klbr/khoj
1
0
Fork 0
mirror of https://github.com/khoaliber/khoj.git synced 2026-03-09 13:25:11 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix to calculate usage from openai api streaming completion

Browse Source 
During streaming chunk.chunk contains usage data. This regression must
have appeared while tuning openai stream processors
This commit is contained in:
Debanjum
2025-08-18 18:53:30 -07:00
parent 573c6a32e1
commit e504141c07
1 changed files with 4 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
src/khoj/processor/conversation/openai/utils.py
View File

@@ -219,6 +219,10 @@ def completion_with_backoff(
# Json dump tool calls into aggregated response # Json dump tool calls into aggregated response
aggregated_response = json.dumps([tool_call.__dict__ for tool_call in tool_calls]) aggregated_response = json.dumps([tool_call.__dict__ for tool_call in tool_calls])
# Align chunk definition with non-streaming mode for post stream completion usage
if hasattr(chunk, "chunk"):
chunk = chunk.chunk
# Calculate cost of chat # Calculate cost of chat
input_tokens = chunk.usage.prompt_tokens if hasattr(chunk, "usage") and chunk.usage else 0 input_tokens = chunk.usage.prompt_tokens if hasattr(chunk, "usage") and chunk.usage else 0
output_tokens = chunk.usage.completion_tokens if hasattr(chunk, "usage") and chunk.usage else 0 output_tokens = chunk.usage.completion_tokens if hasattr(chunk, "usage") and chunk.usage else 0