ocrd.cli.workspace.set_id() - Code Metrics - Inspection of "ocrd workspace add: check file exists with -C/--ch..." - OCR-D/core - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Pull Request — master (#498)

by Konstantin

created 2020-06-08 09:46 UTC

ocrd.cli.workspace.set_id() A

↳ Parent: ocrd.cli.workspace

Complexity

Conditions

Size

Total Lines	14
Code Lines	7

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	1
eloc	7
nop	2
dl	0
loc	14
rs	10
c	0
b	0
f	0

import os
from os.path import relpath, exists, join
import sys

import click

from ocrd import Resolver, Workspace, WorkspaceValidator, WorkspaceBackupManager
from ocrd_utils import getLogger, pushd_popd

log = getLogger('ocrd.cli.workspace')

class WorkspaceCtx():

    def __init__(self, directory, mets_basename, automatic_backup):
        self.directory = directory
        self.resolver = Resolver()
        self.mets_basename = mets_basename
        self.automatic_backup = automatic_backup

pass_workspace = click.make_pass_decorator(WorkspaceCtx)

# ----------------------------------------------------------------------
# ocrd workspace
# ----------------------------------------------------------------------

@click.group("workspace")
@click.option('-d', '--directory', envvar='WORKSPACE_DIR', default='.', type=click.Path(file_okay=False), metavar='WORKSPACE_DIR', help='Changes the workspace folder location.', show_default=True)
@click.option('-M', '--mets-basename', default="mets.xml", help='The basename of the METS file.', show_default=True)
@click.option('--backup', default=False, help="Backup mets.xml whenever it is saved.", is_flag=True)
@click.pass_context
def workspace_cli(ctx, directory, mets_basename, backup):
    """
    Working with workspace
    """
    ctx.obj = WorkspaceCtx(os.path.abspath(directory), mets_basename, automatic_backup=backup)

# ----------------------------------------------------------------------
# ocrd workspace validate
# ----------------------------------------------------------------------

@workspace_cli.command('validate', help='''

    Validate a workspace

''')
@pass_workspace
@click.option('-a', '--download', is_flag=True, help="Download all files")
@click.option('-s', '--skip', help="Tests to skip", default=[], multiple=True, type=click.Choice(['imagefilename', 'dimension', 'mets_unique_identifier', 'mets_file_group_names', 'mets_files', 'pixel_density', 'page', 'url']))
@click.option('--page-textequiv-consistency', '--page-strictness', help="How strict to check PAGE multi-level textequiv consistency", type=click.Choice(['strict', 'lax', 'fix', 'off']), default='strict')
@click.option('--page-coordinate-consistency', help="How fierce to check PAGE multi-level coordinate consistency", type=click.Choice(['poly', 'baseline', 'both', 'off']), default='poly')
@click.argument('mets_url', nargs=-1)
def validate_workspace(ctx, mets_url, download, skip, page_textequiv_consistency, page_coordinate_consistency):
    if not mets_url:
        mets_url = 'mets.xml'
    else:
        mets_url = mets_url[0]
    report = WorkspaceValidator.validate(
        ctx.resolver,
        mets_url,
        src_dir=ctx.directory,
        skip=skip,
        download=download,
        page_strictness=page_textequiv_consistency,
        page_coordinate_consistency=page_coordinate_consistency
    )
    print(report.to_xml())
    if not report.is_valid:
        sys.exit(128)

# ----------------------------------------------------------------------
# ocrd workspace clone
# ----------------------------------------------------------------------

@workspace_cli.command('clone')
@click.option('-f', '--clobber-mets', help="Overwrite existing METS file", default=False, is_flag=True)
@click.option('-a', '--download', is_flag=True, help="Download all files and change location in METS file after cloning")
@click.argument('mets_url')
@click.argument('workspace_dir', default=None, required=False)
@pass_workspace
def workspace_clone(ctx, clobber_mets, download, mets_url, workspace_dir):
    """
    Create a workspace from a METS_URL and return the directory

    METS_URL can be a URL, an absolute path or a path relative to $PWD.

    If WORKSPACE_DIR is not provided, use the current working directory
    """
    if not workspace_dir:
        workspace_dir = '.'
    workspace = ctx.resolver.workspace_from_url(
        mets_url,
        dst_dir=os.path.abspath(workspace_dir),
        mets_basename=ctx.mets_basename,
        clobber_mets=clobber_mets,
        download=download,
    )
    workspace.save_mets()
    print(workspace.directory)

# ----------------------------------------------------------------------
# ocrd workspace init
# ----------------------------------------------------------------------

@workspace_cli.command('init')
@click.option('-f', '--clobber-mets', help="Clobber mets.xml if it exists", is_flag=True, default=False)
@click.argument('directory', required=False)
@pass_workspace
def workspace_create(ctx, clobber_mets, directory):
    """
    Create a workspace with an empty METS file in DIRECTORY.

    Use '.' for $PWD"
    """
    if not directory:
        directory = '.'
    workspace = ctx.resolver.workspace_from_nothing(
        directory=os.path.abspath(directory),
        mets_basename=ctx.mets_basename,
        clobber_mets=clobber_mets
    )
    workspace.save_mets()
    print(workspace.directory)

# ----------------------------------------------------------------------
# ocrd workspace add
# ----------------------------------------------------------------------

@workspace_cli.command('add')
@click.option('-G', '--file-grp', help="fileGrp USE", required=True)
@click.option('-i', '--file-id', help="ID for the file", required=True)
@click.option('-m', '--mimetype', help="Media type of the file", required=True)
@click.option('-g', '--page-id', help="ID of the physical page")
@click.option('-C', '--check-file-exists', help="Whether to ensure FNAME exists", is_flag=True, default=False)
@click.option('--force', help="If file with ID already exists, replace it", default=False, is_flag=True)
@click.argument('fname', required=True)
@pass_workspace
def workspace_add_file(ctx, file_grp, file_id, mimetype, page_id, check_file_exists, force, fname):
    """
    Add a file or http(s) URL FNAME to METS in a workspace.
    If FNAME is not an http(s) URL and is not a workspace-local existing file, try to copy to workspace.
    """
    workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup)

    kwargs = {'fileGrp': file_grp, 'ID': file_id, 'mimetype': mimetype, 'pageId': page_id, 'force': force}
    log = getLogger('ocrd.cli.workspace.add')
    if not (fname.startswith('http://') or fname.startswith('https://')):
        if not fname.startswith(ctx.directory):
            if exists(join(ctx.directory, fname)):
                fname = join(ctx.directory, fname)
            else:
                log.debug("File '%s' is not in workspace, copying", fname)
                try:
                    fname = ctx.resolver.download_to_directory(ctx.directory, fname, subdir=file_grp)
                except FileNotFoundError as e:
                    if check_file_exists:
                        log.error("File '%s' does not exist, halt execution!" % fname)
                        sys.exit(1)
        if check_file_exists and not exists(fname):
            log.error("File '%s' does not exist, halt execution!" % fname)
            sys.exit(1)
        if fname.startswith(ctx.directory):
            fname = relpath(fname, ctx.directory)
        kwargs['local_filename'] = fname

    kwargs['url'] = fname
    workspace.mets.add_file(**kwargs)
    workspace.save_mets()

# ----------------------------------------------------------------------
# ocrd workspace find
# ----------------------------------------------------------------------

@workspace_cli.command('find')
@click.option('-G', '--file-grp', help="fileGrp USE")
@click.option('-m', '--mimetype', help="Media type to look for")
@click.option('-g', '--page-id', help="Page ID")
@click.option('-i', '--file-id', help="ID")
# pylint: disable=bad-continuation
@click.option('-k', '--output-field', help="Output field. Repeat for multiple fields, will be joined with tab",
        default=['url'],
        multiple=True,
        type=click.Choice([
            'url',
            'mimetype',
            'pageId',
            'ID',
            'fileGrp',
            'basename',
            'basename_without_extension',
            'local_filename',
        ]))
@click.option('--download', is_flag=True, help="Download found files to workspace and change location in METS file ")
@pass_workspace
def workspace_find(ctx, file_grp, mimetype, page_id, file_id, output_field, download):
    """
    Find files.
    """
    modified_mets = False
    ret = list()
    workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename)
    for f in workspace.mets.find_files(
            ID=file_id,
            fileGrp=file_grp,
            mimetype=mimetype,
            pageId=page_id,
        ):
        if download and not f.local_filename:
            workspace.download_file(f)
            modified_mets = True
        ret.append([f.ID if field == 'pageId' else getattr(f, field) or ''
                    for field in output_field])
    if modified_mets:
        workspace.save_mets()
    if 'pageId' in output_field:
        idx = output_field.index('pageId')
        fileIds = list(map(lambda fields: fields[idx], ret))

        pages = workspace.mets.get_physical_pages(for_fileIds=fileIds)
        for fields, page in zip(ret, pages):
            fields[idx] = page or ''
    for fields in ret:
        print('\t'.join(fields))

# ----------------------------------------------------------------------
# ocrd workspace remove
# ----------------------------------------------------------------------

@workspace_cli.command('remove')
@click.option('-k', '--keep-file', help="Do not delete file from file system", default=False, is_flag=True)
@click.option('-f', '--force', help="Continue even if mets:file or file on file system does not exist", default=False, is_flag=True)
@click.argument('ID', nargs=-1)
@pass_workspace
def workspace_remove_file(ctx, id, force, keep_file):  # pylint: disable=redefined-builtin
    """
    Delete file by ID from mets.xml
    """
    workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup)
    for i in id:
        workspace.remove_file(i, force=force, keep_file=keep_file)
    workspace.save_mets()


# ----------------------------------------------------------------------
# ocrd workspace remove-group
# ----------------------------------------------------------------------

@workspace_cli.command('remove-group', help="""

    Delete a file group

""")
@click.option('-r', '--recursive', help="Delete any files in the group before the group itself", default=False, is_flag=True)
@click.option('-f', '--force', help="Continue removing even if group or containing files not found in METS", default=False, is_flag=True)
@click.option('-k', '--keep-files', help="Do not delete files from file system", default=False, is_flag=True)
@click.argument('GROUP', nargs=-1)
@pass_workspace
def remove_group(ctx, group, recursive, force, keep_files):
    workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename)
    for g in group:
        workspace.remove_file_group(g, recursive=recursive, force=force, keep_files=keep_files)
    workspace.save_mets()

# ----------------------------------------------------------------------
# ocrd workspace prune-files
# ----------------------------------------------------------------------

@workspace_cli.command('prune-files', help="""

    Removes mets:files that point to non-existing local files

""")
@pass_workspace
def prune_files(ctx):
    workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename)
    with pushd_popd(workspace.directory):
        for f in workspace.mets.find_files():
            try:
                if not f.local_filename or not exists(f.local_filename):
                    workspace.mets.remove_file(f.ID)
            except Exception as e:
                log.exception("Error removing %f: %s", f, e)
                raise(e)
        workspace.save_mets()

# ----------------------------------------------------------------------
# ocrd workspace list-group
# ----------------------------------------------------------------------

@workspace_cli.command('list-group', help="""

    List fileGrp USE attributes

""")
@pass_workspace
def list_groups(ctx):
    workspace = Workspace(ctx.resolver, directory=ctx.directory)
    print("\n".join(workspace.mets.file_groups))

# ----------------------------------------------------------------------
# ocrd workspace list-pages
# ----------------------------------------------------------------------

@workspace_cli.command('list-page', help="""

    List page IDs

""")
@pass_workspace
def list_pages(ctx):
    workspace = Workspace(ctx.resolver, directory=ctx.directory)
    print("\n".join(workspace.mets.physical_pages))

# ----------------------------------------------------------------------
# ocrd workspace get-id
# ----------------------------------------------------------------------

@workspace_cli.command('get-id', help="""

    Get METS id if any

""")
@pass_workspace
def get_id(ctx):
    workspace = Workspace(ctx.resolver, directory=ctx.directory)
    ID = workspace.mets.unique_identifier
    if ID:
        print(ID)

# ----------------------------------------------------------------------
# ocrd workspace set-id
# ----------------------------------------------------------------------

@workspace_cli.command('set-id', help="""

    Set METS ID.

    If one of the supported identifier mechanisms is used, will set this identifier.

    Otherwise will create a new <mods:identifier type="purl">{{ ID }}</mods:identifier>.
""")
@click.argument('ID')
@pass_workspace
def set_id(ctx, id):   # pylint: disable=redefined-builtin
    workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup)
    workspace.mets.unique_identifier = id
    workspace.save_mets()

# ----------------------------------------------------------------------
# ocrd workspace backup
# ----------------------------------------------------------------------

@workspace_cli.group('backup')
@click.pass_context
def workspace_backup_cli(ctx): # pylint: disable=unused-argument
    """
    Backing and restoring workspaces - dev edition
    """

@workspace_backup_cli.command('add')
@pass_workspace
def workspace_backup_add(ctx):
    """
    Create a new backup
    """
    backup_manager = WorkspaceBackupManager(Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup))
    backup_manager.add()

@workspace_backup_cli.command('list')
@pass_workspace
def workspace_backup_list(ctx):
    """
    List backups
    """
    backup_manager = WorkspaceBackupManager(Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup))
    for b in backup_manager.list():
        print(b)

@workspace_backup_cli.command('restore')
@click.option('-f', '--choose-first', help="Restore first matching version if more than one", is_flag=True)
@click.argument('bak') #, type=click.Path(dir_okay=False, readable=True, resolve_path=True))
@pass_workspace
def workspace_backup_restore(ctx, choose_first, bak):
    """
    Restore backup BAK
    """
    backup_manager = WorkspaceBackupManager(Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup))
    backup_manager.restore(bak, choose_first)

@workspace_backup_cli.command('undo')
@pass_workspace
def workspace_backup_undo(ctx):
    """
    Restore the last backup
    """
    backup_manager = WorkspaceBackupManager(Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup))
    backup_manager.undo()


1			import os
2			from os.path import relpath, exists, join
3			import sys
4
5			import click
6
7			from ocrd import Resolver, Workspace, WorkspaceValidator, WorkspaceBackupManager
8			from ocrd_utils import getLogger, pushd_popd
9
10			log = getLogger('ocrd.cli.workspace')
11
12			class WorkspaceCtx():
13
14			def __init__(self, directory, mets_basename, automatic_backup):
15			self.directory = directory
16			self.resolver = Resolver()
17			self.mets_basename = mets_basename
18			self.automatic_backup = automatic_backup
19
20			pass_workspace = click.make_pass_decorator(WorkspaceCtx)
21
22			# ----------------------------------------------------------------------
23			# ocrd workspace
24			# ----------------------------------------------------------------------
25
26			@click.group("workspace")
27			@click.option('-d', '--directory', envvar='WORKSPACE_DIR', default='.', type=click.Path(file_okay=False), metavar='WORKSPACE_DIR', help='Changes the workspace folder location.', show_default=True)
28			@click.option('-M', '--mets-basename', default="mets.xml", help='The basename of the METS file.', show_default=True)
29			@click.option('--backup', default=False, help="Backup mets.xml whenever it is saved.", is_flag=True)
30			@click.pass_context
31			def workspace_cli(ctx, directory, mets_basename, backup):
32			"""
33			Working with workspace
34			"""
35			ctx.obj = WorkspaceCtx(os.path.abspath(directory), mets_basename, automatic_backup=backup)
36
37			# ----------------------------------------------------------------------
38			# ocrd workspace validate
39			# ----------------------------------------------------------------------
40
41			@workspace_cli.command('validate', help='''
42
43			Validate a workspace
44
45			''')
46			@pass_workspace
47			@click.option('-a', '--download', is_flag=True, help="Download all files")
48			@click.option('-s', '--skip', help="Tests to skip", default=[], multiple=True, type=click.Choice(['imagefilename', 'dimension', 'mets_unique_identifier', 'mets_file_group_names', 'mets_files', 'pixel_density', 'page', 'url']))
49			@click.option('--page-textequiv-consistency', '--page-strictness', help="How strict to check PAGE multi-level textequiv consistency", type=click.Choice(['strict', 'lax', 'fix', 'off']), default='strict')
50			@click.option('--page-coordinate-consistency', help="How fierce to check PAGE multi-level coordinate consistency", type=click.Choice(['poly', 'baseline', 'both', 'off']), default='poly')
51			@click.argument('mets_url', nargs=-1)
52			def validate_workspace(ctx, mets_url, download, skip, page_textequiv_consistency, page_coordinate_consistency):
53			if not mets_url:
54			mets_url = 'mets.xml'
55			else:
56			mets_url = mets_url[0]
57			report = WorkspaceValidator.validate(
58			ctx.resolver,
59			mets_url,
60			src_dir=ctx.directory,
61			skip=skip,
62			download=download,
63			page_strictness=page_textequiv_consistency,
64			page_coordinate_consistency=page_coordinate_consistency
65			)
66			print(report.to_xml())
67			if not report.is_valid:
68			sys.exit(128)
69
70			# ----------------------------------------------------------------------
71			# ocrd workspace clone
72			# ----------------------------------------------------------------------
73
74			@workspace_cli.command('clone')
75			@click.option('-f', '--clobber-mets', help="Overwrite existing METS file", default=False, is_flag=True)
76			@click.option('-a', '--download', is_flag=True, help="Download all files and change location in METS file after cloning")
77			@click.argument('mets_url')
78			@click.argument('workspace_dir', default=None, required=False)
79			@pass_workspace
80			def workspace_clone(ctx, clobber_mets, download, mets_url, workspace_dir):
81			"""
82			Create a workspace from a METS_URL and return the directory
83
84			METS_URL can be a URL, an absolute path or a path relative to $PWD.
85
86			If WORKSPACE_DIR is not provided, use the current working directory
87			"""
88			if not workspace_dir:
89			workspace_dir = '.'
90			workspace = ctx.resolver.workspace_from_url(
91			mets_url,
92			dst_dir=os.path.abspath(workspace_dir),
93			mets_basename=ctx.mets_basename,
94			clobber_mets=clobber_mets,
95			download=download,
96			)
97			workspace.save_mets()
98			print(workspace.directory)
99
100			# ----------------------------------------------------------------------
101			# ocrd workspace init
102			# ----------------------------------------------------------------------
103
104			@workspace_cli.command('init')
105			@click.option('-f', '--clobber-mets', help="Clobber mets.xml if it exists", is_flag=True, default=False)
106			@click.argument('directory', required=False)
107			@pass_workspace
108			def workspace_create(ctx, clobber_mets, directory):
109			"""
110			Create a workspace with an empty METS file in DIRECTORY.
111
112			Use '.' for $PWD"
113			"""
114			if not directory:
115			directory = '.'
116			workspace = ctx.resolver.workspace_from_nothing(
117			directory=os.path.abspath(directory),
118			mets_basename=ctx.mets_basename,
119			clobber_mets=clobber_mets
120			)
121			workspace.save_mets()
122			print(workspace.directory)
123
124			# ----------------------------------------------------------------------
125			# ocrd workspace add
126			# ----------------------------------------------------------------------
127
128			@workspace_cli.command('add')
129			@click.option('-G', '--file-grp', help="fileGrp USE", required=True)
130			@click.option('-i', '--file-id', help="ID for the file", required=True)
131			@click.option('-m', '--mimetype', help="Media type of the file", required=True)
132			@click.option('-g', '--page-id', help="ID of the physical page")
133			@click.option('-C', '--check-file-exists', help="Whether to ensure FNAME exists", is_flag=True, default=False)
134			@click.option('--force', help="If file with ID already exists, replace it", default=False, is_flag=True)
135			@click.argument('fname', required=True)
136			@pass_workspace
137			def workspace_add_file(ctx, file_grp, file_id, mimetype, page_id, check_file_exists, force, fname):
138			"""
139			Add a file or http(s) URL FNAME to METS in a workspace.
140			If FNAME is not an http(s) URL and is not a workspace-local existing file, try to copy to workspace.
141			"""
142			workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup)
143
144			kwargs = {'fileGrp': file_grp, 'ID': file_id, 'mimetype': mimetype, 'pageId': page_id, 'force': force}
145			log = getLogger('ocrd.cli.workspace.add')
146			if not (fname.startswith('http://') or fname.startswith('https://')):
147			if not fname.startswith(ctx.directory):
148			if exists(join(ctx.directory, fname)):
149			fname = join(ctx.directory, fname)
150			else:
151			log.debug("File '%s' is not in workspace, copying", fname)
152			try:
153			fname = ctx.resolver.download_to_directory(ctx.directory, fname, subdir=file_grp)
154			except FileNotFoundError as e:
155			if check_file_exists:
156			log.error("File '%s' does not exist, halt execution!" % fname)
157			sys.exit(1)
158			if check_file_exists and not exists(fname):
159			log.error("File '%s' does not exist, halt execution!" % fname)
160			sys.exit(1)
161			if fname.startswith(ctx.directory):
162			fname = relpath(fname, ctx.directory)
163			kwargs['local_filename'] = fname
164
165			kwargs['url'] = fname
166			workspace.mets.add_file(**kwargs)
167			workspace.save_mets()
168
169			# ----------------------------------------------------------------------
170			# ocrd workspace find
171			# ----------------------------------------------------------------------
172
173			@workspace_cli.command('find')
174			@click.option('-G', '--file-grp', help="fileGrp USE")
175			@click.option('-m', '--mimetype', help="Media type to look for")
176			@click.option('-g', '--page-id', help="Page ID")
177			@click.option('-i', '--file-id', help="ID")
178			# pylint: disable=bad-continuation
179			@click.option('-k', '--output-field', help="Output field. Repeat for multiple fields, will be joined with tab",
180			default=['url'],
181			multiple=True,
182			type=click.Choice([
183			'url',
184			'mimetype',
185			'pageId',
186			'ID',
187			'fileGrp',
188			'basename',
189			'basename_without_extension',
190			'local_filename',
191			]))
192			@click.option('--download', is_flag=True, help="Download found files to workspace and change location in METS file ")
193			@pass_workspace
194			def workspace_find(ctx, file_grp, mimetype, page_id, file_id, output_field, download):
195			"""
196			Find files.
197			"""
198			modified_mets = False
199			ret = list()
200			workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename)
201			for f in workspace.mets.find_files(
202			ID=file_id,
203			fileGrp=file_grp,
204			mimetype=mimetype,
205			pageId=page_id,
206			):
207			if download and not f.local_filename:
208			workspace.download_file(f)
209			modified_mets = True
210			ret.append([f.ID if field == 'pageId' else getattr(f, field) or ''
211			for field in output_field])
212			if modified_mets:
213			workspace.save_mets()
214			if 'pageId' in output_field:
215			idx = output_field.index('pageId')
216			fileIds = list(map(lambda fields: fields[idx], ret))
			0 ignored issues – show introduced 2020-05-14 19:31 UTC by Report Bug Copy Issue Report The variable `idx` does not seem to be defined in case `'pageId' in output_field` on line `214` is `False`. Are you sure this can never be the case? Loading history...
217			pages = workspace.mets.get_physical_pages(for_fileIds=fileIds)
218			for fields, page in zip(ret, pages):
219			fields[idx] = page or ''
220			for fields in ret:
221			print('\t'.join(fields))
222
223			# ----------------------------------------------------------------------
224			# ocrd workspace remove
225			# ----------------------------------------------------------------------
226
227			@workspace_cli.command('remove')
228			@click.option('-k', '--keep-file', help="Do not delete file from file system", default=False, is_flag=True)
229			@click.option('-f', '--force', help="Continue even if mets:file or file on file system does not exist", default=False, is_flag=True)
230			@click.argument('ID', nargs=-1)
231			@pass_workspace
232			def workspace_remove_file(ctx, id, force, keep_file): # pylint: disable=redefined-builtin
233			"""
234			Delete file by ID from mets.xml
235			"""
236			workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup)
237			for i in id:
238			workspace.remove_file(i, force=force, keep_file=keep_file)
239			workspace.save_mets()
240
241
242			# ----------------------------------------------------------------------
243			# ocrd workspace remove-group
244			# ----------------------------------------------------------------------
245
246			@workspace_cli.command('remove-group', help="""
247
248			Delete a file group
249
250			""")
251			@click.option('-r', '--recursive', help="Delete any files in the group before the group itself", default=False, is_flag=True)
252			@click.option('-f', '--force', help="Continue removing even if group or containing files not found in METS", default=False, is_flag=True)
253			@click.option('-k', '--keep-files', help="Do not delete files from file system", default=False, is_flag=True)
254			@click.argument('GROUP', nargs=-1)
255			@pass_workspace
256			def remove_group(ctx, group, recursive, force, keep_files):
257			workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename)
258			for g in group:
259			workspace.remove_file_group(g, recursive=recursive, force=force, keep_files=keep_files)
260			workspace.save_mets()
261
262			# ----------------------------------------------------------------------
263			# ocrd workspace prune-files
264			# ----------------------------------------------------------------------
265
266			@workspace_cli.command('prune-files', help="""
267
268			Removes mets:files that point to non-existing local files
269
270			""")
271			@pass_workspace
272			def prune_files(ctx):
273			workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename)
274			with pushd_popd(workspace.directory):
275			for f in workspace.mets.find_files():
276			try:
277			if not f.local_filename or not exists(f.local_filename):
278			workspace.mets.remove_file(f.ID)
279			except Exception as e:
280			log.exception("Error removing %f: %s", f, e)
281			raise(e)
282			workspace.save_mets()
283
284			# ----------------------------------------------------------------------
285			# ocrd workspace list-group
286			# ----------------------------------------------------------------------
287
288			@workspace_cli.command('list-group', help="""
289
290			List fileGrp USE attributes
291
292			""")
293			@pass_workspace
294			def list_groups(ctx):
295			workspace = Workspace(ctx.resolver, directory=ctx.directory)
296			print("\n".join(workspace.mets.file_groups))
297
298			# ----------------------------------------------------------------------
299			# ocrd workspace list-pages
300			# ----------------------------------------------------------------------
301
302			@workspace_cli.command('list-page', help="""
303
304			List page IDs
305
306			""")
307			@pass_workspace
308			def list_pages(ctx):
309			workspace = Workspace(ctx.resolver, directory=ctx.directory)
310			print("\n".join(workspace.mets.physical_pages))
311
312			# ----------------------------------------------------------------------
313			# ocrd workspace get-id
314			# ----------------------------------------------------------------------
315
316			@workspace_cli.command('get-id', help="""
317
318			Get METS id if any
319
320			""")
321			@pass_workspace
322			def get_id(ctx):
323			workspace = Workspace(ctx.resolver, directory=ctx.directory)
324			ID = workspace.mets.unique_identifier
325			if ID:
326			print(ID)
327
328			# ----------------------------------------------------------------------
329			# ocrd workspace set-id
330			# ----------------------------------------------------------------------
331
332			@workspace_cli.command('set-id', help="""
333
334			Set METS ID.
335
336			If one of the supported identifier mechanisms is used, will set this identifier.
337
338			Otherwise will create a new <mods:identifier type="purl">{{ ID }}</mods:identifier>.
339			""")
340			@click.argument('ID')
341			@pass_workspace
342			def set_id(ctx, id): # pylint: disable=redefined-builtin
343			workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup)
344			workspace.mets.unique_identifier = id
345			workspace.save_mets()
346
347			# ----------------------------------------------------------------------
348			# ocrd workspace backup
349			# ----------------------------------------------------------------------
350
351			@workspace_cli.group('backup')
352			@click.pass_context
353			def workspace_backup_cli(ctx): # pylint: disable=unused-argument
354			"""
355			Backing and restoring workspaces - dev edition
356			"""
357
358			@workspace_backup_cli.command('add')
359			@pass_workspace
360			def workspace_backup_add(ctx):
361			"""
362			Create a new backup
363			"""
364			backup_manager = WorkspaceBackupManager(Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup))
365			backup_manager.add()
366
367			@workspace_backup_cli.command('list')
368			@pass_workspace
369			def workspace_backup_list(ctx):
370			"""
371			List backups
372			"""
373			backup_manager = WorkspaceBackupManager(Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup))
374			for b in backup_manager.list():
375			print(b)
376
377			@workspace_backup_cli.command('restore')
378			@click.option('-f', '--choose-first', help="Restore first matching version if more than one", is_flag=True)
379			@click.argument('bak') #, type=click.Path(dir_okay=False, readable=True, resolve_path=True))
380			@pass_workspace
381			def workspace_backup_restore(ctx, choose_first, bak):
382			"""
383			Restore backup BAK
384			"""
385			backup_manager = WorkspaceBackupManager(Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup))
386			backup_manager.restore(bak, choose_first)
387
388			@workspace_backup_cli.command('undo')
389			@pass_workspace
390			def workspace_backup_undo(ctx):
391			"""
392			Restore the last backup
393			"""
394			backup_manager = WorkspaceBackupManager(Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup))
395			backup_manager.undo()
396

OCR-D / core

Pull Request — master (#498)

ocrd.cli.workspace.set_id() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like