def post(self):
json_data = request.get_json(force=True)
app_context.init()
app_context.application_context = json_data
log_info(
"Resource Layout_Detector_WF Layout_Detector service started",
app_context.application_context)
task_id = str("LD-" + str(time.time()).replace('.', '')[0:13])
task_starttime = eval(str(time.time()).replace('.', '')[0:13])
#json_data = request.get_json(force = True)
try:
error_validator = ValidationResponse(DOWNLOAD_FOLDER)
if error_validator.format_error(json_data) is True:
response_gen = Response(json_data, DOWNLOAD_FOLDER)
response = response_gen.workflow_response(
task_id, task_starttime)
log_info(
"Resource Layout_Detector_WF Layout_Detector api response completed",
app_context.application_context)
return jsonify(response)
except FormatError as e:
log_error(
"Resource Layout_Detector_WF Input json format is not correct or dict_key is missing",
app_context.application_context, e)
return Status.ERR_request_input_format.value
def nonwf_response(self):
log_info("non workflow response started the response generation",
app_context.application_context)
input_files = self.json_data['input']['inputs']
app_context.init()
app_context.application_context = self.json_data
error_validator = ValidationResponse(self.DOWNLOAD_FOLDER)
try:
error_validator.inputfile_list_error(input_files)
# output_file_response = list()
# for item in input_files:
# input_filename, in_file_type, identifier = file_ops.accessing_files(item['file'])
# output_json_data = DocumentStructure(None, input_filename)
# output_filename_json = file_ops.writing_json_file(i, output_json_data, self.DOWNLOAD_FOLDER)
# file_res = file_ops.one_filename_response(input_filename, output_filename_json, in_file_type)
# output_file_response.append(file_res)
response_true = Status.SUCCESS.value
#response_true['output'] = output_file_response
output_json_data = Service(app_context=app_context)
output_filename_json = file_ops.writing_json_file(
0, output_json_data, self.DOWNLOAD_FOLDER)
response_true = file_ops.one_filename_response(
output_filename_json, langs=' ')
log_info(
"non workflow_response successfully generated response for rest server",
app_context.application_context)
response_true = copy.deepcopy(response_true)
return response_true
except FileErrors as e:
response_custom = Status.ERR_STATUS.value
response_custom['message'] = e.message
response = file_ops.error_handler(response_custom, e.code, False)
log_exception(
"non workflow_response some error occured while validating file",
app_context.application_context, e)
response = copy.deepcopy(response)
return response
except ServiceError as e:
response_custom = Status.ERR_STATUS.value
response_custom['message'] = str(e)
response = file_ops.error_handler(response_custom, "SERVICE_ERROR",
False)
log_exception(
"non workflow_response Something went wrong during ocr.",
app_context.application_context, e)
response = copy.deepcopy(response)
return response
def post(self):
json_data = request.get_json(force=True)
app_context.init()
app_context.application_context = json_data
log_info("Resource Vision_OCR service started", app_context.application_context)
try:
error_validator = ValidationResponse(DOWNLOAD_FOLDER)
if error_validator.format_error(json_data) is True:
response_gen = Response(json_data, DOWNLOAD_FOLDER)
response = response_gen.nonwf_response()
log_info("Resource Vision_OCR api response completed", app_context.application_context)
return jsonify(response)
except FormatError as e:
log_error("Resource Vision_OCR Input json format is not correct or dict_key is missing", app_context.application_context, e)
return Status.ERR_request_input_format.value
def workflow_response(self, task_id, task_starttime, debug_flush=False):
app_context.init()
app_context.application_context = {}
input_files, workflow_id, jobid, tool_name, step_order = file_ops.json_input_format(self.json_data)
log_info("workflow_response started the response generation", app_context.application_context)
error_validator = ValidationResponse(self.DOWNLOAD_FOLDER)
try:
error_validator.wf_keyerror(jobid, workflow_id, tool_name, step_order)
error_validator.inputfile_list_error(input_files)
output_file_response = list()
for i, item in enumerate(input_files):
input_filename, in_file_type, in_locale = file_ops.accessing_files(item)
self.json_data['taskID'] = task_id
app_context.application_context = self.json_data
if debug_flush == False:
bm_response = DocumentStructure(app_context=app_context, file_name=input_filename, lang=in_locale)
if bm_response['code'] == 200:
output_filename_json = file_ops.writing_json_file(i, bm_response['rsp'], self.DOWNLOAD_FOLDER)
file_res = file_ops.one_filename_response(input_filename, output_filename_json, in_locale, in_file_type)
output_file_response.append(file_res)
task_endtime = str(time.time()).replace('.', '')
response_true = CustomResponse(Status.SUCCESS.value, jobid, task_id)
response_success = response_true.success_response(workflow_id, task_starttime, task_endtime, tool_name, step_order, output_file_response)
response = copy.deepcopy(response_success)
log_info("successfully generated response for workflow", app_context.application_context)
return response
else:
post_error_wf(bm_response.code, bm_response.message, app_context.application_context, None)
return None
else:
log_info('flushing queue data, not handling file {}'.format(input_files), app_context.application_context)
post_error_wf(400, 'flushing queue data, not handling file {}'.format(input_files), app_context.application_context, None)
return None
except WorkflowkeyError as e:
response_custom = CustomResponse(Status.ERR_STATUS.value, jobid, task_id)
response_custom.status_code['message'] = str(e)
response = file_ops.error_handler(response_custom.status_code, "WORKFLOWKEY-ERROR", True)
log_exception("workflow_response workflow key error: key value missing", app_context.application_context, e)
response = copy.deepcopy(response)
return response
except FileErrors as e:
response_custom = CustomResponse(Status.ERR_STATUS.value, jobid, task_id)
response_custom.status_code['message'] = e.message
response = file_ops.error_handler(response_custom.status_code, e.code, True)
log_exception("workflow_response some error occured while validating file", app_context.application_context, e)
response = copy.deepcopy(response)
return response
except ServiceError as e:
response_custom = CustomResponse(Status.ERR_STATUS.value, jobid, task_id)
response_custom.status_code['message'] = str(e)
response = file_ops.error_handler(response_custom.status_code, "SERVICE_ERROR", True)
log_exception("workflow_response Something went wrong during pdf to block conversion.", app_context.application_context, e)
response = copy.deepcopy(response)
return response
from anuvaad_auditor.loghandler import log_info
from anuvaad_auditor.loghandler import log_error
from src.services import main
from src.services.get_underline import get_underline
from services import get_xml
from src.services.child_text_unify_to_parent import ChildTextUnify
from services.preprocess import prepocess_pdf_regions
from services.get_tables import page_num_correction, get_text_table_line_df
from src.services.ocr_text_utilities import tesseract_ocr
from src.services.get_response import process_image_df, process_table_df, df_to_json, process_line_df
from utilities.filesystem import (create_directory, read_directory_files)
import config
import src.utilities.app_context as app_context
app_context.init()
from src.services import main
def run_test(document_directory):
start_time = time.time()
files = read_directory_files(document_directory, "*.pdf")
for file in files:
app_context.application_context = dict({
'task_id':
'BM-DUMMY',
'filename':
os.path.basename(file)
})
def workflow_response(self, task_id, task_starttime):
app_context.init()
app_context.application_context = self.json_data
input_params, workflow_id, jobid, tool_name, step_order = file_ops.json_input_format(
self.json_data)
log_info("workflow_response started the response generation",
app_context.application_context)
error_validator = ValidationResponse(self.DOWNLOAD_FOLDER)
try:
error_validator.wf_keyerror(jobid, workflow_id, tool_name,
step_order)
# --------------------------
result = process_incoming_request(app_context, input_params[0],
jobid, workflow_id)
# --------------------------
task_endtime = eval(str(time.time()).replace('.', '')[0:13])
response_true = CustomResponse(Status.SUCCESS.value, jobid,
task_id)
response_success = response_true.success_response(
workflow_id, task_starttime, task_endtime, tool_name,
step_order, result)
log_info(
"workflow_response : successfully generated response for workflow",
app_context.application_context)
return response_success
except WorkflowkeyError as e:
response_custom = CustomResponse(Status.ERR_STATUS.value, jobid,
task_id)
response_custom.status_code['message'] = str(e)
response = file_ops.error_handler(response_custom.status_code,
"WORKFLOWKEY-ERROR", True)
log_exception(
"workflow_response workflow key error: key value missing",
app_context.application_context, e)
response = copy.deepcopy(response)
return response
except FileErrors as e:
response_custom = CustomResponse(Status.ERR_STATUS.value, jobid,
task_id)
response_custom.status_code['message'] = e.message
response = file_ops.error_handler(response_custom.status_code,
e.code, True)
log_exception(
"workflow_response some error occured while validating file",
app_context.application_context, e)
response = copy.deepcopy(response)
return response
except ServiceError as e:
response_custom = CustomResponse(Status.ERR_STATUS.value, jobid,
task_id)
response_custom.status_code['message'] = str(e)
response = file_ops.error_handler(response_custom.status_code,
"SERVICE_ERROR", True)
log_exception("workflow_response Something went wrong during ocr.",
app_context.application_context, e)
response = copy.deepcopy(response)
return response
