jacob-mathison/doczyai-pipelines
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Activity

Merged dev into feature/upDatepipelinesFile

Browse Source
This commit is contained in:
Sujit Deokar
2026-04-16 12:07:54 +00:00
parent 72ccdcd341 b9ecdf8dfc
commit 7f3879470a
1 changed files with 5 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/pipelines/saas/file_processing.py
+5
View File
@@ -54,6 +54,11 @@ def process_file(file_object, constants: Constants, run_timestamp):
with timing_utils.timed_block("preprocess", context=filename): with timing_utils.timed_block("preprocess", context=filename):
contract_text = preprocess.clean_text(contract_text) contract_text = preprocess.clean_text(contract_text)
text_dict, top_sheet_dict = preprocess.split_text(contract_text) text_dict, top_sheet_dict = preprocess.split_text(contract_text)
if not text_dict:
logging.warning(
f"{datetime_str()} All pages removed as cover sheets, skipping processing - {filename}"
)
return final_results, pd.DataFrame([{"FILE_NAME": filename}])
text_dict, removal_metadata = preprocess.clean_header_footer(text_dict) text_dict, removal_metadata = preprocess.clean_header_footer(text_dict)
# ONE TO N PROCESSING # ONE TO N PROCESSING
one_to_n_results = pd.DataFrame() # Initialize empty DataFrame for one_to_n_results one_to_n_results = pd.DataFrame() # Initialize empty DataFrame for one_to_n_results