/* Copyright (C) 2002-2005 RealVNC Ltd. All Rights Reserved. * Copyright 2011-2019 Pierre Ossman for Cendio AB * * This 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 software 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 software; if not, write to the Free Software * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, * USA. */ #ifdef HAVE_CONFIG_H #include #endif #include #include #include #include namespace rfb { void CharArray::format(const char *fmt, ...) { va_list ap; int len; va_start(ap, fmt); len = vsnprintf(NULL, 0, fmt, ap); va_end(ap); delete [] buf; if (len < 0) { buf = new char[1]; buf[0] = '\0'; return; } buf = new char[len+1]; va_start(ap, fmt); vsnprintf(buf, len+1, fmt, ap); va_end(ap); } char* strDup(const char* s) { if (!s) return 0; int l = strlen(s); char* r = new char[l+1]; memcpy(r, s, l+1); return r; }; void strFree(char* s) { delete [] s; } void strFree(wchar_t* s) { delete [] s; } bool strSplit(const char* src, const char limiter, char** out1, char** out2, bool fromEnd) { CharArray out1old, out2old; if (out1) out1old.buf = *out1; if (out2) out2old.buf = *out2; int len = strlen(src); int i=0, increment=1, limit=len; if (fromEnd) { i=len-1; increment = -1; limit = -1; } while (i!=limit) { if (src[i] == limiter) { if (out1) { *out1 = new char[i+1]; if (i) memcpy(*out1, src, i); (*out1)[i] = 0; } if (out2) { *out2 = new char[len-i]; if (len-i-1) memcpy(*out2, &src[i+1], len-i-1); (*out2)[len-i-1] = 0; } return true; } i+=increment; } if (out1) *out1 = strDup(src); if (out2) *out2 = 0; return false; } bool strContains(const char* src, char c) { int l=strlen(src); for (int i=0; i 0) && (*in != '\0')) { if (*in != '\r') { sz++; in++; in_len--; continue; } if ((in_len < 2) || (*(in+1) != '\n')) sz++; in++; in_len--; } // Alloc buffer = new char[sz]; memset(buffer, 0, sz); // And convert out = buffer; in = src; in_len = bytes; while ((in_len > 0) && (*in != '\0')) { if (*in != '\r') { *out++ = *in++; in_len--; continue; } if ((in_len < 2) || (*(in+1) != '\n')) *out++ = '\n'; in++; in_len--; } return buffer; } char* convertCRLF(const char* src, size_t bytes) { char* buffer; size_t sz; char* out; const char* in; size_t in_len; // Always include space for a NULL sz = 1; // Compute output size in = src; in_len = bytes; while ((in_len > 0) && (*in != '\0')) { sz++; if (*in == '\r') { if ((in_len < 2) || (*(in+1) != '\n')) sz++; } else if (*in == '\n') { if ((in == src) || (*(in-1) != '\r')) sz++; } in++; in_len--; } // Alloc buffer = new char[sz]; memset(buffer, 0, sz); // And convert out = buffer; in = src; in_len = bytes; while ((in_len > 0) && (*in != '\0')) { if (*in == '\n') { if ((in == src) || (*(in-1) != '\r')) *out++ = '\r'; } *out = *in; if (*in == '\r') { if ((in_len < 2) || (*(in+1) != '\n')) { out++; *out = '\n'; } } out++; in++; in_len--; } return buffer; } size_t ucs4ToUTF8(unsigned src, char* dst) { if (src < 0x80) { *dst++ = src; *dst++ = '\0'; return 1; } else if (src < 0x800) { *dst++ = 0xc0 | (src >> 6); *dst++ = 0x80 | (src & 0x3f); *dst++ = '\0'; return 2; } else if (src < 0x10000) { *dst++ = 0xe0 | (src >> 12); *dst++ = 0x80 | ((src >> 6) & 0x3f); *dst++ = 0x80 | (src & 0x3f); *dst++ = '\0'; return 3; } else if (src < 0x110000) { *dst++ = 0xf0 | (src >> 18); *dst++ = 0x80 | ((src >> 12) & 0x3f); *dst++ = 0x80 | ((src >> 6) & 0x3f); *dst++ = 0x80 | (src & 0x3f); *dst++ = '\0'; return 4; } else { return ucs4ToUTF8(0xfffd, dst); } } size_t utf8ToUCS4(const char* src, size_t max, unsigned* dst) { size_t count, consumed; *dst = 0xfffd; if (max == 0) return 0; consumed = 1; if ((*src & 0x80) == 0) { *dst = *src; count = 0; } else if ((*src & 0xe0) == 0xc0) { *dst = *src & 0x1f; count = 1; } else if ((*src & 0xf0) == 0xe0) { *dst = *src & 0x0f; count = 2; } else if ((*src & 0xf8) == 0xf0) { *dst = *src & 0x07; count = 3; } else { // Invalid sequence, consume all continuation characters src++; max--; while ((max-- > 0) && ((*src++ & 0xc0) == 0x80)) consumed++; return consumed; } src++; max--; while (count--) { consumed++; // Invalid or truncated sequence? if ((max == 0) || ((*src & 0xc0) != 0x80)) { *dst = 0xfffd; return consumed; } *dst <<= 6; *dst |= *src & 0x3f; src++; max--; } return consumed; } size_t ucs4ToUTF16(unsigned src, wchar_t* dst) { if ((src < 0xd800) || ((src >= 0xe000) && (src < 0x10000))) { *dst++ = src; *dst++ = L'\0'; return 1; } else if ((src >= 0x10000) && (src < 0x110000)) { src -= 0x10000; *dst++ = 0xd800 | ((src >> 10) & 0x03ff); *dst++ = 0xdc00 | (src & 0x03ff); *dst++ = L'\0'; return 2; } else { return ucs4ToUTF16(0xfffd, dst); } } size_t utf16ToUCS4(const wchar_t* src, size_t max, unsigned* dst) { *dst = 0xfffd; if (max == 0) return 0; if ((*src < 0xd800) || (*src >= 0xe000)) { *dst = *src; return 1; } if (*src & 0x0400) { size_t consumed; // Invalid sequence, consume all continuation characters consumed = 0; while ((max > 0) && (*src & 0x0400)) { src++; max--; consumed++; } return consumed; } *dst = *src++; max--; // Invalid or truncated sequence? if ((max == 0) || ((*src & 0xfc00) != 0xdc00)) { *dst = 0xfffd; return 1; } *dst = 0x10000 + ((*dst & 0x03ff) << 10); *dst |= *src & 0x3ff; return 2; } char* latin1ToUTF8(const char* src, size_t bytes) { char* buffer; size_t sz; char* out; const char* in; size_t in_len; // Always include space for a NULL sz = 1; // Compute output size in = src; in_len = bytes; while ((in_len > 0) && (*in != '\0')) { char buf[5]; sz += ucs4ToUTF8(*(const unsigned char*)in, buf); in++; in_len--; } // Alloc buffer = new char[sz]; memset(buffer, 0, sz); // And convert out = buffer; in = src; in_len = bytes; while ((in_len > 0) && (*in != '\0')) { out += ucs4ToUTF8(*(const unsigned char*)in, out); in++; in_len--; } return buffer; } char* utf8ToLatin1(const char* src, size_t bytes) { char* buffer; size_t sz; char* out; const char* in; size_t in_len; // Always include space for a NULL sz = 1; // Compute output size in = src; in_len = bytes; while ((in_len > 0) && (*in != '\0')) { size_t len; unsigned ucs; len = utf8ToUCS4(in, in_len, &ucs); in += len; in_len -= len; sz++; } // Alloc buffer = new char[sz]; memset(buffer, 0, sz); // And convert out = buffer; in = src; in_len = bytes; while ((in_len > 0) && (*in != '\0')) { size_t len; unsigned ucs; len = utf8ToUCS4(in, in_len, &ucs); in += len; in_len -= len; if (ucs > 0xff) *out++ = '?'; else *out++ = (unsigned char)ucs; } return buffer; } char* utf16ToUTF8(const wchar_t* src, size_t units) { char* buffer; size_t sz; char* out; const wchar_t* in; size_t in_len; // Always include space for a NULL sz = 1; // Compute output size in = src; in_len = units; while ((in_len > 0) && (*in != '\0')) { size_t len; unsigned ucs; char buf[5]; len = utf16ToUCS4(in, in_len, &ucs); in += len; in_len -= len; sz += ucs4ToUTF8(ucs, buf); } // Alloc buffer = new char[sz]; memset(buffer, 0, sz); // And convert out = buffer; in = src; in_len = units; while ((in_len > 0) && (*in != '\0')) { size_t len; unsigned ucs; len = utf16ToUCS4(in, in_len, &ucs); in += len; in_len -= len; out += ucs4ToUTF8(ucs, out); } return buffer; } wchar_t* utf8ToUTF16(const char* src, size_t bytes) { wchar_t* buffer; size_t sz; wchar_t* out; const char* in; size_t in_len; // Always include space for a NULL sz = 1; // Compute output size in = src; in_len = bytes; while ((in_len > 0) && (*in != '\0')) { size_t len; unsigned ucs; wchar_t buf[3]; len = utf8ToUCS4(in, in_len, &ucs); in += len; in_len -= len; sz += ucs4ToUTF16(ucs, buf); } // Alloc buffer = new wchar_t[sz]; memset(buffer, 0, sz * sizeof(wchar_t)); // And convert out = buffer; in = src; in_len = bytes; while ((in_len > 0) && (*in != '\0')) { size_t len; unsigned ucs; len = utf8ToUCS4(in, in_len, &ucs); in += len; in_len -= len; out += ucs4ToUTF16(ucs, out); } return buffer; } unsigned msBetween(const struct timeval *first, const struct timeval *second) { unsigned diff; diff = (second->tv_sec - first->tv_sec) * 1000; diff += second->tv_usec / 1000; diff -= first->tv_usec / 1000; return diff; } unsigned msSince(const struct timeval *then) { struct timeval now; gettimeofday(&now, NULL); return msBetween(then, &now); } bool isBefore(const struct timeval *first, const struct timeval *second) { if (first->tv_sec < second->tv_sec) return true; if (first->tv_sec > second->tv_sec) return false; if (first->tv_usec < second->tv_usec) return true; return false; } static size_t doPrefix(long long value, const char *unit, char *buffer, size_t maxlen, unsigned divisor, const char **prefixes, size_t prefixCount, int precision) { double newValue; size_t prefix, len; newValue = value; prefix = 0; while (newValue >= divisor) { if (prefix >= prefixCount) break; newValue /= divisor; prefix++; } len = snprintf(buffer, maxlen, "%.*g %s%s", precision, newValue, (prefix == 0) ? "" : prefixes[prefix-1], unit); buffer[maxlen-1] = '\0'; return len; } static const char *siPrefixes[] = { "k", "M", "G", "T", "P", "E", "Z", "Y" }; static const char *iecPrefixes[] = { "Ki", "Mi", "Gi", "Ti", "Pi", "Ei", "Zi", "Yi" }; size_t siPrefix(long long value, const char *unit, char *buffer, size_t maxlen, int precision) { return doPrefix(value, unit, buffer, maxlen, 1000, siPrefixes, sizeof(siPrefixes)/sizeof(*siPrefixes), precision); } size_t iecPrefix(long long value, const char *unit, char *buffer, size_t maxlen, int precision) { return doPrefix(value, unit, buffer, maxlen, 1024, iecPrefixes, sizeof(iecPrefixes)/sizeof(*iecPrefixes), precision); } };