def upload_sample_documents(self):
repo = CmisClient(self.settings["cmis_url"], self.settings["cmis_username"], self.settings["cmis_password"]).defaultRepository
try:
repo.getObjectByPath(self.settings["capture_folder"])
self.LOGGER.debug("Capture folder available")
except:
self.LOGGER.debug("CMIS folder not present. Uploading demo files.")
parent_folder_name = "/".join(self.settings["capture_folder"].split("/")[:-1])
parent_folder = repo.getObjectByPath(parent_folder_name)
demoFolder = repo.createFolder(parent_folder, self.settings["capture_folder"].split("/")[-1])
demoFolder.createDocument('Invoice01.pdf', contentFile=open("test/data-barcode-skeleton/doc001/Invoice01.pdf"))
demoFolder.createDocument('Invoice02.pdf', contentFile=open("test/data-barcode-skeleton/doc002/Invoice02.pdf"))
def download_from_cmis(self):
if "cmis_url" not in self.settings:
self.LOGGER.warn("No CMIS repo specified. Processing documents is datadir.")
return
repo = CmisClient(self.settings["cmis_url"], self.settings["cmis_username"], self.settings["cmis_password"]).defaultRepository
for document in repo.getObjectByPath(self.settings["capture_folder"]).getChildren():
document_name = document.getName()
docid = document.getObjectId()
docid = docid.replace("workspace://SpacesStore/", "")
# remove version identifier ;1.0
if ";" in docid:
docid = docid.split(";")[0]
document_workdir = os.path.join(self.settings["datadir"], docid)
if not os.path.exists(document_workdir):
os.makedirs(document_workdir)
local_document_path = os.path.join(document_workdir, document_name)
if os.path.exists(local_document_path):
self.LOGGER.debug("{} exists. skipping download from CMIS".format(local_document_path))
else:
self.LOGGER.debug("downloading {}".format(document_name))
with open(local_document_path, "w") as fh:
fh.write(document.getContentStream().read())
