def main():
parser = ArgumentParser(description="{description}". \
format(description = __description__),
epilog="{copyright}; ". \
format(copyright = __copyright__) + \
"written by {author} <{email}>.". \
format(author = __author__,
email = __email__))
parser.add_argument("-v", help="See the version of this program",
action="version", version=__version__)
parser.add_argument( \
'-b','-verbose',help="set verbose logging",
action='store_const',dest='log_level',
const=INFO \
)
parser.add_argument( \
'-d','--debugging',help="set debugging logging",
action='store_const',dest='log_level',
const=DEBUG \
)
parser.add_argument( \
'-l','--log_loc',help="save logging to a file",
action="store_const",dest="log_loc",
const='./{progname}.log'. \
format(progname=__file__) \
)
parser.add_argument( \
'--db_url',help="Enter a db url",action='store' \
)
parser.add_argument( \
'--root',help="Enter the root of the repository",
action='store')
parser.add_argument( \
'--object_pattern',help="Enter the regex pattern " + \
"to match an object",
action='store')
parser.add_argument( \
'--page_pattern',help="Enter the regex pattern " + \
"to match a page",
action='store')
parser.add_argument( \
'accessions',nargs="*",action='store',
help="Enter 1 or more accession " + \
"identifiers to process" \
)
args = parser.parse_args()
log_format = Formatter( \
"[%(levelname)s] %(asctime)s " + \
"= %(message)s",
datefmt="%Y-%m-%dT%H:%M:%S" \
)
global logger
logger = getLogger( \
"lib.uchicago.repository.logger" \
)
ch = StreamHandler()
ch.setFormatter(log_format)
try:
logger.setLevel(args.log_level)
except TypeError:
logger.setLevel(INFO)
if args.log_loc:
fh = FileHandler(args.log_loc)
fh.setFormatter(log_format)
logger.addHandler(fh)
logger.addHandler(ch)
db = Database(args.db_url, ['record','file'])
class Record(db.base):
__table__ = Table('record', db.metadata, autoload=True)
class File(db.base):
__table__ = Table('file', db.metadata, autoload=True)
query = db.session.query(File).filter(File.accession.in_(args.accessions))
if args.root:
batch = Batch(args.root, query = query)
items = batch.find_items(from_db = True)
batch.set_items(items)
else:
raise ValueError("need to include a root")
try:
all_objects = []
for item in batch.get_items():
accession = item.find_file_accession()
item.set_accession(accession)
canon = item.find_canonical_filepath()
item.set_canonical_filepath(canon)
search_pattern = item.find_matching_object_pattern( \
re_compile("(mvol)/(\w{4})/(\w{4})/(\w{4})/" +
"(mvol)-(\w{4})-(\w{4})-(\w{4})"))
if search_pattern.status == True:
potential_identifier = '-'.join(search_pattern.data.groups())
is_an_object_already_present = [x for x in all_objects \
if x.identifier == \
potential_identifier]
if is_an_object_already_present:
logger.debug("found this id already")
else:
logger.debug("this id is new!")
new_object = DigitalObject(potential_identifier)
all_objects.append(new_object)
logger.debug(potential_identifier)
return 0
except KeyboardInterrupt:
logger.error("Program aborted manually")
return 131
def main():
parser = ArgumentParser(description="{description}". \
format(description = __description__),
epilog="{copyright}; ". \
format(copyright=__copyright__) + \
"written by {name} ".format(name=__author__) + \
" <{email}> ".format(email=__email__) + \
"University of Chicago")
parser.add_argument("-v", help="See the version of this program",
action="version", version=__version__)
parser.add_argument( \
'-b','-verbose',help="set verbose logging",
action='store_const',dest='log_level',
const=INFO \
)
parser.add_argument( \
'-d','--debugging',help="set debugging logging",
action='store_const',dest='log_level',
const=DEBUG \
)
parser.add_argument( \
'-l','--log_loc',help="save logging to a file",
action="store_const",dest="log_loc",
const='./{progname}.log'. \
format(progname=argv[0]) \
)
parser.add_argument("location_root",help="Enter the root " + \
"of the directory path",
action="store")
parser.add_argument("directory_path",
help="Enter a directory that you need to work on ",
action='store')
args = parser.parse_args()
log_format = Formatter( \
"[%(levelname)s] %(asctime)s " + \
"= %(message)s",
datefmt="%Y-%m-%dT%H:%M:%S" \
)
global logger
logger = getLogger( \
"lib.uchicago.repository.logger" \
)
ch = StreamHandler()
ch.setFormatter(log_format)
try:
logger.setLevel(args.log_level)
except TypeError:
logger.setLevel(INFO)
if args.log_loc:
fh = FileHandler(args.log_loc)
fh.setFormatter(log_format)
logger.addHandler(fh)
logger.addHandler(ch)
try:
b = Batch(args.location_root, directory = args.directory_path)
generator_object = b.find_items(from_directory=True)
logger.debug(generator_object)
b.set_items(generator_object)
stdout.write("begin transaction;\n")
for a_file in b.get_items():
if a_file.test_readability():
file_hash = a_file.find_hash_of_file(sha256)
mime = a_file.find_file_mime_type()
size = a_file.find_file_size()
accession = a_file.find_file_accession()
a_file.set_file_mime_type(mime)
a_file.set_file_size(size)
a_file.set_hash(file_hash)
a_file.set_accession(accession)
out_string = "insert into file (filepath,accession," + \
"mimetype,size,checksum) values (" + \
"\"{path}\",\"{accession}\",\"{mimetype}\"". \
format(path = a_file.filepath,
accession = a_file.get_accession(),
mimetype = a_file.get_file_mime_type()) + \
",{filesize},\"{filehash}\");\n". \
format(filesize = a_file.get_file_size(),
filehash = a_file.get_hash())
stdout.write(out_string)
else:
logger.error("{path} could not be read". \
format(path=a_file.filepath))
stdout.write("commit;\n")
return 0
except KeyboardInterrupt:
logger.warn("Program aborted manually")
return 131
