-
Notifications
You must be signed in to change notification settings - Fork 0
/
suss.go
682 lines (638 loc) · 15.9 KB
/
suss.go
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
// Package suss (Full name Suspicion) is a property-based testing library
//
// Property-based testing uses random generation of data to find edge cases
// that violate some property. Suspicion implements a state-of-the-art shrinker
// to find minimal examples of these edge cases.
//
// Suspicion was heavily-inspired by the python project Hypothesis and its internal component, conjecture.
// Users curious about internal workings can find more info at http://hypothesis.works/
package suss
import (
"bytes"
"fmt"
"io"
"math/rand"
"os"
"sort"
"testing"
"time"
)
// Runner is the main entry point to a Suspicion test.
type Runner struct {
rnd *rand.Rand
seeder *rand.Rand
t *testing.T
buf *buffer
lastBuf *buffer
tree *bufTree
testfunc func()
startTime time.Time
change int
}
// NewTest returns a Runner that runs a suspicion test.
func NewTest(t *testing.T) *Runner {
rnd := rand.New(rand.NewSource(time.Now().UnixNano()))
r := &Runner{
t: t,
seeder: rnd,
lastBuf: &buffer{},
tree: newBufTree(),
}
return r
}
func (r *Runner) newData() {
r.rnd = rand.New(rand.NewSource(int64(r.seeder.Uint64())))
r.buf = newBuffer(maxsize, r.regularDraw)
}
const maxsize = 8 << 10
// Run is the main entry point to a suspicion test.
// To run a suspicion test, give it a function that verifies some
// property and calls Runner.Fatalf if it's violated.
// The function is executed multiple times with different
// data to find a failing test.
//
// If data is found that causes the test to fail, then we
// will attempt to "shrink" the data. Shrinking involves
// making changes to the data, executing the test again
// and seeing if the test still fails.
//
// The function given should be a self contained function that can
// be called multiple times. This can be done by either making the
// function side-effect free or making the function implement setup and
// teardown logic. Since Suspicion uses panics as control flow,
// teardown should be done using defers.
func (r *Runner) Run(f func()) {
r.startTime = time.Now()
r.testfunc = f
r.newData()
mutations := 0
for !r.tree.dead[0] {
r.runOnce()
r.tree.add(r.buf)
if r.buf.status == statusInteresting {
r.lastBuf = r.buf
break
}
if time.Since(r.startTime) > 1*time.Second {
r.buf.discard()
return
}
if mutations >= 10 {
r.buf.discard()
r.newData()
mutations = 0
continue
}
mutations++
if r.considerNewBuffer(r.buf) {
r.lastBuf = r.buf
}
// this is not an interesting buffer, so we
// can discard its stdout regardless
r.buf.discard()
mut := r.newMutator()
r.buf = newBuffer(maxsize, mut)
}
// if we got here, that means that we have an interestinr.buffer
// That usually means a failing test, now try shrinking it
if r.buf.status != statusInteresting {
r.buf.discard()
return
}
r.lastBuf.finalize()
r.shrink()
// ok, open the output file and dump it on stdout
stdfile, err := os.Open(r.lastBuf.stdout)
if err != nil {
fmt.Println("could not open stdout: %v\n", err)
}
io.Copy(os.Stdout, stdfile)
stdfile.Close()
os.Remove(r.lastBuf.stdout)
r.t.FailNow()
}
func (r *Runner) shrink() {
r.startTime = time.Now()
change := -1
for r.change > change {
change = r.change
// structured interval delete
k := len(r.lastBuf.sortedInter) / 2
for k > 0 {
i := 0
for i+k <= len(r.lastBuf.sortedInter) {
// if I were clever, i'd use some sort of tree
// for this
elide := make([]bool, len(r.lastBuf.buf))
for _, v := range r.lastBuf.sortedInter[i : i+k] {
for t := v[0]; t < v[1]; t++ {
elide[t] = true
}
}
byt := make([]byte, 0, len(r.lastBuf.buf))
for i, v := range r.lastBuf.buf {
if elide[i] {
continue
}
byt = append(byt, v)
}
if !r.tryShrink(byt) {
i += k
}
}
k /= 2
}
r.zeroBlocks()
// entire buffer minimization
minimize(r.lastBuf.buf, r.tryShrink, true)
if change != r.change {
continue
}
// bulk replacing blocks with simpler blocks
i := 0
// can't use range here, lastbuf might change
for i < len(r.lastBuf.blocks) {
uv := r.lastBuf.blocks[i]
u, v := uv[0], uv[1]
buf := r.lastBuf.buf
block := buf[u:v]
n := v - u
byt := make([]byte, 0, len(r.lastBuf.buf))
for _, v := range r.lastBuf.blocks {
l := v[1] - v[0]
origblock := r.lastBuf.buf[v[0]:v[1]]
if l == n && (bytes.Compare(origblock, block) > 0) {
byt = append(byt, block...)
} else {
byt = append(byt, origblock...)
}
}
r.tryShrink(byt)
i++
}
// replace individual blocks with simpler blocks
i = 0
for i < len(r.lastBuf.blocks) {
uv := r.lastBuf.blocks[i]
u, v := uv[0], uv[1]
buf := r.lastBuf.buf
block := buf[u:v]
n := v - u
otherblocks := r.lastBuf.blockStarts[n]
// find all the blocks simpler than this
j := sort.Search(len(otherblocks), func(idx int) bool {
v := otherblocks[idx]
byt := r.lastBuf.buf[v : v+n]
return bytes.Compare(byt, block) >= 0
})
otherblocks = otherblocks[:j]
for _, b := range otherblocks {
byt := append([]byte(nil), r.lastBuf.buf...)
copy(byt[u:v], r.lastBuf.buf[b:b+n])
if r.tryShrink(byt) {
break
}
}
i++
}
// shrinking of duplicated blocks
blockChanged := -1
for blockChanged != r.change {
blockChanged = r.change
blocks := make(map[string][][2]int)
buf := append([]byte(nil), r.lastBuf.buf...)
for _, v := range r.lastBuf.blocks {
s := string(r.lastBuf.buf[v[0]:v[1]])
blocks[s] = append(blocks[s], v)
}
for k, v := range blocks {
if len(v) == 1 {
delete(blocks, k)
}
}
for k, s := range blocks {
minimize([]byte(k), func(b []byte) bool {
for _, v := range s {
copy(buf[v[0]:v[1]], b)
}
return r.tryShrink(buf)
}, false)
}
}
if change != r.change {
continue
}
// shrinking of individual blocks
i = 0
for i < len(r.lastBuf.blocks) {
block := r.lastBuf.blocks[i]
u, v := block[0], block[1]
buf := append([]byte(nil), r.lastBuf.buf[u:v]...)
minimize(buf, func(b []byte) bool {
byt := append([]byte(nil), r.lastBuf.buf...)
copy(byt[u:v], b)
return r.tryShrink(byt)
}, false)
i++
}
if change != r.change {
continue
}
// reordering blocks
blockLengths := make([]int, 0, len(r.lastBuf.blockStarts))
for k := range r.lastBuf.blockStarts {
blockLengths = append(blockLengths, k)
}
// Sort in descending order
sort.Slice(blockLengths, func(i, j int) bool {
return blockLengths[i] > blockLengths[j]
})
for _, n := range blockLengths {
i := 1
starts := startsByLoc(r.lastBuf, n)
for i < len(starts) {
j := i
for j > 0 {
as := starts[j-1]
bs := starts[i]
// Use lastbuf for reading and write
// into byt
a := r.lastBuf.buf[as : as+n]
b := r.lastBuf.buf[bs : bs+n]
if bytes.Compare(a, b) <= 0 {
break
}
byt := append([]byte(nil), r.lastBuf.buf...)
copy(byt[as:], b)
copy(byt[bs:], a)
if r.tryShrink(byt) {
starts = startsByLoc(r.lastBuf, n)
j -= 1
} else {
break
}
}
i += 1
}
}
// TODO: implement suffix shuffling while
// minimizing prefixes. This requires bind_points
// to be implemented and given the strategies involved
// in hypothesis, I suspect they might not be worth it
}
}
func startsByLoc(b *buffer, length int) []int {
// finalization of buffer sorts by simplicity of
// block, we want by start here
starts := append([]int(nil), b.blockStarts[length]...)
sort.Slice(starts, func(i, j int) bool {
return starts[i] < starts[j]
})
return starts
}
func (r *Runner) runOnce() {
testfail := true
defer func() {
rec := recover()
if rec == nil {
if testfail {
// we tell users to not use t.FailNow
// but if they do use it
// give them an error
panic("use of t.FailNow, t.Fatalf or similar")
}
return
}
switch rec.(type) {
case *eos:
r.buf.status = statusOverrun
return
case *failed:
r.buf.status = statusInteresting
return
case *invalid:
r.buf.status = statusInvalid
return
}
panic(r)
}()
f, closefunc, err := redirectOutput()
if err != nil {
panic("could not redirect output:" + err.Error())
}
defer closefunc()
r.buf.stdout = f.Name()
f.Close()
r.testfunc()
r.buf.status = statusValid
testfail = false
}
func (r *Runner) tryShrink(byt []byte) bool {
if r.lastBuf.status != statusInteresting {
panic("whoa")
}
s := r.lastBuf.index
if len(byt) > s {
byt = byt[:s]
}
i := 0
noveledge := false
for _, b := range byt {
if r.tree.dead[i] {
return false
}
var ok bool
i, ok = r.tree.nodes[i].edges[b]
if !ok {
noveledge = true
break
}
}
if !noveledge {
return false
}
r.buf = bufFromBytes(byt)
r.runOnce()
r.tree.add(r.buf)
r.buf.finalize()
if r.considerNewBuffer(r.buf) {
r.lastBuf.discard()
r.change += 1
r.lastBuf = r.buf
return true
}
r.buf.discard()
return false
}
func (r *Runner) zeroBlocks() {
lo := 0
numBlocks := len(r.lastBuf.blocks)
hi := numBlocks
for lo < hi {
mid := lo + (hi-lo)/2
byt := append([]byte(nil), r.lastBuf.buf...)
u := r.lastBuf.blocks[mid][0]
for i := u; i < len(byt); i++ {
byt[i] = 0
}
if r.tryShrink(byt) {
// TODO: figure out if this is right
// if we changed the number of blocks drawn
// then we could potentially run into out-of-bounds
// and linear time probing
if len(r.lastBuf.blocks) != numBlocks {
break
}
hi = mid
} else {
lo = mid + 1
}
}
for i := len(r.lastBuf.blocks) - 1; i >= 0; i-- {
// shrinking might change number of blocks in the
// last buffer
if i >= len(r.lastBuf.blocks) {
i = len(r.lastBuf.blocks)
continue
}
byt := append([]byte(nil), r.lastBuf.buf...)
block := r.lastBuf.blocks[i]
u, v := block[0], block[1]
for i := u; i < v; i++ {
byt[i] = 0
}
r.tryShrink(byt)
}
}
// Fatalf signals to the test runner that this test has failed.
// It takes a fmt.Printf format string that is printed
// when a minimal failing example has been found.
func (r *Runner) Fatalf(format string, i ...interface{}) {
// TODO: make this hook into the shrinking and gofuzz
fmt.Printf(format, i...)
fmt.Println()
panic(new(failed))
}
// Draw takes a generator and fills it with data. This is
// used to get the data that might cause a failing example.
func (r *Runner) Draw(g Generator) {
r.buf.StartExample()
g.Fill(r.buf)
r.buf.EndExample()
}
// Invalid signals to the test runner that the current data is
// invalid and should no longer be considered. This is useful
// for setting up assumptions like, "This float cannot be a NaN"
// or "This string must be at least 5 bytes long"
//
// Invalid calls panic internally. The test function should
// be aware of the non-local control-flow and use defers for
// cleanup.
func Invalid() {
panic(new(invalid))
}
func (r *Runner) regularDraw(b *buffer, n int, smp Sample) []byte {
res := smp(r.rnd, n)
return r.rewriteNovelty(b, res)
}
// The Sample type is a function, used to return sample values
// during the draw process. This is used to guide shrinking
// towards values which are meaningful and interesting.
//
// Meaningful means "can be interpreted to become a value".
// A good example of this is UTF-8 strings, where some
// byte sequences aren't valid values.
//
// Interesting means, "may cause a failure". A good example
// of an interesting value is floating point NaNs, which are
// known to cause failure in many bits of code.
type Sample func(r *rand.Rand, n int) []byte
// Uniform is a Sample function that return uninterpreted random bytes
// It's a convenience function for when any byte sequence is valid.
func Uniform(r *rand.Rand, n int) []byte {
b := make([]byte, n)
r.Read(b)
return b
}
// in case that we happen across a prefix or extension of a buffer
// we generated before, rewrite it with something we haven't seen before
func (r *Runner) rewriteNovelty(b *buffer, result []byte) []byte {
idx := b.nodeIndex
if idx == -1 {
if len(b.buf) != 0 {
fmt.Println(b.buf)
panic("invalid node index")
}
b.nodeIndex = 0
idx = 0
}
// we were novel before, stop the search
if b.hitNovelty == true {
return result
}
// any opportunity for us to become a dead node
// goes through previous nodes and we should have
// rewritten that.
if r.tree.dead[idx] {
panic("dead node")
}
n := r.tree.nodes[idx]
// walk the tree, looking for places where we
// would become dead and inserting new values there
for i, v := range result {
next, ok := n.edges[v]
if !ok {
b.hitNovelty = true
return result
}
nextn := r.tree.nodes[next]
if r.tree.dead[next] {
for c := 0; c < 256; c++ {
if _, ok := n.edges[byte(c)]; !ok {
result[i] = byte(c)
b.hitNovelty = true
return result
}
next = n.edges[byte(c)]
nextn = r.tree.nodes[next]
if !r.tree.dead[next] {
result[i] = byte(c)
break
}
}
}
idx = next
n = nextn
}
b.nodeIndex = idx
return result
}
func (r *Runner) newMutator() drawFunc {
mutateLibrary := []drawFunc{
r.drawNew,
r.drawExisting,
r.drawLarger,
r.drawSmaller,
r.drawZero,
r.drawConstant,
r.flipBit,
}
// choose 3 mutation functions and choose randomly
// between them on each draw
// This is the mutation scheme used by conjecture
perm := r.rnd.Perm(len(mutateLibrary))
mutateDraws := make([]drawFunc, 3)
for i := 0; i < 3; i++ {
mutateDraws[i] = mutateLibrary[perm[i]]
}
return func(b *buffer, n int, smp Sample) []byte {
var res []byte
if b.index+n > len(r.lastBuf.buf) {
res = smp(r.rnd, n)
} else {
d := r.seeder.Intn(len(mutateDraws))
res = mutateDraws[d](b, n, smp)
}
return r.rewriteNovelty(b, res)
}
}
func (r *Runner) drawLarger(b *buffer, n int, smp Sample) []byte {
exist := r.lastBuf.buf[b.index : b.index+n]
sample := smp(r.rnd, n)
if bytes.Compare(sample, exist) >= 0 {
return sample
}
return r.larger(exist)
}
func (r *Runner) drawSmaller(b *buffer, n int, smp Sample) []byte {
exist := r.lastBuf.buf[b.index : b.index+n]
sample := smp(r.rnd, n)
if bytes.Compare(sample, exist) <= 0 {
return sample
}
return r.smaller(exist)
}
func (r *Runner) drawNew(b *buffer, n int, smp Sample) []byte {
return smp(r.rnd, n)
}
func (r *Runner) drawExisting(b *buffer, n int, smp Sample) []byte {
ret := make([]byte, n)
copy(ret, r.lastBuf.buf[b.index:b.index+n])
return ret
}
func (r *Runner) drawZero(b *buffer, n int, smp Sample) []byte {
return make([]byte, n)
}
func (r *Runner) drawConstant(b *buffer, n int, smp Sample) []byte {
v := byte(r.rnd.Intn(256))
byt := make([]byte, n)
for i := 0; i < len(byt); i++ {
byt[i] = v
}
return byt
}
func (r *Runner) flipBit(b *buffer, n int, smp Sample) []byte {
byt := make([]byte, n)
copy(byt, r.lastBuf.buf[b.index:b.index+n])
i := r.rnd.Intn(n)
k := r.rnd.Intn(8)
byt[i] ^= 1 << byte(k)
return byt
}
func (r *Runner) larger(b []byte) []byte {
rnd := make([]byte, len(b))
drewlarger := false
for i := 0; i < len(b); i++ {
if !drewlarger {
v := 256 - int(b[i])
rnd[i] = b[i] + byte(r.rnd.Intn(v))
if rnd[i] > b[i] {
drewlarger = true
}
} else {
rnd[i] = byte(r.rnd.Intn(256))
}
}
return rnd
}
func (r *Runner) smaller(b []byte) []byte {
rnd := make([]byte, len(b))
drewsmaller := false
for i := 0; i < len(b); i++ {
if !drewsmaller {
rnd[i] = byte(r.rnd.Intn(int(b[i]) + 1))
if rnd[i] < b[i] {
drewsmaller = true
}
} else {
rnd[i] = byte(r.rnd.Intn(256))
}
}
return rnd
}
func (r *Runner) considerNewBuffer(b *buffer) bool {
if bytes.Compare(r.lastBuf.buf, b.buf) == 0 {
return false
}
if r.lastBuf.status != b.status {
return b.status > r.lastBuf.status
}
if b.status == statusInvalid {
return b.index >= r.lastBuf.index
}
if b.status == statusOverrun {
return b.overdraw < r.lastBuf.overdraw
}
if b.status == statusInteresting {
if len(b.buf) > len(r.lastBuf.buf) {
panic("buffer grew in size")
}
if len(b.buf) == len(r.lastBuf.buf) && bytes.Compare(b.buf, r.lastBuf.buf) >= 0 {
panic("buffer grew in value")
}
}
return true
}
type eos struct{}
type failed struct{}
type invalid struct{}