Skip to content

LLM-Driven Extraction of Unstructured Data — Built for API Deployments & ETL Pipeline Workflows

FIX: Fixed missing keys in highlight data (#861)

Deepak Kb67768c7748e
diff --git a/prompt-service/src/unstract/prompt_service/helper.py b/prompt-service/src/unstract/prompt_service/helper.py
index 913fb118d..e451a2030 100644
--- a/prompt-service/src/unstract/prompt_service/helper.py
+++ b/prompt-service/src/unstract/prompt_service/helper.py
@@ -327,13 +327,10 @@ def run_completion(
extract_json=prompt_type.lower() != PSKeys.TEXT,
)
answer: str = completion[PSKeys.RESPONSE].text
- highlight_data = completion.get(PSKeys.HIGHLIGHT_DATA)
+ highlight_data = completion.get(PSKeys.HIGHLIGHT_DATA, [])
confidence_data = completion.get(PSKeys.CONFIDENCE_DATA)
if metadata is not None and prompt_key:
- if highlight_data:
- metadata.setdefault(PSKeys.HIGHLIGHT_DATA, {})[
- prompt_key
- ] = highlight_data
+ metadata.setdefault(PSKeys.HIGHLIGHT_DATA, {})[prompt_key] = highlight_data
if confidence_data:
metadata.setdefault(PSKeys.CONFIDENCE_DATA, {})[