Update to Tesseract 5.0.0, Set version 0.1.0

This commit is contained in:
Patrick Jentsch
2022-01-04 11:42:55 +01:00
parent a0760487ae
commit e1b78b6ba4
7 changed files with 574 additions and 338 deletions

View File

@ -6,9 +6,10 @@ import os
import subprocess
import sys
CONTAINER_IMAGE = 'gitlab.ub.uni-bielefeld.de:4567/sfb1288inf/ocr:1.0.0'
CONTAINER_IMAGE = 'gitlab.ub.uni-bielefeld.de:4567/sfb1288inf/ocr:0.1.0'
CONTAINER_INPUT_DIR = '/input'
CONTAINER_OUTPUT_DIR = '/output'
CONTAINER_MODELS_DIR = '/usr/local/share/tessdata'
CONTAINER_LOG_DIR = '/logs'
UID = str(os.getuid())
GID = str(os.getgid())
@ -16,20 +17,25 @@ GID = str(os.getgid())
parser = ArgumentParser(add_help=False)
parser.add_argument('-i', '--input-dir')
parser.add_argument('-o', '--output-dir')
parser.add_argument('-m', '--model', action='extend', dest='models', nargs='+')
parser.add_argument('--log-dir')
args, remaining_args = parser.parse_known_args()
cmd = ['docker', 'run', '--rm', '-it', '-u', '{}:{}'.format(UID, GID)]
cmd = ['docker', 'run', '--rm', '-it', '-u', f'{UID}:{GID}']
if args.input_dir is not None:
mapping = os.path.abspath(args.input_dir) + ':' + CONTAINER_INPUT_DIR
mapping = f'{os.path.abspath(args.input_dir)}:{CONTAINER_INPUT_DIR}'
cmd += ['-v', mapping]
remaining_args += ['-i', CONTAINER_INPUT_DIR]
if args.output_dir is not None:
mapping = os.path.abspath(args.output_dir) + ':' + CONTAINER_OUTPUT_DIR
mapping = f'{os.path.abspath(args.output_dir)}:{CONTAINER_OUTPUT_DIR}'
cmd += ['-v', mapping]
remaining_args += ['-o', CONTAINER_OUTPUT_DIR]
if args.models is not None:
for model in args.models:
mapping = f'{os.path.abspath(model)}:{CONTAINER_MODELS_DIR}/{os.path.basename(model)}' # noqa
cmd += ['-v', mapping]
if args.log_dir is not None:
mapping = os.path.abspath(args.log_dir) + ':' + CONTAINER_LOG_DIR
mapping = '{os.path.abspath(args.log_dir)}:{CONTAINER_LOG_DIR}'
cmd += ['-v', mapping]
remaining_args += ['--log-dir', CONTAINER_LOG_DIR]
cmd.append(CONTAINER_IMAGE)