fileutil.cpp

ncbi源码

string MakeFileName(const string& fname, size_t addLength)
{
    string name = Identifier(fname);
    size_t fullLength = name.size() + addLength;
    if ( fullLength <= MAX_FILE_NAME_LENGTH )
        return name;
    size_t remove = fullLength - MAX_FILE_NAME_LENGTH;
    // we'll have to truncate very long filename

    _TRACE("MakeFileName(\""<<fname<<"\", "<<addLength<<") remove="<<remove);
    // 1st step: parse name dividing by '_' sorting elements by their size
    SIZE_TYPE removable = 0; // removable part of string
    typedef set<SSubString, SSubString::ByLength> TByLength;
    TByLength byLength;
    {
        SIZE_TYPE curr = 0; // current element position in string
        size_t order = 0; // current element order
        for (;;) {
            SIZE_TYPE und = name.find('_', curr);
            if ( und == NPOS ) {
                // end of string
                break;
            }
            _TRACE("MakeFileName: \""<<name.substr(curr, und - curr)<<"\"");
            removable += (und - curr);
            byLength.insert(SSubString(name.substr(curr, und - curr), order));
            curr = und + 1;
            ++order;
        }
        _TRACE("MakeFileName: \""<<name.substr(curr)<<"\"");
        removable += name.size() - curr;
        byLength.insert(SSubString(name.substr(curr), order));
    }
    _TRACE("MakeFileName: removable="<<removable);

    // if removable part of string too small...
    if ( removable - remove < size_t(MAX_FILE_NAME_LENGTH - addLength) / 2 ) {
        // we'll do plain truncate
        _TRACE("MakeFileName: return \""<<name.substr(0, MAX_FILE_NAME_LENGTH - addLength)<<"\"");
        return name.substr(0, MAX_FILE_NAME_LENGTH - addLength);
    }
    
    // 2nd step: shorten elementes beginning with longest
    while ( remove > 0 ) {
        // extract most long element
        SSubString s = *byLength.begin();
        _TRACE("MakeFileName: shorten \""<<s.value<<"\"");
        byLength.erase(byLength.begin());
        // shorten it by one symbol
        s.value = s.value.substr(0, s.value.size() - 1);
        // insert it back
        byLength.insert(s);
        // decrement progress counter
        remove--;
    }
    // 3rd step: reorder elements by their relative order in original string
    typedef set<SSubString, SSubString::ByOrder> TByOrder;
    TByOrder byOrder;
    {
        ITERATE ( TByLength, i, byLength ) {
            byOrder.insert(*i);
        }
    }
    // 4th step: join elements in resulting string
    name.erase();
    {
        ITERATE ( TByOrder, i, byOrder ) {
            if ( !name.empty() )
                name += '_';
            name += i->value;
        }
    }
    _TRACE("MakeFileName: return \""<<name<<"\"");
    return name;
}

CDelayedOfstream::CDelayedOfstream(const string& fileName)
{
    open(fileName);
}

CDelayedOfstream::~CDelayedOfstream(void)
{
    close();
}

void CDelayedOfstream::open(const string& fileName)
{
    close();
    clear();
    seekp(0, IOS_BASE::beg);
    clear(); // eof set?
    m_FileName = fileName;
    m_Istream.reset(new CNcbiIfstream(fileName.c_str()));
    if ( !*m_Istream ) {
        _TRACE("cannot open " << m_FileName);
        m_Istream.reset(0);
        m_Ostream.reset(new CNcbiOfstream(fileName.c_str()));
        if ( !*m_Ostream ) {
            _TRACE("cannot create " << m_FileName);
            setstate(m_Ostream->rdstate());
            m_Ostream.reset(0);
            m_FileName.erase();
        }
    }
}

void CDelayedOfstream::close(void)
{
    if ( !is_open() )
        return;
    if ( !equals() ) {
        if ( !rewrite() )
            setstate(m_Ostream->rdstate());
        m_Ostream.reset(0);
    }
    m_Istream.reset(0);
    m_FileName.erase();
}

bool CDelayedOfstream::equals(void)
{
    if ( !m_Istream.get() )
        return false;
    streamsize count = pcount();
    const char* ptr = str();
    freeze(false);
    while ( count > 0 ) {
        char buffer[BUFFER_SIZE];
        streamsize c = count;
        if ( c > BUFFER_SIZE )
            c = BUFFER_SIZE;
        if ( !m_Istream->read(buffer, c) ) {
            _TRACE("read fault " << m_FileName <<
                   " need: " << c << " was: " << m_Istream->gcount());
            return false;
        }
        if ( memcmp(buffer, ptr, c) != 0 ) {
            _TRACE("file differs " << m_FileName);
            return false;
        }
        ptr += c;
        count -= c;
    }
    if ( m_Istream->get() != -1 ) {
        _TRACE("file too long " << m_FileName);
        return false;
    }
    return true;
}

bool CDelayedOfstream::rewrite(void)
{
    if ( !m_Ostream.get() ) {
        m_Ostream.reset(new CNcbiOfstream(m_FileName.c_str()));
        if ( !*m_Ostream ) {
            _TRACE("rewrite fault " << m_FileName);
            return false;
        }
    }
    streamsize count = pcount();
    const char* ptr = str();
    freeze(false);
    if ( !m_Ostream->write(ptr, count) ) {
        _TRACE("write fault " << m_FileName);
        return false;
    }
    m_Ostream->close();
    if ( !*m_Ostream ) {
        _TRACE("close fault " << m_FileName);
        return false;
    }
    return true;
}

bool Empty(const CNcbiOstrstream& src)
{
    return const_cast<CNcbiOstrstream&>(src).pcount() == 0;
}

CNcbiOstream& Write(CNcbiOstream& out, const CNcbiOstrstream& src)
{
    CNcbiOstrstream& source = const_cast<CNcbiOstrstream&>(src);
    size_t size = source.pcount();
    if ( size != 0 ) {
        out.write(source.str(), size);
        source.freeze(false);
    }
    return out;
}

CNcbiOstream& WriteTabbed(CNcbiOstream& out, const CNcbiOstrstream& code,
                          const char* tab)
{
    CNcbiOstrstream& source = const_cast<CNcbiOstrstream&>(code);
    size_t size = source.pcount();
    if ( size != 0 ) {
        if ( !tab )
            tab = "    ";
        const char* ptr = source.str();
        source.freeze(false);
        while ( size > 0 ) {
            out << tab;
            const char* endl =
                reinterpret_cast<const char*>(memchr(ptr, '\n', size));
            if ( !endl ) { // no more '\n'
                out.write(ptr, size) << '\n';
                break;
            }
            ++endl; // skip '\n'
            size_t lineSize = endl - ptr;
            out.write(ptr, lineSize);
            ptr = endl;
            size -= lineSize;
        }
    }
    return out;
}

END_NCBI_SCOPE

/*
*
* ---------------------------------------------------------------------------
* $Log: fileutil.cpp,v $
* Revision 1000.1  2004/06/01 19:43:09  gouriano
* PRODUCTION: UPGRADED [GCC34_MSVC7] Dev-tree R1.28
*
* Revision 1.28  2004/05/17 21:03:14  gorelenk
* Added include of PCH ncbi_pch.hpp
*
* Revision 1.27  2004/04/29 20:12:36  gouriano
* GetStdPath tweaked
*
* Revision 1.26  2003/03/11 20:06:47  kuznets
* iterate -> ITERATE
*
* Revision 1.25  2002/10/15 13:54:58  gouriano
* addded "GetType" method - module file type by extension
*
* Revision 1.24  2002/08/06 17:03:49  ucko
* Let -opm take a comma-delimited list; move relevant CVS logs to end.
*
* Revision 1.23  2001/09/19 19:03:59  ucko
* Use SIZE_TYPE to index strings.
*
* Revision 1.22  2001/08/31 20:05:46  ucko
* Fix ICC build.
*
* Revision 1.21  2001/08/16 13:19:26  grichenk
* Removed extra-defines for MAC, rearranged code for MAC paths
*
* Revision 1.20  2001/08/15 20:26:32  juran
* Correctly assemble Mac pathnames.  (Omit duplicate ':'.)
* IsLocalPath(): A Mac pathname beginning with ':' is relative.
* GetStdPath(): Convert native pathnames to unix-style.
*
* Revision 1.19  2001/05/17 15:07:12  lavr
* Typos corrected
*
* Revision 1.18  2001/02/02 20:50:35  vasilche
* Fixed defines on Mac
*
* Revision 1.17  2001/02/02 16:20:00  vasilche
* Fixed file path processing on Mac
*
* Revision 1.16  2000/11/29 17:42:44  vasilche
* Added CComment class for storing/printing ASN.1/XML module comments.
* Added srcutil.hpp file to reduce file dependency.
*
* Revision 1.15  2000/11/20 17:26:32  vasilche
* Fixed warnings on 64 bit platforms.
* Updated names of config variables.
*
* Revision 1.14  2000/11/15 20:34:54  vasilche
* Added user comments to ENUMERATED types.
* Added storing of user comments to ASN.1 module definition.
*
* Revision 1.13  2000/11/14 21:41:24  vasilche
* Added preserving of ASN.1 definition comments.
*
* Revision 1.12  2000/08/25 15:59:22  vasilche
* Renamed directory tool -> datatool.
*
* Revision 1.11  2000/04/10 20:01:31  vakatov
* Typo fixed
*
* Revision 1.10  2000/04/10 19:34:02  vakatov
* Get rid of a minor compiler warning
*
* Revision 1.9  2000/04/07 19:26:27  vasilche
* Added namespace support to datatool.
* By default with argument -oR datatool will generate objects in namespace
* NCBI_NS_NCBI::objects (aka ncbi::objects).
* Datatool's classes also moved to NCBI namespace.
*
* Revision 1.8  2000/03/29 15:52:27  vasilche
* Generated files names limited to 31 symbols due to limitations of Mac.
* Removed unions with only one member.
*
* Revision 1.7  2000/02/01 21:47:59  vasilche
* Added CGeneratedChoiceTypeInfo for generated choice classes.
* Removed CMemberInfo subclasses.
* Added support for DEFAULT/OPTIONAL members.
* Changed class generation.
* Moved datatool headers to include/internal/serial/tool.
*
* Revision 1.6  2000/01/05 19:34:53  vasilche
* Fixed warning on MS VC
*
* Revision 1.5  1999/12/30 21:33:39  vasilche
* Changed arguments - more structured.
* Added intelligence in detection of source directories.
*
* Revision 1.4  1999/12/28 18:55:57  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 1.3  1999/12/21 17:44:18  vasilche
* Fixed compilation on SunPro C++
*
* Revision 1.2  1999/12/21 17:18:34  vasilche
* Added CDelayedFostream class which rewrites file only if contents is changed.
*
* Revision 1.1  1999/12/20 21:00:17  vasilche
* Added generation of sources in different directories.
*
* ===========================================================================
*/