This commit is contained in:
Julius Unverfehrt 2022-03-17 12:50:40 +01:00
parent 549b89952e
commit b01e5e05e2

View File

@ -42,21 +42,21 @@ def make_connection() -> pika.BlockingConnection:
return connection
def update_message(message_dict, analyse_container_type):
if analyse_container_type == "detr" or analyse_container_type == "image":
message_dict.update({"targetFileExtension": "ORIGIN.pdf.gz", "responseFileExtension": "IMAGE_INFO.json.gz"})
if analyse_container_type == "ner":
message_dict.update({"targetFileExtension": "TEXT.json.gz", "responseFileExtension": "NER_ENTITIES.json.gz"})
return message_dict
def build_message_bodies(analyse_container_type):
def update_message(message_dict):
if analyse_container_type == "detr" or analyse_container_type == "image":
message_dict.update({"targetFileExtension": "ORIGIN.pdf.gz", "responseFileExtension": "IMAGE_INFO.json.gz"})
if analyse_container_type == "ner":
message_dict.update(
{"targetFileExtension": "TEXT.json.gz", "responseFileExtension": "NER_ENTITIES.json.gz"})
return message_dict
storage = get_s3_storage()
for bucket_name, pdf_name in storage.get_all_object_names(CONFIG.storage.bucket):
file_id = pdf_name.split(".")[0]
dossier_id, file_id = file_id.split("/")
message_dict = {"dossierId": dossier_id, "fileId": file_id}
update_message(message_dict, analyse_container_type)
update_message(message_dict)
yield json.dumps(message_dict).encode()