Bug Summary

File:include/llvm/ADT/edit_distance.h
Warning:line 97, column 3
Assigned value is garbage or undefined

Annotated Source Code

Press '?' to see keyboard shortcuts

clang -cc1 -triple x86_64-pc-linux-gnu -analyze -disable-free -disable-llvm-verifier -discard-value-names -main-file-name StringRef.cpp -analyzer-store=region -analyzer-opt-analyze-nested-blocks -analyzer-checker=core -analyzer-checker=apiModeling -analyzer-checker=unix -analyzer-checker=deadcode -analyzer-checker=cplusplus -analyzer-checker=security.insecureAPI.UncheckedReturn -analyzer-checker=security.insecureAPI.getpw -analyzer-checker=security.insecureAPI.gets -analyzer-checker=security.insecureAPI.mktemp -analyzer-checker=security.insecureAPI.mkstemp -analyzer-checker=security.insecureAPI.vfork -analyzer-checker=nullability.NullPassedToNonnull -analyzer-checker=nullability.NullReturnedFromNonnull -analyzer-output plist -w -analyzer-config-compatibility-mode=true -mrelocation-model pic -pic-level 2 -mthread-model posix -fmath-errno -masm-verbose -mconstructor-aliases -munwind-tables -fuse-init-array -target-cpu x86-64 -dwarf-column-info -debugger-tuning=gdb -momit-leaf-frame-pointer -ffunction-sections -fdata-sections -resource-dir /usr/lib/llvm-8/lib/clang/8.0.0 -D _DEBUG -D _GNU_SOURCE -D __STDC_CONSTANT_MACROS -D __STDC_FORMAT_MACROS -D __STDC_LIMIT_MACROS -I /build/llvm-toolchain-snapshot-8~svn350071/build-llvm/lib/Support -I /build/llvm-toolchain-snapshot-8~svn350071/lib/Support -I /build/llvm-toolchain-snapshot-8~svn350071/build-llvm/include -I /build/llvm-toolchain-snapshot-8~svn350071/include -U NDEBUG -internal-isystem /usr/lib/gcc/x86_64-linux-gnu/6.3.0/../../../../include/c++/6.3.0 -internal-isystem /usr/lib/gcc/x86_64-linux-gnu/6.3.0/../../../../include/x86_64-linux-gnu/c++/6.3.0 -internal-isystem /usr/lib/gcc/x86_64-linux-gnu/6.3.0/../../../../include/x86_64-linux-gnu/c++/6.3.0 -internal-isystem /usr/lib/gcc/x86_64-linux-gnu/6.3.0/../../../../include/c++/6.3.0/backward -internal-isystem /usr/include/clang/8.0.0/include/ -internal-isystem /usr/local/include -internal-isystem /usr/lib/llvm-8/lib/clang/8.0.0/include -internal-externc-isystem /usr/include/x86_64-linux-gnu -internal-externc-isystem /include -internal-externc-isystem /usr/include -O2 -Wno-unused-parameter -Wwrite-strings -Wno-missing-field-initializers -Wno-long-long -Wno-maybe-uninitialized -Wno-comment -std=c++11 -fdeprecated-macro -fdebug-compilation-dir /build/llvm-toolchain-snapshot-8~svn350071/build-llvm/lib/Support -fdebug-prefix-map=/build/llvm-toolchain-snapshot-8~svn350071=. -ferror-limit 19 -fmessage-length 0 -fvisibility-inlines-hidden -stack-protector 2 -fobjc-runtime=gcc -fdiagnostics-show-option -vectorize-loops -vectorize-slp -analyzer-output=html -analyzer-config stable-report-filename=true -o /tmp/scan-build-2018-12-27-042839-1215-1 -x c++ /build/llvm-toolchain-snapshot-8~svn350071/lib/Support/StringRef.cpp -faddrsig

/build/llvm-toolchain-snapshot-8~svn350071/lib/Support/StringRef.cpp

→
1//===-- StringRef.cpp - Lightweight String References ---------------------===//
2//
3// The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9
10#include "llvm/ADT/StringRef.h"
11#include "llvm/ADT/APFloat.h"
12#include "llvm/ADT/APInt.h"
13#include "llvm/ADT/Hashing.h"
14#include "llvm/ADT/StringExtras.h"
15#include "llvm/ADT/edit_distance.h"
16#include <bitset>
17
18using namespace llvm;
19
20// MSVC emits references to this into the translation units which reference it.
21#ifndef _MSC_VER
22const size_t StringRef::npos;
23#endif
24
25// strncasecmp() is not available on non-POSIX systems, so define an
26// alternative function here.
27static int ascii_strncasecmp(const char *LHS, const char *RHS, size_t Length) {
28 for (size_t I = 0; I < Length; ++I) {
29 unsigned char LHC = toLower(LHS[I]);
30 unsigned char RHC = toLower(RHS[I]);
31 if (LHC != RHC)
32 return LHC < RHC ? -1 : 1;
33 }
34 return 0;
35}
36
37/// compare_lower - Compare strings, ignoring case.
38int StringRef::compare_lower(StringRef RHS) const {
39 if (int Res = ascii_strncasecmp(Data, RHS.Data, std::min(Length, RHS.Length)))
40 return Res;
41 if (Length == RHS.Length)
42 return 0;
43 return Length < RHS.Length ? -1 : 1;
44}
45
46/// Check if this string starts with the given \p Prefix, ignoring case.
47bool StringRef::startswith_lower(StringRef Prefix) const {
48 return Length >= Prefix.Length &&
49 ascii_strncasecmp(Data, Prefix.Data, Prefix.Length) == 0;
50}
51
52/// Check if this string ends with the given \p Suffix, ignoring case.
53bool StringRef::endswith_lower(StringRef Suffix) const {
54 return Length >= Suffix.Length &&
55 ascii_strncasecmp(end() - Suffix.Length, Suffix.Data, Suffix.Length) == 0;
56}
57
58size_t StringRef::find_lower(char C, size_t From) const {
59 char L = toLower(C);
60 return find_if([L](char D) { return toLower(D) == L; }, From);
61}
62
63/// compare_numeric - Compare strings, handle embedded numbers.
64int StringRef::compare_numeric(StringRef RHS) const {
65 for (size_t I = 0, E = std::min(Length, RHS.Length); I != E; ++I) {
66 // Check for sequences of digits.
67 if (isDigit(Data[I]) && isDigit(RHS.Data[I])) {
68 // The longer sequence of numbers is considered larger.
69 // This doesn't really handle prefixed zeros well.
70 size_t J;
71 for (J = I + 1; J != E + 1; ++J) {
72 bool ld = J < Length && isDigit(Data[J]);
73 bool rd = J < RHS.Length && isDigit(RHS.Data[J]);
74 if (ld != rd)
75 return rd ? -1 : 1;
76 if (!rd)
77 break;
78 }
79 // The two number sequences have the same length (J-I), just memcmp them.
80 if (int Res = compareMemory(Data + I, RHS.Data + I, J - I))
81 return Res < 0 ? -1 : 1;
82 // Identical number sequences, continue search after the numbers.
83 I = J - 1;
84 continue;
85 }
86 if (Data[I] != RHS.Data[I])
87 return (unsigned char)Data[I] < (unsigned char)RHS.Data[I] ? -1 : 1;
88 }
89 if (Length == RHS.Length)
90 return 0;
91 return Length < RHS.Length ? -1 : 1;
92}
93
94// Compute the edit distance between the two given strings.
95unsigned StringRef::edit_distance(llvm::StringRef Other,
96 bool AllowReplacements,
97 unsigned MaxEditDistance) const {
98 return llvm::ComputeEditDistance(
1
Calling 'ComputeEditDistance<char>'
→
99 makeArrayRef(data(), size()),
100 makeArrayRef(Other.data(), Other.size()),
101 AllowReplacements, MaxEditDistance);
102}
103
104//===----------------------------------------------------------------------===//
105// String Operations
106//===----------------------------------------------------------------------===//
107
108std::string StringRef::lower() const {
109 std::string Result(size(), char());
110 for (size_type i = 0, e = size(); i != e; ++i) {
111 Result[i] = toLower(Data[i]);
112 }
113 return Result;
114}
115
116std::string StringRef::upper() const {
117 std::string Result(size(), char());
118 for (size_type i = 0, e = size(); i != e; ++i) {
119 Result[i] = toUpper(Data[i]);
120 }
121 return Result;
122}
123
124//===----------------------------------------------------------------------===//
125// String Searching
126//===----------------------------------------------------------------------===//
127
128
129/// find - Search for the first string \arg Str in the string.
130///
131/// \return - The index of the first occurrence of \arg Str, or npos if not
132/// found.
133size_t StringRef::find(StringRef Str, size_t From) const {
134 if (From > Length)
135 return npos;
136
137 const char *Start = Data + From;
138 size_t Size = Length - From;
139
140 const char *Needle = Str.data();
141 size_t N = Str.size();
142 if (N == 0)
143 return From;
144 if (Size < N)
145 return npos;
146 if (N == 1) {
147 const char *Ptr = (const char *)::memchr(Start, Needle[0], Size);
148 return Ptr == nullptr ? npos : Ptr - Data;
149 }
150
151 const char *Stop = Start + (Size - N + 1);
152
153 // For short haystacks or unsupported needles fall back to the naive algorithm
154 if (Size < 16 || N > 255) {
155 do {
156 if (std::memcmp(Start, Needle, N) == 0)
157 return Start - Data;
158 ++Start;
159 } while (Start < Stop);
160 return npos;
161 }
162
163 // Build the bad char heuristic table, with uint8_t to reduce cache thrashing.
164 uint8_t BadCharSkip[256];
165 std::memset(BadCharSkip, N, 256);
166 for (unsigned i = 0; i != N-1; ++i)
167 BadCharSkip[(uint8_t)Str[i]] = N-1-i;
168
169 do {
170 uint8_t Last = Start[N - 1];
171 if (LLVM_UNLIKELY(Last == (uint8_t)Needle[N - 1])__builtin_expect((bool)(Last == (uint8_t)Needle[N - 1]), false
))
172 if (std::memcmp(Start, Needle, N - 1) == 0)
173 return Start - Data;
174
175 // Otherwise skip the appropriate number of bytes.
176 Start += BadCharSkip[Last];
177 } while (Start < Stop);
178
179 return npos;
180}
181
182size_t StringRef::find_lower(StringRef Str, size_t From) const {
183 StringRef This = substr(From);
184 while (This.size() >= Str.size()) {
185 if (This.startswith_lower(Str))
186 return From;
187 This = This.drop_front();
188 ++From;
189 }
190 return npos;
191}
192
193size_t StringRef::rfind_lower(char C, size_t From) const {
194 From = std::min(From, Length);
195 size_t i = From;
196 while (i != 0) {
197 --i;
198 if (toLower(Data[i]) == toLower(C))
199 return i;
200 }
201 return npos;
202}
203
204/// rfind - Search for the last string \arg Str in the string.
205///
206/// \return - The index of the last occurrence of \arg Str, or npos if not
207/// found.
208size_t StringRef::rfind(StringRef Str) const {
209 size_t N = Str.size();
210 if (N > Length)
211 return npos;
212 for (size_t i = Length - N + 1, e = 0; i != e;) {
213 --i;
214 if (substr(i, N).equals(Str))
215 return i;
216 }
217 return npos;
218}
219
220size_t StringRef::rfind_lower(StringRef Str) const {
221 size_t N = Str.size();
222 if (N > Length)
223 return npos;
224 for (size_t i = Length - N + 1, e = 0; i != e;) {
225 --i;
226 if (substr(i, N).equals_lower(Str))
227 return i;
228 }
229 return npos;
230}
231
232/// find_first_of - Find the first character in the string that is in \arg
233/// Chars, or npos if not found.
234///
235/// Note: O(size() + Chars.size())
236StringRef::size_type StringRef::find_first_of(StringRef Chars,
237 size_t From) const {
238 std::bitset<1 << CHAR_BIT8> CharBits;
239 for (size_type i = 0; i != Chars.size(); ++i)
240 CharBits.set((unsigned char)Chars[i]);
241
242 for (size_type i = std::min(From, Length), e = Length; i != e; ++i)
243 if (CharBits.test((unsigned char)Data[i]))
244 return i;
245 return npos;
246}
247
248/// find_first_not_of - Find the first character in the string that is not
249/// \arg C or npos if not found.
250StringRef::size_type StringRef::find_first_not_of(char C, size_t From) const {
251 for (size_type i = std::min(From, Length), e = Length; i != e; ++i)
252 if (Data[i] != C)
253 return i;
254 return npos;
255}
256
257/// find_first_not_of - Find the first character in the string that is not
258/// in the string \arg Chars, or npos if not found.
259///
260/// Note: O(size() + Chars.size())
261StringRef::size_type StringRef::find_first_not_of(StringRef Chars,
262 size_t From) const {
263 std::bitset<1 << CHAR_BIT8> CharBits;
264 for (size_type i = 0; i != Chars.size(); ++i)
265 CharBits.set((unsigned char)Chars[i]);
266
267 for (size_type i = std::min(From, Length), e = Length; i != e; ++i)
268 if (!CharBits.test((unsigned char)Data[i]))
269 return i;
270 return npos;
271}
272
273/// find_last_of - Find the last character in the string that is in \arg C,
274/// or npos if not found.
275///
276/// Note: O(size() + Chars.size())
277StringRef::size_type StringRef::find_last_of(StringRef Chars,
278 size_t From) const {
279 std::bitset<1 << CHAR_BIT8> CharBits;
280 for (size_type i = 0; i != Chars.size(); ++i)
281 CharBits.set((unsigned char)Chars[i]);
282
283 for (size_type i = std::min(From, Length) - 1, e = -1; i != e; --i)
284 if (CharBits.test((unsigned char)Data[i]))
285 return i;
286 return npos;
287}
288
289/// find_last_not_of - Find the last character in the string that is not
290/// \arg C, or npos if not found.
291StringRef::size_type StringRef::find_last_not_of(char C, size_t From) const {
292 for (size_type i = std::min(From, Length) - 1, e = -1; i != e; --i)
293 if (Data[i] != C)
294 return i;
295 return npos;
296}
297
298/// find_last_not_of - Find the last character in the string that is not in
299/// \arg Chars, or npos if not found.
300///
301/// Note: O(size() + Chars.size())
302StringRef::size_type StringRef::find_last_not_of(StringRef Chars,
303 size_t From) const {
304 std::bitset<1 << CHAR_BIT8> CharBits;
305 for (size_type i = 0, e = Chars.size(); i != e; ++i)
306 CharBits.set((unsigned char)Chars[i]);
307
308 for (size_type i = std::min(From, Length) - 1, e = -1; i != e; --i)
309 if (!CharBits.test((unsigned char)Data[i]))
310 return i;
311 return npos;
312}
313
314void StringRef::split(SmallVectorImpl<StringRef> &A,
315 StringRef Separator, int MaxSplit,
316 bool KeepEmpty) const {
317 StringRef S = *this;
318
319 // Count down from MaxSplit. When MaxSplit is -1, this will just split
320 // "forever". This doesn't support splitting more than 2^31 times
321 // intentionally; if we ever want that we can make MaxSplit a 64-bit integer
322 // but that seems unlikely to be useful.
323 while (MaxSplit-- != 0) {
324 size_t Idx = S.find(Separator);
325 if (Idx == npos)
326 break;
327
328 // Push this split.
329 if (KeepEmpty || Idx > 0)
330 A.push_back(S.slice(0, Idx));
331
332 // Jump forward.
333 S = S.slice(Idx + Separator.size(), npos);
334 }
335
336 // Push the tail.
337 if (KeepEmpty || !S.empty())
338 A.push_back(S);
339}
340
341void StringRef::split(SmallVectorImpl<StringRef> &A, char Separator,
342 int MaxSplit, bool KeepEmpty) const {
343 StringRef S = *this;
344
345 // Count down from MaxSplit. When MaxSplit is -1, this will just split
346 // "forever". This doesn't support splitting more than 2^31 times
347 // intentionally; if we ever want that we can make MaxSplit a 64-bit integer
348 // but that seems unlikely to be useful.
349 while (MaxSplit-- != 0) {
350 size_t Idx = S.find(Separator);
351 if (Idx == npos)
352 break;
353
354 // Push this split.
355 if (KeepEmpty || Idx > 0)
356 A.push_back(S.slice(0, Idx));
357
358 // Jump forward.
359 S = S.slice(Idx + 1, npos);
360 }
361
362 // Push the tail.
363 if (KeepEmpty || !S.empty())
364 A.push_back(S);
365}
366
367//===----------------------------------------------------------------------===//
368// Helpful Algorithms
369//===----------------------------------------------------------------------===//
370
371/// count - Return the number of non-overlapped occurrences of \arg Str in
372/// the string.
373size_t StringRef::count(StringRef Str) const {
374 size_t Count = 0;
375 size_t N = Str.size();
376 if (N > Length)
377 return 0;
378 for (size_t i = 0, e = Length - N + 1; i != e; ++i)
379 if (substr(i, N).equals(Str))
380 ++Count;
381 return Count;
382}
383
384static unsigned GetAutoSenseRadix(StringRef &Str) {
385 if (Str.empty())
386 return 10;
387
388 if (Str.startswith("0x") || Str.startswith("0X")) {
389 Str = Str.substr(2);
390 return 16;
391 }
392
393 if (Str.startswith("0b") || Str.startswith("0B")) {
394 Str = Str.substr(2);
395 return 2;
396 }
397
398 if (Str.startswith("0o")) {
399 Str = Str.substr(2);
400 return 8;
401 }
402
403 if (Str[0] == '0' && Str.size() > 1 && isDigit(Str[1])) {
404 Str = Str.substr(1);
405 return 8;
406 }
407
408 return 10;
409}
410
411bool llvm::consumeUnsignedInteger(StringRef &Str, unsigned Radix,
412 unsigned long long &Result) {
413 // Autosense radix if not specified.
414 if (Radix == 0)
415 Radix = GetAutoSenseRadix(Str);
416
417 // Empty strings (after the radix autosense) are invalid.
418 if (Str.empty()) return true;
419
420 // Parse all the bytes of the string given this radix. Watch for overflow.
421 StringRef Str2 = Str;
422 Result = 0;
423 while (!Str2.empty()) {
424 unsigned CharVal;
425 if (Str2[0] >= '0' && Str2[0] <= '9')
426 CharVal = Str2[0] - '0';
427 else if (Str2[0] >= 'a' && Str2[0] <= 'z')
428 CharVal = Str2[0] - 'a' + 10;
429 else if (Str2[0] >= 'A' && Str2[0] <= 'Z')
430 CharVal = Str2[0] - 'A' + 10;
431 else
432 break;
433
434 // If the parsed value is larger than the integer radix, we cannot
435 // consume any more characters.
436 if (CharVal >= Radix)
437 break;
438
439 // Add in this character.
440 unsigned long long PrevResult = Result;
441 Result = Result * Radix + CharVal;
442
443 // Check for overflow by shifting back and seeing if bits were lost.
444 if (Result / Radix < PrevResult)
445 return true;
446
447 Str2 = Str2.substr(1);
448 }
449
450 // We consider the operation a failure if no characters were consumed
451 // successfully.
452 if (Str.size() == Str2.size())
453 return true;
454
455 Str = Str2;
456 return false;
457}
458
459bool llvm::consumeSignedInteger(StringRef &Str, unsigned Radix,
460 long long &Result) {
461 unsigned long long ULLVal;
462
463 // Handle positive strings first.
464 if (Str.empty() || Str.front() != '-') {
465 if (consumeUnsignedInteger(Str, Radix, ULLVal) ||
466 // Check for value so large it overflows a signed value.
467 (long long)ULLVal < 0)
468 return true;
469 Result = ULLVal;
470 return false;
471 }
472
473 // Get the positive part of the value.
474 StringRef Str2 = Str.drop_front(1);
475 if (consumeUnsignedInteger(Str2, Radix, ULLVal) ||
476 // Reject values so large they'd overflow as negative signed, but allow
477 // "-0". This negates the unsigned so that the negative isn't undefined
478 // on signed overflow.
479 (long long)-ULLVal > 0)
480 return true;
481
482 Str = Str2;
483 Result = -ULLVal;
484 return false;
485}
486
487/// GetAsUnsignedInteger - Workhorse method that converts a integer character
488/// sequence of radix up to 36 to an unsigned long long value.
489bool llvm::getAsUnsignedInteger(StringRef Str, unsigned Radix,
490 unsigned long long &Result) {
491 if (consumeUnsignedInteger(Str, Radix, Result))
492 return true;
493
494 // For getAsUnsignedInteger, we require the whole string to be consumed or
495 // else we consider it a failure.
496 return !Str.empty();
497}
498
499bool llvm::getAsSignedInteger(StringRef Str, unsigned Radix,
500 long long &Result) {
501 if (consumeSignedInteger(Str, Radix, Result))
502 return true;
503
504 // For getAsSignedInteger, we require the whole string to be consumed or else
505 // we consider it a failure.
506 return !Str.empty();
507}
508
509bool StringRef::getAsInteger(unsigned Radix, APInt &Result) const {
510 StringRef Str = *this;
511
512 // Autosense radix if not specified.
513 if (Radix == 0)
514 Radix = GetAutoSenseRadix(Str);
515
516 assert(Radix > 1 && Radix <= 36)((Radix > 1 && Radix <= 36) ? static_cast<void
> (0) : __assert_fail ("Radix > 1 && Radix <= 36"
, "/build/llvm-toolchain-snapshot-8~svn350071/lib/Support/StringRef.cpp"
, 516, __PRETTY_FUNCTION__));
517
518 // Empty strings (after the radix autosense) are invalid.
519 if (Str.empty()) return true;
520
521 // Skip leading zeroes. This can be a significant improvement if
522 // it means we don't need > 64 bits.
523 while (!Str.empty() && Str.front() == '0')
524 Str = Str.substr(1);
525
526 // If it was nothing but zeroes....
527 if (Str.empty()) {
528 Result = APInt(64, 0);
529 return false;
530 }
531
532 // (Over-)estimate the required number of bits.
533 unsigned Log2Radix = 0;
534 while ((1U << Log2Radix) < Radix) Log2Radix++;
535 bool IsPowerOf2Radix = ((1U << Log2Radix) == Radix);
536
537 unsigned BitWidth = Log2Radix * Str.size();
538 if (BitWidth < Result.getBitWidth())
539 BitWidth = Result.getBitWidth(); // don't shrink the result
540 else if (BitWidth > Result.getBitWidth())
541 Result = Result.zext(BitWidth);
542
543 APInt RadixAP, CharAP; // unused unless !IsPowerOf2Radix
544 if (!IsPowerOf2Radix) {
545 // These must have the same bit-width as Result.
546 RadixAP = APInt(BitWidth, Radix);
547 CharAP = APInt(BitWidth, 0);
548 }
549
550 // Parse all the bytes of the string given this radix.
551 Result = 0;
552 while (!Str.empty()) {
553 unsigned CharVal;
554 if (Str[0] >= '0' && Str[0] <= '9')
555 CharVal = Str[0]-'0';
556 else if (Str[0] >= 'a' && Str[0] <= 'z')
557 CharVal = Str[0]-'a'+10;
558 else if (Str[0] >= 'A' && Str[0] <= 'Z')
559 CharVal = Str[0]-'A'+10;
560 else
561 return true;
562
563 // If the parsed value is larger than the integer radix, the string is
564 // invalid.
565 if (CharVal >= Radix)
566 return true;
567
568 // Add in this character.
569 if (IsPowerOf2Radix) {
570 Result <<= Log2Radix;
571 Result |= CharVal;
572 } else {
573 Result *= RadixAP;
574 CharAP = CharVal;
575 Result += CharAP;
576 }
577
578 Str = Str.substr(1);
579 }
580
581 return false;
582}
583
584bool StringRef::getAsDouble(double &Result, bool AllowInexact) const {
585 APFloat F(0.0);
586 APFloat::opStatus Status =
587 F.convertFromString(*this, APFloat::rmNearestTiesToEven);
588 if (Status != APFloat::opOK) {
589 if (!AllowInexact || !(Status & APFloat::opInexact))
590 return true;
591 }
592
593 Result = F.convertToDouble();
594 return false;
595}
596
597// Implementation of StringRef hashing.
598hash_code llvm::hash_value(StringRef S) {
599 return hash_combine_range(S.begin(), S.end());
600}
←

/build/llvm-toolchain-snapshot-8~svn350071/include/llvm/ADT/edit_distance.h

1//===-- llvm/ADT/edit_distance.h - Array edit distance function --- C++ -*-===//
2//
3// The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9//
10// This file defines a Levenshtein distance function that works for any two
11// sequences, with each element of each sequence being analogous to a character
12// in a string.
13//
14//===----------------------------------------------------------------------===//
15
16#ifndef LLVM_ADT_EDIT_DISTANCE_H
17#define LLVM_ADT_EDIT_DISTANCE_H
18
19#include "llvm/ADT/ArrayRef.h"
20#include <algorithm>
21#include <memory>
22
23namespace llvm {
24
25/// Determine the edit distance between two sequences.
26///
27/// \param FromArray the first sequence to compare.
28///
29/// \param ToArray the second sequence to compare.
30///
31/// \param AllowReplacements whether to allow element replacements (change one
32/// element into another) as a single operation, rather than as two operations
33/// (an insertion and a removal).
34///
35/// \param MaxEditDistance If non-zero, the maximum edit distance that this
36/// routine is allowed to compute. If the edit distance will exceed that
37/// maximum, returns \c MaxEditDistance+1.
38///
39/// \returns the minimum number of element insertions, removals, or (if
40/// \p AllowReplacements is \c true) replacements needed to transform one of
41/// the given sequences into the other. If zero, the sequences are identical.
42template<typename T>
43unsigned ComputeEditDistance(ArrayRef<T> FromArray, ArrayRef<T> ToArray,
44 bool AllowReplacements = true,
45 unsigned MaxEditDistance = 0) {
46 // The algorithm implemented below is the "classic"
47 // dynamic-programming algorithm for computing the Levenshtein
48 // distance, which is described here:
49 //
50 // http://en.wikipedia.org/wiki/Levenshtein_distance
51 //
52 // Although the algorithm is typically described using an m x n
53 // array, only one row plus one element are used at a time, so this
54 // implementation just keeps one vector for the row. To update one entry,
55 // only the entries to the left, top, and top-left are needed. The left
56 // entry is in Row[x-1], the top entry is what's in Row[x] from the last
57 // iteration, and the top-left entry is stored in Previous.
58 typename ArrayRef<T>::size_type m = FromArray.size();
59 typename ArrayRef<T>::size_type n = ToArray.size();
60
61 const unsigned SmallBufferSize = 64;
62 unsigned SmallBuffer[SmallBufferSize];
63 std::unique_ptr<unsigned[]> Allocated;
64 unsigned *Row = SmallBuffer;
65 if (n + 1 > SmallBufferSize) {
2
←
Assuming the condition is false
→
3
←
Taking false branch
→
66 Row = new unsigned[n + 1];
67 Allocated.reset(Row);
68 }
69
70 for (unsigned i = 1; i <= n; ++i)
4
←
Assuming 'i' is > 'n'
→
5
←
Loop condition is false. Execution continues on line 73
→
71 Row[i] = i;
72
73 for (typename ArrayRef<T>::size_type y = 1; y <= m; ++y) {
6
←
Assuming 'y' is > 'm'
→
7
←
Loop condition is false. Execution continues on line 97
→
74 Row[0] = y;
75 unsigned BestThisRow = Row[0];
76
77 unsigned Previous = y - 1;
78 for (typename ArrayRef<T>::size_type x = 1; x <= n; ++x) {
79 int OldRow = Row[x];
80 if (AllowReplacements) {
81 Row[x] = std::min(
82 Previous + (FromArray[y-1] == ToArray[x-1] ? 0u : 1u),
83 std::min(Row[x-1], Row[x])+1);
84 }
85 else {
86 if (FromArray[y-1] == ToArray[x-1]) Row[x] = Previous;
87 else Row[x] = std::min(Row[x-1], Row[x]) + 1;
88 }
89 Previous = OldRow;
90 BestThisRow = std::min(BestThisRow, Row[x]);
91 }
92
93 if (MaxEditDistance && BestThisRow > MaxEditDistance)
94 return MaxEditDistance + 1;
95 }
96
97 unsigned Result = Row[n];
8
←
Assigned value is garbage or undefined
98 return Result;
99}
100
101} // End llvm namespace
102
103#endif