|
1
|
|
|
""" |
|
2
|
|
|
OCR-D CLI: management of processor resources |
|
3
|
|
|
|
|
4
|
|
|
.. click:: ocrd.cli.resmgr:resmgr_cli |
|
5
|
|
|
:prog: ocrd resmgr |
|
6
|
|
|
:nested: full |
|
7
|
|
|
""" |
|
8
|
|
|
import sys |
|
9
|
|
|
from os import environ |
|
10
|
|
|
from pathlib import Path |
|
11
|
|
|
from distutils.spawn import find_executable as which |
|
12
|
|
|
|
|
13
|
|
|
import requests |
|
14
|
|
|
import click |
|
15
|
|
|
|
|
16
|
|
|
from ocrd_utils import ( |
|
17
|
|
|
initLogging, |
|
18
|
|
|
getLogger, |
|
19
|
|
|
RESOURCE_LOCATIONS |
|
20
|
|
|
) |
|
21
|
|
|
|
|
22
|
|
|
from ..resource_manager import OcrdResourceManager |
|
23
|
|
|
|
|
24
|
|
|
def print_resources(executable, reslist, resmgr): |
|
25
|
|
|
print('%s' % executable) |
|
26
|
|
|
for resdict in reslist: |
|
27
|
|
|
print('- %s %s (%s)\n %s' % ( |
|
28
|
|
|
resdict['name'], |
|
29
|
|
|
'@ %s' % resmgr.resource_dir_to_location(resdict['path']) if 'path' in resdict else '', |
|
30
|
|
|
resdict['url'], |
|
31
|
|
|
resdict['description'] |
|
32
|
|
|
)) |
|
33
|
|
|
print() |
|
34
|
|
|
|
|
35
|
|
|
@click.group("resmgr") |
|
36
|
|
|
def resmgr_cli(): |
|
37
|
|
|
""" |
|
38
|
|
|
Managing processor resources |
|
39
|
|
|
""" |
|
40
|
|
|
initLogging() |
|
41
|
|
|
|
|
42
|
|
|
@resmgr_cli.command('discover') |
|
43
|
|
|
@click.option('-e', '--executable', 'glob', default='ocrd-*', help="Glob pattern for executables to inspect") |
|
44
|
|
|
@click.option('-n', '--dry-run', is_flag=True, default=False, help="Do not save to user resource_list.yml") |
|
45
|
|
|
def discover(glob, dry_run): |
|
46
|
|
|
resmgr = OcrdResourceManager() |
|
47
|
|
|
ocrd_processors = [] |
|
48
|
|
|
for executable, reslist in resmgr.discover(glob=glob, dry_run=dry_run): |
|
49
|
|
|
print_resources(executable, reslist, resmgr) |
|
50
|
|
|
|
|
51
|
|
|
@resmgr_cli.command('list-available') |
|
52
|
|
|
@click.option('-e', '--executable', help='Show only resources for executable EXEC', metavar='EXEC') |
|
53
|
|
|
def list_available(executable=None): |
|
54
|
|
|
""" |
|
55
|
|
|
List available resources |
|
56
|
|
|
""" |
|
57
|
|
|
resmgr = OcrdResourceManager() |
|
58
|
|
|
for executable, reslist in resmgr.list_available(executable): |
|
59
|
|
|
print_resources(executable, reslist, resmgr) |
|
60
|
|
|
|
|
61
|
|
|
@resmgr_cli.command('list-installed') |
|
62
|
|
|
@click.option('-e', '--executable', help='Show only resources for executable EXEC', metavar='EXEC') |
|
63
|
|
|
def list_installed(executable=None): |
|
64
|
|
|
""" |
|
65
|
|
|
List installed resources |
|
66
|
|
|
""" |
|
67
|
|
|
resmgr = OcrdResourceManager() |
|
68
|
|
|
ret = [] |
|
69
|
|
|
for executable, reslist in resmgr.list_installed(executable): |
|
70
|
|
|
print_resources(executable, reslist, resmgr) |
|
71
|
|
|
|
|
72
|
|
|
@resmgr_cli.command('download') |
|
73
|
|
|
@click.option('-n', '--any-url', help='Allow downloading/copying unregistered resources', is_flag=True) |
|
74
|
|
|
@click.option('-a', '--allow-uninstalled', help="Allow installing resources for uninstalled processors", is_flag=True) |
|
75
|
|
|
@click.option('-o', '--overwrite', help='Overwrite existing resources', is_flag=True) |
|
76
|
|
|
@click.option('-l', '--location', help='Where to store resources', type=click.Choice(RESOURCE_LOCATIONS), default='data', show_default=True) |
|
77
|
|
|
@click.argument('executable', required=True) |
|
78
|
|
|
@click.argument('url_or_name', required=False) |
|
79
|
|
|
def download(any_url, allow_uninstalled, overwrite, location, executable, url_or_name): |
|
80
|
|
|
""" |
|
81
|
|
|
Download resource URL_OR_NAME for processor EXECUTABLE. |
|
82
|
|
|
|
|
83
|
|
|
URL_OR_NAME can either be the ``name`` or ``url`` of a registered resource. |
|
84
|
|
|
|
|
85
|
|
|
If URL_OR_NAME is '*' (asterisk), download all known resources for this processor |
|
86
|
|
|
|
|
87
|
|
|
If ``--any-url`` is given, also accepts URL or filenames of non-registered resources for ``URL_OR_NAME``. |
|
88
|
|
|
""" |
|
89
|
|
|
log = getLogger('ocrd.cli.resmgr') |
|
90
|
|
|
resmgr = OcrdResourceManager() |
|
91
|
|
|
basedir = resmgr.location_to_resource_dir(location) |
|
92
|
|
|
if executable != '*' and not url_or_name: |
|
93
|
|
|
log.error("Unless EXECUTABLE ('%s') is the '*' wildcard, URL_OR_NAME is required" % executable) |
|
94
|
|
|
sys.exit(1) |
|
95
|
|
|
elif executable == '*': |
|
96
|
|
|
executable = None |
|
97
|
|
|
is_url = (url_or_name.startswith('https://') or url_or_name.startswith('http://')) if url_or_name else False |
|
98
|
|
|
is_filename = Path(url_or_name).exists() if url_or_name else False |
|
99
|
|
|
if executable and not which(executable): |
|
100
|
|
|
if not allow_uninstalled: |
|
101
|
|
|
log.error("Executable %s is not installed. Is there a typo in the executable? " \ |
|
102
|
|
|
"To install resources for uninstalled processor, use the -a/--allow-uninstalled flag" % executable) |
|
103
|
|
|
sys.exit(1) |
|
104
|
|
|
else: |
|
105
|
|
|
log.warning("Executable %s is not installed but -a/--allow-uninstalled was given, so proceeding" % executable) |
|
106
|
|
|
find_kwargs = {'executable': executable} |
|
107
|
|
|
if url_or_name and url_or_name != '*': |
|
108
|
|
|
find_kwargs['url' if is_url else 'name'] = url_or_name |
|
109
|
|
|
reslist = resmgr.find_resources(**find_kwargs) |
|
110
|
|
|
if not reslist: |
|
111
|
|
|
log.info("No resources found in registry") |
|
112
|
|
|
if any_url and (is_url or is_filename): |
|
113
|
|
|
log.info("%s unregistered resource %s" % ("Downloading" if is_url else "Copying", url_or_name)) |
|
114
|
|
|
if is_url: |
|
115
|
|
|
with requests.get(url_or_name, stream=True) as r: |
|
116
|
|
|
content_length = int(r.headers.get('content-length')) |
|
117
|
|
|
else: |
|
118
|
|
|
url_or_name = str(Path(url_or_name).resolve()) |
|
119
|
|
|
content_length = Path(url_or_name).stat().st_size |
|
120
|
|
|
with click.progressbar(length=content_length, label="Downloading" if is_url else "Copying") as bar: |
|
121
|
|
|
fpath = resmgr.download( |
|
122
|
|
|
executable, |
|
123
|
|
|
url_or_name, |
|
124
|
|
|
overwrite=overwrite, |
|
125
|
|
|
basedir=basedir, |
|
126
|
|
|
no_subdir=location == 'cwd', |
|
127
|
|
|
progress_cb=lambda delta: bar.update(delta)) |
|
128
|
|
|
log.info("%s resource '%s' (%s) not a known resource, creating stub in %s'" % (executable, fpath.name, url_or_name, resmgr.user_list)) |
|
129
|
|
|
resmgr.add_to_user_database(executable, fpath, url_or_name) |
|
130
|
|
|
log.info("%s %s to %s" % ("Downloaded" if is_url else "Copied", url_or_name, fpath)) |
|
131
|
|
|
log.info("Use in parameters as '%s'" % fpath.name) |
|
132
|
|
|
else: |
|
133
|
|
|
sys.exit(1) |
|
134
|
|
|
else: |
|
135
|
|
|
for executable, resdict in reslist: |
|
136
|
|
|
if not allow_uninstalled and not which(executable): |
|
137
|
|
|
log.info("Skipping installing resources for %s as it is not installed. (Use -a/--allow-uninstalled to force)") |
|
138
|
|
|
continue |
|
139
|
|
|
if resdict['url'] == '???': |
|
140
|
|
|
log.info("Cannot download user resource %s" % (resdict['name'])), |
|
141
|
|
|
continue |
|
142
|
|
|
log.info("Downloading resource %s" % resdict) |
|
143
|
|
|
with click.progressbar(length=resdict['size']) as bar: |
|
144
|
|
|
fpath = resmgr.download( |
|
145
|
|
|
executable, |
|
146
|
|
|
resdict['url'], |
|
147
|
|
|
name=resdict['name'], |
|
148
|
|
|
resource_type=resdict['type'], |
|
149
|
|
|
path_in_archive=resdict.get('path_in_archive', '.'), |
|
150
|
|
|
overwrite=overwrite, |
|
151
|
|
|
size=resdict['size'], |
|
152
|
|
|
no_subdir=location == 'cwd', |
|
153
|
|
|
basedir=basedir, |
|
154
|
|
|
progress_cb=lambda delta: bar.update(delta) |
|
155
|
|
|
) |
|
156
|
|
|
log.info("Downloaded %s to %s" % (resdict['url'], fpath)) |
|
157
|
|
|
log.info("Use in parameters as '%s'" % resmgr.parameter_usage(resdict['name'], usage=resdict['parameter_usage'])) |
|
158
|
|
|
|
|
159
|
|
|
|
OCR-D /
core
