| 1 | #!/usr/bin/env python |
|---|
| 2 | |
|---|
| 3 | #************************************************************************************** |
|---|
| 4 | #csmldataIface.py |
|---|
| 5 | #contains classes for interfacing with various files |
|---|
| 6 | #currently supports cdunif (NetCDF, PP, Grib(untested)) And Nappy (NASAAmes) |
|---|
| 7 | #use by instantiating the factory class: DataInterface |
|---|
| 8 | #v0.00 30th November 2005 |
|---|
| 9 | #Dominic Lowe, BADC |
|---|
| 10 | #************************************************************************************** |
|---|
| 11 | |
|---|
| 12 | |
|---|
| 13 | import cdms |
|---|
| 14 | import nappy |
|---|
| 15 | import string |
|---|
| 16 | import sys |
|---|
| 17 | |
|---|
| 18 | class DataInterface(object): |
|---|
| 19 | #Use DataInterface and setInterfaceType to instantiate the correct |
|---|
| 20 | #subclass for data |
|---|
| 21 | def __init__(self): |
|---|
| 22 | self.iface ='None' |
|---|
| 23 | |
|---|
| 24 | def setInterfaceType(self,interfaceType): |
|---|
| 25 | # function returns approprate data interface |
|---|
| 26 | #set interfaceType: should correspond to available datainterface subclasses |
|---|
| 27 | self.iface =interfaceType |
|---|
| 28 | if self.iface == 'nappy': |
|---|
| 29 | return NappyInterface() |
|---|
| 30 | elif self.iface == 'cdunif': |
|---|
| 31 | return cdunifInterface() |
|---|
| 32 | |
|---|
| 33 | |
|---|
| 34 | def getUnknownInterfaceType(self, filename): |
|---|
| 35 | #if the interface type is not known at the time of instantiation, then use |
|---|
| 36 | #this function to examine the file and return the correct interface (if it exists). |
|---|
| 37 | fileExtension = str(filename)[-3:] |
|---|
| 38 | if fileExtension == '.nc': |
|---|
| 39 | return cdunifInterface() |
|---|
| 40 | if fileExtension == '.qxf': |
|---|
| 41 | return cdunifInterface() |
|---|
| 42 | elif fileExtension == '.pp': |
|---|
| 43 | return cdunifInterface() |
|---|
| 44 | elif fileExtension == 'ctl': |
|---|
| 45 | return cdunifInterface() |
|---|
| 46 | else: |
|---|
| 47 | try: |
|---|
| 48 | nappy.readFFI(filename) in [1001,1010,1020,2010,2110,2160,2310,3010,4010] |
|---|
| 49 | return NappyInterface() |
|---|
| 50 | except: |
|---|
| 51 | print "Could not establish file type" |
|---|
| 52 | print "See csmldataiface.py" |
|---|
| 53 | |
|---|
| 54 | |
|---|
| 55 | |
|---|
| 56 | class AbstractDI(object): |
|---|
| 57 | #Abstract data interface class |
|---|
| 58 | #does nothing, but contains templates for methods required for a data interface class |
|---|
| 59 | #individual interfaces (e.g NappyInterface) should override these methods |
|---|
| 60 | |
|---|
| 61 | def __init__(self): |
|---|
| 62 | self.extractType='' |
|---|
| 63 | self.extractPrefix = '' |
|---|
| 64 | |
|---|
| 65 | def openFile(self, filename): |
|---|
| 66 | #opens file |
|---|
| 67 | self.file='' |
|---|
| 68 | |
|---|
| 69 | def closeFile(self): |
|---|
| 70 | #closes file |
|---|
| 71 | self.file.close() |
|---|
| 72 | |
|---|
| 73 | |
|---|
| 74 | |
|---|
| 75 | class NappyInterface(AbstractDI): |
|---|
| 76 | # Data Interface for Nappy (NASA Ames Processing in Python) |
|---|
| 77 | |
|---|
| 78 | def __init__(self): |
|---|
| 79 | self.extractType='NASAAmesExtract' |
|---|
| 80 | self.extractPrefix = '_naextract_' |
|---|
| 81 | |
|---|
| 82 | def openFile(self, filename): |
|---|
| 83 | #print 'opening NA file: ' + str(filename) |
|---|
| 84 | self.file=nappy.openNAFile(filename) |
|---|
| 85 | #print 'reading data....' |
|---|
| 86 | #self.file.readData() |
|---|
| 87 | #print 'nappyopen ' + filename |
|---|
| 88 | |
|---|
| 89 | def getListOfAxes(self): |
|---|
| 90 | axes=self.file.XNAME |
|---|
| 91 | #print 'before units stripped' + str(axes) |
|---|
| 92 | axes=self.stripunits(axes) |
|---|
| 93 | #print 'after units stripped' + str(axes) |
|---|
| 94 | return axes |
|---|
| 95 | |
|---|
| 96 | def setAxis(self,axis): |
|---|
| 97 | axes = self.getListOfAxes() |
|---|
| 98 | self.axisstub=axes.index(axis) |
|---|
| 99 | |
|---|
| 100 | def getAxisAttribute(self, att): |
|---|
| 101 | #need to do something here...? maybe |
|---|
| 102 | pass |
|---|
| 103 | return attValue |
|---|
| 104 | |
|---|
| 105 | def getTimeUnits(self): |
|---|
| 106 | axes = self.getListOfAxes() |
|---|
| 107 | for axis in axes: |
|---|
| 108 | if string.find(string.upper(axis),'SECONDS SINCE') != -1: |
|---|
| 109 | #found possible time axis. |
|---|
| 110 | if axis[-3:]=='UTC': |
|---|
| 111 | units =string.lower(axis[:-4]) #hack! |
|---|
| 112 | units=units.replace('/','-') #try and clean it up |
|---|
| 113 | else: |
|---|
| 114 | units=string.lower(axis) |
|---|
| 115 | break |
|---|
| 116 | elif string.find(string.upper(axis),'HOURS SINCE') != -1: |
|---|
| 117 | #found possible time axis. |
|---|
| 118 | units =(str(axis)) |
|---|
| 119 | break |
|---|
| 120 | elif string.find(string.upper(axis),'DAYS SINCE') != -1: |
|---|
| 121 | #found possible time axis. |
|---|
| 122 | units =(str(axis)) |
|---|
| 123 | break |
|---|
| 124 | |
|---|
| 125 | #revisit with udunits python library? |
|---|
| 126 | return units |
|---|
| 127 | |
|---|
| 128 | |
|---|
| 129 | def getDataForAxis(self): |
|---|
| 130 | |
|---|
| 131 | if self.file.X == None: |
|---|
| 132 | #print 'reading data....' |
|---|
| 133 | self.file.readData() |
|---|
| 134 | |
|---|
| 135 | if type(self.file.X[1])==list: |
|---|
| 136 | #if len(self.file.X) > 0: |
|---|
| 137 | data = self.file.X[self.axisstub] |
|---|
| 138 | else: |
|---|
| 139 | data =self.file.X |
|---|
| 140 | return data |
|---|
| 141 | |
|---|
| 142 | def getSizeOfAxis(self,axis): |
|---|
| 143 | |
|---|
| 144 | #check this function is okay |
|---|
| 145 | #is time always the first dimension in NA?? |
|---|
| 146 | axes = self.getListOfAxes() |
|---|
| 147 | axisPosition=axes.index(axis) |
|---|
| 148 | #print "axis position" + str( axisPosition) |
|---|
| 149 | #print "NX" + str(self.file.NX) |
|---|
| 150 | try : |
|---|
| 151 | axisSize=self.file.NX[axisPosition-1] |
|---|
| 152 | except: |
|---|
| 153 | axisSize ='Unknown axis size' |
|---|
| 154 | return axisSize |
|---|
| 155 | |
|---|
| 156 | def getListofVariables(self): |
|---|
| 157 | variableList=self.stripunits(self.file.VNAME) |
|---|
| 158 | return variableList |
|---|
| 159 | |
|---|
| 160 | def setVariable(self,varname): |
|---|
| 161 | vlist=self.getListofVariables() |
|---|
| 162 | self.varstub=vlist.index(varname) |
|---|
| 163 | |
|---|
| 164 | def getVariableAxes(self): |
|---|
| 165 | #hmm, now with Nasa Ames the Axis will be the same for all variables. |
|---|
| 166 | #so just use the getListOfAxes function again |
|---|
| 167 | #I think... check this! |
|---|
| 168 | varAxesList=self.getListOfAxes() |
|---|
| 169 | return varAxesList |
|---|
| 170 | |
|---|
| 171 | def getVariableAttribute(self,attName): |
|---|
| 172 | if attName =='units': |
|---|
| 173 | #strip the units (attribute) from the variable |
|---|
| 174 | unitslist=self.getUnits(self.file.VNAME) |
|---|
| 175 | attribValue = unitslist[self.varstub] |
|---|
| 176 | try: |
|---|
| 177 | attribValue = unitslist[self.varstub] |
|---|
| 178 | except: |
|---|
| 179 | attribValue = 'unknown' |
|---|
| 180 | else: |
|---|
| 181 | attribValue = 'unknown' |
|---|
| 182 | return attribValue |
|---|
| 183 | |
|---|
| 184 | def getDataForVar(self): |
|---|
| 185 | #NOTE TO SELF: |
|---|
| 186 | #Review this function (and in fact all of nasa ames data interface...) |
|---|
| 187 | if self.file.V == None: |
|---|
| 188 | #print 'reading data....' |
|---|
| 189 | self.file.readData() |
|---|
| 190 | |
|---|
| 191 | try: |
|---|
| 192 | if type(self.file.V[1])==list: |
|---|
| 193 | data = self.file.V[self.varstub] |
|---|
| 194 | #else: |
|---|
| 195 | # data =self.file.X |
|---|
| 196 | # print data |
|---|
| 197 | return data |
|---|
| 198 | except: |
|---|
| 199 | data = self.file.X |
|---|
| 200 | # print data |
|---|
| 201 | return data |
|---|
| 202 | |
|---|
| 203 | def getArraySizeOfVar(self): |
|---|
| 204 | #iterates through all dimensions in variable to get array size i.e a 3x3x3 array would have a size of 27 |
|---|
| 205 | |
|---|
| 206 | dimlist=self.file.NX |
|---|
| 207 | varsize =1 |
|---|
| 208 | for item in dimlist: |
|---|
| 209 | varsize = varsize * item |
|---|
| 210 | #print "VARSISZE" + str(varsize) |
|---|
| 211 | return varsize |
|---|
| 212 | |
|---|
| 213 | def getShapeOfVar(self): |
|---|
| 214 | #this should return a list. |
|---|
| 215 | varShape = [] |
|---|
| 216 | for item in self.file.NX: |
|---|
| 217 | varShape.append(item) |
|---|
| 218 | return varShape |
|---|
| 219 | |
|---|
| 220 | def getLowLimits(self): |
|---|
| 221 | lowlims = "" |
|---|
| 222 | for i in range (0, len(self.file.NX)): |
|---|
| 223 | #for now, assume low limit is always of form 1 1 1 .. |
|---|
| 224 | lowlims =lowlims + str(1) +' ' |
|---|
| 225 | return lowlims |
|---|
| 226 | |
|---|
| 227 | def getHighLimits(self): |
|---|
| 228 | highlims = "" |
|---|
| 229 | for i in range (0, len(self.file.NX)): |
|---|
| 230 | dimValue = self.file.NX[i] |
|---|
| 231 | highlims =highlims + str(dimValue) +' ' |
|---|
| 232 | return highlims |
|---|
| 233 | |
|---|
| 234 | |
|---|
| 235 | def stripunits(self,listtostrip): |
|---|
| 236 | #strips units of measure from list |
|---|
| 237 | #eg ['Universal time (hours)', 'Altitude (km)', 'Latitude (degrees)', 'Longitude (degrees)'] |
|---|
| 238 | #becomes ['Universal time', 'Altitude', 'Latitude', 'Longitude'] |
|---|
| 239 | cleanlist = [] |
|---|
| 240 | for item in listtostrip: |
|---|
| 241 | openbracket=string.find(item,'(') |
|---|
| 242 | if openbracket != -1: |
|---|
| 243 | #if brackets exist, strip units. |
|---|
| 244 | item=item[:openbracket-1] |
|---|
| 245 | cleanlist.append(item) |
|---|
| 246 | return cleanlist |
|---|
| 247 | |
|---|
| 248 | def getUnits(self,listwithunits): |
|---|
| 249 | #gets units from list |
|---|
| 250 | #eg ['Universal time (hours)', 'Altitude (km)', 'Latitude (degrees)', 'Longitude (degrees)'] |
|---|
| 251 | #becomes ['hours', 'km', 'degrees', 'degrees'] |
|---|
| 252 | unitlist=[] |
|---|
| 253 | for item in listwithunits: |
|---|
| 254 | openbracket=string.find(item,'(') |
|---|
| 255 | item = item[openbracket+1:-1] |
|---|
| 256 | unitlist.append(item) |
|---|
| 257 | return unitlist |
|---|
| 258 | |
|---|
| 259 | def getTimes(self): |
|---|
| 260 | #This function attempts to determine the time axis and read the time data |
|---|
| 261 | #it may well not manage it. |
|---|
| 262 | axes = self.getListOfAxes() |
|---|
| 263 | for axis in axes: |
|---|
| 264 | if string.find(string.upper(axis),'TIME') != -1: |
|---|
| 265 | #found possible time axis. |
|---|
| 266 | self.setAxis(axis) |
|---|
| 267 | times=self.getDataForAxis() |
|---|
| 268 | break |
|---|
| 269 | elif string.find(string.upper(axis),'SECONDS SINCE') != -1: |
|---|
| 270 | #found possible time axis. |
|---|
| 271 | self.setAxis(axis) |
|---|
| 272 | times=self.getDataForAxis() |
|---|
| 273 | break |
|---|
| 274 | elif string.find(string.upper(axis),'HOURS SINCE') != -1: |
|---|
| 275 | #found possible time axis. |
|---|
| 276 | self.setAxis(axis) |
|---|
| 277 | times=self.getDataForAxis() |
|---|
| 278 | break |
|---|
| 279 | elif string.find(string.upper(axis),'DAYS SINCE') != -1: |
|---|
| 280 | #found possible time axis. |
|---|
| 281 | self.setAxis(axis) |
|---|
| 282 | times=self.getDataForAxis() |
|---|
| 283 | break |
|---|
| 284 | return times |
|---|
| 285 | |
|---|
| 286 | |
|---|
| 287 | |
|---|
| 288 | class cdunifInterface(AbstractDI): |
|---|
| 289 | #Data Interface for cdunif (netcdf & pp formats & grib (not tested with grib) |
|---|
| 290 | |
|---|
| 291 | def __init__(self): |
|---|
| 292 | #these are just temporary values until we can determine whether the |
|---|
| 293 | #file is netcdf pp or grib |
|---|
| 294 | self.extractType='cdunifExtract' |
|---|
| 295 | self.extractPrefix = '_cdunifextract_' |
|---|
| 296 | |
|---|
| 297 | def openFile(self, filename): |
|---|
| 298 | self.file=cdms.open(filename) |
|---|
| 299 | #print 'cdunifopen ' + filename |
|---|
| 300 | |
|---|
| 301 | #now we have the file name can properly determine extractType/Prefix |
|---|
| 302 | fileExtension = str(filename)[-3:] |
|---|
| 303 | if fileExtension == '.nc': |
|---|
| 304 | self.extractType = 'NetCDFExtract' |
|---|
| 305 | self.extractPrefix = '_ncextract_' |
|---|
| 306 | elif fileExtension == '.qxf': |
|---|
| 307 | self.extractType = 'NetCDFExtract' |
|---|
| 308 | self.extractPrefix = '_ncextract_' |
|---|
| 309 | elif fileExtension == '.pp': |
|---|
| 310 | self.extractType = 'PPExtract' |
|---|
| 311 | self.extractPrefix = '_ppextract_' |
|---|
| 312 | elif fileExtension == 'ctl': |
|---|
| 313 | self.extractType = 'GRIBExtract' |
|---|
| 314 | self.extractPrefix = '_gribextract_' |
|---|
| 315 | |
|---|
| 316 | def getListOfAxes(self): |
|---|
| 317 | axes=self.file.dimensions.keys() |
|---|
| 318 | return axes |
|---|
| 319 | |
|---|
| 320 | def getSizeOfAxis(self,axis): |
|---|
| 321 | axisSize=self.file.dimensions[axis] |
|---|
| 322 | return axisSize |
|---|
| 323 | |
|---|
| 324 | def getListofVariables(self): |
|---|
| 325 | variableList=self.file.variables.keys() |
|---|
| 326 | return variableList |
|---|
| 327 | |
|---|
| 328 | def setAxis(self,axis): |
|---|
| 329 | self.axisobj=self.file.getAxis(axis) |
|---|
| 330 | |
|---|
| 331 | def getAxisAttribute(self, att): |
|---|
| 332 | attValue=self.axisobj.attributes[att] |
|---|
| 333 | return attValue |
|---|
| 334 | |
|---|
| 335 | def getTimeUnits(self): |
|---|
| 336 | #this does the same as getAxisAttribute, but is a separate function as different formats handle time differently. |
|---|
| 337 | return self.getAxisAttribute('units') |
|---|
| 338 | |
|---|
| 339 | def getDataForAxis(self): |
|---|
| 340 | data = self.axisobj.getValue() |
|---|
| 341 | return data |
|---|
| 342 | |
|---|
| 343 | def setVariable(self,varname): |
|---|
| 344 | self.varobj=self.file.variables[varname] |
|---|
| 345 | |
|---|
| 346 | def getVariableAxes(self): |
|---|
| 347 | varAxesList=self.varobj.getAxisIds() |
|---|
| 348 | return varAxesList |
|---|
| 349 | |
|---|
| 350 | def getVariableAttribute(self,attName): |
|---|
| 351 | |
|---|
| 352 | #atts=self.varobj.getattributes(attName) |
|---|
| 353 | #print atts |
|---|
| 354 | if attName == 'long_name': |
|---|
| 355 | try: |
|---|
| 356 | attribValue = self.varobj.long_name |
|---|
| 357 | except: |
|---|
| 358 | attribValue='' |
|---|
| 359 | elif attName == 'units': |
|---|
| 360 | try: |
|---|
| 361 | attribValue = self.varobj.units |
|---|
| 362 | except: |
|---|
| 363 | attribValue='' |
|---|
| 364 | |
|---|
| 365 | return attribValue |
|---|
| 366 | |
|---|
| 367 | def getDataForVar(self): |
|---|
| 368 | data = self.varobj.getValue() |
|---|
| 369 | return data |
|---|
| 370 | |
|---|
| 371 | def getSubsetOfDataForVar(self, **kwargs): |
|---|
| 372 | #takes keyword args defining subset eg |
|---|
| 373 | #subset=getSubsetOfDataForVar(latitude=(0.,10.0), longitude=(90, 100.0)) |
|---|
| 374 | for key in kwargs: |
|---|
| 375 | if key == 'longitude': #this test needs to be more robust... |
|---|
| 376 | print kwargs[key] |
|---|
| 377 | if kwargs[key][0] > kwargs[key][1]: |
|---|
| 378 | #subsetting greenwich meridian around 0 |
|---|
| 379 | lonMin = kwargs[key][0] |
|---|
| 380 | lonMax =kwargs[key][1] |
|---|
| 381 | kwargs[key]=(0.0, lonMax) |
|---|
| 382 | sel=cdms.selectors.Selector(**kwargs) |
|---|
| 383 | subset1=self.file(self.varobj.name,sel) |
|---|
| 384 | kwargs[key]=(lonMin,359.9999) |
|---|
| 385 | sel=cdms.selectors.Selector(**kwargs) |
|---|
| 386 | subset2=self.file(self.varobj.name,sel) |
|---|
| 387 | #concatenate arrays along longitude |
|---|
| 388 | longitudeAxis=subset1.getAxisIndex('longitude') # this needs to be more robust test. |
|---|
| 389 | subset = cdms.MV.concatenate([subset1,subset2],axis=longitudeAxis) |
|---|
| 390 | else: |
|---|
| 391 | sel=cdms.selectors.Selector(**kwargs) |
|---|
| 392 | subset=self.file(self.varobj.name,sel) |
|---|
| 393 | data = subset |
|---|
| 394 | return data |
|---|
| 395 | |
|---|
| 396 | def getArraySizeOfVar(self): |
|---|
| 397 | #iterates through all dimensions in variable to get array size i.e a 3x3x3 array would have a size of 27 |
|---|
| 398 | var = self.varobj |
|---|
| 399 | size = var.shape |
|---|
| 400 | varsize = 1 |
|---|
| 401 | for item in size: |
|---|
| 402 | varsize = item *varsize |
|---|
| 403 | return varsize |
|---|
| 404 | |
|---|
| 405 | def getShapeOfVar(self): |
|---|
| 406 | varShape = [] |
|---|
| 407 | for item in self.varobj.shape: |
|---|
| 408 | varShape.append(item) |
|---|
| 409 | return varShape |
|---|
| 410 | |
|---|
| 411 | def getLowLimits(self): |
|---|
| 412 | dimNames = self.varobj.getAxisIds() |
|---|
| 413 | lowlims = "" |
|---|
| 414 | for i in range (1, len(dimNames)): |
|---|
| 415 | #for now, assume low limit is always of form 1 1 1 .. |
|---|
| 416 | lowlims =lowlims + str(1) +' ' |
|---|
| 417 | return lowlims |
|---|
| 418 | |
|---|
| 419 | def getHighLimits(self): |
|---|
| 420 | dimNames = self.varobj.getAxisIds() |
|---|
| 421 | highlims = "" |
|---|
| 422 | for i in range (1, len(dimNames)): |
|---|
| 423 | dimValue = self.file.dimensions[dimNames[i]] |
|---|
| 424 | highlims =highlims + str(dimValue) +' ' |
|---|
| 425 | return highlims |
|---|
| 426 | |
|---|
| 427 | |
|---|
| 428 | |
|---|
| 429 | |
|---|
