1
|
|
|
# -*- coding: utf-8 -*- |
2
|
|
|
|
3
|
|
|
import os |
4
|
|
|
import shutil |
5
|
|
|
|
6
|
|
|
from pathlib import ( |
7
|
|
|
Path |
8
|
|
|
) |
9
|
|
|
from requests import Session |
10
|
|
|
from unittest.mock import ( |
11
|
|
|
patch, |
12
|
|
|
Mock |
13
|
|
|
) |
14
|
|
|
from PIL import ( |
15
|
|
|
Image |
16
|
|
|
) |
17
|
|
|
|
18
|
|
|
from ocrd_models.ocrd_page import OcrdPage |
19
|
|
|
|
20
|
|
|
import pytest |
21
|
|
|
|
22
|
|
|
from tests.base import ( |
23
|
|
|
assets, |
24
|
|
|
main |
25
|
|
|
) |
26
|
|
|
|
27
|
|
|
from ocrd.resolver import Resolver |
28
|
|
|
from ocrd_utils import pushd_popd |
29
|
|
|
|
30
|
|
|
|
31
|
|
|
# set pylint once on module level |
32
|
|
|
# pylint: disable=protected-access |
33
|
|
|
|
34
|
|
|
METS_HEROLD = assets.url_of('SBB0000F29300010000/data/mets.xml') |
35
|
|
|
FOLDER_KANT = assets.path_to('kant_aufklaerung_1784') |
36
|
|
|
DATA_KANT = {'mets.xml': (os.path.join(FOLDER_KANT, 'data', 'mets.xml'), 'text/xml'), |
37
|
|
|
'INPUT_0017.tif': (os.path.join(FOLDER_KANT, 'data', 'OCR-D-IMG', 'INPUT_0017.tif'), 'image/tiff'), |
38
|
|
|
'INPUT_0020.tif': (os.path.join(FOLDER_KANT, 'data', 'OCR-D-IMG', 'INPUT_0020.tif'), 'image/tiff'), |
39
|
|
|
'PAGE_0017_ALTO.xml': (os.path.join(FOLDER_KANT, 'data', 'OCR-D-GT-ALTO', 'PAGE_0017_ALTO.xml'), 'text/xml'), |
40
|
|
|
'PAGE_0020_ALTO.xml': (os.path.join(FOLDER_KANT, 'data', 'OCR-D-GT-ALTO', 'PAGE_0020_ALTO.xml'), 'text/xml'), |
41
|
|
|
'PAGE_0017_PAGE.xml': (os.path.join(FOLDER_KANT, 'data', 'OCR-D-GT-PAGE', 'PAGE_0017_PAGE.xml'), 'text/xml'), |
42
|
|
|
'PAGE_0020_PAGE.xml': (os.path.join(FOLDER_KANT, 'data', 'OCR-D-GT-PAGE', 'PAGE_0020_PAGE.xml'), 'text/xml'), |
43
|
|
|
} |
44
|
|
|
|
45
|
|
|
|
46
|
|
|
def _get_kant_data(key): |
47
|
|
|
if key in DATA_KANT.keys(): |
48
|
|
|
(path, mime) = DATA_KANT[key] |
49
|
|
|
with open(path, mode='rb') as _file: |
50
|
|
|
return (_file.read(), mime) |
51
|
|
|
|
52
|
|
|
|
53
|
|
|
def request_behavior(*args, **kwargs): |
54
|
|
|
resp = Mock() |
55
|
|
|
resp.status_code = 200 |
56
|
|
|
resp.headers = {} |
57
|
|
|
the_key = args[0].split('/')[-1] |
58
|
|
|
if the_key in DATA_KANT: |
59
|
|
|
(cnt, mime) = _get_kant_data(the_key) |
60
|
|
|
resp.content = cnt |
61
|
|
|
resp.headers = {'Content-Type': mime} |
62
|
|
|
return resp |
63
|
|
|
|
64
|
|
|
|
65
|
|
|
def test_workspace_from_url_bad(): |
66
|
|
|
with pytest.raises(Exception) as exc: |
67
|
|
|
Resolver().workspace_from_url(None) |
68
|
|
|
|
69
|
|
|
# check exception |
70
|
|
|
assert "Must pass 'mets_url'" in str(exc) |
71
|
|
|
|
72
|
|
|
|
73
|
|
|
@patch.object(Session, "get") |
74
|
|
|
def test_workspace_from_url_kant(mock_request, tmp_path): |
75
|
|
|
|
76
|
|
|
# arrange |
77
|
|
|
url_src = 'https://raw.githubusercontent.com/OCR-D/assets/master/data/kant_aufklaerung_1784/data/mets.xml' |
78
|
|
|
mock_request.side_effect = request_behavior |
79
|
|
|
dst_dir = tmp_path / 'workspace_kant' |
80
|
|
|
dst_dir.mkdir() |
81
|
|
|
|
82
|
|
|
# act |
83
|
|
|
resolver = Resolver() |
84
|
|
|
resolver.workspace_from_url(url_src, mets_basename='foo.xml', dst_dir=dst_dir) |
85
|
|
|
|
86
|
|
|
# assert |
87
|
|
|
local_path = dst_dir / 'foo.xml' |
88
|
|
|
assert os.path.isfile(str(local_path)) |
89
|
|
|
# 1 time data was requested |
90
|
|
|
assert mock_request.call_count == 1 |
91
|
|
|
|
92
|
|
|
|
93
|
|
|
@patch.object(Session, "get") |
94
|
|
|
def test_workspace_from_url_kant_with_resources(mock_request, tmp_path): |
95
|
|
|
|
96
|
|
|
# arrange |
97
|
|
|
url_src = 'https://raw.githubusercontent.com/OCR-D/assets/master/data/kant_aufklaerung_1784/data/mets.xml' |
98
|
|
|
mock_request.side_effect = request_behavior |
99
|
|
|
dst_dir = tmp_path / 'workspace_kant' |
100
|
|
|
dst_dir.mkdir() |
101
|
|
|
|
102
|
|
|
# act |
103
|
|
|
resolver = Resolver() |
104
|
|
|
resolver.workspace_from_url(url_src, mets_basename='kant_aufklaerung_1784.xml', dst_dir=dst_dir, download=True) |
105
|
|
|
|
106
|
|
|
# assert files present under local tmp_path |
107
|
|
|
local_path_mets = dst_dir / 'kant_aufklaerung_1784.xml' |
108
|
|
|
assert os.path.isfile(str(local_path_mets)) |
109
|
|
|
local_path_img1 = dst_dir / 'OCR-D-IMG' / 'INPUT_0017.tif' |
110
|
|
|
assert os.path.isfile(str(local_path_img1)) |
111
|
|
|
local_path_page1 = dst_dir / 'OCR-D-GT-PAGE' / 'PAGE_0017_PAGE.xml' |
112
|
|
|
assert os.path.isfile(str(local_path_page1)) |
113
|
|
|
|
114
|
|
|
# 1 METS/MODS + 2 images + 4 OCR files = 7 requests |
115
|
|
|
assert mock_request.call_count == 7 |
116
|
|
|
|
117
|
|
|
|
118
|
|
|
@patch.object(Session, "get") |
119
|
|
|
def test_workspace_from_url_kant_with_resources_existing_local(mock_request, tmp_path): |
120
|
|
|
|
121
|
|
|
# arrange |
122
|
|
|
url_src = 'https://raw.githubusercontent.com/OCR-D/assets/master/data/kant_aufklaerung_1784/data/mets.xml' |
123
|
|
|
mock_request.side_effect = request_behavior |
124
|
|
|
dst_dir = tmp_path / 'workspace_kant' |
125
|
|
|
dst_dir.mkdir() |
126
|
|
|
src_mets = Path(assets.path_to('kant_aufklaerung_1784-binarized/data/mets.xml')) |
127
|
|
|
dst_mets = Path(dst_dir, 'mets.xml') |
128
|
|
|
shutil.copyfile(src_mets, dst_mets) |
129
|
|
|
|
130
|
|
|
# act |
131
|
|
|
Resolver().workspace_from_url(url_src, clobber_mets=False, dst_dir=dst_dir) |
132
|
|
|
|
133
|
|
|
# assert |
134
|
|
|
# no real request was made, since mets already present |
135
|
|
|
assert mock_request.call_count == 0 |
136
|
|
|
|
137
|
|
|
|
138
|
|
|
@patch.object(Session, "get") |
139
|
|
|
def test_workspace_from_url_404(mock_request): |
140
|
|
|
"""Expected behavior when try create workspace from invalid online target |
141
|
|
|
""" |
142
|
|
|
|
143
|
|
|
# arrange |
144
|
|
|
url_404 = 'https://raw.githubusercontent.com/OCR-D/assets/master/data/kant_aufklaerung_1784/data/mets.xmlX' |
145
|
|
|
mock_request.side_effect = Exception('HTTP request failed') |
146
|
|
|
|
147
|
|
|
with pytest.raises(Exception) as exc: |
148
|
|
|
Resolver().workspace_from_url(mets_url=url_404) |
149
|
|
|
|
150
|
|
|
# assert |
151
|
|
|
assert "HTTP request failed" in str(exc) |
152
|
|
|
assert mock_request.call_count == 1 |
153
|
|
|
|
154
|
|
|
|
155
|
|
|
def test_workspace_from_url_with_rel_dir(tmp_path): |
156
|
|
|
bogus_dst_dir = '../../../../../../../../../../../../../../../../%s' % str(tmp_path)[1:] |
157
|
|
|
|
158
|
|
|
# act |
159
|
|
|
with pushd_popd(FOLDER_KANT): |
160
|
|
|
ws1 = Resolver().workspace_from_url('data/mets.xml', dst_dir=bogus_dst_dir) |
161
|
|
|
|
162
|
|
|
# assert |
163
|
|
|
assert os.path.join(tmp_path, 'mets.xml') == ws1.mets_target |
164
|
|
|
assert str(tmp_path) == ws1.directory |
165
|
|
|
|
166
|
|
|
|
167
|
|
|
def test_workspace_from_url0(): |
168
|
|
|
|
169
|
|
|
# act |
170
|
|
|
workspace = Resolver().workspace_from_url(METS_HEROLD) |
171
|
|
|
input_files = workspace.mets.find_all_files(fileGrp='OCR-D-IMG') |
172
|
|
|
image_file = input_files[0] |
173
|
|
|
f = workspace.download_file(image_file) |
174
|
|
|
|
175
|
|
|
# assert |
176
|
|
|
assert '%s.tif' % f.ID == 'FILE_0001_IMAGE.tif' |
177
|
|
|
assert f.local_filename == 'OCR-D-IMG/FILE_0001_IMAGE.tif' |
178
|
|
|
|
179
|
|
|
|
180
|
|
|
def test_resolve_image0(): |
181
|
|
|
workspace = Resolver().workspace_from_url(METS_HEROLD) |
182
|
|
|
input_files = workspace.mets.find_all_files(fileGrp='OCR-D-IMG') |
183
|
|
|
f = input_files[0] |
184
|
|
|
print(f) |
185
|
|
|
img_pil1 = workspace._resolve_image_as_pil(f.local_filename) |
186
|
|
|
assert img_pil1.size == (2875, 3749) |
187
|
|
|
img_pil2 = workspace._resolve_image_as_pil(f.local_filename, [[0, 0], [1, 1]]) |
188
|
|
|
assert img_pil2.size == (1, 1) |
189
|
|
|
|
190
|
|
|
|
191
|
|
|
@pytest.mark.parametrize( |
192
|
|
|
"image_url,size_pil", |
193
|
|
|
[('OCR-D-IMG-NRM/OCR-D-IMG-NRM_0017.png', (1, 1)), |
194
|
|
|
('OCR-D-IMG-1BIT/OCR-D-IMG-1BIT_0017.png', (1, 1)), |
195
|
|
|
]) |
196
|
|
|
def test_resolve_image_as_pil(image_url, size_pil): |
197
|
|
|
url_path = assets.url_of('kant_aufklaerung_1784-binarized/data/mets.xml') |
198
|
|
|
workspace = Resolver().workspace_from_url(url_path) |
199
|
|
|
img_pil = workspace._resolve_image_as_pil(image_url, [[0, 0], [1, 1]]) |
200
|
|
|
assert img_pil.size == size_pil |
201
|
|
|
|
202
|
|
|
|
203
|
|
|
def test_resolve_image_as_pil_deprecated(): |
204
|
|
|
url_path = os.path.join(assets.url_of('kant_aufklaerung_1784-binarized'), 'data/mets.xml') |
205
|
|
|
workspace = Resolver().workspace_from_url(url_path) |
206
|
|
|
with pytest.warns(DeprecationWarning) as record: |
207
|
|
|
workspace.resolve_image_as_pil('OCR-D-IMG-NRM/OCR-D-IMG-NRM_0017.png') |
208
|
|
|
|
209
|
|
|
# assert |
210
|
|
|
assert len(record) == 1 |
211
|
|
|
assert 'Call to deprecated method resolve_image_as_pil.' in str(record[0].message) |
212
|
|
|
|
213
|
|
|
|
214
|
|
|
def test_workspace_from_nothing(): |
215
|
|
|
ws1 = Resolver().workspace_from_nothing(None) |
216
|
|
|
assert ws1.mets |
217
|
|
|
|
218
|
|
|
|
219
|
|
|
def test_workspace_from_nothing_makedirs(tmp_path): |
220
|
|
|
non_existent_dir = tmp_path / 'target' |
221
|
|
|
ws1 = Resolver().workspace_from_nothing(non_existent_dir) |
222
|
|
|
assert ws1.directory == non_existent_dir |
223
|
|
|
|
224
|
|
|
|
225
|
|
|
def test_workspace_from_nothing_noclobber(tmp_path): |
226
|
|
|
"""Attempt to re-create workspace shall fail because already created |
227
|
|
|
""" |
228
|
|
|
|
229
|
|
|
ws2 = Resolver().workspace_from_nothing(tmp_path) |
230
|
|
|
assert ws2.directory == tmp_path |
231
|
|
|
|
232
|
|
|
with pytest.raises(Exception) as exc: |
233
|
|
|
Resolver().workspace_from_nothing(tmp_path) |
234
|
|
|
|
235
|
|
|
# assert |
236
|
|
|
the_msg = "METS 'mets.xml' already exists in '%s' and clobber_mets not set" % tmp_path |
237
|
|
|
assert the_msg in str(exc) |
238
|
|
|
|
239
|
|
|
|
240
|
|
|
@pytest.mark.parametrize("url,basename,exc_msg", |
241
|
|
|
[(None, None, "'url' must be a non-empty string"), |
242
|
|
|
(None, 'foo', "'directory' must be a non-empty string")] |
243
|
|
|
) |
244
|
|
|
def test_download_to_directory_with_badargs(url, basename, exc_msg): |
245
|
|
|
|
246
|
|
|
with pytest.raises(ValueError) as exc: |
247
|
|
|
Resolver().download_to_directory(url, basename) |
248
|
|
|
|
249
|
|
|
# assert exception message contained |
250
|
|
|
assert exc_msg in str(exc) |
251
|
|
|
|
252
|
|
|
|
253
|
|
|
@pytest.fixture(name='fixture_copy_kant') |
254
|
|
|
def _fixture_copy_kant(tmp_path): |
255
|
|
|
temporary_phil = tmp_path / 'kant_aufklaerung_1784' |
256
|
|
|
shutil.copytree(FOLDER_KANT, temporary_phil) |
257
|
|
|
yield temporary_phil |
258
|
|
|
|
259
|
|
|
|
260
|
|
|
def test_download_to_directory_default(fixture_copy_kant): |
261
|
|
|
tmp_root = fixture_copy_kant.parent |
262
|
|
|
phil_data = fixture_copy_kant / 'data' / 'mets.xml' |
263
|
|
|
fn = Resolver().download_to_directory(str(tmp_root), str(phil_data)) |
264
|
|
|
assert Path(tmp_root, fn).exists() |
265
|
|
|
assert fn == 'mets.xml' |
266
|
|
|
|
267
|
|
|
|
268
|
|
|
def test_download_to_directory_basename(fixture_copy_kant): |
269
|
|
|
tmp_root = fixture_copy_kant.parent |
270
|
|
|
phil_data = fixture_copy_kant / 'data' / 'mets.xml' |
271
|
|
|
fn = Resolver().download_to_directory(str(tmp_root), str(phil_data), basename='foo') |
272
|
|
|
assert Path(tmp_root, fn).exists() |
273
|
|
|
assert fn == 'foo' |
274
|
|
|
|
275
|
|
|
|
276
|
|
|
def test_download_to_directory_subdir(fixture_copy_kant): |
277
|
|
|
tmp_root = fixture_copy_kant.parent |
278
|
|
|
phil_data = fixture_copy_kant / 'data' / 'mets.xml' |
279
|
|
|
fn = Resolver().download_to_directory(str(tmp_root), str(phil_data), subdir='baz') |
280
|
|
|
assert Path(tmp_root, fn).exists() |
281
|
|
|
assert fn == 'baz/mets.xml' |
282
|
|
|
|
283
|
|
|
|
284
|
|
|
def test_resolve_mets_arguments(): |
285
|
|
|
""" |
286
|
|
|
https://github.com/OCR-D/core/issues/693 |
287
|
|
|
https://github.com/OCR-D/core/issues/517 |
288
|
|
|
""" |
289
|
|
|
resolver = Resolver() |
290
|
|
|
assert resolver.resolve_mets_arguments(None, None, None, None) == (str(Path.cwd()), str(Path.cwd() / 'mets.xml'), 'mets.xml', None) |
291
|
|
|
assert resolver.resolve_mets_arguments('/', None, 'mets.xml', None) == ('/', '/mets.xml', 'mets.xml', None) |
292
|
|
|
assert resolver.resolve_mets_arguments('/foo', '/foo/foo.xml', None, None) == ('/foo', '/foo/foo.xml', 'foo.xml', None) |
293
|
|
|
assert resolver.resolve_mets_arguments(None, '/foo/foo.xml', None, None) == ('/foo', '/foo/foo.xml', 'foo.xml', None) |
294
|
|
|
assert resolver.resolve_mets_arguments('/foo', 'foo.xml', None, None) == ('/foo', '/foo/foo.xml', 'foo.xml', None) |
295
|
|
|
assert resolver.resolve_mets_arguments('/foo', 'http://bar/foo.xml', None, None) == ('/foo', 'http://bar/foo.xml', 'foo.xml', None) |
296
|
|
|
with pytest.raises(ValueError, match="Use either --mets or --mets-basename, not both"): |
297
|
|
|
resolver.resolve_mets_arguments('/', '/foo/bar', 'foo.xml', None) |
298
|
|
|
with pytest.raises(ValueError, match="inconsistent with --directory"): |
299
|
|
|
resolver.resolve_mets_arguments('/foo', '/bar/foo.xml', None, None) |
300
|
|
|
with pytest.warns(DeprecationWarning): |
301
|
|
|
resolver.resolve_mets_arguments('/foo', None, 'not_mets.xml', None) |
302
|
|
|
with pytest.raises(ValueError, match=r"--mets is an http\(s\) URL but no --directory was given"): |
303
|
|
|
resolver.resolve_mets_arguments(None, 'http://bar/foo.xml', None, None) |
304
|
|
|
|
305
|
|
|
if __name__ == '__main__': |
306
|
|
|
main(__file__) |
307
|
|
|
|