def step(self):
'''
This class checks the output from a LIMB or OCR process
to see if it matches certain criteria
Require params process_title from command line
Other values taken from config.ini
In the case of errors a message will be returned
and sent to the previous step.
'''
error = None
try:
self.getVariables()
tools.ensureDirsExist(self.bw_pdf_input_dir,
self.color_pdf_input_dir,self.input_files_dir)
# Check if color pdf is ok
if not limb_tools.pageCountMatches(self.color_pdf_input_dir,self.input_files_dir,self.valid_exts):
raise DataError('PDF page count does not match input picture count in "{0}"!'.format(self.color_pdf_input_dir))
# Check if bw pdf is ok
if not limb_tools.pageCountMatches(self.bw_pdf_input_dir,self.preprocessed_input_files,self.valid_exts):
raise DataError('PDF page count does not match input picture count in "{0}"!'.format(self.bw_pdf_input_dir))
except IOError as e:
error = "IOError - {0}".format(e.strerror)
except DataError as e:
error = "Validation error - {0}.".format(e.strerror)
return error
def ocrIsReady(self):
'''
Check to see if OCR is finished
return boolean
'''
try:
# raises error if one of our directories is missing
tools.ensureDirsExist(self.pdf_input_dir, self.input_files)
except IOError as e:
msg = ('One of the output folder from OCR is not yet created.'
' Waiting for OCR to be ready. Error: {0}')
msg = msg.format(e.strerror)
self.debug_message(msg)
return False
# legr: we can use limb_tools generally - they are not Limb specific
# we should rename them someday
pdf_ok = limb_tools.pageCountMatches(self.pdf_input_dir,
self.input_files,
self.valid_exts)
if pdf_ok:
return True
return False