Chris@16: #ifndef _DATE_TIME_DATE_PARSING_HPP___
Chris@16: #define _DATE_TIME_DATE_PARSING_HPP___
Chris@16: 
Chris@16: /* Copyright (c) 2002,2003,2005 CrystalClear Software, Inc.
Chris@16:  * Use, modification and distribution is subject to the
Chris@16:  * Boost Software License, Version 1.0. (See accompanying
Chris@16:  * file LICENSE_1_0.txt or http://www.boost.org/LICENSE_1_0.txt)
Chris@16:  * Author: Jeff Garland, Bart Garst
Chris@101:  * $Date$
Chris@16:  */
Chris@16: 
Chris@16: #include <string>
Chris@16: #include <iterator>
Chris@16: #include <algorithm>
Chris@16: #include <boost/tokenizer.hpp>
Chris@16: #include <boost/lexical_cast.hpp>
Chris@16: #include <boost/date_time/compiler_config.hpp>
Chris@16: #include <boost/date_time/parse_format_base.hpp>
Chris@16: 
Chris@16: #if defined(BOOST_DATE_TIME_NO_LOCALE)
Chris@16: #include <cctype> // ::tolower(int)
Chris@16: #else
Chris@16: #include <locale> // std::tolower(char, locale)
Chris@16: #endif
Chris@16: 
Chris@16: namespace boost {
Chris@16: namespace date_time {
Chris@16: 
Chris@16:   //! A function to replace the std::transform( , , ,tolower) construct
Chris@16:   /*! This function simply takes a string, and changes all the characters
Chris@16:    * in that string to lowercase (according to the default system locale).
Chris@16:    * In the event that a compiler does not support locales, the old
Chris@16:    * C style tolower() is used.
Chris@16:    */
Chris@16:   inline
Chris@16:   std::string
Chris@16:   convert_to_lower(std::string inp)
Chris@16:   {
Chris@16: #if !defined(BOOST_DATE_TIME_NO_LOCALE)
Chris@16:     const std::locale loc(std::locale::classic());
Chris@16: #endif
Chris@16:     std::string::size_type i = 0, n = inp.length();
Chris@16:     for (; i < n; ++i) {
Chris@16:       inp[i] =
Chris@16: #if defined(BOOST_DATE_TIME_NO_LOCALE)
Chris@16:         static_cast<char>(std::tolower(inp[i]));
Chris@16: #else
Chris@16:         // tolower and others were brought in to std for borland >= v564
Chris@16:         // in compiler_config.hpp
Chris@16:         std::tolower(inp[i], loc);
Chris@16: #endif
Chris@16:     }
Chris@16:     return inp;
Chris@16:   }
Chris@16: 
Chris@16:     //! Helper function for parse_date.
Chris@16:     /* Used by-value parameter because we change the string and may
Chris@16:      * want to preserve the original argument */
Chris@16:     template<class month_type>
Chris@16:     inline unsigned short
Chris@16:     month_str_to_ushort(std::string const& s) {
Chris@16:       if((s.at(0) >= '0') && (s.at(0) <= '9')) {
Chris@16:         return boost::lexical_cast<unsigned short>(s);
Chris@16:       }
Chris@16:       else {
Chris@16:         std::string str = convert_to_lower(s);
Chris@16:         typename month_type::month_map_ptr_type ptr = month_type::get_month_map_ptr();
Chris@16:         typename month_type::month_map_type::iterator iter = ptr->find(str);
Chris@16:         if(iter != ptr->end()) { // required for STLport
Chris@16:           return iter->second;
Chris@16:         }
Chris@16:       }
Chris@16:       return 13; // intentionally out of range - name not found
Chris@16:     }
Chris@16:  
Chris@16:     //! Find index of a string in either of 2 arrays
Chris@16:     /*! find_match searches both arrays for a match to 's'. Both arrays
Chris@16:      * must contain 'size' elements. The index of the match is returned.
Chris@16:      * If no match is found, 'size' is returned.
Chris@16:      * Ex. "Jan" returns 0, "Dec" returns 11, "Tue" returns 2.
Chris@16:      * 'size' can be sent in with: (greg_month::max)() (which 12),
Chris@16:      * (greg_weekday::max)() + 1 (which is 7) or date_time::NumSpecialValues */
Chris@16:     template<class charT>
Chris@16:     short find_match(const charT* const* short_names,
Chris@16:                      const charT* const* long_names,
Chris@16:                      short size,
Chris@16:                      const std::basic_string<charT>& s) {
Chris@16:       for(short i = 0; i < size; ++i){
Chris@16:         if(short_names[i] == s || long_names[i] == s){
Chris@16:           return i;
Chris@16:         }
Chris@16:       }
Chris@16:       return size; // not-found, return a value out of range
Chris@16:     }
Chris@16: 
Chris@16:     //! Generic function to parse a delimited date (eg: 2002-02-10)
Chris@16:     /*! Accepted formats are: "2003-02-10" or " 2003-Feb-10" or
Chris@16:      * "2003-Feburary-10"
Chris@16:      * The order in which the Month, Day, & Year appear in the argument
Chris@16:      * string can be accomodated by passing in the appropriate ymd_order_spec
Chris@16:      */
Chris@16:     template<class date_type>
Chris@16:     date_type
Chris@16:     parse_date(const std::string& s, int order_spec = ymd_order_iso) {
Chris@16:       std::string spec_str;
Chris@16:       if(order_spec == ymd_order_iso) {
Chris@16:         spec_str = "ymd";
Chris@16:       }
Chris@16:       else if(order_spec == ymd_order_dmy) {
Chris@16:         spec_str = "dmy";
Chris@16:       }
Chris@16:       else { // (order_spec == ymd_order_us)
Chris@16:         spec_str = "mdy";
Chris@16:       }
Chris@16: 
Chris@16:       typedef typename date_type::month_type month_type;
Chris@16:       unsigned pos = 0;
Chris@16:       unsigned short year(0), month(0), day(0);
Chris@16:       typedef typename std::basic_string<char>::traits_type traits_type;
Chris@16:       typedef boost::char_separator<char, traits_type> char_separator_type;
Chris@16:       typedef boost::tokenizer<char_separator_type,
Chris@16:                                std::basic_string<char>::const_iterator,
Chris@16:                                std::basic_string<char> > tokenizer;
Chris@16:       typedef boost::tokenizer<char_separator_type,
Chris@16:                                std::basic_string<char>::const_iterator,
Chris@16:                                std::basic_string<char> >::iterator tokenizer_iterator;
Chris@16:       // may need more delimiters, these work for the regression tests
Chris@16:       const char sep_char[] = {',','-','.',' ','/','\0'};
Chris@16:       char_separator_type sep(sep_char);
Chris@16:       tokenizer tok(s,sep);
Chris@16:       for(tokenizer_iterator beg=tok.begin();
Chris@16:           beg!=tok.end() && pos < spec_str.size();
Chris@16:           ++beg, ++pos) {
Chris@16:         switch(spec_str.at(pos)) {
Chris@16:           case 'y':
Chris@16:           {
Chris@16:             year = boost::lexical_cast<unsigned short>(*beg);
Chris@16:             break;
Chris@16:           }
Chris@16:           case 'm':
Chris@16:           {
Chris@16:             month = month_str_to_ushort<month_type>(*beg);
Chris@16:             break;
Chris@16:           }
Chris@16:           case 'd':
Chris@16:           {
Chris@16:             day = boost::lexical_cast<unsigned short>(*beg);
Chris@16:             break;
Chris@16:           }
Chris@16:           default: break;
Chris@16:         } //switch
Chris@16:       }
Chris@16:       return date_type(year, month, day);
Chris@16:     }
Chris@16: 
Chris@16:     //! Generic function to parse undelimited date (eg: 20020201)
Chris@16:     template<class date_type>
Chris@16:     date_type
Chris@16:     parse_undelimited_date(const std::string& s) {
Chris@16:       int offsets[] = {4,2,2};
Chris@16:       int pos = 0;
Chris@16:       //typename date_type::ymd_type ymd((year_type::min)(),1,1);
Chris@16:       unsigned short y = 0, m = 0, d = 0;
Chris@16: 
Chris@16:       /* The two bool arguments state that parsing will not wrap
Chris@16:        * (only the first 8 characters will be parsed) and partial
Chris@16:        * strings will not be parsed.
Chris@16:        * Ex:
Chris@16:        * "2005121" will parse 2005 & 12, but not the "1" */
Chris@16:       boost::offset_separator osf(offsets, offsets+3, false, false);
Chris@16: 
Chris@16:       typedef typename boost::tokenizer<boost::offset_separator,
Chris@16:                                         std::basic_string<char>::const_iterator,
Chris@16:                                         std::basic_string<char> > tokenizer_type;
Chris@16:       tokenizer_type tok(s, osf);
Chris@16:       for(typename tokenizer_type::iterator ti=tok.begin(); ti!=tok.end();++ti) {
Chris@16:         unsigned short i = boost::lexical_cast<unsigned short>(*ti);
Chris@16:         switch(pos) {
Chris@16:         case 0: y = i; break;
Chris@16:         case 1: m = i; break;
Chris@16:         case 2: d = i; break;
Chris@16:         default:       break;
Chris@16:         }
Chris@16:         pos++;
Chris@16:       }
Chris@16:       return date_type(y,m,d);
Chris@16:     }
Chris@16: 
Chris@16:     //! Helper function for 'date gregorian::from_stream()'
Chris@16:     /*! Creates a string from the iterators that reference the
Chris@16:      * begining & end of a char[] or string. All elements are
Chris@16:      * used in output string */
Chris@16:     template<class date_type, class iterator_type>
Chris@16:     inline
Chris@16:     date_type
Chris@16:     from_stream_type(iterator_type& beg,
Chris@16:                      iterator_type const& end,
Chris@16:                      char)
Chris@16:     {
Chris@16:       std::ostringstream ss;
Chris@16:       while(beg != end) {
Chris@16:         ss << *beg++;
Chris@16:       }
Chris@16:       return parse_date<date_type>(ss.str());
Chris@16:     }
Chris@16: 
Chris@16:     //! Helper function for 'date gregorian::from_stream()'
Chris@16:     /*! Returns the first string found in the stream referenced by the
Chris@16:      * begining & end iterators */
Chris@16:     template<class date_type, class iterator_type>
Chris@16:     inline
Chris@16:     date_type
Chris@16:     from_stream_type(iterator_type& beg,
Chris@16:                      iterator_type const& /* end */,
Chris@16:                      std::string const&)
Chris@16:     {
Chris@16:       return parse_date<date_type>(*beg);
Chris@16:     }
Chris@16: 
Chris@16:     /* I believe the wchar stuff would be best elsewhere, perhaps in
Chris@16:      * parse_date<>()? In the mean time this gets us started... */
Chris@16:     //! Helper function for 'date gregorian::from_stream()'
Chris@16:     /*! Creates a string from the iterators that reference the
Chris@16:      * begining & end of a wstring. All elements are
Chris@16:      * used in output string */
Chris@16:     template<class date_type, class iterator_type>
Chris@16:     inline
Chris@16:     date_type from_stream_type(iterator_type& beg,
Chris@16:                                iterator_type const& end,
Chris@16:                                wchar_t)
Chris@16:     {
Chris@16:       std::ostringstream ss;
Chris@16: #if !defined(BOOST_DATE_TIME_NO_LOCALE)
Chris@16:       std::locale loc;
Chris@16:       std::ctype<wchar_t> const& fac = std::use_facet<std::ctype<wchar_t> >(loc);
Chris@16:       while(beg != end) {
Chris@16:         ss << fac.narrow(*beg++, 'X'); // 'X' will cause exception to be thrown
Chris@16:       }
Chris@16: #else
Chris@16:       while(beg != end) {
Chris@16:         char c = 'X'; // 'X' will cause exception to be thrown
Chris@16:         const wchar_t wc = *beg++;
Chris@16:         if (wc >= 0 && wc <= 127)
Chris@16:           c = static_cast< char >(wc);
Chris@16:         ss << c;
Chris@16:       }
Chris@16: #endif
Chris@16:       return parse_date<date_type>(ss.str());
Chris@16:     }
Chris@16: #ifndef BOOST_NO_STD_WSTRING
Chris@16:     //! Helper function for 'date gregorian::from_stream()'
Chris@16:     /*! Creates a string from the first wstring found in the stream
Chris@16:      * referenced by the begining & end iterators */
Chris@16:     template<class date_type, class iterator_type>
Chris@16:     inline
Chris@16:     date_type
Chris@16:     from_stream_type(iterator_type& beg,
Chris@16:                      iterator_type const& /* end */,
Chris@16:                      std::wstring const&) {
Chris@16:       std::wstring ws = *beg;
Chris@16:       std::ostringstream ss;
Chris@16:       std::wstring::iterator wsb = ws.begin(), wse = ws.end();
Chris@16: #if !defined(BOOST_DATE_TIME_NO_LOCALE)
Chris@16:       std::locale loc;
Chris@16:       std::ctype<wchar_t> const& fac = std::use_facet<std::ctype<wchar_t> >(loc);
Chris@16:       while(wsb != wse) {
Chris@16:         ss << fac.narrow(*wsb++, 'X'); // 'X' will cause exception to be thrown
Chris@16:       }
Chris@16: #else
Chris@16:       while(wsb != wse) {
Chris@16:         char c = 'X'; // 'X' will cause exception to be thrown
Chris@16:         const wchar_t wc = *wsb++;
Chris@16:         if (wc >= 0 && wc <= 127)
Chris@16:           c = static_cast< char >(wc);
Chris@16:         ss << c;
Chris@16:       }
Chris@16: #endif
Chris@16:       return parse_date<date_type>(ss.str());
Chris@16:     }
Chris@16: #endif // BOOST_NO_STD_WSTRING
Chris@16: #if (defined(BOOST_MSVC) && (_MSC_VER < 1300))
Chris@16:     // This function cannot be compiled with MSVC 6.0 due to internal compiler shorcomings
Chris@16: #else
Chris@16:     //! function called by wrapper functions: date_period_from_(w)string()
Chris@16:     template<class date_type, class charT>
Chris@16:     period<date_type, typename date_type::duration_type>
Chris@16:     from_simple_string_type(const std::basic_string<charT>& s){
Chris@16:       typedef typename std::basic_string<charT>::traits_type traits_type;
Chris@16:       typedef typename boost::char_separator<charT, traits_type> char_separator;
Chris@16:       typedef typename boost::tokenizer<char_separator,
Chris@16:                                         typename std::basic_string<charT>::const_iterator,
Chris@16:                                         std::basic_string<charT> > tokenizer;
Chris@16:       const charT sep_list[4] = {'[','/',']','\0'};
Chris@16:       char_separator sep(sep_list);
Chris@16:       tokenizer tokens(s, sep);
Chris@16:       typename tokenizer::iterator tok_it = tokens.begin();
Chris@16:       std::basic_string<charT> date_string = *tok_it;
Chris@16:       // get 2 string iterators and generate a date from them
Chris@16:       typename std::basic_string<charT>::iterator date_string_start = date_string.begin(),
Chris@16:                                                   date_string_end = date_string.end();
Chris@16:       typedef typename std::iterator_traits<typename std::basic_string<charT>::iterator>::value_type value_type;
Chris@16:       date_type d1 = from_stream_type<date_type>(date_string_start, date_string_end, value_type());
Chris@16:       date_string = *(++tok_it); // next token
Chris@16:       date_string_start = date_string.begin(), date_string_end = date_string.end();
Chris@16:       date_type d2 = from_stream_type<date_type>(date_string_start, date_string_end, value_type());
Chris@16:       return period<date_type, typename date_type::duration_type>(d1, d2);
Chris@16:     }
Chris@16: #endif
Chris@16: 
Chris@16: } } //namespace date_time
Chris@16: 
Chris@16: 
Chris@16: 
Chris@16: 
Chris@16: #endif
Chris@16: