/* * Copyright (c) 2020 - 2024 the ThorVG project. All rights reserved. * Permission is hereby granted, free of charge, to any person obtaining a copy * of this software and associated documentation files (the "Software"), to deal * in the Software without restriction, including without limitation the rights * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell * copies of the Software, and to permit persons to whom the Software is * furnished to do so, subject to the following conditions: * The above copyright notice and this permission notice shall be included in all * copies or substantial portions of the Software. * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. */ #include "config.h" #include #include #include #include "tvgMath.h" #include "tvgStr.h" /************************************************************************/ /* Internal Class Implementation */ /************************************************************************/ static inline bool _floatExact(float a, float b) { return memcmp(&a, &b, sizeof(float)) == 0; } /************************************************************************/ /* External Class Implementation */ /************************************************************************/ namespace tvg { /* * https://docs.microsoft.com/en-us/cpp/c-runtime-library/reference/strtof-strtof-l-wcstof-wcstof-l?view=msvc-160 * * src should be one of the following form : * * [whitespace] [sign] {digits [radix digits] | radix digits} [{e | E} [sign] digits] * [whitespace] [sign] {INF | INFINITY} * [whitespace] [sign] NAN [sequence] * * No hexadecimal form supported * no sequence supported after NAN */ float strToFloat(const char *nPtr, char **endPtr) { if (endPtr) *endPtr = (char *) (nPtr); if (!nPtr) return 0.0f; auto a = nPtr; auto iter = nPtr; auto val = 0.0f; unsigned long long integerPart = 0; int minus = 1; //ignore leading whitespaces while (isspace(*iter)) iter++; //signed or not if (*iter == '-') { minus = -1; iter++; } else if (*iter == '+') { iter++; } if (tolower(*iter) == 'i') { if ((tolower(*(iter + 1)) == 'n') && (tolower(*(iter + 2)) == 'f')) iter += 3; else goto error; if (tolower(*(iter)) == 'i') { if ((tolower(*(iter + 1)) == 'n') && (tolower(*(iter + 2)) == 'i') && (tolower(*(iter + 3)) == 't') && (tolower(*(iter + 4)) == 'y')) iter += 5; else goto error; } if (endPtr) *endPtr = (char *) (iter); return (minus == -1) ? -INFINITY : INFINITY; } if (tolower(*iter) == 'n') { if ((tolower(*(iter + 1)) == 'a') && (tolower(*(iter + 2)) == 'n')) iter += 3; else goto error; if (endPtr) *endPtr = (char *) (iter); return (minus == -1) ? -NAN : NAN; } //Optional: integer part before dot if (isdigit(*iter)) { for (; isdigit(*iter); iter++) { integerPart = integerPart * 10ULL + (unsigned long long) (*iter - '0'); } a = iter; } else if (*iter != '.') { goto success; } val = static_cast(integerPart); //Optional: decimal part after dot if (*iter == '.') { unsigned long long decimalPart = 0; unsigned long long pow10 = 1; int count = 0; iter++; if (isdigit(*iter)) { for (; isdigit(*iter); iter++, count++) { if (count < 19) { decimalPart = decimalPart * 10ULL + +static_cast(*iter - '0'); pow10 *= 10ULL; } } } else if (isspace(*iter)) { //skip if there is a space after the dot. a = iter; goto success; } val += static_cast(decimalPart) / static_cast(pow10); a = iter; } //Optional: exponent if (*iter == 'e' || *iter == 'E') { ++iter; //Exception: svg may have 'em' unit for fonts. ex) 5em, 10.5em if ((*iter == 'm') || (*iter == 'M')) { //TODO: We don't support font em unit now, but has to multiply val * font size later... a = iter + 1; goto success; } //signed or not int minus_e = 1; if (*iter == '-') { minus_e = -1; ++iter; } else if (*iter == '+') { iter++; } unsigned int exponentPart = 0; if (isdigit(*iter)) { while (*iter == '0') iter++; for (; isdigit(*iter); iter++) { exponentPart = exponentPart * 10U + static_cast(*iter - '0'); } } else if (!isdigit(*(a - 1))) { a = nPtr; goto success; } else if (*iter == 0) { goto success; } //if ((_floatExact(val, 2.2250738585072011f)) && ((minus_e * static_cast(exponentPart)) <= -308)) { if ((_floatExact(val, 1.175494351f)) && ((minus_e * static_cast(exponentPart)) <= -38)) { //val *= 1.0e-308f; val *= 1.0e-38f; a = iter; goto success; } a = iter; auto scale = 1.0f; while (exponentPart >= 8U) { scale *= 1E8; exponentPart -= 8U; } while (exponentPart > 0U) { scale *= 10.0f; exponentPart--; } val = (minus_e == -1) ? (val / scale) : (val * scale); } else if ((iter > nPtr) && !isdigit(*(iter - 1))) { a = nPtr; goto success; } success: if (endPtr) *endPtr = (char *)(a); if (!std::isfinite(val)) return 0.0f; return minus * val; error: if (endPtr) *endPtr = (char *)(nPtr); return 0.0f; } int str2int(const char* str, size_t n) { int ret = 0; for(size_t i = 0; i < n; ++i) { ret = ret * 10 + (str[i] - '0'); } return ret; } char* strDuplicate(const char *str, size_t n) { auto len = strlen(str); if (len < n) n = len; auto ret = (char *) malloc(n + 1); if (!ret) return nullptr; ret[n] = '\0'; return (char *) memcpy(ret, str, n); } char* strDirname(const char* path) { const char *ptr = strrchr(path, '/'); #ifdef _WIN32 if (ptr) ptr = strrchr(ptr + 1, '\\'); #endif int len = int(ptr + 1 - path); // +1 to include '/' return strDuplicate(path, len); } }