ncbistr.cpp

ncbi源码

                        "Wrong UTF8 format",
                        s_DiffPtr(srcBuf,c_str()));
        }
    }
    result.reserve( needed+1);
    for (srcBuf = c_str(); *srcBuf; ++srcBuf) {
        Uint1 chRes;
        size_t more;
        Uint1 ch = *srcBuf;
        if ((ch & 0x80) == 0) {
            chRes = ch;
            more = 0;
        } else {
            chRes = (ch & 0x1F);
            more = 1;
        }
        while (more--) {
            ch = *(++srcBuf);
            chRes = (chRes << 6) | (ch & 0x3F);
        }
        result += chRes;
    }
    return result;
}


#if defined(HAVE_WSTRING)
wstring CStringUTF8::AsUnicode(void) const
{
    wstring result;
    const char* srcBuf;
    size_t needed = 0;
    bool bad = false;
    for (srcBuf = c_str(); *srcBuf; ++srcBuf) {
        Uint1 ch = *srcBuf;
        if ((ch & 0x80) == 0) {
            ++needed;
        } else if ((ch & 0xE0) == 0xC0) {
            ++needed;
            ch = *(++srcBuf);
            bad = (ch & 0xC0) != 0x80;
        } else if ((ch & 0xF0) == 0xE0) {
            ++needed;
            ch = *(++srcBuf);
            bad = (ch & 0xC0) != 0x80;
            if (!bad) {
                ch = *(++srcBuf);
                bad = (ch & 0xC0) != 0x80;
            }
        } else {
            bad = true;
        }
        if (bad) {
            NCBI_THROW2(CStringException, eFormat,
                        "Wrong UTF8 format",
                        s_DiffPtr(srcBuf,c_str()));
        }
    }
    result.reserve( needed+1);
    for (srcBuf = c_str(); *srcBuf; ++srcBuf) {
        Uint2 chRes;
        size_t more;
        Uint1 ch = *srcBuf;
        if ((ch & 0x80) == 0) {
            chRes = ch;
            more = 0;
        } else if ((ch & 0xE0) == 0xC0) {
            chRes = (ch & 0x1F);
            more = 1;
        } else {
            chRes = (ch & 0x0F);
            more = 2;
        }
        while (more--) {
            ch = *(++srcBuf);
            chRes = (chRes << 6) | (ch & 0x3F);
        }
        result += chRes;
    }
    return result;
}
#endif // HAVE_WSTRING


END_NCBI_SCOPE


/*
 * ===========================================================================
 * $Log: ncbistr.cpp,v $
 * Revision 1000.6  2004/06/01 19:09:21  gouriano
 * PRODUCTION: UPGRADED [GCC34_MSVC7] Dev-tree R1.109
 *
 * Revision 1.109  2004/05/26 19:21:25  ucko
 * FindNoCase: avoid looping in eLastMode when there aren't any full
 * matches but the first character of the string matches the first
 * character of the pattern.
 *
 * Revision 1.108  2004/05/14 13:59:27  gorelenk
 * Added include of ncbi_pch.hpp
 *
 * Revision 1.107  2004/03/11 18:49:48  gorelenk
 * Removed(condionaly) implementation of class CNcbiEmptyString.
 *
 * Revision 1.106  2004/03/05 12:26:43  ivanov
 * Moved CDirEntry::MatchesMask() to NStr class.
 *
 * Revision 1.105  2004/03/04 13:38:57  kuznets
 * + set of ToString conversion functions taking outout string as a parameter,
 * not a return value (should give a performance advantage in some cases)
 *
 * Revision 1.104  2004/02/19 16:44:55  vasilche
 * WorkShop compiler doesn't support static templates.
 *
 * Revision 1.103  2004/02/18 20:54:47  shomrat
 * bug fix (pos -> pos2)
 *
 * Revision 1.102  2003/12/12 20:06:44  rsmith
 * Take out un-needed include of stdarg.h (included in ncbistr.hpp).
 *
 * Revision 1.101  2003/12/12 20:04:24  rsmith
 * make sure stdarg.h is included to define va_list.
 *
 * Revision 1.100  2003/12/12 17:26:54  ucko
 * +FormatVarargs
 *
 * Revision 1.99  2003/12/01 20:45:47  ucko
 * Extend Join to handle vectors as well as lists (common code templatized).
 * Add ParseEscapes (inverse of PrintableString).
 *
 * Revision 1.98  2003/10/31 13:15:20  lavr
 * Fix typos in the log of the previous commit :-)
 *
 * Revision 1.97  2003/10/31 12:59:46  lavr
 * Better diagnostics messages from exceptions; some other cosmetic changes
 *
 * Revision 1.96  2003/10/03 15:16:02  ucko
 * NStr::Join: preallocate as much space as we need for result.
 *
 * Revision 1.95  2003/09/17 15:18:29  vasilche
 * Reduce memory allocations in NStr::PrintableString()
 *
 * Revision 1.94  2003/08/19 15:17:20  rsmith
 * Add NStr::SplitInTwo() function.
 *
 * Revision 1.93  2003/06/16 15:19:03  ucko
 * FindNoCase: always honor both start and end (oops).
 *
 * Revision 1.92  2003/05/22 20:09:29  gouriano
 * added UTF8 strings
 *
 * Revision 1.91  2003/05/14 21:52:09  ucko
 * Move FindNoCase out of line and reimplement it to avoid making
 * lowercase copies of both strings.
 *
 * Revision 1.90  2003/03/25 22:15:40  lavr
 * NStr::PrintableString():: Print NUL char as \x00 instead of \0
 *
 * Revision 1.89  2003/03/20 13:27:52  dicuccio
 * Oops.  Removed old code wrapped in #if 0...#endif.
 *
 * Revision 1.88  2003/03/20 13:27:11  dicuccio
 * Changed NStr::StringToPtr() - now symmetric with NSrt::PtrToString (there were
 * too many special cases).
 *
 * Revision 1.87  2003/03/17 12:49:26  dicuccio
 * Fixed indexing error in NStr::PtrToString() - buffer is 0-based index, not
 * 1-based
 *
 * Revision 1.86  2003/03/11 16:57:12  ucko
 * Process backspaces in NStr::Wrap, allowing in particular the use of
 * " \b" ("-\b") to indicate mid-word break (hyphenation) points.
 *
 * Revision 1.85  2003/03/10 18:57:08  kuznets
 * iterate->ITERATE
 *
 * Revision 1.84  2003/03/04 00:02:21  vakatov
 * NStr::PtrToString() -- use runtime check for "0x"
 * NStr::StringToPtr() -- minor polishing
 *
 * Revision 1.83  2003/02/27 15:34:01  lavr
 * Bugfix in converting string to double [spurious dots], some reformatting
 *
 * Revision 1.82  2003/02/26 21:07:52  siyan
 * Remove const for base parameter for StringToUInt8
 *
 * Revision 1.81  2003/02/26 20:34:11  siyan
 * Added/deleted whitespaces to conform to existing coding style
 *
 * Revision 1.80  2003/02/26 16:45:53  siyan
 * Reimplemented NStr::StringToUInt8 to support additional base parameters
 * that can take radix values such as 10(default), 16, 8, 2.
 * Reimplemented StringToPtr to support 64 bit addresses.
 *
 * Revision 1.79  2003/02/25 19:14:53  kuznets
 * NStr::StringToBool changed to understand YES/NO
 *
 * Revision 1.78  2003/02/25 15:43:40  dicuccio
 * Added #ifdef'd hack for MSVC's non-standard sprintf() in PtrToString(),
 * '%p' lacks a leading '0x'
 *
 * Revision 1.77  2003/02/25 14:43:53  dicuccio
 * Added handling of special NULL pointer encoding in StringToPtr()
 *
 * Revision 1.76  2003/02/24 20:25:59  gouriano
 * use template-based errno and parse exceptions
 *
 * Revision 1.75  2003/02/21 21:20:01  vakatov
 * Fixed some types, did some casts to avoid compilation warnings in 64-bit
 *
 * Revision 1.74  2003/02/20 18:41:28  dicuccio
 * Added NStr::StringToPtr()
 *
 * Revision 1.73  2003/02/11 22:11:03  ucko
 * Make NStr::WrapList a no-op if the input list is empty.
 *
 * Revision 1.72  2003/02/06 21:31:35  ucko
 * Fixed an off-by-one error in NStr::Wrap.
 *
 * Revision 1.71  2003/02/04 21:54:12  ucko
 * NStr::Wrap: when breaking on punctuation, try to position the break
 * *after* everything but opening delimiters.
 *
 * Revision 1.70  2003/01/31 03:39:11  lavr
 * Heed int->bool performance warning
 *
 * Revision 1.69  2003/01/27 20:06:59  ivanov
 * Get rid of compilation warnings in StringToUInt8() and DoubleToString()
 *
 * Revision 1.68  2003/01/24 16:59:27  ucko
 * Add an optional parameter to Split and Tokenize indicating whether to
 * merge adjacent delimiters; clean up WrapList slightly.
 *
 * Revision 1.67  2003/01/21 23:22:22  vakatov
 * NStr::Tokenize() to return reference, and not a new "vector<>".
 *
 * Revision 1.66  2003/01/21 20:08:01  ivanov
 * Added function NStr::DoubleToString(value, precision, buf, buf_size)
 *
 * Revision 1.65  2003/01/14 22:13:56  kuznets
 * Overflow check reimplemented for NStr::StringToInt
 *
 * Revision 1.64  2003/01/14 21:16:46  kuznets
 * +Nstr::Tokenize
 *
 * Revision 1.63  2003/01/13 14:47:16  kuznets
 * Implemented overflow checking for StringToInt8 function
 *
 * Revision 1.62  2003/01/10 22:17:06  kuznets
 * Implemented NStr::String2Int8
 *
 * Revision 1.61  2003/01/10 16:49:54  kuznets
 * Cosmetics
 *
 * Revision 1.60  2003/01/10 15:27:12  kuznets
 * Eliminated int -> bool performance warning
 *
 * Revision 1.59  2003/01/10 00:08:17  vakatov
 * + Int8ToString(),  UInt8ToString()
 *
 * Revision 1.58  2003/01/06 16:42:45  ivanov
 * + DoubleToString() with 'precision'
 *
 * Revision 1.57  2002/10/18 20:48:56  lavr
 * +ENewLine_Mode and '\n' translation in NStr::PrintableString()
 *
 * Revision 1.56  2002/10/17 14:41:20  ucko
 * * Make s_EndOf{Tag,Reference} actually static (oops).
 * * Pull width-determination code from WrapList into a separate function
 *   (s_VisibleWidth) and make Wrap{,List} call it for everything rather
 *   than assuming prefixes and delimiters to be plain text.
 * * Add a column variable to WrapList, as it may not equal s.size().
 *
 * Revision 1.55  2002/10/16 19:30:36  ucko
 * Add support for wrapping HTML <PRE> blocks.  (Not yet tested, but
 * behavior without fWrap_HTMLPre should stay the same.)
 *
 * Revision 1.54  2002/10/11 19:41:48  ucko
 * Clean up NStr::Wrap a bit more, doing away with the "limit" variables
 * for ease of potential extension.
 *
 * Revision 1.53  2002/10/03 14:44:35  ucko
 * Tweak the interfaces to NStr::Wrap* to avoid publicly depending on
 * kEmptyStr, removing the need for fWrap_UsePrefix1 in the process; also
 * drop fWrap_FavorPunct, as WrapList should be a better choice for such
 * situations.
 *
 * Revision 1.52  2002/10/02 20:15:09  ucko
 * Add Join, Wrap, and WrapList functions to NStr::.
 *
 * Revision 1.51  2002/09/04 15:16:57  lavr
 * Backslashed double quote (\") in PrintableString()
 *
 * Revision 1.50  2002/07/15 18:17:25  gouriano
 * renamed CNcbiException and its descendents
 *
 * Revision 1.49  2002/07/11 14:18:28  gouriano
 * exceptions replaced by CNcbiException-type ones
 *
 * Revision 1.48  2002/05/02 15:25:37  ivanov
 * Added new parameter to String-to-X functions for skipping the check
 * the end of string on zero
 *
 * Revision 1.47  2002/04/11 21:08:03  ivanov
 * CVS log moved to end of the file
 *
 * Revision 1.46  2002/02/22 17:50:52  ivanov
 * Added compatible compare functions strcmp, strncmp, strcasecmp, strncasecmp.
 * Was speed-up some Compare..() functions.
 *
 * Revision 1.45  2001/08/30 00:36:45  vakatov
 * + NStr::StringToNumeric()
 * Also, well-groomed the code and get rid of some compilation warnings.
 *
 * Revision 1.44  2001/05/30 15:56:25  vakatov
 * NStr::CompareNocase, NStr::CompareCase -- get rid of the possible
 * compilation warning (ICC compiler:  "return statement missing").
 *
 * Revision 1.43  2001/05/17 15:04:59  lavr
 * Typos corrected
 *
 * Revision 1.42  2001/04/12 21:39:44  vakatov
 * NStr::Replace() -- check against source and dest. strings being the same
 *
 * Revision 1.41  2001/04/11 20:15:29  vakatov
 * NStr::PrintableString() -- cast "char" to "unsigned char".
 *
 * Revision 1.40  2001/03/16 19:38:35  grichenk
 * Added NStr::Split()
 *
 * Revision 1.39  2001/01/03 17:45:35  vakatov
 * + <ncbi_limits.h>
 *
 * Revision 1.38  2000/12/15 15:36:41  vasilche
 * Added header corelib/ncbistr.hpp for all string utility functions.
 * Optimized string utility functions.
 * Added assignment operator to CRef<> and CConstRef<>.
 * Add Upcase() and Locase() methods for automatic conversion.
 *
 * Revision 1.37  2000/12/12 14:20:36  vasilche
 * Added operator bool to CArgValue.
 * Various NStr::Compare() methods made faster.
 * Added class Upcase for printing strings to ostream with automatic conversion
 *
 * Revision 1.36  2000/12/11 20:42:50  vakatov
 * + NStr::PrintableString()
 *
 * Revision 1.35  2000/11/16 23:52:41  vakatov
 * Porting to Mac...
 *
 * Revision 1.34  2000/11/07 04:06:08  vakatov
 * kEmptyCStr (equiv. to NcbiEmptyCStr)
 *
 * Revision 1.33  2000/10/11 21:03:49  vakatov
 * Cleanup to avoid 64-bit to 32-bit values truncation, etc.
 * (reported by Forte6 Patch 109490-01)
 *
 * Revision 1.32  2000/08/03 20:21:29  golikov
 * Added predicate PCase for AStrEquiv
 * PNocase, PCase goes through NStr::Compare now
 *
 * Revision 1.31  2000/07/19 19:03:55  vakatov
 * StringToBool() -- short and case-insensitive versions of "true"/"false"
 * ToUpper/ToLower(string&) -- fixed
 *
 * Revision 1.30  2000/06/01 19:05:40  vasilche
 * NStr::StringToInt now reports errors for tailing symbols in release version
 * too
 *
 * Revision 1.29  2000/05/01 19:02:25  vasilche
 * Force argument in NStr::StringToInt() etc to be full number.
 * This check will be in DEBUG version for month.
 *
 * Revision 1.28  2000/04/19 18:36:04  vakatov
 * Fixed for non-zero "pos" in s_Compare()
 *
 * Revision 1.27  2000/04/17 04:15:08  vakatov
 * NStr::  extended Compare(), and allow case-insensitive string comparison
 * NStr::  added ToLower() and ToUpper()
 *
 * Revision 1.26  2000/04/04 22:28:09  vakatov
 * NStr::  added conversions for "long"
 *
 * Revision 1.25  2000/02/01 16:48:09  vakatov
 * CNcbiEmptyString::  more dancing around the Sun "feature" (see also R1.24)
 *
 * Revision 1.24  2000/01/20 16:24:42  vakatov
 * Kludging around the "NcbiEmptyString" to ensure its initialization when
 * it is used by the constructor of a statically allocated object
 * (I believe that it is actually just another Sun WorkShop compiler "feature")
 *
 * Revision 1.23  1999/12/28 18:55:43  vasilche
 * Reduced size of compiled object files:
 * 1. avoid inline or implicit virtual methods (especially destructors).
 * 2. avoid std::string's methods usage in inline methods.
 * 3. avoid string literals ("xxx") in inline methods.
 *
 * Revision 1.22  1999/12/17 19:04:09  vasilche
 * NcbiEmptyString made extern.
 *
 * Revision 1.21  1999/11/26 19:29:09  golikov
 * fix
 *
 * Revision 1.20  1999/11/26 18:45:17  golikov
 * NStr::Replace added
 *
 * Revision 1.19  1999/11/17 22:05:04  vakatov
 * [!HAVE_STRDUP]  Emulate "strdup()" -- it's missing on some platforms
 *
 * Revision 1.18  1999/10/13 16:30:30  vasilche
 * Fixed bug in PNocase which appears under GCC implementation of STL.
 *
 * Revision 1.17  1999/07/08 16:10:14  vakatov
 * Fixed a warning in NStr::StringToUInt()
 *
 * Revision 1.16  1999/07/06 15:21:06  vakatov
 * + NStr::TruncateSpaces(const string& str, ETrunc where=eTrunc_Both)
 *
 * Revision 1.15  1999/06/15 20:50:05  vakatov
 * NStr::  +BoolToString, +StringToBool
 *
 * Revision 1.14  1999/05/27 15:21:40  vakatov
 * Fixed all StringToXXX() functions
 *
 * Revision 1.13  1999/05/17 20:10:36  vasilche
 * Fixed bug in NStr::StringToUInt which cause an exception.
 *
 * Revision 1.12  1999/05/04 00:03:13  vakatov
 * Removed the redundant severity arg from macro ERR_POST()
 *
 * Revision 1.11  1999/04/22 14:19:04  vasilche
 * Added _TRACE_THROW() macro, which can be configured to produce coredump
 * at a point of throwing an exception.
 *
 * Revision 1.10  1999/04/14 21:20:33  vakatov
 * Dont use "snprintf()" as it is not quite portable yet
 *
 * Revision 1.9  1999/04/14 19:57:36  vakatov
 * Use limits from <ncbitype.h> rather than from <limits>.
 * [MSVC++]  fix for "snprintf()" in <ncbistd.hpp>.
 *
 * Revision 1.8  1999/04/09 19:51:37  sandomir
 * minor changes in NStr::StringToXXX - base added
 *
 * Revision 1.7  1999/01/21 16:18:04  sandomir
 * minor changes due to NStr namespace to contain string utility functions
 *
 * Revision 1.6  1999/01/11 22:05:50  vasilche
 * Fixed CHTML_font size.
 * Added CHTML_image input element.
 *
 * Revision 1.5  1998/12/28 17:56:39  vakatov
 * New CVS and development tree structure for the NCBI C++ projects
 *
 * Revision 1.4  1998/12/21 17:19:37  sandomir
 * VC++ fixes in ncbistd; minor fixes in Resource
 *
 * Revision 1.3  1998/12/17 21:50:45  sandomir
 * CNCBINode fixed in Resource; case insensitive string comparison predicate
 * added
 *
 * Revision 1.2  1998/12/15 17:36:33  vasilche
 * Fixed "double colon" bug in multithreaded version of headers.
 *
 * Revision 1.1  1998/12/15 15:43:22  vasilche
 * Added utilities to convert string <> int.
 * ===========================================================================
 */