|
1
|
|
|
""" |
|
2
|
|
|
OCR-D CLI: workspace management |
|
3
|
|
|
|
|
4
|
|
|
.. click:: ocrd.cli.workspace:workspace_cli |
|
5
|
|
|
:prog: ocrd workspace |
|
6
|
|
|
:nested: full |
|
7
|
|
|
""" |
|
8
|
|
|
from os import rmdir, unlink |
|
9
|
|
|
from os.path import dirname, relpath, normpath, exists, join, isabs, isdir |
|
10
|
|
|
from pathlib import Path |
|
11
|
|
|
from json import loads, dumps |
|
12
|
|
|
import sys |
|
13
|
|
|
from glob import glob # XXX pathlib.Path.glob does not support absolute globs |
|
14
|
|
|
import re |
|
15
|
|
|
import time |
|
16
|
|
|
|
|
17
|
|
|
import click |
|
18
|
|
|
|
|
19
|
|
|
from ocrd import Resolver, Workspace, WorkspaceValidator, WorkspaceBackupManager |
|
20
|
|
|
from ocrd.mets_server import OcrdMetsServer |
|
21
|
|
|
from ocrd_utils import ( |
|
22
|
|
|
getLogger, |
|
23
|
|
|
initLogging, |
|
24
|
|
|
pushd_popd, |
|
25
|
|
|
EXT_TO_MIME, |
|
26
|
|
|
safe_filename, |
|
27
|
|
|
parse_json_string_or_file, |
|
28
|
|
|
partition_list, |
|
29
|
|
|
DEFAULT_METS_BASENAME, |
|
30
|
|
|
) |
|
31
|
|
|
from ocrd.decorators import mets_find_options |
|
32
|
|
|
from . import command_with_replaced_help |
|
33
|
|
|
from ocrd_models.constants import METS_PAGE_DIV_ATTRIBUTE |
|
34
|
|
|
|
|
35
|
|
|
|
|
36
|
|
|
class WorkspaceCtx(): |
|
37
|
|
|
|
|
38
|
|
|
def __init__(self, directory, mets_url, mets_basename=DEFAULT_METS_BASENAME, mets_server_url=None, automatic_backup=False): |
|
39
|
|
|
self.log = getLogger('ocrd.cli.workspace') |
|
40
|
|
|
if mets_basename: |
|
41
|
|
|
self.log.warning(DeprecationWarning('--mets-basename is deprecated. Use --mets/--directory instead.')) |
|
42
|
|
|
self.resolver = Resolver() |
|
43
|
|
|
self.directory, self.mets_url, self.mets_basename, self.mets_server_url = \ |
|
44
|
|
|
self.resolver.resolve_mets_arguments(directory, mets_url, mets_basename, mets_server_url) |
|
45
|
|
|
self.automatic_backup = automatic_backup |
|
46
|
|
|
|
|
47
|
|
|
def workspace(self): |
|
48
|
|
|
return Workspace( |
|
49
|
|
|
self.resolver, |
|
50
|
|
|
directory=self.directory, |
|
51
|
|
|
mets_basename=self.mets_basename, |
|
52
|
|
|
automatic_backup=self.automatic_backup, |
|
53
|
|
|
mets_server_url=self.mets_server_url, |
|
54
|
|
|
) |
|
55
|
|
|
|
|
56
|
|
|
def backup_manager(self): |
|
57
|
|
|
return WorkspaceBackupManager(self.workspace()) |
|
58
|
|
|
|
|
59
|
|
|
|
|
60
|
|
|
pass_workspace = click.make_pass_decorator(WorkspaceCtx) |
|
61
|
|
|
|
|
62
|
|
|
|
|
63
|
|
|
# ---------------------------------------------------------------------- |
|
64
|
|
|
# ocrd workspace |
|
65
|
|
|
# ---------------------------------------------------------------------- |
|
66
|
|
|
|
|
67
|
|
|
@click.group("workspace") |
|
68
|
|
|
@click.option('-d', '--directory', envvar='WORKSPACE_DIR', type=click.Path(file_okay=False), metavar='WORKSPACE_DIR', |
|
69
|
|
|
help='Changes the workspace folder location [default: METS_URL directory or .]"') |
|
70
|
|
|
@click.option('-M', '--mets-basename', default=None, help='METS file basename. Deprecated, use --mets/--directory') |
|
71
|
|
|
@click.option('-m', '--mets', default=None, metavar="METS_URL", |
|
72
|
|
|
help='The path/URL of the METS file [default: WORKSPACE_DIR/mets.xml]') |
|
73
|
|
|
@click.option('-U', '--mets-server-url', 'mets_server_url', help="TCP host URI or UDS path of METS server") |
|
74
|
|
|
@click.option('--backup', default=False, help="Backup mets.xml whenever it is saved.", is_flag=True) |
|
75
|
|
|
@click.pass_context |
|
76
|
|
|
def workspace_cli(ctx, directory, mets, mets_basename, mets_server_url, backup): |
|
77
|
|
|
""" |
|
78
|
|
|
Managing workspaces |
|
79
|
|
|
|
|
80
|
|
|
A workspace comprises a METS file and a directory as point of reference. |
|
81
|
|
|
|
|
82
|
|
|
Operates on the file system directly or via a METS server |
|
83
|
|
|
(already running via some prior `server start` subcommand). |
|
84
|
|
|
""" |
|
85
|
|
|
initLogging() |
|
86
|
|
|
ctx.obj = WorkspaceCtx( |
|
87
|
|
|
directory, |
|
88
|
|
|
mets_url=mets, |
|
89
|
|
|
mets_basename=mets_basename, |
|
90
|
|
|
mets_server_url=mets_server_url, |
|
91
|
|
|
automatic_backup=backup |
|
92
|
|
|
) |
|
93
|
|
|
|
|
94
|
|
|
|
|
95
|
|
|
# ---------------------------------------------------------------------- |
|
96
|
|
|
# ocrd workspace validate |
|
97
|
|
|
# ---------------------------------------------------------------------- |
|
98
|
|
|
|
|
99
|
|
|
@workspace_cli.command('validate', cls=command_with_replaced_help( |
|
100
|
|
|
(r' \[METS_URL\]', ''))) # XXX deprecated argument |
|
101
|
|
|
@pass_workspace |
|
102
|
|
|
@click.option('-a', '--download', is_flag=True, help="Download all files") |
|
103
|
|
|
@click.option('-s', '--skip', help="Tests to skip", default=[], multiple=True, type=click.Choice( |
|
104
|
|
|
['imagefilename', 'alternativeimage_filename', 'alternativeimage_comments', 'dimension', 'pixel_density', |
|
105
|
|
|
'page', 'page_xsd', 'url', 'mets_fileid_page_pcgtsid', 'mets_unique_identifier', 'mets_files', 'mets_xsd'])) |
|
106
|
|
|
@click.option('--page-textequiv-consistency', '--page-strictness', type=click.Choice(['strict', 'lax', 'fix', 'off']), |
|
107
|
|
|
default='strict', help="How strict to check PAGE multi-level textequiv consistency") |
|
108
|
|
|
@click.option('--page-coordinate-consistency', help="How fierce to check PAGE multi-level coordinate consistency", |
|
109
|
|
|
type=click.Choice(['poly', 'baseline', 'both', 'off']), default='poly') |
|
110
|
|
|
@click.argument('mets_url', default=None, required=False) |
|
111
|
|
|
def workspace_validate(ctx, mets_url, download, skip, page_textequiv_consistency, page_coordinate_consistency): |
|
112
|
|
|
""" |
|
113
|
|
|
Validate a workspace |
|
114
|
|
|
|
|
115
|
|
|
METS_URL can be a URL, an absolute path or a path relative to $PWD. |
|
116
|
|
|
If not given, use --mets accordingly. |
|
117
|
|
|
|
|
118
|
|
|
Check that the METS and its referenced file contents |
|
119
|
|
|
abide by the OCR-D specifications. |
|
120
|
|
|
""" |
|
121
|
|
|
LOG = getLogger('ocrd.cli.workspace.validate') |
|
122
|
|
|
if mets_url: |
|
123
|
|
|
LOG.warning(DeprecationWarning("Use 'ocrd workspace --mets METS init' instead of " |
|
124
|
|
|
"argument 'METS_URL' ('%s')" % mets_url)) |
|
125
|
|
|
else: |
|
126
|
|
|
mets_url = ctx.mets_url |
|
127
|
|
|
report = WorkspaceValidator.validate( |
|
128
|
|
|
ctx.resolver, |
|
129
|
|
|
mets_url, |
|
130
|
|
|
src_dir=ctx.directory, |
|
131
|
|
|
skip=skip, |
|
132
|
|
|
download=download, |
|
133
|
|
|
page_strictness=page_textequiv_consistency, |
|
134
|
|
|
page_coordinate_consistency=page_coordinate_consistency |
|
135
|
|
|
) |
|
136
|
|
|
print(report.to_xml()) |
|
137
|
|
|
if not report.is_valid: |
|
138
|
|
|
sys.exit(128) |
|
139
|
|
|
|
|
140
|
|
|
|
|
141
|
|
|
# ---------------------------------------------------------------------- |
|
142
|
|
|
# ocrd workspace clone |
|
143
|
|
|
# ---------------------------------------------------------------------- |
|
144
|
|
|
|
|
145
|
|
|
@workspace_cli.command('clone', cls=command_with_replaced_help( |
|
146
|
|
|
(r' \[WORKSPACE_DIR\]', ''))) # XXX deprecated argument |
|
147
|
|
|
@click.option('-f', '--clobber-mets', help="Overwrite existing METS file", default=False, is_flag=True) |
|
148
|
|
|
@click.option('-a', '--download', is_flag=True, help="Download all selected files and add local FLocat " |
|
149
|
|
|
"path references in METS file afterwards") |
|
150
|
|
|
@click.argument('mets_url') |
|
151
|
|
|
@mets_find_options |
|
152
|
|
|
# XXX deprecated |
|
153
|
|
|
@click.argument('workspace_dir', default=None, required=False) |
|
154
|
|
|
@pass_workspace |
|
155
|
|
|
def workspace_clone(ctx, clobber_mets, download, file_grp, file_id, page_id, mimetype, |
|
156
|
|
|
include_fileGrp, exclude_fileGrp, mets_url, workspace_dir): |
|
157
|
|
|
""" |
|
158
|
|
|
Create a workspace from METS_URL and return the directory |
|
159
|
|
|
|
|
160
|
|
|
METS_URL can be a URL, an absolute path or a path relative to $PWD. |
|
161
|
|
|
METS_URL can also be an OAI-PMH GetRecord URL wrapping a METS file. |
|
162
|
|
|
|
|
163
|
|
|
Additional options pertain to the selection of files / fileGrps / pages |
|
164
|
|
|
to be downloaded, if --download is used. |
|
165
|
|
|
""" |
|
166
|
|
|
LOG = getLogger('ocrd.cli.workspace.clone') |
|
167
|
|
|
if workspace_dir: |
|
168
|
|
|
LOG.warning(DeprecationWarning("Use 'ocrd workspace --directory DIR clone' instead of " |
|
169
|
|
|
"argument 'WORKSPACE_DIR' ('%s')" % workspace_dir)) |
|
170
|
|
|
ctx.directory = workspace_dir |
|
171
|
|
|
|
|
172
|
|
|
assert not ctx.mets_server_url, \ |
|
173
|
|
|
f"clone cannot be performed with METS Server - stop server, rerun without -U {ctx.mets_server_url}" |
|
174
|
|
|
workspace = ctx.resolver.workspace_from_url( |
|
175
|
|
|
mets_url, |
|
176
|
|
|
dst_dir=ctx.directory, |
|
177
|
|
|
mets_basename=ctx.mets_basename, |
|
178
|
|
|
clobber_mets=clobber_mets, |
|
179
|
|
|
download=download, |
|
180
|
|
|
fileGrp=file_grp, |
|
181
|
|
|
ID=file_id, |
|
182
|
|
|
pageId=page_id, |
|
183
|
|
|
mimetype=mimetype, |
|
184
|
|
|
include_fileGrp=include_fileGrp, |
|
185
|
|
|
exclude_fileGrp=exclude_fileGrp, |
|
186
|
|
|
) |
|
187
|
|
|
workspace.save_mets() |
|
188
|
|
|
print(workspace.directory) |
|
189
|
|
|
|
|
190
|
|
|
|
|
191
|
|
|
# ---------------------------------------------------------------------- |
|
192
|
|
|
# ocrd workspace init |
|
193
|
|
|
# ---------------------------------------------------------------------- |
|
194
|
|
|
|
|
195
|
|
|
@workspace_cli.command('init', cls=command_with_replaced_help( |
|
196
|
|
|
(r' \[DIRECTORY\]', ''))) # XXX deprecated argument |
|
197
|
|
|
@click.option('-f', '--clobber-mets', help="Clobber mets.xml if it exists", is_flag=True, default=False) |
|
198
|
|
|
# XXX deprecated |
|
199
|
|
|
@click.argument('directory', default=None, required=False) |
|
200
|
|
|
@pass_workspace |
|
201
|
|
|
def workspace_init(ctx, clobber_mets, directory): |
|
202
|
|
|
""" |
|
203
|
|
|
Create a workspace with an empty METS file in DIRECTORY or CWD. |
|
204
|
|
|
|
|
205
|
|
|
""" |
|
206
|
|
|
LOG = getLogger('ocrd.cli.workspace.init') |
|
207
|
|
|
if directory: |
|
208
|
|
|
LOG.warning(DeprecationWarning("Use 'ocrd workspace --directory DIR init' instead of " |
|
209
|
|
|
"argument 'DIRECTORY' ('%s')" % directory)) |
|
210
|
|
|
ctx.directory = directory |
|
211
|
|
|
assert not ctx.mets_server_url, \ |
|
212
|
|
|
f"init cannot be performed with METS Server - stop server, rerun without -U {ctx.mets_server_url}" |
|
213
|
|
|
workspace = ctx.resolver.workspace_from_nothing( |
|
214
|
|
|
directory=ctx.directory, |
|
215
|
|
|
mets_basename=ctx.mets_basename, |
|
216
|
|
|
clobber_mets=clobber_mets, |
|
217
|
|
|
) |
|
218
|
|
|
workspace.save_mets() |
|
219
|
|
|
print(workspace.directory) |
|
220
|
|
|
|
|
221
|
|
|
|
|
222
|
|
|
# ---------------------------------------------------------------------- |
|
223
|
|
|
# ocrd workspace add |
|
224
|
|
|
# ---------------------------------------------------------------------- |
|
225
|
|
|
|
|
226
|
|
|
@workspace_cli.command('add') |
|
227
|
|
|
@click.option('-G', '--file-grp', help="fileGrp USE", required=True, metavar='FILE_GRP') |
|
228
|
|
|
@click.option('-i', '--file-id', help="ID for the file", required=True, metavar='FILE_ID') |
|
229
|
|
|
@click.option('-m', '--mimetype', help="Media type of the file. Guessed from extension if not provided", |
|
230
|
|
|
required=False, metavar='TYPE') |
|
231
|
|
|
@click.option('-g', '--page-id', help="ID of the physical page", metavar='PAGE_ID') |
|
232
|
|
|
@click.option('-C', '--check-file-exists', help="Whether to ensure FNAME exists", is_flag=True, default=False) |
|
233
|
|
|
@click.option('--ignore', help="Do not check whether file exists.", default=False, is_flag=True) |
|
234
|
|
|
@click.option('--force', help="If file with ID already exists, replace it. No effect if --ignore is set.", |
|
235
|
|
|
default=False, is_flag=True) |
|
236
|
|
|
@click.argument('fname', required=True) |
|
237
|
|
|
@pass_workspace |
|
238
|
|
|
def workspace_add_file(ctx, file_grp, file_id, mimetype, page_id, ignore, check_file_exists, force, fname): |
|
239
|
|
|
""" |
|
240
|
|
|
Add a file or http(s) URL FNAME to METS in a workspace. |
|
241
|
|
|
If FNAME is not an http(s) URL and is not a workspace-local existing file, try to copy to workspace. |
|
242
|
|
|
""" |
|
243
|
|
|
workspace = ctx.workspace() |
|
244
|
|
|
|
|
245
|
|
|
log = getLogger('ocrd.cli.workspace.add') |
|
246
|
|
|
if not mimetype: |
|
247
|
|
|
try: |
|
248
|
|
|
mimetype = EXT_TO_MIME[Path(fname).suffix] |
|
249
|
|
|
log.info("Guessed mimetype to be %s" % mimetype) |
|
250
|
|
|
except KeyError: |
|
251
|
|
|
log.error("Cannot guess mimetype from extension '%s' for '%s'. " |
|
252
|
|
|
"Set --mimetype explicitly" % (Path(fname).suffix, fname)) |
|
253
|
|
|
|
|
254
|
|
|
log.debug("Adding '%s'", fname) |
|
255
|
|
|
local_filename = None |
|
256
|
|
|
if not (fname.startswith('http://') or fname.startswith('https://')): |
|
257
|
|
|
if not fname.startswith(ctx.directory): |
|
258
|
|
|
if not isabs(fname) and exists(join(ctx.directory, fname)): |
|
259
|
|
|
fname = join(ctx.directory, fname) |
|
260
|
|
|
else: |
|
261
|
|
|
log.debug("File '%s' is not in workspace, copying", fname) |
|
262
|
|
|
try: |
|
263
|
|
|
fname = ctx.resolver.download_to_directory(ctx.directory, fname, subdir=file_grp) |
|
264
|
|
|
except FileNotFoundError: |
|
265
|
|
|
if check_file_exists: |
|
266
|
|
|
log.error("File '%s' does not exist, halt execution!" % fname) |
|
267
|
|
|
sys.exit(1) |
|
268
|
|
|
if check_file_exists and not exists(fname): |
|
269
|
|
|
log.error("File '%s' does not exist, halt execution!" % fname) |
|
270
|
|
|
sys.exit(1) |
|
271
|
|
|
if fname.startswith(ctx.directory): |
|
272
|
|
|
fname = relpath(fname, ctx.directory) |
|
273
|
|
|
local_filename = fname |
|
274
|
|
|
|
|
275
|
|
|
if not page_id: |
|
276
|
|
|
log.warning("You did not provide '--page-id/-g', so the file you added is not linked to a specific page.") |
|
277
|
|
|
kwargs = { |
|
278
|
|
|
'file_id': file_id, |
|
279
|
|
|
'mimetype': mimetype, |
|
280
|
|
|
'page_id': page_id, |
|
281
|
|
|
'force': force, |
|
282
|
|
|
'ignore': ignore, |
|
283
|
|
|
'local_filename': local_filename, |
|
284
|
|
|
'url': fname |
|
285
|
|
|
} |
|
286
|
|
|
workspace.add_file(file_grp, **kwargs) |
|
287
|
|
|
workspace.save_mets() |
|
288
|
|
|
|
|
289
|
|
|
|
|
290
|
|
|
# ---------------------------------------------------------------------- |
|
291
|
|
|
# ocrd workspace bulk-add |
|
292
|
|
|
# ---------------------------------------------------------------------- |
|
293
|
|
|
|
|
294
|
|
|
# pylint: disable=broad-except |
|
295
|
|
|
@workspace_cli.command('bulk-add') |
|
296
|
|
|
@click.option('-r', '--regex', help="Regular expression matching the FILE_GLOB filesystem paths " |
|
297
|
|
|
"to define named captures usable in the other parameters", required=True) |
|
298
|
|
|
@click.option('-m', '--mimetype', help="Media type of the file. If not provided, guess from filename", required=False) |
|
299
|
|
|
@click.option('-g', '--page-id', help="physical page ID of the file", required=False) |
|
300
|
|
|
@click.option('-i', '--file-id', help="ID of the file. If not provided, derive from fileGrp and filename", required=False) |
|
301
|
|
|
@click.option('-u', '--url', help="Remote URL of the file", required=False) |
|
302
|
|
|
@click.option('-l', '--local-filename', help="Local filesystem path in the workspace directory " |
|
303
|
|
|
"(copied from source file if different)", required=False) |
|
304
|
|
|
@click.option('-G', '--file-grp', help="File group USE of the file", required=True) |
|
305
|
|
|
@click.option('-n', '--dry-run', help="Don't actually do anything to the METS or filesystem, just preview", |
|
306
|
|
|
default=False, is_flag=True) |
|
307
|
|
|
@click.option('-S', '--source-path', 'src_path_option', help="File path to copy from (if different from FILE_GLOB values)", |
|
308
|
|
|
required=False) |
|
309
|
|
|
@click.option('-I', '--ignore', help="Disable checking for existing file entries (faster)", default=False, is_flag=True) |
|
310
|
|
|
@click.option('-f', '--force', help="Replace existing file entries with the same ID (no effect when --ignore is set, too)", |
|
311
|
|
|
default=False, is_flag=True) |
|
312
|
|
|
@click.option('-s', '--skip', help="Skip files not matching --regex (instead of failing)", default=False, is_flag=True) |
|
313
|
|
|
@click.argument('file_glob', nargs=-1, required=True) |
|
314
|
|
|
@pass_workspace |
|
315
|
|
|
def workspace_cli_bulk_add(ctx, regex, mimetype, page_id, file_id, url, local_filename, file_grp, dry_run, |
|
316
|
|
|
file_glob, src_path_option, ignore, force, skip): |
|
317
|
|
|
""" |
|
318
|
|
|
Add files in bulk to an OCR-D workspace. |
|
319
|
|
|
|
|
320
|
|
|
FILE_GLOB can either be a shell glob expression to match file names, |
|
321
|
|
|
or a list of expressions or '-', in which case expressions are read from STDIN. |
|
322
|
|
|
|
|
323
|
|
|
After globbing, --regex is matched against each expression resulting from FILE_GLOB, and can |
|
324
|
|
|
define named groups reusable in the --page-id, --file-id, --mimetype, --url, --source-path and |
|
325
|
|
|
--file-grp options, e.g. by referencing the group name 'grp' from the regex as '{{ grp }}'. |
|
326
|
|
|
|
|
327
|
|
|
If the FILE_GLOB expressions do not denote the file names themselves |
|
328
|
|
|
(but arbitrary strings for --regex matching), then use --source-path to set |
|
329
|
|
|
the actual file paths to use. (This could involve fixed strings or group references.) |
|
330
|
|
|
|
|
331
|
|
|
\b |
|
332
|
|
|
Examples: |
|
333
|
|
|
ocrd workspace bulk-add \\ |
|
334
|
|
|
--regex '(?P<fileGrp>[^/]+)/page_(?P<pageid>.*)\\.[^.]+' \\ |
|
335
|
|
|
--page-id 'PHYS_{{ pageid }}' \\ |
|
336
|
|
|
--file-grp "{{ fileGrp }}" \\ |
|
337
|
|
|
path/to/files/*/*.* |
|
338
|
|
|
\b |
|
339
|
|
|
echo "path/to/src/file.xml SEG/page_p0001.xml" \\ |
|
340
|
|
|
| ocrd workspace bulk-add \\ |
|
341
|
|
|
--regex '(?P<src>.*?) (?P<fileGrp>.+?)/page_(?P<pageid>.*)\\.(?P<ext>[^\\.]*)' \\ |
|
342
|
|
|
--file-id 'FILE_{{ fileGrp }}_{{ pageid }}' \\ |
|
343
|
|
|
--page-id 'PHYS_{{ pageid }}' \\ |
|
344
|
|
|
--file-grp "{{ fileGrp }}" \\ |
|
345
|
|
|
--local-filename '{{ fileGrp }}/FILE_{{ pageid }}.{{ ext }}' \\ |
|
346
|
|
|
- |
|
347
|
|
|
|
|
348
|
|
|
\b |
|
349
|
|
|
{ echo PHYS_0001 BIN FILE_0001_BIN.IMG-wolf BIN/FILE_0001_BIN.IMG-wolf.png; \\ |
|
350
|
|
|
echo PHYS_0001 BIN FILE_0001_BIN BIN/FILE_0001_BIN.xml; \\ |
|
351
|
|
|
echo PHYS_0002 BIN FILE_0002_BIN.IMG-wolf BIN/FILE_0002_BIN.IMG-wolf.png; \\ |
|
352
|
|
|
echo PHYS_0002 BIN FILE_0002_BIN BIN/FILE_0002_BIN.xml; \\ |
|
353
|
|
|
} | ocrd workspace bulk-add -r '(?P<pageid>.*) (?P<filegrp>.*) (?P<fileid>.*) (?P<local_filename>.*)' \\ |
|
354
|
|
|
-G '{{ filegrp }}' -g '{{ pageid }}' -i '{{ fileid }}' -S '{{ local_filename }}' - |
|
355
|
|
|
|
|
356
|
|
|
""" |
|
357
|
|
|
log = getLogger('ocrd.cli.workspace.bulk-add') # pylint: disable=redefined-outer-name |
|
358
|
|
|
workspace = ctx.workspace() |
|
359
|
|
|
|
|
360
|
|
|
try: |
|
361
|
|
|
pat = re.compile(regex) |
|
362
|
|
|
except Exception as e: |
|
363
|
|
|
log.error("Invalid regex: %s" % e) |
|
364
|
|
|
sys.exit(1) |
|
365
|
|
|
|
|
366
|
|
|
file_paths = [] |
|
367
|
|
|
from_stdin = file_glob == ('-',) |
|
368
|
|
|
if from_stdin: |
|
369
|
|
|
file_paths += [Path(x.strip('\n')) for x in sys.stdin.readlines()] |
|
370
|
|
|
else: |
|
371
|
|
|
for fglob in file_glob: |
|
372
|
|
|
expanded = glob(fglob) |
|
373
|
|
|
if not expanded: |
|
374
|
|
|
file_paths += [Path(fglob)] |
|
375
|
|
|
else: |
|
376
|
|
|
file_paths += [Path(x) for x in expanded] |
|
377
|
|
|
|
|
378
|
|
|
for i, file_path in enumerate(file_paths): |
|
379
|
|
|
log.info("[%4d/%d] %s" % (i + 1, len(file_paths), file_path)) |
|
380
|
|
|
|
|
381
|
|
|
# match regex |
|
382
|
|
|
m = pat.match(str(file_path)) |
|
383
|
|
|
if not m: |
|
384
|
|
|
if skip: |
|
385
|
|
|
continue |
|
386
|
|
|
log.error("File '%s' not matched by regex: '%s'" % (file_path, regex)) |
|
387
|
|
|
sys.exit(1) |
|
388
|
|
|
group_dict = m.groupdict() |
|
389
|
|
|
|
|
390
|
|
|
# set up file info |
|
391
|
|
|
file_dict = {'local_filename': local_filename, |
|
392
|
|
|
'url': url, |
|
393
|
|
|
'mimetype': mimetype, |
|
394
|
|
|
'file_id': file_id, |
|
395
|
|
|
'page_id': page_id, |
|
396
|
|
|
'file_grp': file_grp} |
|
397
|
|
|
|
|
398
|
|
|
# Flag to track whether 'local_filename' should be 'src' |
|
399
|
|
|
local_filename_is_src = False |
|
400
|
|
|
|
|
401
|
|
|
# expand templates |
|
402
|
|
|
for param_name in file_dict: |
|
403
|
|
|
if not file_dict[param_name]: |
|
404
|
|
|
if param_name == 'local_filename': |
|
405
|
|
|
local_filename_is_src = True |
|
406
|
|
|
continue |
|
407
|
|
|
elif param_name in ['mimetype', 'file_id']: |
|
408
|
|
|
# auto-filled below once the other |
|
409
|
|
|
# replacements have happened |
|
410
|
|
|
continue |
|
411
|
|
|
elif param_name == 'url': |
|
412
|
|
|
# Remote URL is not required |
|
413
|
|
|
continue |
|
414
|
|
|
raise ValueError(f"OcrdFile attribute '{param_name}' unset ({file_dict})") |
|
415
|
|
|
for group_name in group_dict: |
|
416
|
|
|
file_dict[param_name] = file_dict[param_name].replace('{{ %s }}' % group_name, group_dict[group_name]) |
|
417
|
|
|
|
|
418
|
|
|
# Where to copy from |
|
419
|
|
|
if src_path_option: |
|
420
|
|
|
src_path = src_path_option |
|
421
|
|
|
for group_name in group_dict: |
|
422
|
|
|
src_path = src_path.replace('{{ %s }}' % group_name, group_dict[group_name]) |
|
423
|
|
|
srcpath = Path(src_path) |
|
424
|
|
|
else: |
|
425
|
|
|
srcpath = file_path |
|
426
|
|
|
|
|
427
|
|
|
# derive --file-id from filename if not --file-id not explicitly set |
|
428
|
|
|
if not file_id: |
|
429
|
|
|
id_field = srcpath.stem if file_path != srcpath else file_path.stem |
|
430
|
|
|
file_dict['file_id'] = safe_filename('%s_%s' % (file_dict['file_grp'], id_field)) |
|
431
|
|
|
if not mimetype: |
|
432
|
|
|
try: |
|
433
|
|
|
file_dict['mimetype'] = EXT_TO_MIME[srcpath.suffix] |
|
434
|
|
|
except KeyError: |
|
435
|
|
|
log.error("Cannot guess MIME type from extension '%s' for '%s'. " |
|
436
|
|
|
"Set --mimetype explicitly" % (srcpath.suffix, srcpath)) |
|
437
|
|
|
|
|
438
|
|
|
# copy files if src != url |
|
439
|
|
|
if local_filename_is_src: |
|
440
|
|
|
file_dict['local_filename'] = srcpath |
|
441
|
|
|
else: |
|
442
|
|
|
destpath = Path(workspace.directory, file_dict['local_filename']) |
|
443
|
|
|
if srcpath != destpath and not destpath.exists(): |
|
444
|
|
|
log.info("cp '%s' '%s'", srcpath, destpath) |
|
445
|
|
|
if not dry_run: |
|
446
|
|
|
if not destpath.parent.is_dir(): |
|
447
|
|
|
destpath.parent.mkdir() |
|
448
|
|
|
destpath.write_bytes(srcpath.read_bytes()) |
|
449
|
|
|
|
|
450
|
|
|
# Add to workspace (or not) |
|
451
|
|
|
fileGrp = file_dict.pop('file_grp') |
|
452
|
|
|
if dry_run: |
|
453
|
|
|
log.info('workspace.add_file(%s)' % file_dict) |
|
454
|
|
|
else: |
|
455
|
|
|
workspace.add_file(fileGrp, ignore=ignore, force=force, **file_dict) # pylint: disable=redundant-keyword-arg |
|
456
|
|
|
|
|
457
|
|
|
# save changes to disk |
|
458
|
|
|
workspace.save_mets() |
|
459
|
|
|
|
|
460
|
|
|
|
|
461
|
|
|
# ---------------------------------------------------------------------- |
|
462
|
|
|
# ocrd workspace find |
|
463
|
|
|
# ---------------------------------------------------------------------- |
|
464
|
|
|
|
|
465
|
|
|
@workspace_cli.command('find') |
|
466
|
|
|
@mets_find_options |
|
467
|
|
|
@click.option('-k', '--output-field', help="Output field. Repeat for multiple fields, will be joined with tab", |
|
468
|
|
|
default=['local_filename'], |
|
469
|
|
|
show_default=True, |
|
470
|
|
|
multiple=True, |
|
471
|
|
|
type=click.Choice([ |
|
472
|
|
|
'url', |
|
473
|
|
|
'mimetype', |
|
474
|
|
|
'page_id', |
|
475
|
|
|
'pageId', |
|
476
|
|
|
'file_id', |
|
477
|
|
|
'ID', |
|
478
|
|
|
'file_grp', |
|
479
|
|
|
'fileGrp', |
|
480
|
|
|
'basename', |
|
481
|
|
|
'basename_without_extension', |
|
482
|
|
|
'local_filename', |
|
483
|
|
|
])) |
|
484
|
|
|
@click.option('--download', is_flag=True, help="Download found files to workspace and change location in METS file") |
|
485
|
|
|
@click.option('--undo-download', is_flag=True, help="Remove all downloaded files from the METS and workspace") |
|
486
|
|
|
@click.option('--keep-files', is_flag=True, help="Do not remove downloaded files from the workspace with --undo-download") |
|
487
|
|
|
@click.option('--wait', type=int, default=0, help="Wait this many seconds between download requests") |
|
488
|
|
|
@pass_workspace |
|
489
|
|
|
def workspace_find(ctx, file_grp, mimetype, page_id, file_id, output_field, |
|
490
|
|
|
include_fileGrp, exclude_fileGrp, download, undo_download, keep_files, wait): |
|
491
|
|
|
""" |
|
492
|
|
|
Find files. |
|
493
|
|
|
|
|
494
|
|
|
(If any ``FILTER`` starts with ``//``, then its remainder |
|
495
|
|
|
will be interpreted as a regular expression.) |
|
496
|
|
|
""" |
|
497
|
|
|
snake_to_camel = {"file_id": "ID", "page_id": "pageId", "file_grp": "fileGrp"} |
|
498
|
|
|
output_field = [snake_to_camel.get(x, x) for x in output_field] |
|
499
|
|
|
modified_mets = False |
|
500
|
|
|
ret = [] |
|
501
|
|
|
workspace = ctx.workspace() |
|
502
|
|
|
with pushd_popd(workspace.directory): |
|
503
|
|
|
for f in workspace.find_files( |
|
504
|
|
|
file_id=file_id, |
|
505
|
|
|
file_grp=file_grp, |
|
506
|
|
|
mimetype=mimetype, |
|
507
|
|
|
page_id=page_id, |
|
508
|
|
|
include_fileGrp=include_fileGrp, |
|
509
|
|
|
exclude_fileGrp=exclude_fileGrp, |
|
510
|
|
|
): |
|
511
|
|
|
if download and not f.local_filename: |
|
512
|
|
|
workspace.download_file(f) |
|
513
|
|
|
modified_mets = True |
|
514
|
|
|
if wait: |
|
515
|
|
|
time.sleep(wait) |
|
516
|
|
|
if undo_download and f.url and f.local_filename: |
|
517
|
|
|
modified_mets = True |
|
518
|
|
|
if not keep_files: |
|
519
|
|
|
ctx.log.debug("rm %s [cwd=%s]", f.local_filename, workspace.directory) |
|
520
|
|
|
unlink(f.local_filename) |
|
521
|
|
|
f.local_filename = None |
|
522
|
|
|
ret_entry = [f.ID if field == 'pageId' else str(getattr(f, field)) or '' for field in output_field] |
|
523
|
|
|
ret.append(ret_entry) |
|
524
|
|
|
if modified_mets: |
|
525
|
|
|
workspace.save_mets() |
|
526
|
|
|
if 'pageId' in output_field: |
|
527
|
|
|
idx = output_field.index('pageId') |
|
528
|
|
|
fileIds = list(map(lambda fields: fields[idx], ret)) |
|
529
|
|
|
pages = workspace.mets.get_physical_pages(for_fileIds=fileIds) |
|
530
|
|
|
for fields, page in zip(ret, pages): |
|
531
|
|
|
fields[idx] = page or '' |
|
532
|
|
|
for fields in ret: |
|
533
|
|
|
print('\t'.join(fields)) |
|
534
|
|
|
|
|
535
|
|
|
|
|
536
|
|
|
# ---------------------------------------------------------------------- |
|
537
|
|
|
# ocrd workspace remove |
|
538
|
|
|
# ---------------------------------------------------------------------- |
|
539
|
|
|
|
|
540
|
|
|
@workspace_cli.command('remove') |
|
541
|
|
|
@click.option('-k', '--keep-file', help="Do not delete file from file system", default=False, is_flag=True) |
|
542
|
|
|
@click.option('-f', '--force', help="Continue even if mets:file or file on file system does not exist", |
|
543
|
|
|
default=False, is_flag=True) |
|
544
|
|
|
@click.argument('ID', nargs=-1) |
|
545
|
|
|
@pass_workspace |
|
546
|
|
|
def workspace_remove_file(ctx, id, force, keep_file): # pylint: disable=redefined-builtin |
|
547
|
|
|
""" |
|
548
|
|
|
Delete files (given by their ID attribute ``ID``). |
|
549
|
|
|
|
|
550
|
|
|
(If any ``ID`` starts with ``//``, then its remainder |
|
551
|
|
|
will be interpreted as a regular expression.) |
|
552
|
|
|
""" |
|
553
|
|
|
assert not ctx.mets_server_url, \ |
|
554
|
|
|
f"remove cannot be performed with METS Server - stop server, rerun without -U {ctx.mets_server_url}" |
|
555
|
|
|
workspace = ctx.workspace() |
|
556
|
|
|
for i in id: |
|
557
|
|
|
workspace.remove_file(i, force=force, keep_file=keep_file) |
|
558
|
|
|
workspace.save_mets() |
|
559
|
|
|
|
|
560
|
|
|
|
|
561
|
|
|
# ---------------------------------------------------------------------- |
|
562
|
|
|
# ocrd workspace rename-group |
|
563
|
|
|
# ---------------------------------------------------------------------- |
|
564
|
|
|
|
|
565
|
|
|
@workspace_cli.command('rename-group') |
|
566
|
|
|
@click.argument('OLD', nargs=1) |
|
567
|
|
|
@click.argument('NEW', nargs=1) |
|
568
|
|
|
@pass_workspace |
|
569
|
|
|
def rename_group(ctx, old, new): |
|
570
|
|
|
""" |
|
571
|
|
|
Rename fileGrp (USE attribute ``NEW`` to ``OLD``). |
|
572
|
|
|
""" |
|
573
|
|
|
assert not ctx.mets_server_url, \ |
|
574
|
|
|
f"rename-group cannot be performed with METS Server - stop server, rerun without -U {ctx.mets_server_url}" |
|
575
|
|
|
workspace = ctx.workspace() |
|
576
|
|
|
workspace.rename_file_group(old, new) |
|
577
|
|
|
workspace.save_mets() |
|
578
|
|
|
|
|
579
|
|
|
|
|
580
|
|
|
# ---------------------------------------------------------------------- |
|
581
|
|
|
# ocrd workspace remove-group |
|
582
|
|
|
# ---------------------------------------------------------------------- |
|
583
|
|
|
|
|
584
|
|
|
@workspace_cli.command('remove-group') |
|
585
|
|
|
@click.option('-r', '--recursive', help="Delete any files in the group before the group itself", |
|
586
|
|
|
default=False, is_flag=True) |
|
587
|
|
|
@click.option('-f', '--force', help="Continue removing even if group or containing files not found in METS", |
|
588
|
|
|
default=False, is_flag=True) |
|
589
|
|
|
@click.option('-k', '--keep-files', help="Do not delete files from file system", default=False, is_flag=True) |
|
590
|
|
|
@click.argument('GROUP', nargs=-1) |
|
591
|
|
|
@pass_workspace |
|
592
|
|
|
def remove_group(ctx, group, recursive, force, keep_files): |
|
593
|
|
|
""" |
|
594
|
|
|
Delete fileGrps (given by their USE attribute ``GROUP``). |
|
595
|
|
|
|
|
596
|
|
|
(If any ``GROUP`` starts with ``//``, then its remainder |
|
597
|
|
|
will be interpreted as a regular expression.) |
|
598
|
|
|
""" |
|
599
|
|
|
assert not ctx.mets_server_url, \ |
|
600
|
|
|
f"remove-group cannot be performed with METS Server - stop server, rerun without -U {ctx.mets_server_url}" |
|
601
|
|
|
workspace = ctx.workspace() |
|
602
|
|
|
for g in group: |
|
603
|
|
|
workspace.remove_file_group(g, recursive=recursive, force=force, keep_files=keep_files) |
|
604
|
|
|
workspace.save_mets() |
|
605
|
|
|
|
|
606
|
|
|
|
|
607
|
|
|
# ---------------------------------------------------------------------- |
|
608
|
|
|
# ocrd workspace prune-files |
|
609
|
|
|
# ---------------------------------------------------------------------- |
|
610
|
|
|
|
|
611
|
|
|
@workspace_cli.command('prune-files') |
|
612
|
|
|
@click.option('-G', '--file-grp', help="fileGrp USE", metavar='FILTER') |
|
613
|
|
|
@click.option('-m', '--mimetype', help="Media type to look for", metavar='FILTER') |
|
614
|
|
|
@click.option('-g', '--page-id', help="Page ID", metavar='FILTER') |
|
615
|
|
|
@click.option('-i', '--file-id', help="ID", metavar='FILTER') |
|
616
|
|
|
@pass_workspace |
|
617
|
|
|
def prune_files(ctx, file_grp, mimetype, page_id, file_id): |
|
618
|
|
|
""" |
|
619
|
|
|
Removes mets:files that point to non-existing local files |
|
620
|
|
|
|
|
621
|
|
|
(If any ``FILTER`` starts with ``//``, then its remainder |
|
622
|
|
|
will be interpreted as a regular expression.) |
|
623
|
|
|
""" |
|
624
|
|
|
assert not ctx.mets_server_url, \ |
|
625
|
|
|
f"prune-files cannot be performed with METS Server - stop server, rerun without -U {ctx.mets_server_url}" |
|
626
|
|
|
workspace = ctx.workspace() |
|
627
|
|
|
with pushd_popd(workspace.directory): |
|
628
|
|
|
for f in workspace.find_files( |
|
629
|
|
|
file_id=file_id, |
|
630
|
|
|
file_grp=file_grp, |
|
631
|
|
|
mimetype=mimetype, |
|
632
|
|
|
page_id=page_id, |
|
633
|
|
|
): |
|
634
|
|
|
try: |
|
635
|
|
|
if not f.local_filename or not exists(f.local_filename): |
|
636
|
|
|
workspace.mets.remove_file(f.ID) |
|
637
|
|
|
except Exception as e: |
|
638
|
|
|
ctx.log.exception("Error removing %f: %s", f, e) |
|
639
|
|
|
raise e |
|
640
|
|
|
workspace.save_mets() |
|
641
|
|
|
|
|
642
|
|
|
|
|
643
|
|
|
# ---------------------------------------------------------------------- |
|
644
|
|
|
# ocrd workspace clean |
|
645
|
|
|
# ---------------------------------------------------------------------- |
|
646
|
|
|
|
|
647
|
|
|
@workspace_cli.command('clean') |
|
648
|
|
|
@click.option('-n', '--dry-run', help="Don't actually do anything to the filesystem, just preview", |
|
649
|
|
|
default=False, is_flag=True) |
|
650
|
|
|
@click.option('-d', '--directories', help="Remove untracked directories in addition to untracked files", |
|
651
|
|
|
default=False, is_flag=True) |
|
652
|
|
|
@click.argument('path_glob', nargs=-1, required=False) |
|
653
|
|
|
@pass_workspace |
|
654
|
|
|
def clean(ctx, dry_run, directories, path_glob): |
|
655
|
|
|
""" |
|
656
|
|
|
Removes files and directories from the workspace that are not |
|
657
|
|
|
referenced by any mets:files. |
|
658
|
|
|
|
|
659
|
|
|
PATH_GLOB can be a shell glob expression to match file names, |
|
660
|
|
|
directory names (recursively), or plain paths. All paths are |
|
661
|
|
|
resolved w.r.t. the workspace. |
|
662
|
|
|
|
|
663
|
|
|
If no PATH_GLOB are specified, then all files and directories |
|
664
|
|
|
may match. |
|
665
|
|
|
""" |
|
666
|
|
|
workspace = ctx.workspace() |
|
667
|
|
|
allowed_files = [normpath(f.local_filename) for f in workspace.find_files(local_only=True)] |
|
668
|
|
|
allowed_files.append(relpath(workspace.mets_target, start=workspace.directory)) |
|
669
|
|
|
allowed_dirs = set(dirname(path) for path in allowed_files) |
|
670
|
|
|
with pushd_popd(workspace.directory): |
|
671
|
|
|
if len(path_glob): |
|
672
|
|
|
paths = [] |
|
673
|
|
|
for expression in path_glob: |
|
674
|
|
|
if isabs(expression): |
|
675
|
|
|
expression = relpath(expression) |
|
676
|
|
|
paths += glob(expression, recursive=True) or [expression] |
|
677
|
|
|
else: |
|
678
|
|
|
paths = glob('**', recursive=True) |
|
679
|
|
|
file_paths = [path for path in paths if not isdir(path)] |
|
680
|
|
|
for path in file_paths: |
|
681
|
|
|
if normpath(path) in allowed_files: |
|
682
|
|
|
continue |
|
683
|
|
|
if dry_run: |
|
684
|
|
|
ctx.log.info('unlink(%s)' % path) |
|
685
|
|
|
else: |
|
686
|
|
|
unlink(path) |
|
687
|
|
|
if not directories: |
|
688
|
|
|
return |
|
689
|
|
|
dir_paths = [path for path in paths if isdir(path)] |
|
690
|
|
|
for path in sorted(dir_paths, key=lambda p: p.count('/'), reverse=True): |
|
691
|
|
|
if normpath(path) in allowed_dirs: |
|
692
|
|
|
continue |
|
693
|
|
|
if dry_run: |
|
694
|
|
|
ctx.log.info('rmdir(%s)' % path) |
|
695
|
|
|
else: |
|
696
|
|
|
rmdir(path) |
|
697
|
|
|
|
|
698
|
|
|
|
|
699
|
|
|
# ---------------------------------------------------------------------- |
|
700
|
|
|
# ocrd workspace list-group |
|
701
|
|
|
# ---------------------------------------------------------------------- |
|
702
|
|
|
|
|
703
|
|
|
@workspace_cli.command('list-group') |
|
704
|
|
|
@pass_workspace |
|
705
|
|
|
def list_groups(ctx): |
|
706
|
|
|
""" |
|
707
|
|
|
List fileGrp USE attributes |
|
708
|
|
|
""" |
|
709
|
|
|
workspace = ctx.workspace() |
|
710
|
|
|
print("\n".join(workspace.mets.file_groups)) |
|
711
|
|
|
|
|
712
|
|
|
|
|
713
|
|
|
# ---------------------------------------------------------------------- |
|
714
|
|
|
# ocrd workspace list-page |
|
715
|
|
|
# ---------------------------------------------------------------------- |
|
716
|
|
|
|
|
717
|
|
|
@workspace_cli.command('list-page') |
|
718
|
|
|
@click.option('-k', '--output-field', help="Output field. Repeat for multiple fields, will be joined with tab", |
|
719
|
|
|
default=['ID'], |
|
720
|
|
|
show_default=True, |
|
721
|
|
|
multiple=True, |
|
722
|
|
|
type=click.Choice(METS_PAGE_DIV_ATTRIBUTE.names())) |
|
723
|
|
|
@click.option('-f', '--output-format', help="Output format", type=click.Choice(['one-per-line', 'comma-separated', 'json']), |
|
724
|
|
|
default='one-per-line') |
|
725
|
|
|
@click.option('-D', '--chunk-number', help="Partition the return value into n roughly equally sized chunks", |
|
726
|
|
|
default=1, type=int) |
|
727
|
|
|
@click.option('-C', '--chunk-index', help="Output the nth chunk of results, -1 for all of them.", default=None, type=int) |
|
728
|
|
|
@click.option('-r', '--page-id-range', help="Restrict the pages to those matching the provided range, " |
|
729
|
|
|
"based on the @ID attribute. Separate start/end with ..") |
|
730
|
|
|
@click.option('-R', '--numeric-range', help="Restrict the pages to those in the range, in numerical document order. " |
|
731
|
|
|
"Separate start/end with ..") |
|
732
|
|
|
@pass_workspace |
|
733
|
|
|
def list_pages(ctx, output_field, output_format, chunk_number, chunk_index, page_id_range, numeric_range): |
|
734
|
|
|
""" |
|
735
|
|
|
List physical page IDs |
|
736
|
|
|
|
|
737
|
|
|
(If any ``FILTER`` starts with ``//``, then its remainder |
|
738
|
|
|
will be interpreted as a regular expression.) |
|
739
|
|
|
""" |
|
740
|
|
|
workspace = ctx.workspace() |
|
741
|
|
|
ret = [] |
|
742
|
|
|
if page_id_range or list(output_field) != ['ID']: |
|
743
|
|
|
for i, page_div in enumerate(workspace.mets.get_physical_pages(for_pageIds=page_id_range, return_divs=True)): |
|
744
|
|
|
ret.append([]) |
|
745
|
|
|
for k in output_field: |
|
746
|
|
|
ret[i].append(page_div.get(k, 'None')) |
|
747
|
|
|
else: |
|
748
|
|
|
for page_id in workspace.mets.physical_pages: |
|
749
|
|
|
ret.append([page_id]) |
|
750
|
|
|
|
|
751
|
|
|
if numeric_range: |
|
752
|
|
|
start, end = map(int, numeric_range.split('..')) |
|
753
|
|
|
ret = ret[start-1:end] |
|
754
|
|
|
|
|
755
|
|
|
chunks = partition_list(ret, chunk_number, chunk_index) |
|
756
|
|
|
lines = [] |
|
757
|
|
|
if output_format == 'one-per-line': |
|
758
|
|
|
for chunk in chunks: |
|
759
|
|
|
line_strs = [] |
|
760
|
|
|
for entry in chunk: |
|
761
|
|
|
line_strs.append("\t".join(entry)) |
|
762
|
|
|
lines.append('\n'.join(line_strs)) |
|
763
|
|
|
elif output_format == 'comma-separated': |
|
764
|
|
|
for chunk in chunks: |
|
765
|
|
|
line_strs = [] |
|
766
|
|
|
for entry in chunk: |
|
767
|
|
|
line_strs.append("\t".join(entry)) |
|
768
|
|
|
lines.append(','.join(line_strs)) |
|
769
|
|
|
elif output_format == 'json': |
|
770
|
|
|
lines.append(dumps(chunks)) |
|
771
|
|
|
print('\n'.join(lines)) |
|
772
|
|
|
|
|
773
|
|
|
|
|
774
|
|
|
# ---------------------------------------------------------------------- |
|
775
|
|
|
# ocrd workspace get-id |
|
776
|
|
|
# ---------------------------------------------------------------------- |
|
777
|
|
|
|
|
778
|
|
|
@workspace_cli.command('get-id') |
|
779
|
|
|
@pass_workspace |
|
780
|
|
|
def get_id(ctx): |
|
781
|
|
|
""" |
|
782
|
|
|
Get METS id if any |
|
783
|
|
|
""" |
|
784
|
|
|
workspace = ctx.workspace() |
|
785
|
|
|
ID = workspace.mets.unique_identifier |
|
786
|
|
|
if ID: |
|
787
|
|
|
print(ID) |
|
788
|
|
|
|
|
789
|
|
|
|
|
790
|
|
|
# ---------------------------------------------------------------------- |
|
791
|
|
|
# ocrd workspace set-id |
|
792
|
|
|
# ---------------------------------------------------------------------- |
|
793
|
|
|
|
|
794
|
|
|
@workspace_cli.command('set-id') |
|
795
|
|
|
@click.argument('ID') |
|
796
|
|
|
@pass_workspace |
|
797
|
|
|
def set_id(ctx, id): # pylint: disable=redefined-builtin |
|
798
|
|
|
""" |
|
799
|
|
|
Set METS ID. |
|
800
|
|
|
|
|
801
|
|
|
If one of the supported identifier mechanisms is used, will set this identifier. |
|
802
|
|
|
|
|
803
|
|
|
Otherwise will create a new <mods:identifier type="purl">{{ ID }}</mods:identifier>. |
|
804
|
|
|
""" |
|
805
|
|
|
workspace = ctx.workspace() |
|
806
|
|
|
workspace.mets.unique_identifier = id |
|
807
|
|
|
workspace.save_mets() |
|
808
|
|
|
|
|
809
|
|
|
|
|
810
|
|
|
@workspace_cli.command('update-page') |
|
811
|
|
|
@click.option('--set', 'attr_value_pairs', help="set mets:div ATTR to VALUE", metavar="ATTR VALUE", |
|
812
|
|
|
type=(click.Choice(METS_PAGE_DIV_ATTRIBUTE.names()), str), nargs=2, multiple=True) |
|
813
|
|
|
@click.option('--order', help="[DEPRECATED - use --set ATTR VALUE", metavar='ORDER') |
|
814
|
|
|
@click.option('--orderlabel', help="DEPRECATED - use --set ATTR VALUE", metavar='ORDERLABEL') |
|
815
|
|
|
@click.option('--contentids', help="DEPRECATED - use --set ATTR VALUE", metavar='ORDERLABEL') |
|
816
|
|
|
@click.argument('PAGE_ID') |
|
817
|
|
|
@pass_workspace |
|
818
|
|
|
def update_page(ctx, attr_value_pairs, order, orderlabel, contentids, page_id): |
|
819
|
|
|
""" |
|
820
|
|
|
Update the @ID, @ORDER, @ORDERLABEL, @LABEL or @CONTENTIDS attributes of the mets:div with @ID=PAGE_ID |
|
821
|
|
|
""" |
|
822
|
|
|
update_kwargs = dict(attr_value_pairs) |
|
823
|
|
|
if order: |
|
824
|
|
|
update_kwargs['ORDER'] = order |
|
825
|
|
|
if orderlabel: |
|
826
|
|
|
update_kwargs['ORDERLABEL'] = orderlabel |
|
827
|
|
|
if contentids: |
|
828
|
|
|
update_kwargs['CONTENTIDS'] = contentids |
|
829
|
|
|
try: |
|
830
|
|
|
assert not ctx.mets_server_url, \ |
|
831
|
|
|
f"update-page cannot be performed with METS Server - stop server, rerun without -U {ctx.mets_server_url}" |
|
832
|
|
|
workspace = ctx.workspace() |
|
833
|
|
|
workspace.mets.update_physical_page_attributes(page_id, **update_kwargs) |
|
834
|
|
|
workspace.save_mets() |
|
835
|
|
|
except Exception as err: |
|
836
|
|
|
print(f"Error: {err}") |
|
837
|
|
|
sys.exit(1) |
|
838
|
|
|
|
|
839
|
|
|
|
|
840
|
|
|
# ---------------------------------------------------------------------- |
|
841
|
|
|
# ocrd workspace merge |
|
842
|
|
|
# ---------------------------------------------------------------------- |
|
843
|
|
|
|
|
844
|
|
|
def _handle_json_option(ctx, param, value): |
|
845
|
|
|
return parse_json_string_or_file(value) if value else None |
|
846
|
|
|
|
|
847
|
|
|
|
|
848
|
|
|
@workspace_cli.command('merge') |
|
849
|
|
|
@click.argument('METS_PATH') |
|
850
|
|
|
@click.option('--overwrite/--no-overwrite', is_flag=True, default=False, |
|
851
|
|
|
help="Overwrite on-disk file in case of file name conflicts with data from METS_PATH") |
|
852
|
|
|
@click.option('--force/--no-force', is_flag=True, default=False, |
|
853
|
|
|
help="Overwrite mets:file from --mets with mets:file from METS_PATH if IDs clash") |
|
854
|
|
|
@click.option('--copy-files/--no-copy-files', is_flag=True, help="Copy files as well", default=True, show_default=True) |
|
855
|
|
|
@click.option('--fileGrp-mapping', help="JSON object mapping src to dest fileGrp", callback=_handle_json_option) |
|
856
|
|
|
@click.option('--fileId-mapping', help="JSON object mapping src to dest file ID", callback=_handle_json_option) |
|
857
|
|
|
@click.option('--pageId-mapping', help="JSON object mapping src to dest page ID", callback=_handle_json_option) |
|
858
|
|
|
@mets_find_options |
|
859
|
|
|
@pass_workspace |
|
860
|
|
|
def merge(ctx, overwrite, force, copy_files, filegrp_mapping, fileid_mapping, pageid_mapping, |
|
861
|
|
|
file_grp, file_id, page_id, mimetype, include_fileGrp, exclude_fileGrp, mets_path): # pylint: disable=redefined-builtin |
|
862
|
|
|
""" |
|
863
|
|
|
Merges this workspace with the workspace that contains ``METS_PATH`` |
|
864
|
|
|
|
|
865
|
|
|
Pass a JSON string or file to ``--fileGrp-mapping``, ``--fileId-mapping`` or ``--pageId-mapping`` |
|
866
|
|
|
in order to rename all fileGrp, file ID or page ID values, respectively. |
|
867
|
|
|
|
|
868
|
|
|
The ``--file-id``, ``--page-id``, ``--mimetype`` and ``--file-grp`` options have |
|
869
|
|
|
the same semantics as in ``ocrd workspace find``, see ``ocrd workspace find --help`` |
|
870
|
|
|
for an explanation. |
|
871
|
|
|
""" |
|
872
|
|
|
mets_path = Path(mets_path) |
|
873
|
|
|
if filegrp_mapping: |
|
874
|
|
|
filegrp_mapping = loads(filegrp_mapping) |
|
875
|
|
|
assert not ctx.mets_server_url, \ |
|
876
|
|
|
f"merge cannot be performed with METS Server - stop server, rerun without -U {ctx.mets_server_url}" |
|
877
|
|
|
workspace = ctx.workspace() |
|
878
|
|
|
other_workspace = Workspace(ctx.resolver, directory=str(mets_path.parent), mets_basename=str(mets_path.name)) |
|
879
|
|
|
workspace.merge( |
|
880
|
|
|
other_workspace, |
|
881
|
|
|
force=force, |
|
882
|
|
|
overwrite=overwrite, |
|
883
|
|
|
copy_files=copy_files, |
|
884
|
|
|
fileGrp_mapping=filegrp_mapping, |
|
885
|
|
|
fileId_mapping=fileid_mapping, |
|
886
|
|
|
pageId_mapping=pageid_mapping, |
|
887
|
|
|
file_grp=file_grp, |
|
888
|
|
|
file_id=file_id, |
|
889
|
|
|
page_id=page_id, |
|
890
|
|
|
mimetype=mimetype, |
|
891
|
|
|
include_fileGrp=include_fileGrp, |
|
892
|
|
|
exclude_fileGrp=exclude_fileGrp, |
|
893
|
|
|
) |
|
894
|
|
|
workspace.save_mets() |
|
895
|
|
|
|
|
896
|
|
|
|
|
897
|
|
|
# ---------------------------------------------------------------------- |
|
898
|
|
|
# ocrd workspace backup |
|
899
|
|
|
# ---------------------------------------------------------------------- |
|
900
|
|
|
|
|
901
|
|
|
@workspace_cli.group('backup') |
|
902
|
|
|
@pass_workspace |
|
903
|
|
|
def workspace_backup_cli(ctx): # pylint: disable=unused-argument |
|
904
|
|
|
""" |
|
905
|
|
|
Backing and restoring workspaces - dev edition |
|
906
|
|
|
""" |
|
907
|
|
|
assert not ctx.mets_server_url, "Workspace backups currently not interoperable with METS Server" |
|
908
|
|
|
|
|
909
|
|
|
|
|
910
|
|
|
@workspace_backup_cli.command('add') |
|
911
|
|
|
@pass_workspace |
|
912
|
|
|
def workspace_backup_add(ctx): |
|
913
|
|
|
""" |
|
914
|
|
|
Create a new backup |
|
915
|
|
|
""" |
|
916
|
|
|
backup_manager = ctx.backup_manager() |
|
917
|
|
|
backup_manager.add() |
|
918
|
|
|
|
|
919
|
|
|
|
|
920
|
|
|
@workspace_backup_cli.command('list') |
|
921
|
|
|
@pass_workspace |
|
922
|
|
|
def workspace_backup_list(ctx): |
|
923
|
|
|
""" |
|
924
|
|
|
List backups |
|
925
|
|
|
""" |
|
926
|
|
|
backup_manager = ctx.backup_manager() |
|
927
|
|
|
for b in backup_manager.list(): |
|
928
|
|
|
print(b) |
|
929
|
|
|
|
|
930
|
|
|
|
|
931
|
|
|
@workspace_backup_cli.command('restore') |
|
932
|
|
|
@click.option('-f', '--choose-first', help="Restore first matching version if more than one", is_flag=True) |
|
933
|
|
|
@click.argument('bak') # type=click.Path(dir_okay=False, readable=True, resolve_path=True)) |
|
934
|
|
|
@pass_workspace |
|
935
|
|
|
def workspace_backup_restore(ctx, choose_first, bak): |
|
936
|
|
|
""" |
|
937
|
|
|
Restore backup BAK |
|
938
|
|
|
""" |
|
939
|
|
|
backup_manager = ctx.backup_manager() |
|
940
|
|
|
backup_manager.restore(bak, choose_first) |
|
941
|
|
|
|
|
942
|
|
|
|
|
943
|
|
|
@workspace_backup_cli.command('undo') |
|
944
|
|
|
@pass_workspace |
|
945
|
|
|
def workspace_backup_undo(ctx): |
|
946
|
|
|
""" |
|
947
|
|
|
Restore the last backup |
|
948
|
|
|
""" |
|
949
|
|
|
backup_manager = ctx.backup_manager() |
|
950
|
|
|
backup_manager.undo() |
|
951
|
|
|
|
|
952
|
|
|
|
|
953
|
|
|
# ---------------------------------------------------------------------- |
|
954
|
|
|
# ocrd workspace server |
|
955
|
|
|
# ---------------------------------------------------------------------- |
|
956
|
|
|
|
|
957
|
|
|
@workspace_cli.group('server') |
|
958
|
|
|
@pass_workspace |
|
959
|
|
|
def workspace_serve_cli(ctx): # pylint: disable=unused-argument |
|
960
|
|
|
"""Control a METS server for this workspace""" |
|
961
|
|
|
assert ctx.mets_server_url, "For METS server commands, you must provide '-U/--mets-server-url'" |
|
962
|
|
|
|
|
963
|
|
|
|
|
964
|
|
|
@workspace_serve_cli.command('stop') |
|
965
|
|
|
@pass_workspace |
|
966
|
|
|
def workspace_serve_stop(ctx): # pylint: disable=unused-argument |
|
967
|
|
|
"""Stop the METS server (saving changes to disk)""" |
|
968
|
|
|
workspace = ctx.workspace() |
|
969
|
|
|
workspace.mets.stop() |
|
970
|
|
|
|
|
971
|
|
|
|
|
972
|
|
|
@workspace_serve_cli.command('reload') |
|
973
|
|
|
@pass_workspace |
|
974
|
|
|
def workspace_serve_reload(ctx): # pylint: disable=unused-argument |
|
975
|
|
|
"""Reload the METS server from disk""" |
|
976
|
|
|
workspace = ctx.workspace() |
|
977
|
|
|
workspace.mets.reload() |
|
978
|
|
|
|
|
979
|
|
|
|
|
980
|
|
|
@workspace_serve_cli.command('save') |
|
981
|
|
|
@pass_workspace |
|
982
|
|
|
def workspace_serve_save(ctx): # pylint: disable=unused-argument |
|
983
|
|
|
"""Save the METS changes to disk""" |
|
984
|
|
|
workspace = ctx.workspace() |
|
985
|
|
|
workspace.mets.save() |
|
986
|
|
|
|
|
987
|
|
|
|
|
988
|
|
|
@workspace_serve_cli.command('start') |
|
989
|
|
|
@pass_workspace |
|
990
|
|
|
def workspace_serve_start(ctx): # pylint: disable=unused-argument |
|
991
|
|
|
""" |
|
992
|
|
|
Start a METS server |
|
993
|
|
|
|
|
994
|
|
|
(For TCP backend, pass a network interface to bind to as the '-U/--mets-server-url' parameter.) |
|
995
|
|
|
""" |
|
996
|
|
|
OcrdMetsServer( |
|
997
|
|
|
workspace=Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename), |
|
998
|
|
|
url=ctx.mets_server_url, |
|
999
|
|
|
).startup() |
|
1000
|
|
|
|
OCR-D /
core
