|
1 #ifndef _DATE_TIME_DATE_PARSING_HPP___ |
|
2 #define _DATE_TIME_DATE_PARSING_HPP___ |
|
3 |
|
4 /* Copyright (c) 2002,2003,2005 CrystalClear Software, Inc. |
|
5 * |
|
6 * Distributed under the Boost Software License, Version 1.0. |
|
7 * (See accompanying file LICENSE_1_0.txt or copy at |
|
8 * http://www.boost.org/LICENSE_1_0.txt) |
|
9 * |
|
10 * Author: Jeff Garland, Bart Garst |
|
11 * $Date: 2006/07/17 03:48:31 $ |
|
12 */ |
|
13 |
|
14 #include "boost/tokenizer.hpp" |
|
15 #include "boost/lexical_cast.hpp" |
|
16 #include "boost/date_time/compiler_config.hpp" |
|
17 #include "boost/date_time/parse_format_base.hpp" |
|
18 #include <string> |
|
19 #include <iterator> |
|
20 #include <algorithm> |
|
21 |
|
22 #if defined(BOOST_NO_STD_LOCALE) |
|
23 #include <cctype> // ::tolower(int) |
|
24 #else |
|
25 #include <locale> // std::tolower(char, locale) |
|
26 #endif |
|
27 |
|
28 namespace boost { |
|
29 namespace date_time { |
|
30 |
|
31 //! A function to replace the std::transform( , , ,tolower) construct |
|
32 /*! This function simply takes a string, and changes all the characters |
|
33 * in that string to lowercase (according to the default system locale). |
|
34 * In the event that a compiler does not support locales, the old |
|
35 * C style tolower() is used. |
|
36 */ |
|
37 inline |
|
38 std::string |
|
39 convert_to_lower(const std::string& inp) { |
|
40 std::string tmp; |
|
41 unsigned i = 0; |
|
42 #if defined(BOOST_NO_STD_LOCALE) |
|
43 while(i < inp.length()) { |
|
44 tmp += static_cast<char>(std::tolower(inp.at(i++))); |
|
45 #else |
|
46 static const std::locale loc(std::locale::classic()); |
|
47 while(i < inp.length()) { |
|
48 // tolower and others were brought in to std for borland >= v564 |
|
49 // in compiler_config.hpp |
|
50 std::string::value_type c(inp.at(i++)); |
|
51 tmp += std::tolower(c, loc); |
|
52 #endif |
|
53 |
|
54 } |
|
55 return tmp; |
|
56 } |
|
57 |
|
58 //! Helper function for parse_date. |
|
59 /* Used by-value parameter because we change the string and may |
|
60 * want to preserve the original argument */ |
|
61 template<class month_type> |
|
62 unsigned short |
|
63 month_str_to_ushort(std::string s) { |
|
64 if((s.at(0) >= '0') && (s.at(0) <= '9')) { |
|
65 return boost::lexical_cast<unsigned short>(s); |
|
66 } |
|
67 else { |
|
68 s = convert_to_lower(s); |
|
69 typename month_type::month_map_ptr_type ptr = month_type::get_month_map_ptr(); |
|
70 typename month_type::month_map_type::iterator iter = ptr->find(s); |
|
71 if(iter != ptr->end()) { // required for STLport |
|
72 return iter->second; |
|
73 } |
|
74 } |
|
75 return 13; // intentionally out of range - name not found |
|
76 } |
|
77 |
|
78 //! Find index of a string in either of 2 arrays |
|
79 /*! find_match searches both arrays for a match to 's'. Indexing of the |
|
80 * arrays is from 0 to 'limit'. The index of the match is returned. |
|
81 * Ex. "Jan" returns 0, "Dec" returns 11, "Tue" returns 2. |
|
82 * 'limit' can be sent in with: (greg_month::max)(), |
|
83 * (greg_weekday::max)() or date_time::NumSpecialValues */ |
|
84 template<class charT> |
|
85 short find_match(const charT* const* short_names, |
|
86 const charT* const* long_names, |
|
87 short limit, |
|
88 const std::basic_string<charT>& s) { |
|
89 for(short i = 0; i <= limit; ++i){ |
|
90 if(short_names[i] == s || long_names[i] == s){ |
|
91 return i; |
|
92 } |
|
93 } |
|
94 return static_cast<short>(limit + 1); // not-found, return a value out of range |
|
95 } |
|
96 |
|
97 //! Generic function to parse a delimited date (eg: 2002-02-10) |
|
98 /*! Accepted formats are: "2003-02-10" or " 2003-Feb-10" or |
|
99 * "2003-Feburary-10" |
|
100 * The order in which the Month, Day, & Year appear in the argument |
|
101 * string can be accomodated by passing in the appropriate ymd_order_spec |
|
102 */ |
|
103 template<class date_type> |
|
104 date_type |
|
105 parse_date(const std::string& s, int order_spec = ymd_order_iso) { |
|
106 std::string spec_str(""); |
|
107 if(order_spec == ymd_order_iso) { |
|
108 spec_str = "ymd"; |
|
109 } |
|
110 else if(order_spec == ymd_order_dmy) { |
|
111 spec_str = "dmy"; |
|
112 } |
|
113 else { // (order_spec == ymd_order_us) |
|
114 spec_str = "mdy"; |
|
115 } |
|
116 |
|
117 typedef typename date_type::year_type year_type; |
|
118 typedef typename date_type::month_type month_type; |
|
119 unsigned pos = 0; |
|
120 unsigned short year(0), month(0), day(0); |
|
121 typedef typename std::basic_string<char>::traits_type traits_type; |
|
122 typedef boost::char_separator<char, traits_type> char_separator_type; |
|
123 typedef boost::tokenizer<char_separator_type, |
|
124 std::basic_string<char>::const_iterator, |
|
125 std::basic_string<char> > tokenizer; |
|
126 typedef boost::tokenizer<char_separator_type, |
|
127 std::basic_string<char>::const_iterator, |
|
128 std::basic_string<char> >::iterator tokenizer_iterator; |
|
129 // may need more delimiters, these work for the regression tests |
|
130 const char sep_char[] = {',','-','.',' ','/','\0'}; |
|
131 char_separator_type sep(sep_char); |
|
132 tokenizer tok(s,sep); |
|
133 for(tokenizer_iterator beg=tok.begin(); |
|
134 beg!=tok.end() && pos < spec_str.size(); |
|
135 ++beg, ++pos) { |
|
136 switch(spec_str.at(pos)) { |
|
137 case 'y': |
|
138 { |
|
139 year = boost::lexical_cast<unsigned short>(*beg); |
|
140 break; |
|
141 } |
|
142 case 'm': |
|
143 { |
|
144 month = month_str_to_ushort<month_type>(*beg); |
|
145 break; |
|
146 } |
|
147 case 'd': |
|
148 { |
|
149 day = boost::lexical_cast<unsigned short>(*beg); |
|
150 break; |
|
151 } |
|
152 } //switch |
|
153 } |
|
154 return date_type(year, month, day); |
|
155 } |
|
156 |
|
157 //! Generic function to parse undelimited date (eg: 20020201) |
|
158 template<class date_type> |
|
159 date_type |
|
160 parse_undelimited_date(const std::string& s) { |
|
161 int offsets[] = {4,2,2}; |
|
162 int pos = 0; |
|
163 typedef typename date_type::year_type year_type; |
|
164 //typename date_type::ymd_type ymd((year_type::min)(),1,1); |
|
165 unsigned short y = 0, m = 0, d = 0; |
|
166 |
|
167 /* The two bool arguments state that parsing will not wrap |
|
168 * (only the first 8 characters will be parsed) and partial |
|
169 * strings will not be parsed. |
|
170 * Ex: |
|
171 * "2005121" will parse 2005 & 12, but not the "1" */ |
|
172 boost::offset_separator osf(offsets, offsets+3, false, false); |
|
173 |
|
174 typedef typename boost::tokenizer<boost::offset_separator, |
|
175 std::basic_string<char>::const_iterator, |
|
176 std::basic_string<char> > tokenizer_type; |
|
177 tokenizer_type tok(s, osf); |
|
178 for(typename tokenizer_type::iterator ti=tok.begin(); ti!=tok.end();++ti) { |
|
179 unsigned short i = boost::lexical_cast<unsigned short>(*ti); |
|
180 switch(pos) { |
|
181 case 0: y = i; break; |
|
182 case 1: m = i; break; |
|
183 case 2: d = i; break; |
|
184 } |
|
185 pos++; |
|
186 } |
|
187 return date_type(y,m,d); |
|
188 } |
|
189 |
|
190 //! Helper function for 'date gregorian::from_stream()' |
|
191 /*! Creates a string from the iterators that reference the |
|
192 * begining & end of a char[] or string. All elements are |
|
193 * used in output string */ |
|
194 template<class date_type, class iterator_type> |
|
195 inline |
|
196 date_type |
|
197 from_stream_type(iterator_type& beg, |
|
198 iterator_type& end, |
|
199 char) |
|
200 { |
|
201 std::stringstream ss(""); |
|
202 while(beg != end) { |
|
203 ss << *beg++; |
|
204 } |
|
205 return parse_date<date_type>(ss.str()); |
|
206 } |
|
207 |
|
208 //! Helper function for 'date gregorian::from_stream()' |
|
209 /*! Returns the first string found in the stream referenced by the |
|
210 * begining & end iterators */ |
|
211 template<class date_type, class iterator_type> |
|
212 inline |
|
213 date_type |
|
214 from_stream_type(iterator_type& beg, |
|
215 iterator_type& end, |
|
216 std::string) |
|
217 { |
|
218 return parse_date<date_type>(*beg); |
|
219 } |
|
220 |
|
221 /* I believe the wchar stuff would be best elsewhere, perhaps in |
|
222 * parse_date<>()? In the mean time this gets us started... */ |
|
223 //! Helper function for 'date gregorian::from_stream()' |
|
224 /*! Creates a string from the iterators that reference the |
|
225 * begining & end of a wstring. All elements are |
|
226 * used in output string */ |
|
227 template<class date_type, class iterator_type> |
|
228 inline |
|
229 date_type from_stream_type(iterator_type& beg, |
|
230 iterator_type& end, |
|
231 wchar_t) |
|
232 { |
|
233 std::stringstream ss(""); |
|
234 while(beg != end) { |
|
235 #if !defined(BOOST_DATE_TIME_NO_LOCALE) |
|
236 ss << std::use_facet<std::ctype<wchar_t> >(std::locale()).narrow(*beg++, 'X'); // 'X' will cause exception to be thrown |
|
237 #else |
|
238 ss << ss.narrow(*beg++, 'X'); |
|
239 #endif |
|
240 } |
|
241 return parse_date<date_type>(ss.str()); |
|
242 } |
|
243 #ifndef BOOST_NO_STD_WSTRING |
|
244 //! Helper function for 'date gregorian::from_stream()' |
|
245 /*! Creates a string from the first wstring found in the stream |
|
246 * referenced by the begining & end iterators */ |
|
247 template<class date_type, class iterator_type> |
|
248 inline |
|
249 date_type |
|
250 from_stream_type(iterator_type& beg, |
|
251 iterator_type& end, |
|
252 std::wstring) { |
|
253 std::wstring ws = *beg; |
|
254 std::stringstream ss(""); |
|
255 std::wstring::iterator wsb = ws.begin(), wse = ws.end(); |
|
256 while(wsb != wse) { |
|
257 #if !defined(BOOST_DATE_TIME_NO_LOCALE) |
|
258 ss << std::use_facet<std::ctype<wchar_t> >(std::locale()).narrow(*wsb++, 'X'); // 'X' will cause exception to be thrown |
|
259 #else |
|
260 ss << ss.narrow(*wsb++, 'X'); // 'X' will cause exception to be thrown |
|
261 #endif |
|
262 } |
|
263 return parse_date<date_type>(ss.str()); |
|
264 } |
|
265 #endif // BOOST_NO_STD_WSTRING |
|
266 #if (defined(BOOST_MSVC) && (_MSC_VER < 1300)) |
|
267 // This function cannot be compiled with MSVC 6.0 due to internal compiler shorcomings |
|
268 #else |
|
269 //! function called by wrapper functions: date_period_from_(w)string() |
|
270 template<class date_type, class charT> |
|
271 period<date_type, typename date_type::duration_type> |
|
272 from_simple_string_type(const std::basic_string<charT>& s){ |
|
273 typedef typename std::basic_string<charT>::traits_type traits_type; |
|
274 typedef typename boost::char_separator<charT, traits_type> char_separator; |
|
275 typedef typename boost::tokenizer<char_separator, |
|
276 typename std::basic_string<charT>::const_iterator, |
|
277 std::basic_string<charT> > tokenizer; |
|
278 const charT sep_list[4] = {'[','/',']','\0'}; |
|
279 char_separator sep(sep_list); |
|
280 tokenizer tokens(s, sep); |
|
281 typename tokenizer::iterator tok_it = tokens.begin(); |
|
282 std::basic_string<charT> date_string = *tok_it; |
|
283 // get 2 string iterators and generate a date from them |
|
284 typename std::basic_string<charT>::iterator date_string_start = date_string.begin(), |
|
285 date_string_end = date_string.end(); |
|
286 typedef typename std::iterator_traits<typename std::basic_string<charT>::iterator>::value_type value_type; |
|
287 date_type d1 = from_stream_type<date_type>(date_string_start, date_string_end, value_type()); |
|
288 date_string = *(++tok_it); // next token |
|
289 date_string_start = date_string.begin(), date_string_end = date_string.end(); |
|
290 date_type d2 = from_stream_type<date_type>(date_string_start, date_string_end, value_type()); |
|
291 return period<date_type, typename date_type::duration_type>(d1, d2); |
|
292 } |
|
293 #endif |
|
294 |
|
295 } } //namespace date_time |
|
296 |
|
297 |
|
298 |
|
299 |
|
300 #endif |
|
301 |