00001
00002
00003
00004
00005
00006
00007
00008
00009
00012 #include "stdafx.h"
00013 #include "debug.h"
00014 #include "core/alloc_func.hpp"
00015 #include "core/math_func.hpp"
00016 #include "string_func.h"
00017
00018 #include "table/control_codes.h"
00019
00020 #include <stdarg.h>
00021 #include <ctype.h>
00022
00023 #ifdef _MSC_VER
00024 #include <errno.h>
00025 #endif
00026
00027 #ifdef WITH_ICU
00028
00029 #include <unicode/ustring.h>
00030 #include "language.h"
00031 #include "gfx_func.h"
00032 #endif
00033
00044 static int CDECL vseprintf(char *str, const char *last, const char *format, va_list ap)
00045 {
00046 ptrdiff_t diff = last - str;
00047 if (diff < 0) return 0;
00048 return min((int)diff, vsnprintf(str, diff + 1, format, ap));
00049 }
00050
00065 void ttd_strlcat(char *dst, const char *src, size_t size)
00066 {
00067 assert(size > 0);
00068 while (size > 0 && *dst != '\0') {
00069 size--;
00070 dst++;
00071 }
00072
00073 ttd_strlcpy(dst, src, size);
00074 }
00075
00076
00091 void ttd_strlcpy(char *dst, const char *src, size_t size)
00092 {
00093 assert(size > 0);
00094 while (--size > 0 && *src != '\0') {
00095 *dst++ = *src++;
00096 }
00097 *dst = '\0';
00098 }
00099
00100
00117 char *strecat(char *dst, const char *src, const char *last)
00118 {
00119 assert(dst <= last);
00120 while (*dst != '\0') {
00121 if (dst == last) return dst;
00122 dst++;
00123 }
00124
00125 return strecpy(dst, src, last);
00126 }
00127
00128
00145 char *strecpy(char *dst, const char *src, const char *last)
00146 {
00147 assert(dst <= last);
00148 while (dst != last && *src != '\0') {
00149 *dst++ = *src++;
00150 }
00151 *dst = '\0';
00152
00153 if (dst == last && *src != '\0') {
00154 #if defined(STRGEN) || defined(SETTINGSGEN)
00155 error("String too long for destination buffer");
00156 #else
00157 DEBUG(misc, 0, "String too long for destination buffer");
00158 #endif
00159 }
00160 return dst;
00161 }
00162
00168 char *CDECL str_fmt(const char *str, ...)
00169 {
00170 char buf[4096];
00171 va_list va;
00172
00173 va_start(va, str);
00174 int len = vseprintf(buf, lastof(buf), str, va);
00175 va_end(va);
00176 char *p = MallocT<char>(len + 1);
00177 memcpy(p, buf, len + 1);
00178 return p;
00179 }
00180
00181
00190 void str_validate(char *str, const char *last, bool allow_newlines, bool ignore)
00191 {
00192
00193
00194 char *dst = str;
00195 while (str <= last && *str != '\0') {
00196 size_t len = Utf8EncodedCharLen(*str);
00197
00198
00199
00200
00201
00202 if ((len == 0 && str + 4 > last) || str + len > last) break;
00203
00204 WChar c;
00205 len = Utf8Decode(&c, str);
00206
00207
00208
00209 if (c == '\0') break;
00210
00211 if (IsPrintable(c) && (c < SCC_SPRITE_START || c > SCC_SPRITE_END)) {
00212
00213
00214
00215 do {
00216 *dst++ = *str++;
00217 } while (--len != 0);
00218 } else if (allow_newlines && c == '\n') {
00219 *dst++ = *str++;
00220 } else {
00221 if (allow_newlines && c == '\r' && str[1] == '\n') {
00222 str += len;
00223 continue;
00224 }
00225
00226 str += len;
00227 if (!ignore) *dst++ = '?';
00228
00229
00230
00231
00232
00233 if (c == SCC_SETX) {
00234 str++;
00235 } else if (c == SCC_SETXY) {
00236 str += 2;
00237 }
00238 }
00239 }
00240
00241 *dst = '\0';
00242 }
00243
00251 bool StrValid(const char *str, const char *last)
00252 {
00253
00254
00255 while (str <= last && *str != '\0') {
00256 size_t len = Utf8EncodedCharLen(*str);
00257
00258
00259
00260
00261 if (len == 0 || str + len > last) return false;
00262
00263 WChar c;
00264 len = Utf8Decode(&c, str);
00265 if (!IsPrintable(c) || (c >= SCC_SPRITE_START && c <= SCC_SPRITE_END)) {
00266 return false;
00267 }
00268
00269 str += len;
00270 }
00271
00272 return *str == '\0';
00273 }
00274
00276 void str_strip_colours(char *str)
00277 {
00278 char *dst = str;
00279 WChar c;
00280 size_t len;
00281
00282 for (len = Utf8Decode(&c, str); c != '\0'; len = Utf8Decode(&c, str)) {
00283 if (c < SCC_BLUE || c > SCC_BLACK) {
00284
00285
00286
00287 do {
00288 *dst++ = *str++;
00289 } while (--len != 0);
00290 } else {
00291
00292 str += len;
00293 }
00294 }
00295 *dst = '\0';
00296 }
00297
00304 size_t Utf8StringLength(const char *s)
00305 {
00306 size_t len = 0;
00307 const char *t = s;
00308 while (Utf8Consume(&t) != 0) len++;
00309 return len;
00310 }
00311
00312
00323 void strtolower(char *str)
00324 {
00325 for (; *str != '\0'; str++) *str = tolower(*str);
00326 }
00327
00335 bool IsValidChar(WChar key, CharSetFilter afilter)
00336 {
00337 switch (afilter) {
00338 case CS_ALPHANUMERAL: return IsPrintable(key);
00339 case CS_NUMERAL: return (key >= '0' && key <= '9');
00340 case CS_NUMERAL_SPACE: return (key >= '0' && key <= '9') || key == ' ';
00341 case CS_ALPHA: return IsPrintable(key) && !(key >= '0' && key <= '9');
00342 case CS_HEXADECIMAL: return (key >= '0' && key <= '9') || (key >= 'a' && key <= 'f') || (key >= 'A' && key <= 'F');
00343 }
00344
00345 return false;
00346 }
00347
00348 #ifdef WIN32
00349
00350 #if (__MINGW32_MAJOR_VERSION < 3) || ((__MINGW32_MAJOR_VERSION == 3) && (__MINGW32_MINOR_VERSION < 14))
00351 int CDECL snprintf(char *str, size_t size, const char *format, ...)
00352 {
00353 va_list ap;
00354 int ret;
00355
00356 va_start(ap, format);
00357 ret = vsnprintf(str, size, format, ap);
00358 va_end(ap);
00359 return ret;
00360 }
00361 #endif
00362
00363 #ifdef _MSC_VER
00364
00371 int CDECL vsnprintf(char *str, size_t size, const char *format, va_list ap)
00372 {
00373 if (size == 0) return 0;
00374
00375 errno = 0;
00376 int ret = _vsnprintf(str, size, format, ap);
00377
00378 if (ret < 0) {
00379 if (errno != ERANGE) {
00380
00381
00382 NOT_REACHED();
00383 }
00384 } else if ((size_t)ret < size) {
00385
00386
00387
00388 return ret;
00389 }
00390
00391
00392
00393 str[size - 1] = '\0';
00394 return (int)size;
00395 }
00396 #endif
00397
00398 #endif
00399
00409 int CDECL seprintf(char *str, const char *last, const char *format, ...)
00410 {
00411 va_list ap;
00412
00413 va_start(ap, format);
00414 int ret = vseprintf(str, last, format, ap);
00415 va_end(ap);
00416 return ret;
00417 }
00418
00419
00427 char *md5sumToString(char *buf, const char *last, const uint8 md5sum[16])
00428 {
00429 char *p = buf;
00430
00431 for (uint i = 0; i < 16; i++) {
00432 p += seprintf(p, last, "%02X", md5sum[i]);
00433 }
00434
00435 return p;
00436 }
00437
00438
00439
00440
00441
00448 size_t Utf8Decode(WChar *c, const char *s)
00449 {
00450 assert(c != NULL);
00451
00452 if (!HasBit(s[0], 7)) {
00453
00454 *c = s[0];
00455 return 1;
00456 } else if (GB(s[0], 5, 3) == 6) {
00457 if (IsUtf8Part(s[1])) {
00458
00459 *c = GB(s[0], 0, 5) << 6 | GB(s[1], 0, 6);
00460 if (*c >= 0x80) return 2;
00461 }
00462 } else if (GB(s[0], 4, 4) == 14) {
00463 if (IsUtf8Part(s[1]) && IsUtf8Part(s[2])) {
00464
00465 *c = GB(s[0], 0, 4) << 12 | GB(s[1], 0, 6) << 6 | GB(s[2], 0, 6);
00466 if (*c >= 0x800) return 3;
00467 }
00468 } else if (GB(s[0], 3, 5) == 30) {
00469 if (IsUtf8Part(s[1]) && IsUtf8Part(s[2]) && IsUtf8Part(s[3])) {
00470
00471 *c = GB(s[0], 0, 3) << 18 | GB(s[1], 0, 6) << 12 | GB(s[2], 0, 6) << 6 | GB(s[3], 0, 6);
00472 if (*c >= 0x10000 && *c <= 0x10FFFF) return 4;
00473 }
00474 }
00475
00476
00477 *c = '?';
00478 return 1;
00479 }
00480
00481
00488 size_t Utf8Encode(char *buf, WChar c)
00489 {
00490 if (c < 0x80) {
00491 *buf = c;
00492 return 1;
00493 } else if (c < 0x800) {
00494 *buf++ = 0xC0 + GB(c, 6, 5);
00495 *buf = 0x80 + GB(c, 0, 6);
00496 return 2;
00497 } else if (c < 0x10000) {
00498 *buf++ = 0xE0 + GB(c, 12, 4);
00499 *buf++ = 0x80 + GB(c, 6, 6);
00500 *buf = 0x80 + GB(c, 0, 6);
00501 return 3;
00502 } else if (c < 0x110000) {
00503 *buf++ = 0xF0 + GB(c, 18, 3);
00504 *buf++ = 0x80 + GB(c, 12, 6);
00505 *buf++ = 0x80 + GB(c, 6, 6);
00506 *buf = 0x80 + GB(c, 0, 6);
00507 return 4;
00508 }
00509
00510
00511 *buf = '?';
00512 return 1;
00513 }
00514
00522 size_t Utf8TrimString(char *s, size_t maxlen)
00523 {
00524 size_t length = 0;
00525
00526 for (const char *ptr = strchr(s, '\0'); *s != '\0';) {
00527 size_t len = Utf8EncodedCharLen(*s);
00528
00529 if (len == 0) len = 1;
00530
00531
00532
00533 if (length + len >= maxlen || (s + len > ptr)) break;
00534 s += len;
00535 length += len;
00536 }
00537
00538 *s = '\0';
00539 return length;
00540 }
00541
00542 #ifdef DEFINE_STRNDUP
00543 #include "core/math_func.hpp"
00544 char *strndup(const char *s, size_t len)
00545 {
00546 len = min(strlen(s), len);
00547 char *tmp = CallocT<char>(len + 1);
00548 memcpy(tmp, s, len);
00549 return tmp;
00550 }
00551 #endif
00552
00553 #ifdef DEFINE_STRCASESTR
00554 char *strcasestr(const char *haystack, const char *needle)
00555 {
00556 size_t hay_len = strlen(haystack);
00557 size_t needle_len = strlen(needle);
00558 while (hay_len >= needle_len) {
00559 if (strncasecmp(haystack, needle, needle_len) == 0) return const_cast<char *>(haystack);
00560
00561 haystack++;
00562 hay_len--;
00563 }
00564
00565 return NULL;
00566 }
00567 #endif
00568
00576 int strnatcmp(const char *s1, const char *s2)
00577 {
00578 #ifdef WITH_ICU
00579 if (_current_collator != NULL) {
00580 UErrorCode status = U_ZERO_ERROR;
00581 int result;
00582
00583
00584 #if U_ICU_VERSION_MAJOR_NUM > 4 || (U_ICU_VERSION_MAJOR_NUM == 4 && U_ICU_VERSION_MINOR_NUM >= 2)
00585
00586 result = _current_collator->compareUTF8(s1, s2, status);
00587 #else
00588 UChar buffer1[DRAW_STRING_BUFFER];
00589 u_strFromUTF8Lenient(buffer1, lengthof(buffer1), NULL, s1, -1, &status);
00590 UChar buffer2[DRAW_STRING_BUFFER];
00591 u_strFromUTF8Lenient(buffer2, lengthof(buffer2), NULL, s2, -1, &status);
00592
00593 result = _current_collator->compare(buffer1, buffer2, status);
00594 #endif
00595 if (U_SUCCESS(status)) return result;
00596 }
00597
00598 #endif
00599
00600
00601 return strcasecmp(s1, s2);
00602 }