1 #include "first.h" 2 3 #include "buffer.h" 4 5 #include <stdlib.h> 6 #include <string.h> 7 #include "sys-time.h" /* strftime() */ 8 9 static const char hex_chars_lc[] = "0123456789abcdef"; 10 static const char hex_chars_uc[] = "0123456789ABCDEF"; 11 12 13 __attribute_noinline__ 14 buffer* buffer_init(void) { 15 buffer * const b = calloc(1, sizeof(*b)); 16 force_assert(b); 17 return b; 18 } 19 20 buffer *buffer_init_buffer(const buffer *src) { 21 buffer * const b = buffer_init(); 22 buffer_copy_string_len(b, BUF_PTR_LEN(src)); 23 return b; 24 } 25 26 buffer *buffer_init_string(const char *str) { 27 buffer * const b = buffer_init(); 28 buffer_copy_string(b, str); 29 return b; 30 } 31 32 void buffer_free(buffer *b) { 33 if (NULL == b) return; 34 35 free(b->ptr); 36 free(b); 37 } 38 39 void buffer_free_ptr(buffer *b) { 40 free(b->ptr); 41 b->ptr = NULL; 42 b->used = 0; 43 b->size = 0; 44 } 45 46 void buffer_move(buffer * restrict b, buffer * restrict src) { 47 buffer tmp; 48 buffer_clear(b); 49 tmp = *src; *src = *b; *b = tmp; 50 } 51 52 /* make sure buffer is at least "size" big + 1 for '\0'. keep old data */ 53 __attribute_cold__ 54 __attribute_noinline__ 55 __attribute_nonnull__() 56 __attribute_returns_nonnull__ 57 static char* buffer_realloc(buffer * const restrict b, const size_t len) { 58 #define BUFFER_PIECE_SIZE 64uL /*(must be power-of-2)*/ 59 size_t sz = (len + 1 + BUFFER_PIECE_SIZE-1) & ~(BUFFER_PIECE_SIZE-1); 60 force_assert(sz > len); 61 if ((sz & (sz-1)) && sz < INT_MAX) {/* not power-2; huge val not expected */ 62 /*(optimizer should recognize this and use ffs or clz or equivalent)*/ 63 const size_t psz = sz; 64 for (sz = 256; sz < psz; sz <<= 1) ; 65 } 66 sz |= 1; /*(extra +1 for '\0' when needed buffer size is exact power-2)*/ 67 68 b->size = sz; 69 b->ptr = realloc(b->ptr, sz); 70 71 force_assert(NULL != b->ptr); 72 return b->ptr; 73 } 74 75 __attribute_cold__ 76 __attribute_noinline__ 77 __attribute_nonnull__() 78 __attribute_returns_nonnull__ 79 static char* buffer_alloc_replace(buffer * const restrict b, const size_t size) { 80 /*(discard old data so realloc() does not copy)*/ 81 if (NULL != b->ptr) { 82 free(b->ptr); 83 b->ptr = NULL; 84 } 85 /*(note: if size larger than one lshift, use size instead of power-2)*/ 86 const size_t bsize2x = (b->size & ~1uL) << 1; 87 return buffer_realloc(b, bsize2x > size ? bsize2x-1 : size); 88 } 89 90 char* buffer_string_prepare_copy(buffer * const b, const size_t size) { 91 b->used = 0; 92 #ifdef __COVERITY__ /*(b->ptr is not NULL if b->size is not 0)*/ 93 force_assert(size >= b->size || b->ptr); 94 #endif 95 return (size < b->size) 96 ? b->ptr 97 : buffer_alloc_replace(b, size); 98 } 99 100 __attribute_cold__ 101 __attribute_noinline__ 102 __attribute_nonnull__() 103 __attribute_returns_nonnull__ 104 static char* buffer_string_prepare_append_resize(buffer * const restrict b, const size_t size) { 105 if (b->used < 2) { /* buffer_is_blank(b) */ 106 char * const s = buffer_string_prepare_copy(b, size); 107 *s = '\0'; /*(for case (1 == b->used))*/ 108 return s; 109 } 110 111 /* not empty, b->used already includes a terminating 0 */ 112 /*(note: if size larger than one lshift, use size instead of power-2)*/ 113 const size_t bsize2x = (b->size & ~1uL) << 1; 114 const size_t req_size = (bsize2x - b->used > size) 115 ? bsize2x-1 116 : b->used + size; 117 118 /* check for overflow: unsigned overflow is defined to wrap around */ 119 force_assert(req_size >= b->used); 120 121 return buffer_realloc(b, req_size) + b->used - 1; 122 } 123 124 char* buffer_string_prepare_append(buffer * const b, const size_t size) { 125 const uint32_t len = b->used ? b->used-1 : 0; 126 return (b->size - len >= size + 1) 127 ? b->ptr + len 128 : buffer_string_prepare_append_resize(b, size); 129 } 130 131 /*(prefer smaller code than inlining buffer_extend in many places in buffer.c)*/ 132 __attribute_noinline__ 133 char* 134 buffer_extend (buffer * const b, const size_t x) 135 { 136 /* extend buffer to append x (reallocate by power-2 (or larger), if needed) 137 * (combine buffer_string_prepare_append() and buffer_commit()) 138 * (future: might make buffer.h static inline func for HTTP/1.1 performance) 139 * pre-sets '\0' byte and b->used (unlike buffer_string_prepare_append())*/ 140 #if 0 141 char * const s = buffer_string_prepare_append(b, x); 142 b->used += x + (0 == b->used); 143 #else 144 const uint32_t len = b->used ? b->used-1 : 0; 145 char * const s = (b->size - len >= x + 1) 146 ? b->ptr + len 147 : buffer_string_prepare_append_resize(b, x); 148 b->used = len+x+1; 149 #endif 150 s[x] = '\0'; 151 return s; 152 } 153 154 void buffer_commit(buffer *b, size_t size) 155 { 156 size_t sz = b->used; 157 if (0 == sz) sz = 1; 158 159 if (size > 0) { 160 /* check for overflow: unsigned overflow is defined to wrap around */ 161 sz += size; 162 force_assert(sz > size); 163 } 164 165 b->used = sz; 166 b->ptr[sz - 1] = '\0'; 167 } 168 169 void buffer_copy_string(buffer * restrict b, const char * restrict s) { 170 buffer_copy_string_len(b, s, NULL != s ? strlen(s) : 0); 171 } 172 173 void buffer_copy_string_len(buffer * const restrict b, const char * const restrict s, const size_t len) { 174 b->used = len + 1; 175 char * const restrict d = (len < b->size) 176 ? b->ptr 177 : buffer_alloc_replace(b, len); 178 d[len] = '\0'; 179 memcpy(d, s, len); 180 } 181 182 void buffer_append_string(buffer * restrict b, const char * restrict s) { 183 buffer_append_string_len(b, s, NULL != s ? strlen(s) : 0); 184 } 185 186 /** 187 * append a string to the end of the buffer 188 * 189 * the resulting buffer is terminated with a '\0' 190 * s is treated as a un-terminated string (a \0 is handled a normal character) 191 * 192 * @param b a buffer 193 * @param s the string 194 * @param s_len size of the string (without the terminating \0) 195 */ 196 197 void buffer_append_string_len(buffer * const restrict b, const char * const restrict s, const size_t len) { 198 memcpy(buffer_extend(b, len), s, len); 199 } 200 201 void buffer_append_str2(buffer * const restrict b, const char * const s1, const size_t len1, const char * const s2, const size_t len2) { 202 char * const restrict s = buffer_extend(b, len1+len2); 203 #ifdef HAVE_MEMPCPY 204 mempcpy(mempcpy(s, s1, len1), s2, len2); 205 #else 206 memcpy(s, s1, len1); 207 memcpy(s+len1, s2, len2); 208 #endif 209 } 210 211 void buffer_append_str3(buffer * const restrict b, const char * const s1, const size_t len1, const char * const s2, const size_t len2, const char * const s3, const size_t len3) { 212 char * restrict s = buffer_extend(b, len1+len2+len3); 213 #ifdef HAVE_MEMPCPY 214 mempcpy(mempcpy(mempcpy(s, s1, len1), s2, len2), s3, len3); 215 #else 216 memcpy(s, s1, len1); 217 memcpy((s+=len1), s2, len2); 218 memcpy((s+=len2), s3, len3); 219 #endif 220 } 221 222 void buffer_append_iovec(buffer * const restrict b, const struct const_iovec * const iov, const size_t n) { 223 size_t len = 0; 224 for (size_t i = 0; i < n; ++i) 225 len += iov[i].iov_len; 226 char *s = buffer_extend(b, len); 227 for (size_t i = 0; i < n; ++i) { 228 if (0 == iov[i].iov_len) continue; 229 #ifdef HAVE_MEMPCPY 230 s = mempcpy(s, iov[i].iov_base, iov[i].iov_len); 231 #else 232 memcpy(s, iov[i].iov_base, iov[i].iov_len); 233 s += iov[i].iov_len; 234 #endif 235 } 236 } 237 238 void buffer_append_path_len(buffer * restrict b, const char * restrict a, size_t alen) { 239 char * restrict s = buffer_string_prepare_append(b, alen+1); 240 const int aslash = (alen && a[0] == '/'); 241 if (b->used > 1 && s[-1] == '/') { 242 if (aslash) { 243 ++a; 244 --alen; 245 } 246 } 247 else { 248 if (0 == b->used) b->used = 1; 249 if (!aslash) { 250 *s++ = '/'; 251 ++b->used; 252 } 253 } 254 b->used += alen; 255 s[alen] = '\0'; 256 memcpy(s, a, alen); 257 } 258 259 void 260 buffer_copy_path_len2 (buffer * const restrict b, const char * const restrict s1, size_t len1, const char * const restrict s2, size_t len2) 261 { 262 /*(similar to buffer_copy_string_len(b, s1, len1) but combined allocation)*/ 263 memcpy(buffer_string_prepare_copy(b, len1+len2+1), s1, len1); 264 b->used = len1 + 1; /*('\0' byte will be written below)*/ 265 266 buffer_append_path_len(b, s2, len2);/*(choice: not inlined, special-cased)*/ 267 } 268 269 void 270 buffer_copy_string_len_lc (buffer * const restrict b, const char * const restrict s, const size_t len) 271 { 272 char * const restrict d = buffer_string_prepare_copy(b, len); 273 b->used = len+1; 274 d[len] = '\0'; 275 for (size_t i = 0; i < len; ++i) 276 d[i] = (!light_isupper(s[i])) ? s[i] : s[i] | 0x20; 277 } 278 279 void buffer_append_uint_hex_lc(buffer *b, uintmax_t value) { 280 char *buf; 281 unsigned int shift = 0; 282 283 { 284 uintmax_t copy = value; 285 do { 286 copy >>= 8; 287 shift += 8; /* counting bits */ 288 } while (0 != copy); 289 } 290 291 buf = buffer_extend(b, shift >> 2); /*nibbles (4 bits)*/ 292 293 while (shift > 0) { 294 shift -= 4; 295 *(buf++) = hex_chars_lc[(value >> shift) & 0x0F]; 296 } 297 } 298 299 __attribute_nonnull__() 300 __attribute_returns_nonnull__ 301 static char* utostr(char buf[LI_ITOSTRING_LENGTH], uintmax_t val) { 302 char *cur = buf+LI_ITOSTRING_LENGTH; 303 uintmax_t x; 304 do { 305 *(--cur) = (char) ('0' + (int)(val - (x = val/10) * 10)); 306 } while (0 != (val = x)); /* val % 10 */ 307 return cur; 308 } 309 310 __attribute_nonnull__() 311 __attribute_returns_nonnull__ 312 static char* itostr(char buf[LI_ITOSTRING_LENGTH], intmax_t val) { 313 /* absolute value not defined for INTMAX_MIN, but can take absolute 314 * value of any negative number via twos complement cast to unsigned. 315 * negative sign is prepended after (now unsigned) value is converted 316 * to string */ 317 uintmax_t uval = val >= 0 ? (uintmax_t)val : ((uintmax_t)~val) + 1; 318 char *cur = utostr(buf, uval); 319 if (val < 0) *(--cur) = '-'; 320 321 return cur; 322 } 323 324 void buffer_append_int(buffer *b, intmax_t val) { 325 char buf[LI_ITOSTRING_LENGTH]; 326 const char * const str = itostr(buf, val); 327 buffer_append_string_len(b, str, buf+sizeof(buf) - str); 328 } 329 330 void buffer_append_strftime(buffer * const restrict b, const char * const restrict format, const struct tm * const restrict tm) { 331 /*(localtime_r() or gmtime_r() producing tm should not have failed)*/ 332 if (__builtin_expect( (NULL == tm), 0)) return; 333 334 /*(expecting typical format strings to result in < 64 bytes needed; 335 * skipping buffer_string_space() calculation and providing fixed size)*/ 336 size_t rv = strftime(buffer_string_prepare_append(b, 63), 64, format, tm); 337 338 /* 0 (in some apis) signals the string may have been too small; 339 * but the format could also just have lead to an empty string */ 340 if (__builtin_expect( (0 == rv), 0) || __builtin_expect( (rv > 63), 0)) { 341 /* unexpected; give it a second try with a larger string */ 342 rv = strftime(buffer_string_prepare_append(b, 4095), 4096, format, tm); 343 if (__builtin_expect( (rv > 4095), 0))/*(input format was ridiculous)*/ 344 return; 345 } 346 347 /*buffer_commit(b, rv);*/ 348 b->used += (uint32_t)rv + (0 == b->used); 349 } 350 351 352 size_t li_itostrn(char *buf, size_t buf_len, intmax_t val) { 353 char p_buf[LI_ITOSTRING_LENGTH]; 354 char* const str = itostr(p_buf, val); 355 size_t len = (size_t)(p_buf+sizeof(p_buf)-str); 356 force_assert(len <= buf_len); 357 memcpy(buf, str, len); 358 return len; 359 } 360 361 size_t li_utostrn(char *buf, size_t buf_len, uintmax_t val) { 362 char p_buf[LI_ITOSTRING_LENGTH]; 363 char* const str = utostr(p_buf, val); 364 size_t len = (size_t)(p_buf+sizeof(p_buf)-str); 365 force_assert(len <= buf_len); 366 memcpy(buf, str, len); 367 return len; 368 } 369 370 #define li_ntox_lc(n) ((n) <= 9 ? (n) + '0' : (n) + 'a' - 10) 371 372 /* c (char) and n (nibble) MUST be unsigned integer types */ 373 #define li_cton(c,n) \ 374 (((n) = (c) - '0') <= 9 || (((n) = ((c)&0xdf) - 'A') <= 5 ? ((n) += 10) : 0)) 375 376 /* converts hex char (0-9, A-Z, a-z) to decimal. 377 * returns 0xFF on invalid input. 378 */ 379 char hex2int(unsigned char hex) { 380 unsigned char n; 381 return li_cton(hex,n) ? (char)n : 0xFF; 382 } 383 384 int li_hex2bin (unsigned char * const bin, const size_t binlen, const char * const hexstr, const size_t len) 385 { 386 /* validate and transform 32-byte MD5 hex string to 16-byte binary MD5, 387 * or 64-byte SHA-256 or SHA-512-256 hex string to 32-byte binary digest */ 388 if (len > (binlen << 1)) return -1; 389 for (int i = 0, ilen = (int)len; i < ilen; i+=2) { 390 int hi = hexstr[i]; 391 int lo = hexstr[i+1]; 392 if ('0' <= hi && hi <= '9') hi -= '0'; 393 else if ((uint32_t)(hi |= 0x20)-'a' <= 'f'-'a')hi += -'a' + 10; 394 else return -1; 395 if ('0' <= lo && lo <= '9') lo -= '0'; 396 else if ((uint32_t)(lo |= 0x20)-'a' <= 'f'-'a')lo += -'a' + 10; 397 else return -1; 398 bin[(i >> 1)] = (unsigned char)((hi << 4) | lo); 399 } 400 return 0; 401 } 402 403 404 __attribute_noinline__ 405 int buffer_eq_icase_ssn(const char * const a, const char * const b, const size_t len) { 406 for (size_t i = 0; i < len; ++i) { 407 unsigned int ca = ((unsigned char *)a)[i]; 408 unsigned int cb = ((unsigned char *)b)[i]; 409 if (ca != cb) { 410 ca |= 0x20; 411 cb |= 0x20; 412 if (ca != cb) return 0; 413 if (!light_islower(ca)) return 0; 414 if (!light_islower(cb)) return 0; 415 } 416 } 417 return 1; 418 } 419 420 int buffer_eq_icase_ss(const char * const a, const size_t alen, const char * const b, const size_t blen) { 421 /* 1 = equal; 0 = not equal */ /* short string sizes expected (< INT_MAX) */ 422 return (alen == blen) ? buffer_eq_icase_ssn(a, b, blen) : 0; 423 } 424 425 int buffer_eq_icase_slen(const buffer * const b, const char * const s, const size_t slen) { 426 /* Note: b must be initialized, i.e. 0 != b->used; uninitialized is not eq*/ 427 /* 1 = equal; 0 = not equal */ /* short string sizes expected (< INT_MAX) */ 428 return (b->used == slen + 1) ? buffer_eq_icase_ssn(b->ptr, s, slen) : 0; 429 } 430 431 int buffer_eq_slen(const buffer * const b, const char * const s, const size_t slen) { 432 /* Note: b must be initialized, i.e. 0 != b->used; uninitialized is not eq*/ 433 /* 1 = equal; 0 = not equal */ /* short string sizes expected (< INT_MAX) */ 434 return (b->used == slen + 1 && 0 == memcmp(b->ptr, s, slen)); 435 } 436 437 438 /** 439 * check if two buffer contain the same data 440 */ 441 442 int buffer_is_equal(const buffer *a, const buffer *b) { 443 /* 1 = equal; 0 = not equal */ 444 return (a->used == b->used && 0 == memcmp(a->ptr, b->ptr, a->used)); 445 } 446 447 448 void li_tohex_lc(char * const restrict buf, size_t buf_len, const char * const restrict s, size_t s_len) { 449 force_assert(2 * s_len > s_len); 450 force_assert(2 * s_len < buf_len); 451 452 for (size_t i = 0; i < s_len; ++i) { 453 buf[2*i] = hex_chars_lc[(s[i] >> 4) & 0x0F]; 454 buf[2*i+1] = hex_chars_lc[s[i] & 0x0F]; 455 } 456 buf[2*s_len] = '\0'; 457 } 458 459 void li_tohex_uc(char * const restrict buf, size_t buf_len, const char * const restrict s, size_t s_len) { 460 force_assert(2 * s_len > s_len); 461 force_assert(2 * s_len < buf_len); 462 463 for (size_t i = 0; i < s_len; ++i) { 464 buf[2*i] = hex_chars_uc[(s[i] >> 4) & 0x0F]; 465 buf[2*i+1] = hex_chars_uc[s[i] & 0x0F]; 466 } 467 buf[2*s_len] = '\0'; 468 } 469 470 471 void buffer_substr_replace (buffer * const restrict b, const size_t offset, 472 const size_t len, const buffer * const restrict replace) 473 { 474 const size_t blen = buffer_clen(b); 475 const size_t rlen = buffer_clen(replace); 476 477 if (rlen > len) { 478 buffer_extend(b, blen-len+rlen); 479 memmove(b->ptr+offset+rlen, b->ptr+offset+len, blen-offset-len); 480 } 481 482 memcpy(b->ptr+offset, replace->ptr, rlen); 483 484 if (rlen < len) { 485 memmove(b->ptr+offset+rlen, b->ptr+offset+len, blen-offset-len); 486 buffer_truncate(b, blen-len+rlen); 487 } 488 } 489 490 491 void buffer_append_string_encoded_hex_lc(buffer * const restrict b, const char * const restrict s, size_t len) { 492 unsigned char * const p = (unsigned char *)buffer_extend(b, len*2); 493 for (size_t i = 0; i < len; ++i) { 494 p[(i<<1)] = hex_chars_lc[(s[i] >> 4) & 0x0F]; 495 p[(i<<1)+1] = hex_chars_lc[(s[i]) & 0x0F]; 496 } 497 } 498 499 void buffer_append_string_encoded_hex_uc(buffer * const restrict b, const char * const restrict s, size_t len) { 500 unsigned char * const p = (unsigned char *)buffer_extend(b, len*2); 501 for (size_t i = 0; i < len; ++i) { 502 p[(i<<1)] = hex_chars_uc[(s[i] >> 4) & 0x0F]; 503 p[(i<<1)+1] = hex_chars_uc[(s[i]) & 0x0F]; 504 } 505 } 506 507 508 /* everything except: ! ( ) * - . 0-9 A-Z _ a-z */ 509 static const char encoded_chars_rel_uri_part[] = { 510 /* 511 0 1 2 3 4 5 6 7 8 9 A B C D E F 512 */ 513 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 00 - 0F control chars */ 514 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 10 - 1F */ 515 1, 0, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 0, 0, 1, /* 20 - 2F space " # $ % & ' + , / */ 516 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, /* 30 - 3F : ; < = > ? */ 517 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 40 - 4F @ */ 518 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0, /* 50 - 5F [ \ ] ^ */ 519 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 60 - 6F ` */ 520 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 1, /* 70 - 7F { | } DEL */ 521 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 80 - 8F */ 522 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 90 - 9F */ 523 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* A0 - AF */ 524 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* B0 - BF */ 525 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* C0 - CF */ 526 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* D0 - DF */ 527 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* E0 - EF */ 528 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* F0 - FF */ 529 }; 530 531 /* everything except: ! ( ) * - . / 0-9 A-Z _ a-z */ 532 static const char encoded_chars_rel_uri[] = { 533 /* 534 0 1 2 3 4 5 6 7 8 9 A B C D E F 535 */ 536 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 00 - 0F control chars */ 537 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 10 - 1F */ 538 1, 0, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 0, 0, 0, /* 20 - 2F space " # $ % & ' + , */ 539 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, /* 30 - 3F : ; < = > ? */ 540 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 40 - 4F @ */ 541 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0, /* 50 - 5F [ \ ] ^ */ 542 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 60 - 6F ` */ 543 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 1, /* 70 - 7F { | } DEL */ 544 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 80 - 8F */ 545 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 90 - 9F */ 546 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* A0 - AF */ 547 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* B0 - BF */ 548 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* C0 - CF */ 549 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* D0 - DF */ 550 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* E0 - EF */ 551 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* F0 - FF */ 552 }; 553 554 static const char encoded_chars_html[] = { 555 /* 556 0 1 2 3 4 5 6 7 8 9 A B C D E F 557 */ 558 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 00 - 0F control chars */ 559 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 10 - 1F */ 560 0, 0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, /* 20 - 2F " & ' */ 561 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, /* 30 - 3F < > */ 562 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 40 - 4F */ 563 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 50 - 5F */ 564 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 60 - 6F ` */ 565 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, /* 70 - 7F DEL */ 566 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 80 - 8F */ 567 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 90 - 9F */ 568 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* A0 - AF */ 569 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* B0 - BF */ 570 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* C0 - CF */ 571 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* D0 - DF */ 572 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* E0 - EF */ 573 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* F0 - FF */ 574 }; 575 576 static const char encoded_chars_minimal_xml[] = { 577 /* 578 0 1 2 3 4 5 6 7 8 9 A B C D E F 579 */ 580 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 00 - 0F control chars */ 581 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 10 - 1F */ 582 0, 0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, /* 20 - 2F " & ' */ 583 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, /* 30 - 3F < > */ 584 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 40 - 4F */ 585 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 50 - 5F */ 586 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 60 - 6F ` */ 587 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, /* 70 - 7F DEL */ 588 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 80 - 8F */ 589 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 90 - 9F */ 590 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* A0 - AF */ 591 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* B0 - BF */ 592 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* C0 - CF */ 593 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* D0 - DF */ 594 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* E0 - EF */ 595 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* F0 - FF */ 596 }; 597 598 599 600 void buffer_append_string_encoded(buffer * const restrict b, const char * const restrict s, size_t s_len, buffer_encoding_t encoding) { 601 unsigned char *ds, *d; 602 size_t d_len, ndx; 603 const char *map = NULL; 604 605 switch(encoding) { 606 case ENCODING_REL_URI: 607 map = encoded_chars_rel_uri; 608 break; 609 case ENCODING_REL_URI_PART: 610 map = encoded_chars_rel_uri_part; 611 break; 612 case ENCODING_HTML: 613 map = encoded_chars_html; 614 break; 615 case ENCODING_MINIMAL_XML: 616 map = encoded_chars_minimal_xml; 617 break; 618 } 619 620 /* count to-be-encoded-characters */ 621 for (ds = (unsigned char *)s, d_len = 0, ndx = 0; ndx < s_len; ds++, ndx++) { 622 if (map[*ds & 0xFF]) { 623 switch(encoding) { 624 case ENCODING_REL_URI: 625 case ENCODING_REL_URI_PART: 626 d_len += 3; 627 break; 628 case ENCODING_HTML: 629 case ENCODING_MINIMAL_XML: 630 d_len += 6; 631 break; 632 } 633 } else { 634 d_len++; 635 } 636 } 637 638 d = (unsigned char*) buffer_extend(b, d_len); 639 640 if (d_len == s_len) { /*(short-circuit; nothing to encoded)*/ 641 memcpy(d, s, s_len); 642 return; 643 } 644 645 for (ds = (unsigned char *)s, d_len = 0, ndx = 0; ndx < s_len; ds++, ndx++) { 646 if (map[*ds & 0xFF]) { 647 switch(encoding) { 648 case ENCODING_REL_URI: 649 case ENCODING_REL_URI_PART: 650 d[d_len++] = '%'; 651 d[d_len++] = hex_chars_uc[((*ds) >> 4) & 0x0F]; 652 d[d_len++] = hex_chars_uc[(*ds) & 0x0F]; 653 break; 654 case ENCODING_HTML: 655 case ENCODING_MINIMAL_XML: 656 d[d_len++] = '&'; 657 d[d_len++] = '#'; 658 d[d_len++] = 'x'; 659 d[d_len++] = hex_chars_uc[((*ds) >> 4) & 0x0F]; 660 d[d_len++] = hex_chars_uc[(*ds) & 0x0F]; 661 d[d_len++] = ';'; 662 break; 663 } 664 } else { 665 d[d_len++] = *ds; 666 } 667 } 668 } 669 670 void buffer_append_string_encoded_json(buffer * const restrict b, const char * const restrict s, const size_t len) { 671 const unsigned char * const restrict ds = (unsigned char *)s; 672 size_t dlen = 0; 673 674 /* calculate space needed for string including encodings */ 675 for (size_t i = 0; i < len; ++i) { 676 int c = ds[i]; 677 if (c == '"' || c == '\\' || c < 0x20 || c == 0x7f) { 678 switch (c) { 679 case '\b': 680 case '\t': 681 case '\n': 682 case '\f': 683 case '\r': 684 case '"': 685 case '\\': 686 dlen += 2; 687 break; 688 default: 689 dlen += 6; /* \uCCCC */ 690 break; 691 } 692 } 693 else { 694 ++dlen; 695 } 696 } 697 698 unsigned char * const d = (unsigned char *)buffer_extend(b, dlen); 699 700 if (__builtin_expect( (dlen == len), 1)) {/*(short-circuit; nothing to encode)*/ 701 memcpy(d, ds, len); 702 return; 703 } 704 705 dlen = 0; 706 for (size_t i = 0; i < len; ++i) { 707 int c = ds[i]; 708 if (c == '"' || c == '\\' || c < 0x20 || c == 0x7f) { 709 d[dlen++] = '\\'; 710 switch (c) { 711 case '\b': 712 d[dlen++] = 'b'; 713 break; 714 case '\t': 715 d[dlen++] = 't'; 716 break; 717 case '\n': 718 d[dlen++] = 'n'; 719 break; 720 case '\f': 721 d[dlen++] = 'f'; 722 break; 723 case '\r': 724 d[dlen++] = 'r'; 725 break; 726 case '"': 727 d[dlen++] = '"'; 728 break; 729 case '\\': 730 d[dlen++] = '\\'; 731 break; 732 default: 733 d[dlen ] = 'u'; 734 d[dlen+1] = '0'; 735 d[dlen+2] = '0'; 736 d[dlen+3] = hex_chars_lc[(c >> 4) & 0x0F]; 737 d[dlen+4] = hex_chars_lc[c & 0x0F]; 738 dlen += 5; 739 break; 740 } 741 } 742 else { 743 d[dlen++] = c; 744 } 745 } 746 } 747 748 749 void buffer_append_string_c_escaped(buffer * const restrict b, const char * const restrict s, size_t s_len) { 750 unsigned char *ds, *d; 751 size_t d_len, ndx; 752 753 /* count to-be-encoded-characters */ 754 for (ds = (unsigned char *)s, d_len = 0, ndx = 0; ndx < s_len; ds++, ndx++) { 755 if ((*ds < 0x20) /* control character */ 756 || (*ds >= 0x7f)) { /* DEL + non-ASCII characters */ 757 switch (*ds) { 758 case '\t': 759 case '\r': 760 case '\n': 761 d_len += 2; 762 break; 763 default: 764 d_len += 4; /* \xCC */ 765 break; 766 } 767 } else { 768 d_len++; 769 } 770 } 771 772 d = (unsigned char*) buffer_extend(b, d_len); 773 774 if (d_len == s_len) { /*(short-circuit; nothing to encoded)*/ 775 memcpy(d, s, s_len); 776 return; 777 } 778 779 for (ds = (unsigned char *)s, d_len = 0, ndx = 0; ndx < s_len; ds++, ndx++) { 780 if ((*ds < 0x20) /* control character */ 781 || (*ds >= 0x7f)) { /* DEL + non-ASCII characters */ 782 d[d_len++] = '\\'; 783 switch (*ds) { 784 case '\t': 785 d[d_len++] = 't'; 786 break; 787 case '\r': 788 d[d_len++] = 'r'; 789 break; 790 case '\n': 791 d[d_len++] = 'n'; 792 break; 793 default: 794 d[d_len++] = 'x'; 795 d[d_len++] = hex_chars_lc[((*ds) >> 4) & 0x0F]; 796 d[d_len++] = hex_chars_lc[(*ds) & 0x0F]; 797 break; 798 } 799 } else { 800 d[d_len++] = *ds; 801 } 802 } 803 } 804 805 806 /* decodes url-special-chars inplace. 807 * replaces non-printable characters with '_' 808 * (If this is used on a portion of query string, then query string should be 809 * split on '&', and '+' replaced with ' ' before calling this routine) 810 */ 811 812 void buffer_urldecode_path(buffer * const b) { 813 const size_t len = buffer_clen(b); 814 char *src = len ? memchr(b->ptr, '%', len) : NULL; 815 if (NULL == src) return; 816 817 char *dst = src; 818 do { 819 /* *src == '%' */ 820 unsigned char high = ((unsigned char *)src)[1]; 821 unsigned char low = high ? hex2int(((unsigned char *)src)[2]) : 0xFF; 822 if (0xFF != (high = hex2int(high)) && 0xFF != low) { 823 high = (high << 4) | low; /* map ctrls to '_' */ 824 *dst = (high >= 32 && high != 127) ? high : '_'; 825 src += 2; 826 } /* else ignore this '%'; leave as-is and move on */ 827 828 while ((*++dst = *++src) != '%' && *src) ; 829 } while (*src); 830 b->used = (dst - b->ptr) + 1; 831 } 832 833 int buffer_is_valid_UTF8(const buffer *b) { 834 /* https://www.w3.org/International/questions/qa-forms-utf-8 */ 835 /*assert(b->used);*//*(b->ptr must exist and be '\0'-terminated)*/ 836 const unsigned char *c = (unsigned char *)b->ptr; 837 while (*c) { 838 839 /*(note: includes ctrls)*/ 840 if ( c[0] < 0x80 ) { ++c; continue; } 841 842 if ( 0xc2 <= c[0] && c[0] <= 0xdf 843 && 0x80 <= c[1] && c[1] <= 0xbf ) { c+=2; continue; } 844 845 if ( ( ( 0xe0 == c[0] 846 && 0xa0 <= c[1] && c[1] <= 0xbf) 847 || ( 0xe1 <= c[0] && c[0] <= 0xef && c[0] != 0xed 848 && 0x80 <= c[1] && c[1] <= 0xbf) 849 || ( 0xed == c[0] 850 && 0x80 <= c[1] && c[1] <= 0x9f) ) 851 && 0x80 <= c[2] && c[2] <= 0xbf ) { c+=3; continue; } 852 853 if ( ( ( 0xf0 == c[0] 854 && 0x90 <= c[1] && c[1] <= 0xbf) 855 || ( 0xf1 <= c[0] && c[0] <= 0xf3 856 && 0x80 <= c[1] && c[1] <= 0xbf) 857 || ( 0xf4 == c[0] 858 && 0x80 <= c[1] && c[1] <= 0x8f) ) 859 && 0x80 <= c[2] && c[2] <= 0xbf 860 && 0x80 <= c[3] && c[3] <= 0xbf ) { c+=4; continue; } 861 862 return 0; /* invalid */ 863 } 864 return 1; /* valid */ 865 } 866 867 /* - special case: empty string returns empty string 868 * - on windows or cygwin: replace \ with / 869 * - strip leading spaces 870 * - prepends "/" if not present already 871 * - resolve "/../", "//" and "/./" the usual way: 872 * the first one removes a preceding component, the other two 873 * get compressed to "/". 874 * - "/." and "/.." at the end are similar, but always leave a trailing 875 * "/" 876 * 877 * /blah/.. gets / 878 * /blah/../foo gets /foo 879 * /abc/./xyz gets /abc/xyz 880 * /abc//xyz gets /abc/xyz 881 */ 882 883 void buffer_path_simplify(buffer *b) 884 { 885 char *out = b->ptr; 886 char * const end = b->ptr + b->used - 1; 887 888 if (__builtin_expect( (buffer_is_blank(b)), 0)) { 889 buffer_blank(b); 890 return; 891 } 892 893 #if defined(__WIN32) || defined(__CYGWIN__) 894 /* cygwin is treating \ and / the same, so we have to that too */ 895 for (char *p = b->ptr; *p; p++) { 896 if (*p == '\\') *p = '/'; 897 } 898 #endif 899 900 *end = '/'; /*(end of path modified to avoid need to check '\0')*/ 901 902 char *walk = out; 903 if (__builtin_expect( (*walk == '/'), 1)) { 904 /* scan to detect (potential) need for path simplification 905 * (repeated '/' or "/.") */ 906 do { 907 if (*++walk == '.' || *walk == '/') 908 break; 909 do { ++walk; } while (*walk != '/'); 910 } while (walk != end); 911 if (__builtin_expect( (walk == end), 1)) { 912 /* common case: no repeated '/' or "/." */ 913 *end = '\0'; /* overwrite extra '/' added to end of path */ 914 return; 915 } 916 out = walk-1; 917 } 918 else { 919 if (walk[0] == '.' && walk[1] == '/') 920 *out = *++walk; 921 else if (walk[0] == '.' && walk[1] == '.' && walk[2] == '/') 922 *out = *(walk += 2); 923 else { 924 while (*++walk != '/') ; 925 out = walk; 926 } 927 ++walk; 928 } 929 930 while (walk <= end) { 931 /* previous char is '/' at this point (or start of string w/o '/') */ 932 if (__builtin_expect( (walk[0] == '/'), 0)) { 933 /* skip repeated '/' (e.g. "///" -> "/") */ 934 if (++walk < end) 935 continue; 936 else { 937 ++out; 938 break; 939 } 940 } 941 else if (__builtin_expect( (walk[0] == '.'), 0)) { 942 /* handle "./" and "../" */ 943 if (walk[1] == '.' && walk[2] == '/') { 944 /* handle "../" */ 945 while (out > b->ptr && *--out != '/') ; 946 *out = '/'; /*(in case path had not started with '/')*/ 947 if ((walk += 3) >= end) { 948 ++out; 949 break; 950 } 951 else 952 continue; 953 } 954 else if (walk[1] == '/') { 955 /* handle "./" */ 956 if ((walk += 2) >= end) { 957 ++out; 958 break; 959 } 960 continue; 961 } 962 else { 963 /* accept "." if not part of "../" or "./" */ 964 *++out = '.'; 965 ++walk; 966 } 967 } 968 969 while ((*++out = *walk++) != '/') ; 970 } 971 *out = *end = '\0'; /* overwrite extra '/' added to end of path */ 972 b->used = (out - b->ptr) + 1; 973 /*buffer_truncate(b, out - b->ptr);*/ 974 } 975 976 void buffer_to_lower(buffer * const b) { 977 unsigned char * const restrict s = (unsigned char *)b->ptr; 978 const uint_fast32_t used = b->used; 979 for (uint_fast32_t i = 0; i < used; ++i) { 980 if (light_isupper(s[i])) s[i] |= 0x20; 981 } 982 } 983 984 985 void buffer_to_upper(buffer * const b) { 986 unsigned char * const restrict s = (unsigned char *)b->ptr; 987 const uint_fast32_t used = b->used; 988 for (uint_fast32_t i = 0; i < used; ++i) { 989 if (light_islower(s[i])) s[i] &= 0xdf; 990 } 991 } 992