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 | * Use, modification and distribution is subject to the |
6 | * Boost Software License, Version 1.0. (See accompanying |
7 | * file LICENSE_1_0.txt or http://www.boost.org/LICENSE_1_0.txt) |
8 | * Author: Jeff Garland, Bart Garst |
9 | * $Date$ |
10 | */ |
11 | |
12 | #include <string> |
13 | #include <iterator> |
14 | #include <algorithm> |
15 | #include <boost/tokenizer.hpp> |
16 | #include <boost/lexical_cast.hpp> |
17 | #include <boost/date_time/compiler_config.hpp> |
18 | #include <boost/date_time/parse_format_base.hpp> |
19 | |
20 | #if defined(BOOST_DATE_TIME_NO_LOCALE) |
21 | #include <cctype> // ::tolower(int) |
22 | #else |
23 | #include <locale> // std::tolower(char, locale) |
24 | #endif |
25 | |
26 | namespace boost { |
27 | namespace date_time { |
28 | |
29 | //! A function to replace the std::transform( , , ,tolower) construct |
30 | /*! This function simply takes a string, and changes all the characters |
31 | * in that string to lowercase (according to the default system locale). |
32 | * In the event that a compiler does not support locales, the old |
33 | * C style tolower() is used. |
34 | */ |
35 | inline |
36 | std::string |
37 | convert_to_lower(std::string inp) |
38 | { |
39 | #if !defined(BOOST_DATE_TIME_NO_LOCALE) |
40 | const std::locale loc(std::locale::classic()); |
41 | #endif |
42 | std::string::size_type i = 0, n = inp.length(); |
43 | for (; i < n; ++i) { |
44 | inp[i] = |
45 | #if defined(BOOST_DATE_TIME_NO_LOCALE) |
46 | static_cast<char>(std::tolower(inp[i])); |
47 | #else |
48 | // tolower and others were brought in to std for borland >= v564 |
49 | // in compiler_config.hpp |
50 | std::tolower(c: inp[i], loc: loc); |
51 | #endif |
52 | } |
53 | return inp; |
54 | } |
55 | |
56 | //! Helper function for parse_date. |
57 | /* Used by-value parameter because we change the string and may |
58 | * want to preserve the original argument */ |
59 | template<class month_type> |
60 | inline unsigned short |
61 | month_str_to_ushort(std::string const& s) { |
62 | if((s.at(n: 0) >= '0') && (s.at(n: 0) <= '9')) { |
63 | return boost::lexical_cast<unsigned short>(arg: s); |
64 | } |
65 | else { |
66 | std::string str = convert_to_lower(inp: s); |
67 | typename month_type::month_map_ptr_type ptr = month_type::get_month_map_ptr(); |
68 | typename month_type::month_map_type::iterator iter = ptr->find(str); |
69 | if(iter != ptr->end()) { // required for STLport |
70 | return iter->second; |
71 | } |
72 | } |
73 | return 13; // intentionally out of range - name not found |
74 | } |
75 | |
76 | //! Find index of a string in either of 2 arrays |
77 | /*! find_match searches both arrays for a match to 's'. Both arrays |
78 | * must contain 'size' elements. The index of the match is returned. |
79 | * If no match is found, 'size' is returned. |
80 | * Ex. "Jan" returns 0, "Dec" returns 11, "Tue" returns 2. |
81 | * 'size' can be sent in with: (greg_month::max)() (which 12), |
82 | * (greg_weekday::max)() + 1 (which is 7) or date_time::NumSpecialValues */ |
83 | template<class charT> |
84 | short find_match(const charT* const* short_names, |
85 | const charT* const* long_names, |
86 | short size, |
87 | const std::basic_string<charT>& s) { |
88 | for(short i = 0; i < size; ++i){ |
89 | if(short_names[i] == s || long_names[i] == s){ |
90 | return i; |
91 | } |
92 | } |
93 | return size; // not-found, return a value out of range |
94 | } |
95 | |
96 | //! Generic function to parse a delimited date (eg: 2002-02-10) |
97 | /*! Accepted formats are: "2003-02-10" or " 2003-Feb-10" or |
98 | * "2003-Feburary-10" |
99 | * The order in which the Month, Day, & Year appear in the argument |
100 | * string can be accomodated by passing in the appropriate ymd_order_spec |
101 | */ |
102 | template<class date_type> |
103 | date_type |
104 | parse_date(const std::string& s, int order_spec = ymd_order_iso) { |
105 | std::string spec_str; |
106 | if(order_spec == ymd_order_iso) { |
107 | spec_str = "ymd" ; |
108 | } |
109 | else if(order_spec == ymd_order_dmy) { |
110 | spec_str = "dmy" ; |
111 | } |
112 | else { // (order_spec == ymd_order_us) |
113 | spec_str = "mdy" ; |
114 | } |
115 | |
116 | typedef typename date_type::month_type month_type; |
117 | unsigned pos = 0; |
118 | unsigned short year(0), month(0), day(0); |
119 | typedef typename std::basic_string<char>::traits_type traits_type; |
120 | typedef boost::char_separator<char, traits_type> char_separator_type; |
121 | typedef boost::tokenizer<char_separator_type, |
122 | std::basic_string<char>::const_iterator, |
123 | std::basic_string<char> > tokenizer; |
124 | typedef boost::tokenizer<char_separator_type, |
125 | std::basic_string<char>::const_iterator, |
126 | std::basic_string<char> >::iterator tokenizer_iterator; |
127 | // may need more delimiters, these work for the regression tests |
128 | const char sep_char[] = {',','-','.',' ','/','\0'}; |
129 | char_separator_type sep(sep_char); |
130 | tokenizer tok(s,sep); |
131 | for(tokenizer_iterator beg=tok.begin(); |
132 | beg!=tok.end() && pos < spec_str.size(); |
133 | ++beg, ++pos) { |
134 | switch(spec_str.at(n: pos)) { |
135 | case 'y': |
136 | { |
137 | year = boost::lexical_cast<unsigned short>(arg: *beg); |
138 | break; |
139 | } |
140 | case 'm': |
141 | { |
142 | month = month_str_to_ushort<month_type>(*beg); |
143 | break; |
144 | } |
145 | case 'd': |
146 | { |
147 | day = boost::lexical_cast<unsigned short>(arg: *beg); |
148 | break; |
149 | } |
150 | default: break; |
151 | } //switch |
152 | } |
153 | return date_type(year, month, day); |
154 | } |
155 | |
156 | //! Generic function to parse undelimited date (eg: 20020201) |
157 | template<class date_type> |
158 | date_type |
159 | parse_undelimited_date(const std::string& s) { |
160 | int offsets[] = {4,2,2}; |
161 | int pos = 0; |
162 | //typename date_type::ymd_type ymd((year_type::min)(),1,1); |
163 | unsigned short y = 0, m = 0, d = 0; |
164 | |
165 | /* The two bool arguments state that parsing will not wrap |
166 | * (only the first 8 characters will be parsed) and partial |
167 | * strings will not be parsed. |
168 | * Ex: |
169 | * "2005121" will parse 2005 & 12, but not the "1" */ |
170 | boost::offset_separator osf(offsets, offsets+3, false, false); |
171 | |
172 | typedef typename boost::tokenizer<boost::offset_separator, |
173 | std::basic_string<char>::const_iterator, |
174 | std::basic_string<char> > tokenizer_type; |
175 | tokenizer_type tok(s, osf); |
176 | for(typename tokenizer_type::iterator ti=tok.begin(); ti!=tok.end();++ti) { |
177 | unsigned short i = boost::lexical_cast<unsigned short>(arg: *ti); |
178 | switch(pos) { |
179 | case 0: y = i; break; |
180 | case 1: m = i; break; |
181 | case 2: d = i; break; |
182 | default: break; |
183 | } |
184 | pos++; |
185 | } |
186 | return date_type(y,m,d); |
187 | } |
188 | |
189 | //! Helper function for 'date gregorian::from_stream()' |
190 | /*! Creates a string from the iterators that reference the |
191 | * begining & end of a char[] or string. All elements are |
192 | * used in output string */ |
193 | template<class date_type, class iterator_type> |
194 | inline |
195 | date_type |
196 | from_stream_type(iterator_type& beg, |
197 | iterator_type const& end, |
198 | char) |
199 | { |
200 | std::ostringstream ss; |
201 | while(beg != end) { |
202 | ss << *beg++; |
203 | } |
204 | return parse_date<date_type>(ss.str()); |
205 | } |
206 | |
207 | //! Helper function for 'date gregorian::from_stream()' |
208 | /*! Returns the first string found in the stream referenced by the |
209 | * begining & end iterators */ |
210 | template<class date_type, class iterator_type> |
211 | inline |
212 | date_type |
213 | from_stream_type(iterator_type& beg, |
214 | iterator_type const& /* end */, |
215 | std::string const&) |
216 | { |
217 | return parse_date<date_type>(*beg); |
218 | } |
219 | |
220 | /* I believe the wchar stuff would be best elsewhere, perhaps in |
221 | * parse_date<>()? In the mean time this gets us started... */ |
222 | //! Helper function for 'date gregorian::from_stream()' |
223 | /*! Creates a string from the iterators that reference the |
224 | * begining & end of a wstring. All elements are |
225 | * used in output string */ |
226 | template<class date_type, class iterator_type> |
227 | inline |
228 | date_type from_stream_type(iterator_type& beg, |
229 | iterator_type const& end, |
230 | wchar_t) |
231 | { |
232 | std::ostringstream ss; |
233 | #if !defined(BOOST_DATE_TIME_NO_LOCALE) |
234 | std::locale loc; |
235 | std::ctype<wchar_t> const& fac = std::use_facet<std::ctype<wchar_t> >(loc: loc); |
236 | while(beg != end) { |
237 | ss << fac.narrow(*beg++, 'X'); // 'X' will cause exception to be thrown |
238 | } |
239 | #else |
240 | while(beg != end) { |
241 | char c = 'X'; // 'X' will cause exception to be thrown |
242 | const wchar_t wc = *beg++; |
243 | if (wc >= 0 && wc <= 127) |
244 | c = static_cast< char >(wc); |
245 | ss << c; |
246 | } |
247 | #endif |
248 | return parse_date<date_type>(ss.str()); |
249 | } |
250 | #ifndef BOOST_NO_STD_WSTRING |
251 | //! Helper function for 'date gregorian::from_stream()' |
252 | /*! Creates a string from the first wstring found in the stream |
253 | * referenced by the begining & end iterators */ |
254 | template<class date_type, class iterator_type> |
255 | inline |
256 | date_type |
257 | from_stream_type(iterator_type& beg, |
258 | iterator_type const& /* end */, |
259 | std::wstring const&) { |
260 | std::wstring ws = *beg; |
261 | std::ostringstream ss; |
262 | std::wstring::iterator wsb = ws.begin(), wse = ws.end(); |
263 | #if !defined(BOOST_DATE_TIME_NO_LOCALE) |
264 | std::locale loc; |
265 | std::ctype<wchar_t> const& fac = std::use_facet<std::ctype<wchar_t> >(loc: loc); |
266 | while(wsb != wse) { |
267 | ss << fac.narrow(c: *wsb++, dfault: 'X'); // 'X' will cause exception to be thrown |
268 | } |
269 | #else |
270 | while(wsb != wse) { |
271 | char c = 'X'; // 'X' will cause exception to be thrown |
272 | const wchar_t wc = *wsb++; |
273 | if (wc >= 0 && wc <= 127) |
274 | c = static_cast< char >(wc); |
275 | ss << c; |
276 | } |
277 | #endif |
278 | return parse_date<date_type>(ss.str()); |
279 | } |
280 | #endif // BOOST_NO_STD_WSTRING |
281 | #if (defined(BOOST_MSVC) && (_MSC_VER < 1300)) |
282 | // This function cannot be compiled with MSVC 6.0 due to internal compiler shorcomings |
283 | #else |
284 | //! function called by wrapper functions: date_period_from_(w)string() |
285 | template<class date_type, class charT> |
286 | period<date_type, typename date_type::duration_type> |
287 | from_simple_string_type(const std::basic_string<charT>& s){ |
288 | typedef typename std::basic_string<charT>::traits_type traits_type; |
289 | typedef typename boost::char_separator<charT, traits_type> char_separator; |
290 | typedef typename boost::tokenizer<char_separator, |
291 | typename std::basic_string<charT>::const_iterator, |
292 | std::basic_string<charT> > tokenizer; |
293 | const charT sep_list[4] = {'[','/',']','\0'}; |
294 | char_separator sep(sep_list); |
295 | tokenizer tokens(s, sep); |
296 | typename tokenizer::iterator tok_it = tokens.begin(); |
297 | std::basic_string<charT> date_string = *tok_it; |
298 | // get 2 string iterators and generate a date from them |
299 | typename std::basic_string<charT>::iterator date_string_start = date_string.begin(), |
300 | date_string_end = date_string.end(); |
301 | typedef typename std::iterator_traits<typename std::basic_string<charT>::iterator>::value_type value_type; |
302 | date_type d1 = from_stream_type<date_type>(date_string_start, date_string_end, value_type()); |
303 | date_string = *(++tok_it); // next token |
304 | date_string_start = date_string.begin(), date_string_end = date_string.end(); |
305 | date_type d2 = from_stream_type<date_type>(date_string_start, date_string_end, value_type()); |
306 | return period<date_type, typename date_type::duration_type>(d1, d2); |
307 | } |
308 | #endif |
309 | |
310 | } } //namespace date_time |
311 | |
312 | |
313 | |
314 | |
315 | #endif |
316 | |
317 | |