From ca773bf4b9839101a17fcd2ac3db563330d0639b Mon Sep 17 00:00:00 2001 From: Katon Minhas Date: Mon, 9 Feb 2026 18:05:58 -0500 Subject: [PATCH] Update other lists --- src/pipelines/shared/extraction/tin_npi_funcs.py | 16 +++++++++++----- 1 file changed, 11 insertions(+), 5 deletions(-) diff --git a/src/pipelines/shared/extraction/tin_npi_funcs.py b/src/pipelines/shared/extraction/tin_npi_funcs.py index b272f6c..9544101 100644 --- a/src/pipelines/shared/extraction/tin_npi_funcs.py +++ b/src/pipelines/shared/extraction/tin_npi_funcs.py @@ -35,8 +35,10 @@ def run_provider_info_fields( ) relevant_pages = get_relevant_pages(all_tins, all_npis, text_dict, filename) + print("relevant_pages: ", relevant_pages) prov_info_json = get_prov_info_json(relevant_pages, text_dict, filename, payer_name) + print("prov_info_json: ", prov_info_json) # Create columns one_to_one_results["PROV_INFO_JSON"] = prov_info_json @@ -248,7 +250,7 @@ def deunknown_and_deduplicate( other_npis = [ v for v in other_npis if v not in group_npis and not string_utils.is_empty(v) ] - + group_names = list(dict.fromkeys(group_names)) other_names = list(dict.fromkeys(other_names)) other_names = [ @@ -275,6 +277,7 @@ def merge_provider_info_with_one_to_one(one_to_one_results: dict, filename: str) other_tins, other_npis, other_names = [], [], [] provider_name = one_to_one_results.get("PROVIDER_NAME") # Can be List or string + print("provider_name (1:1): ", provider_name) # Normalize provider_name to list: if string, convert to list to avoid character splitting if isinstance(provider_name, str): @@ -315,9 +318,12 @@ def merge_provider_info_with_one_to_one(one_to_one_results: dict, filename: str) group_names.add(name) else: provider_dict["IS_GROUP"] = "N" - other_tins.extend(tins) - other_npis.extend(npis) - other_names.extend(names) + for tin in tins: + other_tins.append(tin) + for npi in npis: + other_npis.append(npi) + for name in names: + other_names.append(name) # If provider_name is not null and no match was found, add a new record if provider_name and not found_match: @@ -440,7 +446,7 @@ def flag_missed_prov_info(page_text, providers, page_num, filename): missed_npis = [npi for npi in regex_npis if npi not in llm_found_npis] if missed_tins or missed_npis: # WARN ONLY if there are missed identifiers - logging.warning( + logging.debug( f"LLM missed identifiers on page {page_num} of {filename}. Missed TINs: {missed_tins}, Missed NPIs: {missed_npis}" )