def process_submitted_files():
version = utils.get_tesseract_version()
oldest_file_path = get_oldest_file(BATCHSUBMITED)
logger.info(['I am inside processsubmitted files', oldest_file_path])
starttime = time.strftime("%c")
modifiedtime = time.strftime("%c")
filepath = oldest_file_path
filealtpath = os.path.join(BATCHINPROGRESS, os.path.basename(filepath))
filefinalpath = os.path.join(BATCHPROCESSED, os.path.basename(filealtpath))
logger.info([filepath])
logger.info([filealtpath])
jsonobj = json.load(open(filepath, 'r'))
logger.info([jsonobj])
logger.info([filepath, 'Constructing Header'])
jsonobj['header'] = {}
logger.info([filealtpath, 'adding start and modified time to header'])
jsonobj['header']['start time'] = starttime
jsonobj['header']['modified time'] = modifiedtime
logger.info([filealtpath, 'adding status to in progress'])
jsonobj['header']['status'] = "in progress"
logger.info([filealtpath, 'adding total count of subjects'])
jsonobj['header']['total'] = len(jsonobj['subjects'])
logger.info([filealtpath, 'adding complete flag and setting it to 0'])
jsonobj['header']['complete'] = 0
logger.info([filealtpath, 'adding tesseract version'])
jsonobj['header']['OCR engine'] = version
json.dump(jsonobj, open(filealtpath, "w"))
logger.info([filepath, ' Removing from batchsubmited folder '])
os.remove(filepath)
process_files(filealtpath)
def cron_jobs():
"""
logic for doing the cron job of copying over the file
from batchsubmited to batchprocessed. Later, building
url for each identifier in json file and doing a get
request to single file ocr url, saving the ocr returned
and dumping back the json file
"""
version = utils.get_tesseract_version()
if isBatchInProgress(process_path):
#logger.info(['Batch in progress'])
return
'''inprogress_file_path = get_oldest_file(BATCHINPROGRESS)
logger.info([inprogress_file_path, 'Inprogress file path'])
oldest_file_path = get_oldest_file(BATCHSUBMITED)
logger.info([oldest_file_path, 'Oldest File Path'])'''
#if oldest_file_path is None:
#logger.info(['no files found for batch processing'])
# return
try:
inprogress_file_path = get_oldest_file(BATCHINPROGRESS)
logger.info([inprogress_file_path, 'Inprogress file path'])
oldest_file_path = get_oldest_file(BATCHSUBMITED)
logger.info([oldest_file_path, 'Oldest File Path'])
if inprogress_file_path is not None:
logger.info('calling process_files')
process_files(inprogress_file_path)
elif oldest_file_path is not None:
logger.info('calling process_submitted_files')
logger.info('about to call process_submitted_files')
process_submitted_files()
except Exception as cronexcept:
logger.info([filealtpath, 'Exception: {0}'.format(cronexcept)])
logger.info([filealtpath, traceback.format_exc()])