|
1
|
|
|
from pathlib import Path |
|
2
|
|
|
import re |
|
3
|
|
|
from shutil import copyfileobj, copytree |
|
4
|
|
|
from tempfile import TemporaryFile |
|
5
|
|
|
from tarfile import open as open_tarfile |
|
6
|
|
|
|
|
7
|
|
|
import requests |
|
8
|
|
|
from yaml import safe_load |
|
9
|
|
|
|
|
10
|
|
|
from .constants import RESOURCE_LIST_FILENAME |
|
11
|
|
|
|
|
12
|
|
|
from ocrd_validators import OcrdResourceListValidator |
|
13
|
|
|
from ocrd_utils import getLogger |
|
14
|
|
|
from ocrd_utils.constants import HOME, XDG_CACHE_HOME |
|
15
|
|
|
from ocrd_utils.os import list_resource_candidates, list_all_resources, pushd_popd |
|
16
|
|
|
|
|
17
|
|
|
builtin_list_filename = Path(RESOURCE_LIST_FILENAME) |
|
18
|
|
|
user_list_filename = Path(HOME, 'ocrd', 'resources.yml') |
|
19
|
|
|
|
|
20
|
|
|
class OcrdResourceManager(): |
|
21
|
|
|
|
|
22
|
|
|
""" |
|
23
|
|
|
Managing processor resources |
|
24
|
|
|
""" |
|
25
|
|
|
def __init__(self): |
|
26
|
|
|
self.log = getLogger('ocrd.resource_manager') |
|
27
|
|
|
self.database = {} |
|
28
|
|
|
self.load_resource_list(builtin_list_filename) |
|
29
|
|
|
self.load_resource_list(user_list_filename) |
|
30
|
|
|
|
|
31
|
|
|
def load_resource_list(self, list_filename): |
|
32
|
|
|
if list_filename.is_file(): |
|
33
|
|
|
with open(list_filename, 'r', encoding='utf-8') as f: |
|
34
|
|
|
list_loaded = safe_load(f) |
|
35
|
|
|
report = OcrdResourceListValidator.validate(list_loaded) |
|
36
|
|
|
if not report.is_valid: |
|
37
|
|
|
self.log.error('\n'.join(report.errors)) |
|
38
|
|
|
raise ValueError("Resource list %s is invalid!" % (list_filename)) |
|
39
|
|
|
for executable, resource_list in list_loaded.items(): |
|
40
|
|
|
if executable not in self.database: |
|
41
|
|
|
self.database[executable] = [] |
|
42
|
|
|
# Prepend, so user provided is sorted before builtin |
|
43
|
|
|
self.database[executable] = list_loaded[executable] + self.database[executable] |
|
44
|
|
|
|
|
45
|
|
|
def list_available(self, executable=None): |
|
46
|
|
|
""" |
|
47
|
|
|
List models available for download by processor |
|
48
|
|
|
""" |
|
49
|
|
|
if executable: |
|
50
|
|
|
return [(executable, self.database[executable])] |
|
51
|
|
|
return [(x, y) for x, y in self.database.items()] |
|
52
|
|
|
|
|
53
|
|
|
def list_installed(self, executable=None): |
|
54
|
|
|
""" |
|
55
|
|
|
List installed resources, matching with registry by ``name`` |
|
56
|
|
|
""" |
|
57
|
|
|
ret = [] |
|
58
|
|
|
for executable in [executable] if executable else self.database.keys(): |
|
59
|
|
|
reslist = [] |
|
60
|
|
|
for res_filename in list_all_resources(executable): |
|
61
|
|
|
res_name = Path(res_filename).name |
|
62
|
|
|
resdict = [x for x in self.database[executable] if x['name'] == res_name] |
|
63
|
|
|
if not resdict: |
|
64
|
|
|
# TODO handle gracefully |
|
65
|
|
|
resdict = [{'name': res_name, 'url': '???', 'description': '???', 'version_range': '???'}] |
|
66
|
|
|
reslist.append(resdict[0]) |
|
67
|
|
|
ret.append((executable, reslist)) |
|
68
|
|
|
return ret |
|
69
|
|
|
|
|
70
|
|
|
def find_resources(self, executable=None, name=None, url=None): |
|
71
|
|
|
""" |
|
72
|
|
|
Find resources in the registry |
|
73
|
|
|
""" |
|
74
|
|
|
ret = [] |
|
75
|
|
|
if executable and executable not in self.database.keys(): |
|
76
|
|
|
return ret |
|
77
|
|
|
for executable in [executable] if executable else self.database.keys(): |
|
78
|
|
|
for resdict in self.database[executable]: |
|
79
|
|
|
if not name and not url: |
|
80
|
|
|
ret.append((executable, resdict)) |
|
81
|
|
|
elif url and url == resdict['url']: |
|
82
|
|
|
ret.append((executable, resdict)) |
|
83
|
|
|
elif name and name == resdict['name']: |
|
84
|
|
|
ret.append((executable, resdict)) |
|
85
|
|
|
return ret |
|
86
|
|
|
|
|
87
|
|
|
def parameter_usage(self, name, usage='as-is'): |
|
88
|
|
|
if usage == 'as-is': |
|
89
|
|
|
return name |
|
90
|
|
|
if usage == 'without-extension': |
|
91
|
|
|
return Path(name).stem |
|
92
|
|
|
|
|
93
|
|
|
def _download_impl(self, url, filename, progress_cb=None): |
|
94
|
|
|
with open(filename, 'wb') as f: |
|
95
|
|
|
with requests.get(url, stream=True) as r: |
|
96
|
|
|
total = int(r.headers.get('content-length')) |
|
97
|
|
|
# copyfileobj(r.raw, f_write_tar) |
|
98
|
|
|
for data in r.iter_content(chunk_size=4096): |
|
99
|
|
|
if progress_cb: |
|
100
|
|
|
progress_cb(len(data)) |
|
101
|
|
|
f.write(data) |
|
102
|
|
|
|
|
103
|
|
|
# TODO Proper caching (make head request for size, If-Modified etc) |
|
104
|
|
|
def download( |
|
105
|
|
|
self, |
|
106
|
|
|
executable, |
|
107
|
|
|
url, |
|
108
|
|
|
overwrite=False, |
|
109
|
|
|
basedir=XDG_CACHE_HOME, |
|
110
|
|
|
name=None, |
|
111
|
|
|
resource_type='file', |
|
112
|
|
|
path_in_archive='.', |
|
113
|
|
|
progress_cb=None, |
|
114
|
|
|
): |
|
115
|
|
|
""" |
|
116
|
|
|
Download a resource by URL |
|
117
|
|
|
""" |
|
118
|
|
|
log = getLogger('ocrd.resource_manager.download') |
|
119
|
|
|
destdir = Path(basedir, executable) |
|
120
|
|
|
if not name: |
|
121
|
|
|
name = re.sub('[^A-Za-z0-9]', '', url) |
|
122
|
|
|
fpath = Path(destdir, name) |
|
123
|
|
|
if fpath.exists() and not overwrite: |
|
124
|
|
|
log.info("%s to be downloaded to %s which already exists and overwrite is False" % (url, fpath)) |
|
125
|
|
|
return fpath |
|
126
|
|
|
destdir.mkdir(parents=True, exist_ok=True) |
|
127
|
|
|
if resource_type == 'file': |
|
128
|
|
|
self._download_impl(url, fpath, progress_cb) |
|
129
|
|
|
elif resource_type == 'tarball': |
|
130
|
|
|
with pushd_popd(tempdir=True): |
|
131
|
|
|
log.info("Downloading %s" % url) |
|
132
|
|
|
self._download_impl(url, 'download.tar.xx', progress_cb) |
|
133
|
|
|
Path('out').mkdir() |
|
134
|
|
|
with pushd_popd('out'): |
|
135
|
|
|
log.info("Extracting tarball") |
|
136
|
|
|
with open_tarfile('../download.tar.xx', 'r:*') as tar: |
|
137
|
|
|
tar.extractall() |
|
138
|
|
|
log.info("Copying '%s' from tarball to %s" % (path_in_archive, fpath)) |
|
139
|
|
|
copytree(path_in_archive, str(fpath)) |
|
140
|
|
|
# TODO |
|
141
|
|
|
# elif resource_type == 'github-dir': |
|
142
|
|
|
return fpath |
|
143
|
|
|
|
OCR-D /
core
