1 //===- StringRef.h - Constant String Reference Wrapper ----------*- 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 #ifndef LLVM_ADT_STRINGREF_H 11 #define LLVM_ADT_STRINGREF_H 12 13 #include "llvm/ADT/STLExtras.h" 14 #include "llvm/ADT/iterator_range.h" 15 #include "llvm/Support/Compiler.h" 16 #include <algorithm> 17 #include <cassert> 18 #include <cstddef> 19 #include <cstring> 20 #include <limits> 21 #include <string> 22 #include <type_traits> 23 #include <utility> 24 25 namespace llvm { 26 27 class APInt; 28 class hash_code; 29 template <typename T> class SmallVectorImpl; 30 class StringRef; 31 32 /// Helper functions for StringRef::getAsInteger. 33 bool getAsUnsignedInteger(StringRef Str, unsigned Radix, 34 unsigned long long &Result); 35 36 bool getAsSignedInteger(StringRef Str, unsigned Radix, long long &Result); 37 38 bool consumeUnsignedInteger(StringRef &Str, unsigned Radix, 39 unsigned long long &Result); 40 bool consumeSignedInteger(StringRef &Str, unsigned Radix, long long &Result); 41 42 /// StringRef - Represent a constant reference to a string, i.e. a character 43 /// array and a length, which need not be null terminated. 44 /// 45 /// This class does not own the string data, it is expected to be used in 46 /// situations where the character data resides in some other buffer, whose 47 /// lifetime extends past that of the StringRef. For this reason, it is not in 48 /// general safe to store a StringRef. 49 class StringRef { 50 public: 51 static const size_t npos = ~size_t(0); 52 53 using iterator = const char *; 54 using const_iterator = const char *; 55 using size_type = size_t; 56 57 private: 58 /// The start of the string, in an external buffer. 59 const char *Data = nullptr; 60 61 /// The length of the string. 62 size_t Length = 0; 63 64 // Workaround memcmp issue with null pointers (undefined behavior) 65 // by providing a specialized version 66 LLVM_ATTRIBUTE_ALWAYS_INLINE compareMemory(const char * Lhs,const char * Rhs,size_t Length)67 static int compareMemory(const char *Lhs, const char *Rhs, size_t Length) { 68 if (Length == 0) { return 0; } 69 return ::memcmp(Lhs,Rhs,Length); 70 } 71 72 public: 73 /// @name Constructors 74 /// @{ 75 76 /// Construct an empty string ref. 77 /*implicit*/ StringRef() = default; 78 79 /// Disable conversion from nullptr. This prevents things like 80 /// if (S == nullptr) 81 StringRef(std::nullptr_t) = delete; 82 83 /// Construct a string ref from a cstring. 84 LLVM_ATTRIBUTE_ALWAYS_INLINE StringRef(const char * Str)85 /*implicit*/ StringRef(const char *Str) 86 : Data(Str), Length(Str ? ::strlen(Str) : 0) {} 87 88 /// Construct a string ref from a pointer and length. 89 LLVM_ATTRIBUTE_ALWAYS_INLINE StringRef(const char * data,size_t length)90 /*implicit*/ constexpr StringRef(const char *data, size_t length) 91 : Data(data), Length(length) {} 92 93 /// Construct a string ref from an std::string. 94 LLVM_ATTRIBUTE_ALWAYS_INLINE StringRef(const std::string & Str)95 /*implicit*/ StringRef(const std::string &Str) 96 : Data(Str.data()), Length(Str.length()) {} 97 withNullAsEmpty(const char * data)98 static StringRef withNullAsEmpty(const char *data) { 99 return StringRef(data ? data : ""); 100 } 101 102 /// @} 103 /// @name Iterators 104 /// @{ 105 begin()106 iterator begin() const { return Data; } 107 end()108 iterator end() const { return Data + Length; } 109 bytes_begin()110 const unsigned char *bytes_begin() const { 111 return reinterpret_cast<const unsigned char *>(begin()); 112 } bytes_end()113 const unsigned char *bytes_end() const { 114 return reinterpret_cast<const unsigned char *>(end()); 115 } bytes()116 iterator_range<const unsigned char *> bytes() const { 117 return make_range(bytes_begin(), bytes_end()); 118 } 119 120 /// @} 121 /// @name String Operations 122 /// @{ 123 124 /// data - Get a pointer to the start of the string (which may not be null 125 /// terminated). 126 LLVM_NODISCARD 127 LLVM_ATTRIBUTE_ALWAYS_INLINE data()128 const char *data() const { return Data; } 129 130 /// empty - Check if the string is empty. 131 LLVM_NODISCARD 132 LLVM_ATTRIBUTE_ALWAYS_INLINE empty()133 bool empty() const { return Length == 0; } 134 135 /// size - Get the string size. 136 LLVM_NODISCARD 137 LLVM_ATTRIBUTE_ALWAYS_INLINE size()138 size_t size() const { return Length; } 139 140 /// front - Get the first character in the string. 141 LLVM_NODISCARD front()142 char front() const { 143 assert(!empty()); 144 return Data[0]; 145 } 146 147 /// back - Get the last character in the string. 148 LLVM_NODISCARD back()149 char back() const { 150 assert(!empty()); 151 return Data[Length-1]; 152 } 153 154 // copy - Allocate copy in Allocator and return StringRef to it. 155 template <typename Allocator> copy(Allocator & A)156 LLVM_NODISCARD StringRef copy(Allocator &A) const { 157 // Don't request a length 0 copy from the allocator. 158 if (empty()) 159 return StringRef(); 160 char *S = A.template Allocate<char>(Length); 161 std::copy(begin(), end(), S); 162 return StringRef(S, Length); 163 } 164 165 /// equals - Check for string equality, this is more efficient than 166 /// compare() when the relative ordering of inequal strings isn't needed. 167 LLVM_NODISCARD 168 LLVM_ATTRIBUTE_ALWAYS_INLINE equals(StringRef RHS)169 bool equals(StringRef RHS) const { 170 return (Length == RHS.Length && 171 compareMemory(Data, RHS.Data, RHS.Length) == 0); 172 } 173 174 /// equals_lower - Check for string equality, ignoring case. 175 LLVM_NODISCARD equals_lower(StringRef RHS)176 bool equals_lower(StringRef RHS) const { 177 return Length == RHS.Length && compare_lower(RHS) == 0; 178 } 179 180 /// compare - Compare two strings; the result is -1, 0, or 1 if this string 181 /// is lexicographically less than, equal to, or greater than the \p RHS. 182 LLVM_NODISCARD 183 LLVM_ATTRIBUTE_ALWAYS_INLINE compare(StringRef RHS)184 int compare(StringRef RHS) const { 185 // Check the prefix for a mismatch. 186 if (int Res = compareMemory(Data, RHS.Data, std::min(Length, RHS.Length))) 187 return Res < 0 ? -1 : 1; 188 189 // Otherwise the prefixes match, so we only need to check the lengths. 190 if (Length == RHS.Length) 191 return 0; 192 return Length < RHS.Length ? -1 : 1; 193 } 194 195 /// compare_lower - Compare two strings, ignoring case. 196 LLVM_NODISCARD 197 int compare_lower(StringRef RHS) const; 198 199 /// compare_numeric - Compare two strings, treating sequences of digits as 200 /// numbers. 201 LLVM_NODISCARD 202 int compare_numeric(StringRef RHS) const; 203 204 /// Determine the edit distance between this string and another 205 /// string. 206 /// 207 /// \param Other the string to compare this string against. 208 /// 209 /// \param AllowReplacements whether to allow character 210 /// replacements (change one character into another) as a single 211 /// operation, rather than as two operations (an insertion and a 212 /// removal). 213 /// 214 /// \param MaxEditDistance If non-zero, the maximum edit distance that 215 /// this routine is allowed to compute. If the edit distance will exceed 216 /// that maximum, returns \c MaxEditDistance+1. 217 /// 218 /// \returns the minimum number of character insertions, removals, 219 /// or (if \p AllowReplacements is \c true) replacements needed to 220 /// transform one of the given strings into the other. If zero, 221 /// the strings are identical. 222 LLVM_NODISCARD 223 unsigned edit_distance(StringRef Other, bool AllowReplacements = true, 224 unsigned MaxEditDistance = 0) const; 225 226 /// str - Get the contents as an std::string. 227 LLVM_NODISCARD str()228 std::string str() const { 229 if (!Data) return std::string(); 230 return std::string(Data, Length); 231 } 232 233 /// @} 234 /// @name Operator Overloads 235 /// @{ 236 237 LLVM_NODISCARD 238 char operator[](size_t Index) const { 239 assert(Index < Length && "Invalid index!"); 240 return Data[Index]; 241 } 242 243 /// Disallow accidental assignment from a temporary std::string. 244 /// 245 /// The declaration here is extra complicated so that `stringRef = {}` 246 /// and `stringRef = "abc"` continue to select the move assignment operator. 247 template <typename T> 248 typename std::enable_if<std::is_same<T, std::string>::value, 249 StringRef>::type & 250 operator=(T &&Str) = delete; 251 252 /// @} 253 /// @name Type Conversions 254 /// @{ 255 string()256 operator std::string() const { 257 return str(); 258 } 259 260 /// @} 261 /// @name String Predicates 262 /// @{ 263 264 /// Check if this string starts with the given \p Prefix. 265 LLVM_NODISCARD 266 LLVM_ATTRIBUTE_ALWAYS_INLINE startswith(StringRef Prefix)267 bool startswith(StringRef Prefix) const { 268 return Length >= Prefix.Length && 269 compareMemory(Data, Prefix.Data, Prefix.Length) == 0; 270 } 271 272 /// Check if this string starts with the given \p Prefix, ignoring case. 273 LLVM_NODISCARD 274 bool startswith_lower(StringRef Prefix) const; 275 276 /// Check if this string ends with the given \p Suffix. 277 LLVM_NODISCARD 278 LLVM_ATTRIBUTE_ALWAYS_INLINE endswith(StringRef Suffix)279 bool endswith(StringRef Suffix) const { 280 return Length >= Suffix.Length && 281 compareMemory(end() - Suffix.Length, Suffix.Data, Suffix.Length) == 0; 282 } 283 284 /// Check if this string ends with the given \p Suffix, ignoring case. 285 LLVM_NODISCARD 286 bool endswith_lower(StringRef Suffix) const; 287 288 /// @} 289 /// @name String Searching 290 /// @{ 291 292 /// Search for the first character \p C in the string. 293 /// 294 /// \returns The index of the first occurrence of \p C, or npos if not 295 /// found. 296 LLVM_NODISCARD 297 LLVM_ATTRIBUTE_ALWAYS_INLINE 298 size_t find(char C, size_t From = 0) const { 299 size_t FindBegin = std::min(From, Length); 300 if (FindBegin < Length) { // Avoid calling memchr with nullptr. 301 // Just forward to memchr, which is faster than a hand-rolled loop. 302 if (const void *P = ::memchr(Data + FindBegin, C, Length - FindBegin)) 303 return static_cast<const char *>(P) - Data; 304 } 305 return npos; 306 } 307 308 /// Search for the first character \p C in the string, ignoring case. 309 /// 310 /// \returns The index of the first occurrence of \p C, or npos if not 311 /// found. 312 LLVM_NODISCARD 313 size_t find_lower(char C, size_t From = 0) const; 314 315 /// Search for the first character satisfying the predicate \p F 316 /// 317 /// \returns The index of the first character satisfying \p F starting from 318 /// \p From, or npos if not found. 319 LLVM_NODISCARD 320 LLVM_ATTRIBUTE_ALWAYS_INLINE 321 size_t find_if(function_ref<bool(char)> F, size_t From = 0) const { 322 StringRef S = drop_front(From); 323 while (!S.empty()) { 324 if (F(S.front())) 325 return size() - S.size(); 326 S = S.drop_front(); 327 } 328 return npos; 329 } 330 331 /// Search for the first character not satisfying the predicate \p F 332 /// 333 /// \returns The index of the first character not satisfying \p F starting 334 /// from \p From, or npos if not found. 335 LLVM_NODISCARD 336 LLVM_ATTRIBUTE_ALWAYS_INLINE 337 size_t find_if_not(function_ref<bool(char)> F, size_t From = 0) const { 338 return find_if([F](char c) { return !F(c); }, From); 339 } 340 341 /// Search for the first string \p Str in the string. 342 /// 343 /// \returns The index of the first occurrence of \p Str, or npos if not 344 /// found. 345 LLVM_NODISCARD 346 size_t find(StringRef Str, size_t From = 0) const; 347 348 /// Search for the first string \p Str in the string, ignoring case. 349 /// 350 /// \returns The index of the first occurrence of \p Str, or npos if not 351 /// found. 352 LLVM_NODISCARD 353 size_t find_lower(StringRef Str, size_t From = 0) const; 354 355 /// Search for the last character \p C in the string. 356 /// 357 /// \returns The index of the last occurrence of \p C, or npos if not 358 /// found. 359 LLVM_NODISCARD 360 size_t rfind(char C, size_t From = npos) const { 361 From = std::min(From, Length); 362 size_t i = From; 363 while (i != 0) { 364 --i; 365 if (Data[i] == C) 366 return i; 367 } 368 return npos; 369 } 370 371 /// Search for the last character \p C in the string, ignoring case. 372 /// 373 /// \returns The index of the last occurrence of \p C, or npos if not 374 /// found. 375 LLVM_NODISCARD 376 size_t rfind_lower(char C, size_t From = npos) const; 377 378 /// Search for the last string \p Str in the string. 379 /// 380 /// \returns The index of the last occurrence of \p Str, or npos if not 381 /// found. 382 LLVM_NODISCARD 383 size_t rfind(StringRef Str) const; 384 385 /// Search for the last string \p Str in the string, ignoring case. 386 /// 387 /// \returns The index of the last occurrence of \p Str, or npos if not 388 /// found. 389 LLVM_NODISCARD 390 size_t rfind_lower(StringRef Str) const; 391 392 /// Find the first character in the string that is \p C, or npos if not 393 /// found. Same as find. 394 LLVM_NODISCARD 395 size_t find_first_of(char C, size_t From = 0) const { 396 return find(C, From); 397 } 398 399 /// Find the first character in the string that is in \p Chars, or npos if 400 /// not found. 401 /// 402 /// Complexity: O(size() + Chars.size()) 403 LLVM_NODISCARD 404 size_t find_first_of(StringRef Chars, size_t From = 0) const; 405 406 /// Find the first character in the string that is not \p C or npos if not 407 /// found. 408 LLVM_NODISCARD 409 size_t find_first_not_of(char C, size_t From = 0) const; 410 411 /// Find the first character in the string that is not in the string 412 /// \p Chars, or npos if not found. 413 /// 414 /// Complexity: O(size() + Chars.size()) 415 LLVM_NODISCARD 416 size_t find_first_not_of(StringRef Chars, size_t From = 0) const; 417 418 /// Find the last character in the string that is \p C, or npos if not 419 /// found. 420 LLVM_NODISCARD 421 size_t find_last_of(char C, size_t From = npos) const { 422 return rfind(C, From); 423 } 424 425 /// Find the last character in the string that is in \p C, or npos if not 426 /// found. 427 /// 428 /// Complexity: O(size() + Chars.size()) 429 LLVM_NODISCARD 430 size_t find_last_of(StringRef Chars, size_t From = npos) const; 431 432 /// Find the last character in the string that is not \p C, or npos if not 433 /// found. 434 LLVM_NODISCARD 435 size_t find_last_not_of(char C, size_t From = npos) const; 436 437 /// Find the last character in the string that is not in \p Chars, or 438 /// npos if not found. 439 /// 440 /// Complexity: O(size() + Chars.size()) 441 LLVM_NODISCARD 442 size_t find_last_not_of(StringRef Chars, size_t From = npos) const; 443 444 /// Return true if the given string is a substring of *this, and false 445 /// otherwise. 446 LLVM_NODISCARD 447 LLVM_ATTRIBUTE_ALWAYS_INLINE contains(StringRef Other)448 bool contains(StringRef Other) const { return find(Other) != npos; } 449 450 /// Return true if the given character is contained in *this, and false 451 /// otherwise. 452 LLVM_NODISCARD 453 LLVM_ATTRIBUTE_ALWAYS_INLINE contains(char C)454 bool contains(char C) const { return find_first_of(C) != npos; } 455 456 /// Return true if the given string is a substring of *this, and false 457 /// otherwise. 458 LLVM_NODISCARD 459 LLVM_ATTRIBUTE_ALWAYS_INLINE contains_lower(StringRef Other)460 bool contains_lower(StringRef Other) const { 461 return find_lower(Other) != npos; 462 } 463 464 /// Return true if the given character is contained in *this, and false 465 /// otherwise. 466 LLVM_NODISCARD 467 LLVM_ATTRIBUTE_ALWAYS_INLINE contains_lower(char C)468 bool contains_lower(char C) const { return find_lower(C) != npos; } 469 470 /// @} 471 /// @name Helpful Algorithms 472 /// @{ 473 474 /// Return the number of occurrences of \p C in the string. 475 LLVM_NODISCARD count(char C)476 size_t count(char C) const { 477 size_t Count = 0; 478 for (size_t i = 0, e = Length; i != e; ++i) 479 if (Data[i] == C) 480 ++Count; 481 return Count; 482 } 483 484 /// Return the number of non-overlapped occurrences of \p Str in 485 /// the string. 486 size_t count(StringRef Str) const; 487 488 /// Parse the current string as an integer of the specified radix. If 489 /// \p Radix is specified as zero, this does radix autosensing using 490 /// extended C rules: 0 is octal, 0x is hex, 0b is binary. 491 /// 492 /// If the string is invalid or if only a subset of the string is valid, 493 /// this returns true to signify the error. The string is considered 494 /// erroneous if empty or if it overflows T. 495 template <typename T> 496 typename std::enable_if<std::numeric_limits<T>::is_signed, bool>::type getAsInteger(unsigned Radix,T & Result)497 getAsInteger(unsigned Radix, T &Result) const { 498 long long LLVal; 499 if (getAsSignedInteger(*this, Radix, LLVal) || 500 static_cast<T>(LLVal) != LLVal) 501 return true; 502 Result = LLVal; 503 return false; 504 } 505 506 template <typename T> 507 typename std::enable_if<!std::numeric_limits<T>::is_signed, bool>::type getAsInteger(unsigned Radix,T & Result)508 getAsInteger(unsigned Radix, T &Result) const { 509 unsigned long long ULLVal; 510 // The additional cast to unsigned long long is required to avoid the 511 // Visual C++ warning C4805: '!=' : unsafe mix of type 'bool' and type 512 // 'unsigned __int64' when instantiating getAsInteger with T = bool. 513 if (getAsUnsignedInteger(*this, Radix, ULLVal) || 514 static_cast<unsigned long long>(static_cast<T>(ULLVal)) != ULLVal) 515 return true; 516 Result = ULLVal; 517 return false; 518 } 519 520 /// Parse the current string as an integer of the specified radix. If 521 /// \p Radix is specified as zero, this does radix autosensing using 522 /// extended C rules: 0 is octal, 0x is hex, 0b is binary. 523 /// 524 /// If the string does not begin with a number of the specified radix, 525 /// this returns true to signify the error. The string is considered 526 /// erroneous if empty or if it overflows T. 527 /// The portion of the string representing the discovered numeric value 528 /// is removed from the beginning of the string. 529 template <typename T> 530 typename std::enable_if<std::numeric_limits<T>::is_signed, bool>::type consumeInteger(unsigned Radix,T & Result)531 consumeInteger(unsigned Radix, T &Result) { 532 long long LLVal; 533 if (consumeSignedInteger(*this, Radix, LLVal) || 534 static_cast<long long>(static_cast<T>(LLVal)) != LLVal) 535 return true; 536 Result = LLVal; 537 return false; 538 } 539 540 template <typename T> 541 typename std::enable_if<!std::numeric_limits<T>::is_signed, bool>::type consumeInteger(unsigned Radix,T & Result)542 consumeInteger(unsigned Radix, T &Result) { 543 unsigned long long ULLVal; 544 if (consumeUnsignedInteger(*this, Radix, ULLVal) || 545 static_cast<unsigned long long>(static_cast<T>(ULLVal)) != ULLVal) 546 return true; 547 Result = ULLVal; 548 return false; 549 } 550 551 /// Parse the current string as an integer of the specified \p Radix, or of 552 /// an autosensed radix if the \p Radix given is 0. The current value in 553 /// \p Result is discarded, and the storage is changed to be wide enough to 554 /// store the parsed integer. 555 /// 556 /// \returns true if the string does not solely consist of a valid 557 /// non-empty number in the appropriate base. 558 /// 559 /// APInt::fromString is superficially similar but assumes the 560 /// string is well-formed in the given radix. 561 bool getAsInteger(unsigned Radix, APInt &Result) const; 562 563 /// Parse the current string as an IEEE double-precision floating 564 /// point value. The string must be a well-formed double. 565 /// 566 /// If \p AllowInexact is false, the function will fail if the string 567 /// cannot be represented exactly. Otherwise, the function only fails 568 /// in case of an overflow or underflow. 569 bool getAsDouble(double &Result, bool AllowInexact = true) const; 570 571 /// @} 572 /// @name String Operations 573 /// @{ 574 575 // Convert the given ASCII string to lowercase. 576 LLVM_NODISCARD 577 std::string lower() const; 578 579 /// Convert the given ASCII string to uppercase. 580 LLVM_NODISCARD 581 std::string upper() const; 582 583 /// @} 584 /// @name Substring Operations 585 /// @{ 586 587 /// Return a reference to the substring from [Start, Start + N). 588 /// 589 /// \param Start The index of the starting character in the substring; if 590 /// the index is npos or greater than the length of the string then the 591 /// empty substring will be returned. 592 /// 593 /// \param N The number of characters to included in the substring. If N 594 /// exceeds the number of characters remaining in the string, the string 595 /// suffix (starting with \p Start) will be returned. 596 LLVM_NODISCARD 597 LLVM_ATTRIBUTE_ALWAYS_INLINE 598 StringRef substr(size_t Start, size_t N = npos) const { 599 Start = std::min(Start, Length); 600 return StringRef(Data + Start, std::min(N, Length - Start)); 601 } 602 603 /// Return a StringRef equal to 'this' but with only the first \p N 604 /// elements remaining. If \p N is greater than the length of the 605 /// string, the entire string is returned. 606 LLVM_NODISCARD 607 LLVM_ATTRIBUTE_ALWAYS_INLINE 608 StringRef take_front(size_t N = 1) const { 609 if (N >= size()) 610 return *this; 611 return drop_back(size() - N); 612 } 613 614 /// Return a StringRef equal to 'this' but with only the last \p N 615 /// elements remaining. If \p N is greater than the length of the 616 /// string, the entire string is returned. 617 LLVM_NODISCARD 618 LLVM_ATTRIBUTE_ALWAYS_INLINE 619 StringRef take_back(size_t N = 1) const { 620 if (N >= size()) 621 return *this; 622 return drop_front(size() - N); 623 } 624 625 /// Return the longest prefix of 'this' such that every character 626 /// in the prefix satisfies the given predicate. 627 LLVM_NODISCARD 628 LLVM_ATTRIBUTE_ALWAYS_INLINE take_while(function_ref<bool (char)> F)629 StringRef take_while(function_ref<bool(char)> F) const { 630 return substr(0, find_if_not(F)); 631 } 632 633 /// Return the longest prefix of 'this' such that no character in 634 /// the prefix satisfies the given predicate. 635 LLVM_NODISCARD 636 LLVM_ATTRIBUTE_ALWAYS_INLINE take_until(function_ref<bool (char)> F)637 StringRef take_until(function_ref<bool(char)> F) const { 638 return substr(0, find_if(F)); 639 } 640 641 /// Return a StringRef equal to 'this' but with the first \p N elements 642 /// dropped. 643 LLVM_NODISCARD 644 LLVM_ATTRIBUTE_ALWAYS_INLINE 645 StringRef drop_front(size_t N = 1) const { 646 assert(size() >= N && "Dropping more elements than exist"); 647 return substr(N); 648 } 649 650 /// Return a StringRef equal to 'this' but with the last \p N elements 651 /// dropped. 652 LLVM_NODISCARD 653 LLVM_ATTRIBUTE_ALWAYS_INLINE 654 StringRef drop_back(size_t N = 1) const { 655 assert(size() >= N && "Dropping more elements than exist"); 656 return substr(0, size()-N); 657 } 658 659 /// Return a StringRef equal to 'this', but with all characters satisfying 660 /// the given predicate dropped from the beginning of the string. 661 LLVM_NODISCARD 662 LLVM_ATTRIBUTE_ALWAYS_INLINE drop_while(function_ref<bool (char)> F)663 StringRef drop_while(function_ref<bool(char)> F) const { 664 return substr(find_if_not(F)); 665 } 666 667 /// Return a StringRef equal to 'this', but with all characters not 668 /// satisfying the given predicate dropped from the beginning of the string. 669 LLVM_NODISCARD 670 LLVM_ATTRIBUTE_ALWAYS_INLINE drop_until(function_ref<bool (char)> F)671 StringRef drop_until(function_ref<bool(char)> F) const { 672 return substr(find_if(F)); 673 } 674 675 /// Returns true if this StringRef has the given prefix and removes that 676 /// prefix. 677 LLVM_ATTRIBUTE_ALWAYS_INLINE consume_front(StringRef Prefix)678 bool consume_front(StringRef Prefix) { 679 if (!startswith(Prefix)) 680 return false; 681 682 *this = drop_front(Prefix.size()); 683 return true; 684 } 685 686 /// Returns true if this StringRef has the given suffix and removes that 687 /// suffix. 688 LLVM_ATTRIBUTE_ALWAYS_INLINE consume_back(StringRef Suffix)689 bool consume_back(StringRef Suffix) { 690 if (!endswith(Suffix)) 691 return false; 692 693 *this = drop_back(Suffix.size()); 694 return true; 695 } 696 697 /// Return a reference to the substring from [Start, End). 698 /// 699 /// \param Start The index of the starting character in the substring; if 700 /// the index is npos or greater than the length of the string then the 701 /// empty substring will be returned. 702 /// 703 /// \param End The index following the last character to include in the 704 /// substring. If this is npos or exceeds the number of characters 705 /// remaining in the string, the string suffix (starting with \p Start) 706 /// will be returned. If this is less than \p Start, an empty string will 707 /// be returned. 708 LLVM_NODISCARD 709 LLVM_ATTRIBUTE_ALWAYS_INLINE slice(size_t Start,size_t End)710 StringRef slice(size_t Start, size_t End) const { 711 Start = std::min(Start, Length); 712 End = std::min(std::max(Start, End), Length); 713 return StringRef(Data + Start, End - Start); 714 } 715 716 /// Split into two substrings around the first occurrence of a separator 717 /// character. 718 /// 719 /// If \p Separator is in the string, then the result is a pair (LHS, RHS) 720 /// such that (*this == LHS + Separator + RHS) is true and RHS is 721 /// maximal. If \p Separator is not in the string, then the result is a 722 /// pair (LHS, RHS) where (*this == LHS) and (RHS == ""). 723 /// 724 /// \param Separator The character to split on. 725 /// \returns The split substrings. 726 LLVM_NODISCARD split(char Separator)727 std::pair<StringRef, StringRef> split(char Separator) const { 728 return split(StringRef(&Separator, 1)); 729 } 730 731 /// Split into two substrings around the first occurrence of a separator 732 /// string. 733 /// 734 /// If \p Separator is in the string, then the result is a pair (LHS, RHS) 735 /// such that (*this == LHS + Separator + RHS) is true and RHS is 736 /// maximal. If \p Separator is not in the string, then the result is a 737 /// pair (LHS, RHS) where (*this == LHS) and (RHS == ""). 738 /// 739 /// \param Separator - The string to split on. 740 /// \return - The split substrings. 741 LLVM_NODISCARD split(StringRef Separator)742 std::pair<StringRef, StringRef> split(StringRef Separator) const { 743 size_t Idx = find(Separator); 744 if (Idx == npos) 745 return std::make_pair(*this, StringRef()); 746 return std::make_pair(slice(0, Idx), slice(Idx + Separator.size(), npos)); 747 } 748 749 /// Split into two substrings around the last occurrence of a separator 750 /// string. 751 /// 752 /// If \p Separator is in the string, then the result is a pair (LHS, RHS) 753 /// such that (*this == LHS + Separator + RHS) is true and RHS is 754 /// minimal. If \p Separator is not in the string, then the result is a 755 /// pair (LHS, RHS) where (*this == LHS) and (RHS == ""). 756 /// 757 /// \param Separator - The string to split on. 758 /// \return - The split substrings. 759 LLVM_NODISCARD rsplit(StringRef Separator)760 std::pair<StringRef, StringRef> rsplit(StringRef Separator) const { 761 size_t Idx = rfind(Separator); 762 if (Idx == npos) 763 return std::make_pair(*this, StringRef()); 764 return std::make_pair(slice(0, Idx), slice(Idx + Separator.size(), npos)); 765 } 766 767 /// Split into substrings around the occurrences of a separator string. 768 /// 769 /// Each substring is stored in \p A. If \p MaxSplit is >= 0, at most 770 /// \p MaxSplit splits are done and consequently <= \p MaxSplit + 1 771 /// elements are added to A. 772 /// If \p KeepEmpty is false, empty strings are not added to \p A. They 773 /// still count when considering \p MaxSplit 774 /// An useful invariant is that 775 /// Separator.join(A) == *this if MaxSplit == -1 and KeepEmpty == true 776 /// 777 /// \param A - Where to put the substrings. 778 /// \param Separator - The string to split on. 779 /// \param MaxSplit - The maximum number of times the string is split. 780 /// \param KeepEmpty - True if empty substring should be added. 781 void split(SmallVectorImpl<StringRef> &A, 782 StringRef Separator, int MaxSplit = -1, 783 bool KeepEmpty = true) const; 784 785 /// Split into substrings around the occurrences of a separator character. 786 /// 787 /// Each substring is stored in \p A. If \p MaxSplit is >= 0, at most 788 /// \p MaxSplit splits are done and consequently <= \p MaxSplit + 1 789 /// elements are added to A. 790 /// If \p KeepEmpty is false, empty strings are not added to \p A. They 791 /// still count when considering \p MaxSplit 792 /// An useful invariant is that 793 /// Separator.join(A) == *this if MaxSplit == -1 and KeepEmpty == true 794 /// 795 /// \param A - Where to put the substrings. 796 /// \param Separator - The string to split on. 797 /// \param MaxSplit - The maximum number of times the string is split. 798 /// \param KeepEmpty - True if empty substring should be added. 799 void split(SmallVectorImpl<StringRef> &A, char Separator, int MaxSplit = -1, 800 bool KeepEmpty = true) const; 801 802 /// Split into two substrings around the last occurrence of a separator 803 /// character. 804 /// 805 /// If \p Separator is in the string, then the result is a pair (LHS, RHS) 806 /// such that (*this == LHS + Separator + RHS) is true and RHS is 807 /// minimal. If \p Separator is not in the string, then the result is a 808 /// pair (LHS, RHS) where (*this == LHS) and (RHS == ""). 809 /// 810 /// \param Separator - The character to split on. 811 /// \return - The split substrings. 812 LLVM_NODISCARD rsplit(char Separator)813 std::pair<StringRef, StringRef> rsplit(char Separator) const { 814 return rsplit(StringRef(&Separator, 1)); 815 } 816 817 /// Return string with consecutive \p Char characters starting from the 818 /// the left removed. 819 LLVM_NODISCARD ltrim(char Char)820 StringRef ltrim(char Char) const { 821 return drop_front(std::min(Length, find_first_not_of(Char))); 822 } 823 824 /// Return string with consecutive characters in \p Chars starting from 825 /// the left removed. 826 LLVM_NODISCARD 827 StringRef ltrim(StringRef Chars = " \t\n\v\f\r") const { 828 return drop_front(std::min(Length, find_first_not_of(Chars))); 829 } 830 831 /// Return string with consecutive \p Char characters starting from the 832 /// right removed. 833 LLVM_NODISCARD rtrim(char Char)834 StringRef rtrim(char Char) const { 835 return drop_back(Length - std::min(Length, find_last_not_of(Char) + 1)); 836 } 837 838 /// Return string with consecutive characters in \p Chars starting from 839 /// the right removed. 840 LLVM_NODISCARD 841 StringRef rtrim(StringRef Chars = " \t\n\v\f\r") const { 842 return drop_back(Length - std::min(Length, find_last_not_of(Chars) + 1)); 843 } 844 845 /// Return string with consecutive \p Char characters starting from the 846 /// left and right removed. 847 LLVM_NODISCARD trim(char Char)848 StringRef trim(char Char) const { 849 return ltrim(Char).rtrim(Char); 850 } 851 852 /// Return string with consecutive characters in \p Chars starting from 853 /// the left and right removed. 854 LLVM_NODISCARD 855 StringRef trim(StringRef Chars = " \t\n\v\f\r") const { 856 return ltrim(Chars).rtrim(Chars); 857 } 858 859 /// @} 860 }; 861 862 /// A wrapper around a string literal that serves as a proxy for constructing 863 /// global tables of StringRefs with the length computed at compile time. 864 /// In order to avoid the invocation of a global constructor, StringLiteral 865 /// should *only* be used in a constexpr context, as such: 866 /// 867 /// constexpr StringLiteral S("test"); 868 /// 869 class StringLiteral : public StringRef { 870 private: StringLiteral(const char * Str,size_t N)871 constexpr StringLiteral(const char *Str, size_t N) : StringRef(Str, N) { 872 } 873 874 public: 875 template <size_t N> StringLiteral(const char (& Str)[N])876 constexpr StringLiteral(const char (&Str)[N]) 877 #if defined(__clang__) && __has_attribute(enable_if) 878 #pragma clang diagnostic push 879 #pragma clang diagnostic ignored "-Wgcc-compat" 880 __attribute((enable_if(__builtin_strlen(Str) == N - 1, 881 "invalid string literal"))) 882 #pragma clang diagnostic pop 883 #endif 884 : StringRef(Str, N - 1) { 885 } 886 887 // Explicit construction for strings like "foo\0bar". 888 template <size_t N> withInnerNUL(const char (& Str)[N])889 static constexpr StringLiteral withInnerNUL(const char (&Str)[N]) { 890 return StringLiteral(Str, N - 1); 891 } 892 }; 893 894 /// @name StringRef Comparison Operators 895 /// @{ 896 897 LLVM_ATTRIBUTE_ALWAYS_INLINE 898 inline bool operator==(StringRef LHS, StringRef RHS) { 899 return LHS.equals(RHS); 900 } 901 902 LLVM_ATTRIBUTE_ALWAYS_INLINE 903 inline bool operator!=(StringRef LHS, StringRef RHS) { return !(LHS == RHS); } 904 905 inline bool operator<(StringRef LHS, StringRef RHS) { 906 return LHS.compare(RHS) == -1; 907 } 908 909 inline bool operator<=(StringRef LHS, StringRef RHS) { 910 return LHS.compare(RHS) != 1; 911 } 912 913 inline bool operator>(StringRef LHS, StringRef RHS) { 914 return LHS.compare(RHS) == 1; 915 } 916 917 inline bool operator>=(StringRef LHS, StringRef RHS) { 918 return LHS.compare(RHS) != -1; 919 } 920 921 inline std::string &operator+=(std::string &buffer, StringRef string) { 922 return buffer.append(string.data(), string.size()); 923 } 924 925 /// @} 926 927 /// Compute a hash_code for a StringRef. 928 LLVM_NODISCARD 929 hash_code hash_value(StringRef S); 930 931 // StringRefs can be treated like a POD type. 932 template <typename T> struct isPodLike; 933 template <> struct isPodLike<StringRef> { static const bool value = true; }; 934 935 } // end namespace llvm 936 937 #endif // LLVM_ADT_STRINGREF_H 938