mbed I/F binding for mruby

Dependents:   mruby_mbed_web mirb_mbed

mbed-mruby

How to use

Class

Committer:
mzta
Date:
Mon Apr 13 05:20:15 2015 +0000
Revision:
1:8ccd1d494a4b
Parent:
0:158c61bb030f
- code refactoring.; - add SPI, SPISlave, I2C class to mruby-mbed (Incomplete).

Who changed what in which revision?

UserRevisionLine numberNew contents of line
mzta 0:158c61bb030f 1 /*
mzta 0:158c61bb030f 2 ** string.c - String class
mzta 0:158c61bb030f 3 **
mzta 0:158c61bb030f 4 ** See Copyright Notice in mruby.h
mzta 0:158c61bb030f 5 */
mzta 0:158c61bb030f 6
mzta 0:158c61bb030f 7 #include <ctype.h>
mzta 0:158c61bb030f 8 #include <float.h>
mzta 0:158c61bb030f 9 #include <limits.h>
mzta 0:158c61bb030f 10 #include <stddef.h>
mzta 0:158c61bb030f 11 #include <stdlib.h>
mzta 0:158c61bb030f 12 #include <string.h>
mzta 0:158c61bb030f 13 #include "mruby.h"
mzta 0:158c61bb030f 14 #include "mruby/array.h"
mzta 0:158c61bb030f 15 #include "mruby/class.h"
mzta 0:158c61bb030f 16 #include "mruby/range.h"
mzta 0:158c61bb030f 17 #include "mruby/string.h"
mzta 0:158c61bb030f 18 #include "mruby/re.h"
mzta 0:158c61bb030f 19
mzta 0:158c61bb030f 20 const char mrb_digitmap[] = "0123456789abcdefghijklmnopqrstuvwxyz";
mzta 0:158c61bb030f 21
mzta 0:158c61bb030f 22 typedef struct mrb_shared_string {
mzta 0:158c61bb030f 23 mrb_bool nofree : 1;
mzta 0:158c61bb030f 24 int refcnt;
mzta 0:158c61bb030f 25 char *ptr;
mzta 0:158c61bb030f 26 mrb_int len;
mzta 0:158c61bb030f 27 } mrb_shared_string;
mzta 0:158c61bb030f 28
mzta 0:158c61bb030f 29 static mrb_value str_replace(mrb_state *mrb, struct RString *s1, struct RString *s2);
mzta 0:158c61bb030f 30 static mrb_value mrb_str_subseq(mrb_state *mrb, mrb_value str, mrb_int beg, mrb_int len);
mzta 0:158c61bb030f 31
mzta 0:158c61bb030f 32 MRB_API mrb_int
mzta 0:158c61bb030f 33 mrb_str_strlen(mrb_state *mrb, struct RString *s)
mzta 0:158c61bb030f 34 {
mzta 0:158c61bb030f 35 mrb_int i, max = RSTR_LEN(s);
mzta 0:158c61bb030f 36 char *p = RSTR_PTR(s);
mzta 0:158c61bb030f 37
mzta 0:158c61bb030f 38 if (!p) return 0;
mzta 0:158c61bb030f 39 for (i=0; i<max; i++) {
mzta 0:158c61bb030f 40 if (p[i] == '\0') {
mzta 0:158c61bb030f 41 mrb_raise(mrb, E_ARGUMENT_ERROR, "string contains null byte");
mzta 0:158c61bb030f 42 }
mzta 0:158c61bb030f 43 }
mzta 0:158c61bb030f 44 return max;
mzta 0:158c61bb030f 45 }
mzta 0:158c61bb030f 46
mzta 0:158c61bb030f 47 static inline void
mzta 0:158c61bb030f 48 resize_capa(mrb_state *mrb, struct RString *s, mrb_int capacity)
mzta 0:158c61bb030f 49 {
mzta 0:158c61bb030f 50 if (RSTR_EMBED_P(s)) {
mzta 0:158c61bb030f 51 if (RSTRING_EMBED_LEN_MAX < capacity) {
mzta 0:158c61bb030f 52 char *const tmp = (char *)mrb_malloc(mrb, capacity+1);
mzta 0:158c61bb030f 53 const mrb_int len = RSTR_EMBED_LEN(s);
mzta 0:158c61bb030f 54 memcpy(tmp, s->as.ary, len);
mzta 0:158c61bb030f 55 RSTR_UNSET_EMBED_FLAG(s);
mzta 0:158c61bb030f 56 s->as.heap.ptr = tmp;
mzta 0:158c61bb030f 57 s->as.heap.len = len;
mzta 0:158c61bb030f 58 s->as.heap.aux.capa = capacity;
mzta 0:158c61bb030f 59 }
mzta 0:158c61bb030f 60 }
mzta 0:158c61bb030f 61 else {
mzta 0:158c61bb030f 62 s->as.heap.ptr = (char *)mrb_realloc(mrb, RSTR_PTR(s), capacity+1);
mzta 0:158c61bb030f 63 s->as.heap.aux.capa = capacity;
mzta 0:158c61bb030f 64 }
mzta 0:158c61bb030f 65 }
mzta 0:158c61bb030f 66
mzta 0:158c61bb030f 67 static void
mzta 0:158c61bb030f 68 str_decref(mrb_state *mrb, mrb_shared_string *shared)
mzta 0:158c61bb030f 69 {
mzta 0:158c61bb030f 70 shared->refcnt--;
mzta 0:158c61bb030f 71 if (shared->refcnt == 0) {
mzta 0:158c61bb030f 72 if (!shared->nofree) {
mzta 0:158c61bb030f 73 mrb_free(mrb, shared->ptr);
mzta 0:158c61bb030f 74 }
mzta 0:158c61bb030f 75 mrb_free(mrb, shared);
mzta 0:158c61bb030f 76 }
mzta 0:158c61bb030f 77 }
mzta 0:158c61bb030f 78
mzta 0:158c61bb030f 79 MRB_API void
mzta 0:158c61bb030f 80 mrb_str_modify(mrb_state *mrb, struct RString *s)
mzta 0:158c61bb030f 81 {
mzta 0:158c61bb030f 82 if (RSTR_SHARED_P(s)) {
mzta 0:158c61bb030f 83 mrb_shared_string *shared = s->as.heap.aux.shared;
mzta 0:158c61bb030f 84
mzta 0:158c61bb030f 85 if (shared->refcnt == 1 && s->as.heap.ptr == shared->ptr) {
mzta 0:158c61bb030f 86 s->as.heap.ptr = shared->ptr;
mzta 0:158c61bb030f 87 s->as.heap.aux.capa = shared->len;
mzta 0:158c61bb030f 88 RSTR_PTR(s)[s->as.heap.len] = '\0';
mzta 0:158c61bb030f 89 mrb_free(mrb, shared);
mzta 0:158c61bb030f 90 }
mzta 0:158c61bb030f 91 else {
mzta 0:158c61bb030f 92 char *ptr, *p;
mzta 0:158c61bb030f 93 mrb_int len;
mzta 0:158c61bb030f 94
mzta 0:158c61bb030f 95 p = RSTR_PTR(s);
mzta 0:158c61bb030f 96 len = s->as.heap.len;
mzta 0:158c61bb030f 97 ptr = (char *)mrb_malloc(mrb, (size_t)len + 1);
mzta 0:158c61bb030f 98 if (p) {
mzta 0:158c61bb030f 99 memcpy(ptr, p, len);
mzta 0:158c61bb030f 100 }
mzta 0:158c61bb030f 101 ptr[len] = '\0';
mzta 0:158c61bb030f 102 s->as.heap.ptr = ptr;
mzta 0:158c61bb030f 103 s->as.heap.aux.capa = len;
mzta 0:158c61bb030f 104 str_decref(mrb, shared);
mzta 0:158c61bb030f 105 }
mzta 0:158c61bb030f 106 RSTR_UNSET_SHARED_FLAG(s);
mzta 0:158c61bb030f 107 return;
mzta 0:158c61bb030f 108 }
mzta 0:158c61bb030f 109 if (RSTR_NOFREE_P(s)) {
mzta 0:158c61bb030f 110 char *p = s->as.heap.ptr;
mzta 0:158c61bb030f 111
mzta 0:158c61bb030f 112 s->as.heap.ptr = (char *)mrb_malloc(mrb, (size_t)s->as.heap.len+1);
mzta 0:158c61bb030f 113 if (p) {
mzta 0:158c61bb030f 114 memcpy(RSTR_PTR(s), p, s->as.heap.len);
mzta 0:158c61bb030f 115 }
mzta 0:158c61bb030f 116 RSTR_PTR(s)[s->as.heap.len] = '\0';
mzta 0:158c61bb030f 117 s->as.heap.aux.capa = s->as.heap.len;
mzta 0:158c61bb030f 118 RSTR_UNSET_NOFREE_FLAG(s);
mzta 0:158c61bb030f 119 return;
mzta 0:158c61bb030f 120 }
mzta 0:158c61bb030f 121 }
mzta 0:158c61bb030f 122
mzta 0:158c61bb030f 123 MRB_API mrb_value
mzta 0:158c61bb030f 124 mrb_str_resize(mrb_state *mrb, mrb_value str, mrb_int len)
mzta 0:158c61bb030f 125 {
mzta 0:158c61bb030f 126 mrb_int slen;
mzta 0:158c61bb030f 127 struct RString *s = mrb_str_ptr(str);
mzta 0:158c61bb030f 128
mzta 0:158c61bb030f 129 mrb_str_modify(mrb, s);
mzta 0:158c61bb030f 130 slen = RSTR_LEN(s);
mzta 0:158c61bb030f 131 if (len != slen) {
mzta 0:158c61bb030f 132 if (slen < len || slen - len > 256) {
mzta 0:158c61bb030f 133 resize_capa(mrb, s, len);
mzta 0:158c61bb030f 134 }
mzta 0:158c61bb030f 135 RSTR_SET_LEN(s, len);
mzta 0:158c61bb030f 136 RSTR_PTR(s)[len] = '\0'; /* sentinel */
mzta 0:158c61bb030f 137 }
mzta 0:158c61bb030f 138 return str;
mzta 0:158c61bb030f 139 }
mzta 0:158c61bb030f 140
mzta 0:158c61bb030f 141 #define mrb_obj_alloc_string(mrb) ((struct RString*)mrb_obj_alloc((mrb), MRB_TT_STRING, (mrb)->string_class))
mzta 0:158c61bb030f 142
mzta 0:158c61bb030f 143 static struct RString*
mzta 0:158c61bb030f 144 str_new_static(mrb_state *mrb, const char *p, size_t len)
mzta 0:158c61bb030f 145 {
mzta 0:158c61bb030f 146 struct RString *s;
mzta 0:158c61bb030f 147
mzta 0:158c61bb030f 148 if (len >= MRB_INT_MAX) {
mzta 0:158c61bb030f 149 mrb_raise(mrb, E_ARGUMENT_ERROR, "string size too big");
mzta 0:158c61bb030f 150 }
mzta 0:158c61bb030f 151 s = mrb_obj_alloc_string(mrb);
mzta 0:158c61bb030f 152 s->as.heap.len = len;
mzta 0:158c61bb030f 153 s->as.heap.aux.capa = 0; /* nofree */
mzta 0:158c61bb030f 154 s->as.heap.ptr = (char *)p;
mzta 0:158c61bb030f 155 s->flags = MRB_STR_NOFREE;
mzta 0:158c61bb030f 156
mzta 0:158c61bb030f 157 return s;
mzta 0:158c61bb030f 158 }
mzta 0:158c61bb030f 159
mzta 0:158c61bb030f 160 static struct RString*
mzta 0:158c61bb030f 161 str_new(mrb_state *mrb, const char *p, size_t len)
mzta 0:158c61bb030f 162 {
mzta 0:158c61bb030f 163 struct RString *s;
mzta 0:158c61bb030f 164
mzta 0:158c61bb030f 165 if (mrb_ro_data_p(p)) {
mzta 0:158c61bb030f 166 return str_new_static(mrb, p, len);
mzta 0:158c61bb030f 167 }
mzta 0:158c61bb030f 168 s = mrb_obj_alloc_string(mrb);
mzta 0:158c61bb030f 169 if (len < RSTRING_EMBED_LEN_MAX) {
mzta 0:158c61bb030f 170 RSTR_SET_EMBED_FLAG(s);
mzta 0:158c61bb030f 171 RSTR_SET_EMBED_LEN(s, len);
mzta 0:158c61bb030f 172 if (p) {
mzta 0:158c61bb030f 173 memcpy(s->as.ary, p, len);
mzta 0:158c61bb030f 174 }
mzta 0:158c61bb030f 175 } else {
mzta 0:158c61bb030f 176 if (len >= MRB_INT_MAX) {
mzta 0:158c61bb030f 177 mrb_raise(mrb, E_ARGUMENT_ERROR, "string size too big");
mzta 0:158c61bb030f 178 }
mzta 0:158c61bb030f 179 s->as.heap.len = len;
mzta 0:158c61bb030f 180 s->as.heap.aux.capa = len;
mzta 0:158c61bb030f 181 s->as.heap.ptr = (char *)mrb_malloc(mrb, len+1);
mzta 0:158c61bb030f 182 if (p) {
mzta 0:158c61bb030f 183 memcpy(s->as.heap.ptr, p, len);
mzta 0:158c61bb030f 184 }
mzta 0:158c61bb030f 185 }
mzta 0:158c61bb030f 186 RSTR_PTR(s)[len] = '\0';
mzta 0:158c61bb030f 187 return s;
mzta 0:158c61bb030f 188 }
mzta 0:158c61bb030f 189
mzta 0:158c61bb030f 190 static inline void
mzta 0:158c61bb030f 191 str_with_class(mrb_state *mrb, struct RString *s, mrb_value obj)
mzta 0:158c61bb030f 192 {
mzta 0:158c61bb030f 193 s->c = mrb_str_ptr(obj)->c;
mzta 0:158c61bb030f 194 }
mzta 0:158c61bb030f 195
mzta 0:158c61bb030f 196 static mrb_value
mzta 0:158c61bb030f 197 mrb_str_new_empty(mrb_state *mrb, mrb_value str)
mzta 0:158c61bb030f 198 {
mzta 0:158c61bb030f 199 struct RString *s = str_new(mrb, 0, 0);
mzta 0:158c61bb030f 200
mzta 0:158c61bb030f 201 str_with_class(mrb, s, str);
mzta 0:158c61bb030f 202 return mrb_obj_value(s);
mzta 0:158c61bb030f 203 }
mzta 0:158c61bb030f 204
mzta 0:158c61bb030f 205 #ifndef MRB_STR_BUF_MIN_SIZE
mzta 0:158c61bb030f 206 # define MRB_STR_BUF_MIN_SIZE 128
mzta 0:158c61bb030f 207 #endif
mzta 0:158c61bb030f 208
mzta 0:158c61bb030f 209 MRB_API mrb_value
mzta 0:158c61bb030f 210 mrb_str_buf_new(mrb_state *mrb, size_t capa)
mzta 0:158c61bb030f 211 {
mzta 0:158c61bb030f 212 struct RString *s;
mzta 0:158c61bb030f 213
mzta 0:158c61bb030f 214 s = mrb_obj_alloc_string(mrb);
mzta 0:158c61bb030f 215
mzta 0:158c61bb030f 216 if (capa >= MRB_INT_MAX) {
mzta 0:158c61bb030f 217 mrb_raise(mrb, E_ARGUMENT_ERROR, "string capacity size too big");
mzta 0:158c61bb030f 218 }
mzta 0:158c61bb030f 219 if (capa < MRB_STR_BUF_MIN_SIZE) {
mzta 0:158c61bb030f 220 capa = MRB_STR_BUF_MIN_SIZE;
mzta 0:158c61bb030f 221 }
mzta 0:158c61bb030f 222 s->as.heap.len = 0;
mzta 0:158c61bb030f 223 s->as.heap.aux.capa = capa;
mzta 0:158c61bb030f 224 s->as.heap.ptr = (char *)mrb_malloc(mrb, capa+1);
mzta 0:158c61bb030f 225 RSTR_PTR(s)[0] = '\0';
mzta 0:158c61bb030f 226
mzta 0:158c61bb030f 227 return mrb_obj_value(s);
mzta 0:158c61bb030f 228 }
mzta 0:158c61bb030f 229
mzta 0:158c61bb030f 230 static void
mzta 0:158c61bb030f 231 str_buf_cat(mrb_state *mrb, struct RString *s, const char *ptr, size_t len)
mzta 0:158c61bb030f 232 {
mzta 0:158c61bb030f 233 size_t capa;
mzta 0:158c61bb030f 234 size_t total;
mzta 0:158c61bb030f 235 ptrdiff_t off = -1;
mzta 0:158c61bb030f 236
mzta 0:158c61bb030f 237 if (len == 0) return;
mzta 0:158c61bb030f 238 mrb_str_modify(mrb, s);
mzta 0:158c61bb030f 239 if (ptr >= RSTR_PTR(s) && ptr <= RSTR_PTR(s) + (size_t)RSTR_LEN(s)) {
mzta 0:158c61bb030f 240 off = ptr - RSTR_PTR(s);
mzta 0:158c61bb030f 241 }
mzta 0:158c61bb030f 242
mzta 0:158c61bb030f 243 if (RSTR_EMBED_P(s))
mzta 0:158c61bb030f 244 capa = RSTRING_EMBED_LEN_MAX;
mzta 0:158c61bb030f 245 else
mzta 0:158c61bb030f 246 capa = s->as.heap.aux.capa;
mzta 0:158c61bb030f 247
mzta 0:158c61bb030f 248 if (RSTR_LEN(s) >= MRB_INT_MAX - (mrb_int)len) {
mzta 0:158c61bb030f 249 mrb_raise(mrb, E_ARGUMENT_ERROR, "string size too big");
mzta 0:158c61bb030f 250 }
mzta 0:158c61bb030f 251 total = RSTR_LEN(s)+len;
mzta 0:158c61bb030f 252 if (capa <= total) {
mzta 0:158c61bb030f 253 while (total > capa) {
mzta 0:158c61bb030f 254 if (capa + 1 >= MRB_INT_MAX / 2) {
mzta 0:158c61bb030f 255 capa = (total + 4095) / 4096;
mzta 0:158c61bb030f 256 break;
mzta 0:158c61bb030f 257 }
mzta 0:158c61bb030f 258 capa = (capa + 1) * 2;
mzta 0:158c61bb030f 259 }
mzta 0:158c61bb030f 260 resize_capa(mrb, s, capa);
mzta 0:158c61bb030f 261 }
mzta 0:158c61bb030f 262 if (off != -1) {
mzta 0:158c61bb030f 263 ptr = RSTR_PTR(s) + off;
mzta 0:158c61bb030f 264 }
mzta 0:158c61bb030f 265 memcpy(RSTR_PTR(s) + RSTR_LEN(s), ptr, len);
mzta 0:158c61bb030f 266 mrb_assert_int_fit(size_t, total, mrb_int, MRB_INT_MAX);
mzta 0:158c61bb030f 267 RSTR_SET_LEN(s, total);
mzta 0:158c61bb030f 268 RSTR_PTR(s)[total] = '\0'; /* sentinel */
mzta 0:158c61bb030f 269 }
mzta 0:158c61bb030f 270
mzta 0:158c61bb030f 271 MRB_API mrb_value
mzta 0:158c61bb030f 272 mrb_str_new(mrb_state *mrb, const char *p, size_t len)
mzta 0:158c61bb030f 273 {
mzta 0:158c61bb030f 274 return mrb_obj_value(str_new(mrb, p, len));
mzta 0:158c61bb030f 275 }
mzta 0:158c61bb030f 276
mzta 0:158c61bb030f 277 /*
mzta 0:158c61bb030f 278 * call-seq: (Caution! NULL string)
mzta 0:158c61bb030f 279 * String.new(str="") => new_str
mzta 0:158c61bb030f 280 *
mzta 0:158c61bb030f 281 * Returns a new string object containing a copy of <i>str</i>.
mzta 0:158c61bb030f 282 */
mzta 0:158c61bb030f 283
mzta 0:158c61bb030f 284 MRB_API mrb_value
mzta 0:158c61bb030f 285 mrb_str_new_cstr(mrb_state *mrb, const char *p)
mzta 0:158c61bb030f 286 {
mzta 0:158c61bb030f 287 struct RString *s;
mzta 0:158c61bb030f 288 size_t len;
mzta 0:158c61bb030f 289
mzta 0:158c61bb030f 290 if (p) {
mzta 0:158c61bb030f 291 len = strlen(p);
mzta 0:158c61bb030f 292 }
mzta 0:158c61bb030f 293 else {
mzta 0:158c61bb030f 294 len = 0;
mzta 0:158c61bb030f 295 }
mzta 0:158c61bb030f 296
mzta 0:158c61bb030f 297 s = str_new(mrb, p, len);
mzta 0:158c61bb030f 298
mzta 0:158c61bb030f 299 return mrb_obj_value(s);
mzta 0:158c61bb030f 300 }
mzta 0:158c61bb030f 301
mzta 0:158c61bb030f 302 MRB_API mrb_value
mzta 0:158c61bb030f 303 mrb_str_new_static(mrb_state *mrb, const char *p, size_t len)
mzta 0:158c61bb030f 304 {
mzta 0:158c61bb030f 305 struct RString *s = str_new_static(mrb, p, len);
mzta 0:158c61bb030f 306 return mrb_obj_value(s);
mzta 0:158c61bb030f 307 }
mzta 0:158c61bb030f 308
mzta 0:158c61bb030f 309 void
mzta 0:158c61bb030f 310 mrb_gc_free_str(mrb_state *mrb, struct RString *str)
mzta 0:158c61bb030f 311 {
mzta 0:158c61bb030f 312 if (RSTR_EMBED_P(str))
mzta 0:158c61bb030f 313 /* no code */;
mzta 0:158c61bb030f 314 else if (RSTR_SHARED_P(str))
mzta 0:158c61bb030f 315 str_decref(mrb, str->as.heap.aux.shared);
mzta 0:158c61bb030f 316 else if (!RSTR_NOFREE_P(str))
mzta 0:158c61bb030f 317 mrb_free(mrb, str->as.heap.ptr);
mzta 0:158c61bb030f 318 }
mzta 0:158c61bb030f 319
mzta 0:158c61bb030f 320 MRB_API char*
mzta 0:158c61bb030f 321 mrb_str_to_cstr(mrb_state *mrb, mrb_value str0)
mzta 0:158c61bb030f 322 {
mzta 0:158c61bb030f 323 struct RString *s;
mzta 0:158c61bb030f 324
mzta 0:158c61bb030f 325 if (!mrb_string_p(str0)) {
mzta 0:158c61bb030f 326 mrb_raise(mrb, E_TYPE_ERROR, "expected String");
mzta 0:158c61bb030f 327 }
mzta 0:158c61bb030f 328
mzta 0:158c61bb030f 329 s = str_new(mrb, RSTRING_PTR(str0), RSTRING_LEN(str0));
mzta 0:158c61bb030f 330 if ((strlen(RSTR_PTR(s)) ^ RSTR_LEN(s)) != 0) {
mzta 0:158c61bb030f 331 mrb_raise(mrb, E_ARGUMENT_ERROR, "string contains null byte");
mzta 0:158c61bb030f 332 }
mzta 0:158c61bb030f 333 return RSTR_PTR(s);
mzta 0:158c61bb030f 334 }
mzta 0:158c61bb030f 335
mzta 0:158c61bb030f 336 static void
mzta 0:158c61bb030f 337 str_make_shared(mrb_state *mrb, struct RString *s)
mzta 0:158c61bb030f 338 {
mzta 0:158c61bb030f 339 if (!RSTR_SHARED_P(s)) {
mzta 0:158c61bb030f 340 mrb_shared_string *shared = (mrb_shared_string *)mrb_malloc(mrb, sizeof(mrb_shared_string));
mzta 0:158c61bb030f 341
mzta 0:158c61bb030f 342 shared->refcnt = 1;
mzta 0:158c61bb030f 343 if (RSTR_EMBED_P(s)) {
mzta 0:158c61bb030f 344 const mrb_int len = RSTR_EMBED_LEN(s);
mzta 0:158c61bb030f 345 char *const tmp = (char *)mrb_malloc(mrb, len+1);
mzta 0:158c61bb030f 346 memcpy(tmp, s->as.ary, len);
mzta 0:158c61bb030f 347 tmp[len] = '\0';
mzta 0:158c61bb030f 348 RSTR_UNSET_EMBED_FLAG(s);
mzta 0:158c61bb030f 349 s->as.heap.ptr = tmp;
mzta 0:158c61bb030f 350 s->as.heap.len = len;
mzta 0:158c61bb030f 351 shared->nofree = FALSE;
mzta 0:158c61bb030f 352 shared->ptr = s->as.heap.ptr;
mzta 0:158c61bb030f 353 }
mzta 0:158c61bb030f 354 else if (RSTR_NOFREE_P(s)) {
mzta 0:158c61bb030f 355 shared->nofree = TRUE;
mzta 0:158c61bb030f 356 shared->ptr = s->as.heap.ptr;
mzta 0:158c61bb030f 357 RSTR_UNSET_NOFREE_FLAG(s);
mzta 0:158c61bb030f 358 }
mzta 0:158c61bb030f 359 else {
mzta 0:158c61bb030f 360 shared->nofree = FALSE;
mzta 0:158c61bb030f 361 if (s->as.heap.aux.capa > s->as.heap.len) {
mzta 0:158c61bb030f 362 s->as.heap.ptr = shared->ptr = (char *)mrb_realloc(mrb, s->as.heap.ptr, s->as.heap.len+1);
mzta 0:158c61bb030f 363 }
mzta 0:158c61bb030f 364 else {
mzta 0:158c61bb030f 365 shared->ptr = s->as.heap.ptr;
mzta 0:158c61bb030f 366 }
mzta 0:158c61bb030f 367 }
mzta 0:158c61bb030f 368 shared->len = s->as.heap.len;
mzta 0:158c61bb030f 369 s->as.heap.aux.shared = shared;
mzta 0:158c61bb030f 370 RSTR_SET_SHARED_FLAG(s);
mzta 0:158c61bb030f 371 }
mzta 0:158c61bb030f 372 }
mzta 0:158c61bb030f 373
mzta 0:158c61bb030f 374 /*
mzta 0:158c61bb030f 375 * call-seq: (Caution! String("abcd") change)
mzta 0:158c61bb030f 376 * String("abcdefg") = String("abcd") + String("efg")
mzta 0:158c61bb030f 377 *
mzta 0:158c61bb030f 378 * Returns a new string object containing a copy of <i>str</i>.
mzta 0:158c61bb030f 379 */
mzta 0:158c61bb030f 380 MRB_API void
mzta 0:158c61bb030f 381 mrb_str_concat(mrb_state *mrb, mrb_value self, mrb_value other)
mzta 0:158c61bb030f 382 {
mzta 0:158c61bb030f 383 struct RString *s1 = mrb_str_ptr(self), *s2;
mzta 0:158c61bb030f 384 mrb_int len;
mzta 0:158c61bb030f 385
mzta 0:158c61bb030f 386 mrb_str_modify(mrb, s1);
mzta 0:158c61bb030f 387 if (!mrb_string_p(other)) {
mzta 0:158c61bb030f 388 other = mrb_str_to_str(mrb, other);
mzta 0:158c61bb030f 389 }
mzta 0:158c61bb030f 390 s2 = mrb_str_ptr(other);
mzta 0:158c61bb030f 391 len = RSTR_LEN(s1) + RSTR_LEN(s2);
mzta 0:158c61bb030f 392
mzta 0:158c61bb030f 393 if (RSTRING_CAPA(self) < len) {
mzta 0:158c61bb030f 394 resize_capa(mrb, s1, len);
mzta 0:158c61bb030f 395 }
mzta 0:158c61bb030f 396 memcpy(RSTR_PTR(s1)+RSTR_LEN(s1), RSTR_PTR(s2), RSTR_LEN(s2));
mzta 0:158c61bb030f 397 RSTR_SET_LEN(s1, len);
mzta 0:158c61bb030f 398 RSTR_PTR(s1)[len] = '\0';
mzta 0:158c61bb030f 399 }
mzta 0:158c61bb030f 400
mzta 0:158c61bb030f 401 /*
mzta 0:158c61bb030f 402 * call-seq: (Caution! String("abcd") remain)
mzta 0:158c61bb030f 403 * String("abcdefg") = String("abcd") + String("efg")
mzta 0:158c61bb030f 404 *
mzta 0:158c61bb030f 405 * Returns a new string object containing a copy of <i>str</i>.
mzta 0:158c61bb030f 406 */
mzta 0:158c61bb030f 407 MRB_API mrb_value
mzta 0:158c61bb030f 408 mrb_str_plus(mrb_state *mrb, mrb_value a, mrb_value b)
mzta 0:158c61bb030f 409 {
mzta 0:158c61bb030f 410 struct RString *s = mrb_str_ptr(a);
mzta 0:158c61bb030f 411 struct RString *s2 = mrb_str_ptr(b);
mzta 0:158c61bb030f 412 struct RString *t;
mzta 0:158c61bb030f 413
mzta 0:158c61bb030f 414 t = str_new(mrb, 0, RSTR_LEN(s) + RSTR_LEN(s2));
mzta 0:158c61bb030f 415 memcpy(RSTR_PTR(t), RSTR_PTR(s), RSTR_LEN(s));
mzta 0:158c61bb030f 416 memcpy(RSTR_PTR(t) + RSTR_LEN(s), RSTR_PTR(s2), RSTR_LEN(s2));
mzta 0:158c61bb030f 417
mzta 0:158c61bb030f 418 return mrb_obj_value(t);
mzta 0:158c61bb030f 419 }
mzta 0:158c61bb030f 420
mzta 0:158c61bb030f 421 /* 15.2.10.5.2 */
mzta 0:158c61bb030f 422
mzta 0:158c61bb030f 423 /*
mzta 0:158c61bb030f 424 * call-seq: (Caution! String("abcd") remain) for stack_argument
mzta 0:158c61bb030f 425 * String("abcdefg") = String("abcd") + String("efg")
mzta 0:158c61bb030f 426 *
mzta 0:158c61bb030f 427 * Returns a new string object containing a copy of <i>str</i>.
mzta 0:158c61bb030f 428 */
mzta 0:158c61bb030f 429 static mrb_value
mzta 0:158c61bb030f 430 mrb_str_plus_m(mrb_state *mrb, mrb_value self)
mzta 0:158c61bb030f 431 {
mzta 0:158c61bb030f 432 mrb_value str;
mzta 0:158c61bb030f 433
mzta 0:158c61bb030f 434 mrb_get_args(mrb, "S", &str);
mzta 0:158c61bb030f 435 return mrb_str_plus(mrb, self, str);
mzta 0:158c61bb030f 436 }
mzta 0:158c61bb030f 437
mzta 0:158c61bb030f 438 /* 15.2.10.5.26 */
mzta 0:158c61bb030f 439 /* 15.2.10.5.33 */
mzta 0:158c61bb030f 440 /*
mzta 0:158c61bb030f 441 * call-seq:
mzta 0:158c61bb030f 442 * len = strlen(String("abcd"))
mzta 0:158c61bb030f 443 *
mzta 0:158c61bb030f 444 * Returns the length of string.
mzta 0:158c61bb030f 445 */
mzta 0:158c61bb030f 446 static mrb_value
mzta 0:158c61bb030f 447 mrb_str_size(mrb_state *mrb, mrb_value self)
mzta 0:158c61bb030f 448 {
mzta 0:158c61bb030f 449 struct RString *s = mrb_str_ptr(self);
mzta 0:158c61bb030f 450 return mrb_fixnum_value(RSTR_LEN(s));
mzta 0:158c61bb030f 451 }
mzta 0:158c61bb030f 452
mzta 0:158c61bb030f 453 /* 15.2.10.5.1 */
mzta 0:158c61bb030f 454 /*
mzta 0:158c61bb030f 455 * call-seq:
mzta 0:158c61bb030f 456 * str * integer => new_str
mzta 0:158c61bb030f 457 *
mzta 0:158c61bb030f 458 * Copy---Returns a new <code>String</code> containing <i>integer</i> copies of
mzta 0:158c61bb030f 459 * the receiver.
mzta 0:158c61bb030f 460 *
mzta 0:158c61bb030f 461 * "Ho! " * 3 #=> "Ho! Ho! Ho! "
mzta 0:158c61bb030f 462 */
mzta 0:158c61bb030f 463 static mrb_value
mzta 0:158c61bb030f 464 mrb_str_times(mrb_state *mrb, mrb_value self)
mzta 0:158c61bb030f 465 {
mzta 0:158c61bb030f 466 mrb_int n,len,times;
mzta 0:158c61bb030f 467 struct RString *str2;
mzta 0:158c61bb030f 468 char *p;
mzta 0:158c61bb030f 469
mzta 0:158c61bb030f 470 mrb_get_args(mrb, "i", &times);
mzta 0:158c61bb030f 471 if (times < 0) {
mzta 0:158c61bb030f 472 mrb_raise(mrb, E_ARGUMENT_ERROR, "negative argument");
mzta 0:158c61bb030f 473 }
mzta 0:158c61bb030f 474 if (times && MRB_INT_MAX / times < RSTRING_LEN(self)) {
mzta 0:158c61bb030f 475 mrb_raise(mrb, E_ARGUMENT_ERROR, "argument too big");
mzta 0:158c61bb030f 476 }
mzta 0:158c61bb030f 477
mzta 0:158c61bb030f 478 len = RSTRING_LEN(self)*times;
mzta 0:158c61bb030f 479 str2 = str_new(mrb, 0, len);
mzta 0:158c61bb030f 480 str_with_class(mrb, str2, self);
mzta 0:158c61bb030f 481 p = RSTR_PTR(str2);
mzta 0:158c61bb030f 482 if (len > 0) {
mzta 0:158c61bb030f 483 n = RSTRING_LEN(self);
mzta 0:158c61bb030f 484 memcpy(p, RSTRING_PTR(self), n);
mzta 0:158c61bb030f 485 while (n <= len/2) {
mzta 0:158c61bb030f 486 memcpy(p + n, p, n);
mzta 0:158c61bb030f 487 n *= 2;
mzta 0:158c61bb030f 488 }
mzta 0:158c61bb030f 489 memcpy(p + n, p, len-n);
mzta 0:158c61bb030f 490 }
mzta 0:158c61bb030f 491 p[RSTR_LEN(str2)] = '\0';
mzta 0:158c61bb030f 492
mzta 0:158c61bb030f 493 return mrb_obj_value(str2);
mzta 0:158c61bb030f 494 }
mzta 0:158c61bb030f 495 /* -------------------------------------------------------------- */
mzta 0:158c61bb030f 496
mzta 0:158c61bb030f 497 #define lesser(a,b) (((a)>(b))?(b):(a))
mzta 0:158c61bb030f 498
mzta 0:158c61bb030f 499 /* ---------------------------*/
mzta 0:158c61bb030f 500 /*
mzta 0:158c61bb030f 501 * call-seq:
mzta 0:158c61bb030f 502 * mrb_value str1 <=> mrb_value str2 => int
mzta 0:158c61bb030f 503 * > 1
mzta 0:158c61bb030f 504 * = 0
mzta 0:158c61bb030f 505 * < -1
mzta 0:158c61bb030f 506 */
mzta 0:158c61bb030f 507 MRB_API int
mzta 0:158c61bb030f 508 mrb_str_cmp(mrb_state *mrb, mrb_value str1, mrb_value str2)
mzta 0:158c61bb030f 509 {
mzta 0:158c61bb030f 510 mrb_int len;
mzta 0:158c61bb030f 511 mrb_int retval;
mzta 0:158c61bb030f 512 struct RString *s1 = mrb_str_ptr(str1);
mzta 0:158c61bb030f 513 struct RString *s2 = mrb_str_ptr(str2);
mzta 0:158c61bb030f 514
mzta 0:158c61bb030f 515 len = lesser(RSTR_LEN(s1), RSTR_LEN(s2));
mzta 0:158c61bb030f 516 retval = memcmp(RSTR_PTR(s1), RSTR_PTR(s2), len);
mzta 0:158c61bb030f 517 if (retval == 0) {
mzta 0:158c61bb030f 518 if (RSTR_LEN(s1) == RSTR_LEN(s2)) return 0;
mzta 0:158c61bb030f 519 if (RSTR_LEN(s1) > RSTR_LEN(s2)) return 1;
mzta 0:158c61bb030f 520 return -1;
mzta 0:158c61bb030f 521 }
mzta 0:158c61bb030f 522 if (retval > 0) return 1;
mzta 0:158c61bb030f 523 return -1;
mzta 0:158c61bb030f 524 }
mzta 0:158c61bb030f 525
mzta 0:158c61bb030f 526 /* 15.2.10.5.3 */
mzta 0:158c61bb030f 527
mzta 0:158c61bb030f 528 /*
mzta 0:158c61bb030f 529 * call-seq:
mzta 0:158c61bb030f 530 * str <=> other_str => -1, 0, +1
mzta 0:158c61bb030f 531 *
mzta 0:158c61bb030f 532 * Comparison---Returns -1 if <i>other_str</i> is less than, 0 if
mzta 0:158c61bb030f 533 * <i>other_str</i> is equal to, and +1 if <i>other_str</i> is greater than
mzta 0:158c61bb030f 534 * <i>str</i>. If the strings are of different lengths, and the strings are
mzta 0:158c61bb030f 535 * equal when compared up to the shortest length, then the longer string is
mzta 0:158c61bb030f 536 * considered greater than the shorter one. If the variable <code>$=</code> is
mzta 0:158c61bb030f 537 * <code>false</code>, the comparison is based on comparing the binary values
mzta 0:158c61bb030f 538 * of each character in the string. In older versions of Ruby, setting
mzta 0:158c61bb030f 539 * <code>$=</code> allowed case-insensitive comparisons; this is now deprecated
mzta 0:158c61bb030f 540 * in favor of using <code>String#casecmp</code>.
mzta 0:158c61bb030f 541 *
mzta 0:158c61bb030f 542 * <code><=></code> is the basis for the methods <code><</code>,
mzta 0:158c61bb030f 543 * <code><=</code>, <code>></code>, <code>>=</code>, and <code>between?</code>,
mzta 0:158c61bb030f 544 * included from module <code>Comparable</code>. The method
mzta 0:158c61bb030f 545 * <code>String#==</code> does not use <code>Comparable#==</code>.
mzta 0:158c61bb030f 546 *
mzta 0:158c61bb030f 547 * "abcdef" <=> "abcde" #=> 1
mzta 0:158c61bb030f 548 * "abcdef" <=> "abcdef" #=> 0
mzta 0:158c61bb030f 549 * "abcdef" <=> "abcdefg" #=> -1
mzta 0:158c61bb030f 550 * "abcdef" <=> "ABCDEF" #=> 1
mzta 0:158c61bb030f 551 */
mzta 0:158c61bb030f 552 static mrb_value
mzta 0:158c61bb030f 553 mrb_str_cmp_m(mrb_state *mrb, mrb_value str1)
mzta 0:158c61bb030f 554 {
mzta 0:158c61bb030f 555 mrb_value str2;
mzta 0:158c61bb030f 556 mrb_int result;
mzta 0:158c61bb030f 557
mzta 0:158c61bb030f 558 mrb_get_args(mrb, "o", &str2);
mzta 0:158c61bb030f 559 if (!mrb_string_p(str2)) {
mzta 0:158c61bb030f 560 if (!mrb_respond_to(mrb, str2, mrb_intern_lit(mrb, "to_s"))) {
mzta 0:158c61bb030f 561 return mrb_nil_value();
mzta 0:158c61bb030f 562 }
mzta 0:158c61bb030f 563 else if (!mrb_respond_to(mrb, str2, mrb_intern_lit(mrb, "<=>"))) {
mzta 0:158c61bb030f 564 return mrb_nil_value();
mzta 0:158c61bb030f 565 }
mzta 0:158c61bb030f 566 else {
mzta 0:158c61bb030f 567 mrb_value tmp = mrb_funcall(mrb, str2, "<=>", 1, str1);
mzta 0:158c61bb030f 568
mzta 0:158c61bb030f 569 if (mrb_nil_p(tmp)) return mrb_nil_value();
mzta 0:158c61bb030f 570 if (!mrb_fixnum(tmp)) {
mzta 0:158c61bb030f 571 return mrb_funcall(mrb, mrb_fixnum_value(0), "-", 1, tmp);
mzta 0:158c61bb030f 572 }
mzta 0:158c61bb030f 573 result = -mrb_fixnum(tmp);
mzta 0:158c61bb030f 574 }
mzta 0:158c61bb030f 575 }
mzta 0:158c61bb030f 576 else {
mzta 0:158c61bb030f 577 result = mrb_str_cmp(mrb, str1, str2);
mzta 0:158c61bb030f 578 }
mzta 0:158c61bb030f 579 return mrb_fixnum_value(result);
mzta 0:158c61bb030f 580 }
mzta 0:158c61bb030f 581
mzta 0:158c61bb030f 582 static mrb_bool
mzta 0:158c61bb030f 583 str_eql(mrb_state *mrb, const mrb_value str1, const mrb_value str2)
mzta 0:158c61bb030f 584 {
mzta 0:158c61bb030f 585 const mrb_int len = RSTRING_LEN(str1);
mzta 0:158c61bb030f 586
mzta 0:158c61bb030f 587 if (len != RSTRING_LEN(str2)) return FALSE;
mzta 0:158c61bb030f 588 if (memcmp(RSTRING_PTR(str1), RSTRING_PTR(str2), (size_t)len) == 0)
mzta 0:158c61bb030f 589 return TRUE;
mzta 0:158c61bb030f 590 return FALSE;
mzta 0:158c61bb030f 591 }
mzta 0:158c61bb030f 592
mzta 0:158c61bb030f 593 MRB_API mrb_bool
mzta 0:158c61bb030f 594 mrb_str_equal(mrb_state *mrb, mrb_value str1, mrb_value str2)
mzta 0:158c61bb030f 595 {
mzta 0:158c61bb030f 596 if (mrb_immediate_p(str2)) return FALSE;
mzta 0:158c61bb030f 597 if (!mrb_string_p(str2)) {
mzta 0:158c61bb030f 598 if (mrb_nil_p(str2)) return FALSE;
mzta 0:158c61bb030f 599 if (!mrb_respond_to(mrb, str2, mrb_intern_lit(mrb, "to_str"))) {
mzta 0:158c61bb030f 600 return FALSE;
mzta 0:158c61bb030f 601 }
mzta 0:158c61bb030f 602 str2 = mrb_funcall(mrb, str2, "to_str", 0);
mzta 0:158c61bb030f 603 return mrb_equal(mrb, str2, str1);
mzta 0:158c61bb030f 604 }
mzta 0:158c61bb030f 605 return str_eql(mrb, str1, str2);
mzta 0:158c61bb030f 606 }
mzta 0:158c61bb030f 607
mzta 0:158c61bb030f 608 /* 15.2.10.5.4 */
mzta 0:158c61bb030f 609 /*
mzta 0:158c61bb030f 610 * call-seq:
mzta 0:158c61bb030f 611 * str == obj => true or false
mzta 0:158c61bb030f 612 *
mzta 0:158c61bb030f 613 * Equality---
mzta 0:158c61bb030f 614 * If <i>obj</i> is not a <code>String</code>, returns <code>false</code>.
mzta 0:158c61bb030f 615 * Otherwise, returns <code>false</code> or <code>true</code>
mzta 0:158c61bb030f 616 *
mzta 0:158c61bb030f 617 * caution:if <i>str</i> <code><=></code> <i>obj</i> returns zero.
mzta 0:158c61bb030f 618 */
mzta 0:158c61bb030f 619 static mrb_value
mzta 0:158c61bb030f 620 mrb_str_equal_m(mrb_state *mrb, mrb_value str1)
mzta 0:158c61bb030f 621 {
mzta 0:158c61bb030f 622 mrb_value str2;
mzta 0:158c61bb030f 623
mzta 0:158c61bb030f 624 mrb_get_args(mrb, "o", &str2);
mzta 0:158c61bb030f 625
mzta 0:158c61bb030f 626 return mrb_bool_value(mrb_str_equal(mrb, str1, str2));
mzta 0:158c61bb030f 627 }
mzta 0:158c61bb030f 628 /* ---------------------------------- */
mzta 0:158c61bb030f 629 MRB_API mrb_value
mzta 0:158c61bb030f 630 mrb_str_to_str(mrb_state *mrb, mrb_value str)
mzta 0:158c61bb030f 631 {
mzta 0:158c61bb030f 632 mrb_value s;
mzta 0:158c61bb030f 633
mzta 0:158c61bb030f 634 if (!mrb_string_p(str)) {
mzta 0:158c61bb030f 635 s = mrb_check_convert_type(mrb, str, MRB_TT_STRING, "String", "to_str");
mzta 0:158c61bb030f 636 if (mrb_nil_p(s)) {
mzta 0:158c61bb030f 637 s = mrb_convert_type(mrb, str, MRB_TT_STRING, "String", "to_s");
mzta 0:158c61bb030f 638 }
mzta 0:158c61bb030f 639 return s;
mzta 0:158c61bb030f 640 }
mzta 0:158c61bb030f 641 return str;
mzta 0:158c61bb030f 642 }
mzta 0:158c61bb030f 643
mzta 0:158c61bb030f 644 MRB_API const char*
mzta 0:158c61bb030f 645 mrb_string_value_ptr(mrb_state *mrb, mrb_value ptr)
mzta 0:158c61bb030f 646 {
mzta 0:158c61bb030f 647 mrb_value str = mrb_str_to_str(mrb, ptr);
mzta 0:158c61bb030f 648 return RSTRING_PTR(str);
mzta 0:158c61bb030f 649 }
mzta 0:158c61bb030f 650
mzta 0:158c61bb030f 651 void
mzta 0:158c61bb030f 652 mrb_noregexp(mrb_state *mrb, mrb_value self)
mzta 0:158c61bb030f 653 {
mzta 0:158c61bb030f 654 mrb_raise(mrb, E_NOTIMP_ERROR, "Regexp class not implemented");
mzta 0:158c61bb030f 655 }
mzta 0:158c61bb030f 656
mzta 0:158c61bb030f 657 void
mzta 0:158c61bb030f 658 mrb_regexp_check(mrb_state *mrb, mrb_value obj)
mzta 0:158c61bb030f 659 {
mzta 0:158c61bb030f 660 if (mrb_regexp_p(mrb, obj)) {
mzta 0:158c61bb030f 661 mrb_noregexp(mrb, obj);
mzta 0:158c61bb030f 662 }
mzta 0:158c61bb030f 663 }
mzta 0:158c61bb030f 664
mzta 0:158c61bb030f 665 static inline mrb_int
mzta 0:158c61bb030f 666 mrb_memsearch_qs(const unsigned char *xs, mrb_int m, const unsigned char *ys, mrb_int n)
mzta 0:158c61bb030f 667 {
mzta 0:158c61bb030f 668 const unsigned char *x = xs, *xe = xs + m;
mzta 0:158c61bb030f 669 const unsigned char *y = ys;
mzta 0:158c61bb030f 670 int i, qstable[256];
mzta 0:158c61bb030f 671
mzta 0:158c61bb030f 672 /* Preprocessing */
mzta 0:158c61bb030f 673 for (i = 0; i < 256; ++i)
mzta 0:158c61bb030f 674 qstable[i] = m + 1;
mzta 0:158c61bb030f 675 for (; x < xe; ++x)
mzta 0:158c61bb030f 676 qstable[*x] = xe - x;
mzta 0:158c61bb030f 677 /* Searching */
mzta 0:158c61bb030f 678 for (; y + m <= ys + n; y += *(qstable + y[m])) {
mzta 0:158c61bb030f 679 if (*xs == *y && memcmp(xs, y, m) == 0)
mzta 0:158c61bb030f 680 return y - ys;
mzta 0:158c61bb030f 681 }
mzta 0:158c61bb030f 682 return -1;
mzta 0:158c61bb030f 683 }
mzta 0:158c61bb030f 684
mzta 0:158c61bb030f 685 static mrb_int
mzta 0:158c61bb030f 686 mrb_memsearch(const void *x0, mrb_int m, const void *y0, mrb_int n)
mzta 0:158c61bb030f 687 {
mzta 0:158c61bb030f 688 const unsigned char *x = (const unsigned char *)x0, *y = (const unsigned char *)y0;
mzta 0:158c61bb030f 689
mzta 0:158c61bb030f 690 if (m > n) return -1;
mzta 0:158c61bb030f 691 else if (m == n) {
mzta 0:158c61bb030f 692 return memcmp(x0, y0, m) == 0 ? 0 : -1;
mzta 0:158c61bb030f 693 }
mzta 0:158c61bb030f 694 else if (m < 1) {
mzta 0:158c61bb030f 695 return 0;
mzta 0:158c61bb030f 696 }
mzta 0:158c61bb030f 697 else if (m == 1) {
mzta 0:158c61bb030f 698 const unsigned char *ys = y, *ye = ys + n;
mzta 0:158c61bb030f 699 for (; y < ye; ++y) {
mzta 0:158c61bb030f 700 if (*x == *y)
mzta 0:158c61bb030f 701 return y - ys;
mzta 0:158c61bb030f 702 }
mzta 0:158c61bb030f 703 return -1;
mzta 0:158c61bb030f 704 }
mzta 0:158c61bb030f 705 return mrb_memsearch_qs((const unsigned char *)x0, m, (const unsigned char *)y0, n);
mzta 0:158c61bb030f 706 }
mzta 0:158c61bb030f 707
mzta 0:158c61bb030f 708 static mrb_int
mzta 0:158c61bb030f 709 mrb_str_index(mrb_state *mrb, mrb_value str, mrb_value sub, mrb_int offset)
mzta 0:158c61bb030f 710 {
mzta 0:158c61bb030f 711 mrb_int pos;
mzta 0:158c61bb030f 712 char *s, *sptr;
mzta 0:158c61bb030f 713 mrb_int len, slen;
mzta 0:158c61bb030f 714
mzta 0:158c61bb030f 715 len = RSTRING_LEN(str);
mzta 0:158c61bb030f 716 slen = RSTRING_LEN(sub);
mzta 0:158c61bb030f 717 if (offset < 0) {
mzta 0:158c61bb030f 718 offset += len;
mzta 0:158c61bb030f 719 if (offset < 0) return -1;
mzta 0:158c61bb030f 720 }
mzta 0:158c61bb030f 721 if (len - offset < slen) return -1;
mzta 0:158c61bb030f 722 s = RSTRING_PTR(str);
mzta 0:158c61bb030f 723 if (offset) {
mzta 0:158c61bb030f 724 s += offset;
mzta 0:158c61bb030f 725 }
mzta 0:158c61bb030f 726 if (slen == 0) return offset;
mzta 0:158c61bb030f 727 /* need proceed one character at a time */
mzta 0:158c61bb030f 728 sptr = RSTRING_PTR(sub);
mzta 0:158c61bb030f 729 slen = RSTRING_LEN(sub);
mzta 0:158c61bb030f 730 len = RSTRING_LEN(str) - offset;
mzta 0:158c61bb030f 731 pos = mrb_memsearch(sptr, slen, s, len);
mzta 0:158c61bb030f 732 if (pos < 0) return pos;
mzta 0:158c61bb030f 733 return pos + offset;
mzta 0:158c61bb030f 734 }
mzta 0:158c61bb030f 735
mzta 0:158c61bb030f 736 MRB_API mrb_value
mzta 0:158c61bb030f 737 mrb_str_dup(mrb_state *mrb, mrb_value str)
mzta 0:158c61bb030f 738 {
mzta 0:158c61bb030f 739 struct RString *s = mrb_str_ptr(str);
mzta 0:158c61bb030f 740 struct RString *dup = str_new(mrb, 0, 0);
mzta 0:158c61bb030f 741
mzta 0:158c61bb030f 742 str_with_class(mrb, dup, str);
mzta 0:158c61bb030f 743 return str_replace(mrb, dup, s);
mzta 0:158c61bb030f 744 }
mzta 0:158c61bb030f 745
mzta 0:158c61bb030f 746 static mrb_value
mzta 0:158c61bb030f 747 mrb_str_aref(mrb_state *mrb, mrb_value str, mrb_value indx)
mzta 0:158c61bb030f 748 {
mzta 0:158c61bb030f 749 mrb_int idx;
mzta 0:158c61bb030f 750
mzta 0:158c61bb030f 751 mrb_regexp_check(mrb, indx);
mzta 0:158c61bb030f 752 switch (mrb_type(indx)) {
mzta 0:158c61bb030f 753 case MRB_TT_FIXNUM:
mzta 0:158c61bb030f 754 idx = mrb_fixnum(indx);
mzta 0:158c61bb030f 755
mzta 0:158c61bb030f 756 num_index:
mzta 0:158c61bb030f 757 str = mrb_str_substr(mrb, str, idx, 1);
mzta 0:158c61bb030f 758 if (!mrb_nil_p(str) && RSTRING_LEN(str) == 0) return mrb_nil_value();
mzta 0:158c61bb030f 759 return str;
mzta 0:158c61bb030f 760
mzta 0:158c61bb030f 761 case MRB_TT_STRING:
mzta 0:158c61bb030f 762 if (mrb_str_index(mrb, str, indx, 0) != -1)
mzta 0:158c61bb030f 763 return mrb_str_dup(mrb, indx);
mzta 0:158c61bb030f 764 return mrb_nil_value();
mzta 0:158c61bb030f 765
mzta 0:158c61bb030f 766 case MRB_TT_RANGE:
mzta 0:158c61bb030f 767 /* check if indx is Range */
mzta 0:158c61bb030f 768 {
mzta 0:158c61bb030f 769 mrb_int beg, len;
mzta 0:158c61bb030f 770
mzta 0:158c61bb030f 771 len = RSTRING_LEN(str);
mzta 0:158c61bb030f 772 if (mrb_range_beg_len(mrb, indx, &beg, &len, len)) {
mzta 0:158c61bb030f 773 return mrb_str_subseq(mrb, str, beg, len);
mzta 0:158c61bb030f 774 }
mzta 0:158c61bb030f 775 else {
mzta 0:158c61bb030f 776 return mrb_nil_value();
mzta 0:158c61bb030f 777 }
mzta 0:158c61bb030f 778 }
mzta 0:158c61bb030f 779 default:
mzta 0:158c61bb030f 780 idx = mrb_fixnum(indx);
mzta 0:158c61bb030f 781 goto num_index;
mzta 0:158c61bb030f 782 }
mzta 0:158c61bb030f 783 return mrb_nil_value(); /* not reached */
mzta 0:158c61bb030f 784 }
mzta 0:158c61bb030f 785
mzta 0:158c61bb030f 786 /* 15.2.10.5.6 */
mzta 0:158c61bb030f 787 /* 15.2.10.5.34 */
mzta 0:158c61bb030f 788 /*
mzta 0:158c61bb030f 789 * call-seq:
mzta 0:158c61bb030f 790 * str[fixnum] => fixnum or nil
mzta 0:158c61bb030f 791 * str[fixnum, fixnum] => new_str or nil
mzta 0:158c61bb030f 792 * str[range] => new_str or nil
mzta 0:158c61bb030f 793 * str[regexp] => new_str or nil
mzta 0:158c61bb030f 794 * str[regexp, fixnum] => new_str or nil
mzta 0:158c61bb030f 795 * str[other_str] => new_str or nil
mzta 0:158c61bb030f 796 * str.slice(fixnum) => fixnum or nil
mzta 0:158c61bb030f 797 * str.slice(fixnum, fixnum) => new_str or nil
mzta 0:158c61bb030f 798 * str.slice(range) => new_str or nil
mzta 0:158c61bb030f 799 * str.slice(other_str) => new_str or nil
mzta 0:158c61bb030f 800 *
mzta 0:158c61bb030f 801 * Element Reference---If passed a single <code>Fixnum</code>, returns the code
mzta 0:158c61bb030f 802 * of the character at that position. If passed two <code>Fixnum</code>
mzta 0:158c61bb030f 803 * objects, returns a substring starting at the offset given by the first, and
mzta 0:158c61bb030f 804 * a length given by the second. If given a range, a substring containing
mzta 0:158c61bb030f 805 * characters at offsets given by the range is returned. In all three cases, if
mzta 0:158c61bb030f 806 * an offset is negative, it is counted from the end of <i>str</i>. Returns
mzta 0:158c61bb030f 807 * <code>nil</code> if the initial offset falls outside the string, the length
mzta 0:158c61bb030f 808 * is negative, or the beginning of the range is greater than the end.
mzta 0:158c61bb030f 809 *
mzta 0:158c61bb030f 810 * If a <code>String</code> is given, that string is returned if it occurs in
mzta 0:158c61bb030f 811 * <i>str</i>. In both cases, <code>nil</code> is returned if there is no
mzta 0:158c61bb030f 812 * match.
mzta 0:158c61bb030f 813 *
mzta 0:158c61bb030f 814 * a = "hello there"
mzta 0:158c61bb030f 815 * a[1] #=> 101(1.8.7) "e"(1.9.2)
mzta 0:158c61bb030f 816 * a[1,3] #=> "ell"
mzta 0:158c61bb030f 817 * a[1..3] #=> "ell"
mzta 0:158c61bb030f 818 * a[-3,2] #=> "er"
mzta 0:158c61bb030f 819 * a[-4..-2] #=> "her"
mzta 0:158c61bb030f 820 * a[12..-1] #=> nil
mzta 0:158c61bb030f 821 * a[-2..-4] #=> ""
mzta 0:158c61bb030f 822 * a["lo"] #=> "lo"
mzta 0:158c61bb030f 823 * a["bye"] #=> nil
mzta 0:158c61bb030f 824 */
mzta 0:158c61bb030f 825 static mrb_value
mzta 0:158c61bb030f 826 mrb_str_aref_m(mrb_state *mrb, mrb_value str)
mzta 0:158c61bb030f 827 {
mzta 0:158c61bb030f 828 mrb_value a1, a2;
mzta 0:158c61bb030f 829 int argc;
mzta 0:158c61bb030f 830
mzta 0:158c61bb030f 831 argc = mrb_get_args(mrb, "o|o", &a1, &a2);
mzta 0:158c61bb030f 832 if (argc == 2) {
mzta 0:158c61bb030f 833 mrb_regexp_check(mrb, a1);
mzta 0:158c61bb030f 834 return mrb_str_substr(mrb, str, mrb_fixnum(a1), mrb_fixnum(a2));
mzta 0:158c61bb030f 835 }
mzta 0:158c61bb030f 836 if (argc != 1) {
mzta 0:158c61bb030f 837 mrb_raisef(mrb, E_ARGUMENT_ERROR, "wrong number of arguments (%S for 1)", mrb_fixnum_value(argc));
mzta 0:158c61bb030f 838 }
mzta 0:158c61bb030f 839 return mrb_str_aref(mrb, str, a1);
mzta 0:158c61bb030f 840 }
mzta 0:158c61bb030f 841
mzta 0:158c61bb030f 842 /* 15.2.10.5.8 */
mzta 0:158c61bb030f 843 /*
mzta 0:158c61bb030f 844 * call-seq:
mzta 0:158c61bb030f 845 * str.capitalize! => str or nil
mzta 0:158c61bb030f 846 *
mzta 0:158c61bb030f 847 * Modifies <i>str</i> by converting the first character to uppercase and the
mzta 0:158c61bb030f 848 * remainder to lowercase. Returns <code>nil</code> if no changes are made.
mzta 0:158c61bb030f 849 *
mzta 0:158c61bb030f 850 * a = "hello"
mzta 0:158c61bb030f 851 * a.capitalize! #=> "Hello"
mzta 0:158c61bb030f 852 * a #=> "Hello"
mzta 0:158c61bb030f 853 * a.capitalize! #=> nil
mzta 0:158c61bb030f 854 */
mzta 0:158c61bb030f 855 static mrb_value
mzta 0:158c61bb030f 856 mrb_str_capitalize_bang(mrb_state *mrb, mrb_value str)
mzta 0:158c61bb030f 857 {
mzta 0:158c61bb030f 858 char *p, *pend;
mzta 0:158c61bb030f 859 mrb_bool modify = FALSE;
mzta 0:158c61bb030f 860 struct RString *s = mrb_str_ptr(str);
mzta 0:158c61bb030f 861
mzta 0:158c61bb030f 862 mrb_str_modify(mrb, s);
mzta 0:158c61bb030f 863 if (RSTR_LEN(s) == 0 || !RSTR_PTR(s)) return mrb_nil_value();
mzta 0:158c61bb030f 864 p = RSTR_PTR(s); pend = RSTR_PTR(s) + RSTR_LEN(s);
mzta 0:158c61bb030f 865 if (ISLOWER(*p)) {
mzta 0:158c61bb030f 866 *p = TOUPPER(*p);
mzta 0:158c61bb030f 867 modify = TRUE;
mzta 0:158c61bb030f 868 }
mzta 0:158c61bb030f 869 while (++p < pend) {
mzta 0:158c61bb030f 870 if (ISUPPER(*p)) {
mzta 0:158c61bb030f 871 *p = TOLOWER(*p);
mzta 0:158c61bb030f 872 modify = TRUE;
mzta 0:158c61bb030f 873 }
mzta 0:158c61bb030f 874 }
mzta 0:158c61bb030f 875 if (modify) return str;
mzta 0:158c61bb030f 876 return mrb_nil_value();
mzta 0:158c61bb030f 877 }
mzta 0:158c61bb030f 878
mzta 0:158c61bb030f 879 /* 15.2.10.5.7 */
mzta 0:158c61bb030f 880 /*
mzta 0:158c61bb030f 881 * call-seq:
mzta 0:158c61bb030f 882 * str.capitalize => new_str
mzta 0:158c61bb030f 883 *
mzta 0:158c61bb030f 884 * Returns a copy of <i>str</i> with the first character converted to uppercase
mzta 0:158c61bb030f 885 * and the remainder to lowercase.
mzta 0:158c61bb030f 886 *
mzta 0:158c61bb030f 887 * "hello".capitalize #=> "Hello"
mzta 0:158c61bb030f 888 * "HELLO".capitalize #=> "Hello"
mzta 0:158c61bb030f 889 * "123ABC".capitalize #=> "123abc"
mzta 0:158c61bb030f 890 */
mzta 0:158c61bb030f 891 static mrb_value
mzta 0:158c61bb030f 892 mrb_str_capitalize(mrb_state *mrb, mrb_value self)
mzta 0:158c61bb030f 893 {
mzta 0:158c61bb030f 894 mrb_value str;
mzta 0:158c61bb030f 895
mzta 0:158c61bb030f 896 str = mrb_str_dup(mrb, self);
mzta 0:158c61bb030f 897 mrb_str_capitalize_bang(mrb, str);
mzta 0:158c61bb030f 898 return str;
mzta 0:158c61bb030f 899 }
mzta 0:158c61bb030f 900
mzta 0:158c61bb030f 901 /* 15.2.10.5.10 */
mzta 0:158c61bb030f 902 /*
mzta 0:158c61bb030f 903 * call-seq:
mzta 0:158c61bb030f 904 * str.chomp!(separator=$/) => str or nil
mzta 0:158c61bb030f 905 *
mzta 0:158c61bb030f 906 * Modifies <i>str</i> in place as described for <code>String#chomp</code>,
mzta 0:158c61bb030f 907 * returning <i>str</i>, or <code>nil</code> if no modifications were made.
mzta 0:158c61bb030f 908 */
mzta 0:158c61bb030f 909 static mrb_value
mzta 0:158c61bb030f 910 mrb_str_chomp_bang(mrb_state *mrb, mrb_value str)
mzta 0:158c61bb030f 911 {
mzta 0:158c61bb030f 912 mrb_value rs;
mzta 0:158c61bb030f 913 mrb_int newline;
mzta 0:158c61bb030f 914 char *p, *pp;
mzta 0:158c61bb030f 915 mrb_int rslen;
mzta 0:158c61bb030f 916 mrb_int len;
mzta 0:158c61bb030f 917 struct RString *s = mrb_str_ptr(str);
mzta 0:158c61bb030f 918
mzta 0:158c61bb030f 919 mrb_str_modify(mrb, s);
mzta 0:158c61bb030f 920 len = RSTR_LEN(s);
mzta 0:158c61bb030f 921 if (mrb_get_args(mrb, "|S", &rs) == 0) {
mzta 0:158c61bb030f 922 if (len == 0) return mrb_nil_value();
mzta 0:158c61bb030f 923 smart_chomp:
mzta 0:158c61bb030f 924 if (RSTR_PTR(s)[len-1] == '\n') {
mzta 0:158c61bb030f 925 RSTR_SET_LEN(s, RSTR_LEN(s) - 1);
mzta 0:158c61bb030f 926 if (RSTR_LEN(s) > 0 &&
mzta 0:158c61bb030f 927 RSTR_PTR(s)[RSTR_LEN(s)-1] == '\r') {
mzta 0:158c61bb030f 928 RSTR_SET_LEN(s, RSTR_LEN(s) - 1);
mzta 0:158c61bb030f 929 }
mzta 0:158c61bb030f 930 }
mzta 0:158c61bb030f 931 else if (RSTR_PTR(s)[len-1] == '\r') {
mzta 0:158c61bb030f 932 RSTR_SET_LEN(s, RSTR_LEN(s) - 1);
mzta 0:158c61bb030f 933 }
mzta 0:158c61bb030f 934 else {
mzta 0:158c61bb030f 935 return mrb_nil_value();
mzta 0:158c61bb030f 936 }
mzta 0:158c61bb030f 937 RSTR_PTR(s)[RSTR_LEN(s)] = '\0';
mzta 0:158c61bb030f 938 return str;
mzta 0:158c61bb030f 939 }
mzta 0:158c61bb030f 940
mzta 0:158c61bb030f 941 if (len == 0 || mrb_nil_p(rs)) return mrb_nil_value();
mzta 0:158c61bb030f 942 p = RSTR_PTR(s);
mzta 0:158c61bb030f 943 rslen = RSTRING_LEN(rs);
mzta 0:158c61bb030f 944 if (rslen == 0) {
mzta 0:158c61bb030f 945 while (len>0 && p[len-1] == '\n') {
mzta 0:158c61bb030f 946 len--;
mzta 0:158c61bb030f 947 if (len>0 && p[len-1] == '\r')
mzta 0:158c61bb030f 948 len--;
mzta 0:158c61bb030f 949 }
mzta 0:158c61bb030f 950 if (len < RSTR_LEN(s)) {
mzta 0:158c61bb030f 951 RSTR_SET_LEN(s, len);
mzta 0:158c61bb030f 952 p[len] = '\0';
mzta 0:158c61bb030f 953 return str;
mzta 0:158c61bb030f 954 }
mzta 0:158c61bb030f 955 return mrb_nil_value();
mzta 0:158c61bb030f 956 }
mzta 0:158c61bb030f 957 if (rslen > len) return mrb_nil_value();
mzta 0:158c61bb030f 958 newline = RSTRING_PTR(rs)[rslen-1];
mzta 0:158c61bb030f 959 if (rslen == 1 && newline == '\n')
mzta 0:158c61bb030f 960 newline = RSTRING_PTR(rs)[rslen-1];
mzta 0:158c61bb030f 961 if (rslen == 1 && newline == '\n')
mzta 0:158c61bb030f 962 goto smart_chomp;
mzta 0:158c61bb030f 963
mzta 0:158c61bb030f 964 pp = p + len - rslen;
mzta 0:158c61bb030f 965 if (p[len-1] == newline &&
mzta 0:158c61bb030f 966 (rslen <= 1 ||
mzta 0:158c61bb030f 967 memcmp(RSTRING_PTR(rs), pp, rslen) == 0)) {
mzta 0:158c61bb030f 968 RSTR_SET_LEN(s, len - rslen);
mzta 0:158c61bb030f 969 p[RSTR_LEN(s)] = '\0';
mzta 0:158c61bb030f 970 return str;
mzta 0:158c61bb030f 971 }
mzta 0:158c61bb030f 972 return mrb_nil_value();
mzta 0:158c61bb030f 973 }
mzta 0:158c61bb030f 974
mzta 0:158c61bb030f 975 /* 15.2.10.5.9 */
mzta 0:158c61bb030f 976 /*
mzta 0:158c61bb030f 977 * call-seq:
mzta 0:158c61bb030f 978 * str.chomp(separator=$/) => new_str
mzta 0:158c61bb030f 979 *
mzta 0:158c61bb030f 980 * Returns a new <code>String</code> with the given record separator removed
mzta 0:158c61bb030f 981 * from the end of <i>str</i> (if present). If <code>$/</code> has not been
mzta 0:158c61bb030f 982 * changed from the default Ruby record separator, then <code>chomp</code> also
mzta 0:158c61bb030f 983 * removes carriage return characters (that is it will remove <code>\n</code>,
mzta 0:158c61bb030f 984 * <code>\r</code>, and <code>\r\n</code>).
mzta 0:158c61bb030f 985 *
mzta 0:158c61bb030f 986 * "hello".chomp #=> "hello"
mzta 0:158c61bb030f 987 * "hello\n".chomp #=> "hello"
mzta 0:158c61bb030f 988 * "hello\r\n".chomp #=> "hello"
mzta 0:158c61bb030f 989 * "hello\n\r".chomp #=> "hello\n"
mzta 0:158c61bb030f 990 * "hello\r".chomp #=> "hello"
mzta 0:158c61bb030f 991 * "hello \n there".chomp #=> "hello \n there"
mzta 0:158c61bb030f 992 * "hello".chomp("llo") #=> "he"
mzta 0:158c61bb030f 993 */
mzta 0:158c61bb030f 994 static mrb_value
mzta 0:158c61bb030f 995 mrb_str_chomp(mrb_state *mrb, mrb_value self)
mzta 0:158c61bb030f 996 {
mzta 0:158c61bb030f 997 mrb_value str;
mzta 0:158c61bb030f 998
mzta 0:158c61bb030f 999 str = mrb_str_dup(mrb, self);
mzta 0:158c61bb030f 1000 mrb_str_chomp_bang(mrb, str);
mzta 0:158c61bb030f 1001 return str;
mzta 0:158c61bb030f 1002 }
mzta 0:158c61bb030f 1003
mzta 0:158c61bb030f 1004 /* 15.2.10.5.12 */
mzta 0:158c61bb030f 1005 /*
mzta 0:158c61bb030f 1006 * call-seq:
mzta 0:158c61bb030f 1007 * str.chop! => str or nil
mzta 0:158c61bb030f 1008 *
mzta 0:158c61bb030f 1009 * Processes <i>str</i> as for <code>String#chop</code>, returning <i>str</i>,
mzta 0:158c61bb030f 1010 * or <code>nil</code> if <i>str</i> is the empty string. See also
mzta 0:158c61bb030f 1011 * <code>String#chomp!</code>.
mzta 0:158c61bb030f 1012 */
mzta 0:158c61bb030f 1013 static mrb_value
mzta 0:158c61bb030f 1014 mrb_str_chop_bang(mrb_state *mrb, mrb_value str)
mzta 0:158c61bb030f 1015 {
mzta 0:158c61bb030f 1016 struct RString *s = mrb_str_ptr(str);
mzta 0:158c61bb030f 1017
mzta 0:158c61bb030f 1018 mrb_str_modify(mrb, s);
mzta 0:158c61bb030f 1019 if (RSTR_LEN(s) > 0) {
mzta 0:158c61bb030f 1020 mrb_int len;
mzta 0:158c61bb030f 1021 len = RSTR_LEN(s) - 1;
mzta 0:158c61bb030f 1022 if (RSTR_PTR(s)[len] == '\n') {
mzta 0:158c61bb030f 1023 if (len > 0 &&
mzta 0:158c61bb030f 1024 RSTR_PTR(s)[len-1] == '\r') {
mzta 0:158c61bb030f 1025 len--;
mzta 0:158c61bb030f 1026 }
mzta 0:158c61bb030f 1027 }
mzta 0:158c61bb030f 1028 RSTR_SET_LEN(s, len);
mzta 0:158c61bb030f 1029 RSTR_PTR(s)[len] = '\0';
mzta 0:158c61bb030f 1030 return str;
mzta 0:158c61bb030f 1031 }
mzta 0:158c61bb030f 1032 return mrb_nil_value();
mzta 0:158c61bb030f 1033 }
mzta 0:158c61bb030f 1034
mzta 0:158c61bb030f 1035 /* 15.2.10.5.11 */
mzta 0:158c61bb030f 1036 /*
mzta 0:158c61bb030f 1037 * call-seq:
mzta 0:158c61bb030f 1038 * str.chop => new_str
mzta 0:158c61bb030f 1039 *
mzta 0:158c61bb030f 1040 * Returns a new <code>String</code> with the last character removed. If the
mzta 0:158c61bb030f 1041 * string ends with <code>\r\n</code>, both characters are removed. Applying
mzta 0:158c61bb030f 1042 * <code>chop</code> to an empty string returns an empty
mzta 0:158c61bb030f 1043 * string. <code>String#chomp</code> is often a safer alternative, as it leaves
mzta 0:158c61bb030f 1044 * the string unchanged if it doesn't end in a record separator.
mzta 0:158c61bb030f 1045 *
mzta 0:158c61bb030f 1046 * "string\r\n".chop #=> "string"
mzta 0:158c61bb030f 1047 * "string\n\r".chop #=> "string\n"
mzta 0:158c61bb030f 1048 * "string\n".chop #=> "string"
mzta 0:158c61bb030f 1049 * "string".chop #=> "strin"
mzta 0:158c61bb030f 1050 * "x".chop #=> ""
mzta 0:158c61bb030f 1051 */
mzta 0:158c61bb030f 1052 static mrb_value
mzta 0:158c61bb030f 1053 mrb_str_chop(mrb_state *mrb, mrb_value self)
mzta 0:158c61bb030f 1054 {
mzta 0:158c61bb030f 1055 mrb_value str;
mzta 0:158c61bb030f 1056 str = mrb_str_dup(mrb, self);
mzta 0:158c61bb030f 1057 mrb_str_chop_bang(mrb, str);
mzta 0:158c61bb030f 1058 return str;
mzta 0:158c61bb030f 1059 }
mzta 0:158c61bb030f 1060
mzta 0:158c61bb030f 1061 /* 15.2.10.5.14 */
mzta 0:158c61bb030f 1062 /*
mzta 0:158c61bb030f 1063 * call-seq:
mzta 0:158c61bb030f 1064 * str.downcase! => str or nil
mzta 0:158c61bb030f 1065 *
mzta 0:158c61bb030f 1066 * Downcases the contents of <i>str</i>, returning <code>nil</code> if no
mzta 0:158c61bb030f 1067 * changes were made.
mzta 0:158c61bb030f 1068 */
mzta 0:158c61bb030f 1069 static mrb_value
mzta 0:158c61bb030f 1070 mrb_str_downcase_bang(mrb_state *mrb, mrb_value str)
mzta 0:158c61bb030f 1071 {
mzta 0:158c61bb030f 1072 char *p, *pend;
mzta 0:158c61bb030f 1073 mrb_bool modify = FALSE;
mzta 0:158c61bb030f 1074 struct RString *s = mrb_str_ptr(str);
mzta 0:158c61bb030f 1075
mzta 0:158c61bb030f 1076 mrb_str_modify(mrb, s);
mzta 0:158c61bb030f 1077 p = RSTR_PTR(s);
mzta 0:158c61bb030f 1078 pend = RSTR_PTR(s) + RSTR_LEN(s);
mzta 0:158c61bb030f 1079 while (p < pend) {
mzta 0:158c61bb030f 1080 if (ISUPPER(*p)) {
mzta 0:158c61bb030f 1081 *p = TOLOWER(*p);
mzta 0:158c61bb030f 1082 modify = TRUE;
mzta 0:158c61bb030f 1083 }
mzta 0:158c61bb030f 1084 p++;
mzta 0:158c61bb030f 1085 }
mzta 0:158c61bb030f 1086
mzta 0:158c61bb030f 1087 if (modify) return str;
mzta 0:158c61bb030f 1088 return mrb_nil_value();
mzta 0:158c61bb030f 1089 }
mzta 0:158c61bb030f 1090
mzta 0:158c61bb030f 1091 /* 15.2.10.5.13 */
mzta 0:158c61bb030f 1092 /*
mzta 0:158c61bb030f 1093 * call-seq:
mzta 0:158c61bb030f 1094 * str.downcase => new_str
mzta 0:158c61bb030f 1095 *
mzta 0:158c61bb030f 1096 * Returns a copy of <i>str</i> with all uppercase letters replaced with their
mzta 0:158c61bb030f 1097 * lowercase counterparts. The operation is locale insensitive---only
mzta 0:158c61bb030f 1098 * characters ``A'' to ``Z'' are affected.
mzta 0:158c61bb030f 1099 *
mzta 0:158c61bb030f 1100 * "hEllO".downcase #=> "hello"
mzta 0:158c61bb030f 1101 */
mzta 0:158c61bb030f 1102 static mrb_value
mzta 0:158c61bb030f 1103 mrb_str_downcase(mrb_state *mrb, mrb_value self)
mzta 0:158c61bb030f 1104 {
mzta 0:158c61bb030f 1105 mrb_value str;
mzta 0:158c61bb030f 1106
mzta 0:158c61bb030f 1107 str = mrb_str_dup(mrb, self);
mzta 0:158c61bb030f 1108 mrb_str_downcase_bang(mrb, str);
mzta 0:158c61bb030f 1109 return str;
mzta 0:158c61bb030f 1110 }
mzta 0:158c61bb030f 1111
mzta 0:158c61bb030f 1112 /* 15.2.10.5.16 */
mzta 0:158c61bb030f 1113 /*
mzta 0:158c61bb030f 1114 * call-seq:
mzta 0:158c61bb030f 1115 * str.empty? => true or false
mzta 0:158c61bb030f 1116 *
mzta 0:158c61bb030f 1117 * Returns <code>true</code> if <i>str</i> has a length of zero.
mzta 0:158c61bb030f 1118 *
mzta 0:158c61bb030f 1119 * "hello".empty? #=> false
mzta 0:158c61bb030f 1120 * "".empty? #=> true
mzta 0:158c61bb030f 1121 */
mzta 0:158c61bb030f 1122 static mrb_value
mzta 0:158c61bb030f 1123 mrb_str_empty_p(mrb_state *mrb, mrb_value self)
mzta 0:158c61bb030f 1124 {
mzta 0:158c61bb030f 1125 struct RString *s = mrb_str_ptr(self);
mzta 0:158c61bb030f 1126
mzta 0:158c61bb030f 1127 return mrb_bool_value(RSTR_LEN(s) == 0);
mzta 0:158c61bb030f 1128 }
mzta 0:158c61bb030f 1129
mzta 0:158c61bb030f 1130 /* 15.2.10.5.17 */
mzta 0:158c61bb030f 1131 /*
mzta 0:158c61bb030f 1132 * call-seq:
mzta 0:158c61bb030f 1133 * str.eql?(other) => true or false
mzta 0:158c61bb030f 1134 *
mzta 0:158c61bb030f 1135 * Two strings are equal if the have the same length and content.
mzta 0:158c61bb030f 1136 */
mzta 0:158c61bb030f 1137 static mrb_value
mzta 0:158c61bb030f 1138 mrb_str_eql(mrb_state *mrb, mrb_value self)
mzta 0:158c61bb030f 1139 {
mzta 0:158c61bb030f 1140 mrb_value str2;
mzta 0:158c61bb030f 1141 mrb_bool eql_p;
mzta 0:158c61bb030f 1142
mzta 0:158c61bb030f 1143 mrb_get_args(mrb, "o", &str2);
mzta 0:158c61bb030f 1144 eql_p = (mrb_type(str2) == MRB_TT_STRING) && str_eql(mrb, self, str2);
mzta 0:158c61bb030f 1145
mzta 0:158c61bb030f 1146 return mrb_bool_value(eql_p);
mzta 0:158c61bb030f 1147 }
mzta 0:158c61bb030f 1148
mzta 0:158c61bb030f 1149 static mrb_value
mzta 0:158c61bb030f 1150 mrb_str_subseq(mrb_state *mrb, mrb_value str, mrb_int beg, mrb_int len)
mzta 0:158c61bb030f 1151 {
mzta 0:158c61bb030f 1152 struct RString *orig, *s;
mzta 0:158c61bb030f 1153 mrb_shared_string *shared;
mzta 0:158c61bb030f 1154
mzta 0:158c61bb030f 1155 orig = mrb_str_ptr(str);
mzta 0:158c61bb030f 1156 if (RSTR_EMBED_P(orig)) {
mzta 0:158c61bb030f 1157 s = str_new(mrb, orig->as.ary+beg, len);
mzta 0:158c61bb030f 1158 } else {
mzta 0:158c61bb030f 1159 str_make_shared(mrb, orig);
mzta 0:158c61bb030f 1160 shared = orig->as.heap.aux.shared;
mzta 0:158c61bb030f 1161 s = mrb_obj_alloc_string(mrb);
mzta 0:158c61bb030f 1162 s->as.heap.ptr = orig->as.heap.ptr + beg;
mzta 0:158c61bb030f 1163 s->as.heap.len = len;
mzta 0:158c61bb030f 1164 s->as.heap.aux.shared = shared;
mzta 0:158c61bb030f 1165 RSTR_SET_SHARED_FLAG(s);
mzta 0:158c61bb030f 1166 shared->refcnt++;
mzta 0:158c61bb030f 1167 }
mzta 0:158c61bb030f 1168
mzta 0:158c61bb030f 1169 return mrb_obj_value(s);
mzta 0:158c61bb030f 1170 }
mzta 0:158c61bb030f 1171
mzta 0:158c61bb030f 1172 MRB_API mrb_value
mzta 0:158c61bb030f 1173 mrb_str_substr(mrb_state *mrb, mrb_value str, mrb_int beg, mrb_int len)
mzta 0:158c61bb030f 1174 {
mzta 0:158c61bb030f 1175 if (len < 0) return mrb_nil_value();
mzta 0:158c61bb030f 1176 if (!RSTRING_LEN(str)) {
mzta 0:158c61bb030f 1177 len = 0;
mzta 0:158c61bb030f 1178 }
mzta 0:158c61bb030f 1179 if (beg > RSTRING_LEN(str)) return mrb_nil_value();
mzta 0:158c61bb030f 1180 if (beg < 0) {
mzta 0:158c61bb030f 1181 beg += RSTRING_LEN(str);
mzta 0:158c61bb030f 1182 if (beg < 0) return mrb_nil_value();
mzta 0:158c61bb030f 1183 }
mzta 0:158c61bb030f 1184 if (beg + len > RSTRING_LEN(str))
mzta 0:158c61bb030f 1185 len = RSTRING_LEN(str) - beg;
mzta 0:158c61bb030f 1186 if (len <= 0) {
mzta 0:158c61bb030f 1187 len = 0;
mzta 0:158c61bb030f 1188 }
mzta 0:158c61bb030f 1189 return mrb_str_subseq(mrb, str, beg, len);
mzta 0:158c61bb030f 1190 }
mzta 0:158c61bb030f 1191
mzta 0:158c61bb030f 1192 mrb_int
mzta 0:158c61bb030f 1193 mrb_str_hash(mrb_state *mrb, mrb_value str)
mzta 0:158c61bb030f 1194 {
mzta 0:158c61bb030f 1195 /* 1-8-7 */
mzta 0:158c61bb030f 1196 struct RString *s = mrb_str_ptr(str);
mzta 0:158c61bb030f 1197 mrb_int len = RSTR_LEN(s);
mzta 0:158c61bb030f 1198 char *p = RSTR_PTR(s);
mzta 0:158c61bb030f 1199 mrb_int key = 0;
mzta 0:158c61bb030f 1200
mzta 0:158c61bb030f 1201 while (len--) {
mzta 0:158c61bb030f 1202 key = key*65599 + *p;
mzta 0:158c61bb030f 1203 p++;
mzta 0:158c61bb030f 1204 }
mzta 0:158c61bb030f 1205 return key + (key>>5);
mzta 0:158c61bb030f 1206 }
mzta 0:158c61bb030f 1207
mzta 0:158c61bb030f 1208 /* 15.2.10.5.20 */
mzta 0:158c61bb030f 1209 /*
mzta 0:158c61bb030f 1210 * call-seq:
mzta 0:158c61bb030f 1211 * str.hash => fixnum
mzta 0:158c61bb030f 1212 *
mzta 0:158c61bb030f 1213 * Return a hash based on the string's length and content.
mzta 0:158c61bb030f 1214 */
mzta 0:158c61bb030f 1215 static mrb_value
mzta 0:158c61bb030f 1216 mrb_str_hash_m(mrb_state *mrb, mrb_value self)
mzta 0:158c61bb030f 1217 {
mzta 0:158c61bb030f 1218 mrb_int key = mrb_str_hash(mrb, self);
mzta 0:158c61bb030f 1219 return mrb_fixnum_value(key);
mzta 0:158c61bb030f 1220 }
mzta 0:158c61bb030f 1221
mzta 0:158c61bb030f 1222 /* 15.2.10.5.21 */
mzta 0:158c61bb030f 1223 /*
mzta 0:158c61bb030f 1224 * call-seq:
mzta 0:158c61bb030f 1225 * str.include? other_str => true or false
mzta 0:158c61bb030f 1226 * str.include? fixnum => true or false
mzta 0:158c61bb030f 1227 *
mzta 0:158c61bb030f 1228 * Returns <code>true</code> if <i>str</i> contains the given string or
mzta 0:158c61bb030f 1229 * character.
mzta 0:158c61bb030f 1230 *
mzta 0:158c61bb030f 1231 * "hello".include? "lo" #=> true
mzta 0:158c61bb030f 1232 * "hello".include? "ol" #=> false
mzta 0:158c61bb030f 1233 * "hello".include? ?h #=> true
mzta 0:158c61bb030f 1234 */
mzta 0:158c61bb030f 1235 static mrb_value
mzta 0:158c61bb030f 1236 mrb_str_include(mrb_state *mrb, mrb_value self)
mzta 0:158c61bb030f 1237 {
mzta 0:158c61bb030f 1238 mrb_int i;
mzta 0:158c61bb030f 1239 mrb_value str2;
mzta 0:158c61bb030f 1240 mrb_bool include_p;
mzta 0:158c61bb030f 1241
mzta 0:158c61bb030f 1242 mrb_get_args(mrb, "o", &str2);
mzta 0:158c61bb030f 1243 if (mrb_fixnum_p(str2)) {
mzta 0:158c61bb030f 1244 include_p = (memchr(RSTRING_PTR(self), mrb_fixnum(str2), RSTRING_LEN(self)) != NULL);
mzta 0:158c61bb030f 1245 }
mzta 0:158c61bb030f 1246 else {
mzta 0:158c61bb030f 1247 str2 = mrb_str_to_str(mrb, str2);
mzta 0:158c61bb030f 1248 i = mrb_str_index(mrb, self, str2, 0);
mzta 0:158c61bb030f 1249
mzta 0:158c61bb030f 1250 include_p = (i != -1);
mzta 0:158c61bb030f 1251 }
mzta 0:158c61bb030f 1252
mzta 0:158c61bb030f 1253 return mrb_bool_value(include_p);
mzta 0:158c61bb030f 1254 }
mzta 0:158c61bb030f 1255
mzta 0:158c61bb030f 1256 /* 15.2.10.5.22 */
mzta 0:158c61bb030f 1257 /*
mzta 0:158c61bb030f 1258 * call-seq:
mzta 0:158c61bb030f 1259 * str.index(substring [, offset]) => fixnum or nil
mzta 0:158c61bb030f 1260 * str.index(fixnum [, offset]) => fixnum or nil
mzta 0:158c61bb030f 1261 * str.index(regexp [, offset]) => fixnum or nil
mzta 0:158c61bb030f 1262 *
mzta 0:158c61bb030f 1263 * Returns the index of the first occurrence of the given
mzta 0:158c61bb030f 1264 * <i>substring</i>,
mzta 0:158c61bb030f 1265 * character (<i>fixnum</i>), or pattern (<i>regexp</i>) in <i>str</i>.
mzta 0:158c61bb030f 1266 * Returns
mzta 0:158c61bb030f 1267 * <code>nil</code> if not found.
mzta 0:158c61bb030f 1268 * If the second parameter is present, it
mzta 0:158c61bb030f 1269 * specifies the position in the string to begin the search.
mzta 0:158c61bb030f 1270 *
mzta 0:158c61bb030f 1271 * "hello".index('e') #=> 1
mzta 0:158c61bb030f 1272 * "hello".index('lo') #=> 3
mzta 0:158c61bb030f 1273 * "hello".index('a') #=> nil
mzta 0:158c61bb030f 1274 * "hello".index(101) #=> 1(101=0x65='e')
mzta 0:158c61bb030f 1275 * "hello".index(/[aeiou]/, -3) #=> 4
mzta 0:158c61bb030f 1276 */
mzta 0:158c61bb030f 1277 static mrb_value
mzta 0:158c61bb030f 1278 mrb_str_index_m(mrb_state *mrb, mrb_value str)
mzta 0:158c61bb030f 1279 {
mzta 0:158c61bb030f 1280 mrb_value *argv;
mzta 0:158c61bb030f 1281 mrb_int argc;
mzta 0:158c61bb030f 1282 mrb_value sub;
mzta 0:158c61bb030f 1283 mrb_int pos;
mzta 0:158c61bb030f 1284
mzta 0:158c61bb030f 1285 mrb_get_args(mrb, "*", &argv, &argc);
mzta 0:158c61bb030f 1286 if (argc == 2) {
mzta 0:158c61bb030f 1287 pos = mrb_fixnum(argv[1]);
mzta 0:158c61bb030f 1288 sub = argv[0];
mzta 0:158c61bb030f 1289 }
mzta 0:158c61bb030f 1290 else {
mzta 0:158c61bb030f 1291 pos = 0;
mzta 0:158c61bb030f 1292 if (argc > 0)
mzta 0:158c61bb030f 1293 sub = argv[0];
mzta 0:158c61bb030f 1294 else
mzta 0:158c61bb030f 1295 sub = mrb_nil_value();
mzta 0:158c61bb030f 1296 }
mzta 0:158c61bb030f 1297 mrb_regexp_check(mrb, sub);
mzta 0:158c61bb030f 1298 if (pos < 0) {
mzta 0:158c61bb030f 1299 pos += RSTRING_LEN(str);
mzta 0:158c61bb030f 1300 if (pos < 0) {
mzta 0:158c61bb030f 1301 return mrb_nil_value();
mzta 0:158c61bb030f 1302 }
mzta 0:158c61bb030f 1303 }
mzta 0:158c61bb030f 1304
mzta 0:158c61bb030f 1305 switch (mrb_type(sub)) {
mzta 0:158c61bb030f 1306 case MRB_TT_FIXNUM: {
mzta 0:158c61bb030f 1307 int c = mrb_fixnum(sub);
mzta 0:158c61bb030f 1308 mrb_int len = RSTRING_LEN(str);
mzta 0:158c61bb030f 1309 unsigned char *p = (unsigned char*)RSTRING_PTR(str);
mzta 0:158c61bb030f 1310
mzta 0:158c61bb030f 1311 for (;pos<len;pos++) {
mzta 0:158c61bb030f 1312 if (p[pos] == c) return mrb_fixnum_value(pos);
mzta 0:158c61bb030f 1313 }
mzta 0:158c61bb030f 1314 return mrb_nil_value();
mzta 0:158c61bb030f 1315 }
mzta 0:158c61bb030f 1316
mzta 0:158c61bb030f 1317 default: {
mzta 0:158c61bb030f 1318 mrb_value tmp;
mzta 0:158c61bb030f 1319
mzta 0:158c61bb030f 1320 tmp = mrb_check_string_type(mrb, sub);
mzta 0:158c61bb030f 1321 if (mrb_nil_p(tmp)) {
mzta 0:158c61bb030f 1322 mrb_raisef(mrb, E_TYPE_ERROR, "type mismatch: %S given", sub);
mzta 0:158c61bb030f 1323 }
mzta 0:158c61bb030f 1324 sub = tmp;
mzta 0:158c61bb030f 1325 }
mzta 0:158c61bb030f 1326 /* fall through */
mzta 0:158c61bb030f 1327 case MRB_TT_STRING:
mzta 0:158c61bb030f 1328 pos = mrb_str_index(mrb, str, sub, pos);
mzta 0:158c61bb030f 1329 break;
mzta 0:158c61bb030f 1330 }
mzta 0:158c61bb030f 1331
mzta 0:158c61bb030f 1332 if (pos == -1) return mrb_nil_value();
mzta 0:158c61bb030f 1333 return mrb_fixnum_value(pos);
mzta 0:158c61bb030f 1334 }
mzta 0:158c61bb030f 1335
mzta 0:158c61bb030f 1336 #define STR_REPLACE_SHARED_MIN 10
mzta 0:158c61bb030f 1337
mzta 0:158c61bb030f 1338 static mrb_value
mzta 0:158c61bb030f 1339 str_replace(mrb_state *mrb, struct RString *s1, struct RString *s2)
mzta 0:158c61bb030f 1340 {
mzta 0:158c61bb030f 1341 long len;
mzta 0:158c61bb030f 1342
mzta 0:158c61bb030f 1343 len = RSTR_LEN(s2);
mzta 0:158c61bb030f 1344 if (RSTR_SHARED_P(s1)) {
mzta 0:158c61bb030f 1345 str_decref(mrb, s1->as.heap.aux.shared);
mzta 0:158c61bb030f 1346 }
mzta 0:158c61bb030f 1347 else if (!RSTR_EMBED_P(s1) && !RSTR_NOFREE_P(s1)) {
mzta 0:158c61bb030f 1348 mrb_free(mrb, s1->as.heap.ptr);
mzta 0:158c61bb030f 1349 }
mzta 0:158c61bb030f 1350
mzta 0:158c61bb030f 1351 RSTR_UNSET_NOFREE_FLAG(s1);
mzta 0:158c61bb030f 1352
mzta 0:158c61bb030f 1353 if (RSTR_SHARED_P(s2)) {
mzta 0:158c61bb030f 1354 L_SHARE:
mzta 0:158c61bb030f 1355 RSTR_UNSET_EMBED_FLAG(s1);
mzta 0:158c61bb030f 1356 s1->as.heap.ptr = s2->as.heap.ptr;
mzta 0:158c61bb030f 1357 s1->as.heap.len = len;
mzta 0:158c61bb030f 1358 s1->as.heap.aux.shared = s2->as.heap.aux.shared;
mzta 0:158c61bb030f 1359 RSTR_SET_SHARED_FLAG(s1);
mzta 0:158c61bb030f 1360 s1->as.heap.aux.shared->refcnt++;
mzta 0:158c61bb030f 1361 }
mzta 0:158c61bb030f 1362 else {
mzta 0:158c61bb030f 1363 if (len <= RSTRING_EMBED_LEN_MAX) {
mzta 0:158c61bb030f 1364 RSTR_UNSET_SHARED_FLAG(s1);
mzta 0:158c61bb030f 1365 RSTR_SET_EMBED_FLAG(s1);
mzta 0:158c61bb030f 1366 memcpy(s1->as.ary, RSTR_PTR(s2), len);
mzta 0:158c61bb030f 1367 RSTR_SET_EMBED_LEN(s1, len);
mzta 0:158c61bb030f 1368 }
mzta 0:158c61bb030f 1369 else {
mzta 0:158c61bb030f 1370 str_make_shared(mrb, s2);
mzta 0:158c61bb030f 1371 goto L_SHARE;
mzta 0:158c61bb030f 1372 }
mzta 0:158c61bb030f 1373 }
mzta 0:158c61bb030f 1374
mzta 0:158c61bb030f 1375 return mrb_obj_value(s1);
mzta 0:158c61bb030f 1376 }
mzta 0:158c61bb030f 1377
mzta 0:158c61bb030f 1378 /* 15.2.10.5.24 */
mzta 0:158c61bb030f 1379 /* 15.2.10.5.28 */
mzta 0:158c61bb030f 1380 /*
mzta 0:158c61bb030f 1381 * call-seq:
mzta 0:158c61bb030f 1382 * str.replace(other_str) => str
mzta 0:158c61bb030f 1383 *
mzta 0:158c61bb030f 1384 * s = "hello" #=> "hello"
mzta 0:158c61bb030f 1385 * s.replace "world" #=> "world"
mzta 0:158c61bb030f 1386 */
mzta 0:158c61bb030f 1387 static mrb_value
mzta 0:158c61bb030f 1388 mrb_str_replace(mrb_state *mrb, mrb_value str)
mzta 0:158c61bb030f 1389 {
mzta 0:158c61bb030f 1390 mrb_value str2;
mzta 0:158c61bb030f 1391
mzta 0:158c61bb030f 1392 mrb_get_args(mrb, "S", &str2);
mzta 0:158c61bb030f 1393 return str_replace(mrb, mrb_str_ptr(str), mrb_str_ptr(str2));
mzta 0:158c61bb030f 1394 }
mzta 0:158c61bb030f 1395
mzta 0:158c61bb030f 1396 /* 15.2.10.5.23 */
mzta 0:158c61bb030f 1397 /*
mzta 0:158c61bb030f 1398 * call-seq:
mzta 0:158c61bb030f 1399 * String.new(str="") => new_str
mzta 0:158c61bb030f 1400 *
mzta 0:158c61bb030f 1401 * Returns a new string object containing a copy of <i>str</i>.
mzta 0:158c61bb030f 1402 */
mzta 0:158c61bb030f 1403 static mrb_value
mzta 0:158c61bb030f 1404 mrb_str_init(mrb_state *mrb, mrb_value self)
mzta 0:158c61bb030f 1405 {
mzta 0:158c61bb030f 1406 mrb_value str2;
mzta 0:158c61bb030f 1407
mzta 0:158c61bb030f 1408 if (mrb_get_args(mrb, "|S", &str2) == 1) {
mzta 0:158c61bb030f 1409 str_replace(mrb, mrb_str_ptr(self), mrb_str_ptr(str2));
mzta 0:158c61bb030f 1410 }
mzta 0:158c61bb030f 1411 return self;
mzta 0:158c61bb030f 1412 }
mzta 0:158c61bb030f 1413
mzta 0:158c61bb030f 1414 /* 15.2.10.5.25 */
mzta 0:158c61bb030f 1415 /* 15.2.10.5.41 */
mzta 0:158c61bb030f 1416 /*
mzta 0:158c61bb030f 1417 * call-seq:
mzta 0:158c61bb030f 1418 * str.intern => symbol
mzta 0:158c61bb030f 1419 * str.to_sym => symbol
mzta 0:158c61bb030f 1420 *
mzta 0:158c61bb030f 1421 * Returns the <code>Symbol</code> corresponding to <i>str</i>, creating the
mzta 0:158c61bb030f 1422 * symbol if it did not previously exist. See <code>Symbol#id2name</code>.
mzta 0:158c61bb030f 1423 *
mzta 0:158c61bb030f 1424 * "Koala".intern #=> :Koala
mzta 0:158c61bb030f 1425 * s = 'cat'.to_sym #=> :cat
mzta 0:158c61bb030f 1426 * s == :cat #=> true
mzta 0:158c61bb030f 1427 * s = '@cat'.to_sym #=> :@cat
mzta 0:158c61bb030f 1428 * s == :@cat #=> true
mzta 0:158c61bb030f 1429 *
mzta 0:158c61bb030f 1430 * This can also be used to create symbols that cannot be represented using the
mzta 0:158c61bb030f 1431 * <code>:xxx</code> notation.
mzta 0:158c61bb030f 1432 *
mzta 0:158c61bb030f 1433 * 'cat and dog'.to_sym #=> :"cat and dog"
mzta 0:158c61bb030f 1434 */
mzta 0:158c61bb030f 1435 MRB_API mrb_value
mzta 0:158c61bb030f 1436 mrb_str_intern(mrb_state *mrb, mrb_value self)
mzta 0:158c61bb030f 1437 {
mzta 0:158c61bb030f 1438 return mrb_symbol_value(mrb_intern_str(mrb, self));
mzta 0:158c61bb030f 1439 }
mzta 0:158c61bb030f 1440 /* ---------------------------------- */
mzta 0:158c61bb030f 1441 MRB_API mrb_value
mzta 0:158c61bb030f 1442 mrb_obj_as_string(mrb_state *mrb, mrb_value obj)
mzta 0:158c61bb030f 1443 {
mzta 0:158c61bb030f 1444 mrb_value str;
mzta 0:158c61bb030f 1445
mzta 0:158c61bb030f 1446 if (mrb_string_p(obj)) {
mzta 0:158c61bb030f 1447 return obj;
mzta 0:158c61bb030f 1448 }
mzta 0:158c61bb030f 1449 str = mrb_funcall(mrb, obj, "to_s", 0);
mzta 0:158c61bb030f 1450 if (!mrb_string_p(str))
mzta 0:158c61bb030f 1451 return mrb_any_to_s(mrb, obj);
mzta 0:158c61bb030f 1452 return str;
mzta 0:158c61bb030f 1453 }
mzta 0:158c61bb030f 1454
mzta 0:158c61bb030f 1455 MRB_API mrb_value
mzta 0:158c61bb030f 1456 mrb_ptr_to_str(mrb_state *mrb, void *p)
mzta 0:158c61bb030f 1457 {
mzta 0:158c61bb030f 1458 struct RString *p_str;
mzta 0:158c61bb030f 1459 char *p1;
mzta 0:158c61bb030f 1460 char *p2;
mzta 0:158c61bb030f 1461 uintptr_t n = (uintptr_t)p;
mzta 0:158c61bb030f 1462
mzta 0:158c61bb030f 1463 p_str = str_new(mrb, NULL, 2 + sizeof(uintptr_t) * CHAR_BIT / 4);
mzta 0:158c61bb030f 1464 p1 = RSTR_PTR(p_str);
mzta 0:158c61bb030f 1465 *p1++ = '0';
mzta 0:158c61bb030f 1466 *p1++ = 'x';
mzta 0:158c61bb030f 1467 p2 = p1;
mzta 0:158c61bb030f 1468
mzta 0:158c61bb030f 1469 do {
mzta 0:158c61bb030f 1470 *p2++ = mrb_digitmap[n % 16];
mzta 0:158c61bb030f 1471 n /= 16;
mzta 0:158c61bb030f 1472 } while (n > 0);
mzta 0:158c61bb030f 1473 *p2 = '\0';
mzta 0:158c61bb030f 1474 RSTR_SET_LEN(p_str, (mrb_int)(p2 - RSTR_PTR(p_str)));
mzta 0:158c61bb030f 1475
mzta 0:158c61bb030f 1476 while (p1 < p2) {
mzta 0:158c61bb030f 1477 const char c = *p1;
mzta 0:158c61bb030f 1478 *p1++ = *--p2;
mzta 0:158c61bb030f 1479 *p2 = c;
mzta 0:158c61bb030f 1480 }
mzta 0:158c61bb030f 1481
mzta 0:158c61bb030f 1482 return mrb_obj_value(p_str);
mzta 0:158c61bb030f 1483 }
mzta 0:158c61bb030f 1484
mzta 0:158c61bb030f 1485 MRB_API mrb_value
mzta 0:158c61bb030f 1486 mrb_string_type(mrb_state *mrb, mrb_value str)
mzta 0:158c61bb030f 1487 {
mzta 0:158c61bb030f 1488 return mrb_convert_type(mrb, str, MRB_TT_STRING, "String", "to_str");
mzta 0:158c61bb030f 1489 }
mzta 0:158c61bb030f 1490
mzta 0:158c61bb030f 1491 MRB_API mrb_value
mzta 0:158c61bb030f 1492 mrb_check_string_type(mrb_state *mrb, mrb_value str)
mzta 0:158c61bb030f 1493 {
mzta 0:158c61bb030f 1494 return mrb_check_convert_type(mrb, str, MRB_TT_STRING, "String", "to_str");
mzta 0:158c61bb030f 1495 }
mzta 0:158c61bb030f 1496
mzta 0:158c61bb030f 1497 /* ---------------------------------- */
mzta 0:158c61bb030f 1498 /* 15.2.10.5.29 */
mzta 0:158c61bb030f 1499 /*
mzta 0:158c61bb030f 1500 * call-seq:
mzta 0:158c61bb030f 1501 * str.reverse => new_str
mzta 0:158c61bb030f 1502 *
mzta 0:158c61bb030f 1503 * Returns a new string with the characters from <i>str</i> in reverse order.
mzta 0:158c61bb030f 1504 *
mzta 0:158c61bb030f 1505 * "stressed".reverse #=> "desserts"
mzta 0:158c61bb030f 1506 */
mzta 0:158c61bb030f 1507 static mrb_value
mzta 0:158c61bb030f 1508 mrb_str_reverse(mrb_state *mrb, mrb_value str)
mzta 0:158c61bb030f 1509 {
mzta 0:158c61bb030f 1510 struct RString *s2;
mzta 0:158c61bb030f 1511 char *s, *e, *p;
mzta 0:158c61bb030f 1512
mzta 0:158c61bb030f 1513 if (RSTRING_LEN(str) <= 1) return mrb_str_dup(mrb, str);
mzta 0:158c61bb030f 1514
mzta 0:158c61bb030f 1515 s2 = str_new(mrb, 0, RSTRING_LEN(str));
mzta 0:158c61bb030f 1516 str_with_class(mrb, s2, str);
mzta 0:158c61bb030f 1517 s = RSTRING_PTR(str); e = RSTRING_END(str) - 1;
mzta 0:158c61bb030f 1518 p = RSTR_PTR(s2);
mzta 0:158c61bb030f 1519
mzta 0:158c61bb030f 1520 while (e >= s) {
mzta 0:158c61bb030f 1521 *p++ = *e--;
mzta 0:158c61bb030f 1522 }
mzta 0:158c61bb030f 1523 return mrb_obj_value(s2);
mzta 0:158c61bb030f 1524 }
mzta 0:158c61bb030f 1525
mzta 0:158c61bb030f 1526 /* 15.2.10.5.30 */
mzta 0:158c61bb030f 1527 /*
mzta 0:158c61bb030f 1528 * call-seq:
mzta 0:158c61bb030f 1529 * str.reverse! => str
mzta 0:158c61bb030f 1530 *
mzta 0:158c61bb030f 1531 * Reverses <i>str</i> in place.
mzta 0:158c61bb030f 1532 */
mzta 0:158c61bb030f 1533 static mrb_value
mzta 0:158c61bb030f 1534 mrb_str_reverse_bang(mrb_state *mrb, mrb_value str)
mzta 0:158c61bb030f 1535 {
mzta 0:158c61bb030f 1536 struct RString *s = mrb_str_ptr(str);
mzta 0:158c61bb030f 1537 char *p, *e;
mzta 0:158c61bb030f 1538 char c;
mzta 0:158c61bb030f 1539
mzta 0:158c61bb030f 1540 mrb_str_modify(mrb, s);
mzta 0:158c61bb030f 1541 if (RSTR_LEN(s) > 1) {
mzta 0:158c61bb030f 1542 p = RSTR_PTR(s);
mzta 0:158c61bb030f 1543 e = p + RSTR_LEN(s) - 1;
mzta 0:158c61bb030f 1544 while (p < e) {
mzta 0:158c61bb030f 1545 c = *p;
mzta 0:158c61bb030f 1546 *p++ = *e;
mzta 0:158c61bb030f 1547 *e-- = c;
mzta 0:158c61bb030f 1548 }
mzta 0:158c61bb030f 1549 }
mzta 0:158c61bb030f 1550 return str;
mzta 0:158c61bb030f 1551 }
mzta 0:158c61bb030f 1552
mzta 0:158c61bb030f 1553 /*
mzta 0:158c61bb030f 1554 * call-seq:
mzta 0:158c61bb030f 1555 * str.rindex(substring [, fixnum]) => fixnum or nil
mzta 0:158c61bb030f 1556 * str.rindex(fixnum [, fixnum]) => fixnum or nil
mzta 0:158c61bb030f 1557 * str.rindex(regexp [, fixnum]) => fixnum or nil
mzta 0:158c61bb030f 1558 *
mzta 0:158c61bb030f 1559 * Returns the index of the last occurrence of the given <i>substring</i>,
mzta 0:158c61bb030f 1560 * character (<i>fixnum</i>), or pattern (<i>regexp</i>) in <i>str</i>. Returns
mzta 0:158c61bb030f 1561 * <code>nil</code> if not found. If the second parameter is present, it
mzta 0:158c61bb030f 1562 * specifies the position in the string to end the search---characters beyond
mzta 0:158c61bb030f 1563 * this point will not be considered.
mzta 0:158c61bb030f 1564 *
mzta 0:158c61bb030f 1565 * "hello".rindex('e') #=> 1
mzta 0:158c61bb030f 1566 * "hello".rindex('l') #=> 3
mzta 0:158c61bb030f 1567 * "hello".rindex('a') #=> nil
mzta 0:158c61bb030f 1568 * "hello".rindex(101) #=> 1
mzta 0:158c61bb030f 1569 * "hello".rindex(/[aeiou]/, -2) #=> 1
mzta 0:158c61bb030f 1570 */
mzta 0:158c61bb030f 1571 static mrb_int
mzta 0:158c61bb030f 1572 mrb_str_rindex(mrb_state *mrb, mrb_value str, mrb_value sub, mrb_int pos)
mzta 0:158c61bb030f 1573 {
mzta 0:158c61bb030f 1574 char *s, *sbeg, *t;
mzta 0:158c61bb030f 1575 struct RString *ps = mrb_str_ptr(str);
mzta 0:158c61bb030f 1576 mrb_int len = RSTRING_LEN(sub);
mzta 0:158c61bb030f 1577
mzta 0:158c61bb030f 1578 /* substring longer than string */
mzta 0:158c61bb030f 1579 if (RSTR_LEN(ps) < len) return -1;
mzta 0:158c61bb030f 1580 if (RSTR_LEN(ps) - pos < len) {
mzta 0:158c61bb030f 1581 pos = RSTR_LEN(ps) - len;
mzta 0:158c61bb030f 1582 }
mzta 0:158c61bb030f 1583 sbeg = RSTR_PTR(ps);
mzta 0:158c61bb030f 1584 s = RSTR_PTR(ps) + pos;
mzta 0:158c61bb030f 1585 t = RSTRING_PTR(sub);
mzta 0:158c61bb030f 1586 if (len) {
mzta 0:158c61bb030f 1587 while (sbeg <= s) {
mzta 0:158c61bb030f 1588 if (memcmp(s, t, len) == 0) {
mzta 0:158c61bb030f 1589 return s - RSTR_PTR(ps);
mzta 0:158c61bb030f 1590 }
mzta 0:158c61bb030f 1591 s--;
mzta 0:158c61bb030f 1592 }
mzta 0:158c61bb030f 1593 return -1;
mzta 0:158c61bb030f 1594 }
mzta 0:158c61bb030f 1595 else {
mzta 0:158c61bb030f 1596 return pos;
mzta 0:158c61bb030f 1597 }
mzta 0:158c61bb030f 1598 }
mzta 0:158c61bb030f 1599
mzta 0:158c61bb030f 1600 /* 15.2.10.5.31 */
mzta 0:158c61bb030f 1601 /*
mzta 0:158c61bb030f 1602 * call-seq:
mzta 0:158c61bb030f 1603 * str.rindex(substring [, fixnum]) => fixnum or nil
mzta 0:158c61bb030f 1604 * str.rindex(fixnum [, fixnum]) => fixnum or nil
mzta 0:158c61bb030f 1605 * str.rindex(regexp [, fixnum]) => fixnum or nil
mzta 0:158c61bb030f 1606 *
mzta 0:158c61bb030f 1607 * Returns the index of the last occurrence of the given <i>substring</i>,
mzta 0:158c61bb030f 1608 * character (<i>fixnum</i>), or pattern (<i>regexp</i>) in <i>str</i>. Returns
mzta 0:158c61bb030f 1609 * <code>nil</code> if not found. If the second parameter is present, it
mzta 0:158c61bb030f 1610 * specifies the position in the string to end the search---characters beyond
mzta 0:158c61bb030f 1611 * this point will not be considered.
mzta 0:158c61bb030f 1612 *
mzta 0:158c61bb030f 1613 * "hello".rindex('e') #=> 1
mzta 0:158c61bb030f 1614 * "hello".rindex('l') #=> 3
mzta 0:158c61bb030f 1615 * "hello".rindex('a') #=> nil
mzta 0:158c61bb030f 1616 * "hello".rindex(101) #=> 1
mzta 0:158c61bb030f 1617 * "hello".rindex(/[aeiou]/, -2) #=> 1
mzta 0:158c61bb030f 1618 */
mzta 0:158c61bb030f 1619 static mrb_value
mzta 0:158c61bb030f 1620 mrb_str_rindex_m(mrb_state *mrb, mrb_value str)
mzta 0:158c61bb030f 1621 {
mzta 0:158c61bb030f 1622 mrb_value *argv;
mzta 0:158c61bb030f 1623 mrb_int argc;
mzta 0:158c61bb030f 1624 mrb_value sub;
mzta 0:158c61bb030f 1625 mrb_value vpos;
mzta 0:158c61bb030f 1626 mrb_int pos, len = RSTRING_LEN(str);
mzta 0:158c61bb030f 1627
mzta 0:158c61bb030f 1628 mrb_get_args(mrb, "*", &argv, &argc);
mzta 0:158c61bb030f 1629 if (argc == 2) {
mzta 0:158c61bb030f 1630 sub = argv[0];
mzta 0:158c61bb030f 1631 vpos = argv[1];
mzta 0:158c61bb030f 1632 pos = mrb_fixnum(vpos);
mzta 0:158c61bb030f 1633 if (pos < 0) {
mzta 0:158c61bb030f 1634 pos += len;
mzta 0:158c61bb030f 1635 if (pos < 0) {
mzta 0:158c61bb030f 1636 mrb_regexp_check(mrb, sub);
mzta 0:158c61bb030f 1637 return mrb_nil_value();
mzta 0:158c61bb030f 1638 }
mzta 0:158c61bb030f 1639 }
mzta 0:158c61bb030f 1640 if (pos > len) pos = len;
mzta 0:158c61bb030f 1641 }
mzta 0:158c61bb030f 1642 else {
mzta 0:158c61bb030f 1643 pos = len;
mzta 0:158c61bb030f 1644 if (argc > 0)
mzta 0:158c61bb030f 1645 sub = argv[0];
mzta 0:158c61bb030f 1646 else
mzta 0:158c61bb030f 1647 sub = mrb_nil_value();
mzta 0:158c61bb030f 1648 }
mzta 0:158c61bb030f 1649 mrb_regexp_check(mrb, sub);
mzta 0:158c61bb030f 1650
mzta 0:158c61bb030f 1651 switch (mrb_type(sub)) {
mzta 0:158c61bb030f 1652 case MRB_TT_FIXNUM: {
mzta 0:158c61bb030f 1653 int c = mrb_fixnum(sub);
mzta 0:158c61bb030f 1654 unsigned char *p = (unsigned char*)RSTRING_PTR(str);
mzta 0:158c61bb030f 1655
mzta 0:158c61bb030f 1656 for (pos=len-1;pos>=0;pos--) {
mzta 0:158c61bb030f 1657 if (p[pos] == c) return mrb_fixnum_value(pos);
mzta 0:158c61bb030f 1658 }
mzta 0:158c61bb030f 1659 return mrb_nil_value();
mzta 0:158c61bb030f 1660 }
mzta 0:158c61bb030f 1661
mzta 0:158c61bb030f 1662 default: {
mzta 0:158c61bb030f 1663 mrb_value tmp;
mzta 0:158c61bb030f 1664
mzta 0:158c61bb030f 1665 tmp = mrb_check_string_type(mrb, sub);
mzta 0:158c61bb030f 1666 if (mrb_nil_p(tmp)) {
mzta 0:158c61bb030f 1667 mrb_raisef(mrb, E_TYPE_ERROR, "type mismatch: %S given", sub);
mzta 0:158c61bb030f 1668 }
mzta 0:158c61bb030f 1669 sub = tmp;
mzta 0:158c61bb030f 1670 }
mzta 0:158c61bb030f 1671 /* fall through */
mzta 0:158c61bb030f 1672 case MRB_TT_STRING:
mzta 0:158c61bb030f 1673 pos = mrb_str_rindex(mrb, str, sub, pos);
mzta 0:158c61bb030f 1674 if (pos >= 0) return mrb_fixnum_value(pos);
mzta 0:158c61bb030f 1675 break;
mzta 0:158c61bb030f 1676
mzta 0:158c61bb030f 1677 } /* end of switch (TYPE(sub)) */
mzta 0:158c61bb030f 1678 return mrb_nil_value();
mzta 0:158c61bb030f 1679 }
mzta 0:158c61bb030f 1680
mzta 0:158c61bb030f 1681 /* 15.2.10.5.35 */
mzta 0:158c61bb030f 1682
mzta 0:158c61bb030f 1683 /*
mzta 0:158c61bb030f 1684 * call-seq:
mzta 0:158c61bb030f 1685 * str.split(pattern=$;, [limit]) => anArray
mzta 0:158c61bb030f 1686 *
mzta 0:158c61bb030f 1687 * Divides <i>str</i> into substrings based on a delimiter, returning an array
mzta 0:158c61bb030f 1688 * of these substrings.
mzta 0:158c61bb030f 1689 *
mzta 0:158c61bb030f 1690 * If <i>pattern</i> is a <code>String</code>, then its contents are used as
mzta 0:158c61bb030f 1691 * the delimiter when splitting <i>str</i>. If <i>pattern</i> is a single
mzta 0:158c61bb030f 1692 * space, <i>str</i> is split on whitespace, with leading whitespace and runs
mzta 0:158c61bb030f 1693 * of contiguous whitespace characters ignored.
mzta 0:158c61bb030f 1694 *
mzta 0:158c61bb030f 1695 * If <i>pattern</i> is a <code>Regexp</code>, <i>str</i> is divided where the
mzta 0:158c61bb030f 1696 * pattern matches. Whenever the pattern matches a zero-length string,
mzta 0:158c61bb030f 1697 * <i>str</i> is split into individual characters.
mzta 0:158c61bb030f 1698 *
mzta 0:158c61bb030f 1699 * If <i>pattern</i> is omitted, the value of <code>$;</code> is used. If
mzta 0:158c61bb030f 1700 * <code>$;</code> is <code>nil</code> (which is the default), <i>str</i> is
mzta 0:158c61bb030f 1701 * split on whitespace as if ` ' were specified.
mzta 0:158c61bb030f 1702 *
mzta 0:158c61bb030f 1703 * If the <i>limit</i> parameter is omitted, trailing null fields are
mzta 0:158c61bb030f 1704 * suppressed. If <i>limit</i> is a positive number, at most that number of
mzta 0:158c61bb030f 1705 * fields will be returned (if <i>limit</i> is <code>1</code>, the entire
mzta 0:158c61bb030f 1706 * string is returned as the only entry in an array). If negative, there is no
mzta 0:158c61bb030f 1707 * limit to the number of fields returned, and trailing null fields are not
mzta 0:158c61bb030f 1708 * suppressed.
mzta 0:158c61bb030f 1709 *
mzta 0:158c61bb030f 1710 * " now's the time".split #=> ["now's", "the", "time"]
mzta 0:158c61bb030f 1711 * " now's the time".split(' ') #=> ["now's", "the", "time"]
mzta 0:158c61bb030f 1712 * " now's the time".split(/ /) #=> ["", "now's", "", "the", "time"]
mzta 0:158c61bb030f 1713 * "1, 2.34,56, 7".split(%r{,\s*}) #=> ["1", "2.34", "56", "7"]
mzta 0:158c61bb030f 1714 * "hello".split(//) #=> ["h", "e", "l", "l", "o"]
mzta 0:158c61bb030f 1715 * "hello".split(//, 3) #=> ["h", "e", "llo"]
mzta 0:158c61bb030f 1716 * "hi mom".split(%r{\s*}) #=> ["h", "i", "m", "o", "m"]
mzta 0:158c61bb030f 1717 *
mzta 0:158c61bb030f 1718 * "mellow yellow".split("ello") #=> ["m", "w y", "w"]
mzta 0:158c61bb030f 1719 * "1,2,,3,4,,".split(',') #=> ["1", "2", "", "3", "4"]
mzta 0:158c61bb030f 1720 * "1,2,,3,4,,".split(',', 4) #=> ["1", "2", "", "3,4,,"]
mzta 0:158c61bb030f 1721 * "1,2,,3,4,,".split(',', -4) #=> ["1", "2", "", "3", "4", "", ""]
mzta 0:158c61bb030f 1722 */
mzta 0:158c61bb030f 1723
mzta 0:158c61bb030f 1724 static mrb_value
mzta 0:158c61bb030f 1725 mrb_str_split_m(mrb_state *mrb, mrb_value str)
mzta 0:158c61bb030f 1726 {
mzta 0:158c61bb030f 1727 int argc;
mzta 0:158c61bb030f 1728 mrb_value spat = mrb_nil_value();
mzta 0:158c61bb030f 1729 enum {awk, string, regexp} split_type = string;
mzta 0:158c61bb030f 1730 long i = 0, lim_p;
mzta 0:158c61bb030f 1731 mrb_int beg;
mzta 0:158c61bb030f 1732 mrb_int end;
mzta 0:158c61bb030f 1733 mrb_int lim = 0;
mzta 0:158c61bb030f 1734 mrb_value result, tmp;
mzta 0:158c61bb030f 1735
mzta 0:158c61bb030f 1736 argc = mrb_get_args(mrb, "|oi", &spat, &lim);
mzta 0:158c61bb030f 1737 lim_p = (lim > 0 && argc == 2);
mzta 0:158c61bb030f 1738 if (argc == 2) {
mzta 0:158c61bb030f 1739 if (lim == 1) {
mzta 0:158c61bb030f 1740 if (RSTRING_LEN(str) == 0)
mzta 0:158c61bb030f 1741 return mrb_ary_new_capa(mrb, 0);
mzta 0:158c61bb030f 1742 return mrb_ary_new_from_values(mrb, 1, &str);
mzta 0:158c61bb030f 1743 }
mzta 0:158c61bb030f 1744 i = 1;
mzta 0:158c61bb030f 1745 }
mzta 0:158c61bb030f 1746
mzta 0:158c61bb030f 1747 if (argc == 0 || mrb_nil_p(spat)) {
mzta 0:158c61bb030f 1748 split_type = awk;
mzta 0:158c61bb030f 1749 }
mzta 0:158c61bb030f 1750 else {
mzta 0:158c61bb030f 1751 if (mrb_string_p(spat)) {
mzta 0:158c61bb030f 1752 split_type = string;
mzta 0:158c61bb030f 1753 if (RSTRING_LEN(spat) == 1 && RSTRING_PTR(spat)[0] == ' ') {
mzta 0:158c61bb030f 1754 split_type = awk;
mzta 0:158c61bb030f 1755 }
mzta 0:158c61bb030f 1756 }
mzta 0:158c61bb030f 1757 else {
mzta 0:158c61bb030f 1758 mrb_noregexp(mrb, str);
mzta 0:158c61bb030f 1759 }
mzta 0:158c61bb030f 1760 }
mzta 0:158c61bb030f 1761
mzta 0:158c61bb030f 1762 result = mrb_ary_new(mrb);
mzta 0:158c61bb030f 1763 beg = 0;
mzta 0:158c61bb030f 1764 if (split_type == awk) {
mzta 0:158c61bb030f 1765 char *ptr = RSTRING_PTR(str);
mzta 0:158c61bb030f 1766 char *eptr = RSTRING_END(str);
mzta 0:158c61bb030f 1767 char *bptr = ptr;
mzta 0:158c61bb030f 1768 mrb_bool skip = TRUE;
mzta 0:158c61bb030f 1769 unsigned int c;
mzta 0:158c61bb030f 1770
mzta 0:158c61bb030f 1771 end = beg;
mzta 0:158c61bb030f 1772 while (ptr < eptr) {
mzta 0:158c61bb030f 1773 int ai = mrb_gc_arena_save(mrb);
mzta 0:158c61bb030f 1774 c = (unsigned char)*ptr++;
mzta 0:158c61bb030f 1775 if (skip) {
mzta 0:158c61bb030f 1776 if (ISSPACE(c)) {
mzta 0:158c61bb030f 1777 beg = ptr - bptr;
mzta 0:158c61bb030f 1778 }
mzta 0:158c61bb030f 1779 else {
mzta 0:158c61bb030f 1780 end = ptr - bptr;
mzta 0:158c61bb030f 1781 skip = FALSE;
mzta 0:158c61bb030f 1782 if (lim_p && lim <= i) break;
mzta 0:158c61bb030f 1783 }
mzta 0:158c61bb030f 1784 }
mzta 0:158c61bb030f 1785 else if (ISSPACE(c)) {
mzta 0:158c61bb030f 1786 mrb_ary_push(mrb, result, mrb_str_subseq(mrb, str, beg, end-beg));
mzta 0:158c61bb030f 1787 mrb_gc_arena_restore(mrb, ai);
mzta 0:158c61bb030f 1788 skip = TRUE;
mzta 0:158c61bb030f 1789 beg = ptr - bptr;
mzta 0:158c61bb030f 1790 if (lim_p) ++i;
mzta 0:158c61bb030f 1791 }
mzta 0:158c61bb030f 1792 else {
mzta 0:158c61bb030f 1793 end = ptr - bptr;
mzta 0:158c61bb030f 1794 }
mzta 0:158c61bb030f 1795 }
mzta 0:158c61bb030f 1796 }
mzta 0:158c61bb030f 1797 else if (split_type == string) {
mzta 0:158c61bb030f 1798 char *ptr = RSTRING_PTR(str); /* s->as.ary */
mzta 0:158c61bb030f 1799 char *temp = ptr;
mzta 0:158c61bb030f 1800 char *eptr = RSTRING_END(str);
mzta 0:158c61bb030f 1801 mrb_int slen = RSTRING_LEN(spat);
mzta 0:158c61bb030f 1802
mzta 0:158c61bb030f 1803 if (slen == 0) {
mzta 0:158c61bb030f 1804 int ai = mrb_gc_arena_save(mrb);
mzta 0:158c61bb030f 1805 while (ptr < eptr) {
mzta 0:158c61bb030f 1806 mrb_ary_push(mrb, result, mrb_str_subseq(mrb, str, ptr-temp, 1));
mzta 0:158c61bb030f 1807 mrb_gc_arena_restore(mrb, ai);
mzta 0:158c61bb030f 1808 ptr++;
mzta 0:158c61bb030f 1809 if (lim_p && lim <= ++i) break;
mzta 0:158c61bb030f 1810 }
mzta 0:158c61bb030f 1811 }
mzta 0:158c61bb030f 1812 else {
mzta 0:158c61bb030f 1813 char *sptr = RSTRING_PTR(spat);
mzta 0:158c61bb030f 1814 int ai = mrb_gc_arena_save(mrb);
mzta 0:158c61bb030f 1815
mzta 0:158c61bb030f 1816 while (ptr < eptr &&
mzta 0:158c61bb030f 1817 (end = mrb_memsearch(sptr, slen, ptr, eptr - ptr)) >= 0) {
mzta 0:158c61bb030f 1818 mrb_ary_push(mrb, result, mrb_str_subseq(mrb, str, ptr - temp, end));
mzta 0:158c61bb030f 1819 mrb_gc_arena_restore(mrb, ai);
mzta 0:158c61bb030f 1820 ptr += end + slen;
mzta 0:158c61bb030f 1821 if (lim_p && lim <= ++i) break;
mzta 0:158c61bb030f 1822 }
mzta 0:158c61bb030f 1823 }
mzta 0:158c61bb030f 1824 beg = ptr - temp;
mzta 0:158c61bb030f 1825 }
mzta 0:158c61bb030f 1826 else {
mzta 0:158c61bb030f 1827 mrb_noregexp(mrb, str);
mzta 0:158c61bb030f 1828 }
mzta 0:158c61bb030f 1829 if (RSTRING_LEN(str) > 0 && (lim_p || RSTRING_LEN(str) > beg || lim < 0)) {
mzta 0:158c61bb030f 1830 if (RSTRING_LEN(str) == beg) {
mzta 0:158c61bb030f 1831 tmp = mrb_str_new_empty(mrb, str);
mzta 0:158c61bb030f 1832 }
mzta 0:158c61bb030f 1833 else {
mzta 0:158c61bb030f 1834 tmp = mrb_str_subseq(mrb, str, beg, RSTRING_LEN(str)-beg);
mzta 0:158c61bb030f 1835 }
mzta 0:158c61bb030f 1836 mrb_ary_push(mrb, result, tmp);
mzta 0:158c61bb030f 1837 }
mzta 0:158c61bb030f 1838 if (!lim_p && lim == 0) {
mzta 0:158c61bb030f 1839 mrb_int len;
mzta 0:158c61bb030f 1840 while ((len = RARRAY_LEN(result)) > 0 &&
mzta 0:158c61bb030f 1841 (tmp = RARRAY_PTR(result)[len-1], RSTRING_LEN(tmp) == 0))
mzta 0:158c61bb030f 1842 mrb_ary_pop(mrb, result);
mzta 0:158c61bb030f 1843 }
mzta 0:158c61bb030f 1844
mzta 0:158c61bb030f 1845 return result;
mzta 0:158c61bb030f 1846 }
mzta 0:158c61bb030f 1847
mzta 0:158c61bb030f 1848 MRB_API mrb_value
mzta 0:158c61bb030f 1849 mrb_cstr_to_inum(mrb_state *mrb, const char *str, int base, int badcheck)
mzta 0:158c61bb030f 1850 {
mzta 0:158c61bb030f 1851 const char *p;
mzta 0:158c61bb030f 1852 char sign = 1;
mzta 0:158c61bb030f 1853 int c, uscore;
mzta 0:158c61bb030f 1854 unsigned long n = 0;
mzta 0:158c61bb030f 1855 mrb_int val;
mzta 0:158c61bb030f 1856
mzta 0:158c61bb030f 1857 #define conv_digit(c) \
mzta 0:158c61bb030f 1858 (ISDIGIT(c) ? ((c) - '0') : \
mzta 0:158c61bb030f 1859 ISLOWER(c) ? ((c) - 'a' + 10) : \
mzta 0:158c61bb030f 1860 ISUPPER(c) ? ((c) - 'A' + 10) : \
mzta 0:158c61bb030f 1861 -1)
mzta 0:158c61bb030f 1862
mzta 0:158c61bb030f 1863 if (!str) {
mzta 0:158c61bb030f 1864 if (badcheck) goto bad;
mzta 0:158c61bb030f 1865 return mrb_fixnum_value(0);
mzta 0:158c61bb030f 1866 }
mzta 0:158c61bb030f 1867 while (ISSPACE(*str)) str++;
mzta 0:158c61bb030f 1868
mzta 0:158c61bb030f 1869 if (str[0] == '+') {
mzta 0:158c61bb030f 1870 str++;
mzta 0:158c61bb030f 1871 }
mzta 0:158c61bb030f 1872 else if (str[0] == '-') {
mzta 0:158c61bb030f 1873 str++;
mzta 0:158c61bb030f 1874 sign = 0;
mzta 0:158c61bb030f 1875 }
mzta 0:158c61bb030f 1876 if (str[0] == '+' || str[0] == '-') {
mzta 0:158c61bb030f 1877 if (badcheck) goto bad;
mzta 0:158c61bb030f 1878 return mrb_fixnum_value(0);
mzta 0:158c61bb030f 1879 }
mzta 0:158c61bb030f 1880 if (base <= 0) {
mzta 0:158c61bb030f 1881 if (str[0] == '0') {
mzta 0:158c61bb030f 1882 switch (str[1]) {
mzta 0:158c61bb030f 1883 case 'x': case 'X':
mzta 0:158c61bb030f 1884 base = 16;
mzta 0:158c61bb030f 1885 break;
mzta 0:158c61bb030f 1886 case 'b': case 'B':
mzta 0:158c61bb030f 1887 base = 2;
mzta 0:158c61bb030f 1888 break;
mzta 0:158c61bb030f 1889 case 'o': case 'O':
mzta 0:158c61bb030f 1890 base = 8;
mzta 0:158c61bb030f 1891 break;
mzta 0:158c61bb030f 1892 case 'd': case 'D':
mzta 0:158c61bb030f 1893 base = 10;
mzta 0:158c61bb030f 1894 break;
mzta 0:158c61bb030f 1895 default:
mzta 0:158c61bb030f 1896 base = 8;
mzta 0:158c61bb030f 1897 }
mzta 0:158c61bb030f 1898 }
mzta 0:158c61bb030f 1899 else if (base < -1) {
mzta 0:158c61bb030f 1900 base = -base;
mzta 0:158c61bb030f 1901 }
mzta 0:158c61bb030f 1902 else {
mzta 0:158c61bb030f 1903 base = 10;
mzta 0:158c61bb030f 1904 }
mzta 0:158c61bb030f 1905 }
mzta 0:158c61bb030f 1906 switch (base) {
mzta 0:158c61bb030f 1907 case 2:
mzta 0:158c61bb030f 1908 if (str[0] == '0' && (str[1] == 'b'||str[1] == 'B')) {
mzta 0:158c61bb030f 1909 str += 2;
mzta 0:158c61bb030f 1910 }
mzta 0:158c61bb030f 1911 break;
mzta 0:158c61bb030f 1912 case 3:
mzta 0:158c61bb030f 1913 break;
mzta 0:158c61bb030f 1914 case 8:
mzta 0:158c61bb030f 1915 if (str[0] == '0' && (str[1] == 'o'||str[1] == 'O')) {
mzta 0:158c61bb030f 1916 str += 2;
mzta 0:158c61bb030f 1917 }
mzta 0:158c61bb030f 1918 case 4: case 5: case 6: case 7:
mzta 0:158c61bb030f 1919 break;
mzta 0:158c61bb030f 1920 case 10:
mzta 0:158c61bb030f 1921 if (str[0] == '0' && (str[1] == 'd'||str[1] == 'D')) {
mzta 0:158c61bb030f 1922 str += 2;
mzta 0:158c61bb030f 1923 }
mzta 0:158c61bb030f 1924 case 9: case 11: case 12: case 13: case 14: case 15:
mzta 0:158c61bb030f 1925 break;
mzta 0:158c61bb030f 1926 case 16:
mzta 0:158c61bb030f 1927 if (str[0] == '0' && (str[1] == 'x'||str[1] == 'X')) {
mzta 0:158c61bb030f 1928 str += 2;
mzta 0:158c61bb030f 1929 }
mzta 0:158c61bb030f 1930 break;
mzta 0:158c61bb030f 1931 default:
mzta 0:158c61bb030f 1932 if (base < 2 || 36 < base) {
mzta 0:158c61bb030f 1933 mrb_raisef(mrb, E_ARGUMENT_ERROR, "illegal radix %S", mrb_fixnum_value(base));
mzta 0:158c61bb030f 1934 }
mzta 0:158c61bb030f 1935 break;
mzta 0:158c61bb030f 1936 } /* end of switch (base) { */
mzta 0:158c61bb030f 1937 if (*str == '0') { /* squeeze preceeding 0s */
mzta 0:158c61bb030f 1938 uscore = 0;
mzta 0:158c61bb030f 1939 while ((c = *++str) == '0' || c == '_') {
mzta 0:158c61bb030f 1940 if (c == '_') {
mzta 0:158c61bb030f 1941 if (++uscore >= 2)
mzta 0:158c61bb030f 1942 break;
mzta 0:158c61bb030f 1943 }
mzta 0:158c61bb030f 1944 else
mzta 0:158c61bb030f 1945 uscore = 0;
mzta 0:158c61bb030f 1946 }
mzta 0:158c61bb030f 1947 if (!(c = *str) || ISSPACE(c)) --str;
mzta 0:158c61bb030f 1948 }
mzta 0:158c61bb030f 1949 c = *str;
mzta 0:158c61bb030f 1950 c = conv_digit(c);
mzta 0:158c61bb030f 1951 if (c < 0 || c >= base) {
mzta 0:158c61bb030f 1952 if (badcheck) goto bad;
mzta 0:158c61bb030f 1953 return mrb_fixnum_value(0);
mzta 0:158c61bb030f 1954 }
mzta 0:158c61bb030f 1955
mzta 0:158c61bb030f 1956 uscore = 0;
mzta 0:158c61bb030f 1957 for (p=str;*p;p++) {
mzta 0:158c61bb030f 1958 if (*p == '_') {
mzta 0:158c61bb030f 1959 if (uscore == 0) {
mzta 0:158c61bb030f 1960 uscore++;
mzta 0:158c61bb030f 1961 continue;
mzta 0:158c61bb030f 1962 }
mzta 0:158c61bb030f 1963 if (badcheck) goto bad;
mzta 0:158c61bb030f 1964 break;
mzta 0:158c61bb030f 1965 }
mzta 0:158c61bb030f 1966 uscore = 0;
mzta 0:158c61bb030f 1967 c = conv_digit(*p);
mzta 0:158c61bb030f 1968 if (c < 0 || c >= base) {
mzta 0:158c61bb030f 1969 if (badcheck) goto bad;
mzta 0:158c61bb030f 1970 break;
mzta 0:158c61bb030f 1971 }
mzta 0:158c61bb030f 1972 n *= base;
mzta 0:158c61bb030f 1973 n += c;
mzta 0:158c61bb030f 1974 }
mzta 0:158c61bb030f 1975 if (n > MRB_INT_MAX) {
mzta 0:158c61bb030f 1976 mrb_raisef(mrb, E_ARGUMENT_ERROR, "string (%S) too big for integer", mrb_str_new_cstr(mrb, str));
mzta 0:158c61bb030f 1977 }
mzta 0:158c61bb030f 1978 val = n;
mzta 0:158c61bb030f 1979 if (badcheck) {
mzta 0:158c61bb030f 1980 if (p == str) goto bad; /* no number */
mzta 0:158c61bb030f 1981 while (*p && ISSPACE(*p)) p++;
mzta 0:158c61bb030f 1982 if (*p) goto bad; /* trailing garbage */
mzta 0:158c61bb030f 1983 }
mzta 0:158c61bb030f 1984
mzta 0:158c61bb030f 1985 return mrb_fixnum_value(sign ? val : -val);
mzta 0:158c61bb030f 1986 bad:
mzta 0:158c61bb030f 1987 mrb_raisef(mrb, E_ARGUMENT_ERROR, "invalid string for number(%S)", mrb_str_new_cstr(mrb, str));
mzta 0:158c61bb030f 1988 /* not reached */
mzta 0:158c61bb030f 1989 return mrb_fixnum_value(0);
mzta 0:158c61bb030f 1990 }
mzta 0:158c61bb030f 1991
mzta 0:158c61bb030f 1992 MRB_API const char*
mzta 0:158c61bb030f 1993 mrb_string_value_cstr(mrb_state *mrb, mrb_value *ptr)
mzta 0:158c61bb030f 1994 {
mzta 0:158c61bb030f 1995 struct RString *ps = mrb_str_ptr(*ptr);
mzta 0:158c61bb030f 1996 mrb_int len = mrb_str_strlen(mrb, ps);
mzta 0:158c61bb030f 1997 char *p = RSTR_PTR(ps);
mzta 0:158c61bb030f 1998
mzta 0:158c61bb030f 1999 if (!p || p[len] != '\0') {
mzta 0:158c61bb030f 2000 mrb_str_modify(mrb, ps);
mzta 0:158c61bb030f 2001 return RSTR_PTR(ps);
mzta 0:158c61bb030f 2002 }
mzta 0:158c61bb030f 2003 return p;
mzta 0:158c61bb030f 2004 }
mzta 0:158c61bb030f 2005
mzta 0:158c61bb030f 2006 MRB_API mrb_value
mzta 0:158c61bb030f 2007 mrb_str_to_inum(mrb_state *mrb, mrb_value str, mrb_int base, mrb_bool badcheck)
mzta 0:158c61bb030f 2008 {
mzta 0:158c61bb030f 2009 const char *s;
mzta 0:158c61bb030f 2010 mrb_int len;
mzta 0:158c61bb030f 2011
mzta 0:158c61bb030f 2012 str = mrb_str_to_str(mrb, str);
mzta 0:158c61bb030f 2013 if (badcheck) {
mzta 0:158c61bb030f 2014 s = mrb_string_value_cstr(mrb, &str);
mzta 0:158c61bb030f 2015 }
mzta 0:158c61bb030f 2016 else {
mzta 0:158c61bb030f 2017 s = RSTRING_PTR(str);
mzta 0:158c61bb030f 2018 }
mzta 0:158c61bb030f 2019 if (s) {
mzta 0:158c61bb030f 2020 len = RSTRING_LEN(str);
mzta 0:158c61bb030f 2021 if (s[len]) { /* no sentinel somehow */
mzta 0:158c61bb030f 2022 struct RString *temp_str = str_new(mrb, s, len);
mzta 0:158c61bb030f 2023 s = RSTR_PTR(temp_str);
mzta 0:158c61bb030f 2024 }
mzta 0:158c61bb030f 2025 }
mzta 0:158c61bb030f 2026 return mrb_cstr_to_inum(mrb, s, base, badcheck);
mzta 0:158c61bb030f 2027 }
mzta 0:158c61bb030f 2028
mzta 0:158c61bb030f 2029 /* 15.2.10.5.38 */
mzta 0:158c61bb030f 2030 /*
mzta 0:158c61bb030f 2031 * call-seq:
mzta 0:158c61bb030f 2032 * str.to_i(base=10) => integer
mzta 0:158c61bb030f 2033 *
mzta 0:158c61bb030f 2034 * Returns the result of interpreting leading characters in <i>str</i> as an
mzta 0:158c61bb030f 2035 * integer base <i>base</i> (between 2 and 36). Extraneous characters past the
mzta 0:158c61bb030f 2036 * end of a valid number are ignored. If there is not a valid number at the
mzta 0:158c61bb030f 2037 * start of <i>str</i>, <code>0</code> is returned. This method never raises an
mzta 0:158c61bb030f 2038 * exception.
mzta 0:158c61bb030f 2039 *
mzta 0:158c61bb030f 2040 * "12345".to_i #=> 12345
mzta 0:158c61bb030f 2041 * "99 red balloons".to_i #=> 99
mzta 0:158c61bb030f 2042 * "0a".to_i #=> 0
mzta 0:158c61bb030f 2043 * "0a".to_i(16) #=> 10
mzta 0:158c61bb030f 2044 * "hello".to_i #=> 0
mzta 0:158c61bb030f 2045 * "1100101".to_i(2) #=> 101
mzta 0:158c61bb030f 2046 * "1100101".to_i(8) #=> 294977
mzta 0:158c61bb030f 2047 * "1100101".to_i(10) #=> 1100101
mzta 0:158c61bb030f 2048 * "1100101".to_i(16) #=> 17826049
mzta 0:158c61bb030f 2049 */
mzta 0:158c61bb030f 2050 static mrb_value
mzta 0:158c61bb030f 2051 mrb_str_to_i(mrb_state *mrb, mrb_value self)
mzta 0:158c61bb030f 2052 {
mzta 0:158c61bb030f 2053 mrb_int base = 10;
mzta 0:158c61bb030f 2054
mzta 0:158c61bb030f 2055 mrb_get_args(mrb, "|i", &base);
mzta 0:158c61bb030f 2056 if (base < 0) {
mzta 0:158c61bb030f 2057 mrb_raisef(mrb, E_ARGUMENT_ERROR, "illegal radix %S", mrb_fixnum_value(base));
mzta 0:158c61bb030f 2058 }
mzta 0:158c61bb030f 2059 return mrb_str_to_inum(mrb, self, base, FALSE);
mzta 0:158c61bb030f 2060 }
mzta 0:158c61bb030f 2061
mzta 0:158c61bb030f 2062 MRB_API double
mzta 0:158c61bb030f 2063 mrb_cstr_to_dbl(mrb_state *mrb, const char * p, mrb_bool badcheck)
mzta 0:158c61bb030f 2064 {
mzta 0:158c61bb030f 2065 char *end;
mzta 0:158c61bb030f 2066 double d;
mzta 0:158c61bb030f 2067
mzta 0:158c61bb030f 2068 enum {max_width = 20};
mzta 0:158c61bb030f 2069
mzta 0:158c61bb030f 2070 if (!p) return 0.0;
mzta 0:158c61bb030f 2071 while (ISSPACE(*p)) p++;
mzta 0:158c61bb030f 2072
mzta 0:158c61bb030f 2073 if (!badcheck && p[0] == '0' && (p[1] == 'x' || p[1] == 'X')) {
mzta 0:158c61bb030f 2074 return 0.0;
mzta 0:158c61bb030f 2075 }
mzta 0:158c61bb030f 2076 d = strtod(p, &end);
mzta 0:158c61bb030f 2077 if (p == end) {
mzta 0:158c61bb030f 2078 if (badcheck) {
mzta 0:158c61bb030f 2079 bad:
mzta 0:158c61bb030f 2080 mrb_raisef(mrb, E_ARGUMENT_ERROR, "invalid string for float(%S)", mrb_str_new_cstr(mrb, p));
mzta 0:158c61bb030f 2081 /* not reached */
mzta 0:158c61bb030f 2082 }
mzta 0:158c61bb030f 2083 return d;
mzta 0:158c61bb030f 2084 }
mzta 0:158c61bb030f 2085 if (*end) {
mzta 0:158c61bb030f 2086 char buf[DBL_DIG * 4 + 10];
mzta 0:158c61bb030f 2087 char *n = buf;
mzta 0:158c61bb030f 2088 char *e = buf + sizeof(buf) - 1;
mzta 0:158c61bb030f 2089 char prev = 0;
mzta 0:158c61bb030f 2090
mzta 0:158c61bb030f 2091 while (p < end && n < e) prev = *n++ = *p++;
mzta 0:158c61bb030f 2092 while (*p) {
mzta 0:158c61bb030f 2093 if (*p == '_') {
mzta 0:158c61bb030f 2094 /* remove underscores between digits */
mzta 0:158c61bb030f 2095 if (badcheck) {
mzta 0:158c61bb030f 2096 if (n == buf || !ISDIGIT(prev)) goto bad;
mzta 0:158c61bb030f 2097 ++p;
mzta 0:158c61bb030f 2098 if (!ISDIGIT(*p)) goto bad;
mzta 0:158c61bb030f 2099 }
mzta 0:158c61bb030f 2100 else {
mzta 0:158c61bb030f 2101 while (*++p == '_');
mzta 0:158c61bb030f 2102 continue;
mzta 0:158c61bb030f 2103 }
mzta 0:158c61bb030f 2104 }
mzta 0:158c61bb030f 2105 prev = *p++;
mzta 0:158c61bb030f 2106 if (n < e) *n++ = prev;
mzta 0:158c61bb030f 2107 }
mzta 0:158c61bb030f 2108 *n = '\0';
mzta 0:158c61bb030f 2109 p = buf;
mzta 0:158c61bb030f 2110
mzta 0:158c61bb030f 2111 if (!badcheck && p[0] == '0' && (p[1] == 'x' || p[1] == 'X')) {
mzta 0:158c61bb030f 2112 return 0.0;
mzta 0:158c61bb030f 2113 }
mzta 0:158c61bb030f 2114
mzta 0:158c61bb030f 2115 d = strtod(p, &end);
mzta 0:158c61bb030f 2116 if (badcheck) {
mzta 0:158c61bb030f 2117 if (!end || p == end) goto bad;
mzta 0:158c61bb030f 2118 while (*end && ISSPACE(*end)) end++;
mzta 0:158c61bb030f 2119 if (*end) goto bad;
mzta 0:158c61bb030f 2120 }
mzta 0:158c61bb030f 2121 }
mzta 0:158c61bb030f 2122 return d;
mzta 0:158c61bb030f 2123 }
mzta 0:158c61bb030f 2124
mzta 0:158c61bb030f 2125 MRB_API double
mzta 0:158c61bb030f 2126 mrb_str_to_dbl(mrb_state *mrb, mrb_value str, mrb_bool badcheck)
mzta 0:158c61bb030f 2127 {
mzta 0:158c61bb030f 2128 char *s;
mzta 0:158c61bb030f 2129 mrb_int len;
mzta 0:158c61bb030f 2130
mzta 0:158c61bb030f 2131 str = mrb_str_to_str(mrb, str);
mzta 0:158c61bb030f 2132 s = RSTRING_PTR(str);
mzta 0:158c61bb030f 2133 len = RSTRING_LEN(str);
mzta 0:158c61bb030f 2134 if (s) {
mzta 0:158c61bb030f 2135 if (badcheck && memchr(s, '\0', len)) {
mzta 0:158c61bb030f 2136 mrb_raise(mrb, E_ARGUMENT_ERROR, "string for Float contains null byte");
mzta 0:158c61bb030f 2137 }
mzta 0:158c61bb030f 2138 if (s[len]) { /* no sentinel somehow */
mzta 0:158c61bb030f 2139 struct RString *temp_str = str_new(mrb, s, len);
mzta 0:158c61bb030f 2140 s = RSTR_PTR(temp_str);
mzta 0:158c61bb030f 2141 }
mzta 0:158c61bb030f 2142 }
mzta 0:158c61bb030f 2143 return mrb_cstr_to_dbl(mrb, s, badcheck);
mzta 0:158c61bb030f 2144 }
mzta 0:158c61bb030f 2145
mzta 0:158c61bb030f 2146 /* 15.2.10.5.39 */
mzta 0:158c61bb030f 2147 /*
mzta 0:158c61bb030f 2148 * call-seq:
mzta 0:158c61bb030f 2149 * str.to_f => float
mzta 0:158c61bb030f 2150 *
mzta 0:158c61bb030f 2151 * Returns the result of interpreting leading characters in <i>str</i> as a
mzta 0:158c61bb030f 2152 * floating point number. Extraneous characters past the end of a valid number
mzta 0:158c61bb030f 2153 * are ignored. If there is not a valid number at the start of <i>str</i>,
mzta 0:158c61bb030f 2154 * <code>0.0</code> is returned. This method never raises an exception.
mzta 0:158c61bb030f 2155 *
mzta 0:158c61bb030f 2156 * "123.45e1".to_f #=> 1234.5
mzta 0:158c61bb030f 2157 * "45.67 degrees".to_f #=> 45.67
mzta 0:158c61bb030f 2158 * "thx1138".to_f #=> 0.0
mzta 0:158c61bb030f 2159 */
mzta 0:158c61bb030f 2160 static mrb_value
mzta 0:158c61bb030f 2161 mrb_str_to_f(mrb_state *mrb, mrb_value self)
mzta 0:158c61bb030f 2162 {
mzta 0:158c61bb030f 2163 return mrb_float_value(mrb, mrb_str_to_dbl(mrb, self, FALSE));
mzta 0:158c61bb030f 2164 }
mzta 0:158c61bb030f 2165
mzta 0:158c61bb030f 2166 /* 15.2.10.5.40 */
mzta 0:158c61bb030f 2167 /*
mzta 0:158c61bb030f 2168 * call-seq:
mzta 0:158c61bb030f 2169 * str.to_s => str
mzta 0:158c61bb030f 2170 * str.to_str => str
mzta 0:158c61bb030f 2171 *
mzta 0:158c61bb030f 2172 * Returns the receiver.
mzta 0:158c61bb030f 2173 */
mzta 0:158c61bb030f 2174 static mrb_value
mzta 0:158c61bb030f 2175 mrb_str_to_s(mrb_state *mrb, mrb_value self)
mzta 0:158c61bb030f 2176 {
mzta 0:158c61bb030f 2177 if (mrb_obj_class(mrb, self) != mrb->string_class) {
mzta 0:158c61bb030f 2178 return mrb_str_dup(mrb, self);
mzta 0:158c61bb030f 2179 }
mzta 0:158c61bb030f 2180 return self;
mzta 0:158c61bb030f 2181 }
mzta 0:158c61bb030f 2182
mzta 0:158c61bb030f 2183 /* 15.2.10.5.43 */
mzta 0:158c61bb030f 2184 /*
mzta 0:158c61bb030f 2185 * call-seq:
mzta 0:158c61bb030f 2186 * str.upcase! => str or nil
mzta 0:158c61bb030f 2187 *
mzta 0:158c61bb030f 2188 * Upcases the contents of <i>str</i>, returning <code>nil</code> if no changes
mzta 0:158c61bb030f 2189 * were made.
mzta 0:158c61bb030f 2190 */
mzta 0:158c61bb030f 2191 static mrb_value
mzta 0:158c61bb030f 2192 mrb_str_upcase_bang(mrb_state *mrb, mrb_value str)
mzta 0:158c61bb030f 2193 {
mzta 0:158c61bb030f 2194 struct RString *s = mrb_str_ptr(str);
mzta 0:158c61bb030f 2195 char *p, *pend;
mzta 0:158c61bb030f 2196 mrb_bool modify = FALSE;
mzta 0:158c61bb030f 2197
mzta 0:158c61bb030f 2198 mrb_str_modify(mrb, s);
mzta 0:158c61bb030f 2199 p = RSTRING_PTR(str);
mzta 0:158c61bb030f 2200 pend = RSTRING_END(str);
mzta 0:158c61bb030f 2201 while (p < pend) {
mzta 0:158c61bb030f 2202 if (ISLOWER(*p)) {
mzta 0:158c61bb030f 2203 *p = TOUPPER(*p);
mzta 0:158c61bb030f 2204 modify = TRUE;
mzta 0:158c61bb030f 2205 }
mzta 0:158c61bb030f 2206 p++;
mzta 0:158c61bb030f 2207 }
mzta 0:158c61bb030f 2208
mzta 0:158c61bb030f 2209 if (modify) return str;
mzta 0:158c61bb030f 2210 return mrb_nil_value();
mzta 0:158c61bb030f 2211 }
mzta 0:158c61bb030f 2212
mzta 0:158c61bb030f 2213 /* 15.2.10.5.42 */
mzta 0:158c61bb030f 2214 /*
mzta 0:158c61bb030f 2215 * call-seq:
mzta 0:158c61bb030f 2216 * str.upcase => new_str
mzta 0:158c61bb030f 2217 *
mzta 0:158c61bb030f 2218 * Returns a copy of <i>str</i> with all lowercase letters replaced with their
mzta 0:158c61bb030f 2219 * uppercase counterparts. The operation is locale insensitive---only
mzta 0:158c61bb030f 2220 * characters ``a'' to ``z'' are affected.
mzta 0:158c61bb030f 2221 *
mzta 0:158c61bb030f 2222 * "hEllO".upcase #=> "HELLO"
mzta 0:158c61bb030f 2223 */
mzta 0:158c61bb030f 2224 static mrb_value
mzta 0:158c61bb030f 2225 mrb_str_upcase(mrb_state *mrb, mrb_value self)
mzta 0:158c61bb030f 2226 {
mzta 0:158c61bb030f 2227 mrb_value str;
mzta 0:158c61bb030f 2228
mzta 0:158c61bb030f 2229 str = mrb_str_dup(mrb, self);
mzta 0:158c61bb030f 2230 mrb_str_upcase_bang(mrb, str);
mzta 0:158c61bb030f 2231 return str;
mzta 0:158c61bb030f 2232 }
mzta 0:158c61bb030f 2233
mzta 0:158c61bb030f 2234 #define IS_EVSTR(p,e) ((p) < (e) && (*(p) == '$' || *(p) == '@' || *(p) == '{'))
mzta 0:158c61bb030f 2235
mzta 0:158c61bb030f 2236 /*
mzta 0:158c61bb030f 2237 * call-seq:
mzta 0:158c61bb030f 2238 * str.dump -> new_str
mzta 0:158c61bb030f 2239 *
mzta 0:158c61bb030f 2240 * Produces a version of <i>str</i> with all nonprinting characters replaced by
mzta 0:158c61bb030f 2241 * <code>\nnn</code> notation and all special characters escaped.
mzta 0:158c61bb030f 2242 */
mzta 0:158c61bb030f 2243 mrb_value
mzta 0:158c61bb030f 2244 mrb_str_dump(mrb_state *mrb, mrb_value str)
mzta 0:158c61bb030f 2245 {
mzta 0:158c61bb030f 2246 mrb_int len;
mzta 0:158c61bb030f 2247 const char *p, *pend;
mzta 0:158c61bb030f 2248 char *q;
mzta 0:158c61bb030f 2249 struct RString *result;
mzta 0:158c61bb030f 2250
mzta 0:158c61bb030f 2251 len = 2; /* "" */
mzta 0:158c61bb030f 2252 p = RSTRING_PTR(str); pend = p + RSTRING_LEN(str);
mzta 0:158c61bb030f 2253 while (p < pend) {
mzta 0:158c61bb030f 2254 unsigned char c = *p++;
mzta 0:158c61bb030f 2255 switch (c) {
mzta 0:158c61bb030f 2256 case '"': case '\\':
mzta 0:158c61bb030f 2257 case '\n': case '\r':
mzta 0:158c61bb030f 2258 case '\t': case '\f':
mzta 0:158c61bb030f 2259 case '\013': case '\010': case '\007': case '\033':
mzta 0:158c61bb030f 2260 len += 2;
mzta 0:158c61bb030f 2261 break;
mzta 0:158c61bb030f 2262
mzta 0:158c61bb030f 2263 case '#':
mzta 0:158c61bb030f 2264 len += IS_EVSTR(p, pend) ? 2 : 1;
mzta 0:158c61bb030f 2265 break;
mzta 0:158c61bb030f 2266
mzta 0:158c61bb030f 2267 default:
mzta 0:158c61bb030f 2268 if (ISPRINT(c)) {
mzta 0:158c61bb030f 2269 len++;
mzta 0:158c61bb030f 2270 }
mzta 0:158c61bb030f 2271 else {
mzta 0:158c61bb030f 2272 len += 4; /* \NNN */
mzta 0:158c61bb030f 2273 }
mzta 0:158c61bb030f 2274 break;
mzta 0:158c61bb030f 2275 }
mzta 0:158c61bb030f 2276 }
mzta 0:158c61bb030f 2277
mzta 0:158c61bb030f 2278 result = str_new(mrb, 0, len);
mzta 0:158c61bb030f 2279 str_with_class(mrb, result, str);
mzta 0:158c61bb030f 2280 p = RSTRING_PTR(str); pend = p + RSTRING_LEN(str);
mzta 0:158c61bb030f 2281 q = RSTR_PTR(result);
mzta 0:158c61bb030f 2282 *q++ = '"';
mzta 0:158c61bb030f 2283 while (p < pend) {
mzta 0:158c61bb030f 2284 unsigned char c = *p++;
mzta 0:158c61bb030f 2285
mzta 0:158c61bb030f 2286 switch (c) {
mzta 0:158c61bb030f 2287 case '"':
mzta 0:158c61bb030f 2288 case '\\':
mzta 0:158c61bb030f 2289 *q++ = '\\';
mzta 0:158c61bb030f 2290 *q++ = c;
mzta 0:158c61bb030f 2291 break;
mzta 0:158c61bb030f 2292
mzta 0:158c61bb030f 2293 case '\n':
mzta 0:158c61bb030f 2294 *q++ = '\\';
mzta 0:158c61bb030f 2295 *q++ = 'n';
mzta 0:158c61bb030f 2296 break;
mzta 0:158c61bb030f 2297
mzta 0:158c61bb030f 2298 case '\r':
mzta 0:158c61bb030f 2299 *q++ = '\\';
mzta 0:158c61bb030f 2300 *q++ = 'r';
mzta 0:158c61bb030f 2301 break;
mzta 0:158c61bb030f 2302
mzta 0:158c61bb030f 2303 case '\t':
mzta 0:158c61bb030f 2304 *q++ = '\\';
mzta 0:158c61bb030f 2305 *q++ = 't';
mzta 0:158c61bb030f 2306 break;
mzta 0:158c61bb030f 2307
mzta 0:158c61bb030f 2308 case '\f':
mzta 0:158c61bb030f 2309 *q++ = '\\';
mzta 0:158c61bb030f 2310 *q++ = 'f';
mzta 0:158c61bb030f 2311 break;
mzta 0:158c61bb030f 2312
mzta 0:158c61bb030f 2313 case '\013':
mzta 0:158c61bb030f 2314 *q++ = '\\';
mzta 0:158c61bb030f 2315 *q++ = 'v';
mzta 0:158c61bb030f 2316 break;
mzta 0:158c61bb030f 2317
mzta 0:158c61bb030f 2318 case '\010':
mzta 0:158c61bb030f 2319 *q++ = '\\';
mzta 0:158c61bb030f 2320 *q++ = 'b';
mzta 0:158c61bb030f 2321 break;
mzta 0:158c61bb030f 2322
mzta 0:158c61bb030f 2323 case '\007':
mzta 0:158c61bb030f 2324 *q++ = '\\';
mzta 0:158c61bb030f 2325 *q++ = 'a';
mzta 0:158c61bb030f 2326 break;
mzta 0:158c61bb030f 2327
mzta 0:158c61bb030f 2328 case '\033':
mzta 0:158c61bb030f 2329 *q++ = '\\';
mzta 0:158c61bb030f 2330 *q++ = 'e';
mzta 0:158c61bb030f 2331 break;
mzta 0:158c61bb030f 2332
mzta 0:158c61bb030f 2333 case '#':
mzta 0:158c61bb030f 2334 if (IS_EVSTR(p, pend)) *q++ = '\\';
mzta 0:158c61bb030f 2335 *q++ = '#';
mzta 0:158c61bb030f 2336 break;
mzta 0:158c61bb030f 2337
mzta 0:158c61bb030f 2338 default:
mzta 0:158c61bb030f 2339 if (ISPRINT(c)) {
mzta 0:158c61bb030f 2340 *q++ = c;
mzta 0:158c61bb030f 2341 }
mzta 0:158c61bb030f 2342 else {
mzta 0:158c61bb030f 2343 *q++ = '\\';
mzta 0:158c61bb030f 2344 q[2] = '0' + c % 8; c /= 8;
mzta 0:158c61bb030f 2345 q[1] = '0' + c % 8; c /= 8;
mzta 0:158c61bb030f 2346 q[0] = '0' + c % 8;
mzta 0:158c61bb030f 2347 q += 3;
mzta 0:158c61bb030f 2348 }
mzta 0:158c61bb030f 2349 }
mzta 0:158c61bb030f 2350 }
mzta 0:158c61bb030f 2351 *q = '"';
mzta 0:158c61bb030f 2352 return mrb_obj_value(result);
mzta 0:158c61bb030f 2353 }
mzta 0:158c61bb030f 2354
mzta 0:158c61bb030f 2355 MRB_API mrb_value
mzta 0:158c61bb030f 2356 mrb_str_cat(mrb_state *mrb, mrb_value str, const char *ptr, size_t len)
mzta 0:158c61bb030f 2357 {
mzta 0:158c61bb030f 2358 str_buf_cat(mrb, mrb_str_ptr(str), ptr, len);
mzta 0:158c61bb030f 2359 return str;
mzta 0:158c61bb030f 2360 }
mzta 0:158c61bb030f 2361
mzta 0:158c61bb030f 2362 MRB_API mrb_value
mzta 0:158c61bb030f 2363 mrb_str_cat_cstr(mrb_state *mrb, mrb_value str, const char *ptr)
mzta 0:158c61bb030f 2364 {
mzta 0:158c61bb030f 2365 return mrb_str_cat(mrb, str, ptr, strlen(ptr));
mzta 0:158c61bb030f 2366 }
mzta 0:158c61bb030f 2367
mzta 0:158c61bb030f 2368 MRB_API mrb_value
mzta 0:158c61bb030f 2369 mrb_str_cat_str(mrb_state *mrb, mrb_value str, mrb_value str2)
mzta 0:158c61bb030f 2370 {
mzta 0:158c61bb030f 2371 return mrb_str_cat(mrb, str, RSTRING_PTR(str2), RSTRING_LEN(str2));
mzta 0:158c61bb030f 2372 }
mzta 0:158c61bb030f 2373
mzta 0:158c61bb030f 2374 MRB_API mrb_value
mzta 0:158c61bb030f 2375 mrb_str_append(mrb_state *mrb, mrb_value str, mrb_value str2)
mzta 0:158c61bb030f 2376 {
mzta 0:158c61bb030f 2377 str2 = mrb_str_to_str(mrb, str2);
mzta 0:158c61bb030f 2378 return mrb_str_cat_str(mrb, str, str2);
mzta 0:158c61bb030f 2379 }
mzta 0:158c61bb030f 2380
mzta 0:158c61bb030f 2381 #define CHAR_ESC_LEN 13 /* sizeof(\x{ hex of 32bit unsigned int } \0) */
mzta 0:158c61bb030f 2382
mzta 0:158c61bb030f 2383 /*
mzta 0:158c61bb030f 2384 * call-seq:
mzta 0:158c61bb030f 2385 * str.inspect -> string
mzta 0:158c61bb030f 2386 *
mzta 0:158c61bb030f 2387 * Returns a printable version of _str_, surrounded by quote marks,
mzta 0:158c61bb030f 2388 * with special characters escaped.
mzta 0:158c61bb030f 2389 *
mzta 0:158c61bb030f 2390 * str = "hello"
mzta 0:158c61bb030f 2391 * str[3] = "\b"
mzta 0:158c61bb030f 2392 * str.inspect #=> "\"hel\\bo\""
mzta 0:158c61bb030f 2393 */
mzta 0:158c61bb030f 2394 mrb_value
mzta 0:158c61bb030f 2395 mrb_str_inspect(mrb_state *mrb, mrb_value str)
mzta 0:158c61bb030f 2396 {
mzta 0:158c61bb030f 2397 const char *p, *pend;
mzta 0:158c61bb030f 2398 char buf[CHAR_ESC_LEN + 1];
mzta 0:158c61bb030f 2399 mrb_value result = mrb_str_new_lit(mrb, "\"");
mzta 0:158c61bb030f 2400
mzta 0:158c61bb030f 2401 p = RSTRING_PTR(str); pend = RSTRING_END(str);
mzta 0:158c61bb030f 2402 for (;p < pend; p++) {
mzta 0:158c61bb030f 2403 unsigned char c, cc;
mzta 0:158c61bb030f 2404
mzta 0:158c61bb030f 2405 c = *p;
mzta 0:158c61bb030f 2406 if (c == '"'|| c == '\\' || (c == '#' && IS_EVSTR(p, pend))) {
mzta 0:158c61bb030f 2407 buf[0] = '\\'; buf[1] = c;
mzta 0:158c61bb030f 2408 mrb_str_cat(mrb, result, buf, 2);
mzta 0:158c61bb030f 2409 continue;
mzta 0:158c61bb030f 2410 }
mzta 0:158c61bb030f 2411 if (ISPRINT(c)) {
mzta 0:158c61bb030f 2412 buf[0] = c;
mzta 0:158c61bb030f 2413 mrb_str_cat(mrb, result, buf, 1);
mzta 0:158c61bb030f 2414 continue;
mzta 0:158c61bb030f 2415 }
mzta 0:158c61bb030f 2416 switch (c) {
mzta 0:158c61bb030f 2417 case '\n': cc = 'n'; break;
mzta 0:158c61bb030f 2418 case '\r': cc = 'r'; break;
mzta 0:158c61bb030f 2419 case '\t': cc = 't'; break;
mzta 0:158c61bb030f 2420 case '\f': cc = 'f'; break;
mzta 0:158c61bb030f 2421 case '\013': cc = 'v'; break;
mzta 0:158c61bb030f 2422 case '\010': cc = 'b'; break;
mzta 0:158c61bb030f 2423 case '\007': cc = 'a'; break;
mzta 0:158c61bb030f 2424 case 033: cc = 'e'; break;
mzta 0:158c61bb030f 2425 default: cc = 0; break;
mzta 0:158c61bb030f 2426 }
mzta 0:158c61bb030f 2427 if (cc) {
mzta 0:158c61bb030f 2428 buf[0] = '\\';
mzta 0:158c61bb030f 2429 buf[1] = (char)cc;
mzta 0:158c61bb030f 2430 mrb_str_cat(mrb, result, buf, 2);
mzta 0:158c61bb030f 2431 continue;
mzta 0:158c61bb030f 2432 }
mzta 0:158c61bb030f 2433 else {
mzta 0:158c61bb030f 2434 buf[0] = '\\';
mzta 0:158c61bb030f 2435 buf[3] = '0' + c % 8; c /= 8;
mzta 0:158c61bb030f 2436 buf[2] = '0' + c % 8; c /= 8;
mzta 0:158c61bb030f 2437 buf[1] = '0' + c % 8;
mzta 0:158c61bb030f 2438 mrb_str_cat(mrb, result, buf, 4);
mzta 0:158c61bb030f 2439 continue;
mzta 0:158c61bb030f 2440 }
mzta 0:158c61bb030f 2441 }
mzta 0:158c61bb030f 2442 mrb_str_cat_lit(mrb, result, "\"");
mzta 0:158c61bb030f 2443
mzta 0:158c61bb030f 2444 return result;
mzta 0:158c61bb030f 2445 }
mzta 0:158c61bb030f 2446
mzta 0:158c61bb030f 2447 /*
mzta 0:158c61bb030f 2448 * call-seq:
mzta 0:158c61bb030f 2449 * str.bytes -> array of fixnums
mzta 0:158c61bb030f 2450 *
mzta 0:158c61bb030f 2451 * Returns an array of bytes in _str_.
mzta 0:158c61bb030f 2452 *
mzta 0:158c61bb030f 2453 * str = "hello"
mzta 0:158c61bb030f 2454 * str.bytes #=> [104, 101, 108, 108, 111]
mzta 0:158c61bb030f 2455 */
mzta 0:158c61bb030f 2456 static mrb_value
mzta 0:158c61bb030f 2457 mrb_str_bytes(mrb_state *mrb, mrb_value str)
mzta 0:158c61bb030f 2458 {
mzta 0:158c61bb030f 2459 struct RString *s = mrb_str_ptr(str);
mzta 0:158c61bb030f 2460 mrb_value a = mrb_ary_new_capa(mrb, RSTR_LEN(s));
mzta 0:158c61bb030f 2461 unsigned char *p = (unsigned char *)(RSTR_PTR(s)), *pend = p + RSTR_LEN(s);
mzta 0:158c61bb030f 2462
mzta 0:158c61bb030f 2463 while (p < pend) {
mzta 0:158c61bb030f 2464 mrb_ary_push(mrb, a, mrb_fixnum_value(p[0]));
mzta 0:158c61bb030f 2465 p++;
mzta 0:158c61bb030f 2466 }
mzta 0:158c61bb030f 2467 return a;
mzta 0:158c61bb030f 2468 }
mzta 0:158c61bb030f 2469
mzta 0:158c61bb030f 2470 /* ---------------------------*/
mzta 0:158c61bb030f 2471 void
mzta 0:158c61bb030f 2472 mrb_init_string(mrb_state *mrb)
mzta 0:158c61bb030f 2473 {
mzta 0:158c61bb030f 2474 struct RClass *s;
mzta 0:158c61bb030f 2475
mzta 0:158c61bb030f 2476 mrb_static_assert(RSTRING_EMBED_LEN_MAX < (1 << 5), "pointer size too big for embedded string");
mzta 0:158c61bb030f 2477
mzta 0:158c61bb030f 2478 s = mrb->string_class = mrb_define_class(mrb, "String", mrb->object_class); /* 15.2.10 */
mzta 0:158c61bb030f 2479 MRB_SET_INSTANCE_TT(s, MRB_TT_STRING);
mzta 0:158c61bb030f 2480
mzta 0:158c61bb030f 2481 mrb_define_method(mrb, s, "bytesize", mrb_str_size, MRB_ARGS_NONE());
mzta 0:158c61bb030f 2482
mzta 0:158c61bb030f 2483 mrb_define_method(mrb, s, "<=>", mrb_str_cmp_m, MRB_ARGS_REQ(1)); /* 15.2.10.5.1 */
mzta 0:158c61bb030f 2484 mrb_define_method(mrb, s, "==", mrb_str_equal_m, MRB_ARGS_REQ(1)); /* 15.2.10.5.2 */
mzta 0:158c61bb030f 2485 mrb_define_method(mrb, s, "+", mrb_str_plus_m, MRB_ARGS_REQ(1)); /* 15.2.10.5.4 */
mzta 0:158c61bb030f 2486 mrb_define_method(mrb, s, "*", mrb_str_times, MRB_ARGS_REQ(1)); /* 15.2.10.5.5 */
mzta 0:158c61bb030f 2487 mrb_define_method(mrb, s, "[]", mrb_str_aref_m, MRB_ARGS_ANY()); /* 15.2.10.5.6 */
mzta 0:158c61bb030f 2488 mrb_define_method(mrb, s, "capitalize", mrb_str_capitalize, MRB_ARGS_NONE()); /* 15.2.10.5.7 */
mzta 0:158c61bb030f 2489 mrb_define_method(mrb, s, "capitalize!", mrb_str_capitalize_bang, MRB_ARGS_NONE()); /* 15.2.10.5.8 */
mzta 0:158c61bb030f 2490 mrb_define_method(mrb, s, "chomp", mrb_str_chomp, MRB_ARGS_ANY()); /* 15.2.10.5.9 */
mzta 0:158c61bb030f 2491 mrb_define_method(mrb, s, "chomp!", mrb_str_chomp_bang, MRB_ARGS_ANY()); /* 15.2.10.5.10 */
mzta 0:158c61bb030f 2492 mrb_define_method(mrb, s, "chop", mrb_str_chop, MRB_ARGS_REQ(1)); /* 15.2.10.5.11 */
mzta 0:158c61bb030f 2493 mrb_define_method(mrb, s, "chop!", mrb_str_chop_bang, MRB_ARGS_REQ(1)); /* 15.2.10.5.12 */
mzta 0:158c61bb030f 2494 mrb_define_method(mrb, s, "downcase", mrb_str_downcase, MRB_ARGS_NONE()); /* 15.2.10.5.13 */
mzta 0:158c61bb030f 2495 mrb_define_method(mrb, s, "downcase!", mrb_str_downcase_bang, MRB_ARGS_NONE()); /* 15.2.10.5.14 */
mzta 0:158c61bb030f 2496 mrb_define_method(mrb, s, "empty?", mrb_str_empty_p, MRB_ARGS_NONE()); /* 15.2.10.5.16 */
mzta 0:158c61bb030f 2497 mrb_define_method(mrb, s, "eql?", mrb_str_eql, MRB_ARGS_REQ(1)); /* 15.2.10.5.17 */
mzta 0:158c61bb030f 2498
mzta 0:158c61bb030f 2499 mrb_define_method(mrb, s, "hash", mrb_str_hash_m, MRB_ARGS_NONE()); /* 15.2.10.5.20 */
mzta 0:158c61bb030f 2500 mrb_define_method(mrb, s, "include?", mrb_str_include, MRB_ARGS_REQ(1)); /* 15.2.10.5.21 */
mzta 0:158c61bb030f 2501 mrb_define_method(mrb, s, "index", mrb_str_index_m, MRB_ARGS_ANY()); /* 15.2.10.5.22 */
mzta 0:158c61bb030f 2502 mrb_define_method(mrb, s, "initialize", mrb_str_init, MRB_ARGS_REQ(1)); /* 15.2.10.5.23 */
mzta 0:158c61bb030f 2503 mrb_define_method(mrb, s, "initialize_copy", mrb_str_replace, MRB_ARGS_REQ(1)); /* 15.2.10.5.24 */
mzta 0:158c61bb030f 2504 mrb_define_method(mrb, s, "intern", mrb_str_intern, MRB_ARGS_NONE()); /* 15.2.10.5.25 */
mzta 0:158c61bb030f 2505 mrb_define_method(mrb, s, "length", mrb_str_size, MRB_ARGS_NONE()); /* 15.2.10.5.26 */
mzta 0:158c61bb030f 2506 mrb_define_method(mrb, s, "replace", mrb_str_replace, MRB_ARGS_REQ(1)); /* 15.2.10.5.28 */
mzta 0:158c61bb030f 2507 mrb_define_method(mrb, s, "reverse", mrb_str_reverse, MRB_ARGS_NONE()); /* 15.2.10.5.29 */
mzta 0:158c61bb030f 2508 mrb_define_method(mrb, s, "reverse!", mrb_str_reverse_bang, MRB_ARGS_NONE()); /* 15.2.10.5.30 */
mzta 0:158c61bb030f 2509 mrb_define_method(mrb, s, "rindex", mrb_str_rindex_m, MRB_ARGS_ANY()); /* 15.2.10.5.31 */
mzta 0:158c61bb030f 2510 mrb_define_method(mrb, s, "size", mrb_str_size, MRB_ARGS_NONE()); /* 15.2.10.5.33 */
mzta 0:158c61bb030f 2511 mrb_define_method(mrb, s, "slice", mrb_str_aref_m, MRB_ARGS_ANY()); /* 15.2.10.5.34 */
mzta 0:158c61bb030f 2512 mrb_define_method(mrb, s, "split", mrb_str_split_m, MRB_ARGS_ANY()); /* 15.2.10.5.35 */
mzta 0:158c61bb030f 2513
mzta 0:158c61bb030f 2514 mrb_define_method(mrb, s, "to_f", mrb_str_to_f, MRB_ARGS_NONE()); /* 15.2.10.5.38 */
mzta 0:158c61bb030f 2515 mrb_define_method(mrb, s, "to_i", mrb_str_to_i, MRB_ARGS_ANY()); /* 15.2.10.5.39 */
mzta 0:158c61bb030f 2516 mrb_define_method(mrb, s, "to_s", mrb_str_to_s, MRB_ARGS_NONE()); /* 15.2.10.5.40 */
mzta 0:158c61bb030f 2517 mrb_define_method(mrb, s, "to_str", mrb_str_to_s, MRB_ARGS_NONE());
mzta 0:158c61bb030f 2518 mrb_define_method(mrb, s, "to_sym", mrb_str_intern, MRB_ARGS_NONE()); /* 15.2.10.5.41 */
mzta 0:158c61bb030f 2519 mrb_define_method(mrb, s, "upcase", mrb_str_upcase, MRB_ARGS_NONE()); /* 15.2.10.5.42 */
mzta 0:158c61bb030f 2520 mrb_define_method(mrb, s, "upcase!", mrb_str_upcase_bang, MRB_ARGS_NONE()); /* 15.2.10.5.43 */
mzta 0:158c61bb030f 2521 mrb_define_method(mrb, s, "inspect", mrb_str_inspect, MRB_ARGS_NONE()); /* 15.2.10.5.46(x) */
mzta 0:158c61bb030f 2522 mrb_define_method(mrb, s, "bytes", mrb_str_bytes, MRB_ARGS_NONE());
mzta 0:158c61bb030f 2523 }
mzta 0:158c61bb030f 2524