new files types inferred from file names 🟩

This commit is contained in:
Daniel Roth 2026-05-13 13:32:54 +00:00
parent e3646162de
commit b3a68a264a
2 changed files with 12 additions and 4 deletions

View file

@ -39,4 +39,14 @@ def infer_file_type(filename: str) -> Optional[str]:
for core_file, file_type in CORE_TO_FILETYPE_MAP.items():
if filename.startswith(core_file.value):
return file_type
if CoreFiles.IMPROVEMENT_OPTION_EVALUATION.value in filename:
return CORE_TO_FILETYPE_MAP[CoreFiles.IMPROVEMENT_OPTION_EVALUATION]
if CoreFiles.MEDIUM_TERM_IMPROVEMENT_PLAN.value in filename:
return CORE_TO_FILETYPE_MAP[CoreFiles.MEDIUM_TERM_IMPROVEMENT_PLAN]
if "-OSM-" in filename and "DR-N-A" in filename:
return CORE_TO_FILETYPE_MAP[CoreFiles.RETROFIT_DESIGN_DOC]
return None

View file

@ -1,5 +1,3 @@
import pytest
from backend.pashub_fetcher.core_files import infer_file_type
@ -59,6 +57,6 @@ def test_infer_medium_term_improvement_plan():
assert infer_file_type("60800802 - NG4 4HD - Medium Term Improvement Plan IOE.pdf") == "medium_term_improvement_plan"
@pytest.mark.skip(reason="Retrofit Design Doc filename pattern not yet known")
def test_infer_retrofit_design_doc():
assert infer_file_type("2512-OSM-H56M900-XX-DR-N-A_Radford Road 408.pdf") == "retrofit_design_doc"
assert infer_file_type("2512-OSM-H21M900-XX-DR-N-A_Lord Nelson Street 018.pdf") == "retrofit_design_doc"
assert infer_file_type("2603-OSM-B06M901-XX-DR-N-A_Alvaston Walk 022.pdf") == "retrofit_design_doc"