include/mruby/string.h

   1 /**
   2 ** @file mruby/string.h - String class
   3 **
   4 ** See Copyright Notice in mruby.h
   5 */
   6
   7 #ifndef MRUBY_STRING_H
   8 #define MRUBY_STRING_H
   9
  10 #include "common.h"
  11
  12 /**
  13  * String class
  14  */
  15 MRB_BEGIN_DECL
  16
  17 extern const char mrb_digitmap[];
  18
  19 #define RSTRING_EMBED_LEN_MAX \
  20   ((mrb_int)(sizeof(void*) * 3 + sizeof(void*) - 32 / CHAR_BIT - 1))
  21
  22 struct RString {
  23   MRB_OBJECT_HEADER;
  24   union {
  25     struct {
  26       mrb_ssize len;
  27       union {
  28         mrb_ssize capa;
  29         struct mrb_shared_string *shared;
  30         struct RString *fshared;
  31       } aux;
  32       char *ptr;
  33     } heap;
  34   } as;
  35 };
  36 struct RStringEmbed {
  37   MRB_OBJECT_HEADER;
  38   char ary[RSTRING_EMBED_LEN_MAX+1];
  39 };
  40
  41 #define RSTR_SET_TYPE_FLAG(s, type) (RSTR_UNSET_TYPE_FLAG(s), (s)->flags |= MRB_STR_##type)
  42 #define RSTR_UNSET_TYPE_FLAG(s) ((s)->flags &= ~(MRB_STR_TYPE_MASK|MRB_STR_EMBED_LEN_MASK))
  43
  44 #define RSTR_EMBED_P(s) ((s)->flags & MRB_STR_EMBED)
  45 #define RSTR_SET_EMBED_FLAG(s) ((s)->flags |= MRB_STR_EMBED)
  46 #define RSTR_UNSET_EMBED_FLAG(s) ((s)->flags &= ~(MRB_STR_EMBED|MRB_STR_EMBED_LEN_MASK))
  47 #define RSTR_SET_EMBED_LEN(s, n) do {\
  48   size_t tmp_n = (n);\
  49   (s)->flags &= ~MRB_STR_EMBED_LEN_MASK;\
  50   (s)->flags |= (tmp_n) << MRB_STR_EMBED_LEN_SHIFT;\
  51 } while (0)
  52 #define RSTR_SET_LEN(s, n) do {\
  53   if (RSTR_EMBED_P(s)) {\
  54     RSTR_SET_EMBED_LEN((s),(n));\
  55   }\
  56   else {\
  57     (s)->as.heap.len = (mrb_ssize)(n);\
  58   }\
  59 } while (0)
  60 #define RSTR_EMBED_PTR(s) (((struct RStringEmbed*)(s))->ary)
  61 #define RSTR_EMBED_LEN(s)\
  62   (mrb_int)(((s)->flags & MRB_STR_EMBED_LEN_MASK) >> MRB_STR_EMBED_LEN_SHIFT)
  63 #define RSTR_EMBEDDABLE_P(len) ((len) <= RSTRING_EMBED_LEN_MAX)
  64
  65 #define RSTR_PTR(s) ((RSTR_EMBED_P(s)) ? RSTR_EMBED_PTR(s) : (s)->as.heap.ptr)
  66 #define RSTR_LEN(s) ((RSTR_EMBED_P(s)) ? RSTR_EMBED_LEN(s) : (s)->as.heap.len)
  67 #define RSTR_CAPA(s) (RSTR_EMBED_P(s) ? RSTRING_EMBED_LEN_MAX : (s)->as.heap.aux.capa)
  68
  69 #define RSTR_SHARED_P(s) ((s)->flags & MRB_STR_SHARED)
  70 #define RSTR_SET_SHARED_FLAG(s) ((s)->flags |= MRB_STR_SHARED)
  71 #define RSTR_UNSET_SHARED_FLAG(s) ((s)->flags &= ~MRB_STR_SHARED)
  72
  73 #define RSTR_FSHARED_P(s) ((s)->flags & MRB_STR_FSHARED)
  74 #define RSTR_SET_FSHARED_FLAG(s) ((s)->flags |= MRB_STR_FSHARED)
  75 #define RSTR_UNSET_FSHARED_FLAG(s) ((s)->flags &= ~MRB_STR_FSHARED)
  76
  77 #define RSTR_NOFREE_P(s) ((s)->flags & MRB_STR_NOFREE)
  78 #define RSTR_SET_NOFREE_FLAG(s) ((s)->flags |= MRB_STR_NOFREE)
  79 #define RSTR_UNSET_NOFREE_FLAG(s) ((s)->flags &= ~MRB_STR_NOFREE)
  80
  81 #ifdef MRB_UTF8_STRING
  82 # define RSTR_ASCII_P(s) ((s)->flags & MRB_STR_ASCII)
  83 # define RSTR_SET_ASCII_FLAG(s) ((s)->flags |= MRB_STR_ASCII)
  84 # define RSTR_UNSET_ASCII_FLAG(s) ((s)->flags &= ~MRB_STR_ASCII)
  85 # define RSTR_WRITE_ASCII_FLAG(s, v) (RSTR_UNSET_ASCII_FLAG(s), (s)->flags |= v)
  86 # define RSTR_COPY_ASCII_FLAG(dst, src) RSTR_WRITE_ASCII_FLAG(dst, RSTR_ASCII_P(src))
  87 #else
  88 # define RSTR_ASCII_P(s) (void)0
  89 # define RSTR_SET_ASCII_FLAG(s) (void)0
  90 # define RSTR_UNSET_ASCII_FLAG(s) (void)0
  91 # define RSTR_WRITE_ASCII_FLAG(s, v) (void)0
  92 # define RSTR_COPY_ASCII_FLAG(dst, src) (void)0
  93 #endif
  94
  95 /**
  96  * Returns a pointer from a Ruby string
  97  */
  98 #define mrb_str_ptr(s)       ((struct RString*)(mrb_ptr(s)))
  99 #define RSTRING(s)           mrb_str_ptr(s)
 100 #define RSTRING_PTR(s)       RSTR_PTR(RSTRING(s))
 101 #define RSTRING_EMBED_LEN(s) RSTR_EMBED_LEN(RSTRING(s))
 102 #define RSTRING_LEN(s)       RSTR_LEN(RSTRING(s))
 103 #define RSTRING_CAPA(s)      RSTR_CAPA(RSTRING(s))
 104 #define RSTRING_END(s)       (RSTRING_PTR(s) + RSTRING_LEN(s))
 105 MRB_API mrb_int mrb_str_strlen(mrb_state*, struct RString*);
 106 #define RSTRING_CSTR(mrb,s)  mrb_string_cstr(mrb, s)
 107
 108 #define MRB_STR_SHARED    1
 109 #define MRB_STR_FSHARED   2
 110 #define MRB_STR_NOFREE    4
 111 #define MRB_STR_EMBED     8  /* type flags up to here */
 112 #define MRB_STR_ASCII    16
 113 #define MRB_STR_EMBED_LEN_SHIFT 6
 114 #define MRB_STR_EMBED_LEN_BIT 5
 115 #define MRB_STR_EMBED_LEN_MASK (((1 << MRB_STR_EMBED_LEN_BIT) - 1) << MRB_STR_EMBED_LEN_SHIFT)
 116 #define MRB_STR_TYPE_MASK 15
 117
 118 void mrb_gc_free_str(mrb_state*, struct RString*);
 119
 120 MRB_API void mrb_str_modify(mrb_state *mrb, struct RString *s);
 121 /* mrb_str_modify() with keeping ASCII flag if set */
 122 MRB_API void mrb_str_modify_keep_ascii(mrb_state *mrb, struct RString *s);
 123
 124 /**
 125  * Finds the index of a substring in a string
 126  */
 127 MRB_API mrb_int mrb_str_index(mrb_state *mrb, mrb_value str, const char *p, mrb_int len, mrb_int offset);
 128 #define mrb_str_index_lit(mrb, str, lit, off) mrb_str_index(mrb, str, lit, mrb_strlen_lit(lit), off);
 129
 130 /**
 131  * Appends self to other. Returns self as a concatenated string.
 132  *
 133  *
 134  * Example:
 135  *
 136  *     int
 137  *     main(int argc,
 138  *          char **argv)
 139  *     {
 140  *       // Variable declarations.
 141  *       mrb_value str1;
 142  *       mrb_value str2;
 143  *
 144  *       mrb_state *mrb = mrb_open();
 145  *       if (!mrb)
 146  *       {
 147  *          // handle error
 148  *       }
 149  *
 150  *       // Creates new Ruby strings.
 151  *       str1 = mrb_str_new_lit(mrb, "abc");
 152  *       str2 = mrb_str_new_lit(mrb, "def");
 153  *
 154  *       // Concatenates str2 to str1.
 155  *       mrb_str_concat(mrb, str1, str2);
 156  *
 157  *       // Prints new Concatenated Ruby string.
 158  *       mrb_p(mrb, str1);
 159  *
 160  *       mrb_close(mrb);
 161  *       return 0;
 162  *     }
 163  *
 164  * Result:
 165  *
 166  *     => "abcdef"
 167  *
 168  * @param mrb The current mruby state.
 169  * @param self String to concatenate.
 170  * @param other String to append to self.
 171  * @return [mrb_value] Returns a new String appending other to self.
 172  */
 173 MRB_API void mrb_str_concat(mrb_state *mrb, mrb_value self, mrb_value other);
 174
 175 /**
 176  * Adds two strings together.
 177  *
 178  *
 179  * Example:
 180  *
 181  *     int
 182  *     main(int argc,
 183  *          char **argv)
 184  *     {
 185  *       // Variable declarations.
 186  *       mrb_value a;
 187  *       mrb_value b;
 188  *       mrb_value c;
 189  *
 190  *       mrb_state *mrb = mrb_open();
 191  *       if (!mrb)
 192  *       {
 193  *          // handle error
 194  *       }
 195  *
 196  *       // Creates two Ruby strings from the passed in C strings.
 197  *       a = mrb_str_new_lit(mrb, "abc");
 198  *       b = mrb_str_new_lit(mrb, "def");
 199  *
 200  *       // Prints both C strings.
 201  *       mrb_p(mrb, a);
 202  *       mrb_p(mrb, b);
 203  *
 204  *       // Concatenates both Ruby strings.
 205  *       c = mrb_str_plus(mrb, a, b);
 206  *
 207  *       // Prints new Concatenated Ruby string.
 208  *       mrb_p(mrb, c);
 209  *
 210  *       mrb_close(mrb);
 211  *       return 0;
 212  *     }
 213  *
 214  *
 215  * Result:
 216  *
 217  *     => "abc"  # First string
 218  *     => "def"  # Second string
 219  *     => "abcdef" # First & Second concatenated.
 220  *
 221  * @param mrb The current mruby state.
 222  * @param a First string to concatenate.
 223  * @param b Second string to concatenate.
 224  * @return [mrb_value] Returns a new String containing a concatenated to b.
 225  */
 226 MRB_API mrb_value mrb_str_plus(mrb_state *mrb, mrb_value a, mrb_value b);
 227
 228 /**
 229  * Converts pointer into a Ruby string.
 230  *
 231  * @param mrb The current mruby state.
 232  * @param p The pointer to convert to Ruby string.
 233  * @return [mrb_value] Returns a new Ruby String.
 234  */
 235 MRB_API mrb_value mrb_ptr_to_str(mrb_state *mrb, void *p);
 236
 237 /**
 238  * Returns an object as a Ruby string.
 239  *
 240  * @param mrb The current mruby state.
 241  * @param obj An object to return as a Ruby string.
 242  * @return [mrb_value] An object as a Ruby string.
 243  */
 244 MRB_API mrb_value mrb_obj_as_string(mrb_state *mrb, mrb_value obj);
 245
 246 /**
 247  * Resizes the string's length. Returns the amount of characters
 248  * in the specified by len.
 249  *
 250  * Example:
 251  *
 252  *     int
 253  *     main(int argc,
 254  *          char **argv)
 255  *     {
 256  *         // Variable declaration.
 257  *         mrb_value str;
 258  *
 259  *         mrb_state *mrb = mrb_open();
 260  *         if (!mrb)
 261  *         {
 262  *            // handle error
 263  *         }
 264  *         // Creates a new string.
 265  *         str = mrb_str_new_lit(mrb, "Hello, world!");
 266  *         // Returns 5 characters of
 267  *         mrb_str_resize(mrb, str, 5);
 268  *         mrb_p(mrb, str);
 269  *
 270  *         mrb_close(mrb);
 271  *         return 0;
 272  *      }
 273  *
 274  * Result:
 275  *
 276  *      => "Hello"
 277  *
 278  * @param mrb The current mruby state.
 279  * @param str The Ruby string to resize.
 280  * @param len The length.
 281  * @return [mrb_value] An object as a Ruby string.
 282  */
 283 MRB_API mrb_value mrb_str_resize(mrb_state *mrb, mrb_value str, mrb_int len);
 284
 285 /**
 286  * Returns a sub string.
 287  *
 288  * Example:
 289  *
 290  *     int
 291  *     main(int argc,
 292  *     char const **argv)
 293  *     {
 294  *       // Variable declarations.
 295  *       mrb_value str1;
 296  *       mrb_value str2;
 297  *
 298  *       mrb_state *mrb = mrb_open();
 299  *       if (!mrb)
 300  *       {
 301  *         // handle error
 302  *       }
 303  *       // Creates new string.
 304  *       str1 = mrb_str_new_lit(mrb, "Hello, world!");
 305  *       // Returns a sub-string within the range of 0..2
 306  *       str2 = mrb_str_substr(mrb, str1, 0, 2);
 307  *
 308  *       // Prints sub-string.
 309  *       mrb_p(mrb, str2);
 310  *
 311  *       mrb_close(mrb);
 312  *       return 0;
 313  *     }
 314  *
 315  * Result:
 316  *
 317  *     => "He"
 318  *
 319  * @param mrb The current mruby state.
 320  * @param str Ruby string.
 321  * @param beg The beginning point of the sub-string.
 322  * @param len The end point of the sub-string.
 323  * @return [mrb_value] An object as a Ruby sub-string.
 324  */
 325 MRB_API mrb_value mrb_str_substr(mrb_state *mrb, mrb_value str, mrb_int beg, mrb_int len);
 326
 327 MRB_API mrb_value mrb_str_new_capa(mrb_state *mrb, size_t capa);
 328 #define mrb_str_buf_new(mrb, capa) mrb_str_new_capa(mrb, (capa))
 329
 330 /* NULL terminated C string from mrb_value */
 331 MRB_API const char *mrb_string_cstr(mrb_state *mrb, mrb_value str);
 332 /* NULL terminated C string from mrb_value; `str` will be updated */
 333 MRB_API const char *mrb_string_value_cstr(mrb_state *mrb, mrb_value *str);
 334 /* obsolete: use RSTRING_PTR() */
 335 MRB_API const char *mrb_string_value_ptr(mrb_state *mrb, mrb_value str);
 336 /* obsolete: use RSTRING_LEN() */
 337 MRB_API mrb_int mrb_string_value_len(mrb_state *mrb, mrb_value str);
 338
 339 /**
 340  * Duplicates a string object.
 341  *
 342  *
 343  * @param mrb The current mruby state.
 344  * @param str Ruby string.
 345  * @return [mrb_value] Duplicated Ruby string.
 346  */
 347 MRB_API mrb_value mrb_str_dup(mrb_state *mrb, mrb_value str);
 348
 349 /**
 350  * Returns a symbol from a passed in Ruby string.
 351  *
 352  * @param mrb The current mruby state.
 353  * @param self Ruby string.
 354  * @return [mrb_value] A symbol.
 355  */
 356 MRB_API mrb_value mrb_str_intern(mrb_state *mrb, mrb_value self);
 357
 358 MRB_API mrb_value mrb_str_to_integer(mrb_state *mrb, mrb_value str, mrb_int base, mrb_bool badcheck);
 359 /* obsolete: use mrb_str_to_integer() */
 360 #define mrb_str_to_inum(mrb, str, base, badcheck) mrb_str_to_integer(mrb, str, base, badcheck)
 361 MRB_API double mrb_str_to_dbl(mrb_state *mrb, mrb_value str, mrb_bool badcheck);
 362
 363 /**
 364  * Returns true if the strings match and false if the strings don't match.
 365  *
 366  * @param mrb The current mruby state.
 367  * @param str1 Ruby string to compare.
 368  * @param str2 Ruby string to compare.
 369  * @return [mrb_value] boolean value.
 370  */
 371 MRB_API mrb_bool mrb_str_equal(mrb_state *mrb, mrb_value str1, mrb_value str2);
 372
 373 /**
 374  * Returns a concatenated string comprised of a Ruby string and a C string.
 375  *
 376  * @param mrb The current mruby state.
 377  * @param str Ruby string.
 378  * @param ptr A C string.
 379  * @param len length of C string.
 380  * @return [mrb_value] A Ruby string.
 381  * @see mrb_str_cat_cstr
 382  */
 383 MRB_API mrb_value mrb_str_cat(mrb_state *mrb, mrb_value str, const char *ptr, size_t len);
 384
 385 /**
 386  * Returns a concatenated string comprised of a Ruby string and a C string.
 387  *
 388  * @param mrb The current mruby state.
 389  * @param str Ruby string.
 390  * @param ptr A C string.
 391  * @return [mrb_value] A Ruby string.
 392  * @see mrb_str_cat
 393  */
 394 MRB_API mrb_value mrb_str_cat_cstr(mrb_state *mrb, mrb_value str, const char *ptr);
 395 MRB_API mrb_value mrb_str_cat_str(mrb_state *mrb, mrb_value str, mrb_value str2);
 396 #define mrb_str_cat_lit(mrb, str, lit) mrb_str_cat(mrb, str, lit, mrb_strlen_lit(lit))
 397
 398 /**
 399  * Adds str2 to the end of str1.
 400  */
 401 MRB_API mrb_value mrb_str_append(mrb_state *mrb, mrb_value str, mrb_value str2);
 402
 403 /**
 404  * Returns 0 if both Ruby strings are equal. Returns a value < 0 if Ruby str1 is less than Ruby str2. Returns a value > 0 if Ruby str2 is greater than Ruby str1.
 405  */
 406 MRB_API int mrb_str_cmp(mrb_state *mrb, mrb_value str1, mrb_value str2);
 407
 408 /**
 409  * Returns a newly allocated C string from a Ruby string.
 410  * This is an utility function to pass a Ruby string to C library functions.
 411  *
 412  * - Returned string does not contain any NUL characters (but terminator).
 413  * - It raises an ArgumentError exception if Ruby string contains
 414  *   NUL characters.
 415  * - Returned string will be freed automatically on next GC.
 416  * - Caller can modify returned string without affecting Ruby string
 417  *   (e.g. it can be used for mkstemp(3)).
 418  *
 419  * @param mrb The current mruby state.
 420  * @param str Ruby string. Must be an instance of String.
 421  * @return [char *] A newly allocated C string.
 422  */
 423 MRB_API char *mrb_str_to_cstr(mrb_state *mrb, mrb_value str);
 424
 425 uint32_t mrb_str_hash(mrb_state *mrb, mrb_value str);
 426 mrb_value mrb_str_dump(mrb_state *mrb, mrb_value str);
 427
 428 /**
 429  * Returns a printable version of str, surrounded by quote marks, with special characters escaped.
 430  */
 431 mrb_value mrb_str_inspect(mrb_state *mrb, mrb_value str);
 432
 433 /* For backward compatibility */
 434 #define mrb_str_cat2(mrb, str, ptr) mrb_str_cat_cstr(mrb, str, ptr)
 435 #define mrb_str_buf_cat(mrb, str, ptr, len) mrb_str_cat(mrb, str, ptr, len)
 436 #define mrb_str_buf_append(mrb, str, str2) mrb_str_cat_str(mrb, str, str2)
 437
 438 mrb_bool mrb_str_beg_len(mrb_int str_len, mrb_int *begp, mrb_int *lenp);
 439 mrb_value mrb_str_byte_subseq(mrb_state *mrb, mrb_value str, mrb_int beg, mrb_int len);
 440
 441 #ifdef MRB_UTF8_STRING
 442 mrb_int mrb_utf8len(const char *str, const char *end);
 443 mrb_int mrb_utf8_strlen(const char *str, mrb_int byte_len);
 444 #endif
 445
 446 MRB_END_DECL
 447
 448 #endif  /* MRUBY_STRING_H */