55from ml_cloud_connector .domain .RestCall import RestCall
66from ml_cloud_connector .domain .ServerParameters import ServerParameters
77from ml_cloud_connector .domain .ServerType import ServerType
8- from ml_cloud_connector .ports .CloudProviderRepository import CloudProviderRepository
98from ml_cloud_connector .use_cases .ExecuteOnCloudUseCase import ExecuteOnCloudUseCase
109
1110from configuration import (
2524
2625RETRIES = 3
2726
27+ if not USE_LOCAL_SEGMENTATION :
28+ SERVER_PARAMETERS = ServerParameters (namespace = "google_v2" , server_type = ServerType .DOCUMENT_LAYOUT_ANALYSIS )
29+ CLOUD_PROVIDER = GoogleV2Repository (server_parameters = SERVER_PARAMETERS , service_logger = service_logger )
30+ EXECUTE_ON_CLOUD = ExecuteOnCloudUseCase (cloud_provider = CLOUD_PROVIDER , service_logger = service_logger )
31+
2832
2933def get_xml_name (task : Task ) -> str :
3034 xml_file_name = f"{ task .tenant } __{ task .params .filename .lower ().replace ('.pdf' , '.xml' )} "
@@ -72,10 +76,6 @@ def extract_segments(task: Task, xml_file_name: str = "") -> ExtractionData:
7276
7377
7478def extract_segments_cloud (pdf_file : PdfFile , task : Task , xml_file_name : str = "" ) -> (bool , ExtractionData ):
75- server_parameters = ServerParameters (namespace = "google_v2" , server_type = ServerType .DOCUMENT_LAYOUT_ANALYSIS )
76- cloud_provider = GoogleV2Repository (server_parameters = server_parameters , service_logger = service_logger )
77- execute_on_cloud_use_case = ExecuteOnCloudUseCase (cloud_provider = cloud_provider , service_logger = service_logger )
78-
7979 with open (pdf_file .get_path (task .params .filename ), "rb" ) as stream :
8080 file_content = stream .read ()
8181
@@ -88,11 +88,11 @@ def extract_segments_cloud(pdf_file: PdfFile, task: Task, xml_file_name: str = "
8888 files = files ,
8989 data = {"fast" : "False" },
9090 )
91- response , success , error = execute_on_cloud_use_case .execute (rest_call )
91+ response , success , error = EXECUTE_ON_CLOUD .execute (rest_call )
9292 if not success :
9393 return False , None
9494
95- if not save_cloud_xml_file (execute_on_cloud_use_case , xml_file_name ):
95+ if not save_cloud_xml_file (xml_file_name ):
9696 return False , None
9797
9898 segments : list [SegmentBox ] = [SegmentBox (** segment_box ) for segment_box in response .json ()]
@@ -106,14 +106,14 @@ def extract_segments_cloud(pdf_file: PdfFile, task: Task, xml_file_name: str = "
106106 )
107107
108108
109- def save_cloud_xml_file (execute_on_cloud_use_case : ExecuteOnCloudUseCase , xml_file_name : str ) -> bool :
109+ def save_cloud_xml_file (xml_file_name : str ) -> bool :
110110 try :
111111 rest_call = RestCall (
112112 port = DOCUMENT_LAYOUT_ANALYSIS_PORT ,
113113 endpoint = ["get_xml" , xml_file_name ],
114114 method = "GET" ,
115115 )
116- response , success , error = execute_on_cloud_use_case .execute (rest_call )
116+ response , success , error = EXECUTE_ON_CLOUD .execute (rest_call )
117117
118118 if not success :
119119 return False
0 commit comments