base/string_tokenizer.h

   1 // Copyright (c) 2006-2008 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #ifndef BASE_STRING_TOKENIZER_H_
   6 #define BASE_STRING_TOKENIZER_H_
   7
   8 #include <algorithm>
   9 #include <string>
  10
  11 #include "base/string_piece.h"
  12
  13 // StringTokenizerT is a simple string tokenizer class.  It works like an
  14 // iterator that with each step (see the Advance method) updates members that
  15 // refer to the next token in the input string.  The user may optionally
  16 // configure the tokenizer to return delimiters.
  17 //
  18 // Warning: be careful not to pass a C string into the 2-arg constructor:
  19 // StringTokenizer t("this is a test", " ");  // WRONG
  20 // This will create a temporary std::string, save the begin() and end()
  21 // iterators, and then the string will be freed before we actually start
  22 // tokenizing it.
  23 // Instead, use a std::string or use the 3 arg constructor of CStringTokenizer.
  24 //
  25 //
  26 // EXAMPLE 1:
  27 //
  28 //   char input[] = "this is a test";
  29 //   CStringTokenizer t(input, input + strlen(input), " ");
  30 //   while (t.GetNext()) {
  31 //     printf("%s\n", t.token().c_str());
  32 //   }
  33 //
  34 // Output:
  35 //
  36 //   this
  37 //   is
  38 //   a
  39 //   test
  40 //
  41 //
  42 // EXAMPLE 2:
  43 //
  44 //   std::string input = "no-cache=\"foo, bar\", private";
  45 //   StringTokenizer t(input, ", ");
  46 //   t.set_quote_chars("\"");
  47 //   while (t.GetNext()) {
  48 //     printf("%s\n", t.token().c_str());
  49 //   }
  50 //
  51 // Output:
  52 //
  53 //   no-cache="foo, bar"
  54 //   private
  55 //
  56 //
  57 // EXAMPLE 3:
  58 //
  59 //   bool next_is_option = false, next_is_value = false;
  60 //   std::string input = "text/html; charset=UTF-8; foo=bar";
  61 //   StringTokenizer t(input, "; =");
  62 //   t.set_options(StringTokenizer::RETURN_DELIMS);
  63 //   while (t.GetNext()) {
  64 //     if (t.token_is_delim()) {
  65 //       switch (*t.token_begin()) {
  66 //         case ';':
  67 //           next_is_option = true;
  68 //           break;
  69 //         case '=':
  70 //           next_is_value = true;
  71 //           break;
  72 //       }
  73 //     } else {
  74 //       const char* label;
  75 //       if (next_is_option) {
  76 //         label = "option-name";
  77 //         next_is_option = false;
  78 //       } else if (next_is_value) {
  79 //         label = "option-value";
  80 //         next_is_value = false;
  81 //       } else {
  82 //         label = "mime-type";
  83 //       }
  84 //       printf("%s: %s\n", label, t.token().c_str());
  85 //     }
  86 //   }
  87 //
  88 //
  89 template <class str, class const_iterator>
  90 class StringTokenizerT {
  91  public:
  92   typedef typename str::value_type char_type;
  93
  94   // Options that may be pass to set_options()
  95   enum {
  96     // Specifies the delimiters should be returned as tokens
  97     RETURN_DELIMS = 1 << 0,
  98   };
  99
 100   // The string object must live longer than the tokenizer.  (In particular this
 101   // should not be constructed with a temporary.)
 102   StringTokenizerT(const str& string,
 103                    const str& delims) {
 104     Init(string.begin(), string.end(), delims);
 105   }
 106
 107   StringTokenizerT(const_iterator string_begin,
 108                    const_iterator string_end,
 109                    const str& delims) {
 110     Init(string_begin, string_end, delims);
 111   }
 112
 113   // Set the options for this tokenizer.  By default, this is 0.
 114   void set_options(int options) { options_ = options; }
 115
 116   // Set the characters to regard as quotes.  By default, this is empty.  When
 117   // a quote char is encountered, the tokenizer will switch into a mode where
 118   // it ignores delimiters that it finds.  It switches out of this mode once it
 119   // finds another instance of the quote char.  If a backslash is encountered
 120   // within a quoted string, then the next character is skipped.
 121   void set_quote_chars(const str& quotes) { quotes_ = quotes; }
 122
 123   // Call this method to advance the tokenizer to the next delimiter.  This
 124   // returns false if the tokenizer is complete.  This method must be called
 125   // before calling any of the token* methods.
 126   bool GetNext() {
 127     if (quotes_.empty() && options_ == 0)
 128       return QuickGetNext();
 129     else
 130       return FullGetNext();
 131   }
 132
 133   // Start iterating through tokens from the beginning of the string.
 134   void Reset() {
 135     token_end_ = start_pos_;
 136   }
 137
 138   // Returns true if token is a delimiter.  When the tokenizer is constructed
 139   // with the RETURN_DELIMS option, this method can be used to check if the
 140   // returned token is actually a delimiter.
 141   bool token_is_delim() const { return token_is_delim_; }
 142
 143   // If GetNext() returned true, then these methods may be used to read the
 144   // value of the token.
 145   const_iterator token_begin() const { return token_begin_; }
 146   const_iterator token_end() const { return token_end_; }
 147   str token() const { return str(token_begin_, token_end_); }
 148   base::StringPiece token_piece() const {
 149     return base::StringPiece(&*token_begin_,
 150                              std::distance(token_begin_, token_end_));
 151   }
 152
 153  private:
 154   void Init(const_iterator string_begin,
 155             const_iterator string_end,
 156             const str& delims) {
 157     start_pos_ = string_begin;
 158     token_begin_ = string_begin;
 159     token_end_ = string_begin;
 160     end_ = string_end;
 161     delims_ = delims;
 162     options_ = 0;
 163   }
 164
 165   // Implementation of GetNext() for when we have no quote characters. We have
 166   // two separate implementations because AdvanceOne() is a hot spot in large
 167   // text files with large tokens.
 168   bool QuickGetNext() {
 169     token_is_delim_ = false;
 170     for (;;) {
 171       token_begin_ = token_end_;
 172       if (token_end_ == end_)
 173         return false;
 174       ++token_end_;
 175       if (delims_.find(*token_begin_) == str::npos)
 176         break;
 177       // else skip over delim
 178     }
 179     while (token_end_ != end_ && delims_.find(*token_end_) == str::npos)
 180       ++token_end_;
 181     return true;
 182   }
 183
 184   // Implementation of GetNext() for when we have to take quotes into account.
 185   bool FullGetNext() {
 186     AdvanceState state;
 187     token_is_delim_ = false;
 188     for (;;) {
 189       token_begin_ = token_end_;
 190       if (token_end_ == end_)
 191         return false;
 192       ++token_end_;
 193       if (AdvanceOne(&state, *token_begin_))
 194         break;
 195       if (options_ & RETURN_DELIMS) {
 196         token_is_delim_ = true;
 197         return true;
 198       }
 199       // else skip over delim
 200     }
 201     while (token_end_ != end_ && AdvanceOne(&state, *token_end_))
 202       ++token_end_;
 203     return true;
 204   }
 205
 206   bool IsDelim(char_type c) const {
 207     return delims_.find(c) != str::npos;
 208   }
 209
 210   bool IsQuote(char_type c) const {
 211     return quotes_.find(c) != str::npos;
 212   }
 213
 214   struct AdvanceState {
 215     bool in_quote;
 216     bool in_escape;
 217     char_type quote_char;
 218     AdvanceState() : in_quote(false), in_escape(false) {}
 219   };
 220
 221   // Returns true if a delimiter was not hit.
 222   bool AdvanceOne(AdvanceState* state, char_type c) {
 223     if (state->in_quote) {
 224       if (state->in_escape) {
 225         state->in_escape = false;
 226       } else if (c == '\\') {
 227         state->in_escape = true;
 228       } else if (c == state->quote_char) {
 229         state->in_quote = false;
 230       }
 231     } else {
 232       if (IsDelim(c))
 233         return false;
 234       state->in_quote = IsQuote(state->quote_char = c);
 235     }
 236     return true;
 237   }
 238
 239   const_iterator start_pos_;
 240   const_iterator token_begin_;
 241   const_iterator token_end_;
 242   const_iterator end_;
 243   str delims_;
 244   str quotes_;
 245   int options_;
 246   bool token_is_delim_;
 247 };
 248
 249 typedef StringTokenizerT<std::string, std::string::const_iterator>
 250     StringTokenizer;
 251 typedef StringTokenizerT<std::wstring, std::wstring::const_iterator>
 252     WStringTokenizer;
 253 typedef StringTokenizerT<std::string, const char*> CStringTokenizer;
 254
 255 #endif  // BASE_STRING_TOKENIZER_H_