|
1
|
|
|
""" |
|
2
|
|
|
OCR-D CLI: bash library |
|
3
|
|
|
|
|
4
|
|
|
.. click:: ocrd.cli.bashlib:bashlib_cli |
|
5
|
|
|
:prog: ocrd bashlib |
|
6
|
|
|
:nested: full |
|
7
|
|
|
|
|
8
|
|
|
""" |
|
9
|
|
|
from __future__ import print_function |
|
10
|
|
|
import sys |
|
11
|
|
|
from os.path import isfile |
|
12
|
|
|
import click |
|
13
|
|
|
|
|
14
|
|
|
from ocrd.constants import BASHLIB_FILENAME |
|
15
|
|
|
import ocrd.constants |
|
16
|
|
|
import ocrd_utils.constants |
|
17
|
|
|
import ocrd_models.constants |
|
18
|
|
|
import ocrd_validators.constants |
|
19
|
|
|
from ocrd.decorators import ( |
|
20
|
|
|
parameter_option, |
|
21
|
|
|
parameter_override_option, |
|
22
|
|
|
ocrd_loglevel |
|
23
|
|
|
) |
|
24
|
|
|
from ocrd_utils import ( |
|
25
|
|
|
is_local_filename, |
|
26
|
|
|
get_local_filename, |
|
27
|
|
|
initLogging, |
|
28
|
|
|
make_file_id |
|
29
|
|
|
) |
|
30
|
|
|
from ocrd.resolver import Resolver |
|
31
|
|
|
from ocrd.processor import Processor |
|
32
|
|
|
|
|
33
|
|
|
# ---------------------------------------------------------------------- |
|
34
|
|
|
# ocrd bashlib |
|
35
|
|
|
# ---------------------------------------------------------------------- |
|
36
|
|
|
|
|
37
|
|
|
@click.group('bashlib') |
|
38
|
|
|
def bashlib_cli(): |
|
39
|
|
|
""" |
|
40
|
|
|
Work with bash library |
|
41
|
|
|
""" |
|
42
|
|
|
|
|
43
|
|
|
# ---------------------------------------------------------------------- |
|
44
|
|
|
# ocrd bashlib filename |
|
45
|
|
|
# ---------------------------------------------------------------------- |
|
46
|
|
|
|
|
47
|
|
|
@bashlib_cli.command('filename') |
|
48
|
|
|
def bashlib_filename(): |
|
49
|
|
|
""" |
|
50
|
|
|
Dump the bash library filename for sourcing by shell scripts |
|
51
|
|
|
|
|
52
|
|
|
For functions exported by bashlib, see `<../../README.md>`_ |
|
53
|
|
|
""" |
|
54
|
|
|
print(BASHLIB_FILENAME) |
|
55
|
|
|
|
|
56
|
|
|
@bashlib_cli.command('constants') |
|
57
|
|
|
@click.argument('name') |
|
58
|
|
|
def bashlib_constants(name): |
|
59
|
|
|
""" |
|
60
|
|
|
Query constants from ocrd_utils and ocrd_models |
|
61
|
|
|
""" |
|
62
|
|
|
all_constants = {} |
|
63
|
|
|
for src in [ocrd.constants, ocrd_utils.constants, ocrd_models.constants, ocrd_validators.constants]: |
|
64
|
|
|
for k in src.__all__: |
|
65
|
|
|
all_constants[k] = src.__dict__[k] |
|
66
|
|
|
if name in ['*', 'KEYS', '__all__']: |
|
67
|
|
|
print(sorted(all_constants.keys())) |
|
68
|
|
|
if name not in all_constants: |
|
69
|
|
|
print("ERROR: name '%s' is not a known constant" % name, file=sys.stderr) |
|
70
|
|
|
sys.exit(1) |
|
71
|
|
|
val = all_constants[name] |
|
72
|
|
|
if isinstance(val, dict): |
|
73
|
|
|
# make this bash-friendly (show initialization for associative array) |
|
74
|
|
|
for key in val: |
|
75
|
|
|
print("[%s]=%s" % (key, val[key]), end=' ') |
|
76
|
|
|
else: |
|
77
|
|
|
print(val) |
|
78
|
|
|
|
|
79
|
|
|
@bashlib_cli.command('input-files') |
|
80
|
|
|
@click.option('-m', '--mets', help="METS to process", default="mets.xml") |
|
81
|
|
|
@click.option('-w', '--working-dir', help="Working Directory") |
|
82
|
|
|
@click.option('-I', '--input-file-grp', help='File group(s) used as input.', default='INPUT') |
|
83
|
|
|
@click.option('-O', '--output-file-grp', help='File group(s) used as output.', default='OUTPUT') |
|
84
|
|
|
# repeat some other processor options for convenience (will be ignored here) |
|
85
|
|
|
@click.option('-g', '--page-id', help="ID(s) of the pages to process") |
|
86
|
|
|
@click.option('--overwrite', is_flag=True, default=False, help="Remove output pages/images if they already exist") |
|
87
|
|
|
@parameter_option |
|
88
|
|
|
@parameter_override_option |
|
89
|
|
|
@ocrd_loglevel |
|
90
|
|
|
def bashlib_input_files(**kwargs): |
|
91
|
|
|
""" |
|
92
|
|
|
List input files for processing |
|
93
|
|
|
|
|
94
|
|
|
Instantiate a processor and workspace from the given processing options. |
|
95
|
|
|
Then loop through the input files of the input fileGrp, and for each one, |
|
96
|
|
|
print its `url`, `ID`, `mimetype` and `pageId`, as well as its recommended |
|
97
|
|
|
`outputFileId` (from ``make_file_id``). |
|
98
|
|
|
|
|
99
|
|
|
(The printing format is one associative array initializer per line.) |
|
100
|
|
|
""" |
|
101
|
|
|
initLogging() |
|
102
|
|
|
mets = kwargs.pop('mets') |
|
103
|
|
|
working_dir = kwargs.pop('working_dir') |
|
104
|
|
|
if is_local_filename(mets) and not isfile(get_local_filename(mets)): |
|
105
|
|
|
msg = "File does not exist: %s" % mets |
|
106
|
|
|
raise Exception(msg) |
|
107
|
|
|
resolver = Resolver() |
|
108
|
|
|
workspace = resolver.workspace_from_url(mets, working_dir) |
|
109
|
|
|
processor = Processor(workspace, |
|
110
|
|
|
ocrd_tool=None, |
|
111
|
|
|
page_id=kwargs['page_id'], |
|
112
|
|
|
input_file_grp=kwargs['input_file_grp'], |
|
113
|
|
|
output_file_grp=kwargs['output_file_grp']) |
|
114
|
|
|
for input_file in processor.input_files: |
|
115
|
|
|
for field in ['url', 'ID', 'mimetype', 'pageId']: |
|
116
|
|
|
# make this bash-friendly (show initialization for associative array) |
|
117
|
|
|
print("[%s]='%s'" % (field, getattr(input_file, field)), end=' ') |
|
118
|
|
|
print("[outputFileId]='%s'" % make_file_id(input_file, kwargs['output_file_grp'])) |
|
119
|
|
|
|
OCR-D /
core
