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 <cassert>
14#include <cstring>
15#include <utility>
16#include <string>
17
18namespace llvm {
19  template<typename T>
20  class SmallVectorImpl;
21  class APInt;
22
23  /// StringRef - Represent a constant reference to a string, i.e. a character
24  /// array and a length, which need not be null terminated.
25  ///
26  /// This class does not own the string data, it is expected to be used in
27  /// situations where the character data resides in some other buffer, whose
28  /// lifetime extends past that of the StringRef. For this reason, it is not in
29  /// general safe to store a StringRef.
30  class StringRef {
31  public:
32    typedef const char *iterator;
33    typedef const char *const_iterator;
34    static const size_t npos = ~size_t(0);
35    typedef size_t size_type;
36
37  private:
38    /// The start of the string, in an external buffer.
39    const char *Data;
40
41    /// The length of the string.
42    size_t Length;
43
44    // Workaround PR5482: nearly all gcc 4.x miscompile StringRef and std::min()
45    // Changing the arg of min to be an integer, instead of a reference to an
46    // integer works around this bug.
47    static size_t min(size_t a, size_t b) { return a < b ? a : b; }
48    static size_t max(size_t a, size_t b) { return a > b ? a : b; }
49
50    // Workaround memcmp issue with null pointers (undefined behavior)
51    // by providing a specialized version
52    static int compareMemory(const char *Lhs, const char *Rhs, size_t Length) {
53      if (Length == 0) { return 0; }
54      return ::memcmp(Lhs,Rhs,Length);
55    }
56
57  public:
58    /// @name Constructors
59    /// @{
60
61    /// Construct an empty string ref.
62    /*implicit*/ StringRef() : Data(0), Length(0) {}
63
64    /// Construct a string ref from a cstring.
65    /*implicit*/ StringRef(const char *Str)
66      : Data(Str) {
67        assert(Str && "StringRef cannot be built from a NULL argument");
68        Length = ::strlen(Str); // invoking strlen(NULL) is undefined behavior
69      }
70
71    /// Construct a string ref from a pointer and length.
72    /*implicit*/ StringRef(const char *data, size_t length)
73      : Data(data), Length(length) {
74        assert((data || length == 0) &&
75        "StringRef cannot be built from a NULL argument with non-null length");
76      }
77
78    /// Construct a string ref from an std::string.
79    /*implicit*/ StringRef(const std::string &Str)
80      : Data(Str.data()), Length(Str.length()) {}
81
82    /// @}
83    /// @name Iterators
84    /// @{
85
86    iterator begin() const { return Data; }
87
88    iterator end() const { return Data + Length; }
89
90    /// @}
91    /// @name String Operations
92    /// @{
93
94    /// data - Get a pointer to the start of the string (which may not be null
95    /// terminated).
96    const char *data() const { return Data; }
97
98    /// empty - Check if the string is empty.
99    bool empty() const { return Length == 0; }
100
101    /// size - Get the string size.
102    size_t size() const { return Length; }
103
104    /// front - Get the first character in the string.
105    char front() const {
106      assert(!empty());
107      return Data[0];
108    }
109
110    /// back - Get the last character in the string.
111    char back() const {
112      assert(!empty());
113      return Data[Length-1];
114    }
115
116    /// equals - Check for string equality, this is more efficient than
117    /// compare() when the relative ordering of inequal strings isn't needed.
118    bool equals(StringRef RHS) const {
119      return (Length == RHS.Length &&
120              compareMemory(Data, RHS.Data, RHS.Length) == 0);
121    }
122
123    /// equals_lower - Check for string equality, ignoring case.
124    bool equals_lower(StringRef RHS) const {
125      return Length == RHS.Length && compare_lower(RHS) == 0;
126    }
127
128    /// compare - Compare two strings; the result is -1, 0, or 1 if this string
129    /// is lexicographically less than, equal to, or greater than the \arg RHS.
130    int compare(StringRef RHS) const {
131      // Check the prefix for a mismatch.
132      if (int Res = compareMemory(Data, RHS.Data, min(Length, RHS.Length)))
133        return Res < 0 ? -1 : 1;
134
135      // Otherwise the prefixes match, so we only need to check the lengths.
136      if (Length == RHS.Length)
137        return 0;
138      return Length < RHS.Length ? -1 : 1;
139    }
140
141    /// compare_lower - Compare two strings, ignoring case.
142    int compare_lower(StringRef RHS) const;
143
144    /// compare_numeric - Compare two strings, treating sequences of digits as
145    /// numbers.
146    int compare_numeric(StringRef RHS) const;
147
148    /// \brief Determine the edit distance between this string and another
149    /// string.
150    ///
151    /// \param Other the string to compare this string against.
152    ///
153    /// \param AllowReplacements whether to allow character
154    /// replacements (change one character into another) as a single
155    /// operation, rather than as two operations (an insertion and a
156    /// removal).
157    ///
158    /// \param MaxEditDistance If non-zero, the maximum edit distance that
159    /// this routine is allowed to compute. If the edit distance will exceed
160    /// that maximum, returns \c MaxEditDistance+1.
161    ///
162    /// \returns the minimum number of character insertions, removals,
163    /// or (if \p AllowReplacements is \c true) replacements needed to
164    /// transform one of the given strings into the other. If zero,
165    /// the strings are identical.
166    unsigned edit_distance(StringRef Other, bool AllowReplacements = true,
167                           unsigned MaxEditDistance = 0);
168
169    /// str - Get the contents as an std::string.
170    std::string str() const {
171      if (Data == 0) return std::string();
172      return std::string(Data, Length);
173    }
174
175    /// @}
176    /// @name Operator Overloads
177    /// @{
178
179    char operator[](size_t Index) const {
180      assert(Index < Length && "Invalid index!");
181      return Data[Index];
182    }
183
184    /// @}
185    /// @name Type Conversions
186    /// @{
187
188    operator std::string() const {
189      return str();
190    }
191
192    /// @}
193    /// @name String Predicates
194    /// @{
195
196    /// startswith - Check if this string starts with the given \arg Prefix.
197    bool startswith(StringRef Prefix) const {
198      return Length >= Prefix.Length &&
199             compareMemory(Data, Prefix.Data, Prefix.Length) == 0;
200    }
201
202    /// endswith - Check if this string ends with the given \arg Suffix.
203    bool endswith(StringRef Suffix) const {
204      return Length >= Suffix.Length &&
205        compareMemory(end() - Suffix.Length, Suffix.Data, Suffix.Length) == 0;
206    }
207
208    /// @}
209    /// @name String Searching
210    /// @{
211
212    /// find - Search for the first character \arg C in the string.
213    ///
214    /// \return - The index of the first occurrence of \arg C, or npos if not
215    /// found.
216    size_t find(char C, size_t From = 0) const {
217      for (size_t i = min(From, Length), e = Length; i != e; ++i)
218        if (Data[i] == C)
219          return i;
220      return npos;
221    }
222
223    /// find - Search for the first string \arg Str in the string.
224    ///
225    /// \return - The index of the first occurrence of \arg Str, or npos if not
226    /// found.
227    size_t find(StringRef Str, size_t From = 0) const;
228
229    /// rfind - Search for the last character \arg C in the string.
230    ///
231    /// \return - The index of the last occurrence of \arg C, or npos if not
232    /// found.
233    size_t rfind(char C, size_t From = npos) const {
234      From = min(From, Length);
235      size_t i = From;
236      while (i != 0) {
237        --i;
238        if (Data[i] == C)
239          return i;
240      }
241      return npos;
242    }
243
244    /// rfind - Search for the last string \arg Str in the string.
245    ///
246    /// \return - The index of the last occurrence of \arg Str, or npos if not
247    /// found.
248    size_t rfind(StringRef Str) const;
249
250    /// find_first_of - Find the first character in the string that is \arg C,
251    /// or npos if not found. Same as find.
252    size_type find_first_of(char C, size_t From = 0) const {
253      return find(C, From);
254    }
255
256    /// find_first_of - Find the first character in the string that is in \arg
257    /// Chars, or npos if not found.
258    ///
259    /// Note: O(size() + Chars.size())
260    size_type find_first_of(StringRef Chars, size_t From = 0) const;
261
262    /// find_first_not_of - Find the first character in the string that is not
263    /// \arg C or npos if not found.
264    size_type find_first_not_of(char C, size_t From = 0) const;
265
266    /// find_first_not_of - Find the first character in the string that is not
267    /// in the string \arg Chars, or npos if not found.
268    ///
269    /// Note: O(size() + Chars.size())
270    size_type find_first_not_of(StringRef Chars, size_t From = 0) const;
271
272    /// find_last_of - Find the last character in the string that is \arg C, or
273    /// npos if not found.
274    size_type find_last_of(char C, size_t From = npos) const {
275      return rfind(C, From);
276    }
277
278    /// find_last_of - Find the last character in the string that is in \arg C,
279    /// or npos if not found.
280    ///
281    /// Note: O(size() + Chars.size())
282    size_type find_last_of(StringRef Chars, size_t From = npos) const;
283
284    /// @}
285    /// @name Helpful Algorithms
286    /// @{
287
288    /// count - Return the number of occurrences of \arg C in the string.
289    size_t count(char C) const {
290      size_t Count = 0;
291      for (size_t i = 0, e = Length; i != e; ++i)
292        if (Data[i] == C)
293          ++Count;
294      return Count;
295    }
296
297    /// count - Return the number of non-overlapped occurrences of \arg Str in
298    /// the string.
299    size_t count(StringRef Str) const;
300
301    /// getAsInteger - Parse the current string as an integer of the specified
302    /// radix.  If Radix is specified as zero, this does radix autosensing using
303    /// extended C rules: 0 is octal, 0x is hex, 0b is binary.
304    ///
305    /// If the string is invalid or if only a subset of the string is valid,
306    /// this returns true to signify the error.  The string is considered
307    /// erroneous if empty.
308    ///
309    bool getAsInteger(unsigned Radix, long long &Result) const;
310    bool getAsInteger(unsigned Radix, unsigned long long &Result) const;
311    bool getAsInteger(unsigned Radix, int &Result) const;
312    bool getAsInteger(unsigned Radix, unsigned &Result) const;
313
314    // TODO: Provide overloads for int/unsigned that check for overflow.
315
316    /// getAsInteger - Parse the current string as an integer of the
317    /// specified radix, or of an autosensed radix if the radix given
318    /// is 0.  The current value in Result is discarded, and the
319    /// storage is changed to be wide enough to store the parsed
320    /// integer.
321    ///
322    /// Returns true if the string does not solely consist of a valid
323    /// non-empty number in the appropriate base.
324    ///
325    /// APInt::fromString is superficially similar but assumes the
326    /// string is well-formed in the given radix.
327    bool getAsInteger(unsigned Radix, APInt &Result) const;
328
329    /// @}
330    /// @name Substring Operations
331    /// @{
332
333    /// substr - Return a reference to the substring from [Start, Start + N).
334    ///
335    /// \param Start - The index of the starting character in the substring; if
336    /// the index is npos or greater than the length of the string then the
337    /// empty substring will be returned.
338    ///
339    /// \param N - The number of characters to included in the substring. If N
340    /// exceeds the number of characters remaining in the string, the string
341    /// suffix (starting with \arg Start) will be returned.
342    StringRef substr(size_t Start, size_t N = npos) const {
343      Start = min(Start, Length);
344      return StringRef(Data + Start, min(N, Length - Start));
345    }
346
347    /// slice - Return a reference to the substring from [Start, End).
348    ///
349    /// \param Start - The index of the starting character in the substring; if
350    /// the index is npos or greater than the length of the string then the
351    /// empty substring will be returned.
352    ///
353    /// \param End - The index following the last character to include in the
354    /// substring. If this is npos, or less than \arg Start, or exceeds the
355    /// number of characters remaining in the string, the string suffix
356    /// (starting with \arg Start) will be returned.
357    StringRef slice(size_t Start, size_t End) const {
358      Start = min(Start, Length);
359      End = min(max(Start, End), Length);
360      return StringRef(Data + Start, End - Start);
361    }
362
363    /// split - Split into two substrings around the first occurrence of a
364    /// separator character.
365    ///
366    /// If \arg Separator is in the string, then the result is a pair (LHS, RHS)
367    /// such that (*this == LHS + Separator + RHS) is true and RHS is
368    /// maximal. If \arg Separator is not in the string, then the result is a
369    /// pair (LHS, RHS) where (*this == LHS) and (RHS == "").
370    ///
371    /// \param Separator - The character to split on.
372    /// \return - The split substrings.
373    std::pair<StringRef, StringRef> split(char Separator) const {
374      size_t Idx = find(Separator);
375      if (Idx == npos)
376        return std::make_pair(*this, StringRef());
377      return std::make_pair(slice(0, Idx), slice(Idx+1, npos));
378    }
379
380    /// split - Split into two substrings around the first occurrence of a
381    /// separator string.
382    ///
383    /// If \arg Separator is in the string, then the result is a pair (LHS, RHS)
384    /// such that (*this == LHS + Separator + RHS) is true and RHS is
385    /// maximal. If \arg Separator is not in the string, then the result is a
386    /// pair (LHS, RHS) where (*this == LHS) and (RHS == "").
387    ///
388    /// \param Separator - The string to split on.
389    /// \return - The split substrings.
390    std::pair<StringRef, StringRef> split(StringRef Separator) const {
391      size_t Idx = find(Separator);
392      if (Idx == npos)
393        return std::make_pair(*this, StringRef());
394      return std::make_pair(slice(0, Idx), slice(Idx + Separator.size(), npos));
395    }
396
397    /// split - Split into substrings around the occurrences of a separator
398    /// string.
399    ///
400    /// Each substring is stored in \arg A. If \arg MaxSplit is >= 0, at most
401    /// \arg MaxSplit splits are done and consequently <= \arg MaxSplit
402    /// elements are added to A.
403    /// If \arg KeepEmpty is false, empty strings are not added to \arg A. They
404    /// still count when considering \arg MaxSplit
405    /// An useful invariant is that
406    /// Separator.join(A) == *this if MaxSplit == -1 and KeepEmpty == true
407    ///
408    /// \param A - Where to put the substrings.
409    /// \param Separator - The string to split on.
410    /// \param MaxSplit - The maximum number of times the string is split.
411    /// \param KeepEmpty - True if empty substring should be added.
412    void split(SmallVectorImpl<StringRef> &A,
413               StringRef Separator, int MaxSplit = -1,
414               bool KeepEmpty = true) const;
415
416    /// rsplit - Split into two substrings around the last occurrence of a
417    /// separator character.
418    ///
419    /// If \arg Separator is in the string, then the result is a pair (LHS, RHS)
420    /// such that (*this == LHS + Separator + RHS) is true and RHS is
421    /// minimal. If \arg Separator is not in the string, then the result is a
422    /// pair (LHS, RHS) where (*this == LHS) and (RHS == "").
423    ///
424    /// \param Separator - The character to split on.
425    /// \return - The split substrings.
426    std::pair<StringRef, StringRef> rsplit(char Separator) const {
427      size_t Idx = rfind(Separator);
428      if (Idx == npos)
429        return std::make_pair(*this, StringRef());
430      return std::make_pair(slice(0, Idx), slice(Idx+1, npos));
431    }
432
433    /// @}
434  };
435
436  /// @name StringRef Comparison Operators
437  /// @{
438
439  inline bool operator==(StringRef LHS, StringRef RHS) {
440    return LHS.equals(RHS);
441  }
442
443  inline bool operator!=(StringRef LHS, StringRef RHS) {
444    return !(LHS == RHS);
445  }
446
447  inline bool operator<(StringRef LHS, StringRef RHS) {
448    return LHS.compare(RHS) == -1;
449  }
450
451  inline bool operator<=(StringRef LHS, StringRef RHS) {
452    return LHS.compare(RHS) != 1;
453  }
454
455  inline bool operator>(StringRef LHS, StringRef RHS) {
456    return LHS.compare(RHS) == 1;
457  }
458
459  inline bool operator>=(StringRef LHS, StringRef RHS) {
460    return LHS.compare(RHS) != -1;
461  }
462
463  inline std::string &operator+=(std::string &buffer, llvm::StringRef string) {
464    return buffer.append(string.data(), string.size());
465  }
466
467  /// @}
468
469  // StringRefs can be treated like a POD type.
470  template <typename T> struct isPodLike;
471  template <> struct isPodLike<StringRef> { static const bool value = true; };
472
473}
474
475#endif
476