1 # (c) 2005 Clark C. Evans and contributors 2 # This module is part of the Python Paste Project and is released under 3 # the MIT License: http://www.opensource.org/licenses/mit-license.php 4 # Some of this code was funded by: http://prometheusresearch.com 5 """ 6 Date, Time, and Timespan Parsing Utilities 7 8 This module contains parsing support to create "human friendly" 9 ``datetime`` object parsing. The explicit goal of these routines is 10 to provide a multi-format date/time support not unlike that found in 11 Microsoft Excel. In most approaches, the input is very "strict" to 12 prevent errors -- however, this approach is much more liberal since we 13 are assuming the user-interface is parroting back the normalized value 14 and thus the user has immediate feedback if the data is not typed in 15 correctly. 16 17 ``parse_date`` and ``normalize_date`` 18 19 These functions take a value like '9 jan 2007' and returns either an 20 ``date`` object, or an ISO 8601 formatted date value such 21 as '2007-01-09'. There is an option to provide an Oracle database 22 style output as well, ``09 JAN 2007``, but this is not the default. 23 24 This module always treats '/' delimiters as using US date order 25 (since the author's clients are US based), hence '1/9/2007' is 26 January 9th. Since this module treats the '-' as following 27 European order this supports both modes of data-entry; together 28 with immediate parroting back the result to the screen, the author 29 has found this approach to work well in pratice. 30 31 ``parse_time`` and ``normalize_time`` 32 33 These functions take a value like '1 pm' and returns either an 34 ``time`` object, or an ISO 8601 formatted 24h clock time 35 such as '13:00'. There is an option to provide for US style time 36 values, '1:00 PM', however this is not the default. 37 38 ``parse_datetime`` and ``normalize_datetime`` 39 40 These functions take a value like '9 jan 2007 at 1 pm' and returns 41 either an ``datetime`` object, or an ISO 8601 formatted 42 return (without the T) such as '2007-01-09 13:00'. There is an 43 option to provide for Oracle / US style, '09 JAN 2007 @ 1:00 PM', 44 however this is not the default. 45 46 ``parse_delta`` and ``normalize_delta`` 47 48 These functions take a value like '1h 15m' and returns either an 49 ``timedelta`` object, or an 2-decimal fixed-point 50 numerical value in hours, such as '1.25'. The rationale is to 51 support meeting or time-billing lengths, not to be an accurate 52 representation in mili-seconds. As such not all valid 53 ``timedelta`` values will have a normalized representation. 54 55 """ 56 from datetime import timedelta, time, date 57 from time import localtime 58 59 __all__ = ['parse_timedelta', 'normalize_timedelta', 60 'parse_time', 'normalize_time', 61 'parse_date', 'normalize_date'] 62 63 def _number(val): 64 try: 65 return int(val) 66 except: 67 return None 68 69 # 70 # timedelta 71 # 72 def parse_timedelta(val): 73 """ 74 returns a ``timedelta`` object, or None 75 """ 76 if not val: 77 return None 78 val = val.lower() 79 if "." in val: 80 val = float(val) 81 return timedelta(hours=int(val), minutes=60*(val % 1.0)) 82 fHour = ("h" in val or ":" in val) 83 fMin = ("m" in val or ":" in val) 84 for noise in "minu:teshour()": 85 val = val.replace(noise, ' ') 86 val = val.strip() 87 val = val.split() 88 hr = 0.0 89 mi = 0 90 val.reverse() 91 if fHour: 92 hr = int(val.pop()) 93 if fMin: 94 mi = int(val.pop()) 95 if len(val) > 0 and not hr: 96 hr = int(val.pop()) 97 return timedelta(hours=hr, minutes=mi) 98 99 def normalize_timedelta(val): 100 """ 101 produces a normalized string value of the timedelta 102 103 This module returns a normalized time span value consisting of the 104 number of hours in fractional form. For example '1h 15min' is 105 formatted as 01.25. 106 """ 107 if type(val) == str: 108 val = parse_timedelta(val) 109 if not val: 110 return '' 111 hr = val.seconds/3600 112 mn = (val.seconds % 3600)/60 113 return "%d.%02d" % (hr, mn * 100/60) 114 115 # 116 # time 117 # 118 def parse_time(val): 119 if not val: 120 return None 121 hr = mi = 0 122 val = val.lower() 123 amflag = (-1 != val.find('a')) # set if AM is found 124 pmflag = (-1 != val.find('p')) # set if PM is found 125 for noise in ":amp.": 126 val = val.replace(noise, ' ') 127 val = val.split() 128 if len(val) > 1: 129 hr = int(val[0]) 130 mi = int(val[1]) 131 else: 132 val = val[0] 133 if len(val) < 1: 134 pass 135 elif 'now' == val: 136 tm = localtime() 137 hr = tm[3] 138 mi = tm[4] 139 elif 'noon' == val: 140 hr = 12 141 elif len(val) < 3: 142 hr = int(val) 143 if not amflag and not pmflag and hr < 7: 144 hr += 12 145 elif len(val) < 5: 146 hr = int(val[:-2]) 147 mi = int(val[-2:]) 148 else: 149 hr = int(val[:1]) 150 if amflag and hr >= 12: 151 hr = hr - 12 152 if pmflag and hr < 12: 153 hr = hr + 12 154 return time(hr, mi) 155 156 def normalize_time(value, ampm): 157 if not value: 158 return '' 159 if type(value) == str: 160 value = parse_time(value) 161 if not ampm: 162 return "%02d:%02d" % (value.hour, value.minute) 163 hr = value.hour 164 am = "AM" 165 if hr < 1 or hr > 23: 166 hr = 12 167 elif hr >= 12: 168 am = "PM" 169 if hr > 12: 170 hr = hr - 12 171 return "%02d:%02d %s" % (hr, value.minute, am) 172 173 # 174 # Date Processing 175 # 176 177 _one_day = timedelta(days=1) 178 179 _str2num = {'jan':1, 'feb':2, 'mar':3, 'apr':4, 'may':5, 'jun':6, 180 'jul':7, 'aug':8, 'sep':9, 'oct':10, 'nov':11, 'dec':12 } 181 182 def _month(val): 183 for (key, mon) in _str2num.items(): 184 if key in val: 185 return mon 186 raise TypeError("unknown month '%s'" % val) 187 188 _days_in_month = {1: 31, 2: 28, 3: 31, 4: 30, 5: 31, 6: 30, 189 7: 31, 8: 31, 9: 30, 10: 31, 11: 30, 12: 31, 190 } 191 _num2str = {1: 'Jan', 2: 'Feb', 3: 'Mar', 4: 'Apr', 5: 'May', 6: 'Jun', 192 7: 'Jul', 8: 'Aug', 9: 'Sep', 10: 'Oct', 11: 'Nov', 12: 'Dec', 193 } 194 _wkdy = ("mon", "tue", "wed", "thu", "fri", "sat", "sun") 195 196 def parse_date(val): 197 if not(val): 198 return None 199 val = val.lower() 200 now = None 201 202 # optimized check for YYYY-MM-DD 203 strict = val.split("-") 204 if len(strict) == 3: 205 (y, m, d) = strict 206 if "+" in d: 207 d = d.split("+")[0] 208 if " " in d: 209 d = d.split(" ")[0] 210 try: 211 now = date(int(y), int(m), int(d)) 212 val = "xxx" + val[10:] 213 except ValueError: 214 pass 215 216 # allow for 'now', 'mon', 'tue', etc. 217 if not now: 218 chk = val[:3] 219 if chk in ('now','tod'): 220 now = date.today() 221 elif chk in _wkdy: 222 now = date.today() 223 idx = list(_wkdy).index(chk) + 1 224 while now.isoweekday() != idx: 225 now += _one_day 226 227 # allow dates to be modified via + or - /w number of days, so 228 # that now+3 is three days from now 229 if now: 230 tail = val[3:].strip() 231 tail = tail.replace("+"," +").replace("-"," -") 232 for item in tail.split(): 233 try: 234 days = int(item) 235 except ValueError: 236 pass 237 else: 238 now += timedelta(days=days) 239 return now 240 241 # ok, standard parsing 242 yr = mo = dy = None 243 for noise in ('/', '-', ',', '*'): 244 val = val.replace(noise, ' ') 245 for noise in _wkdy: 246 val = val.replace(noise, ' ') 247 out = [] 248 last = False 249 ldig = False 250 for ch in val: 251 if ch.isdigit(): 252 if last and not ldig: 253 out.append(' ') 254 last = ldig = True 255 else: 256 if ldig: 257 out.append(' ') 258 ldig = False 259 last = True 260 out.append(ch) 261 val = "".join(out).split() 262 if 3 == len(val): 263 a = _number(val[0]) 264 b = _number(val[1]) 265 c = _number(val[2]) 266 if len(val[0]) == 4: 267 yr = a 268 if b: # 1999 6 23 269 mo = b 270 dy = c 271 else: # 1999 Jun 23 272 mo = _month(val[1]) 273 dy = c 274 elif a is not None and a > 0: 275 yr = c 276 if len(val[2]) < 4: 277 raise TypeError("four digit year required") 278 if b: # 6 23 1999 279 dy = b 280 mo = a 281 else: # 23 Jun 1999 282 dy = a 283 mo = _month(val[1]) 284 else: # Jun 23, 2000 285 dy = b 286 yr = c 287 if len(val[2]) < 4: 288 raise TypeError("four digit year required") 289 mo = _month(val[0]) 290 elif 2 == len(val): 291 a = _number(val[0]) 292 b = _number(val[1]) 293 if a is not None and a > 999: 294 yr = a 295 dy = 1 296 if b is not None and b > 0: # 1999 6 297 mo = b 298 else: # 1999 Jun 299 mo = _month(val[1]) 300 elif a is not None and a > 0: 301 if b is not None and b > 999: # 6 1999 302 mo = a 303 yr = b 304 dy = 1 305 elif b is not None and b > 0: # 6 23 306 mo = a 307 dy = b 308 else: # 23 Jun 309 dy = a 310 mo = _month(val[1]) 311 else: 312 if b > 999: # Jun 2001 313 yr = b 314 dy = 1 315 else: # Jun 23 316 dy = b 317 mo = _month(val[0]) 318 elif 1 == len(val): 319 val = val[0] 320 if not val.isdigit(): 321 mo = _month(val) 322 if mo is not None: 323 dy = 1 324 else: 325 v = _number(val) 326 val = str(v) 327 if 8 == len(val): # 20010623 328 yr = _number(val[:4]) 329 mo = _number(val[4:6]) 330 dy = _number(val[6:]) 331 elif len(val) in (3,4): 332 if v is not None and v > 1300: # 2004 333 yr = v 334 mo = 1 335 dy = 1 336 else: # 1202 337 mo = _number(val[:-2]) 338 dy = _number(val[-2:]) 339 elif v < 32: 340 dy = v 341 else: 342 raise TypeError("four digit year required") 343 tm = localtime() 344 if mo is None: 345 mo = tm[1] 346 if dy is None: 347 dy = tm[2] 348 if yr is None: 349 yr = tm[0] 350 return date(yr, mo, dy) 351 352 def normalize_date(val, iso8601=True): 353 if not val: 354 return '' 355 if type(val) == str: 356 val = parse_date(val) 357 if iso8601: 358 return "%4d-%02d-%02d" % (val.year, val.month, val.day) 359 return "%02d %s %4d" % (val.day, _num2str[val.month], val.year) 360
