string.cpp

Go to the documentation of this file.
00001 /* $Id: string.cpp 21846 2011-01-18 23:09:43Z rubidium $ */
00002 
00003 /*
00004  * This file is part of OpenTTD.
00005  * OpenTTD is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, version 2.
00006  * OpenTTD is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
00007  * See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with OpenTTD. If not, see <http://www.gnu.org/licenses/>.
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> /* required for tolower() */
00022 
00023 #ifdef _MSC_VER
00024 #include <errno.h> // required by vsnprintf implementation for MSVC
00025 #endif
00026 
00027 #ifdef WITH_ICU
00028 /* Required by strnatcmp. */
00029 #include <unicode/ustring.h>
00030 #include "language.h"
00031 #include "gfx_func.h"
00032 #endif /* WITH_ICU */
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 #ifdef STRGEN
00155     error("String too long for destination buffer");
00156 #else /* STRGEN */
00157     DEBUG(misc, 0, "String too long for destination buffer");
00158 #endif /* STRGEN */
00159   }
00160   return dst;
00161 }
00162 
00163 
00164 char *CDECL str_fmt(const char *str, ...)
00165 {
00166   char buf[4096];
00167   va_list va;
00168 
00169   va_start(va, str);
00170   int len = vseprintf(buf, lastof(buf), str, va);
00171   va_end(va);
00172   char *p = MallocT<char>(len + 1);
00173   memcpy(p, buf, len + 1);
00174   return p;
00175 }
00176 
00177 
00186 void str_validate(char *str, const char *last, bool allow_newlines, bool ignore)
00187 {
00188   /* Assume the ABSOLUTE WORST to be in str as it comes from the outside. */
00189 
00190   char *dst = str;
00191   while (str <= last && *str != '\0') {
00192     size_t len = Utf8EncodedCharLen(*str);
00193     /* If the character is unknown, i.e. encoded length is 0
00194      * we assume worst case for the length check.
00195      * The length check is needed to prevent Utf8Decode to read
00196      * over the terminating '\0' if that happens to be placed
00197      * within the encoding of an UTF8 character. */
00198     if ((len == 0 && str + 4 > last) || str + len > last) break;
00199 
00200     WChar c;
00201     len = Utf8Decode(&c, str);
00202     /* It's possible to encode the string termination character
00203      * into a multiple bytes. This prevents those termination
00204      * characters to be skipped */
00205     if (c == '\0') break;
00206 
00207     if (IsPrintable(c) && (c < SCC_SPRITE_START || c > SCC_SPRITE_END)) {
00208       /* Copy the character back. Even if dst is current the same as str
00209        * (i.e. no characters have been changed) this is quicker than
00210        * moving the pointers ahead by len */
00211       do {
00212         *dst++ = *str++;
00213       } while (--len != 0);
00214     } else if (allow_newlines && c == '\n') {
00215       *dst++ = *str++;
00216     } else {
00217       if (allow_newlines && c == '\r' && str[1] == '\n') {
00218         str += len;
00219         continue;
00220       }
00221       /* Replace the undesirable character with a question mark */
00222       str += len;
00223       if (!ignore) *dst++ = '?';
00224 
00225       /* In case of these two special cases assume that they really
00226        * mean SETX/SETXY and also "eat" the paramater. If this was
00227        * not the case the string was broken to begin with and this
00228        * would not break much more. */
00229       if (c == SCC_SETX) {
00230         str++;
00231       } else if (c == SCC_SETXY) {
00232         str += 2;
00233       }
00234     }
00235   }
00236 
00237   *dst = '\0';
00238 }
00239 
00247 bool StrValid(const char *str, const char *last)
00248 {
00249   /* Assume the ABSOLUTE WORST to be in str as it comes from the outside. */
00250 
00251   while (str <= last && *str != '\0') {
00252     size_t len = Utf8EncodedCharLen(*str);
00253     /* Encoded length is 0 if the character isn't known.
00254      * The length check is needed to prevent Utf8Decode to read
00255      * over the terminating '\0' if that happens to be placed
00256      * within the encoding of an UTF8 character. */
00257     if (len == 0 || str + len > last) return false;
00258 
00259     WChar c;
00260     len = Utf8Decode(&c, str);
00261     if (!IsPrintable(c) || (c >= SCC_SPRITE_START && c <= SCC_SPRITE_END)) {
00262       return false;
00263     }
00264 
00265     str += len;
00266   }
00267 
00268   return *str == '\0';
00269 }
00270 
00272 void str_strip_colours(char *str)
00273 {
00274   char *dst = str;
00275   WChar c;
00276   size_t len;
00277 
00278   for (len = Utf8Decode(&c, str); c != '\0'; len = Utf8Decode(&c, str)) {
00279     if (c < SCC_BLUE || c > SCC_BLACK) {
00280       /* Copy the character back. Even if dst is current the same as str
00281        * (i.e. no characters have been changed) this is quicker than
00282        * moving the pointers ahead by len */
00283       do {
00284         *dst++ = *str++;
00285       } while (--len != 0);
00286     } else {
00287       /* Just skip (strip) the colour codes */
00288       str += len;
00289     }
00290   }
00291   *dst = '\0';
00292 }
00293 
00300 size_t Utf8StringLength(const char *s)
00301 {
00302   size_t len = 0;
00303   const char *t = s;
00304   while (Utf8Consume(&t) != 0) len++;
00305   return len;
00306 }
00307 
00308 
00319 void strtolower(char *str)
00320 {
00321   for (; *str != '\0'; str++) *str = tolower(*str);
00322 }
00323 
00331 bool IsValidChar(WChar key, CharSetFilter afilter)
00332 {
00333   switch (afilter) {
00334     case CS_ALPHANUMERAL:  return IsPrintable(key);
00335     case CS_NUMERAL:       return (key >= '0' && key <= '9');
00336     case CS_NUMERAL_SPACE: return (key >= '0' && key <= '9') || key == ' ';
00337     case CS_ALPHA:         return IsPrintable(key) && !(key >= '0' && key <= '9');
00338     case CS_HEXADECIMAL:   return (key >= '0' && key <= '9') || (key >= 'a' && key <= 'f') || (key >= 'A' && key <= 'F');
00339   }
00340 
00341   return false;
00342 }
00343 
00344 #ifdef WIN32
00345 /* Since version 3.14, MinGW Runtime has snprintf() and vsnprintf() conform to C99 but it's not the case for older versions */
00346 #if (__MINGW32_MAJOR_VERSION < 3) || ((__MINGW32_MAJOR_VERSION == 3) && (__MINGW32_MINOR_VERSION < 14))
00347 int CDECL snprintf(char *str, size_t size, const char *format, ...)
00348 {
00349   va_list ap;
00350   int ret;
00351 
00352   va_start(ap, format);
00353   ret = vsnprintf(str, size, format, ap);
00354   va_end(ap);
00355   return ret;
00356 }
00357 #endif /* MinGW Runtime < 3.14 */
00358 
00359 #ifdef _MSC_VER
00360 
00367 int CDECL vsnprintf(char *str, size_t size, const char *format, va_list ap)
00368 {
00369   if (size == 0) return 0;
00370 
00371   errno = 0;
00372   int ret = _vsnprintf(str, size, format, ap);
00373 
00374   if (ret < 0) {
00375     if (errno != ERANGE) {
00376       /* There's a formatting error, better get that looked
00377        * at properly instead of ignoring it. */
00378       NOT_REACHED();
00379     }
00380   } else if ((size_t)ret < size) {
00381     /* The buffer is big enough for the number of
00382      * characers stored (excluding null), i.e.
00383      * the string has been null-terminated. */
00384     return ret;
00385   }
00386 
00387   /* The buffer is too small for _vsnprintf to write the
00388    * null-terminator at its end and return size. */
00389   str[size - 1] = '\0';
00390   return (int)size;
00391 }
00392 #endif /* _MSC_VER */
00393 
00394 #endif /* WIN32 */
00395 
00405 int CDECL seprintf(char *str, const char *last, const char *format, ...)
00406 {
00407   va_list ap;
00408 
00409   va_start(ap, format);
00410   int ret = vseprintf(str, last, format, ap);
00411   va_end(ap);
00412   return ret;
00413 }
00414 
00415 
00423 char *md5sumToString(char *buf, const char *last, const uint8 md5sum[16])
00424 {
00425   char *p = buf;
00426 
00427   for (uint i = 0; i < 16; i++) {
00428     p += seprintf(p, last, "%02X", md5sum[i]);
00429   }
00430 
00431   return p;
00432 }
00433 
00434 
00435 /* UTF-8 handling routines */
00436 
00437 
00444 size_t Utf8Decode(WChar *c, const char *s)
00445 {
00446   assert(c != NULL);
00447 
00448   if (!HasBit(s[0], 7)) {
00449     /* Single byte character: 0xxxxxxx */
00450     *c = s[0];
00451     return 1;
00452   } else if (GB(s[0], 5, 3) == 6) {
00453     if (IsUtf8Part(s[1])) {
00454       /* Double byte character: 110xxxxx 10xxxxxx */
00455       *c = GB(s[0], 0, 5) << 6 | GB(s[1], 0, 6);
00456       if (*c >= 0x80) return 2;
00457     }
00458   } else if (GB(s[0], 4, 4) == 14) {
00459     if (IsUtf8Part(s[1]) && IsUtf8Part(s[2])) {
00460       /* Triple byte character: 1110xxxx 10xxxxxx 10xxxxxx */
00461       *c = GB(s[0], 0, 4) << 12 | GB(s[1], 0, 6) << 6 | GB(s[2], 0, 6);
00462       if (*c >= 0x800) return 3;
00463     }
00464   } else if (GB(s[0], 3, 5) == 30) {
00465     if (IsUtf8Part(s[1]) && IsUtf8Part(s[2]) && IsUtf8Part(s[3])) {
00466       /* 4 byte character: 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */
00467       *c = GB(s[0], 0, 3) << 18 | GB(s[1], 0, 6) << 12 | GB(s[2], 0, 6) << 6 | GB(s[3], 0, 6);
00468       if (*c >= 0x10000 && *c <= 0x10FFFF) return 4;
00469     }
00470   }
00471 
00472   /* DEBUG(misc, 1, "[utf8] invalid UTF-8 sequence"); */
00473   *c = '?';
00474   return 1;
00475 }
00476 
00477 
00484 size_t Utf8Encode(char *buf, WChar c)
00485 {
00486   if (c < 0x80) {
00487     *buf = c;
00488     return 1;
00489   } else if (c < 0x800) {
00490     *buf++ = 0xC0 + GB(c,  6, 5);
00491     *buf   = 0x80 + GB(c,  0, 6);
00492     return 2;
00493   } else if (c < 0x10000) {
00494     *buf++ = 0xE0 + GB(c, 12, 4);
00495     *buf++ = 0x80 + GB(c,  6, 6);
00496     *buf   = 0x80 + GB(c,  0, 6);
00497     return 3;
00498   } else if (c < 0x110000) {
00499     *buf++ = 0xF0 + GB(c, 18, 3);
00500     *buf++ = 0x80 + GB(c, 12, 6);
00501     *buf++ = 0x80 + GB(c,  6, 6);
00502     *buf   = 0x80 + GB(c,  0, 6);
00503     return 4;
00504   }
00505 
00506   /* DEBUG(misc, 1, "[utf8] can't UTF-8 encode value 0x%X", c); */
00507   *buf = '?';
00508   return 1;
00509 }
00510 
00518 size_t Utf8TrimString(char *s, size_t maxlen)
00519 {
00520   size_t length = 0;
00521 
00522   for (const char *ptr = strchr(s, '\0'); *s != '\0';) {
00523     size_t len = Utf8EncodedCharLen(*s);
00524     /* Silently ignore invalid UTF8 sequences, our only concern trimming */
00525     if (len == 0) len = 1;
00526 
00527     /* Take care when a hard cutoff was made for the string and
00528      * the last UTF8 sequence is invalid */
00529     if (length + len >= maxlen || (s + len > ptr)) break;
00530     s += len;
00531     length += len;
00532   }
00533 
00534   *s = '\0';
00535   return length;
00536 }
00537 
00538 #ifdef DEFINE_STRNDUP
00539 #include "core/math_func.hpp"
00540 char *strndup(const char *s, size_t len)
00541 {
00542   len = min(strlen(s), len);
00543   char *tmp = CallocT<char>(len + 1);
00544   memcpy(tmp, s, len);
00545   return tmp;
00546 }
00547 #endif /* DEFINE_STRNDUP */
00548 
00549 #ifdef DEFINE_STRCASESTR
00550 char *strcasestr(const char *haystack, const char *needle)
00551 {
00552   size_t hay_len = strlen(haystack);
00553   size_t needle_len = strlen(needle);
00554   while (hay_len >= needle_len) {
00555     if (strncasecmp(haystack, needle, needle_len) == 0) return const_cast<char *>(haystack);
00556 
00557     haystack++;
00558     hay_len--;
00559   }
00560 
00561   return NULL;
00562 }
00563 #endif /* DEFINE_STRCASESTR */
00564 
00572 int strnatcmp(const char *s1, const char *s2)
00573 {
00574 #ifdef WITH_ICU
00575   if (_current_collator != NULL) {
00576     UErrorCode status = U_ZERO_ERROR;
00577     int result;
00578 
00579     /* We want to use the new faster method for ICU 4.2 and higher. */
00580 #if U_ICU_VERSION_MAJOR_NUM > 4 || (U_ICU_VERSION_MAJOR_NUM == 4 && U_ICU_VERSION_MINOR_NUM >= 2)
00581     /* The StringPiece parameter gets implicitly constructed from the char *. */
00582     result = _current_collator->compareUTF8(s1, s2, status);
00583 #else /* The following for 4.0 and lower. */
00584     UChar buffer1[DRAW_STRING_BUFFER];
00585     u_strFromUTF8Lenient(buffer1, lengthof(buffer1), NULL, s1, -1, &status);
00586     UChar buffer2[DRAW_STRING_BUFFER];
00587     u_strFromUTF8Lenient(buffer2, lengthof(buffer2), NULL, s2, -1, &status);
00588 
00589     result = _current_collator->compare(buffer1, buffer2, status);
00590 #endif /* ICU version check. */
00591     if (U_SUCCESS(status)) return result;
00592   }
00593 
00594 #endif /* WITH_ICU */
00595 
00596   /* Do a normal comparison if ICU is missing or if we cannot create a collator. */
00597   return strcasecmp(s1, s2);
00598 }