/* ScummVM - Graphic Adventure Engine * * ScummVM is the legal property of its developers, whose names * are too numerous to list here. Please refer to the COPYRIGHT * file distributed with this source distribution. * * This program 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; either version 2 * of the License, or (at your option) any later version. * This program 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. See the * GNU General Public License for more details. * You should have received a copy of the GNU General Public License * along with this program; if not, write to the Free Software * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. * */ /* * This file is based on WME Lite. * http://dead-code.org/redir.php?target=wmelite * Copyright (c) 2011 Jan Nedoma */ #include "common/tokenizer.h" #include "engines/wintermute/utils/string_util.h" #include "engines/wintermute/utils/convert_utf.h" namespace Wintermute { ////////////////////////////////////////////////////////////////////////// bool StringUtil::compareNoCase(const AnsiString &str1, const AnsiString &str2) { return (str1.compareToIgnoreCase(str2) == 0); } ////////////////////////////////////////////////////////////////////////// /*bool StringUtil::CompareNoCase(const WideString &str1, const WideString &str2) { WideString str1lc = str1; WideString str2lc = str2; ToLowerCase(str1lc); ToLowerCase(str2lc); return (str1lc == str2lc); }*/ ////////////////////////////////////////////////////////////////////////// WideString StringUtil::utf8ToWide(const Utf8String &Utf8Str) { error("StringUtil::Utf8ToWide - WideString not supported yet"); /* size_t WideSize = Utf8Str.size(); if (sizeof(wchar_t) == 2) { wchar_t *WideStringNative = new wchar_t[WideSize + 1]; const UTF8 *SourceStart = reinterpret_cast(Utf8Str.c_str()); const UTF8 *SourceEnd = SourceStart + WideSize; UTF16 *TargetStart = reinterpret_cast(WideStringNative); UTF16 *TargetEnd = TargetStart + WideSize + 1; ConversionResult res = ConvertUTF8toUTF16(&SourceStart, SourceEnd, &TargetStart, TargetEnd, strictConversion); if (res != conversionOK) { delete[] WideStringNative; return L""; } *TargetStart = 0; WideString ResultString(WideStringNative); delete[] WideStringNative; return ResultString; } else if (sizeof(wchar_t) == 4) { wchar_t *WideStringNative = new wchar_t[WideSize + 1]; const UTF8 *SourceStart = reinterpret_cast(Utf8Str.c_str()); const UTF8 *SourceEnd = SourceStart + WideSize; UTF32 *TargetStart = reinterpret_cast(WideStringNative); UTF32 *TargetEnd = TargetStart + WideSize; ConversionResult res = ConvertUTF8toUTF32(&SourceStart, SourceEnd, &TargetStart, TargetEnd, strictConversion); if (res != conversionOK) { delete[] WideStringNative; return L""; } *TargetStart = 0; WideString ResultString(WideStringNative); delete[] WideStringNative; return ResultString; } else { return L""; }*/ return ""; } ////////////////////////////////////////////////////////////////////////// Utf8String StringUtil::wideToUtf8(const WideString &WideStr) { error("StringUtil::wideToUtf8 - Widestring not supported yet"); /* size_t WideSize = WideStr.length(); if (sizeof(wchar_t) == 2) { size_t utf8Size = 3 * WideSize + 1; char *utf8StringNative = new char[Utf8Size]; const UTF16 *SourceStart = reinterpret_cast(WideStr.c_str()); const UTF16 *SourceEnd = SourceStart + WideSize; UTF8 *TargetStart = reinterpret_cast(Utf8StringNative); UTF8 *TargetEnd = TargetStart + Utf8Size; ConversionResult res = ConvertUTF16toUTF8(&SourceStart, SourceEnd, &TargetStart, TargetEnd, strictConversion); if (res != conversionOK) { delete[] Utf8StringNative; return (Utf8String)""; } *TargetStart = 0; Utf8String ResultString(Utf8StringNative); delete[] Utf8StringNative; return ResultString; } else if (sizeof(wchar_t) == 4) { size_t utf8Size = 4 * WideSize + 1; char *utf8StringNative = new char[Utf8Size]; const UTF32 *SourceStart = reinterpret_cast(WideStr.c_str()); const UTF32 *SourceEnd = SourceStart + WideSize; UTF8 *TargetStart = reinterpret_cast(Utf8StringNative); UTF8 *TargetEnd = TargetStart + Utf8Size; ConversionResult res = ConvertUTF32toUTF8(&SourceStart, SourceEnd, &TargetStart, TargetEnd, strictConversion); if (res != conversionOK) { delete[] Utf8StringNative; return (Utf8String)""; } *TargetStart = 0; Utf8String ResultString(Utf8StringNative); delete[] Utf8StringNative; return ResultString; } else { return (Utf8String)""; }*/ return ""; } ////////////////////////////////////////////////////////////////////////// WideString StringUtil::ansiToWide(const AnsiString &str) { // TODO: This function gets called a lot, so warnings like these drown out the usefull information Common::String converted = ""; uint32 index = 0; while (index != str.size()) { byte c = str[index]; if (c == 146) { converted += (char)39; // Replace right-quote with apostrophe } else if (c == 133) { converted += Common::String("..."); // Replace ...-symbol with ... } else { converted += c; } index++; } // using default os locale! /* setlocale(LC_CTYPE, ""); size_t wideSize = mbstowcs(NULL, str.c_str(), 0) + 1; wchar_t *wstr = new wchar_t[WideSize]; mbstowcs(wstr, str.c_str(), WideSize); WideString ResultString(wstr); delete[] wstr; return ResultString;*/ return WideString(converted); } ////////////////////////////////////////////////////////////////////////// AnsiString StringUtil::wideToAnsi(const WideString &wstr) { // using default os locale! // TODO: This function gets called a lot, so warnings like these drown out the usefull information /* setlocale(LC_CTYPE, ""); size_t wideSize = wcstombs(NULL, wstr.c_str(), 0) + 1; char *str = new char[WideSize]; wcstombs(str, wstr.c_str(), WideSize); AnsiString ResultString(str); delete[] str; return ResultString;*/ return AnsiString(wstr); } ////////////////////////////////////////////////////////////////////////// bool StringUtil::isUtf8BOM(const byte *buffer, uint32 bufferSize) { if (bufferSize > 3 && buffer[0] == 0xEF && buffer[1] == 0xBB && buffer[2] == 0xBF) { return true; } else { return false; } } ////////////////////////////////////////////////////////////////////////// int StringUtil::indexOf(const WideString &str, const WideString &toFind, size_t startFrom) { const char *index = strstr(str.c_str(), toFind.c_str()); if (index == nullptr) { return -1; } else { return index - str.c_str(); } } Common::String StringUtil::encodeSetting(const Common::String &str) { for (uint32 i = 0; i < str.size(); i++) { if ((str[i] < 33) || (str[i] == '=') || (str[i] > 126)) { error("Setting contains illegal characters: %s", str.c_str()); } } return str; } Common::String StringUtil::decodeSetting(const Common::String &str) { return str; } ////////////////////////////////////////////////////////////////////////// AnsiString StringUtil::toString(int val) { return Common::String::format("%d", val); } } // End of namespace Wintermute