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