GeneralBots/botserver
2
1
Fork 0
Code Activity Actions 4

fix: support reasoning models (GLM4.7, Kimi K2.5) - use reasoning_content when content is null
All checks were successful
BotServer CI/CD / build (push) Successful in 3m19s
Details

Browse source 
- GLM4.7 and Kimi K2.5 send response in 'reasoning_content' field, 'content' is null
- Prefer 'content' for normal models, fallback to 'reasoning_content' for reasoning models
- Fixes blank white screen when using z-ai/glm4.7 model

Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
This commit is contained in:
Rodrigo Rodriguez (Pragmatismo) 2026-04-13 15:18:21 -03:00
parent c5d30adebe
commit 3e99235a49
1 changed files with 10 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

12
src/llm/mod.rs
View file

@ -457,8 +457,16 @@ impl LLMProvider for OpenAIClient {
for line in chunk_str.lines() {
if line.starts_with("data: ") && !line.contains("[DONE]") {
if let Ok(data) = serde_json::from_str::<Value>(&line[6..]) {
if let Some(content) = data["choices"][0]["delta"]["content"].as_str() {
let processed = handler.process_content(content);
// Handle reasoning models (GLM4.7, Kimi K2.5): content is null,
// reasoning_content has the actual response
let content = data["choices"][0]["delta"]["content"].as_str();
let reasoning = data["choices"][0]["delta"]["reasoning_content"].as_str();
// Prefer content field (normal models), fallback to reasoning_content
let text_to_use = content.or(reasoning);
if let Some(text) = text_to_use {
let processed = handler.process_content(text);
if !processed.is_empty() {
let _ = tx.send(processed).await;
}