String.h

Go to the documentation of this file.

/*---------------------------------------------------------------------\
|                          ____ _   __ __ ___                          |
|                         |__  / \ / / . \ . \                         |
|                           / / \ V /|  _/  _/                         |
|                          / /__ | | | | | |                           |
|                         /_____||_| |_| |_|                           |
|                                                                      |
\---------------------------------------------------------------------*/
#ifndef ZYPP_BASE_STRING_H
#define ZYPP_BASE_STRING_H

#include <cstring>

#include <iosfwd>
#include <vector>
#include <string>
#include <sstream>
#include <boost/format.hpp>
#include <boost/utility/string_ref.hpp>

#include "zypp/base/Easy.h"
#include "zypp/base/PtrTypes.h"
#include "zypp/base/Function.h"

namespace boost { namespace logic { class tribool; } }
namespace zypp { typedef boost::logic::tribool TriBool; }

namespace zypp
{
  template <class Tp>
  std::string asUserString( const Tp & val_r )
  { return val_r.asUserString(); }

}// namespace zypp

namespace zypp
{ 

  class C_Str
  {
    public:
      typedef std::string::size_type size_type;

    public:
      C_Str()                            : _val( 0 ),             _sze( 0 ) {}
      C_Str( char * c_str_r )            : _val( c_str_r ),       _sze( std::string::npos ) {}
      C_Str( const char * c_str_r )      : _val( c_str_r ),       _sze( std::string::npos ) {}
      C_Str( const std::string & str_r ) : _val( str_r.c_str() ), _sze( str_r.size() ) {}
      C_Str( const boost::string_ref & str_r ) : _val( str_r.data() ), _sze( str_r.size() ) {}

    public:
      bool      isNull()       const { return !_val; }
      bool      empty()        const { return !(_val && *_val); }
      size_type size()         const
      {
        if ( _sze == std::string::npos )
        { _sze = _val ? ::strlen( _val ) : 0; }
        return _sze;
      };

      operator const char *() const { return c_str(); }
      const char * c_str()    const { return _val ? _val : ""; }

    private:
      const char *const _val;
      mutable size_type _sze;
  };

  inline std::ostream & operator<<( std::ostream & str, const C_Str & obj )
  { return str << obj.c_str(); }


  namespace str
  { 


    inline const std::string & asString( const std::string & t )
    { return t; }

#ifndef SWIG // Swig treats it as syntax error
    inline std::string && asString( std::string && t )
    { return std::move(t); }
#endif

    inline std::string asString( const char * t )
    { return t; }

    inline std::string asString( char * t )
    { return t; }

    template<class Tp>
        inline std::string asString( const Tp &t )
        { return t.asString(); }

    template<class Tp>
        inline std::string asString( const intrusive_ptr<Tp> &p )
        { return p->asString(); }

    template<class Tp>
        inline std::string asString( const weak_ptr<Tp> &p )
        { return p->asString(); }

    template<>
        inline std::string asString( const bool &t )
        { return t ? "true" : "false"; }


    std::string form( const char * format, ... )
    __attribute__ ((format (printf, 1, 2)));


    std::string strerror( int errno_r );


    struct SafeBuf
    {
      char * _buf;
      SafeBuf() : _buf( 0 ) {}
      ~SafeBuf() { if ( _buf ) free( _buf ); }
      std::string asString() const
      { return _buf ? std::string(_buf) : std::string(); }
    };

    struct Str
    {
      template<class Tp>
      Str & operator<<( Tp && val )
      { _str << std::forward<Tp>(val); return *this; }

      Str & operator<<( std::ostream& (*iomanip)( std::ostream& ) )
      { _str << iomanip; return *this; }

      operator std::string() const              { return _str.str(); }
      std::string str() const                   { return _str.str(); }

      const std::ostream & stream() const       { return _str; }
      std::ostream & stream()                   { return _str; }

      void clear()                              { _str.str( std::string() ); }

    private:
      std::ostringstream _str;
    };

    inline std::ostream & operator<<( std::ostream & str, const Str & obj )
    { return str << obj.str(); }

    struct Format
    {
      Format() {}
      Format( const std::string & format_r ) : _fmter( format_r ) {}

      template<class Tp>
      Format & operator%( Tp && arg )
      { _fmter % std::forward<Tp>(arg); return *this; }

      operator std::string() const              { return _fmter.str(); }
      std::string asString() const              { return _fmter.str(); }
      std::string str() const                   { return _fmter.str(); }

      const boost::format & fmter() const       { return _fmter; }
      boost::format & fmter()                   { return _fmter; }

    protected:
      boost::format _fmter;
    };

    inline std::ostream & operator<<( std::ostream & str, const Format & obj )
    { return str << obj.fmter(); }

    struct FormatNAC : public Format
    {
      FormatNAC() { relax(); }
      FormatNAC( const std::string & format_r ) : Format( format_r ) { relax(); }

    private:
      void relax()
      {
        using namespace boost::io;
        _fmter.exceptions( all_error_bits ^ ( too_many_args_bit | too_few_args_bit ) );
      }
    };

    inline std::string numstring( char n,               int w = 0 ) { return form( "%*hhd",  w, n ); }
    inline std::string numstring( unsigned char n,      int w = 0 ) { return form( "%*hhu",  w, n ); }
    inline std::string numstring( short n,              int w = 0 ) { return form( "%*hd",   w, n ); }
    inline std::string numstring( unsigned short n,     int w = 0 ) { return form( "%*hu",   w, n ); }
    inline std::string numstring( int n,                int w = 0 ) { return form( "%*d",    w, n ); }
    inline std::string numstring( unsigned n,           int w = 0 ) { return form( "%*u",    w, n ); }
    inline std::string numstring( long n,               int w = 0 ) { return form( "%*ld",   w, n ); }
    inline std::string numstring( unsigned long n,      int w = 0 ) { return form( "%*lu",   w, n ); }
    inline std::string numstring( long long n,          int w = 0 ) { return form( "%*lld",  w, n ); }
    inline std::string numstring( unsigned long long n, int w = 0 ) { return form( "%*llu",  w, n ); }

    template<> inline std::string asString( const char & t )                    { return numstring( t ); }
    template<> inline std::string asString( const unsigned char & t )           { return numstring( t ); }
    template<> inline std::string asString( const short & t )                   { return numstring( t ); }
    template<> inline std::string asString( const unsigned short & t )          { return numstring( t ); }
    template<> inline std::string asString( const int & t )                     { return numstring( t ); }
    template<> inline std::string asString( const unsigned & t )                { return numstring( t ); }
    template<> inline std::string asString( const long & t )                    { return numstring( t ); }
    template<> inline std::string asString( const unsigned long & t )           { return numstring( t ); }
    template<> inline std::string asString( const long long & t )               { return numstring( t ); }
    template<> inline std::string asString( const unsigned long long & t )      { return numstring( t ); }


    inline std::string hexstring( char n,               int w = 4 ) { return form( "%#0*hhx", w, n ); }
    inline std::string hexstring( unsigned char n,      int w = 4 ) { return form( "%#0*hhx", w, n ); }
    inline std::string hexstring( short n,              int w = 10 ){ return form( "%#0*hx",  w, n ); }
    inline std::string hexstring( unsigned short n,     int w = 10 ){ return form( "%#0*hx",  w, n ); }
    inline std::string hexstring( int n,                int w = 10 ){ return form( "%#0*x",   w, n ); }
    inline std::string hexstring( unsigned n,           int w = 10 ){ return form( "%#0*x",   w, n ); }
    inline std::string hexstring( long n,               int w = 10 ){ return form( "%#0*lx",  w, n ); }
    inline std::string hexstring( unsigned long n,      int w = 10 ){ return form( "%#0*lx",  w, n ); }
    inline std::string hexstring( long long n,          int w = 0 ) { return form( "%#0*llx", w, n ); }
    inline std::string hexstring( unsigned long long n, int w = 0 ) { return form( "%#0*llx", w, n ); }


    inline std::string octstring( char n,               int w = 4 ) { return form( "%#0*hho",  w, n ); }
    inline std::string octstring( unsigned char n,      int w = 4 ) { return form( "%#0*hho",  w, n ); }
    inline std::string octstring( short n,              int w = 5 ) { return form( "%#0*ho",   w, n ); }
    inline std::string octstring( unsigned short n,     int w = 5 ) { return form( "%#0*ho",   w, n ); }
    inline std::string octstring( int n,                int w = 5 ) { return form( "%#0*o",    w, n ); }
    inline std::string octstring( unsigned n,           int w = 5 ) { return form( "%#0*o",    w, n ); }
    inline std::string octstring( long n,               int w = 5 ) { return form( "%#0*lo",   w, n ); }
    inline std::string octstring( unsigned long n,      int w = 5 ) { return form( "%#0*lo",   w, n ); }
    inline std::string octstring( long long n,          int w = 0 ) { return form( "%#0*llo",  w, n ); }
    inline std::string octstring( unsigned long long n, int w = 0 ) { return form( "%#0*llo",  w, n ); }



    template <typename TInt>
    std::string binstring( TInt val_r )
    {
      constexpr unsigned bits = sizeof(TInt)*8;
      std::string ret( bits, ' ' );
      TInt bit = 1;
      for ( unsigned pos = bits; pos > 0; )
      { --pos; ret[pos] = ((val_r & bit)?'1':'0'); bit = bit<<1; }
      return ret;
    }
    

    template<typename TInt>
      TInt strtonum( const C_Str & str );

    template<>
      inline short              strtonum( const C_Str & str ) { return ::strtol  ( str, NULL, 0 ); }
    template<>
      inline int                strtonum( const C_Str & str ) { return ::strtol  ( str, NULL, 0 ); }
    template<>
      inline long               strtonum( const C_Str & str ) { return ::strtol  ( str, NULL, 0 ); }
    template<>
      inline long long          strtonum( const C_Str & str ) { return ::strtoll ( str, NULL, 0 ); }

    template<>
      inline unsigned short     strtonum( const C_Str & str ) { return ::strtoul ( str, NULL, 0 ); }
    template<>
      inline unsigned           strtonum( const C_Str & str ) { return ::strtoul ( str, NULL, 0 ); }
    template<>
      inline unsigned long      strtonum( const C_Str & str ) { return ::strtoul ( str, NULL, 0 ); }
    template<>
      inline unsigned long long strtonum( const C_Str & str ) { return ::strtoull( str, NULL, 0 ); }

    template<typename TInt>
      inline TInt strtonum( const C_Str & str, TInt & i )
      { return i = strtonum<TInt>( str ); }


    bool strToTrue( const C_Str & str );

    bool strToFalse( const C_Str & str );

    inline bool strToBool( const C_Str & str, bool default_r )
    { return( default_r ? strToFalse( str ) : strToTrue( str ) ); }

    inline bool strToBoolNodefault( const C_Str & str, bool & return_r )
    {
      if ( strToTrue( str ) ) return (return_r = true);
      if ( !strToFalse( str ) ) return (return_r = false);
      return return_r;
    }

    TriBool strToTriBool( const C_Str & str );


    std::string gsub( const std::string & str_r, const std::string & from_r, const std::string & to_r );

    std::string gsubFun( const std::string & str_r, const std::string & from_r, function<std::string()> to_r );

    std::string & replaceAll( std::string & str_r, const std::string & from_r, const std::string & to_r );

    std::string & replaceAllFun( std::string & str_r, const std::string & from_r, function<std::string()> to_r );

    inline std::string gapify( std::string inp_r, std::string::size_type gap_r = 1, char gapchar = ' ' )
    {
      if ( gap_r &&  inp_r.size() > gap_r )
      {
        inp_r.reserve( inp_r.size() + (inp_r.size()-1)/gap_r );
        for ( std::string::size_type pos = gap_r; pos < inp_r.size(); pos += gap_r+1 )
          inp_r.insert( pos, 1, gapchar );
      }
      return inp_r;
    }


    template<class TOutputIterator>
      unsigned split( const C_Str & line_r, TOutputIterator result_r, const C_Str & sepchars_r = " \t" )
      {
        const char * beg = line_r;
        const char * cur = beg;
        // skip leading sepchars
        while ( *cur && ::strchr( sepchars_r, *cur ) )
          ++cur;
        unsigned ret = 0;
        for ( beg = cur; *beg; beg = cur, ++result_r, ++ret )
          {
            // skip non sepchars
            while( *cur && !::strchr( sepchars_r, *cur ) )
              ++cur;
            // build string
            *result_r = std::string( beg, cur-beg );
            // skip sepchars
            while ( *cur && ::strchr( sepchars_r, *cur ) )
              ++cur;
          }
        return ret;
      }

    template<class TOutputIterator>
      unsigned splitEscaped( const C_Str & line_r, TOutputIterator result_r, const C_Str & sepchars_r = " \t", bool withEmpty = false)
      {
        const char * beg = line_r;
        const char * cur = beg;
        unsigned ret = 0;

        // skip leading sepchars
        while ( *cur && ::strchr( sepchars_r, *cur ) )
        {
          ++cur;
          if (withEmpty)
          {
            *result_r = "";
            ++ret;
          }
        }

        // there were only sepchars in the string
        if (!*cur && withEmpty)
        {
          *result_r = "";
          return ++ret;
        }

        // after the leading sepchars
        enum class Quote { None, Slash, Single, Double, DoubleSlash };
        std::vector<char> buf;
        Quote quoting = Quote::None;
        for ( beg = cur; *beg; beg = cur, ++result_r, ++ret )
        {
          // read next value until unquoted sepchar
          buf.clear();
          quoting = Quote::None;
          do {
            switch ( quoting )
            {
              case Quote::None:
                switch ( *cur )
                {
                  case '\\':    quoting = Quote::Slash;         break;
                  case '\'':    quoting = Quote::Single;        break;
                  case '"':     quoting = Quote::Double;        break;
                  default:      buf.push_back( *cur );          break;
                }
                break;

              case Quote::Slash:
                buf.push_back( *cur );
                quoting = Quote::None;
                break;

              case Quote::Single:
                switch ( *cur )
                {
                  case '\'':    quoting = Quote::None;          break;
                  default:      buf.push_back( *cur );          break;
                }
                break;

              case Quote::Double:
                switch ( *cur )
                {
                  case '\"':    quoting = Quote::None;          break;
                  case '\\':    quoting = Quote::DoubleSlash;   break;
                  default:      buf.push_back( *cur );          break;
                }
                break;

              case Quote::DoubleSlash:
                switch ( *cur )
                {
                  case '\"':    /*fallthrough*/
                  case '\\':    buf.push_back( *cur );          break;
                  default:
                    buf.push_back( '\\' );
                    buf.push_back( *cur );
                    break;
                }
                quoting = Quote::Double;
                break;
            }
            ++cur;
          } while ( *cur && ( quoting != Quote::None || !::strchr( sepchars_r, *cur ) ) );
          *result_r = std::string( buf.begin(), buf.end() );


          // skip sepchars
          if ( *cur && ::strchr( sepchars_r, *cur ) )
            ++cur;
          while ( *cur && ::strchr( sepchars_r, *cur ) )
          {
            ++cur;
            if (withEmpty)
            {
              *result_r = "";
              ++ret;
            }
          }
          // the last was a separator => one more field
          if ( !*cur && withEmpty && ::strchr( sepchars_r, *(cur-1) ) )
          {
            *result_r = "";
            ++ret;
          }
        }
        return ret;
      }

    template<class TOutputIterator>
      unsigned splitFields( const C_Str & line_r, TOutputIterator result_r, const C_Str & sepchars_r = ":" )
      {
        const char * beg = line_r;
        const char * cur = beg;
        unsigned ret = 0;
        for ( beg = cur; *beg; beg = cur, ++result_r )
          {
            // skip non sepchars
            while( *cur && !::strchr( sepchars_r, *cur ) )
            {
              if ( *cur == '\\' && *(cur+1) )
                ++cur;
              ++cur;
            }
            // build string
            *result_r = std::string( beg, cur-beg );
            ++ret;
            // skip sepchar
            if ( *cur )
            {
              ++cur;
              if ( ! *cur )                // ending with sepchar
              {
                *result_r = std::string(); // add final empty field
                ++ret;
                break;
              }
            }
          }
        return ret;
      }

    template<class TOutputIterator>
      unsigned splitFieldsEscaped( const C_Str & line_r, TOutputIterator result_r, const C_Str & sepchars_r = ":" )
      {
        return splitEscaped( line_r, result_r, sepchars_r, true /* withEmpty */ );
      }



    template <class TIterator>
      std::string join( TIterator begin, TIterator end, const C_Str & sep_r = " " )
      {
        std::string res;
        for ( TIterator iter = begin; iter != end; ++ iter )
          {
            if ( iter != begin )
              res += sep_r;
            res += asString(*iter);
          }
        return res;
      }

    template <class TContainer>
      std::string join( const TContainer & cont_r, const C_Str & sep_r = " " )
      { return join( cont_r.begin(), cont_r.end(), sep_r ); }

    template <class TIterator>
      std::string joinEscaped( TIterator begin, TIterator end, const char sep_r = ' ' )
      {
        std::vector<char> buf;
        for ( TIterator iter = begin; iter != end; ++ iter )
        {
          if ( iter != begin )
            buf.push_back( sep_r );

          if ( iter->empty() )
          {
            // empty string goes ""
            buf.push_back( '"' );
            buf.push_back( '"' );
          }
          else
          {
            std::string toadd( asString(*iter) );
            for_( ch, toadd.begin(), toadd.end() )
            {
              switch ( *ch )
              {
                case '"':
                case '\'':
                case '\\':
                  buf.push_back( '\\' );
                  buf.push_back( *ch );
                  break;
                default:
                  if ( *ch == sep_r )
                    buf.push_back( '\\' );
                  buf.push_back( *ch );
              }
            }
          }
        }
        return std::string( buf.begin(), buf.end() );
      }



      inline std::ostream & printIndented( std::ostream & str, const std::string & text_r, const std::string & indent_r = "  ", unsigned maxWitdh_r = 0 )
      {
        if ( maxWitdh_r )
        {
          if ( indent_r.size() >= maxWitdh_r )
            maxWitdh_r = 0;     // nonsense: indent larger than line witdh
          else
            maxWitdh_r -= indent_r.size();
        }
        unsigned width = 0;
        for ( const char * e = text_r.c_str(), * s = e; *e; s = ++e )
        {
          for ( ; *e && *e != '\n'; ++e ) ;/*searching*/
          width = e-s;
          if ( maxWitdh_r && width > maxWitdh_r )
          {
            // must break line
            width = maxWitdh_r;
            for ( e = s+width; e > s && *e != ' '; --e ) ;/*searching*/
            if ( e > s )
              width = e-s;      // on a ' ', replaced by '\n'
            else
              e = s+width-1;    // cut line;
          }
          str << indent_r;
          str.write( s, width );
          str << "\n";
          if ( !*e )    // on '\0'
            break;
        }
        return str;
      }
      inline std::ostream & printIndented( std::ostream & str, const std::string & text_r, unsigned indent_r, char indentch_r = ' ', unsigned maxWitdh_r = 0 )
      { return printIndented( str, text_r, std::string( indent_r, indentch_r ), maxWitdh_r ); }
      inline std::ostream & printIndented( std::ostream & str, const std::string & text_r, unsigned indent_r, unsigned maxWitdh_r, char indentch_r = ' ' )
      { return printIndented( str, text_r, std::string( indent_r, indentch_r ), maxWitdh_r ); }

      inline std::ostream & autoPrefix( std::ostream & str, const std::string & text_r, function<std::string(const char*, const char*)> fnc_r )
      {
        for ( const char * e = text_r.c_str(); *e; ++e )
        {
          const char * s = e;
          for ( ; *e && *e != '\n'; ++e ) /*searching*/;
          str << fnc_r( s, e );
          str.write( s, e-s );
          str << "\n";
          if ( !*e )    // on '\0'
            break;
        }
        return str;
      }
      inline std::ostream & autoPrefix0( std::ostream & str, const std::string & text_r, function<std::string()> fnc_r )
      {
        auto wrap = [&fnc_r]( const char*, const char* )-> std::string {
          return fnc_r();
        };
        return autoPrefix( str, text_r, wrap );
      }

      std::string escape( const C_Str & str_r, const char c = ' ' );

      inline void appendEscaped( std::string & str_r, const C_Str & next_r, const char sep_r = ' ' )
      {
        if ( ! str_r.empty() )
          str_r += sep_r;
        if ( next_r.empty() )
          str_r += "\"\"";
        else
          str_r += escape( next_r, sep_r );
      }

      std::string bEscape( std::string str_r, const C_Str & special_r );

      std::string rxEscapeStr( std::string str_r );

      std::string rxEscapeGlob( std::string str_r );



    std::string hexencode( const C_Str & str_r );
    std::string hexdecode( const C_Str & str_r );

    std::string toLower( const std::string & s );
    std::string toLower( std::string && s );
    inline std::string toLower( const char * s )
    { return( s ? toLower( std::string(s) ) : std::string() ); }

    std::string toUpper( const std::string & s );
    std::string toUpper( std::string && s );
    inline std::string toUpper( const char * s )
    { return( s ? toUpper( std::string(s) ) : std::string() ); }


    inline int compareCI( const C_Str & lhs, const C_Str & rhs )
    { return ::strcasecmp( lhs, rhs ); }

    inline bool contains( const C_Str & str_r, const C_Str & val_r )
    { return ::strstr( str_r, val_r ); }
    inline bool containsCI( const C_Str & str_r, const C_Str & val_r )
    { return ::strcasestr( str_r, val_r ); }


    enum Trim {
      NO_TRIM = 0x00,
      L_TRIM  = 0x01,
      R_TRIM  = 0x02,
      TRIM    = (L_TRIM|R_TRIM)
    };

    std::string trim( const std::string & s, const Trim trim_r = TRIM );
    std::string trim( std::string && s, const Trim trim_r = TRIM );

    inline std::string ltrim( const std::string & s )
    { return trim( s, L_TRIM ); }
    inline std::string ltrim( std::string && s )
    { return trim( std::move(s), L_TRIM ); }

    inline std::string rtrim( const std::string & s )
    { return trim( s, R_TRIM ); }
    inline std::string rtrim( std::string && s )
    { return trim( std::move(s), R_TRIM ); }

    std::string stripFirstWord( std::string & line, const bool ltrim_first = true );

    std::string stripLastWord( std::string & line, const bool rtrim_first = true );

    std::string getline( std::istream & str, bool trim = false );

    std::string getline( std::istream & str, const Trim trim_r );

    std::string receiveUpTo( std::istream & str, const char delim_r, bool returnDelim_r = false );


    inline bool hasPrefix( const C_Str & str_r, const C_Str & prefix_r )
    { return( ::strncmp( str_r, prefix_r, prefix_r.size() ) == 0 ); }

    inline std::string stripPrefix( const C_Str & str_r, const C_Str & prefix_r )
    { return( hasPrefix( str_r, prefix_r ) ? str_r + prefix_r.size() : str_r.c_str() ); }

    inline bool hasSuffix( const C_Str & str_r, const C_Str & suffix_r )
    { return( str_r.size() >= suffix_r.size() && ::strncmp( str_r + str_r.size() - suffix_r.size() , suffix_r, suffix_r.size() ) == 0 ); }

    inline std::string stripSuffix( const C_Str & str_r, const C_Str & suffix_r )
    {
      if ( hasSuffix( str_r, suffix_r ) )
        return std::string( str_r, str_r.size() - suffix_r.size() );
      return str_r.c_str();
    }
    inline std::string::size_type commonPrefix( const C_Str & lhs, const C_Str & rhs )
    {
      const char * lp = lhs.c_str();
      const char * rp = rhs.c_str();
      std::string::size_type ret = 0;
      while ( *lp == *rp && *lp != '\0' )
      { ++lp, ++rp, ++ret; }
      return ret;
    }

    inline bool startsWith( const C_Str & str_r, const C_Str & prefix_r )
    { return hasPrefix( str_r, prefix_r ); }
    inline bool endsWith( const C_Str & str_r, const C_Str & prefix_r )
    { return hasSuffix( str_r, prefix_r ); }
  } // namespace str

  // drag into zypp:: namespace
  using str::asString;

} // namespace zypp
#endif // ZYPP_BASE_STRING_H