cpython/Objects/stringlib/fastsearch.h at main · python/cpython · GitHub

History

830 lines (767 loc) · 26.5 KB

Raw

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

683

684

685

686

687

688

689

690

691

692

693

694

695

696

697

698

699

700

701

702

703

704

705

706

707

708

709

710

711

712

713

714

715

716

717

718

719

720

721

722

723

724

725

726

727

728

729

730

731

732

733

734

735

736

737

738

739

740

741

742

743

744

745

746

747

748

749

750

751

752

753

754

755

756

757

758

759

760

761

762

763

764

765

766

767

768

769

770

771

772

773

774

775

776

777

778

779

780

781

782

783

784

785

786

787

788

789

790

791

792

793

794

795

796

797

798

799

800

801

802

803

804

805

806

807

808

809

810

811

812

813

814

815

816

817

818

819

820

821

822

823

824

825

826

827

828

829

/* stringlib: fastsearch implementation */

#define STRINGLIB_FASTSEARCH_H

/* fast search/count implementation, based on a mix between boyer-

moore and horspool, with a few more bells and whistles on the top.

for some more background, see:

https://web.archive.org/web/20201107074620/http://effbot.org/zone/stringlib.htm */

/* note: fastsearch may access s[n], which isn't a problem when using

Python's ordinary string types, but may cause problems if you're

using this code in other contexts. also, the count mode returns -1

if there cannot possibly be a match in the target string, and 0 if

it has actually checked for matches, but didn't find any. callers

beware! */

/* If the strings are long enough, use Crochemore and Perrin's Two-Way

algorithm, which has worst-case O(n) runtime and best-case O(n/k).

Also compute a table of shifts to achieve O(n/k) in more cases,

and often (data dependent) deduce larger shifts than pure C&P can

deduce. See stringlib_find_two_way_notes.txt in this folder for a

detailed explanation. */

#define FAST_COUNT 0

#define FAST_SEARCH 1

#define FAST_RSEARCH 2

#if LONG_BIT >= 128

#define STRINGLIB_BLOOM_WIDTH 128

#elif LONG_BIT >= 64

#define STRINGLIB_BLOOM_WIDTH 64

#elif LONG_BIT >= 32

#define STRINGLIB_BLOOM_WIDTH 32

#else

#error "LONG_BIT is smaller than 32"

#endif

#define STRINGLIB_BLOOM_ADD(mask, ch) \

((mask |= (1UL << ((ch) & (STRINGLIB_BLOOM_WIDTH -1)))))

#define STRINGLIB_BLOOM(mask, ch) \

((mask & (1UL << ((ch) & (STRINGLIB_BLOOM_WIDTH -1)))))

#ifdef STRINGLIB_FAST_MEMCHR

# define MEMCHR_CUT_OFF 15

#else

# define MEMCHR_CUT_OFF 40

#endif

Py_LOCAL_INLINE(Py_ssize_t)

STRINGLIB(find_char)(const STRINGLIB_CHAR* s, Py_ssize_t n, STRINGLIB_CHAR ch)

{

const STRINGLIB_CHAR *p, *e;

p = s;

e = s + n;

if (n > MEMCHR_CUT_OFF) {

#ifdef STRINGLIB_FAST_MEMCHR

p = STRINGLIB_FAST_MEMCHR(s, ch, n);

if (p != NULL)

return (p - s);

return -1;

#else

/* use memchr if we can choose a needle without too many likely

false positives */

const STRINGLIB_CHAR *s1, *e1;

unsigned char needle = ch & 0xff;

/* If looking for a multiple of 256, we'd have too

many false positives looking for the '\0' byte in UCS2

and UCS4 representations. */

if (needle != 0) {

do {

const void *candidate = memchr(p, needle,

(e - p) * sizeof(STRINGLIB_CHAR));

if (candidate == NULL)

return -1;

s1 = p;

p = (const STRINGLIB_CHAR *)

_Py_ALIGN_DOWN(candidate, sizeof(STRINGLIB_CHAR));

if (*p == ch)

return (p - s);

/* False positive */

p++;

if (p - s1 > MEMCHR_CUT_OFF)

continue;

if (e - p <= MEMCHR_CUT_OFF)

break;

e1 = p + MEMCHR_CUT_OFF;

while (p != e1) {

if (*p == ch)

return (p - s);

p++;

}

while (e - p > MEMCHR_CUT_OFF);

}

#endif

}

while (p < e) {

if (*p == ch)

return (p - s);

p++;

}

return -1;

}

#undef MEMCHR_CUT_OFF

#if STRINGLIB_SIZEOF_CHAR == 1

# define MEMRCHR_CUT_OFF 15

#else

# define MEMRCHR_CUT_OFF 40

#endif

Py_LOCAL_INLINE(Py_ssize_t)

STRINGLIB(rfind_char)(const STRINGLIB_CHAR* s, Py_ssize_t n, STRINGLIB_CHAR ch)

{

const STRINGLIB_CHAR *p;

#ifdef HAVE_MEMRCHR

/* memrchr() is a GNU extension, available since glibc 2.1.91. it

doesn't seem as optimized as memchr(), but is still quite

faster than our hand-written loop below. There is no wmemrchr

for 4-byte chars. */

if (n > MEMRCHR_CUT_OFF) {

#if STRINGLIB_SIZEOF_CHAR == 1

p = memrchr(s, ch, n);

if (p != NULL)

return (p - s);

return -1;

#else

/* use memrchr if we can choose a needle without too many likely

false positives */

const STRINGLIB_CHAR *s1;

Py_ssize_t n1;

unsigned char needle = ch & 0xff;

/* If looking for a multiple of 256, we'd have too

many false positives looking for the '\0' byte in UCS2

and UCS4 representations. */

if (needle != 0) {

do {

void *candidate = memrchr(s, needle,

n * sizeof(STRINGLIB_CHAR));

if (candidate == NULL)

return -1;

n1 = n;

p = (const STRINGLIB_CHAR *)

_Py_ALIGN_DOWN(candidate, sizeof(STRINGLIB_CHAR));

n = p - s;

if (*p == ch)

return n;

/* False positive */

if (n1 - n > MEMRCHR_CUT_OFF)

continue;

if (n <= MEMRCHR_CUT_OFF)

break;

s1 = p - MEMRCHR_CUT_OFF;

while (p > s1) {

p--;

if (*p == ch)

return (p - s);

}

n = p - s;

}

while (n > MEMRCHR_CUT_OFF);

}

#endif

}

#endif /* HAVE_MEMRCHR */

p = s + n;

while (p > s) {

p--;

if (*p == ch)

return (p - s);

}

return -1;

}

#undef MEMRCHR_CUT_OFF

/* Change to a 1 to see logging comments walk through the algorithm. */

#if 0 && STRINGLIB_SIZEOF_CHAR == 1

# define LOG(...) printf(__VA_ARGS__)

# define LOG_STRING(s, n) printf("\"%.*s\"", (int)(n), s)

# define LOG_LINEUP() do { \

LOG("> "); LOG_STRING(haystack, len_haystack); LOG("\n> "); \

LOG("%*s",(int)(window_last - haystack + 1 - len_needle), ""); \

LOG_STRING(needle, len_needle); LOG("\n"); \

} while(0)

#else

# define LOG(...)

# define LOG_STRING(s, n)

# define LOG_LINEUP()

#endif

Py_LOCAL_INLINE(Py_ssize_t)

STRINGLIB(_lex_search)(const STRINGLIB_CHAR *needle, Py_ssize_t len_needle,

Py_ssize_t *return_period, int invert_alphabet)

{

/* Do a lexicographic search. Essentially this:

>>> max(needle[i:] for i in range(len(needle)+1))

Also find the period of the right half. */

Py_ssize_t max_suffix = 0;

Py_ssize_t candidate = 1;

Py_ssize_t k = 0;

// The period of the right half.

Py_ssize_t period = 1;

while (candidate + k < len_needle) {

// each loop increases candidate + k + max_suffix

STRINGLIB_CHAR a = needle[candidate + k];

STRINGLIB_CHAR b = needle[max_suffix + k];

// check if the suffix at candidate is better than max_suffix

if (invert_alphabet ? (b < a) : (a < b)) {

// Fell short of max_suffix.

// The next k + 1 characters are non-increasing

// from candidate, so they won't start a maximal suffix.

candidate += k + 1;

k = 0;

// We've ruled out any period smaller than what's

// been scanned since max_suffix.

period = candidate - max_suffix;

}

else if (a == b) {

if (k + 1 != period) {

// Keep scanning the equal strings

k++;

}

else {

// Matched a whole period.

// Start matching the next period.

candidate += period;

k = 0;

}

else {

// Did better than max_suffix, so replace it.

max_suffix = candidate;

candidate++;

k = 0;

period = 1;

}

*return_period = period;

return max_suffix;

}

Py_LOCAL_INLINE(Py_ssize_t)

STRINGLIB(_factorize)(const STRINGLIB_CHAR *needle,

Py_ssize_t len_needle,

Py_ssize_t *return_period)

{

/* Do a "critical factorization", making it so that:

>>> needle = (left := needle[:cut]) + (right := needle[cut:])

where the "local period" of the cut is maximal.

The local period of the cut is the minimal length of a string w

such that (left endswith w or w endswith left)

and (right startswith w or w startswith right).

The Critical Factorization Theorem says that this maximal local

period is the global period of the string.

Crochemore and Perrin (1991) show that this cut can be computed

as the later of two cuts: one that gives a lexicographically

maximal right half, and one that gives the same with the

with respect to a reversed alphabet-ordering.

This is what we want to happen:

>>> x = "GCAGAGAG"

>>> cut, period = factorize(x)

>>> x[:cut], (right := x[cut:])

('GC', 'AGAGAG')

>>> period # right half period

>>> right[period:] == right[:-period]

True

This is how the local period lines up in the above example:

GC | AGAGAG

AGAGAGC = AGAGAGC

The length of this minimal repetition is 7, which is indeed the

period of the original string. */

Py_ssize_t cut1, period1, cut2, period2, cut, period;

cut1 = STRINGLIB(_lex_search)(needle, len_needle, &period1, 0);

cut2 = STRINGLIB(_lex_search)(needle, len_needle, &period2, 1);

// Take the later cut.

if (cut1 > cut2) {

period = period1;

cut = cut1;

}

else {

period = period2;

cut = cut2;

}

LOG("split: "); LOG_STRING(needle, cut);

LOG(" + "); LOG_STRING(needle + cut, len_needle - cut);

LOG("\n");

*return_period = period;

return cut;

}

#define SHIFT_TYPE uint8_t

#define MAX_SHIFT UINT8_MAX

#define TABLE_SIZE_BITS 6u

#define TABLE_SIZE (1U << TABLE_SIZE_BITS)

#define TABLE_MASK (TABLE_SIZE - 1U)

typedef struct STRINGLIB(_pre) {

const STRINGLIB_CHAR *needle;

Py_ssize_t len_needle;

Py_ssize_t cut;

Py_ssize_t period;

Py_ssize_t gap;

int is_periodic;

SHIFT_TYPE table[TABLE_SIZE];

} STRINGLIB(prework);

static void

STRINGLIB(_preprocess)(const STRINGLIB_CHAR *needle, Py_ssize_t len_needle,

STRINGLIB(prework) *p)

{

p->needle = needle;

p->len_needle = len_needle;

p->cut = STRINGLIB(_factorize)(needle, len_needle, &(p->period));

assert(p->period + p->cut <= len_needle);

p->is_periodic = (0 == memcmp(needle,

needle + p->period,

p->cut * STRINGLIB_SIZEOF_CHAR));

if (p->is_periodic) {

assert(p->cut <= len_needle/2);

assert(p->cut < p->period);

}

else {

// A lower bound on the period

p->period = Py_MAX(p->cut, len_needle - p->cut) + 1;

}

// The gap between the last character and the previous

// occurrence of an equivalent character (modulo TABLE_SIZE)

p->gap = len_needle;

STRINGLIB_CHAR last = needle[len_needle - 1] & TABLE_MASK;

for (Py_ssize_t i = len_needle - 2; i >= 0; i--) {

STRINGLIB_CHAR x = needle[i] & TABLE_MASK;

if (x == last) {

p->gap = len_needle - 1 - i;

break;

}

// Fill up a compressed Boyer-Moore "Bad Character" table

Py_ssize_t not_found_shift = Py_MIN(len_needle, MAX_SHIFT);

for (Py_ssize_t i = 0; i < (Py_ssize_t)TABLE_SIZE; i++) {

p->table[i] = Py_SAFE_DOWNCAST(not_found_shift,

Py_ssize_t, SHIFT_TYPE);

}

for (Py_ssize_t i = len_needle - not_found_shift; i < len_needle; i++) {

SHIFT_TYPE shift = Py_SAFE_DOWNCAST(len_needle - 1 - i,

Py_ssize_t, SHIFT_TYPE);

p->table[needle[i] & TABLE_MASK] = shift;

}

static Py_ssize_t

STRINGLIB(_two_way)(const STRINGLIB_CHAR *haystack, Py_ssize_t len_haystack,

STRINGLIB(prework) *p)

{

// Crochemore and Perrin's (1991) Two-Way algorithm.

// See http://www-igm.univ-mlv.fr/~lecroq/string/node26.html#SECTION00260

const Py_ssize_t len_needle = p->len_needle;

const Py_ssize_t cut = p->cut;

Py_ssize_t period = p->period;

const STRINGLIB_CHAR *const needle = p->needle;

const STRINGLIB_CHAR *window_last = haystack + len_needle - 1;

const STRINGLIB_CHAR *const haystack_end = haystack + len_haystack;

SHIFT_TYPE *table = p->table;

const STRINGLIB_CHAR *window;

LOG("===== Two-way: \"%s\" in \"%s\". =====\n", needle, haystack);

Py_ssize_t gap = p->gap;

Py_ssize_t gap_jump_end = Py_MIN(len_needle, cut + gap);

if (p->is_periodic) {

LOG("Needle is periodic.\n");

Py_ssize_t memory = 0;

periodicwindowloop:

while (window_last < haystack_end) {

assert(memory == 0);

for (;;) {

LOG_LINEUP();

Py_ssize_t shift = table[(*window_last) & TABLE_MASK];

window_last += shift;

if (shift == 0) {

break;

}

if (window_last >= haystack_end) {

return -1;

}

LOG("Horspool skip\n");

}

no_shift:

window = window_last - len_needle + 1;

assert((window[len_needle - 1] & TABLE_MASK) ==

(needle[len_needle - 1] & TABLE_MASK));

Py_ssize_t i = Py_MAX(cut, memory);

for (; i < len_needle; i++) {

if (needle[i] != window[i]) {

if (i < gap_jump_end) {

LOG("Early right half mismatch: jump by gap.\n");

assert(gap >= i - cut + 1);

window_last += gap;

}

else {

LOG("Late right half mismatch: jump by n (>gap)\n");

assert(i - cut + 1 > gap);

window_last += i - cut + 1;

}

memory = 0;

goto periodicwindowloop;

}

for (i = memory; i < cut; i++) {

if (needle[i] != window[i]) {

LOG("Left half does not match.\n");

window_last += period;

memory = len_needle - period;

if (window_last >= haystack_end) {

return -1;

}

Py_ssize_t shift = table[(*window_last) & TABLE_MASK];

if (shift) {

// A mismatch has been identified to the right

// of where i will next start, so we can jump

// at least as far as if the mismatch occurred

// on the first comparison.

Py_ssize_t mem_jump = Py_MAX(cut, memory) - cut + 1;

LOG("Skip with Memory.\n");

memory = 0;

window_last += Py_MAX(shift, mem_jump);

goto periodicwindowloop;

}

goto no_shift;

}

LOG("Found a match!\n");

return window - haystack;

}

else {

period = Py_MAX(gap, period);

LOG("Needle is not periodic.\n");

windowloop:

while (window_last < haystack_end) {

for (;;) {

LOG_LINEUP();

Py_ssize_t shift = table[(*window_last) & TABLE_MASK];

window_last += shift;

if (shift == 0) {

break;

}

if (window_last >= haystack_end) {

return -1;

}

LOG("Horspool skip\n");

}

window = window_last - len_needle + 1;

assert((window[len_needle - 1] & TABLE_MASK) ==

(needle[len_needle - 1] & TABLE_MASK));

Py_ssize_t i = cut;

for (; i < len_needle; i++) {

if (needle[i] != window[i]) {

if (i < gap_jump_end) {

LOG("Early right half mismatch: jump by gap.\n");

assert(gap >= i - cut + 1);

window_last += gap;

}

else {

LOG("Late right half mismatch: jump by n (>gap)\n");

assert(i - cut + 1 > gap);

window_last += i - cut + 1;

}

goto windowloop;

}

for (Py_ssize_t i = 0; i < cut; i++) {

if (needle[i] != window[i]) {

LOG("Left half does not match.\n");

window_last += period;

goto windowloop;

}

LOG("Found a match!\n");

return window - haystack;

}

LOG("Not found. Returning -1.\n");

return -1;

}

static Py_ssize_t

STRINGLIB(_two_way_find)(const STRINGLIB_CHAR *haystack,

Py_ssize_t len_haystack,

const STRINGLIB_CHAR *needle,

Py_ssize_t len_needle)

{

LOG("###### Finding \"%s\" in \"%s\".\n", needle, haystack);

STRINGLIB(prework) p;

STRINGLIB(_preprocess)(needle, len_needle, &p);

return STRINGLIB(_two_way)(haystack, len_haystack, &p);

}

static Py_ssize_t

STRINGLIB(_two_way_count)(const STRINGLIB_CHAR *haystack,

Py_ssize_t len_haystack,

const STRINGLIB_CHAR *needle,

Py_ssize_t len_needle,

Py_ssize_t maxcount)

{

LOG("###### Counting \"%s\" in \"%s\".\n", needle, haystack);

STRINGLIB(prework) p;

STRINGLIB(_preprocess)(needle, len_needle, &p);

Py_ssize_t index = 0, count = 0;

while (1) {

Py_ssize_t result;

result = STRINGLIB(_two_way)(haystack + index,

len_haystack - index, &p);

if (result == -1) {

return count;

}

count++;

if (count == maxcount) {

return maxcount;

}

index += result + len_needle;

}

return count;

}

#undef SHIFT_TYPE

#undef NOT_FOUND

#undef SHIFT_OVERFLOW

#undef TABLE_SIZE_BITS

#undef TABLE_SIZE

#undef TABLE_MASK

#undef LOG

#undef LOG_STRING

#undef LOG_LINEUP

static inline Py_ssize_t

STRINGLIB(default_find)(const STRINGLIB_CHAR* s, Py_ssize_t n,

const STRINGLIB_CHAR* p, Py_ssize_t m,

Py_ssize_t maxcount, int mode)

{

const Py_ssize_t w = n - m;

Py_ssize_t mlast = m - 1, count = 0;

Py_ssize_t gap = mlast;

const STRINGLIB_CHAR last = p[mlast];

const STRINGLIB_CHAR *const ss = &s[mlast];

unsigned long mask = 0;

for (Py_ssize_t i = 0; i < mlast; i++) {

STRINGLIB_BLOOM_ADD(mask, p[i]);

if (p[i] == last) {

gap = mlast - i - 1;

}

STRINGLIB_BLOOM_ADD(mask, last);

for (Py_ssize_t i = 0; i <= w; i++) {

if (ss[i] == last) {

/* candidate match */

Py_ssize_t j;

for (j = 0; j < mlast; j++) {

if (s[i+j] != p[j]) {

break;

}

if (j == mlast) {

/* got a match! */

if (mode != FAST_COUNT) {

return i;

}

count++;

if (count == maxcount) {

return maxcount;

}

i = i + mlast;

continue;

}

/* miss: check if next character is part of pattern */

if (i + 1 <= w && !STRINGLIB_BLOOM(mask, ss[i+1])) {

i = i + m;

}

else {

i = i + gap;

}

else {

/* skip: check if next character is part of pattern */

if (i + 1 <= w && !STRINGLIB_BLOOM(mask, ss[i+1])) {

i = i + m;

}

return mode == FAST_COUNT ? count : -1;

}

static Py_ssize_t

STRINGLIB(adaptive_find)(const STRINGLIB_CHAR* s, Py_ssize_t n,

const STRINGLIB_CHAR* p, Py_ssize_t m,

Py_ssize_t maxcount, int mode)

{

const Py_ssize_t w = n - m;

Py_ssize_t mlast = m - 1, count = 0;

Py_ssize_t gap = mlast;

Py_ssize_t hits = 0, res;

const STRINGLIB_CHAR last = p[mlast];

const STRINGLIB_CHAR *const ss = &s[mlast];

unsigned long mask = 0;

for (Py_ssize_t i = 0; i < mlast; i++) {

STRINGLIB_BLOOM_ADD(mask, p[i]);

if (p[i] == last) {

gap = mlast - i - 1;

}

STRINGLIB_BLOOM_ADD(mask, last);

for (Py_ssize_t i = 0; i <= w; i++) {

if (ss[i] == last) {

/* candidate match */

Py_ssize_t j;

for (j = 0; j < mlast; j++) {

if (s[i+j] != p[j]) {

break;

}

if (j == mlast) {

/* got a match! */

if (mode != FAST_COUNT) {

return i;

}

count++;

if (count == maxcount) {

return maxcount;

}

i = i + mlast;

continue;

}

hits += j + 1;

if (hits > m / 4 && w - i > 2000) {

if (mode == FAST_SEARCH) {

res = STRINGLIB(_two_way_find)(s + i, n - i, p, m);

return res == -1 ? -1 : res + i;

}

else {

res = STRINGLIB(_two_way_count)(s + i, n - i, p, m,

maxcount - count);

return res + count;

}

/* miss: check if next character is part of pattern */

if (i + 1 <= w && !STRINGLIB_BLOOM(mask, ss[i+1])) {

i = i + m;

}

else {

i = i + gap;

}

else {

/* skip: check if next character is part of pattern */

if (i + 1 <= w && !STRINGLIB_BLOOM(mask, ss[i+1])) {

i = i + m;

}

return mode == FAST_COUNT ? count : -1;

}

static Py_ssize_t

STRINGLIB(default_rfind)(const STRINGLIB_CHAR* s, Py_ssize_t n,

const STRINGLIB_CHAR* p, Py_ssize_t m,

Py_ssize_t maxcount, int mode)

{

/* create compressed boyer-moore delta 1 table */

unsigned long mask = 0;

Py_ssize_t i, j, mlast = m - 1, skip = m - 1, w = n - m;

/* process pattern[0] outside the loop */

STRINGLIB_BLOOM_ADD(mask, p[0]);

/* process pattern[:0:-1] */

for (i = mlast; i > 0; i--) {

STRINGLIB_BLOOM_ADD(mask, p[i]);

if (p[i] == p[0]) {

skip = i - 1;

}

for (i = w; i >= 0; i--) {

if (s[i] == p[0]) {

/* candidate match */

for (j = mlast; j > 0; j--) {

if (s[i+j] != p[j]) {

break;

}

if (j == 0) {

/* got a match! */

return i;

}

/* miss: check if previous character is part of pattern */

if (i > 0 && !STRINGLIB_BLOOM(mask, s[i-1])) {

i = i - m;

}

else {

i = i - skip;

}

else {

/* skip: check if previous character is part of pattern */

if (i > 0 && !STRINGLIB_BLOOM(mask, s[i-1])) {

i = i - m;

}

return -1;

}

static inline Py_ssize_t

STRINGLIB(count_char)(const STRINGLIB_CHAR *s, Py_ssize_t n,

const STRINGLIB_CHAR p0, Py_ssize_t maxcount)

{

Py_ssize_t i, count = 0;

for (i = 0; i < n; i++) {

if (s[i] == p0) {

count++;

if (count == maxcount) {

return maxcount;

}

return count;

}

static inline Py_ssize_t

STRINGLIB(count_char_no_maxcount)(const STRINGLIB_CHAR *s, Py_ssize_t n,

const STRINGLIB_CHAR p0)

/* A specialized function of count_char that does not cut off at a maximum.

As a result, the compiler is able to vectorize the loop. */

{

Py_ssize_t count = 0;

for (Py_ssize_t i = 0; i < n; i++) {

if (s[i] == p0) {

count++;

}

return count;

}

Py_LOCAL_INLINE(Py_ssize_t)

FASTSEARCH(const STRINGLIB_CHAR* s, Py_ssize_t n,

const STRINGLIB_CHAR* p, Py_ssize_t m,

Py_ssize_t maxcount, int mode)

{

if (n < m || (mode == FAST_COUNT && maxcount == 0)) {

return -1;

}

/* look for special cases */

if (m <= 1) {

if (m <= 0) {

return -1;

}

/* use special case for 1-character strings */

if (mode == FAST_SEARCH)

return STRINGLIB(find_char)(s, n, p[0]);

else if (mode == FAST_RSEARCH)

return STRINGLIB(rfind_char)(s, n, p[0]);

else {

if (maxcount == PY_SSIZE_T_MAX) {

return STRINGLIB(count_char_no_maxcount)(s, n, p[0]);

}

return STRINGLIB(count_char)(s, n, p[0], maxcount);

}

if (mode != FAST_RSEARCH) {

if (n < 2500 || (m < 100 && n < 30000) || m < 6) {

return STRINGLIB(default_find)(s, n, p, m, maxcount, mode);

}

else if ((m >> 2) * 3 < (n >> 2)) {

/* 33% threshold, but don't overflow. */

/* For larger problems where the needle isn't a huge

percentage of the size of the haystack, the relatively

expensive O(m) startup cost of the two-way algorithm

will surely pay off. */

if (mode == FAST_SEARCH) {

return STRINGLIB(_two_way_find)(s, n, p, m);

}

else {

return STRINGLIB(_two_way_count)(s, n, p, m, maxcount);

}

else {

/* To ensure that we have good worst-case behavior,

here's an adaptive version of the algorithm, where if

we match O(m) characters without any matches of the

entire needle, then we predict that the startup cost of

the two-way algorithm will probably be worth it. */

return STRINGLIB(adaptive_find)(s, n, p, m, maxcount, mode);

}

else {

/* FAST_RSEARCH */

return STRINGLIB(default_rfind)(s, n, p, m, maxcount, mode);

}

cpython/Objects/stringlib/fastsearch.h at main · python/cpython · GitHub

Uh oh!

Uh oh!

cpython/Objects/stringlib/fastsearch.h at main · python/cpython · GitHub

Uh oh!

Uh oh!

Expand file tree

Breadcrumbs

fastsearch.h

Latest commit

History

Breadcrumbs

fastsearch.h

File metadata and controls