatible
+    name: ernie
+    api_base: https://qianfan.baidubce.com/v2
     api_key: xxx
-    secret_key: xxx
 
   # See https://dashscope.aliyun.com/
   - type: openai-compatible
diff --git a/models.yaml b/models.yaml
index a6003aa..4816586 100644
--- a/models.yaml
+++ b/models.yaml
@@ -746,19 +746,19 @@
       max_input_tokens: 128000
       input_price: 0.042
       output_price: 0.084
-    - name: bge_large_zh
+    - name: bge-large-zh
       type: embedding
       input_price: 0.07
       max_tokens_per_chunk: 512
       default_chunk_size: 1000
       max_batch_size: 16
-    - name: bge_large_en
+    - name: bge-large-en
       type: embedding
       input_price: 0.07
       max_tokens_per_chunk: 512
       default_chunk_size: 1000
       max_batch_size: 16
-    - name: bce_reranker_base
+    - name: bce-reranker-base
       type: reranker
       max_input_tokens: 1024
       input_price: 0.07
diff --git a/src/client/common.rs b/src/client/common.rs
index 80f585d..322fc5c 100644
--- a/src/client/common.rs
+++ b/src/client/common.rs
@@ -525,19 +525,6 @@ pub fn json_str_from_map<'a>(
     map.get(field_name).and_then(|v| v.as_str())
 }
 
-pub fn maybe_catch_error(data: &Value) -> Result<()> {
-    if let (Some(code), Some(message)) = (data["code"].as_str(), data["message"].as_str()) {
-        debug!("Invalid response: {}", data);
-        bail!("{message} (code: {code})");
-    } else if let (Some(error_code), Some(error_msg)) =
-        (data["error_code"].as_number(), data["error_msg"].as_str())
-    {
-        debug!("Invalid response: {}", data);
-        bail!("{error_msg} (error_code: {error_code})");
-    }
-    Ok(())
-}
-
 fn set_client_config(list: &[PromptAction], client_config: &mut Value, client: &str) -> Result<()> {
     for (key, desc, help_message) in list {
         let env_name = format!("{client}_{key}").to_ascii_uppercase();
diff --git a/src/client/ernie.rs b/src/client/ernie.rs
deleted file mode 100644
index 9a6b962..0000000
--- a/src/client/ernie.rs
+++ /dev/null
@@ -1,329 +0,0 @@
-use super::access_token::*;
-use super::openai_compatible::*;
-use super::*;
-
-use anyhow::{anyhow, bail, Context, Result};
-use reqwest::{Client as ReqwestClient, RequestBuilder};
-use serde::Deserialize;
-use serde_json::{json, Value};
-
-const API_BASE: &str = "https://aip.baidubce.com/rpc/2.0/ai_custom/v1";
-const ACCESS_TOKEN_URL: &str = "https://aip.baidubce.com/oauth/2.0/token";
-
-#[derive(Debug, Clone, Deserialize, Default)]
-pub struct ErnieConfig {
-    pub name: Option<String>,
-    pub api_key: Option<String>,
-    pub secret_key: Option<String>,
-    #[serde(default)]
-    pub models: Vec<ModelData>,
-    pub patch: Option<RequestPatch>,
-    pub extra: Option<ExtraConfig>,
-}
-
-impl ErnieClient {
-    config_get_fn!(api_key, get_api_key);
-    config_get_fn!(secret_key, get_secret_key);
-    pub const PROMPTS: [PromptAction<'static>; 2] = [
-        ("api_key", "API Key", None),
-        ("secret_key", "Secret Key", None),
-    ];
-}
-
-#[async_trait::async_trait]
-impl Client for ErnieClient {
-    client_common_fns!();
-
-    async fn chat_completions_inner(
-        &self,
-        client: &ReqwestClient,
-        data: ChatCompletionsData,
-    ) -> Result<ChatCompletionsOutput> {
-        prepare_access_token(self, client).await?;
-        let request_data = prepare_chat_completions(self, data)?;
-        let builder = self.request_builder(client, request_data);
-        chat_completions(builder, &self.model).await
-    }
-
-    async fn chat_completions_streaming_inner(
-        &self,
-        client: &ReqwestClient,
-        handler: &mut SseHandler,
-        data: ChatCompletionsData,
-    ) -> Result<()> {
-        prepare_access_token(self, client).await?;
-        let request_data = prepare_chat_completions(self, data)?;
-        let builder = self.request_builder(client, request_data);
-        chat_completions_streaming(builder, handler, &self.model).await
-    }
-
-    async fn embeddings_inner(
-        &self,
-        client: &ReqwestClient,
-        data: &EmbeddingsData,
-    ) -> Result<EmbeddingsOutput> {
-        prepare_access_token(self, client).await?;
-        let request_data = prepare_embeddings(self, data)?;
-        let builder = self.request_builder(client, request_data);
-        embeddings(builder, &self.model).await
-    }
-
-    async fn rerank_inner(
-        &self,
-        client: &ReqwestClient,
-        data: &RerankData,
-    ) -> Result<RerankOutput> {
-        prepare_access_token(self, client).await?;
-        let request_data = prepare_rerank(self, data)?;
-        let builder = self.request_builder(client, request_data);
-        rerank(builder, &self.model).await
-    }
-}
-
-fn prepare_chat_completions(self_: &ErnieClient, data: ChatCompletionsData) -> Result<RequestData> {
-    let access_token = get_access_token(self_.name())?;
-
-    let url = format!(
-        "{API_BASE}/wenxinworkshop/chat/{}?access_token={access_token}",
-        self_.model.name(),
-    );
-
-    let body = build_chat_completions_body(data, &self_.model);
-
-    let request_data = RequestData::new(url, body);
-
-    Ok(request_data)
-}
-
-fn prepare_embeddings(self_: &ErnieClient, data: &EmbeddingsData) -> Result<RequestData> {
-    let access_token = get_access_token(self_.name())?;
-
-    let url = format!(
-        "{API_BASE}/wenxinworkshop/embeddings/{}?access_token={access_token}",
-        self_.model.name(),
-    );
-
-    let body = json!({
-        "input": data.texts,
-    });
-
-    let request_data = RequestData::new(url, body);
-
-    Ok(request_data)
-}
-
-fn prepare_rerank(self_: &ErnieClient, data: &RerankData) -> Result<RequestData> {
-    let access_token = get_access_token(self_.name())?;
-
-    let url = format!(
-        "{API_BASE}/wenxinworkshop/reranker/{}?access_token={access_token}",
-        self_.model.name(),
-    );
-
-    let RerankData {
-        query,
-        documents,
-        top_n,
-    } = data;
-
-    let body = json!({
-        "query": query,
-        "documents": documents,
-        "top_n": top_n
-    });
-
-    let request_data = RequestData::new(url, body);
-
-    Ok(request_data)
-}
-
-async fn prepare_access_token(self_: &ErnieClient, client: &ReqwestClient) -> Result<()> {
-    let client_name = self_.name();
-    if !is_valid_access_token(client_name) {
-        let api_key = self_.get_api_key()?;
-        let secret_key = self_.get_secret_key()?;
-
-        let token = fetch_access_token(client, &api_key, &secret_key)
-            .await
-            .with_context(|| "Failed to fetch access token")?;
-        set_access_token(client_name, token, 86400);
-    }
-    Ok(())
-}
-
-async fn chat_completions(
-    builder: RequestBuilder,
-    _model: &Model,
-) -> Result<ChatCompletionsOutput> {
-    let data: Value = builder.send().await?.json().await?;
-    maybe_catch_error(&data)?;
-    debug!("non-stream-data: {data}");
-    extract_chat_completions_text(&data)
-}
-
-async fn chat_completions_streaming(
-    builder: RequestBuilder,
-    handler: &mut SseHandler,
-    _model: &Model,
-) -> Result<()> {
-    let handle = |message: SseMmessage| -> Result<bool> {
-        let data: Value = serde_json::from_str(&message.data)?;
-        debug!("stream-data: {data}");
-        if let Some(function) = data["function_call"].as_object() {
-            if let (Some(name), Some(arguments)) = (
-                function.get("name").and_then(|v| v.as_str()),
-                function.get("arguments").and_then(|v| v.as_str()),
-            ) {
-                let arguments: Value = arguments.parse().with_context(|| {
-                    format!("Tool call '{name}' have non-JSON arguments '{arguments}'")
-                })?;
-                handler.tool_call(ToolCall::new(name.to_string(), arguments, None))?;
-            }
-        } else if let Some(text) = data["result"].as_str() {
-            handler.text(text)?;
-        }
-        Ok(false)
-    };
-
-    sse_stream(builder, handle).await
-}
-
-async fn embeddings(builder: RequestBuilder, _model: &Model) -> Result<EmbeddingsOutput> {
-    let data: Value = builder.send().await?.json().await?;
-    maybe_catch_error(&data)?;
-    let res_body: EmbeddingsResBody =
-        serde_json::from_value(data).context("Invalid embeddings data")?;
-    let output = res_body.data.into_iter().map(|v| v.embedding).collect();
-    Ok(output)
-}
-
-#[derive(Deserialize)]
-struct EmbeddingsResBody {
-    data: Vec<EmbeddingsResBodyEmbedding>,
-}
-
-#[derive(Deserialize)]
-struct EmbeddingsResBodyEmbedding {
-    embedding: Vec<f32>,
-}
-
-async fn rerank(builder: RequestBuilder, _model: &Model) -> Result<RerankOutput> {
-    let data: Value = builder.send().await?.json().await?;
-    maybe_catch_error(&data)?;
-    let res_body: GenericRerankResBody =
-        serde_json::from_value(data).context("Invalid rerank data")?;
-    Ok(res_body.results)
-}
-
-fn build_chat_completions_body(data: ChatCompletionsData, model: &Model) -> Value {
-    let ChatCompletionsData {
-        mut messages,
-        temperature,
-        top_p,
-        functions,
-        stream,
-    } = data;
-
-    let system_message = extract_system_message(&mut messages);
-
-    let messages: Vec<Value> = messages
-        .into_iter()
-        .flat_map(|message| {
-            let Message { role, content } = message;
-            match content {
-                MessageContent::ToolCalls(MessageContentToolCalls {
-                    tool_results,  ..
-                }) => {
-                    let mut list = vec![];
-                    for tool_result in tool_results {
-                        list.push(json!({
-                            "role": "assistant",
-                            "content": format!("Action: {}\nAction Input: {}", tool_result.call.name, tool_result.call.arguments)
-                        }));
-                        list.push(json!({
-                            "role": "user",
-                            "content": tool_result.output.to_string(),
-                        }))
-
-                    }
-                    list
-                }
-                _ => vec![json!({ "role": role, "content": content })],
-            }
-        })
-        .collect();
-
-    let mut body = json!({
-        "messages": messages,
-    });
-
-    if let Some(v) = system_message {
-        body["system"] = v.into();
-    }
-
-    if let Some(v) = model.max_tokens_param() {
-        body["max_output_tokens"] = v.into();
-    }
-    if let Some(v) = temperature {
-        body["temperature"] = v.into();
-    }
-    if let Some(v) = top_p {
-        body["top_p"] = v.into();
-    }
-
-    if stream {
-        body["stream"] = true.into();
-    }
-
-    if let Some(functions) = functions {
-        body["functions"] = json!(functions);
-    }
-
-    body
-}
-
-fn extract_chat_completions_text(data: &Value) -> Result<ChatCompletionsOutput> {
-    let text = data["result"].as_str().unwrap_or_default();
-
-    let mut tool_calls = vec![];
-    if let Some(call) = data["function_call"].as_object() {
-        if let (Some(name), Some(arguments)) = (
-            call.get("name").and_then(|v| v.as_str()),
-            call.get("arguments").and_then(|v| v.as_str()),
-        ) {
-            let arguments: Value = arguments.parse().with_context(|| {
-                format!("Tool call '{name}' have non-JSON arguments '{arguments}'")
-            })?;
-            tool_calls.push(ToolCall::new(name.to_string(), arguments, None));
-        }
-    }
-
-    if text.is_empty() && tool_calls.is_empty() {
-        bail!("Invalid response data: {data}");
-    }
-    let output = ChatCompletionsOutput {
-        text: text.to_string(),
-        tool_calls,
-        id: data["id"].as_str().map(|v| v.to_string()),
-        input_tokens: data["usage"]["prompt_tokens"].as_u64(),
-        output_tokens: data["usage"]["completion_tokens"].as_u64(),
-    };
-    Ok(output)
-}
-
-async fn fetch_access_token(
-    client: &reqwest::Client,
-    api_key: &str,
-    secret_key: &str,
-) -> Result<String> {
-    let url = format!("{ACCESS_TOKEN_URL}?grant_type=client_credentials&client_id={api_key}&client_secret={secret_key}");
-    let value: Value = client.get(&url).send().await?.json().await?;
-    let result = value["access_token"].as_str().ok_or_else(|| {
-        if let Some(err_msg) = value["error_description"].as_str() {
-            anyhow!("{err_msg}")
-        } else {
-            anyhow!("Invalid response data")
-        }
-    })?;
-    Ok(result.to_string())
-}
diff --git a/src/client/mod.rs b/src/client/mod.rs
index d8b41c6..c87d2b5 100644
--- a/src/client/mod.rs
+++ b/src/client/mod.rs
@@ -31,10 +31,9 @@ register_client!(
     ),
     (vertexai, "vertexai", VertexAIConfig, VertexAIClient),
     (bedrock, "bedrock", BedrockConfig, BedrockClient),
-    (ernie, "ernie", ErnieConfig, ErnieClient),
 );
 
-pub const OPENAI_COMPATIBLE_PROVIDERS: [(&str, &str); 23] = [
+pub const OPENAI_COMPATIBLE_PROVIDERS: [(&str, &str); 24] = [
     ("ai21", "https://api.ai21.com/studio/v1"),
     (
         "cloudflare",
@@ -42,6 +41,7 @@ pub const OPENAI_COMPATIBLE_PROVIDERS: [(&str, &str); 23] = [
     ),
     ("deepinfra", "https://api.deepinfra.com/v1/openai"),
     ("deepseek", "https://api.deepseek.com"),
+    ("ernie", "https://qianfan.baidubce.com/v2"),
     ("fireworks", "https://api.fireworks.ai/inference/v1"),
     ("github", "https://models.inference.ai.azure.com"),
     ("groq", "https://api.groq.com/openai/v1"),
diff --git a/src/client/openai_compatible.rs b/src/client/openai_compatible.rs
index ce1eea3..30bc464 100644
--- a/src/client/openai_compatible.rs
+++ b/src/client/openai_compatible.rs
@@ -79,7 +79,11 @@ fn prepare_rerank(self_: &OpenAICompatibleClient, data: &RerankData) -> Result<R
     let api_key = self_.get_api_key().ok();
     let api_base = get_api_base_ext(self_)?;
 
-    let url = format!("{api_base}/rerank");
+    let url = if self_.name().starts_with("ernie") {
+        format!("{api_base}/rerankers")
+    } else {
+        format!("{api_base}/rerank")
+    };
 
     let body = generic_build_rerank_body(data, &self_.model);
 
@@ -149,7 +153,7 @@ pub fn generic_build_rerank_body(data: &RerankData, model: &Model) -> Value {
         "query": query,
         "documents": documents,
     });
-    if model.client_name() == "voyageai" {
+    if model.client_name().starts_with("voyageai") {
         body["top_k"] = (*top_n).into()
     } else {
         body["top_n"] = (*top_n).into()