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 __attribute_cold__ /*(reduce code size due to inlining)*/ 170 void buffer_copy_string(buffer * restrict b, const char * restrict s) { 171 if (__builtin_expect( (NULL == s), 0)) s = ""; 172 buffer_copy_string_len(b, s, strlen(s)); 173 } 174 175 void buffer_copy_string_len(buffer * const restrict b, const char * const restrict s, const size_t len) { 176 b->used = len + 1; 177 char * const restrict d = (len < b->size) 178 ? b->ptr 179 : buffer_alloc_replace(b, len); 180 d[len] = '\0'; 181 memcpy(d, s, len); 182 } 183 184 __attribute_cold__ /*(reduce code size due to inlining)*/ 185 void buffer_append_string(buffer * restrict b, const char * restrict s) { 186 if (__builtin_expect( (NULL == s), 0)) s = ""; 187 buffer_append_string_len(b, s, strlen(s)); 188 } 189 190 /** 191 * append a string to the end of the buffer 192 * 193 * the resulting buffer is terminated with a '\0' 194 * s is treated as a un-terminated string (a \0 is handled a normal character) 195 * 196 * @param b a buffer 197 * @param s the string 198 * @param s_len size of the string (without the terminating \0) 199 */ 200 201 void buffer_append_string_len(buffer * const restrict b, const char * const restrict s, const size_t len) { 202 memcpy(buffer_extend(b, len), s, len); 203 } 204 205 void buffer_append_str2(buffer * const restrict b, const char * const s1, const size_t len1, const char * const s2, const size_t len2) { 206 char * const restrict s = buffer_extend(b, len1+len2); 207 #ifdef HAVE_MEMPCPY 208 mempcpy(mempcpy(s, s1, len1), s2, len2); 209 #else 210 memcpy(s, s1, len1); 211 memcpy(s+len1, s2, len2); 212 #endif 213 } 214 215 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) { 216 char * restrict s = buffer_extend(b, len1+len2+len3); 217 #ifdef HAVE_MEMPCPY 218 mempcpy(mempcpy(mempcpy(s, s1, len1), s2, len2), s3, len3); 219 #else 220 memcpy(s, s1, len1); 221 memcpy((s+=len1), s2, len2); 222 memcpy((s+=len2), s3, len3); 223 #endif 224 } 225 226 void buffer_append_iovec(buffer * const restrict b, const struct const_iovec * const iov, const size_t n) { 227 size_t len = 0; 228 for (size_t i = 0; i < n; ++i) 229 len += iov[i].iov_len; 230 char *s = buffer_extend(b, len); 231 for (size_t i = 0; i < n; ++i) { 232 if (0 == iov[i].iov_len) continue; 233 #ifdef HAVE_MEMPCPY 234 s = mempcpy(s, iov[i].iov_base, iov[i].iov_len); 235 #else 236 memcpy(s, iov[i].iov_base, iov[i].iov_len); 237 s += iov[i].iov_len; 238 #endif 239 } 240 } 241 242 void buffer_append_path_len(buffer * restrict b, const char * restrict a, size_t alen) { 243 char * restrict s = buffer_string_prepare_append(b, alen+1); 244 const int aslash = (alen && a[0] == '/'); 245 if (b->used > 1 && s[-1] == '/') { 246 if (aslash) { 247 ++a; 248 --alen; 249 } 250 } 251 else { 252 if (0 == b->used) b->used = 1; 253 if (!aslash) { 254 *s++ = '/'; 255 ++b->used; 256 } 257 } 258 b->used += alen; 259 s[alen] = '\0'; 260 memcpy(s, a, alen); 261 } 262 263 void 264 buffer_copy_path_len2 (buffer * const restrict b, const char * const restrict s1, size_t len1, const char * const restrict s2, size_t len2) 265 { 266 /*(similar to buffer_copy_string_len(b, s1, len1) but combined allocation)*/ 267 memcpy(buffer_string_prepare_copy(b, len1+len2+1), s1, len1); 268 b->used = len1 + 1; /*('\0' byte will be written below)*/ 269 270 buffer_append_path_len(b, s2, len2);/*(choice: not inlined, special-cased)*/ 271 } 272 273 void 274 buffer_copy_string_len_lc (buffer * const restrict b, const char * const restrict s, const size_t len) 275 { 276 char * const restrict d = buffer_string_prepare_copy(b, len); 277 b->used = len+1; 278 d[len] = '\0'; 279 for (size_t i = 0; i < len; ++i) 280 d[i] = (!light_isupper(s[i])) ? s[i] : s[i] | 0x20; 281 } 282 283 void buffer_append_uint_hex_lc(buffer *b, uintmax_t value) { 284 char *buf; 285 unsigned int shift = 0; 286 287 { 288 uintmax_t copy = value; 289 do { 290 copy >>= 8; 291 shift += 8; /* counting bits */ 292 } while (0 != copy); 293 } 294 295 buf = buffer_extend(b, shift >> 2); /*nibbles (4 bits)*/ 296 297 while (shift > 0) { 298 shift -= 4; 299 *(buf++) = hex_chars_lc[(value >> shift) & 0x0F]; 300 } 301 } 302 303 __attribute_nonnull__() 304 __attribute_returns_nonnull__ 305 static char* utostr(char buf[LI_ITOSTRING_LENGTH], uintmax_t val) { 306 char *cur = buf+LI_ITOSTRING_LENGTH; 307 uintmax_t x; 308 do { 309 *(--cur) = (char) ('0' + (int)(val - (x = val/10) * 10)); 310 } while (0 != (val = x)); /* val % 10 */ 311 return cur; 312 } 313 314 __attribute_nonnull__() 315 __attribute_returns_nonnull__ 316 static char* itostr(char buf[LI_ITOSTRING_LENGTH], intmax_t val) { 317 /* absolute value not defined for INTMAX_MIN, but can take absolute 318 * value of any negative number via twos complement cast to unsigned. 319 * negative sign is prepended after (now unsigned) value is converted 320 * to string */ 321 uintmax_t uval = val >= 0 ? (uintmax_t)val : ((uintmax_t)~val) + 1; 322 char *cur = utostr(buf, uval); 323 if (val < 0) *(--cur) = '-'; 324 325 return cur; 326 } 327 328 void buffer_append_int(buffer *b, intmax_t val) { 329 char buf[LI_ITOSTRING_LENGTH]; 330 const char * const str = itostr(buf, val); 331 buffer_append_string_len(b, str, buf+sizeof(buf) - str); 332 } 333 334 void buffer_append_strftime(buffer * const restrict b, const char * const restrict format, const struct tm * const restrict tm) { 335 /*(localtime_r() or gmtime_r() producing tm should not have failed)*/ 336 if (__builtin_expect( (NULL == tm), 0)) return; 337 338 /*(expecting typical format strings to result in < 64 bytes needed; 339 * skipping buffer_string_space() calculation and providing fixed size)*/ 340 size_t rv = strftime(buffer_string_prepare_append(b, 63), 64, format, tm); 341 342 /* 0 (in some apis) signals the string may have been too small; 343 * but the format could also just have lead to an empty string */ 344 if (__builtin_expect( (0 == rv), 0) || __builtin_expect( (rv > 63), 0)) { 345 /* unexpected; give it a second try with a larger string */ 346 rv = strftime(buffer_string_prepare_append(b, 4095), 4096, format, tm); 347 if (__builtin_expect( (rv > 4095), 0))/*(input format was ridiculous)*/ 348 return; 349 } 350 351 /*buffer_commit(b, rv);*/ 352 b->used += (uint32_t)rv + (0 == b->used); 353 } 354 355 356 size_t li_itostrn(char *buf, size_t buf_len, intmax_t val) { 357 char p_buf[LI_ITOSTRING_LENGTH]; 358 char* const str = itostr(p_buf, val); 359 size_t len = (size_t)(p_buf+sizeof(p_buf)-str); 360 force_assert(len <= buf_len); 361 memcpy(buf, str, len); 362 return len; 363 } 364 365 size_t li_utostrn(char *buf, size_t buf_len, uintmax_t val) { 366 char p_buf[LI_ITOSTRING_LENGTH]; 367 char* const str = utostr(p_buf, val); 368 size_t len = (size_t)(p_buf+sizeof(p_buf)-str); 369 force_assert(len <= buf_len); 370 memcpy(buf, str, len); 371 return len; 372 } 373 374 #define li_ntox_lc(n) ((n) <= 9 ? (n) + '0' : (n) + 'a' - 10) 375 376 /* c (char) and n (nibble) MUST be unsigned integer types */ 377 #define li_cton(c,n) \ 378 (((n) = (c) - '0') <= 9 || (((n) = ((c)&0xdf) - 'A') <= 5 ? ((n) += 10) : 0)) 379 380 /* converts hex char (0-9, A-Z, a-z) to decimal. 381 * returns 0xFF on invalid input. 382 */ 383 char hex2int(unsigned char hex) { 384 unsigned char n; 385 return li_cton(hex,n) ? (char)n : 0xFF; 386 } 387 388 int li_hex2bin (unsigned char * const bin, const size_t binlen, const char * const hexstr, const size_t len) 389 { 390 /* validate and transform 32-byte MD5 hex string to 16-byte binary MD5, 391 * or 64-byte SHA-256 or SHA-512-256 hex string to 32-byte binary digest */ 392 if (len > (binlen << 1)) return -1; 393 for (int i = 0, ilen = (int)len; i < ilen; i+=2) { 394 int hi = hexstr[i]; 395 int lo = hexstr[i+1]; 396 if ('0' <= hi && hi <= '9') hi -= '0'; 397 else if ((uint32_t)(hi |= 0x20)-'a' <= 'f'-'a')hi += -'a' + 10; 398 else return -1; 399 if ('0' <= lo && lo <= '9') lo -= '0'; 400 else if ((uint32_t)(lo |= 0x20)-'a' <= 'f'-'a')lo += -'a' + 10; 401 else return -1; 402 bin[(i >> 1)] = (unsigned char)((hi << 4) | lo); 403 } 404 return 0; 405 } 406 407 408 __attribute_noinline__ 409 int buffer_eq_icase_ssn(const char * const a, const char * const b, const size_t len) { 410 for (size_t i = 0; i < len; ++i) { 411 unsigned int ca = ((unsigned char *)a)[i]; 412 unsigned int cb = ((unsigned char *)b)[i]; 413 if (ca != cb) { 414 ca |= 0x20; 415 cb |= 0x20; 416 if (ca != cb) return 0; 417 if (!light_islower(ca)) return 0; 418 if (!light_islower(cb)) return 0; 419 } 420 } 421 return 1; 422 } 423 424 int buffer_eq_icase_ss(const char * const a, const size_t alen, const char * const b, const size_t blen) { 425 /* 1 = equal; 0 = not equal */ /* short string sizes expected (< INT_MAX) */ 426 return (alen == blen) ? buffer_eq_icase_ssn(a, b, blen) : 0; 427 } 428 429 int buffer_eq_icase_slen(const buffer * const b, const char * const s, const size_t slen) { 430 /* Note: b must be initialized, i.e. 0 != b->used; uninitialized is not eq*/ 431 /* 1 = equal; 0 = not equal */ /* short string sizes expected (< INT_MAX) */ 432 return (b->used == slen + 1) ? buffer_eq_icase_ssn(b->ptr, s, slen) : 0; 433 } 434 435 int buffer_eq_slen(const buffer * const b, const char * const s, const size_t slen) { 436 /* Note: b must be initialized, i.e. 0 != b->used; uninitialized is not eq*/ 437 /* 1 = equal; 0 = not equal */ /* short string sizes expected (< INT_MAX) */ 438 return (b->used == slen + 1 && 0 == memcmp(b->ptr, s, slen)); 439 } 440 441 442 /** 443 * check if two buffer contain the same data 444 */ 445 446 int buffer_is_equal(const buffer *a, const buffer *b) { 447 /* 1 = equal; 0 = not equal */ 448 return (a->used == b->used && 0 == memcmp(a->ptr, b->ptr, a->used)); 449 } 450 451 452 void li_tohex_lc(char * const restrict buf, size_t buf_len, const char * const restrict s, size_t s_len) { 453 force_assert(2 * s_len > s_len); 454 force_assert(2 * s_len < buf_len); 455 456 for (size_t i = 0; i < s_len; ++i) { 457 buf[2*i] = hex_chars_lc[(s[i] >> 4) & 0x0F]; 458 buf[2*i+1] = hex_chars_lc[s[i] & 0x0F]; 459 } 460 buf[2*s_len] = '\0'; 461 } 462 463 void li_tohex_uc(char * const restrict buf, size_t buf_len, const char * const restrict s, size_t s_len) { 464 force_assert(2 * s_len > s_len); 465 force_assert(2 * s_len < buf_len); 466 467 for (size_t i = 0; i < s_len; ++i) { 468 buf[2*i] = hex_chars_uc[(s[i] >> 4) & 0x0F]; 469 buf[2*i+1] = hex_chars_uc[s[i] & 0x0F]; 470 } 471 buf[2*s_len] = '\0'; 472 } 473 474 475 void buffer_substr_replace (buffer * const restrict b, const size_t offset, 476 const size_t len, const buffer * const restrict replace) 477 { 478 const size_t blen = buffer_clen(b); 479 const size_t rlen = buffer_clen(replace); 480 481 if (rlen > len) { 482 buffer_extend(b, blen-len+rlen); 483 memmove(b->ptr+offset+rlen, b->ptr+offset+len, blen-offset-len); 484 } 485 486 memcpy(b->ptr+offset, replace->ptr, rlen); 487 488 if (rlen < len) { 489 memmove(b->ptr+offset+rlen, b->ptr+offset+len, blen-offset-len); 490 buffer_truncate(b, blen-len+rlen); 491 } 492 } 493 494 495 void buffer_append_string_encoded_hex_lc(buffer * const restrict b, const char * const restrict s, size_t len) { 496 unsigned char * const p = (unsigned char *)buffer_extend(b, len*2); 497 for (size_t i = 0; i < len; ++i) { 498 p[(i<<1)] = hex_chars_lc[(s[i] >> 4) & 0x0F]; 499 p[(i<<1)+1] = hex_chars_lc[(s[i]) & 0x0F]; 500 } 501 } 502 503 void buffer_append_string_encoded_hex_uc(buffer * const restrict b, const char * const restrict s, size_t len) { 504 unsigned char * const p = (unsigned char *)buffer_extend(b, len*2); 505 for (size_t i = 0; i < len; ++i) { 506 p[(i<<1)] = hex_chars_uc[(s[i] >> 4) & 0x0F]; 507 p[(i<<1)+1] = hex_chars_uc[(s[i]) & 0x0F]; 508 } 509 } 510 511 512 /* everything except: ! ( ) * - . 0-9 A-Z _ a-z */ 513 static const char encoded_chars_rel_uri_part[] = { 514 /* 515 0 1 2 3 4 5 6 7 8 9 A B C D E F 516 */ 517 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 00 - 0F control chars */ 518 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 10 - 1F */ 519 1, 0, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 0, 0, 1, /* 20 - 2F space " # $ % & ' + , / */ 520 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, /* 30 - 3F : ; < = > ? */ 521 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 40 - 4F @ */ 522 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0, /* 50 - 5F [ \ ] ^ */ 523 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 60 - 6F ` */ 524 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 1, /* 70 - 7F { | } DEL */ 525 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 80 - 8F */ 526 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 90 - 9F */ 527 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* A0 - AF */ 528 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* B0 - BF */ 529 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* C0 - CF */ 530 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* D0 - DF */ 531 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* E0 - EF */ 532 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* F0 - FF */ 533 }; 534 535 /* everything except: ! ( ) * - . / 0-9 A-Z _ a-z */ 536 static const char encoded_chars_rel_uri[] = { 537 /* 538 0 1 2 3 4 5 6 7 8 9 A B C D E F 539 */ 540 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 00 - 0F control chars */ 541 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 10 - 1F */ 542 1, 0, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 0, 0, 0, /* 20 - 2F space " # $ % & ' + , */ 543 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, /* 30 - 3F : ; < = > ? */ 544 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 40 - 4F @ */ 545 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0, /* 50 - 5F [ \ ] ^ */ 546 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 60 - 6F ` */ 547 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 1, /* 70 - 7F { | } DEL */ 548 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 80 - 8F */ 549 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 90 - 9F */ 550 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* A0 - AF */ 551 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* B0 - BF */ 552 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* C0 - CF */ 553 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* D0 - DF */ 554 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* E0 - EF */ 555 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* F0 - FF */ 556 }; 557 558 static const char encoded_chars_html[] = { 559 /* 560 0 1 2 3 4 5 6 7 8 9 A B C D E F 561 */ 562 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 00 - 0F control chars */ 563 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 10 - 1F */ 564 0, 0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, /* 20 - 2F " & ' */ 565 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, /* 30 - 3F < > */ 566 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 40 - 4F */ 567 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 50 - 5F */ 568 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 60 - 6F ` */ 569 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, /* 70 - 7F DEL */ 570 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 80 - 8F */ 571 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 90 - 9F */ 572 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* A0 - AF */ 573 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* B0 - BF */ 574 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* C0 - CF */ 575 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* D0 - DF */ 576 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* E0 - EF */ 577 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* F0 - FF */ 578 }; 579 580 static const char encoded_chars_minimal_xml[] = { 581 /* 582 0 1 2 3 4 5 6 7 8 9 A B C D E F 583 */ 584 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 00 - 0F control chars */ 585 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 10 - 1F */ 586 0, 0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, /* 20 - 2F " & ' */ 587 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, /* 30 - 3F < > */ 588 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 40 - 4F */ 589 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 50 - 5F */ 590 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 60 - 6F ` */ 591 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, /* 70 - 7F DEL */ 592 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 80 - 8F */ 593 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 90 - 9F */ 594 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* A0 - AF */ 595 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* B0 - BF */ 596 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* C0 - CF */ 597 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* D0 - DF */ 598 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* E0 - EF */ 599 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* F0 - FF */ 600 }; 601 602 603 604 void buffer_append_string_encoded(buffer * const restrict b, const char * const restrict s, size_t s_len, buffer_encoding_t encoding) { 605 unsigned char *ds, *d; 606 size_t d_len, ndx; 607 const char *map = NULL; 608 609 switch(encoding) { 610 case ENCODING_REL_URI: 611 map = encoded_chars_rel_uri; 612 break; 613 case ENCODING_REL_URI_PART: 614 map = encoded_chars_rel_uri_part; 615 break; 616 case ENCODING_HTML: 617 map = encoded_chars_html; 618 break; 619 case ENCODING_MINIMAL_XML: 620 map = encoded_chars_minimal_xml; 621 break; 622 } 623 624 /* count to-be-encoded-characters */ 625 for (ds = (unsigned char *)s, d_len = 0, ndx = 0; ndx < s_len; ds++, ndx++) { 626 if (map[*ds & 0xFF]) { 627 switch(encoding) { 628 case ENCODING_REL_URI: 629 case ENCODING_REL_URI_PART: 630 d_len += 3; 631 break; 632 case ENCODING_HTML: 633 case ENCODING_MINIMAL_XML: 634 d_len += 6; 635 break; 636 } 637 } else { 638 d_len++; 639 } 640 } 641 642 d = (unsigned char*) buffer_extend(b, d_len); 643 644 if (d_len == s_len) { /*(short-circuit; nothing to encoded)*/ 645 memcpy(d, s, s_len); 646 return; 647 } 648 649 for (ds = (unsigned char *)s, d_len = 0, ndx = 0; ndx < s_len; ds++, ndx++) { 650 if (map[*ds & 0xFF]) { 651 switch(encoding) { 652 case ENCODING_REL_URI: 653 case ENCODING_REL_URI_PART: 654 d[d_len++] = '%'; 655 d[d_len++] = hex_chars_uc[((*ds) >> 4) & 0x0F]; 656 d[d_len++] = hex_chars_uc[(*ds) & 0x0F]; 657 break; 658 case ENCODING_HTML: 659 case ENCODING_MINIMAL_XML: 660 d[d_len++] = '&'; 661 d[d_len++] = '#'; 662 d[d_len++] = 'x'; 663 d[d_len++] = hex_chars_uc[((*ds) >> 4) & 0x0F]; 664 d[d_len++] = hex_chars_uc[(*ds) & 0x0F]; 665 d[d_len++] = ';'; 666 break; 667 } 668 } else { 669 d[d_len++] = *ds; 670 } 671 } 672 } 673 674 void buffer_append_string_encoded_json(buffer * const restrict b, const char * const restrict s, const size_t len) { 675 const unsigned char * const restrict ds = (unsigned char *)s; 676 size_t dlen = 0; 677 678 /* calculate space needed for string including encodings */ 679 for (size_t i = 0; i < len; ++i) { 680 int c = ds[i]; 681 if (c == '"' || c == '\\' || c < 0x20 || c == 0x7f) { 682 switch (c) { 683 case '\b': 684 case '\t': 685 case '\n': 686 case '\f': 687 case '\r': 688 case '"': 689 case '\\': 690 dlen += 2; 691 break; 692 default: 693 dlen += 6; /* \uCCCC */ 694 break; 695 } 696 } 697 else { 698 ++dlen; 699 } 700 } 701 702 unsigned char * const d = (unsigned char *)buffer_extend(b, dlen); 703 704 if (__builtin_expect( (dlen == len), 1)) {/*(short-circuit; nothing to encode)*/ 705 memcpy(d, ds, len); 706 return; 707 } 708 709 dlen = 0; 710 for (size_t i = 0; i < len; ++i) { 711 int c = ds[i]; 712 if (c == '"' || c == '\\' || c < 0x20 || c == 0x7f) { 713 d[dlen++] = '\\'; 714 switch (c) { 715 case '\b': 716 d[dlen++] = 'b'; 717 break; 718 case '\t': 719 d[dlen++] = 't'; 720 break; 721 case '\n': 722 d[dlen++] = 'n'; 723 break; 724 case '\f': 725 d[dlen++] = 'f'; 726 break; 727 case '\r': 728 d[dlen++] = 'r'; 729 break; 730 case '"': 731 d[dlen++] = '"'; 732 break; 733 case '\\': 734 d[dlen++] = '\\'; 735 break; 736 default: 737 d[dlen ] = 'u'; 738 d[dlen+1] = '0'; 739 d[dlen+2] = '0'; 740 d[dlen+3] = hex_chars_lc[(c >> 4) & 0x0F]; 741 d[dlen+4] = hex_chars_lc[c & 0x0F]; 742 dlen += 5; 743 break; 744 } 745 } 746 else { 747 d[dlen++] = c; 748 } 749 } 750 } 751 752 753 void buffer_append_string_c_escaped(buffer * const restrict b, const char * const restrict s, size_t s_len) { 754 unsigned char *ds, *d; 755 size_t d_len, ndx; 756 757 /* count to-be-encoded-characters */ 758 for (ds = (unsigned char *)s, d_len = 0, ndx = 0; ndx < s_len; ds++, ndx++) { 759 if ((*ds < 0x20) /* control character */ 760 || (*ds >= 0x7f)) { /* DEL + non-ASCII characters */ 761 switch (*ds) { 762 case '\t': 763 case '\r': 764 case '\n': 765 d_len += 2; 766 break; 767 default: 768 d_len += 4; /* \xCC */ 769 break; 770 } 771 } else { 772 d_len++; 773 } 774 } 775 776 d = (unsigned char*) buffer_extend(b, d_len); 777 778 if (d_len == s_len) { /*(short-circuit; nothing to encoded)*/ 779 memcpy(d, s, s_len); 780 return; 781 } 782 783 for (ds = (unsigned char *)s, d_len = 0, ndx = 0; ndx < s_len; ds++, ndx++) { 784 if ((*ds < 0x20) /* control character */ 785 || (*ds >= 0x7f)) { /* DEL + non-ASCII characters */ 786 d[d_len++] = '\\'; 787 switch (*ds) { 788 case '\t': 789 d[d_len++] = 't'; 790 break; 791 case '\r': 792 d[d_len++] = 'r'; 793 break; 794 case '\n': 795 d[d_len++] = 'n'; 796 break; 797 default: 798 d[d_len++] = 'x'; 799 d[d_len++] = hex_chars_lc[((*ds) >> 4) & 0x0F]; 800 d[d_len++] = hex_chars_lc[(*ds) & 0x0F]; 801 break; 802 } 803 } else { 804 d[d_len++] = *ds; 805 } 806 } 807 } 808 809 810 /* decodes url-special-chars inplace. 811 * replaces non-printable characters with '_' 812 * (If this is used on a portion of query string, then query string should be 813 * split on '&', and '+' replaced with ' ' before calling this routine) 814 */ 815 816 void buffer_urldecode_path(buffer * const b) { 817 const size_t len = buffer_clen(b); 818 char *src = len ? memchr(b->ptr, '%', len) : NULL; 819 if (NULL == src) return; 820 821 char *dst = src; 822 do { 823 /* *src == '%' */ 824 unsigned char high = ((unsigned char *)src)[1]; 825 unsigned char low = high ? hex2int(((unsigned char *)src)[2]) : 0xFF; 826 if (0xFF != (high = hex2int(high)) && 0xFF != low) { 827 high = (high << 4) | low; /* map ctrls to '_' */ 828 *dst = (high >= 32 && high != 127) ? high : '_'; 829 src += 2; 830 } /* else ignore this '%'; leave as-is and move on */ 831 832 while ((*++dst = *++src) != '%' && *src) ; 833 } while (*src); 834 b->used = (dst - b->ptr) + 1; 835 } 836 837 int buffer_is_valid_UTF8(const buffer *b) { 838 /* https://www.w3.org/International/questions/qa-forms-utf-8 */ 839 /*assert(b->used);*//*(b->ptr must exist and be '\0'-terminated)*/ 840 const unsigned char *c = (unsigned char *)b->ptr; 841 while (*c) { 842 843 /*(note: includes ctrls)*/ 844 if ( c[0] < 0x80 ) { ++c; continue; } 845 846 if ( 0xc2 <= c[0] && c[0] <= 0xdf 847 && 0x80 <= c[1] && c[1] <= 0xbf ) { c+=2; continue; } 848 849 if ( ( ( 0xe0 == c[0] 850 && 0xa0 <= c[1] && c[1] <= 0xbf) 851 || ( 0xe1 <= c[0] && c[0] <= 0xef && c[0] != 0xed 852 && 0x80 <= c[1] && c[1] <= 0xbf) 853 || ( 0xed == c[0] 854 && 0x80 <= c[1] && c[1] <= 0x9f) ) 855 && 0x80 <= c[2] && c[2] <= 0xbf ) { c+=3; continue; } 856 857 if ( ( ( 0xf0 == c[0] 858 && 0x90 <= c[1] && c[1] <= 0xbf) 859 || ( 0xf1 <= c[0] && c[0] <= 0xf3 860 && 0x80 <= c[1] && c[1] <= 0xbf) 861 || ( 0xf4 == c[0] 862 && 0x80 <= c[1] && c[1] <= 0x8f) ) 863 && 0x80 <= c[2] && c[2] <= 0xbf 864 && 0x80 <= c[3] && c[3] <= 0xbf ) { c+=4; continue; } 865 866 return 0; /* invalid */ 867 } 868 return 1; /* valid */ 869 } 870 871 /* - special case: empty string returns empty string 872 * - on windows or cygwin: replace \ with / 873 * - strip leading spaces 874 * - prepends "/" if not present already 875 * - resolve "/../", "//" and "/./" the usual way: 876 * the first one removes a preceding component, the other two 877 * get compressed to "/". 878 * - "/." and "/.." at the end are similar, but always leave a trailing 879 * "/" 880 * 881 * /blah/.. gets / 882 * /blah/../foo gets /foo 883 * /abc/./xyz gets /abc/xyz 884 * /abc//xyz gets /abc/xyz 885 */ 886 887 void buffer_path_simplify(buffer *b) 888 { 889 char *out = b->ptr; 890 char * const end = b->ptr + b->used - 1; 891 892 if (__builtin_expect( (buffer_is_blank(b)), 0)) { 893 buffer_blank(b); 894 return; 895 } 896 897 #if defined(__WIN32) || defined(__CYGWIN__) 898 /* cygwin is treating \ and / the same, so we have to that too */ 899 for (char *p = b->ptr; *p; p++) { 900 if (*p == '\\') *p = '/'; 901 } 902 #endif 903 904 *end = '/'; /*(end of path modified to avoid need to check '\0')*/ 905 906 char *walk = out; 907 if (__builtin_expect( (*walk == '/'), 1)) { 908 /* scan to detect (potential) need for path simplification 909 * (repeated '/' or "/.") */ 910 do { 911 if (*++walk == '.' || *walk == '/') 912 break; 913 do { ++walk; } while (*walk != '/'); 914 } while (walk != end); 915 if (__builtin_expect( (walk == end), 1)) { 916 /* common case: no repeated '/' or "/." */ 917 *end = '\0'; /* overwrite extra '/' added to end of path */ 918 return; 919 } 920 out = walk-1; 921 } 922 else { 923 if (walk[0] == '.' && walk[1] == '/') 924 *out = *++walk; 925 else if (walk[0] == '.' && walk[1] == '.' && walk[2] == '/') 926 *out = *(walk += 2); 927 else { 928 while (*++walk != '/') ; 929 out = walk; 930 } 931 ++walk; 932 } 933 934 while (walk <= end) { 935 /* previous char is '/' at this point (or start of string w/o '/') */ 936 if (__builtin_expect( (walk[0] == '/'), 0)) { 937 /* skip repeated '/' (e.g. "///" -> "/") */ 938 if (++walk < end) 939 continue; 940 else { 941 ++out; 942 break; 943 } 944 } 945 else if (__builtin_expect( (walk[0] == '.'), 0)) { 946 /* handle "./" and "../" */ 947 if (walk[1] == '.' && walk[2] == '/') { 948 /* handle "../" */ 949 while (out > b->ptr && *--out != '/') ; 950 *out = '/'; /*(in case path had not started with '/')*/ 951 if ((walk += 3) >= end) { 952 ++out; 953 break; 954 } 955 else 956 continue; 957 } 958 else if (walk[1] == '/') { 959 /* handle "./" */ 960 if ((walk += 2) >= end) { 961 ++out; 962 break; 963 } 964 continue; 965 } 966 else { 967 /* accept "." if not part of "../" or "./" */ 968 *++out = '.'; 969 ++walk; 970 } 971 } 972 973 while ((*++out = *walk++) != '/') ; 974 } 975 *out = *end = '\0'; /* overwrite extra '/' added to end of path */ 976 b->used = (out - b->ptr) + 1; 977 /*buffer_truncate(b, out - b->ptr);*/ 978 } 979 980 void buffer_to_lower(buffer * const b) { 981 unsigned char * const restrict s = (unsigned char *)b->ptr; 982 const uint_fast32_t used = b->used; 983 for (uint_fast32_t i = 0; i < used; ++i) { 984 if (light_isupper(s[i])) s[i] |= 0x20; 985 } 986 } 987 988 989 void buffer_to_upper(buffer * const b) { 990 unsigned char * const restrict s = (unsigned char *)b->ptr; 991 const uint_fast32_t used = b->used; 992 for (uint_fast32_t i = 0; i < used; ++i) { 993 if (light_islower(s[i])) s[i] &= 0xdf; 994 } 995 } 996