| 1 | import re |
|---|
| 2 | import math |
|---|
| 3 | import string |
|---|
| 4 | from asap._asap import selector as _selector, srctype |
|---|
| 5 | from asap.utils import unique, _to_list |
|---|
| 6 | |
|---|
| 7 | class selector(_selector): |
|---|
| 8 | """ |
|---|
| 9 | A selection object to be applied to scantables to restrict the |
|---|
| 10 | scantables to specific rows. |
|---|
| 11 | """ |
|---|
| 12 | fields = ["pols", "ifs", "beams", "scans", "cycles", "name", "query", "types", "rows"] |
|---|
| 13 | |
|---|
| 14 | def __init__(self, *args, **kw): |
|---|
| 15 | if len(args) == 1: |
|---|
| 16 | if isinstance(args[0], self.__class__) \ |
|---|
| 17 | or isinstance(args[0], _selector): |
|---|
| 18 | _selector.__init__(self, args[0]) |
|---|
| 19 | else: |
|---|
| 20 | raise TypeError("Argument can only be a selector object") |
|---|
| 21 | else: |
|---|
| 22 | _selector.__init__(self) |
|---|
| 23 | for k,v in kw.items(): |
|---|
| 24 | if k in self.fields: |
|---|
| 25 | func = getattr(self, "set_%s" % k) |
|---|
| 26 | func(v) |
|---|
| 27 | |
|---|
| 28 | def reset(self): |
|---|
| 29 | """ |
|---|
| 30 | Unset all selections. |
|---|
| 31 | """ |
|---|
| 32 | self._reset() |
|---|
| 33 | |
|---|
| 34 | def is_empty(self): |
|---|
| 35 | """ |
|---|
| 36 | Has anything been set? |
|---|
| 37 | """ |
|---|
| 38 | return self._empty() |
|---|
| 39 | |
|---|
| 40 | def set_polarisations(self, pols=[]): |
|---|
| 41 | """ |
|---|
| 42 | Set the polarisations to be selected in the scantable. |
|---|
| 43 | Parameters: |
|---|
| 44 | pols: a list of integers of 0-3, or strings, e.g ["I","Q"]. |
|---|
| 45 | Default [] is no selection |
|---|
| 46 | Example: |
|---|
| 47 | sel = selector() |
|---|
| 48 | # These are equivalent if data is 'linear' |
|---|
| 49 | sel.set_polarisations(["XX","Re(XY)"]) |
|---|
| 50 | sel.set_polarisations([0,2]) |
|---|
| 51 | # reset the polarisation selection |
|---|
| 52 | sel.set_polarisations() |
|---|
| 53 | |
|---|
| 54 | """ |
|---|
| 55 | vec = _to_list(pols, str) or _to_list(pols, int) |
|---|
| 56 | if isinstance(vec, list): # is an empty and/or valid vector |
|---|
| 57 | if len(vec) and isinstance(vec[-1],str): |
|---|
| 58 | self._setpolstrings(vec) |
|---|
| 59 | return |
|---|
| 60 | self._setpols(vec) |
|---|
| 61 | else: |
|---|
| 62 | raise TypeError('Unknown pol type. Please use [0,1...] or ["XX","YY"...]') |
|---|
| 63 | |
|---|
| 64 | # for the americans |
|---|
| 65 | set_polarizations = set_polarisations |
|---|
| 66 | # for the lazy |
|---|
| 67 | set_pols = set_polarisations |
|---|
| 68 | |
|---|
| 69 | def set_ifs(self, ifs=[]): |
|---|
| 70 | """ |
|---|
| 71 | Set a sequence of IF numbers (0-based). |
|---|
| 72 | Parameters: |
|---|
| 73 | ifs: a list of integers. Default [] is to unset the selection. |
|---|
| 74 | """ |
|---|
| 75 | vec = _to_list(ifs, int) |
|---|
| 76 | if isinstance(vec,list): |
|---|
| 77 | self._setifs(vec) |
|---|
| 78 | else: |
|---|
| 79 | raise TypeError('Unknown IFno type. Use lists of integers.') |
|---|
| 80 | |
|---|
| 81 | def set_scans(self, scans=[]): |
|---|
| 82 | """ |
|---|
| 83 | Set a sequence of Scan numbers (0-based). |
|---|
| 84 | Parameters: |
|---|
| 85 | scans: a list of integers. Default [] is to unset the selection. |
|---|
| 86 | """ |
|---|
| 87 | vec = _to_list(scans, int) |
|---|
| 88 | if isinstance(vec,list): |
|---|
| 89 | self._setscans(vec) |
|---|
| 90 | else: |
|---|
| 91 | raise TypeError('Unknown Scan number type. Use lists of integers.') |
|---|
| 92 | |
|---|
| 93 | def set_beams(self, beams=[]): |
|---|
| 94 | """ |
|---|
| 95 | Set a sequence of Beam numbers (0-based). |
|---|
| 96 | Parameters: |
|---|
| 97 | beams: a list of integers. Default [] is to unset the selection. |
|---|
| 98 | """ |
|---|
| 99 | vec = _to_list(beams, int) |
|---|
| 100 | if isinstance(vec,list): |
|---|
| 101 | self._setbeams(vec) |
|---|
| 102 | else: |
|---|
| 103 | raise TypeError('Unknown Beam number type. Use lists of integers.') |
|---|
| 104 | |
|---|
| 105 | def set_cycles(self, cycles=[]): |
|---|
| 106 | """ |
|---|
| 107 | Set a sequence of IF numbers (0-based). |
|---|
| 108 | Parameters: |
|---|
| 109 | cycless: a list of integers. Default [] is to unset the selection. |
|---|
| 110 | """ |
|---|
| 111 | vec = _to_list(cycles, int) |
|---|
| 112 | if isinstance(vec,list): |
|---|
| 113 | self._setcycles(vec) |
|---|
| 114 | else: |
|---|
| 115 | raise TypeError('Unknown Cycle number type. Use lists of integers.') |
|---|
| 116 | |
|---|
| 117 | |
|---|
| 118 | def set_name(self, name): |
|---|
| 119 | """ |
|---|
| 120 | Set a selection based on a name. This can be a unix pattern , e.g. "*_R" |
|---|
| 121 | Parameters: |
|---|
| 122 | name: a string containing a source name or pattern |
|---|
| 123 | Examples: |
|---|
| 124 | # select all reference scans which start with "Orion" |
|---|
| 125 | selection.set_name("Orion*_R") |
|---|
| 126 | """ |
|---|
| 127 | if isinstance(name, str): |
|---|
| 128 | self._setname(name) |
|---|
| 129 | else: |
|---|
| 130 | raise TypeError('name must be a string') |
|---|
| 131 | |
|---|
| 132 | def set_tsys(self, tsysmin=0.0, tsysmax=None): |
|---|
| 133 | """ |
|---|
| 134 | Select by Tsys range. |
|---|
| 135 | Parameters: |
|---|
| 136 | tsysmin: the lower threshold. Default 0.0 |
|---|
| 137 | tsysmax: the upper threshold. Default None. |
|---|
| 138 | Examples: |
|---|
| 139 | # select all spectra with Tsys <= 500.0 |
|---|
| 140 | selection.set_tsys(tsysmax=500.0) |
|---|
| 141 | |
|---|
| 142 | """ |
|---|
| 143 | taql = "SELECT FROM $1 WHERE TSYS[0] >= %f" % (tsysmin) |
|---|
| 144 | if isinstance(tsysmax, float): |
|---|
| 145 | taql = taql + " AND TSYS[0] <= %f" % ( tsysmax) |
|---|
| 146 | self._settaql(taql) |
|---|
| 147 | |
|---|
| 148 | def set_query(self, query): |
|---|
| 149 | """ |
|---|
| 150 | Select by Column query. Power users only! |
|---|
| 151 | Example: |
|---|
| 152 | # select all off scans with integration times over 60 seconds. |
|---|
| 153 | selection.set_query("SRCTYPE == PSOFF AND INTERVAL > 60.0") |
|---|
| 154 | """ |
|---|
| 155 | rx = re.compile("((SRCTYPE *[!=][=] *)([a-zA-Z.]+))", re.I) |
|---|
| 156 | for r in rx.findall(query): |
|---|
| 157 | sval = None |
|---|
| 158 | stype = r[-1].lower() |
|---|
| 159 | if stype.find('srctype.') == -1: |
|---|
| 160 | stype = ".".join(["srctype", stype]) |
|---|
| 161 | try: |
|---|
| 162 | sval = eval(stype) |
|---|
| 163 | sval = "%s%d" % (r[1], sval) |
|---|
| 164 | except: |
|---|
| 165 | continue |
|---|
| 166 | query = query.replace(r[0], sval) |
|---|
| 167 | taql = "SELECT FROM $1 WHERE " + query |
|---|
| 168 | self._settaql(taql) |
|---|
| 169 | |
|---|
| 170 | def set_order(self, order): |
|---|
| 171 | """ |
|---|
| 172 | Set the order the scantable should be sorted by. |
|---|
| 173 | Parameters: |
|---|
| 174 | order: The list of column names to sort by in order |
|---|
| 175 | """ |
|---|
| 176 | self._setorder(order) |
|---|
| 177 | |
|---|
| 178 | def set_rows(self, rows=[]): |
|---|
| 179 | """ |
|---|
| 180 | Set a sequence of row numbers (0-based). Power users Only! |
|---|
| 181 | NOTICE row numbers can be changed easily by sorting, |
|---|
| 182 | prior selection, etc. |
|---|
| 183 | Parameters: |
|---|
| 184 | rows: a list of integers. Default [] is to unset the selection. |
|---|
| 185 | """ |
|---|
| 186 | vec = _to_list(rows, int) |
|---|
| 187 | if isinstance(vec,list): |
|---|
| 188 | self._setrows(vec) |
|---|
| 189 | else: |
|---|
| 190 | raise TypeError('Unknown row number type. Use lists of integers.') |
|---|
| 191 | |
|---|
| 192 | def set_types(self, types=[]): |
|---|
| 193 | """ |
|---|
| 194 | Set a sequence of source types. |
|---|
| 195 | Parameters: |
|---|
| 196 | types: a list of integers. Default [] is to unset the selection. |
|---|
| 197 | """ |
|---|
| 198 | vec = _to_list(types, int) |
|---|
| 199 | if isinstance(vec,list): |
|---|
| 200 | self._settypes(vec) |
|---|
| 201 | else: |
|---|
| 202 | raise TypeError('Unknown row number type. Use lists of integers.') |
|---|
| 203 | |
|---|
| 204 | def set_msselection_field(self, selection): |
|---|
| 205 | """ |
|---|
| 206 | Set a field selection in msselection syntax. The msselection |
|---|
| 207 | suppports the following syntax: |
|---|
| 208 | |
|---|
| 209 | pattern match: |
|---|
| 210 | - UNIX style pattern match for source name using '*' |
|---|
| 211 | (compatible with set_name) |
|---|
| 212 | |
|---|
| 213 | field id selection: |
|---|
| 214 | - simple number in string ('0', '1', etc.) |
|---|
| 215 | - range specification using '~' ('0~1', etc.) |
|---|
| 216 | - range specification using '>' or '<' in combination |
|---|
| 217 | with '=' ('>=1', '<3', etc.) |
|---|
| 218 | |
|---|
| 219 | comma separated multiple selection: |
|---|
| 220 | - selections can be combined by using ',' ('0,>1', |
|---|
| 221 | 'mysource*,2~4', etc.) |
|---|
| 222 | """ |
|---|
| 223 | selection_list = map(string.strip, selection.split(',')) |
|---|
| 224 | query_list = list(self.generate_query(selection_list)) |
|---|
| 225 | if len(query_list) > 0: |
|---|
| 226 | query = 'SELECT FROM $1 WHERE ' + ' || '.join(query_list) |
|---|
| 227 | self._settaql(query) |
|---|
| 228 | |
|---|
| 229 | def generate_query(self, selection_list): |
|---|
| 230 | for s in selection_list: |
|---|
| 231 | if s.isdigit() or re.match('^[<>]=?[0-9]*$', s) or \ |
|---|
| 232 | re.match('^[0-9]+~[0-9]+$', s): |
|---|
| 233 | #print '"%s" is ID selection using < or <='%(s) |
|---|
| 234 | a = FieldIdRegexGenerator(s) |
|---|
| 235 | yield '(%s)'%(a.get_regex()) |
|---|
| 236 | elif len(s) > 0: |
|---|
| 237 | #print '"%s" is UNIX style pattern match'%(s) |
|---|
| 238 | yield '(SRCNAME == pattern(\'%s\'))'%(s) |
|---|
| 239 | |
|---|
| 240 | def get_scans(self): |
|---|
| 241 | return list(self._getscans()) |
|---|
| 242 | def get_cycles(self): |
|---|
| 243 | return list(self._getcycles()) |
|---|
| 244 | def get_beams(self): |
|---|
| 245 | return list(self._getbeams()) |
|---|
| 246 | def get_ifs(self): |
|---|
| 247 | return list(self._getifs()) |
|---|
| 248 | def get_pols(self): |
|---|
| 249 | return list(self._getpols()) |
|---|
| 250 | def get_poltypes(self): |
|---|
| 251 | return list(self._getpoltypes()) |
|---|
| 252 | def get_order(self): |
|---|
| 253 | return list(self._getorder()) |
|---|
| 254 | def get_types(self): |
|---|
| 255 | return list(self._gettypes()) |
|---|
| 256 | def get_rows(self): |
|---|
| 257 | return list(self._getrows()) |
|---|
| 258 | def get_query(self): |
|---|
| 259 | prefix = "SELECT FROM $1 WHERE " |
|---|
| 260 | return self._gettaql().replace(prefix, "") |
|---|
| 261 | |
|---|
| 262 | def get_name(self): |
|---|
| 263 | print "NYI" |
|---|
| 264 | s = self._gettaql() |
|---|
| 265 | return |
|---|
| 266 | def __str__(self): |
|---|
| 267 | out = "" |
|---|
| 268 | d = {"SCANNO": self.get_scans(), |
|---|
| 269 | "CYCLENO": self.get_cycles(), |
|---|
| 270 | "BEAMNO": self.get_beams(), |
|---|
| 271 | "IFNO": self.get_ifs(), |
|---|
| 272 | "Pol Type": self.get_poltypes(), |
|---|
| 273 | "POLNO": self.get_pols(), |
|---|
| 274 | "QUERY": self.get_query(), |
|---|
| 275 | "SRCTYPE": self.get_types(), |
|---|
| 276 | "ROWS": self.get_rows(), |
|---|
| 277 | "Sort Order": self.get_order() |
|---|
| 278 | } |
|---|
| 279 | for k,v in d.iteritems(): |
|---|
| 280 | if v: |
|---|
| 281 | out += "%s: %s\n" % (k, v) |
|---|
| 282 | if len(out): |
|---|
| 283 | return out[:-1] |
|---|
| 284 | else: |
|---|
| 285 | return out |
|---|
| 286 | |
|---|
| 287 | def __add__(self, other): |
|---|
| 288 | """ |
|---|
| 289 | Merge two selections. |
|---|
| 290 | """ |
|---|
| 291 | if self.is_empty(): |
|---|
| 292 | return selector(other) |
|---|
| 293 | elif other.is_empty(): |
|---|
| 294 | return selector(self) |
|---|
| 295 | union = selector() |
|---|
| 296 | gets = [[self._getscans(), other._getscans(), union._setscans], |
|---|
| 297 | [self._getcycles(), other._getcycles(),union._setcycles], |
|---|
| 298 | [self._getbeams(), other._getbeams(), union._setbeams], |
|---|
| 299 | [self._getifs(), other._getifs(), union._setifs], |
|---|
| 300 | [self._getpols(), other._getpols(), union._setpols]] |
|---|
| 301 | for v in gets: |
|---|
| 302 | vec = list(v[0]+v[1]) |
|---|
| 303 | vec.sort() |
|---|
| 304 | v[2](unique(vec)) |
|---|
| 305 | q = other.get_query() |
|---|
| 306 | qs = self.get_query() |
|---|
| 307 | if len(q) and len(qs): |
|---|
| 308 | union.set_query(qs +" AND " + q) |
|---|
| 309 | else: |
|---|
| 310 | if len(q): |
|---|
| 311 | union.set_query(q) |
|---|
| 312 | elif len(qs): |
|---|
| 313 | union.set_query(qs) |
|---|
| 314 | return union |
|---|
| 315 | |
|---|
| 316 | class FieldIdRegexGenerator(object): |
|---|
| 317 | def __init__(self, pattern): |
|---|
| 318 | if pattern.isdigit(): |
|---|
| 319 | self.regex = 'FIELDNAME == regex(\'.+__%s$\')'%(pattern) |
|---|
| 320 | else: |
|---|
| 321 | self.regex = None |
|---|
| 322 | ineq = None |
|---|
| 323 | if pattern.find('<') >= 0: |
|---|
| 324 | ineq = '<' |
|---|
| 325 | s = pattern.strip().lstrip(ineq).lstrip('=') |
|---|
| 326 | if not s.isdigit(): |
|---|
| 327 | raise RuntimeError('Invalid syntax: %s'%(pattern)) |
|---|
| 328 | self.id = int(s) + (-1 if pattern.find('=') < 0 else 0) |
|---|
| 329 | self.template = string.Template('FIELDNAME == regex(\'.+__${reg}$\')') |
|---|
| 330 | elif pattern.find('>') >= 0: |
|---|
| 331 | ineq = '>' |
|---|
| 332 | s = pattern.strip().lstrip(ineq).lstrip('=') |
|---|
| 333 | if not s.isdigit(): |
|---|
| 334 | raise RuntimeError('Invalid syntax: %s'%(pattern)) |
|---|
| 335 | self.id = int(s) + (-1 if pattern.find('=') >= 0 else 0) |
|---|
| 336 | self.template = string.Template('FIELDNAME == regex(\'.+__[0-9]+$\') && FIELDNAME != regex(\'.+__${reg}$\')') |
|---|
| 337 | elif pattern.find('~') >= 0: |
|---|
| 338 | s = map(string.strip, pattern.split('~')) |
|---|
| 339 | if len(s) == 2 and s[0].isdigit() and s[1].isdigit(): |
|---|
| 340 | self.id = [int(s[0])-1,int(s[1])] |
|---|
| 341 | else: |
|---|
| 342 | raise RuntimeError('Invalid syntax: %s'%(pattern)) |
|---|
| 343 | self.template = string.Template('FIELDNAME == regex(\'.+__${reg}$\') && FIELDNAME != regex(\'.+__${optreg}$\')') |
|---|
| 344 | else: |
|---|
| 345 | raise RuntimeError('Invalid syntax: %s'%(pattern)) |
|---|
| 346 | #print 'self.id=',self.id |
|---|
| 347 | |
|---|
| 348 | def get_regex(self): |
|---|
| 349 | if self.regex is not None: |
|---|
| 350 | # 'X' |
|---|
| 351 | return self.regex |
|---|
| 352 | elif isinstance(self.id, list): |
|---|
| 353 | # 'X~Y' |
|---|
| 354 | return self.template.safe_substitute(reg=self.__compile(self.id[1]), |
|---|
| 355 | optreg=self.__compile(self.id[0])) |
|---|
| 356 | else: |
|---|
| 357 | # '<(=)X' or '>(=)X' |
|---|
| 358 | return self.template.safe_substitute(reg=self.__compile(self.id)) |
|---|
| 359 | |
|---|
| 360 | def __compile(self, idx): |
|---|
| 361 | pattern = '' |
|---|
| 362 | if idx >= 0: |
|---|
| 363 | if idx < 10: |
|---|
| 364 | num_digits = 1 |
|---|
| 365 | else: |
|---|
| 366 | num_digits = int(math.log10(idx)) + 1 |
|---|
| 367 | numerics = [] |
|---|
| 368 | modulo = 10 |
|---|
| 369 | divider = 1 |
|---|
| 370 | for i in xrange(num_digits): |
|---|
| 371 | numerics.append(int((idx % modulo) / divider)) |
|---|
| 372 | modulo *= 10 |
|---|
| 373 | divider *= 10 |
|---|
| 374 | #print 'numerics=',numerics |
|---|
| 375 | if num_digits == 1: |
|---|
| 376 | if numerics[0] == 0: |
|---|
| 377 | pattern = '0' |
|---|
| 378 | else: |
|---|
| 379 | pattern = '[0-%s]'%(numerics[0]) |
|---|
| 380 | elif num_digits == 2: |
|---|
| 381 | pattern_list = ['[0-9]'] |
|---|
| 382 | pattern_list.append('[1-%s][0-9]'%(numerics[1]-1)) |
|---|
| 383 | if numerics[0] == 0: |
|---|
| 384 | pattern_list.append('%s%s'%(numerics[1],numerics[0])) |
|---|
| 385 | else: |
|---|
| 386 | pattern_list.append('%s[0-%s]'%(numerics[1],numerics[0])) |
|---|
| 387 | pattern = '(%s)'%('|'.join(pattern_list)) |
|---|
| 388 | elif num_digits == 3: |
|---|
| 389 | pattern_list = ['[0-9]','[1-9][0-9]'] |
|---|
| 390 | if numerics[2] == 2: |
|---|
| 391 | pattern_list.append('1[0-9][0-9]') |
|---|
| 392 | elif numerics[2] > 2: |
|---|
| 393 | pattern_list.append('[1-%s][0-9][0-9]'%(numerics[2]-1)) |
|---|
| 394 | if numerics[1] == 0: |
|---|
| 395 | if numerics[0] == 0: |
|---|
| 396 | pattern_list.append('%s00'%(numerics[2])) |
|---|
| 397 | else: |
|---|
| 398 | pattern_list.append('%s0[0-%s]'%(numerics[2],numerics[0])) |
|---|
| 399 | else: |
|---|
| 400 | pattern_list.append('%s[0-%s][0-9]'%(numerics[2],numerics[1]-1)) |
|---|
| 401 | if numerics[0] == 0: |
|---|
| 402 | pattern_list.append('%s%s%s'%(numerics[2],numerics[1],numerics[0])) |
|---|
| 403 | else: |
|---|
| 404 | pattern_list.append('%s%s[0-%s]'%(numerics[2],numerics[1],numerics[0])) |
|---|
| 405 | pattern = '(%s)'%('|'.join(pattern_list)) |
|---|
| 406 | else: |
|---|
| 407 | raise RuntimeError('ID > 999 is not supported') |
|---|
| 408 | pattern = '' |
|---|
| 409 | else: |
|---|
| 410 | raise RuntimeError('ID must be >= 0') |
|---|
| 411 | return pattern |
|---|
