|
1
|
"""Tests for navegador.ingestion.bash — BashParser internal methods.""" |
|
2
|
|
|
3
|
from unittest.mock import MagicMock, patch |
|
4
|
|
|
5
|
import pytest |
|
6
|
|
|
7
|
from navegador.graph.schema import EdgeType, NodeLabel |
|
8
|
|
|
9
|
|
|
10
|
class MockNode: |
|
11
|
_id_counter = 0 |
|
12
|
|
|
13
|
def __init__( |
|
14
|
self, |
|
15
|
type_: str, |
|
16
|
text: bytes = b"", |
|
17
|
children: list = None, |
|
18
|
start_byte: int = 0, |
|
19
|
end_byte: int = 0, |
|
20
|
start_point: tuple = (0, 0), |
|
21
|
end_point: tuple = (0, 0), |
|
22
|
parent=None, |
|
23
|
): |
|
24
|
MockNode._id_counter += 1 |
|
25
|
self.id = MockNode._id_counter |
|
26
|
self.type = type_ |
|
27
|
self._text = text |
|
28
|
self.children = children or [] |
|
29
|
self.start_byte = start_byte |
|
30
|
self.end_byte = end_byte |
|
31
|
self.start_point = start_point |
|
32
|
self.end_point = end_point |
|
33
|
self.parent = parent |
|
34
|
self._fields: dict = {} |
|
35
|
for child in self.children: |
|
36
|
child.parent = self |
|
37
|
|
|
38
|
def child_by_field_name(self, name: str): |
|
39
|
return self._fields.get(name) |
|
40
|
|
|
41
|
def set_field(self, name: str, node): |
|
42
|
self._fields[name] = node |
|
43
|
node.parent = self |
|
44
|
return self |
|
45
|
|
|
46
|
|
|
47
|
def _text_node(text: bytes, type_: str = "identifier") -> MockNode: |
|
48
|
return MockNode(type_, text, start_byte=0, end_byte=len(text)) |
|
49
|
|
|
50
|
|
|
51
|
def _make_store(): |
|
52
|
store = MagicMock() |
|
53
|
store.query.return_value = MagicMock(result_set=[]) |
|
54
|
return store |
|
55
|
|
|
56
|
|
|
57
|
def _make_parser(): |
|
58
|
from navegador.ingestion.bash import BashParser |
|
59
|
|
|
60
|
parser = BashParser.__new__(BashParser) |
|
61
|
parser._parser = MagicMock() |
|
62
|
return parser |
|
63
|
|
|
64
|
|
|
65
|
class TestBashGetLanguage: |
|
66
|
def test_raises_when_not_installed(self): |
|
67
|
from navegador.ingestion.bash import _get_bash_language |
|
68
|
|
|
69
|
with patch.dict( |
|
70
|
"sys.modules", |
|
71
|
{ |
|
72
|
"tree_sitter_bash": None, |
|
73
|
"tree_sitter": None, |
|
74
|
}, |
|
75
|
): |
|
76
|
with pytest.raises(ImportError, match="tree-sitter-bash"): |
|
77
|
_get_bash_language() |
|
78
|
|
|
79
|
def test_returns_language_object(self): |
|
80
|
from navegador.ingestion.bash import _get_bash_language |
|
81
|
|
|
82
|
mock_tsbash = MagicMock() |
|
83
|
mock_ts = MagicMock() |
|
84
|
with patch.dict( |
|
85
|
"sys.modules", |
|
86
|
{ |
|
87
|
"tree_sitter_bash": mock_tsbash, |
|
88
|
"tree_sitter": mock_ts, |
|
89
|
}, |
|
90
|
): |
|
91
|
result = _get_bash_language() |
|
92
|
assert result is mock_ts.Language.return_value |
|
93
|
|
|
94
|
|
|
95
|
class TestBashNodeText: |
|
96
|
def test_extracts_bytes(self): |
|
97
|
from navegador.ingestion.bash import _node_text |
|
98
|
|
|
99
|
source = b"#!/bin/bash\nmy_func() {" |
|
100
|
node = MockNode( |
|
101
|
"identifier", |
|
102
|
start_byte=12, |
|
103
|
end_byte=19, |
|
104
|
) |
|
105
|
assert _node_text(node, source) == "my_func" |
|
106
|
|
|
107
|
|
|
108
|
class TestBashHandleFunction: |
|
109
|
def test_creates_function_node(self): |
|
110
|
parser = _make_parser() |
|
111
|
store = _make_store() |
|
112
|
source = b"deploy" |
|
113
|
name_node = MockNode( |
|
114
|
"word", |
|
115
|
start_byte=0, |
|
116
|
end_byte=6, |
|
117
|
) |
|
118
|
node = MockNode( |
|
119
|
"function_definition", |
|
120
|
start_point=(0, 0), |
|
121
|
end_point=(5, 1), |
|
122
|
) |
|
123
|
node.set_field("name", name_node) |
|
124
|
stats = {"functions": 0, "classes": 0, "edges": 0} |
|
125
|
parser._handle_function(node, source, "deploy.sh", store, stats) |
|
126
|
assert stats["functions"] == 1 |
|
127
|
assert stats["edges"] == 1 |
|
128
|
label = store.create_node.call_args[0][0] |
|
129
|
props = store.create_node.call_args[0][1] |
|
130
|
assert label == NodeLabel.Function |
|
131
|
assert props["name"] == "deploy" |
|
132
|
assert props["semantic_type"] == "shell_function" |
|
133
|
|
|
134
|
def test_skips_if_no_name_node(self): |
|
135
|
parser = _make_parser() |
|
136
|
store = _make_store() |
|
137
|
node = MockNode( |
|
138
|
"function_definition", |
|
139
|
start_point=(0, 0), |
|
140
|
end_point=(0, 5), |
|
141
|
) |
|
142
|
stats = {"functions": 0, "classes": 0, "edges": 0} |
|
143
|
parser._handle_function(node, b"", "test.sh", store, stats) |
|
144
|
assert stats["functions"] == 0 |
|
145
|
store.create_node.assert_not_called() |
|
146
|
|
|
147
|
def test_extracts_calls_from_body(self): |
|
148
|
parser = _make_parser() |
|
149
|
store = _make_store() |
|
150
|
source = b"deploy helper" |
|
151
|
name_node = MockNode( |
|
152
|
"word", |
|
153
|
start_byte=0, |
|
154
|
end_byte=6, |
|
155
|
) |
|
156
|
callee_name = MockNode( |
|
157
|
"word", |
|
158
|
start_byte=7, |
|
159
|
end_byte=13, |
|
160
|
) |
|
161
|
cmd = MockNode("command") |
|
162
|
cmd.set_field("name", callee_name) |
|
163
|
body = MockNode( |
|
164
|
"compound_statement", |
|
165
|
children=[cmd], |
|
166
|
) |
|
167
|
node = MockNode( |
|
168
|
"function_definition", |
|
169
|
start_point=(0, 0), |
|
170
|
end_point=(5, 1), |
|
171
|
) |
|
172
|
node.set_field("name", name_node) |
|
173
|
node.set_field("body", body) |
|
174
|
stats = {"functions": 0, "classes": 0, "edges": 0} |
|
175
|
parser._handle_function(node, source, "deploy.sh", store, stats) |
|
176
|
# 1 CONTAINS edge + 1 CALLS edge |
|
177
|
assert stats["edges"] == 2 |
|
178
|
|
|
179
|
|
|
180
|
class TestBashHandleVariable: |
|
181
|
def test_creates_variable_node_for_top_level(self): |
|
182
|
parser = _make_parser() |
|
183
|
store = _make_store() |
|
184
|
source = b'VERSION="1.0"' |
|
185
|
name_node = MockNode( |
|
186
|
"variable_name", |
|
187
|
start_byte=0, |
|
188
|
end_byte=7, |
|
189
|
) |
|
190
|
value_node = MockNode( |
|
191
|
"string", |
|
192
|
start_byte=8, |
|
193
|
end_byte=13, |
|
194
|
) |
|
195
|
program = MockNode("program") |
|
196
|
node = MockNode( |
|
197
|
"variable_assignment", |
|
198
|
start_point=(0, 0), |
|
199
|
end_point=(0, 13), |
|
200
|
parent=program, |
|
201
|
) |
|
202
|
node.set_field("name", name_node) |
|
203
|
node.set_field("value", value_node) |
|
204
|
# Re-set parent after construction since constructor |
|
205
|
# overwrites it |
|
206
|
node.parent = program |
|
207
|
stats = {"functions": 0, "classes": 0, "edges": 0} |
|
208
|
parser._handle_variable(node, source, "env.sh", store, stats) |
|
209
|
assert stats["edges"] == 1 |
|
210
|
label = store.create_node.call_args[0][0] |
|
211
|
props = store.create_node.call_args[0][1] |
|
212
|
assert label == NodeLabel.Variable |
|
213
|
assert props["name"] == "VERSION" |
|
214
|
assert props["semantic_type"] == "shell_variable" |
|
215
|
|
|
216
|
def test_skips_non_top_level_variable(self): |
|
217
|
parser = _make_parser() |
|
218
|
store = _make_store() |
|
219
|
source = b"x=1" |
|
220
|
name_node = MockNode( |
|
221
|
"variable_name", |
|
222
|
start_byte=0, |
|
223
|
end_byte=1, |
|
224
|
) |
|
225
|
func_parent = MockNode("function_definition") |
|
226
|
node = MockNode( |
|
227
|
"variable_assignment", |
|
228
|
start_point=(0, 0), |
|
229
|
end_point=(0, 3), |
|
230
|
parent=func_parent, |
|
231
|
) |
|
232
|
node.set_field("name", name_node) |
|
233
|
node.parent = func_parent |
|
234
|
stats = {"functions": 0, "classes": 0, "edges": 0} |
|
235
|
parser._handle_variable(node, source, "test.sh", store, stats) |
|
236
|
assert stats["edges"] == 0 |
|
237
|
store.create_node.assert_not_called() |
|
238
|
|
|
239
|
def test_skips_variable_without_name(self): |
|
240
|
parser = _make_parser() |
|
241
|
store = _make_store() |
|
242
|
program = MockNode("program") |
|
243
|
node = MockNode( |
|
244
|
"variable_assignment", |
|
245
|
start_point=(0, 0), |
|
246
|
end_point=(0, 3), |
|
247
|
parent=program, |
|
248
|
) |
|
249
|
node.parent = program |
|
250
|
stats = {"functions": 0, "classes": 0, "edges": 0} |
|
251
|
parser._handle_variable(node, b"", "test.sh", store, stats) |
|
252
|
store.create_node.assert_not_called() |
|
253
|
|
|
254
|
|
|
255
|
class TestBashHandleSource: |
|
256
|
def test_creates_import_for_source_command(self): |
|
257
|
parser = _make_parser() |
|
258
|
store = _make_store() |
|
259
|
source = b"source ./lib.sh" |
|
260
|
name_node = MockNode( |
|
261
|
"word", |
|
262
|
start_byte=0, |
|
263
|
end_byte=6, |
|
264
|
) |
|
265
|
arg_node = MockNode( |
|
266
|
"word", |
|
267
|
start_byte=7, |
|
268
|
end_byte=15, |
|
269
|
) |
|
270
|
node = MockNode( |
|
271
|
"command", |
|
272
|
children=[name_node, arg_node], |
|
273
|
start_point=(0, 0), |
|
274
|
end_point=(0, 15), |
|
275
|
) |
|
276
|
node.set_field("name", name_node) |
|
277
|
stats = {"functions": 0, "classes": 0, "edges": 0} |
|
278
|
parser._handle_command(node, source, "main.sh", store, stats) |
|
279
|
assert stats["edges"] == 1 |
|
280
|
label = store.create_node.call_args[0][0] |
|
281
|
props = store.create_node.call_args[0][1] |
|
282
|
assert label == NodeLabel.Import |
|
283
|
assert props["name"] == "./lib.sh" |
|
284
|
assert props["semantic_type"] == "shell_source" |
|
285
|
|
|
286
|
def test_creates_import_for_dot_command(self): |
|
287
|
parser = _make_parser() |
|
288
|
store = _make_store() |
|
289
|
source = b". /etc/profile" |
|
290
|
name_node = MockNode( |
|
291
|
"word", |
|
292
|
start_byte=0, |
|
293
|
end_byte=1, |
|
294
|
) |
|
295
|
arg_node = MockNode( |
|
296
|
"word", |
|
297
|
start_byte=2, |
|
298
|
end_byte=14, |
|
299
|
) |
|
300
|
node = MockNode( |
|
301
|
"command", |
|
302
|
children=[name_node, arg_node], |
|
303
|
start_point=(0, 0), |
|
304
|
end_point=(0, 14), |
|
305
|
) |
|
306
|
node.set_field("name", name_node) |
|
307
|
stats = {"functions": 0, "classes": 0, "edges": 0} |
|
308
|
parser._handle_command(node, source, "main.sh", store, stats) |
|
309
|
assert stats["edges"] == 1 |
|
310
|
props = store.create_node.call_args[0][1] |
|
311
|
assert props["name"] == "/etc/profile" |
|
312
|
|
|
313
|
def test_ignores_non_source_commands(self): |
|
314
|
parser = _make_parser() |
|
315
|
store = _make_store() |
|
316
|
source = b"echo hello" |
|
317
|
name_node = MockNode( |
|
318
|
"word", |
|
319
|
start_byte=0, |
|
320
|
end_byte=4, |
|
321
|
) |
|
322
|
node = MockNode( |
|
323
|
"command", |
|
324
|
children=[name_node], |
|
325
|
start_point=(0, 0), |
|
326
|
end_point=(0, 10), |
|
327
|
) |
|
328
|
node.set_field("name", name_node) |
|
329
|
stats = {"functions": 0, "classes": 0, "edges": 0} |
|
330
|
parser._handle_command(node, source, "main.sh", store, stats) |
|
331
|
assert stats["edges"] == 0 |
|
332
|
store.create_node.assert_not_called() |
|
333
|
|
|
334
|
def test_skips_source_without_arguments(self): |
|
335
|
parser = _make_parser() |
|
336
|
store = _make_store() |
|
337
|
source = b"source" |
|
338
|
name_node = MockNode( |
|
339
|
"word", |
|
340
|
start_byte=0, |
|
341
|
end_byte=6, |
|
342
|
) |
|
343
|
node = MockNode( |
|
344
|
"command", |
|
345
|
children=[name_node], |
|
346
|
start_point=(0, 0), |
|
347
|
end_point=(0, 6), |
|
348
|
) |
|
349
|
node.set_field("name", name_node) |
|
350
|
stats = {"functions": 0, "classes": 0, "edges": 0} |
|
351
|
parser._handle_command(node, source, "main.sh", store, stats) |
|
352
|
assert stats["edges"] == 0 |
|
353
|
store.create_node.assert_not_called() |
|
354
|
|
|
355
|
|
|
356
|
class TestBashExtractCalls: |
|
357
|
def test_finds_command_calls(self): |
|
358
|
parser = _make_parser() |
|
359
|
store = _make_store() |
|
360
|
source = b"build_app" |
|
361
|
callee = MockNode( |
|
362
|
"word", |
|
363
|
start_byte=0, |
|
364
|
end_byte=9, |
|
365
|
) |
|
366
|
cmd = MockNode("command") |
|
367
|
cmd.set_field("name", callee) |
|
368
|
body = MockNode( |
|
369
|
"compound_statement", |
|
370
|
children=[cmd], |
|
371
|
) |
|
372
|
fn_node = MockNode("function_definition") |
|
373
|
fn_node.set_field("body", body) |
|
374
|
stats = {"functions": 0, "classes": 0, "edges": 0} |
|
375
|
parser._extract_calls(fn_node, source, "deploy.sh", "deploy", store, stats) |
|
376
|
assert stats["edges"] == 1 |
|
377
|
edge_call = store.create_edge.call_args[0] |
|
378
|
assert edge_call[2] == EdgeType.CALLS |
|
379
|
assert edge_call[4]["name"] == "build_app" |
|
380
|
|
|
381
|
def test_skips_builtins(self): |
|
382
|
parser = _make_parser() |
|
383
|
store = _make_store() |
|
384
|
source = b"echo" |
|
385
|
callee = MockNode( |
|
386
|
"word", |
|
387
|
start_byte=0, |
|
388
|
end_byte=4, |
|
389
|
) |
|
390
|
cmd = MockNode("command") |
|
391
|
cmd.set_field("name", callee) |
|
392
|
body = MockNode( |
|
393
|
"compound_statement", |
|
394
|
children=[cmd], |
|
395
|
) |
|
396
|
fn_node = MockNode("function_definition") |
|
397
|
fn_node.set_field("body", body) |
|
398
|
stats = {"functions": 0, "classes": 0, "edges": 0} |
|
399
|
parser._extract_calls(fn_node, source, "test.sh", "myfunc", store, stats) |
|
400
|
assert stats["edges"] == 0 |
|
401
|
|
|
402
|
def test_no_calls_in_empty_body(self): |
|
403
|
parser = _make_parser() |
|
404
|
store = _make_store() |
|
405
|
fn_node = MockNode("function_definition") |
|
406
|
fn_node.set_field("body", MockNode("compound_statement")) |
|
407
|
stats = {"functions": 0, "classes": 0, "edges": 0} |
|
408
|
parser._extract_calls(fn_node, b"", "test.sh", "myfunc", store, stats) |
|
409
|
assert stats["edges"] == 0 |
|
410
|
|
|
411
|
def test_no_body_means_no_calls(self): |
|
412
|
parser = _make_parser() |
|
413
|
store = _make_store() |
|
414
|
fn_node = MockNode("function_definition") |
|
415
|
stats = {"functions": 0, "classes": 0, "edges": 0} |
|
416
|
parser._extract_calls(fn_node, b"", "test.sh", "myfunc", store, stats) |
|
417
|
assert stats["edges"] == 0 |
|
418
|
|
|
419
|
|
|
420
|
class TestBashWalkDispatch: |
|
421
|
def test_walk_handles_function_definition(self): |
|
422
|
parser = _make_parser() |
|
423
|
store = _make_store() |
|
424
|
source = b"deploy" |
|
425
|
name_node = MockNode( |
|
426
|
"word", |
|
427
|
start_byte=0, |
|
428
|
end_byte=6, |
|
429
|
) |
|
430
|
fn = MockNode( |
|
431
|
"function_definition", |
|
432
|
start_point=(0, 0), |
|
433
|
end_point=(5, 1), |
|
434
|
) |
|
435
|
fn.set_field("name", name_node) |
|
436
|
root = MockNode("program", children=[fn]) |
|
437
|
stats = {"functions": 0, "classes": 0, "edges": 0} |
|
438
|
parser._walk(root, source, "deploy.sh", store, stats) |
|
439
|
assert stats["functions"] == 1 |
|
440
|
|
|
441
|
def test_walk_handles_variable_assignment(self): |
|
442
|
parser = _make_parser() |
|
443
|
store = _make_store() |
|
444
|
source = b"VERSION" |
|
445
|
name_node = MockNode( |
|
446
|
"variable_name", |
|
447
|
start_byte=0, |
|
448
|
end_byte=7, |
|
449
|
) |
|
450
|
program = MockNode("program") |
|
451
|
var = MockNode( |
|
452
|
"variable_assignment", |
|
453
|
start_point=(0, 0), |
|
454
|
end_point=(0, 13), |
|
455
|
) |
|
456
|
var.set_field("name", name_node) |
|
457
|
program.children = [var] |
|
458
|
for child in program.children: |
|
459
|
child.parent = program |
|
460
|
stats = {"functions": 0, "classes": 0, "edges": 0} |
|
461
|
parser._walk(program, source, "env.sh", store, stats) |
|
462
|
assert stats["edges"] == 1 |
|
463
|
|
|
464
|
def test_walk_handles_source_command(self): |
|
465
|
parser = _make_parser() |
|
466
|
store = _make_store() |
|
467
|
source = b"source ./lib.sh" |
|
468
|
name_node = MockNode( |
|
469
|
"word", |
|
470
|
start_byte=0, |
|
471
|
end_byte=6, |
|
472
|
) |
|
473
|
arg_node = MockNode( |
|
474
|
"word", |
|
475
|
start_byte=7, |
|
476
|
end_byte=15, |
|
477
|
) |
|
478
|
cmd = MockNode( |
|
479
|
"command", |
|
480
|
children=[name_node, arg_node], |
|
481
|
start_point=(0, 0), |
|
482
|
end_point=(0, 15), |
|
483
|
) |
|
484
|
cmd.set_field("name", name_node) |
|
485
|
root = MockNode("program", children=[cmd]) |
|
486
|
stats = {"functions": 0, "classes": 0, "edges": 0} |
|
487
|
parser._walk(root, source, "main.sh", store, stats) |
|
488
|
assert stats["edges"] == 1 |
|
489
|
|
|
490
|
def test_walk_recurses_into_children(self): |
|
491
|
parser = _make_parser() |
|
492
|
store = _make_store() |
|
493
|
source = b"deploy" |
|
494
|
name_node = MockNode( |
|
495
|
"word", |
|
496
|
start_byte=0, |
|
497
|
end_byte=6, |
|
498
|
) |
|
499
|
fn = MockNode( |
|
500
|
"function_definition", |
|
501
|
start_point=(0, 0), |
|
502
|
end_point=(5, 1), |
|
503
|
) |
|
504
|
fn.set_field("name", name_node) |
|
505
|
wrapper = MockNode("if_statement", children=[fn]) |
|
506
|
root = MockNode("program", children=[wrapper]) |
|
507
|
stats = {"functions": 0, "classes": 0, "edges": 0} |
|
508
|
parser._walk(root, source, "deploy.sh", store, stats) |
|
509
|
assert stats["functions"] == 1 |
|
510
|
|
|
511
|
|
|
512
|
class TestBashParseFile: |
|
513
|
def test_creates_file_node(self): |
|
514
|
import tempfile |
|
515
|
from pathlib import Path |
|
516
|
|
|
517
|
parser = _make_parser() |
|
518
|
store = _make_store() |
|
519
|
mock_tree = MagicMock() |
|
520
|
mock_tree.root_node.type = "program" |
|
521
|
mock_tree.root_node.children = [] |
|
522
|
parser._parser.parse.return_value = mock_tree |
|
523
|
with tempfile.NamedTemporaryFile(suffix=".sh", delete=False) as f: |
|
524
|
f.write(b"#!/bin/bash\necho hello\n") |
|
525
|
fpath = Path(f.name) |
|
526
|
try: |
|
527
|
parser.parse_file(fpath, fpath.parent, store) |
|
528
|
store.create_node.assert_called_once() |
|
529
|
label = store.create_node.call_args[0][0] |
|
530
|
props = store.create_node.call_args[0][1] |
|
531
|
assert label == NodeLabel.File |
|
532
|
assert props["language"] == "bash" |
|
533
|
finally: |
|
534
|
fpath.unlink() |
|
535
|
|