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 return (size < b->size) 97 ? b->ptr 98 : buffer_alloc_replace(b, size); 99 } 100 101 __attribute_cold__ 102 __attribute_noinline__ 103 __attribute_nonnull__ 104 __attribute_returns_nonnull__ 105 static char* buffer_string_prepare_append_resize(buffer * const restrict b, const size_t size) { 106 if (b->used < 2) { /* buffer_is_blank(b) */ 107 char * const s = buffer_string_prepare_copy(b, size); 108 *s = '\0'; /*(for case (1 == b->used))*/ 109 return s; 110 } 111 112 /* not empty, b->used already includes a terminating 0 */ 113 /*(note: if size larger than one lshift, use size instead of power-2)*/ 114 const size_t bsize2x = (b->size & ~1uL) << 1; 115 const size_t req_size = (bsize2x - b->used > size) 116 ? bsize2x-1 117 : b->used + size; 118 119 /* check for overflow: unsigned overflow is defined to wrap around */ 120 force_assert(req_size >= b->used); 121 122 return buffer_realloc(b, req_size) + b->used - 1; 123 } 124 125 char* buffer_string_prepare_append(buffer * const b, const size_t size) { 126 const uint32_t len = b->used ? b->used-1 : 0; 127 return (b->size - len >= size + 1) 128 ? b->ptr + len 129 : buffer_string_prepare_append_resize(b, size); 130 } 131 132 /*(prefer smaller code than inlining buffer_extend in many places in buffer.c)*/ 133 __attribute_noinline__ 134 char* 135 buffer_extend (buffer * const b, const size_t x) 136 { 137 /* extend buffer to append x (reallocate by power-2 (or larger), if needed) 138 * (combine buffer_string_prepare_append() and buffer_commit()) 139 * (future: might make buffer.h static inline func for HTTP/1.1 performance) 140 * pre-sets '\0' byte and b->used (unlike buffer_string_prepare_append())*/ 141 #if 0 142 char * const s = buffer_string_prepare_append(b, x); 143 b->used += x + (0 == b->used); 144 #else 145 const uint32_t len = b->used ? b->used-1 : 0; 146 char * const s = (b->size - len >= x + 1) 147 ? b->ptr + len 148 : buffer_string_prepare_append_resize(b, x); 149 b->used = len+x+1; 150 #endif 151 s[x] = '\0'; 152 return s; 153 } 154 155 void buffer_commit(buffer *b, size_t size) 156 { 157 size_t sz = b->used; 158 if (0 == sz) sz = 1; 159 160 if (size > 0) { 161 /* check for overflow: unsigned overflow is defined to wrap around */ 162 sz += size; 163 force_assert(sz > size); 164 } 165 166 b->used = sz; 167 b->ptr[sz - 1] = '\0'; 168 } 169 170 void buffer_copy_string(buffer * restrict b, const char * restrict s) { 171 buffer_copy_string_len(b, s, NULL != s ? strlen(s) : 0); 172 } 173 174 void buffer_copy_string_len(buffer * const restrict b, const char * const restrict s, const size_t len) { 175 b->used = len + 1; 176 char * const restrict d = (len < b->size) 177 ? b->ptr 178 : buffer_alloc_replace(b, len); 179 d[len] = '\0'; 180 memcpy(d, s, len); 181 } 182 183 void buffer_append_string(buffer * restrict b, const char * restrict s) { 184 buffer_append_string_len(b, s, NULL != s ? strlen(s) : 0); 185 } 186 187 /** 188 * append a string to the end of the buffer 189 * 190 * the resulting buffer is terminated with a '\0' 191 * s is treated as a un-terminated string (a \0 is handled a normal character) 192 * 193 * @param b a buffer 194 * @param s the string 195 * @param s_len size of the string (without the terminating \0) 196 */ 197 198 void buffer_append_string_len(buffer * const restrict b, const char * const restrict s, const size_t len) { 199 memcpy(buffer_extend(b, len), s, len); 200 } 201 202 void buffer_append_str2(buffer * const restrict b, const char * const s1, const size_t len1, const char * const s2, const size_t len2) { 203 char * const restrict s = buffer_extend(b, len1+len2); 204 #ifdef HAVE_MEMPCPY 205 mempcpy(mempcpy(s, s1, len1), s2, len2); 206 #else 207 memcpy(s, s1, len1); 208 memcpy(s+len1, s2, len2); 209 #endif 210 } 211 212 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) { 213 char * restrict s = buffer_extend(b, len1+len2+len3); 214 #ifdef HAVE_MEMPCPY 215 mempcpy(mempcpy(mempcpy(s, s1, len1), s2, len2), s3, len3); 216 #else 217 memcpy(s, s1, len1); 218 memcpy((s+=len1), s2, len2); 219 memcpy((s+=len2), s3, len3); 220 #endif 221 } 222 223 void buffer_append_iovec(buffer * const restrict b, const struct const_iovec * const iov, const size_t n) { 224 size_t len = 0; 225 for (size_t i = 0; i < n; ++i) 226 len += iov[i].iov_len; 227 char *s = buffer_extend(b, len); 228 for (size_t i = 0; i < n; ++i) { 229 if (0 == iov[i].iov_len) continue; 230 #ifdef HAVE_MEMPCPY 231 s = mempcpy(s, iov[i].iov_base, iov[i].iov_len); 232 #else 233 memcpy(s, iov[i].iov_base, iov[i].iov_len); 234 s += iov[i].iov_len; 235 #endif 236 } 237 } 238 239 void buffer_append_path_len(buffer * restrict b, const char * restrict a, size_t alen) { 240 char * restrict s = buffer_string_prepare_append(b, alen+1); 241 const int aslash = (alen && a[0] == '/'); 242 if (b->used > 1 && s[-1] == '/') { 243 if (aslash) { 244 ++a; 245 --alen; 246 } 247 } 248 else { 249 if (0 == b->used) b->used = 1; 250 if (!aslash) { 251 *s++ = '/'; 252 ++b->used; 253 } 254 } 255 b->used += alen; 256 s[alen] = '\0'; 257 memcpy(s, a, alen); 258 } 259 260 void 261 buffer_copy_path_len2 (buffer * const restrict b, const char * const restrict s1, size_t len1, const char * const restrict s2, size_t len2) 262 { 263 /*(similar to buffer_copy_string_len(b, s1, len1) but combined allocation)*/ 264 memcpy(buffer_string_prepare_copy(b, len1+len2+1), s1, len1); 265 b->used = len1 + 1; /*('\0' byte will be written below)*/ 266 267 buffer_append_path_len(b, s2, len2);/*(choice: not inlined, special-cased)*/ 268 } 269 270 void 271 buffer_copy_string_len_lc (buffer * const restrict b, const char * const restrict s, const size_t len) 272 { 273 char * const restrict d = buffer_string_prepare_copy(b, len); 274 b->used = len+1; 275 d[len] = '\0'; 276 for (size_t i = 0; i < len; ++i) 277 d[i] = (!light_isupper(s[i])) ? s[i] : s[i] | 0x20; 278 } 279 280 void buffer_append_uint_hex_lc(buffer *b, uintmax_t value) { 281 char *buf; 282 unsigned int shift = 0; 283 284 { 285 uintmax_t copy = value; 286 do { 287 copy >>= 8; 288 shift += 8; /* counting bits */ 289 } while (0 != copy); 290 } 291 292 buf = buffer_extend(b, shift >> 2); /*nibbles (4 bits)*/ 293 294 while (shift > 0) { 295 shift -= 4; 296 *(buf++) = hex_chars_lc[(value >> shift) & 0x0F]; 297 } 298 } 299 300 __attribute_nonnull__ 301 __attribute_returns_nonnull__ 302 static char* utostr(char buf[LI_ITOSTRING_LENGTH], uintmax_t val) { 303 char *cur = buf+LI_ITOSTRING_LENGTH; 304 do { 305 int mod = val % 10; 306 val /= 10; 307 /* prepend digit mod */ 308 *(--cur) = (char) ('0' + mod); 309 } while (0 != val); 310 return cur; 311 } 312 313 __attribute_nonnull__ 314 __attribute_returns_nonnull__ 315 static char* itostr(char buf[LI_ITOSTRING_LENGTH], intmax_t val) { 316 /* absolute value not defined for INTMAX_MIN, but can take absolute 317 * value of any negative number via twos complement cast to unsigned. 318 * negative sign is prepended after (now unsigned) value is converted 319 * to string */ 320 uintmax_t uval = val >= 0 ? (uintmax_t)val : ((uintmax_t)~val) + 1; 321 char *cur = utostr(buf, uval); 322 if (val < 0) *(--cur) = '-'; 323 324 return cur; 325 } 326 327 void buffer_append_int(buffer *b, intmax_t val) { 328 char buf[LI_ITOSTRING_LENGTH]; 329 const char * const str = itostr(buf, val); 330 buffer_append_string_len(b, str, buf+sizeof(buf) - str); 331 } 332 333 void buffer_append_strftime(buffer * const restrict b, const char * const restrict format, const struct tm * const restrict tm) { 334 force_assert(NULL != format); 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 if (0 == s_len) return; 610 611 force_assert(NULL != s); 612 613 switch(encoding) { 614 case ENCODING_REL_URI: 615 map = encoded_chars_rel_uri; 616 break; 617 case ENCODING_REL_URI_PART: 618 map = encoded_chars_rel_uri_part; 619 break; 620 case ENCODING_HTML: 621 map = encoded_chars_html; 622 break; 623 case ENCODING_MINIMAL_XML: 624 map = encoded_chars_minimal_xml; 625 break; 626 } 627 628 force_assert(NULL != map); 629 630 /* count to-be-encoded-characters */ 631 for (ds = (unsigned char *)s, d_len = 0, ndx = 0; ndx < s_len; ds++, ndx++) { 632 if (map[*ds & 0xFF]) { 633 switch(encoding) { 634 case ENCODING_REL_URI: 635 case ENCODING_REL_URI_PART: 636 d_len += 3; 637 break; 638 case ENCODING_HTML: 639 case ENCODING_MINIMAL_XML: 640 d_len += 6; 641 break; 642 } 643 } else { 644 d_len++; 645 } 646 } 647 648 d = (unsigned char*) buffer_extend(b, d_len); 649 650 if (d_len == s_len) { /*(short-circuit; nothing to encoded)*/ 651 memcpy(d, s, s_len); 652 return; 653 } 654 655 for (ds = (unsigned char *)s, d_len = 0, ndx = 0; ndx < s_len; ds++, ndx++) { 656 if (map[*ds & 0xFF]) { 657 switch(encoding) { 658 case ENCODING_REL_URI: 659 case ENCODING_REL_URI_PART: 660 d[d_len++] = '%'; 661 d[d_len++] = hex_chars_uc[((*ds) >> 4) & 0x0F]; 662 d[d_len++] = hex_chars_uc[(*ds) & 0x0F]; 663 break; 664 case ENCODING_HTML: 665 case ENCODING_MINIMAL_XML: 666 d[d_len++] = '&'; 667 d[d_len++] = '#'; 668 d[d_len++] = 'x'; 669 d[d_len++] = hex_chars_uc[((*ds) >> 4) & 0x0F]; 670 d[d_len++] = hex_chars_uc[(*ds) & 0x0F]; 671 d[d_len++] = ';'; 672 break; 673 } 674 } else { 675 d[d_len++] = *ds; 676 } 677 } 678 } 679 680 void buffer_append_string_c_escaped(buffer * const restrict b, const char * const restrict s, size_t s_len) { 681 unsigned char *ds, *d; 682 size_t d_len, ndx; 683 684 if (0 == s_len) return; 685 686 /* count to-be-encoded-characters */ 687 for (ds = (unsigned char *)s, d_len = 0, ndx = 0; ndx < s_len; ds++, ndx++) { 688 if ((*ds < 0x20) /* control character */ 689 || (*ds >= 0x7f)) { /* DEL + non-ASCII characters */ 690 switch (*ds) { 691 case '\t': 692 case '\r': 693 case '\n': 694 d_len += 2; 695 break; 696 default: 697 d_len += 4; /* \xCC */ 698 break; 699 } 700 } else { 701 d_len++; 702 } 703 } 704 705 d = (unsigned char*) buffer_extend(b, d_len); 706 707 if (d_len == s_len) { /*(short-circuit; nothing to encoded)*/ 708 memcpy(d, s, s_len); 709 return; 710 } 711 712 for (ds = (unsigned char *)s, d_len = 0, ndx = 0; ndx < s_len; ds++, ndx++) { 713 if ((*ds < 0x20) /* control character */ 714 || (*ds >= 0x7f)) { /* DEL + non-ASCII characters */ 715 d[d_len++] = '\\'; 716 switch (*ds) { 717 case '\t': 718 d[d_len++] = 't'; 719 break; 720 case '\r': 721 d[d_len++] = 'r'; 722 break; 723 case '\n': 724 d[d_len++] = 'n'; 725 break; 726 default: 727 d[d_len++] = 'x'; 728 d[d_len++] = hex_chars_lc[((*ds) >> 4) & 0x0F]; 729 d[d_len++] = hex_chars_lc[(*ds) & 0x0F]; 730 break; 731 } 732 } else { 733 d[d_len++] = *ds; 734 } 735 } 736 } 737 738 739 /* decodes url-special-chars inplace. 740 * replaces non-printable characters with '_' 741 * (If this is used on a portion of query string, then query string should be 742 * split on '&', and '+' replaced with ' ' before calling this routine) 743 */ 744 745 void buffer_urldecode_path(buffer * const b) { 746 const size_t len = buffer_clen(b); 747 char *src = len ? memchr(b->ptr, '%', len) : NULL; 748 if (NULL == src) return; 749 750 char *dst = src; 751 do { 752 /* *src == '%' */ 753 unsigned char high = hex2int(*(src + 1)); 754 unsigned char low = hex2int(*(src + 2)); 755 if (0xFF != high && 0xFF != low) { 756 high = (high << 4) | low; /* map ctrls to '_' */ 757 *dst = (high >= 32 && high != 127) ? high : '_'; 758 src += 2; 759 } /* else ignore this '%'; leave as-is and move on */ 760 761 while ((*++dst = *++src) != '%' && *src) ; 762 } while (*src); 763 b->used = (dst - b->ptr) + 1; 764 } 765 766 int buffer_is_valid_UTF8(const buffer *b) { 767 /* https://www.w3.org/International/questions/qa-forms-utf-8 */ 768 const unsigned char *c = (unsigned char *)b->ptr; 769 while (*c) { 770 771 /*(note: includes ctrls)*/ 772 if ( c[0] < 0x80 ) { ++c; continue; } 773 774 if ( 0xc2 <= c[0] && c[0] <= 0xdf 775 && 0x80 <= c[1] && c[1] <= 0xbf ) { c+=2; continue; } 776 777 if ( ( ( 0xe0 == c[0] 778 && 0xa0 <= c[1] && c[1] <= 0xbf) 779 || ( 0xe1 <= c[0] && c[0] <= 0xef && c[0] != 0xed 780 && 0x80 <= c[1] && c[1] <= 0xbf) 781 || ( 0xed == c[0] 782 && 0x80 <= c[1] && c[1] <= 0x9f) ) 783 && 0x80 <= c[2] && c[2] <= 0xbf ) { c+=3; continue; } 784 785 if ( ( ( 0xf0 == c[0] 786 && 0x90 <= c[1] && c[1] <= 0xbf) 787 || ( 0xf1 <= c[0] && c[0] <= 0xf3 788 && 0x80 <= c[1] && c[1] <= 0xbf) 789 || ( 0xf4 == c[0] 790 && 0x80 <= c[1] && c[1] <= 0x8f) ) 791 && 0x80 <= c[2] && c[2] <= 0xbf 792 && 0x80 <= c[3] && c[3] <= 0xbf ) { c+=4; continue; } 793 794 return 0; /* invalid */ 795 } 796 return 1; /* valid */ 797 } 798 799 /* - special case: empty string returns empty string 800 * - on windows or cygwin: replace \ with / 801 * - strip leading spaces 802 * - prepends "/" if not present already 803 * - resolve "/../", "//" and "/./" the usual way: 804 * the first one removes a preceding component, the other two 805 * get compressed to "/". 806 * - "/." and "/.." at the end are similar, but always leave a trailing 807 * "/" 808 * 809 * /blah/.. gets / 810 * /blah/../foo gets /foo 811 * /abc/./xyz gets /abc/xyz 812 * /abc//xyz gets /abc/xyz 813 */ 814 815 void buffer_path_simplify(buffer *b) 816 { 817 char *out = b->ptr; 818 char * const end = b->ptr + b->used - 1; 819 820 if (__builtin_expect( (buffer_is_blank(b)), 0)) { 821 buffer_blank(b); 822 return; 823 } 824 825 #if defined(__WIN32) || defined(__CYGWIN__) 826 /* cygwin is treating \ and / the same, so we have to that too */ 827 for (char *p = b->ptr; *p; p++) { 828 if (*p == '\\') *p = '/'; 829 } 830 #endif 831 832 *end = '/'; /*(end of path modified to avoid need to check '\0')*/ 833 834 char *walk = out; 835 if (__builtin_expect( (*walk == '/'), 1)) { 836 /* scan to detect (potential) need for path simplification 837 * (repeated '/' or "/.") */ 838 do { 839 if (*++walk == '.' || *walk == '/') 840 break; 841 do { ++walk; } while (*walk != '/'); 842 } while (walk != end); 843 if (__builtin_expect( (walk == end), 1)) { 844 /* common case: no repeated '/' or "/." */ 845 *end = '\0'; /* overwrite extra '/' added to end of path */ 846 return; 847 } 848 out = walk-1; 849 } 850 else { 851 if (walk[0] == '.' && walk[1] == '/') 852 *out = *++walk; 853 else if (walk[0] == '.' && walk[1] == '.' && walk[2] == '/') 854 *out = *(walk += 2); 855 else { 856 while (*++walk != '/') ; 857 out = walk; 858 } 859 ++walk; 860 } 861 862 while (walk <= end) { 863 /* previous char is '/' at this point (or start of string w/o '/') */ 864 if (__builtin_expect( (walk[0] == '/'), 0)) { 865 /* skip repeated '/' (e.g. "///" -> "/") */ 866 if (++walk < end) 867 continue; 868 else { 869 ++out; 870 break; 871 } 872 } 873 else if (__builtin_expect( (walk[0] == '.'), 0)) { 874 /* handle "./" and "../" */ 875 if (walk[1] == '.' && walk[2] == '/') { 876 /* handle "../" */ 877 while (out > b->ptr && *--out != '/') ; 878 *out = '/'; /*(in case path had not started with '/')*/ 879 if ((walk += 3) >= end) { 880 ++out; 881 break; 882 } 883 else 884 continue; 885 } 886 else if (walk[1] == '/') { 887 /* handle "./" */ 888 if ((walk += 2) >= end) { 889 ++out; 890 break; 891 } 892 continue; 893 } 894 else { 895 /* accept "." if not part of "../" or "./" */ 896 *++out = '.'; 897 ++walk; 898 } 899 } 900 901 while ((*++out = *walk++) != '/') ; 902 } 903 *out = *end = '\0'; /* overwrite extra '/' added to end of path */ 904 b->used = (out - b->ptr) + 1; 905 /*buffer_truncate(b, out - b->ptr);*/ 906 } 907 908 void buffer_to_lower(buffer * const b) { 909 unsigned char * const restrict s = (unsigned char *)b->ptr; 910 for (uint32_t i = 0; i < b->used; ++i) { 911 if (light_isupper(s[i])) s[i] |= 0x20; 912 } 913 } 914 915 916 void buffer_to_upper(buffer * const b) { 917 unsigned char * const restrict s = (unsigned char *)b->ptr; 918 for (uint32_t i = 0; i < b->used; ++i) { 919 if (light_islower(s[i])) s[i] &= 0xdf; 920 } 921 } 922