mirror of
https://github.com/katanemo/plano.git
synced 2026-06-08 14:55:14 +02:00
[BUG FIX] Add missing chat completions request hint (#116)
Signed-off-by: José Ulises Niño Rivera <junr03@users.noreply.github.com>
This commit is contained in:
parent
07ef1af24f
commit
10b5c5b42c
3 changed files with 9 additions and 2 deletions
|
|
@ -11,3 +11,4 @@ pub const MODEL_SERVER_NAME: &str = "model_server";
|
|||
pub const ARCH_ROUTING_HEADER: &str = "x-arch-llm-provider";
|
||||
pub const ARCH_MESSAGES_KEY: &str = "arch_messages";
|
||||
pub const ARCH_PROVIDER_HINT_HEADER: &str = "x-arch-llm-provider-hint";
|
||||
pub const CHAT_COMPLETIONS_PATH: &str = "v1/chat/completions";
|
||||
|
|
|
|||
|
|
@ -1,7 +1,8 @@
|
|||
use crate::consts::{
|
||||
ARCH_FC_REQUEST_TIMEOUT_MS, ARCH_MESSAGES_KEY, ARCH_PROVIDER_HINT_HEADER, ARCH_ROUTING_HEADER,
|
||||
ARC_FC_CLUSTER, DEFAULT_EMBEDDING_MODEL, DEFAULT_INTENT_MODEL, DEFAULT_PROMPT_TARGET_THRESHOLD,
|
||||
GPT_35_TURBO, MODEL_SERVER_NAME, RATELIMIT_SELECTOR_HEADER_KEY, SYSTEM_ROLE, USER_ROLE,
|
||||
ARC_FC_CLUSTER, CHAT_COMPLETIONS_PATH, DEFAULT_EMBEDDING_MODEL, DEFAULT_INTENT_MODEL,
|
||||
DEFAULT_PROMPT_TARGET_THRESHOLD, GPT_35_TURBO, MODEL_SERVER_NAME,
|
||||
RATELIMIT_SELECTOR_HEADER_KEY, SYSTEM_ROLE, USER_ROLE,
|
||||
};
|
||||
use crate::filter_context::{embeddings_store, WasmMetrics};
|
||||
use crate::llm_providers::LlmProviders;
|
||||
|
|
@ -919,6 +920,9 @@ impl HttpContext for StreamContext {
|
|||
self.delete_content_length_header();
|
||||
self.save_ratelimit_header();
|
||||
|
||||
self.chat_completions_request =
|
||||
self.get_http_request_header(":path").unwrap_or_default() == CHAT_COMPLETIONS_PATH;
|
||||
|
||||
debug!(
|
||||
"S[{}] req_headers={:?}",
|
||||
self.context_id,
|
||||
|
|
|
|||
|
|
@ -53,6 +53,8 @@ fn request_headers_expectations(module: &mut Tester, http_context: i32) {
|
|||
.returning(Some("selector-value"))
|
||||
.expect_get_header_map_pairs(Some(MapType::HttpRequestHeaders))
|
||||
.returning(None)
|
||||
.expect_get_header_map_value(Some(MapType::HttpRequestHeaders), Some(":path"))
|
||||
.returning(Some("/v1/chat/completions"))
|
||||
.expect_log(Some(LogLevel::Debug), None)
|
||||
.execute_and_expect(ReturnType::Action(Action::Continue))
|
||||
.unwrap();
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue