1
|
# SPDX-License-Identifier: CC0-1.0
|
2
|
|
3
|
"""
|
4
|
Haketilo unit tests - URL patterns
|
5
|
"""
|
6
|
|
7
|
# This file is part of Haketilo
|
8
|
#
|
9
|
# Copyright (C) 2021, Wojtek Kosior
|
10
|
#
|
11
|
# This program is free software: you can redistribute it and/or modify
|
12
|
# it under the terms of the CC0 1.0 Universal License as published by
|
13
|
# the Creative Commons Corporation.
|
14
|
#
|
15
|
# This program is distributed in the hope that it will be useful,
|
16
|
# but WITHOUT ANY WARRANTY; without even the implied warranty of
|
17
|
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
18
|
# CC0 1.0 Universal License for more details.
|
19
|
|
20
|
import pytest
|
21
|
|
22
|
from ..script_loader import load_script
|
23
|
|
24
|
@pytest.fixture(scope="session")
|
25
|
def patterns_tree_code():
|
26
|
yield load_script('common/patterns_query_tree.js', ['common'])
|
27
|
|
28
|
def test_modify_branch(execute_in_page, patterns_tree_code):
|
29
|
"""
|
30
|
patterns_query_tree.js contains Pattern Tree data structure that allows
|
31
|
arrays of string labels to be mapped to items.
|
32
|
Verify operations modifying a single branch of such tree work properly.
|
33
|
"""
|
34
|
execute_in_page(patterns_tree_code, page='https://gotmyowndoma.in')
|
35
|
execute_in_page(
|
36
|
'''
|
37
|
let items_added;
|
38
|
let items_removed;
|
39
|
|
40
|
function _item_adder(item, array)
|
41
|
{
|
42
|
items_added++;
|
43
|
return [...(array || []), item];
|
44
|
}
|
45
|
|
46
|
function item_adder(item)
|
47
|
{
|
48
|
items_added = 0;
|
49
|
return array => _item_adder(item, array);
|
50
|
}
|
51
|
|
52
|
function _item_remover(array)
|
53
|
{
|
54
|
if (array !== null) {
|
55
|
items_removed++;
|
56
|
array.pop();
|
57
|
}
|
58
|
return (array && array.length > 0) ? array : null;
|
59
|
}
|
60
|
|
61
|
function item_remover()
|
62
|
{
|
63
|
items_removed = 0;
|
64
|
return _item_remover;
|
65
|
}''')
|
66
|
|
67
|
# Let's construct some tree branch while checking that each addition gives
|
68
|
# the right result.
|
69
|
branch = execute_in_page(
|
70
|
'''{
|
71
|
const branch = empty_node();
|
72
|
modify_sequence(branch, ['com', 'example'], item_adder('some_item'));
|
73
|
returnval(branch);
|
74
|
}''')
|
75
|
assert branch == {
|
76
|
'literal_match': None,
|
77
|
'wildcard_matches': [None, None, None],
|
78
|
'children': {
|
79
|
'com': {
|
80
|
'literal_match': None,
|
81
|
'wildcard_matches': [None, None, None],
|
82
|
'children': {
|
83
|
'example': {
|
84
|
'literal_match': ['some_item'],
|
85
|
'wildcard_matches': [None, None, None],
|
86
|
'children': {
|
87
|
}
|
88
|
}
|
89
|
}
|
90
|
}
|
91
|
}
|
92
|
}
|
93
|
|
94
|
branch, items_added = execute_in_page(
|
95
|
'''{
|
96
|
const branch = arguments[0];
|
97
|
modify_sequence(branch, ['com', 'example'], item_adder('other_item'));
|
98
|
returnval([branch, items_added]);
|
99
|
}''', branch)
|
100
|
assert items_added == 1
|
101
|
assert branch['children']['com']['children']['example']['literal_match'] \
|
102
|
== ['some_item', 'other_item']
|
103
|
|
104
|
for i in range(3):
|
105
|
for expected_array in [['third_item'], ['third_item', '4th_item']]:
|
106
|
wildcard = '*' * (i + 1)
|
107
|
branch, items_added = execute_in_page(
|
108
|
'''{
|
109
|
const branch = arguments[0];
|
110
|
modify_sequence(branch, ['com', 'sample', arguments[1]],
|
111
|
item_adder(arguments[2]));
|
112
|
returnval([branch, items_added]);
|
113
|
}''',
|
114
|
branch, wildcard, expected_array[-1])
|
115
|
assert items_added == 2
|
116
|
sample = branch['children']['com']['children']['sample']
|
117
|
assert sample['wildcard_matches'][i] == expected_array
|
118
|
assert sample['children'][wildcard]['literal_match'] \
|
119
|
== expected_array
|
120
|
|
121
|
branch, items_added = execute_in_page(
|
122
|
'''{
|
123
|
const branch = arguments[0];
|
124
|
modify_sequence(branch, ['org', 'koszko', '***', '123'],
|
125
|
item_adder('5th_item'));
|
126
|
returnval([branch, items_added]);
|
127
|
}''',
|
128
|
branch)
|
129
|
assert items_added == 1
|
130
|
assert branch['children']['org']['children']['koszko']['children']['***']\
|
131
|
['children']['123']['literal_match'] == ['5th_item']
|
132
|
|
133
|
# Let's verify that removing a nonexistent element doesn't modify the tree.
|
134
|
branch2, items_removed = execute_in_page(
|
135
|
'''{
|
136
|
const branch = arguments[0];
|
137
|
modify_sequence(branch, ['com', 'not', 'registered', '*'],
|
138
|
item_remover());
|
139
|
returnval([branch, items_removed]);
|
140
|
}''',
|
141
|
branch)
|
142
|
assert branch == branch2
|
143
|
assert items_removed == 0
|
144
|
|
145
|
# Let's remove all elements in the tree branch while checking that each
|
146
|
# removal gives the right result.
|
147
|
branch, items_removed = execute_in_page(
|
148
|
'''{
|
149
|
const branch = arguments[0];
|
150
|
modify_sequence(branch, ['org', 'koszko', '***', '123'],
|
151
|
item_remover());
|
152
|
returnval([branch, items_removed]);
|
153
|
}''',
|
154
|
branch)
|
155
|
assert items_removed == 1
|
156
|
assert 'org' not in branch['children']
|
157
|
|
158
|
for i in range(3):
|
159
|
for expected_array in [['third_item'], None]:
|
160
|
wildcard = '*' * (i + 1)
|
161
|
branch, items_removed = execute_in_page(
|
162
|
'''{
|
163
|
const branch = arguments[0];
|
164
|
modify_sequence(branch, ['com', 'sample', arguments[1]],
|
165
|
item_remover());
|
166
|
returnval([branch, items_removed]);
|
167
|
}''',
|
168
|
branch, wildcard)
|
169
|
assert items_removed == 2
|
170
|
if i == 2 and expected_array == []:
|
171
|
break
|
172
|
sample = branch['children']['com']['children'].get('sample', {})
|
173
|
assert sample.get('wildcard_matches', [None, None, None])[i] \
|
174
|
== expected_array
|
175
|
assert sample.get('children', {}).get(wildcard, {})\
|
176
|
.get('literal_match') == expected_array
|
177
|
|
178
|
for i in range(2):
|
179
|
branch, items_removed = execute_in_page(
|
180
|
'''{
|
181
|
const branch = arguments[0];
|
182
|
modify_sequence(branch, ['com', 'example'], item_remover());
|
183
|
returnval([branch, items_removed]);
|
184
|
}''',
|
185
|
branch)
|
186
|
assert items_removed == 1
|
187
|
if i == 0:
|
188
|
assert branch['children']['com']['children']['example']\
|
189
|
['literal_match'] == ['some_item']
|
190
|
else:
|
191
|
assert branch == {
|
192
|
'literal_match': None,
|
193
|
'wildcard_matches': [None, None, None],
|
194
|
'children': {
|
195
|
}
|
196
|
}
|
197
|
|
198
|
def test_search_branch(execute_in_page, patterns_tree_code):
|
199
|
"""
|
200
|
patterns_query_tree.js contains Pattern Tree data structure that allows
|
201
|
arrays of string labels to be mapped to items.
|
202
|
Verify searching a single branch of such tree work properly.
|
203
|
"""
|
204
|
execute_in_page(patterns_tree_code, page='https://gotmyowndoma.in')
|
205
|
execute_in_page(
|
206
|
'''
|
207
|
const item_adder = item => (array => [...(array || []), item]);
|
208
|
''')
|
209
|
|
210
|
# Let's construct some tree branch to test on.
|
211
|
execute_in_page(
|
212
|
'''
|
213
|
var branch = empty_node();
|
214
|
|
215
|
for (const [item, sequence] of [
|
216
|
['(root)', []],
|
217
|
['***', ['***']],
|
218
|
['**', ['**']],
|
219
|
['*', ['*']],
|
220
|
|
221
|
['a', ['a']],
|
222
|
['A', ['a']],
|
223
|
['b', ['b']],
|
224
|
|
225
|
['a/***', ['a', '***']],
|
226
|
['A/***', ['a', '***']],
|
227
|
['a/**', ['a', '**']],
|
228
|
['A/**', ['a', '**']],
|
229
|
['a/*', ['a', '*']],
|
230
|
['A/*', ['a', '*']],
|
231
|
['a/sth', ['a', 'sth']],
|
232
|
['A/sth', ['a', 'sth']],
|
233
|
|
234
|
['b/***', ['b', '***']],
|
235
|
['b/**', ['b', '**']],
|
236
|
['b/*', ['b', '*']],
|
237
|
['b/sth', ['b', 'sth']],
|
238
|
])
|
239
|
modify_sequence(branch, sequence, item_adder(item));
|
240
|
''')
|
241
|
|
242
|
# Let's make the actual searches on our testing branch.
|
243
|
for sequence, expected in [
|
244
|
([], [{'(root)'}, {'***'}]),
|
245
|
(['a'], [{'a', 'A'}, {'a/***', 'A/***'}, {'*'}, {'***'}]),
|
246
|
(['b'], [{'b'}, {'b/***'}, {'*'}, {'***'}]),
|
247
|
(['c'], [ {'*'}, {'***'}]),
|
248
|
(['***'], [{'***'}, {'*'} ]),
|
249
|
(['**'], [{'**'}, {'*'}, {'***'}]),
|
250
|
(['**'], [{'**'}, {'*'}, {'***'}]),
|
251
|
(['*'], [{'*'}, {'***'}]),
|
252
|
|
253
|
(['a', 'sth'], [{'a/sth', 'A/sth'}, {'a/*', 'A/*'}, {'a/***', 'A/***'}, {'**'}, {'***'}]),
|
254
|
(['b', 'sth'], [{'b/sth'}, {'b/*'}, {'b/***'}, {'**'}, {'***'}]),
|
255
|
(['a', 'hts'], [ {'a/*', 'A/*'}, {'a/***', 'A/***'}, {'**'}, {'***'}]),
|
256
|
(['b', 'hts'], [ {'b/*'}, {'b/***'}, {'**'}, {'***'}]),
|
257
|
(['a', '***'], [{'a/***', 'A/***'}, {'a/*', 'A/*'}, {'**'}, {'***'}]),
|
258
|
(['b', '***'], [{'b/***'}, {'b/*'}, {'**'}, {'***'}]),
|
259
|
(['a', '**'], [{'a/**', 'A/**'}, {'a/*', 'A/*'}, {'a/***', 'A/***'}, {'**'}, {'***'}]),
|
260
|
(['b', '**'], [{'b/**'}, {'b/*'}, {'b/***'}, {'**'}, {'***'}]),
|
261
|
(['a', '*'], [{'a/*', 'A/*'}, {'a/***', 'A/***'}, {'**'}, {'***'}]),
|
262
|
(['b', '*'], [{'b/*'}, {'b/***'}, {'**'}, {'***'}]),
|
263
|
|
264
|
(['a', 'c', 'd'], [{'a/**', 'A/**'}, {'a/***', 'A/***'}, {'**'}, {'***'}]),
|
265
|
(['b', 'c', 'd'], [{'b/**'}, {'b/***'}, {'**'}, {'***'}])
|
266
|
]:
|
267
|
result = execute_in_page(
|
268
|
'''
|
269
|
returnval([...search_sequence(branch, arguments[0])]);
|
270
|
''',
|
271
|
sequence)
|
272
|
|
273
|
try:
|
274
|
assert len(result) == len(expected)
|
275
|
|
276
|
for expected_set, result_array in zip(expected, result):
|
277
|
assert len(expected_set) == len(result_array)
|
278
|
assert expected_set == set(result_array)
|
279
|
except Exception as e:
|
280
|
import sys
|
281
|
print('sequence:', sequence, '\nexpected:', expected,
|
282
|
'\nresult:', result, file=sys.stderr)
|
283
|
raise e from None
|
284
|
|
285
|
def test_pattern_tree(execute_in_page, patterns_tree_code):
|
286
|
"""
|
287
|
patterns_query_tree.js contains Pattern Tree data structure that allows
|
288
|
arrays of string labels to be mapped to items.
|
289
|
Verify operations on entire such tree work properly.
|
290
|
"""
|
291
|
execute_in_page(patterns_tree_code, page='https://gotmyowndoma.in')
|
292
|
|
293
|
# Perform tests with all possible patterns for a simple URL.
|
294
|
url = 'https://example.com'
|
295
|
patterns = [
|
296
|
'https://example.com',
|
297
|
'https://example.com/***',
|
298
|
'https://***.example.com',
|
299
|
'https://***.example.com/***'
|
300
|
]
|
301
|
bad_patterns = [
|
302
|
'http://example.com',
|
303
|
'https://a.example.com',
|
304
|
'https://*.example.com',
|
305
|
'https://**.example.com',
|
306
|
'https://example.com/a',
|
307
|
'https://example.com/*',
|
308
|
'https://example.com/**',
|
309
|
]
|
310
|
|
311
|
expected = [{'key': p} for p in patterns]
|
312
|
|
313
|
tree, result = execute_in_page(
|
314
|
'''{
|
315
|
const tree = pattern_tree.make();
|
316
|
for (const pattern of arguments[0].concat(arguments[1])) {
|
317
|
pattern_tree.register(tree, pattern, 'key', pattern);
|
318
|
pattern_tree.register(tree, pattern + '/', 'key', pattern + '/');
|
319
|
}
|
320
|
returnval([tree, [...pattern_tree.search(tree, arguments[2])]]);
|
321
|
}''',
|
322
|
patterns, bad_patterns, url)
|
323
|
assert expected == result
|
324
|
|
325
|
# Also verify that deregistering half of the good patterns works correctly.
|
326
|
patterns_removed = [pattern for i, pattern in enumerate(patterns) if i % 2]
|
327
|
patterns = [pattern for i, pattern in enumerate(patterns) if not (i % 2)]
|
328
|
expected = [{'key': p} for p in patterns]
|
329
|
tree, result = execute_in_page(
|
330
|
'''{
|
331
|
const tree = arguments[0];
|
332
|
for (const pattern of arguments[1]) {
|
333
|
pattern_tree.deregister(tree, pattern, 'key');
|
334
|
pattern_tree.deregister(tree, pattern + '/', 'key');
|
335
|
}
|
336
|
returnval([tree, [...pattern_tree.search(tree, arguments[2])]]);
|
337
|
}''',
|
338
|
tree, patterns_removed, url)
|
339
|
assert expected == result
|
340
|
|
341
|
# Also verify that deregistering all the patterns works correctly.
|
342
|
tree = execute_in_page(
|
343
|
'''{
|
344
|
const tree = arguments[0];
|
345
|
for (const pattern of arguments[1].concat(arguments[2])) {
|
346
|
pattern_tree.deregister(tree, pattern, 'key');
|
347
|
pattern_tree.deregister(tree, pattern + '/', 'key');
|
348
|
}
|
349
|
returnval(tree);
|
350
|
}''',
|
351
|
tree, patterns, bad_patterns)
|
352
|
assert tree == {}
|
353
|
|
354
|
# Perform tests with all possible patterns for a complex URL.
|
355
|
url = 'http://settings.query.example.com/google/tries/destroy/adblockers//'
|
356
|
patterns = [
|
357
|
'http://settings.query.example.com/google/tries/destroy/adblockers',
|
358
|
'http://settings.query.example.com/google/tries/destroy/adblockers/***',
|
359
|
'http://settings.query.example.com/google/tries/destroy/*',
|
360
|
'http://settings.query.example.com/google/tries/destroy/***',
|
361
|
'http://settings.query.example.com/google/tries/**',
|
362
|
'http://settings.query.example.com/google/tries/***',
|
363
|
'http://settings.query.example.com/google/**',
|
364
|
'http://settings.query.example.com/google/***',
|
365
|
'http://settings.query.example.com/**',
|
366
|
'http://settings.query.example.com/***',
|
367
|
|
368
|
'http://***.settings.query.example.com/google/tries/destroy/adblockers',
|
369
|
'http://***.settings.query.example.com/google/tries/destroy/adblockers/***',
|
370
|
'http://***.settings.query.example.com/google/tries/destroy/*',
|
371
|
'http://***.settings.query.example.com/google/tries/destroy/***',
|
372
|
'http://***.settings.query.example.com/google/tries/**',
|
373
|
'http://***.settings.query.example.com/google/tries/***',
|
374
|
'http://***.settings.query.example.com/google/**',
|
375
|
'http://***.settings.query.example.com/google/***',
|
376
|
'http://***.settings.query.example.com/**',
|
377
|
'http://***.settings.query.example.com/***',
|
378
|
'http://*.query.example.com/google/tries/destroy/adblockers',
|
379
|
'http://*.query.example.com/google/tries/destroy/adblockers/***',
|
380
|
'http://*.query.example.com/google/tries/destroy/*',
|
381
|
'http://*.query.example.com/google/tries/destroy/***',
|
382
|
'http://*.query.example.com/google/tries/**',
|
383
|
'http://*.query.example.com/google/tries/***',
|
384
|
'http://*.query.example.com/google/**',
|
385
|
'http://*.query.example.com/google/***',
|
386
|
'http://*.query.example.com/**',
|
387
|
'http://*.query.example.com/***',
|
388
|
'http://***.query.example.com/google/tries/destroy/adblockers',
|
389
|
'http://***.query.example.com/google/tries/destroy/adblockers/***',
|
390
|
'http://***.query.example.com/google/tries/destroy/*',
|
391
|
'http://***.query.example.com/google/tries/destroy/***',
|
392
|
'http://***.query.example.com/google/tries/**',
|
393
|
'http://***.query.example.com/google/tries/***',
|
394
|
'http://***.query.example.com/google/**',
|
395
|
'http://***.query.example.com/google/***',
|
396
|
'http://***.query.example.com/**',
|
397
|
'http://***.query.example.com/***',
|
398
|
'http://**.example.com/google/tries/destroy/adblockers',
|
399
|
'http://**.example.com/google/tries/destroy/adblockers/***',
|
400
|
'http://**.example.com/google/tries/destroy/*',
|
401
|
'http://**.example.com/google/tries/destroy/***',
|
402
|
'http://**.example.com/google/tries/**',
|
403
|
'http://**.example.com/google/tries/***',
|
404
|
'http://**.example.com/google/**',
|
405
|
'http://**.example.com/google/***',
|
406
|
'http://**.example.com/**',
|
407
|
'http://**.example.com/***',
|
408
|
'http://***.example.com/google/tries/destroy/adblockers',
|
409
|
'http://***.example.com/google/tries/destroy/adblockers/***',
|
410
|
'http://***.example.com/google/tries/destroy/*',
|
411
|
'http://***.example.com/google/tries/destroy/***',
|
412
|
'http://***.example.com/google/tries/**',
|
413
|
'http://***.example.com/google/tries/***',
|
414
|
'http://***.example.com/google/**',
|
415
|
'http://***.example.com/google/***',
|
416
|
'http://***.example.com/**',
|
417
|
'http://***.example.com/***'
|
418
|
]
|
419
|
bad_patterns = [
|
420
|
'https://settings.query.example.com/google/tries/destroy/adblockers',
|
421
|
'http://settings.query.example.com/google/tries/destroy/adblockers/a',
|
422
|
'http://settings.query.example.com/google/tries/destroy/adblockers/*',
|
423
|
'http://settings.query.example.com/google/tries/destroy/adblockers/**',
|
424
|
'http://settings.query.example.com/google/tries/destroy/a',
|
425
|
'http://settings.query.example.com/google/tries/destroy/**',
|
426
|
'http://settings.query.example.com/google/tries/*',
|
427
|
'http://a.settings.query.example.com/google/tries/destroy/adblockers',
|
428
|
'http://*.settings.query.example.com/google/tries/destroy/adblockers',
|
429
|
'http://**.settings.query.example.com/google/tries/destroy/adblockers',
|
430
|
'http://a.query.example.com/google/tries/destroy/adblockers',
|
431
|
'http://**.query.example.com/google/tries/destroy/adblockers',
|
432
|
'http://*.example.com/google/tries/destroy/adblockers'
|
433
|
]
|
434
|
|
435
|
expected = [{'key': p + s} for p in patterns for s in ['/', '']]
|
436
|
|
437
|
tree, result = execute_in_page(
|
438
|
'''{
|
439
|
const tree = pattern_tree.make();
|
440
|
for (const pattern of arguments[0].concat(arguments[1])) {
|
441
|
pattern_tree.register(tree, pattern, 'key', pattern);
|
442
|
pattern_tree.register(tree, pattern + '/', 'key', pattern + '/');
|
443
|
}
|
444
|
returnval([tree, [...pattern_tree.search(tree, arguments[2])]]);
|
445
|
}''',
|
446
|
patterns, bad_patterns, url)
|
447
|
assert expected == result
|
448
|
|
449
|
# Also verify that deregistering all patterns with trailing slash works
|
450
|
# correctly.
|
451
|
expected = [{'key': p} for p in patterns]
|
452
|
tree, result = execute_in_page(
|
453
|
'''{
|
454
|
const tree = arguments[0];
|
455
|
for (const pattern of arguments[1])
|
456
|
pattern_tree.deregister(tree, pattern + '/', 'key');
|
457
|
returnval([tree, [...pattern_tree.search(tree, arguments[2])]]);
|
458
|
}''',
|
459
|
tree, patterns, url)
|
460
|
assert expected == result
|
461
|
|
462
|
# Also verify that deregistering all the patterns works correctly.
|
463
|
tree = execute_in_page(
|
464
|
'''{
|
465
|
const tree = arguments[0];
|
466
|
for (const pattern of arguments[1])
|
467
|
pattern_tree.deregister(tree, pattern, 'key');
|
468
|
for (const pattern of arguments[2]) {
|
469
|
pattern_tree.deregister(tree, pattern, 'key');
|
470
|
pattern_tree.deregister(tree, pattern + '/', 'key');
|
471
|
}
|
472
|
returnval(tree);
|
473
|
}''',
|
474
|
tree, patterns, bad_patterns)
|
475
|
assert tree == {}
|