def step(self):
error = None
try:
self.getVariables()
# Get the number of valid images
image_count = tools.getFileCountWithExtension(self.image_path, self.valid_exts)
self.debug_message("Der blev optalt {} billeder".format(image_count))
# write the number of images to a Goobi property. It sometimes fail, hence the retry stuff
max_retry = 10
retry = 0
ok = False
while not ok:
ok = self.goobi_com.addProperty(name=self.property_name, value=image_count, overwrite=True)
retry += 1
if not ok:
self.debug_message("Billedantallet kunne ikke gemmes! Proever igen... (retry={})".format(retry))
time.sleep(5)
if retry > max_retry:
error = "Fejl, Kunne ikke gemme billedantallet i Goobi's database!"
self.debug_message(error)
# return error
# Return None for now, we don't want a wep api lockup to stop the workflow
# Until a "sudo service tomcat restart", imagecount will be 0.
# todo: solve problem with web api lockup
return None
self.debug_message("Billedantallet ({}) blev gemt korrekt i forsoeg nr {}".format(image_count, retry))
# not sure which exceptions to expect...
except ValueError as e:
error = str(e.with_traceback)
return error
def altoFileCountMatches(alto_dir, input_files_dir,valid_exts):
'''
Ensure number of alto files is the same as the
number of input files.
Return boolean
'''
numAlto = len(os.listdir(alto_dir))
numInputFiles = tools.getFileCountWithExtension(input_files_dir,valid_exts)
return numAlto == numInputFiles
def pageCountMatches(pdf_input_dir,input_files_dir,valid_exts):
'''
Compare num pages in pdfinfo with pages in input
picture directory.
return boolean
'''
pdf = tools.getFirstFileWithExtension(pdf_input_dir, '.pdf')
pdfInfo = tools.pdfinfo(os.path.join(pdf_input_dir, pdf))
numPages = int(pdfInfo['Pages'])
numInputFiles = tools.getFileCountWithExtension(input_files_dir,valid_exts)
return numPages == numInputFiles
