| 1 | |
|---|
| 2 | from xceptions import baseException |
|---|
| 3 | import shelve, glob, os, uuid, collections, string |
|---|
| 4 | from utils_wb import workbook |
|---|
| 5 | import util_varGroups |
|---|
| 6 | import util_gen |
|---|
| 7 | |
|---|
| 8 | |
|---|
| 9 | tab2freq = {u'CMIP5_cfOff':'subhr', u'CORDEX_mon':'mon', u'SPECS_day':'day', u'CMIP5_day':'day', \ |
|---|
| 10 | u'PMIP3_OImon':'mon', u'CORDEX_day':'day', u'CMIP5_LImon':'mon', u'CMIP5_OImon':'mon', \ |
|---|
| 11 | u'CMIP5_Lmon':'mon', u'CMIP5_3hr':'3hr', u'CMIP5_Omon':'mon', u'PMIP3_OIclim':'monClim', \ |
|---|
| 12 | u'PMIP3_fx':'fx', u'CORDEX_fx':'fx', u'PMIP3_LImon':'mon', u'CMIP5_6hrPlev':'6hr', u'PMIP3_Lmon':'mon', \ |
|---|
| 13 | u'PMIP3_Amon':'mon', u'SPECS_Omon':'mon', u'CCMI1_fixed':'fx', u'PMIP3_Aclim':'monClim', u'CMIP5_6hrLev':'6hr', \ |
|---|
| 14 | u'CMIP5_Oclim':'monClim', u'PMIP3_LIclim':'monClim', u'CCMI1_monthly':'mon', u'CMIP5_fx':'fx', \ |
|---|
| 15 | u'CMIP5_cfDay':'day', u'CORDEX_6h':'6hr', u'PMIP3_day':'day', u'SPECS_OImon':'mon', u'CMIP5_cfMon':'mon', \ |
|---|
| 16 | u'CORDEX_sem':'monClim', u'SPECS_6hr':'6hr', u'CMIP5_cfSites':'subhr', u'CCMI1_hourly':'hr', u'CMIP5_aero':'mon', \ |
|---|
| 17 | u'CMIP5_Amon':'mon', u'PMIP3_Omon':'mon', u'CCMI1_daily':'day', u'SPECS_fx':'fx', u'PMIP3_Lclim':'monClim', \ |
|---|
| 18 | u'DCPP-day':'day', \ |
|---|
| 19 | u'PMIP3_Oclim':'monClim', u'SPECS_Amon':'mon', u'SPECS_Lmon':'mon', u'CMIP5_cf3hr':'3hr', u'CORDEX_3h':'3hr', \ |
|---|
| 20 | 'OImon':'mon', 'Omon_oth':'mon', 'Omon':'mon', 'Oyr':'yr', 'Omon_3d':'mon', \ |
|---|
| 21 | u'CCMI1_annual':'yr', u'CMIP5_Oyr':'yr', \ |
|---|
| 22 | 'Oclim':'monClim', \ |
|---|
| 23 | 'cfMon':'mon', \ |
|---|
| 24 | u'cf3hr':'3hr', u'Amon':'mon', u'aero':'mon', u'aero_3d':'mon', u'6hrPlev':'6hr', u'aero_oth':'mon', \ |
|---|
| 25 | u'cf3hr_grid':'3hr', u'LImon':'mon', u'cfSites':'subhr', u'day_oth':'day', u'cfDay_2d':'day', u'day':'day' } |
|---|
| 26 | |
|---|
| 27 | ## |
|---|
| 28 | ## want to match up with input/vars_... and where is cmv uid fixed ???? |
|---|
| 29 | ## |
|---|
| 30 | ## |
|---|
| 31 | ## need to migrate new vars into input/vars_ ... especially from LUMIP ... |
|---|
| 32 | ## |
|---|
| 33 | ## new variables which need to be added have been identified. No check for modifications (and no clear workflow here)... |
|---|
| 34 | ## ned to check .. a few appear spurious .. and then generate csv records ... for manual copy ... |
|---|
| 35 | ## |
|---|
| 36 | ## |
|---|
| 37 | ## |
|---|
| 38 | ## add AerChemMip and VIACS to stuff scanned and checked below ..... |
|---|
| 39 | |
|---|
| 40 | def getSubGroup(s, var,dims,ttl,ix): |
|---|
| 41 | sg = None |
|---|
| 42 | if s == 'Oyr' and ix < 60: |
|---|
| 43 | sg = 'Oyr_3dtr' |
|---|
| 44 | elif s == 'Amon': |
|---|
| 45 | if string.find( dims, 'lev' ) == -1 and string.find( dims, 'longitude latitude') != -1: |
|---|
| 46 | sg = 'Amon_2d' |
|---|
| 47 | else: |
|---|
| 48 | sg = 'Amon_oth' |
|---|
| 49 | elif s == 'Omon': |
|---|
| 50 | if string.find( dims, 'longitude latitude olevel' ) != -1: |
|---|
| 51 | sg = 'Omon_3d' |
|---|
| 52 | else: |
|---|
| 53 | sg = 'Omon_oth' |
|---|
| 54 | elif s == 'cfMon': |
|---|
| 55 | if ttl[:5] in ['ISCCP','PARAS','CALIP']: |
|---|
| 56 | sg = 'cfMon_sim' |
|---|
| 57 | elif var[-3:] == 'co2': |
|---|
| 58 | if string.find( dims, 'longitude latitude alev' ) != -1: |
|---|
| 59 | sg = 'cfMon_3dmod' |
|---|
| 60 | else: |
|---|
| 61 | sg = 'cfMon_2dmod' |
|---|
| 62 | else: |
|---|
| 63 | sg = 'cfMon_3dstd' |
|---|
| 64 | elif s == 'cfDay': |
|---|
| 65 | bits = '|'.split( dims ) |
|---|
| 66 | |
|---|
| 67 | if string.find( dims, 'longitude|latitude' ) != -1 and len(bits) > 2 and bits[2] in ['alev','alevel','alevhalf','alt40','plev7','alt40']: |
|---|
| 68 | sg = 'cfDay_3d' |
|---|
| 69 | else: |
|---|
| 70 | sg = 'cfDay_2d' |
|---|
| 71 | elif s == 'cf3hr': |
|---|
| 72 | if dims.find( 'longitude|latitude' ) != -1: |
|---|
| 73 | sg = 'cf3hr_grid' |
|---|
| 74 | else: |
|---|
| 75 | sg = 'cf3hr_sim' |
|---|
| 76 | elif s == 'aero': |
|---|
| 77 | if dims.find( 'longitude|latitude|alevel' ) != -1: |
|---|
| 78 | sg = 'aero_3d' |
|---|
| 79 | else: |
|---|
| 80 | sg = 'aero_oth' |
|---|
| 81 | elif s == 'day': |
|---|
| 82 | if ix < 28: |
|---|
| 83 | sg = 'day_ss' |
|---|
| 84 | else: |
|---|
| 85 | sg = 'day_oth' |
|---|
| 86 | return sg |
|---|
| 87 | def getSubGroupO(sn, var,dims,ttl,ix): |
|---|
| 88 | sg = None |
|---|
| 89 | if sn == 'Oyr' and ix < 65: |
|---|
| 90 | sg = 'Oyr_3dtr' |
|---|
| 91 | elif sn == 'Omon': |
|---|
| 92 | if string.find( dims, 'longitude latitude olevel' ) != -1: |
|---|
| 93 | sg = 'Omon_3d' |
|---|
| 94 | else: |
|---|
| 95 | sg = 'Omon_oth' |
|---|
| 96 | elif sn == 'cfMon': |
|---|
| 97 | if ttl[:5] in ['ISCCP','PARAS','CALIP']: |
|---|
| 98 | sg = 'cfMon_sim' |
|---|
| 99 | elif var[-3:] == 'co2': |
|---|
| 100 | if string.find( dims, 'longitude latitude alev' ) != -1: |
|---|
| 101 | sg = 'cfMon_3dmod' |
|---|
| 102 | else: |
|---|
| 103 | sg = 'cfMon_2dmod' |
|---|
| 104 | else: |
|---|
| 105 | sg = 'cfMon_3dstd' |
|---|
| 106 | elif sn == 'cfDay': |
|---|
| 107 | bits = ' '.split( dims ) |
|---|
| 108 | if string.find( dims, 'longitude latitude' ) != -1 and len(bits) > 2 and bits[2] in ['alev','alevel','alevhalf','alt40','plev7','alt40']: |
|---|
| 109 | sg = 'cfDay_3d' |
|---|
| 110 | else: |
|---|
| 111 | sg = 'cfDay_2d' |
|---|
| 112 | elif sn == 'cf3hr': |
|---|
| 113 | if dims.find( 'longitude latitude' ) != -1: |
|---|
| 114 | sg = 'cf3hr_grid' |
|---|
| 115 | else: |
|---|
| 116 | sg = 'cf3hr_sim' |
|---|
| 117 | return sg |
|---|
| 118 | |
|---|
| 119 | class ref(object): |
|---|
| 120 | def __init__(self,sdir='inSh'): |
|---|
| 121 | dir1 = '/data/tmp/svn3/exarch/CMIP6dreqbuild/trunk/src/workbook' |
|---|
| 122 | wb = workbook( '%s/%s' % (dir1,'inputs/vars_20160721.xls') ) |
|---|
| 123 | self.vars = {} |
|---|
| 124 | s1 = wb.book.sheet_by_name(u'var') |
|---|
| 125 | su = set() |
|---|
| 126 | for i in range( s1.nrows ): |
|---|
| 127 | r = s1.row(i) |
|---|
| 128 | assert r[0] not in self.vars, 'DUPLICATE VARIABLE: %s' % str(r) |
|---|
| 129 | self.vars[r[0].value] = (r[9].value,r[10].value) |
|---|
| 130 | fl = sorted( glob.glob( '%s/sh__newVar_*' % sdir ) ) |
|---|
| 131 | for f in fl: |
|---|
| 132 | sh = shelve.open( f, 'r' ) |
|---|
| 133 | ks = [k for k in sh.keys() if k[0] != '_'] |
|---|
| 134 | for k in ks: |
|---|
| 135 | if k not in self.vars: |
|---|
| 136 | print 'MISSING NEW: %s, %s, %s' % (k,f,str(sh[k])) |
|---|
| 137 | |
|---|
| 138 | class cmvRef(object): |
|---|
| 139 | def __init__(self): |
|---|
| 140 | self.byUid = {} |
|---|
| 141 | self.byTab = collections.defaultdict( dict ) |
|---|
| 142 | for l in open( 'ingest/cmvRef_01.beta.32.csv' ).readlines(): |
|---|
| 143 | lab, tab, ttl, uid = l.strip().split( '\t' ) |
|---|
| 144 | assert uid not in self.byUid, 'DUPLICATE UID: %s' % uid |
|---|
| 145 | self.byUid[uid] = (lab,tab,ttl) |
|---|
| 146 | if lab in self.byTab[tab]: |
|---|
| 147 | print 'ERROR.cmvRef.001: duplicate variable name in cmv table: %s,%s' % (tab,lab) |
|---|
| 148 | self.byTab[tab][lab] = uid |
|---|
| 149 | |
|---|
| 150 | class revTabChk(object): |
|---|
| 151 | |
|---|
| 152 | def __init__(self,sdir='inSh'): |
|---|
| 153 | """Review and consolidate revised tables, as ingested into shelves, and annotate records""" |
|---|
| 154 | assert os.path.isdir( sdir ), 'Input directory not found: %s' % inSH |
|---|
| 155 | self.ref = refTabChk() |
|---|
| 156 | fl = sorted( glob.glob( '%s/sh__so_*' % sdir ) ) |
|---|
| 157 | tref = util_varGroups.varGroupSs() |
|---|
| 158 | tref.loadGroups() |
|---|
| 159 | tprobs = set() |
|---|
| 160 | self.sh = shelve.open( '%s/sh__consol01_revTabs' % sdir, 'n' ) |
|---|
| 161 | self.sh['__cols__'] = ['var', 'table', 'mip', 'vid', 'priority','vgid'] |
|---|
| 162 | self.sh['__info__'] = {'label':'revTable', 'title':'Revised variable table records generated by ingest.util_anal.revTabChk'} |
|---|
| 163 | cc = collections.defaultdict( int ) |
|---|
| 164 | vgmaps = {'Oyr':'OMIP-Oyr','OImon':'SIMIP-seaicemon'} |
|---|
| 165 | kkk = 0 |
|---|
| 166 | for f in fl: |
|---|
| 167 | # |
|---|
| 168 | # CMIP5 and OMIP tables go in as reference tables, and so should not be duplicated here. |
|---|
| 169 | # |
|---|
| 170 | ss = set() |
|---|
| 171 | sso = set() |
|---|
| 172 | if f.find( 'CMIP5' ) == -1 and f.find( 'OMIP' ) == -1: |
|---|
| 173 | sh = shelve.open( f, 'r' ) |
|---|
| 174 | for k in sh.keys(): |
|---|
| 175 | if k[0] != '_': |
|---|
| 176 | assert k not in self.sh, 'Duplicate key in revised table items: %s' % k |
|---|
| 177 | |
|---|
| 178 | vgl = sh[k][1].replace( '_', '-' ) |
|---|
| 179 | ss.add(vgl) |
|---|
| 180 | if vgl in ['cf3hr-grid', 'cfMon-3dmod', 'cfMon-2dmod']: |
|---|
| 181 | if vgl not in sso: |
|---|
| 182 | print 'Warning: skipping ',sh[k] |
|---|
| 183 | sso.add( vgl ) |
|---|
| 184 | vgid = '__unset__01__' |
|---|
| 185 | else: |
|---|
| 186 | vgl1 = vgmaps.get( vgl, vgl ) |
|---|
| 187 | if vgl1 not in tref.uidByLabel: |
|---|
| 188 | if 'CMIP5-%s' % vgl1 in tref.uidByLabel: |
|---|
| 189 | vgl1 = 'CMIP5-%s' % vgl1 |
|---|
| 190 | |
|---|
| 191 | if vgl1 not in tref.uidByLabel: |
|---|
| 192 | print 'SEVERE.requestvar.00001: group not found: %s: %s' % (vgl,f) |
|---|
| 193 | vgid = '__vgid_not_found__' |
|---|
| 194 | else: |
|---|
| 195 | vgid = tref.uidByLabel[vgl1] |
|---|
| 196 | cc[vgl1] += 1 |
|---|
| 197 | vid = sh[k][3] |
|---|
| 198 | self.sh[k] = sh[k][:] + [vgid,] |
|---|
| 199 | if vid in self.ref.cmv: |
|---|
| 200 | sg = self.ref.cmv[vid][-1] |
|---|
| 201 | if sg not in [ '',None]: |
|---|
| 202 | sgs = sg.split('_')[-1] |
|---|
| 203 | vgl2 = '%s-%s' % (vgl1,sgs) |
|---|
| 204 | if vgl2 in tref.uidByLabel: |
|---|
| 205 | vg2 = tref.uidByLabel[vgl2] |
|---|
| 206 | rr = sh[k][:] + [vg2,] |
|---|
| 207 | rr[2] = rr[2] + '_' + sgs |
|---|
| 208 | k2 = str( uuid.uuid1() ) |
|---|
| 209 | self.sh[k2] = rr |
|---|
| 210 | else: |
|---|
| 211 | print 'INFO.sg.00050: not found; ',vgl2, sg, sh[k] |
|---|
| 212 | else: |
|---|
| 213 | print 'INFO.sg.00060: no subgroup; ',vgl, sg, sh[k] |
|---|
| 214 | else: |
|---|
| 215 | print 'INFO.sg.00070: no cmv; ',sh[k] |
|---|
| 216 | kkk += 1 |
|---|
| 217 | sh.close() |
|---|
| 218 | print f, ss |
|---|
| 219 | self.sh.close() |
|---|
| 220 | for k in sorted( tref.uidByLabel.keys() ): |
|---|
| 221 | print k, cc[k] |
|---|
| 222 | for k in cc: |
|---|
| 223 | if k not in tref.uidByLabel: |
|---|
| 224 | print 'ERROR: ',k, cc[k] |
|---|
| 225 | print 'TOTAL VARS REQUESTED HERE: ',kkk |
|---|
| 226 | |
|---|
| 227 | class refTabChk(object): |
|---|
| 228 | |
|---|
| 229 | def __init__(self,sdir='inSh'): |
|---|
| 230 | """Review reference tables, as ingested into shelves, and annotate records""" |
|---|
| 231 | assert os.path.isdir( sdir ), 'Input directory not found: %s' % inSH |
|---|
| 232 | fl = sorted( glob.glob( '%s/sh__refso_*' % sdir ) ) |
|---|
| 233 | tprobs = set() |
|---|
| 234 | self.sh = shelve.open( '%s/sh__consol01_refTabs' % sdir, 'n' ) |
|---|
| 235 | self.cmv = dict() |
|---|
| 236 | self.cmvtv = dict() |
|---|
| 237 | self.tabInfo = util_gen.tableInfo() |
|---|
| 238 | self.sh['__cols__'] = self.tabInfo.oh |
|---|
| 239 | self.sh['__info__'] = {'label':'refTable', 'title':'Reference variable table records generated by ingest.util_anal.refTabChk'} |
|---|
| 240 | itab = self.tabInfo.oh.index( 'mipTable' ) |
|---|
| 241 | icmv = self.tabInfo.oh.index( 'var' ) |
|---|
| 242 | iix = self.tabInfo.oh.index( 'rowIndex' ) |
|---|
| 243 | iv = self.tabInfo.oh.index( 'out_name' ) |
|---|
| 244 | ivid = self.tabInfo.oh.index( 'vid' ) |
|---|
| 245 | idim = self.tabInfo.oh.index( 'dimensions' ) |
|---|
| 246 | iln = self.tabInfo.oh.index( 'title' ) |
|---|
| 247 | iss = self.tabInfo.oh.index( 'ssect' ) |
|---|
| 248 | igpid = self.tabInfo.oh.index( 'gpid' ) |
|---|
| 249 | vref = util_varGroups.ref() |
|---|
| 250 | tref = util_varGroups.varGroupSs() |
|---|
| 251 | tref.loadGroups() |
|---|
| 252 | vmsgs = set() |
|---|
| 253 | |
|---|
| 254 | gtm = set() |
|---|
| 255 | sss = set() |
|---|
| 256 | for f in fl: |
|---|
| 257 | self.actions = collections.defaultdict(int) |
|---|
| 258 | self.file = f |
|---|
| 259 | sh = shelve.open( f, 'r' ) |
|---|
| 260 | ks = [k for k in sh if k[0] != '_'] |
|---|
| 261 | for k in ks: |
|---|
| 262 | rr = sh[k][:] |
|---|
| 263 | uid = rr[0] |
|---|
| 264 | try: |
|---|
| 265 | tab = rr[itab] |
|---|
| 266 | tab = tref.mapGroupName(tab,'') |
|---|
| 267 | var = rr[iv] |
|---|
| 268 | if var not in vref.vars: |
|---|
| 269 | if var not in vmsgs: |
|---|
| 270 | print 'SEVERE.vars.00010: var not found: %s [%s]' % (var,tab) |
|---|
| 271 | vmsgs.add( var ) |
|---|
| 272 | vid = '-' |
|---|
| 273 | else: |
|---|
| 274 | vid = vref.vars[var][0] |
|---|
| 275 | |
|---|
| 276 | if tab not in tref.groupset: |
|---|
| 277 | print 'SEVERE.tables.00001: tab not found: %s' % tab |
|---|
| 278 | gpid = '-' |
|---|
| 279 | else: |
|---|
| 280 | gpid = tref.groupset[tab][0] |
|---|
| 281 | gtm.add( (tab,tref.groupset[tab]) ) |
|---|
| 282 | except: |
|---|
| 283 | print k, sh[k] |
|---|
| 284 | print itab,iv |
|---|
| 285 | raise |
|---|
| 286 | if tab not in tab2freq: |
|---|
| 287 | tprobs.add(tab) |
|---|
| 288 | rr[igpid] = gpid |
|---|
| 289 | rr[ivid] = vid |
|---|
| 290 | rr[-1] = int( float(rr[-1]) ) |
|---|
| 291 | sg = '' |
|---|
| 292 | ## check to see if this table has sections, and identify which section this variable is in if appropriate |
|---|
| 293 | print 'INFO.sg.01001: ',tab,tab in self.tabInfo.tabChildren |
|---|
| 294 | if rr[iss] != '': |
|---|
| 295 | sg = rr[iss] |
|---|
| 296 | sg1 = sg.replace( '_', '-' ) |
|---|
| 297 | if sg1 in ['cf3hr-grid','Amon-oth']: |
|---|
| 298 | sgid = '_na_' |
|---|
| 299 | else: |
|---|
| 300 | assert sg1 in tref.groupset, 'Subgroup not found in groupset: %s,%s' % (tab,sg1) |
|---|
| 301 | sgid = tref.groupset[sg1][0] |
|---|
| 302 | rr[iss] = (sg,sgid) |
|---|
| 303 | print 'INFO.sg.00003: ',sg,sgid |
|---|
| 304 | elif tab in self.tabInfo.tabChildren: |
|---|
| 305 | if tab in ['Omon','Oyr']: |
|---|
| 306 | sg = getSubGroupO(tab, var,rr[idim],rr[iln],rr[iix]) |
|---|
| 307 | else: |
|---|
| 308 | sg = getSubGroup(tab, var,rr[idim],rr[iln],rr[iix]) |
|---|
| 309 | print 'INFO.sg.00001: ',tab,var,rr[idim],rr[iix], sg |
|---|
| 310 | if sg != None: |
|---|
| 311 | sg1 = sg.replace( '_', '-' ) |
|---|
| 312 | if sg1 in ['Amon-oth']: |
|---|
| 313 | sgid = '_na_' |
|---|
| 314 | else: |
|---|
| 315 | assert sg1 in tref.groupset, 'Subgroup not found in groupset: %s, %s' % (tab,sg) |
|---|
| 316 | sgid = tref.groupset[sg1][0] |
|---|
| 317 | rr[iss] = (sg,sgid) |
|---|
| 318 | if (sg,sgid) not in sss: |
|---|
| 319 | sss.add( (sg,sgid) ) |
|---|
| 320 | print 'INFO.sg.00002: ',sg,sgid |
|---|
| 321 | else: |
|---|
| 322 | assert rr[iss] == '', 'Corrupted subgroup element: %s,%s,%s' % (iss,rr[iss],str(rr)) |
|---|
| 323 | |
|---|
| 324 | self.cmv[uid] = (tab, var, rr[idim], rr[iix], rr[iln],sg) |
|---|
| 325 | self.cmvtv[(tab,var)] = uid |
|---|
| 326 | if var == 'intppcalc': |
|---|
| 327 | print rr |
|---|
| 328 | if var in {'*','include Oyr 3D tracers'}: |
|---|
| 329 | print 'ERROR.copy.00890: ',f,rr |
|---|
| 330 | if k != uid: |
|---|
| 331 | print 'SEVERE.uid.00020: uid/k mismatch: ',k,rr |
|---|
| 332 | |
|---|
| 333 | if var == 'msftmyz': |
|---|
| 334 | print 'INFO.msftmyz.00001: ',vid, rr |
|---|
| 335 | self.sh[k] = rr[:] |
|---|
| 336 | print gtm |
|---|
| 337 | self.sh['__source__'] = fl |
|---|
| 338 | self.sh.close() |
|---|
| 339 | print 'UNMAPPED tabs (to freq): ',tprobs |
|---|
| 340 | |
|---|
| 341 | class dynGrps(object): |
|---|
| 342 | def __init__(self,rftc): |
|---|
| 343 | sdir = 'inSh' |
|---|
| 344 | sh = shelve.open( '%s/sh__requestScoping' % sdir ) |
|---|
| 345 | sh0 = shelve.open( '%s/sh__requestScoping_0' % sdir, 'r' ) |
|---|
| 346 | for k in sh0: |
|---|
| 347 | mip, rec = sh0[k] |
|---|
| 348 | print mip,rec |
|---|
| 349 | |
|---|
| 350 | class varGroupChk(object): |
|---|
| 351 | cmip5Tables = [u'3hr', u'6hrLev', u'6hrPlev', u'Amon', u'LImon', u'Lmon', u'OImon', u'Oclim', u'Omon', u'Oyr', u'aero', u'cf3hr', u'cfDay', u'cfMon', u'cfOff', u'cfSites', u'day', u'fx'] |
|---|
| 352 | def __init__(self,sdir='inSh'): |
|---|
| 353 | """Review var groups, as ingested into shelves, and annotate records""" |
|---|
| 354 | assert os.path.isdir( sdir ), 'Input directory not found: %s' % inSH |
|---|
| 355 | self.ref = ref() |
|---|
| 356 | self.shl = {} |
|---|
| 357 | self.group2uid = {} |
|---|
| 358 | self.vgss = util_varGroups.varGroupSs() |
|---|
| 359 | self.vgss.loadGroups() |
|---|
| 360 | self.vgss.setUid() |
|---|
| 361 | self.vgss.checkMaps() |
|---|
| 362 | self.varsetoo = None |
|---|
| 363 | self.grps = set() |
|---|
| 364 | self.tbls = set() |
|---|
| 365 | self.frqs = set() |
|---|
| 366 | self.nnew = 0 |
|---|
| 367 | self.nc5 = 0 |
|---|
| 368 | self.noth = 0 |
|---|
| 369 | self.ots = set() |
|---|
| 370 | isFirst = True |
|---|
| 371 | #self.sh = shelve.open( '%s/sh__consol01_grp' % sdir, 'n' ) |
|---|
| 372 | self.sh = shelve.open( '%s/sh__consol01_groupItems' % sdir, 'n' ) |
|---|
| 373 | self.shdvg = shelve.open( '%s/sh__consol01_dynVarGroup_requestVars' % sdir, 'n' ) |
|---|
| 374 | |
|---|
| 375 | self.shdvg['__info__'] = {'label':'dynVarGroup_requestVars', 'title':'Request variable records for dynamically generated variable groups (through variable names listed in request scoping). Generated by ingest.util_anal.varGroupChk'} |
|---|
| 376 | self.shdvg['__cols__'] = ['vid', 'title', 'label', 'priority', 'vgid', 'mip'] |
|---|
| 377 | |
|---|
| 378 | self.sh['__info__'] = {'label':'GroupItemsBeta', 'title':'Group Item records generated by ingest.util_anal.varGroupChk'} |
|---|
| 379 | self.sh['__cols__'] = ['group', 'var', 'table', 'freq', 'descriptionEx', 'shape', 'levels', 'tstyle', 'mask', 'misc', 'mip', 'uid', 'oldv', 'rowIndex', 'var2', 'new', 'gpid', 'vkey', 'vid'] |
|---|
| 380 | self.mg = dict() |
|---|
| 381 | self.checkRequestedGroups() |
|---|
| 382 | |
|---|
| 383 | fl = sorted( glob.glob( '%s/sh__grp_*' % sdir ) ) |
|---|
| 384 | for f in fl: |
|---|
| 385 | self.scanFile(f,isFirst) |
|---|
| 386 | isFirst = False |
|---|
| 387 | |
|---|
| 388 | self.sh.close() |
|---|
| 389 | self.shdvg.close() |
|---|
| 390 | if self.varsetoo != None: |
|---|
| 391 | self.varsetoo.close() |
|---|
| 392 | |
|---|
| 393 | sh = shelve.open( '%s/sh__requestScoping_1' % sdir,'n' ) |
|---|
| 394 | for k in self.group2uid: |
|---|
| 395 | mip, uid, new = self.group2uid[k] |
|---|
| 396 | sh['%s__1' % str(k)] = (mip, uid ) |
|---|
| 397 | sh.close() |
|---|
| 398 | |
|---|
| 399 | for kkk in sorted(self.mg.keys()): |
|---|
| 400 | print 'WARNING: possible problem identifying group: ',self.mg[kkk] |
|---|
| 401 | print 'groups: ',self.grps |
|---|
| 402 | print 'tables: ',self.tbls |
|---|
| 403 | print 'frequencies: ',self.frqs |
|---|
| 404 | print 'nrefs to new: %s, cmip5: %s, other: %s' % (self.nnew, self.nc5, self.noth) |
|---|
| 405 | print 'other tables: ',self.ots |
|---|
| 406 | for k in self.shl: |
|---|
| 407 | self.shl[k].close() |
|---|
| 408 | |
|---|
| 409 | def scanFile(self,f,isFirst): |
|---|
| 410 | self.actions = collections.defaultdict(int) |
|---|
| 411 | self.file = f |
|---|
| 412 | self.shl[f] = shelve.open( f, 'r' ) |
|---|
| 413 | self.mip = self.shl[f]['__info__']['label'] |
|---|
| 414 | if isFirst: |
|---|
| 415 | print self.shl[f]['__cols__'] |
|---|
| 416 | ks = [k for k in self.shl[f].keys() if k[0] != '_'] |
|---|
| 417 | |
|---|
| 418 | for k in ks: |
|---|
| 419 | r = self.shl[f][k] |
|---|
| 420 | self.grps.add(r[0]) |
|---|
| 421 | self.tbls.add(r[2]) |
|---|
| 422 | self.frqs.add(r[3]) |
|---|
| 423 | var = r[1].strip() |
|---|
| 424 | if var != r[1]: |
|---|
| 425 | print 'WARNING.blanks.0001: *%s* and *%s (%s,%s)' % (var,r[1],r[0],f) |
|---|
| 426 | extra = {} |
|---|
| 427 | if r[2] == 'new' or r[2][:2] == 'em': |
|---|
| 428 | extra['mode'] = 'new' |
|---|
| 429 | self.nnew +=1 |
|---|
| 430 | if var not in self.ref.vars: |
|---|
| 431 | if var.find('_') != -1: |
|---|
| 432 | v2 = var.replace('_','') |
|---|
| 433 | if v2 in self.ref.vars: |
|---|
| 434 | print 'WARN: %s should be replaced with %s' % (var,v2) |
|---|
| 435 | print 'ERROR.missing.0001: variable not found: ',var,r[0],f |
|---|
| 436 | else: |
|---|
| 437 | if r[9] == '': |
|---|
| 438 | p1 = -1 |
|---|
| 439 | else: |
|---|
| 440 | p1 = int( r[9] ) |
|---|
| 441 | p2 = int( self.ref.vars[var][1] ) |
|---|
| 442 | if p1 not in [1,2,3]: |
|---|
| 443 | if p2 in [1,2,3]: |
|---|
| 444 | extra['priority'] = p2 |
|---|
| 445 | else: |
|---|
| 446 | print 'ERROR.priority.0001: %s (%s::%s), ref: %s, templ: %s' % (var,r[0],f,p2,p1) |
|---|
| 447 | |
|---|
| 448 | elif r[2] in self.cmip5Tables or r[2][:5] == 'CMIP5' and r[2][6:] in self.cmip5Tables: |
|---|
| 449 | extra['mode'] = 'CMIP5' |
|---|
| 450 | self.nc5 += 1 |
|---|
| 451 | if var not in self.ref.vars: |
|---|
| 452 | print 'ERROR.missing.0003: variable not found: ',var,r[0],f |
|---|
| 453 | else: |
|---|
| 454 | extra['mode'] = 'OTHER MIP' |
|---|
| 455 | self.noth += 1 |
|---|
| 456 | self.ots.add( r[2] ) |
|---|
| 457 | if var not in self.ref.vars: |
|---|
| 458 | print 'ERROR.missing.0002: variable not found: ',var,r[0],f |
|---|
| 459 | ## |
|---|
| 460 | ## but this is the MIP variable id ...... need the CMOR variable ID ?? ... |
|---|
| 461 | ## |
|---|
| 462 | vid = self.ref.vars.get(var,[None,])[0] |
|---|
| 463 | self.consol(k,r,vid,extra) |
|---|
| 464 | for k in sorted( self.actions.keys() ): |
|---|
| 465 | print 'ACTIONS: %s: %s -- %s' % (f,k,self.actions[k]) |
|---|
| 466 | |
|---|
| 467 | |
|---|
| 468 | def varsets(self, thisuid, i1, refuid, i2, title ): |
|---|
| 469 | if self.varsetoo == None: |
|---|
| 470 | self.varsetoo = open('varsets.txt','w' ) |
|---|
| 471 | self.varsetoo.write( '\t'.join( [thisuid, str(i1), refuid, str(i2), title] ) + '\n' ) |
|---|
| 472 | |
|---|
| 473 | def consol(self,rk,rr,vid,extra): |
|---|
| 474 | ##['Short name of group', 'Variable short name', 'Table', 'Frequency', 'Description extension (optional)', 'Shape', 'Levels', 'Time mean, point or climatology', 'Mask (optional)', 'Priority', 'MIP','uid','rowIndex', 'Prev. Var Name'] |
|---|
| 475 | ## |
|---|
| 476 | ## following is provided by sx2.py: |
|---|
| 477 | ##['group', 'var', 'table', 'freq', 'descriptionEx', 'shape', 'levels', 'tstyle', 'mask', 'misc', 'mip', 'uid', 'rowIndex', 'new', 'gpid', 'vkey', 'vid'] |
|---|
| 478 | ## need to add vid (info found above), vgid, "new" flag, (also above), and "vkey". |
|---|
| 479 | |
|---|
| 480 | rset = [] |
|---|
| 481 | r = list(rr) |
|---|
| 482 | if 'priority' in extra: |
|---|
| 483 | r[9] = extra['priority'] |
|---|
| 484 | elif r[9] == '': |
|---|
| 485 | r[9] = -1 |
|---|
| 486 | else: |
|---|
| 487 | r[9] = int( r[9] ) |
|---|
| 488 | |
|---|
| 489 | if not self.vgss.matchGroup(r[0],self.mip): |
|---|
| 490 | print 'ERROR.group.00006: group not found: %s [%s]' % (r[0],self.mip) |
|---|
| 491 | self.mg[r[0]] = (self.mip,r) |
|---|
| 492 | gpid = '__noGroupFound__' |
|---|
| 493 | new = 0 |
|---|
| 494 | else: |
|---|
| 495 | print 'INFO.group.00007: group found: %s (%s):: %s, %s' % (r[0],self.mip,self.vgss.groupName, self.vgss.groupMatch ) |
|---|
| 496 | if r[0] in self.group2uid: |
|---|
| 497 | assert self.vgss.groupMatchRes == self.group2uid[r[0]], 'Mismatch in group lookup ..%s [%s,%s]' % (r[0], str(self.vgss.groupMatchRes), str(self.group2uid[r[0]]) ) |
|---|
| 498 | else: |
|---|
| 499 | self.group2uid[r[0]] = self.vgss.groupMatchRes |
|---|
| 500 | gpid = str(self.vgss.groupMatchRes[1] ) |
|---|
| 501 | isnew = self.vgss.groupMatchRes[2] == 'new' |
|---|
| 502 | if isnew: |
|---|
| 503 | new = 1 |
|---|
| 504 | else: |
|---|
| 505 | new = -1 |
|---|
| 506 | |
|---|
| 507 | il = 6 |
|---|
| 508 | idx = 4 |
|---|
| 509 | ixp = 9 |
|---|
| 510 | iu = 11 |
|---|
| 511 | assert r[iu] == rk, 'CONFUSED ABOUT UIDs? %s, %s' % (rk,r[iu]) |
|---|
| 512 | var2 = r[1] |
|---|
| 513 | if r[il] in {u'17 (or 23 )', u'17 (or 23)', u'17 (or23)', u'10/17/23'}: |
|---|
| 514 | self.actions['Create 17/23 level pair'] += 1 |
|---|
| 515 | if r[il] == u'10/17/23': |
|---|
| 516 | lev0 = 10 |
|---|
| 517 | levsp = [19,23] |
|---|
| 518 | else: |
|---|
| 519 | lev0 = 19 |
|---|
| 520 | levsp = [23,] |
|---|
| 521 | r[ixp] = 1 |
|---|
| 522 | r[il] = lev0 |
|---|
| 523 | var2 = var2 + str( lev0 ) |
|---|
| 524 | p = 1 |
|---|
| 525 | for lp in levsp: |
|---|
| 526 | p += 1 |
|---|
| 527 | rr0 = r + [r[1] + str(lp),] |
|---|
| 528 | rr0[ixp] = p |
|---|
| 529 | rr0[il] = lp |
|---|
| 530 | rr0[iu] = str( uuid.uuid1() ) |
|---|
| 531 | rr = rr0 + [new, gpid, 0, vid] |
|---|
| 532 | if r[1] == 'hus': |
|---|
| 533 | print 'INFO.hus.00202: ',rr |
|---|
| 534 | rset.append( rr ) |
|---|
| 535 | self.varsets(rr0[iu], 2, rr[iu], 1, 'variables at multiple vertical resolutions' ) |
|---|
| 536 | |
|---|
| 537 | elif r[il] in [u'Model levels or plev_27', u'Model levels or 27Plevs',u'27',27.]: |
|---|
| 538 | self.actions['27 level variant of variable name created' ] += 1 |
|---|
| 539 | var2 += '27' |
|---|
| 540 | r[il] = 27 |
|---|
| 541 | elif r[idx].find( '850 hPa' ) != -1: |
|---|
| 542 | if r[1][-3:] != '850': |
|---|
| 543 | self.actions['850mb variant of variable generated'] += 1 |
|---|
| 544 | var2 += '850' |
|---|
| 545 | elif type( r[il] ) == type(1.) and r[il] == 7. and r[10] == 'HighResMIP' and r[5] != 'XYZKT': |
|---|
| 546 | self.actions['Shape modified to XYZKT*'] += 1 |
|---|
| 547 | r[5] = r[5] + '*' |
|---|
| 548 | r[il] = '7h' |
|---|
| 549 | r += [var2,new, gpid, 0, vid] |
|---|
| 550 | rset.append( r ) |
|---|
| 551 | if r[10] != 'OMIP': |
|---|
| 552 | for ri in rset: |
|---|
| 553 | self.consol_sub01( ri, extra ) |
|---|
| 554 | else: |
|---|
| 555 | self.actions['OMIP records skipped'] += 1 |
|---|
| 556 | |
|---|
| 557 | def consol_sub01( self, r, extra ): |
|---|
| 558 | """Write our a requestItem record, which is split into a CMORvar and requestVar in isd3b.py""" |
|---|
| 559 | iu = 11 |
|---|
| 560 | assert r[iu] not in self.sh |
|---|
| 561 | assert len(r) == len( self.sh['__cols__'] ), 'Bad record length .. %s' % str(r) |
|---|
| 562 | self.sh[r[iu]] = r[:] |
|---|
| 563 | |
|---|
| 564 | def checkRequestedGroups(self,sdir='inSh'): |
|---|
| 565 | sh = shelve.open( '%s/sh__requestScoping' % sdir, 'r' ) |
|---|
| 566 | sh0 = shelve.open( '%s/sh__requestScoping_0' % sdir, 'n' ) |
|---|
| 567 | cc = collections.defaultdict( int ) |
|---|
| 568 | ff = set() |
|---|
| 569 | for mip in sorted( sh['__records__'].keys() ): |
|---|
| 570 | reqg = sh['__records__'][mip] |
|---|
| 571 | reqgd = dict() |
|---|
| 572 | self.mip = mip |
|---|
| 573 | for rec in reqg: |
|---|
| 574 | g = rec[0] |
|---|
| 575 | if g[:5] == 'CMIP5': |
|---|
| 576 | g = g[6:] |
|---|
| 577 | g0 = g |
|---|
| 578 | t = self.vgss.recChk(rec,mip) |
|---|
| 579 | dyngrp = t[0] |
|---|
| 580 | if dyngrp: |
|---|
| 581 | mtab, g = t[2] |
|---|
| 582 | if not self.vgss.matchGroup(g,self.mip) and not dyngrp: |
|---|
| 583 | print 'FAILED to match request group: ',self.mip,rec |
|---|
| 584 | ff.add( (self.mip,g) ) |
|---|
| 585 | else: |
|---|
| 586 | print 'INFO.group.00008: group found: %s (%s):: %s, %s' % (g,self.mip,self.vgss.groupName, self.vgss.groupMatch ) |
|---|
| 587 | assert self.vgss.groupMatchRes != None, 'Should have results from matchGroup here!!' |
|---|
| 588 | reqgd[g] = self.vgss.groupMatchRes |
|---|
| 589 | uid = str(self.vgss.groupMatchRes[1] ) |
|---|
| 590 | rec[0] = self.vgss.groupName |
|---|
| 591 | kkk = '%s__%2.2i__%s' % (uid,cc[uid],0) |
|---|
| 592 | cc[uid] += 1 |
|---|
| 593 | assert kkk not in sh0, 'Duplicate key for request link record: %s' % kkk |
|---|
| 594 | sh0[kkk] = [self.mip, rec ] |
|---|
| 595 | print 'INFO.vgrp.00001: ',kkk,self.mip,rec |
|---|
| 596 | if dyngrp: |
|---|
| 597 | self.saveDynGrp(g,mtab,uid) |
|---|
| 598 | ## |
|---|
| 599 | ## need to generate request variables for dynamic groups here (those detected via vgss ) |
|---|
| 600 | ## would be nice to stage things ... create variable database and then add requests .... |
|---|
| 601 | ## "all at once" risks mis-alignments ... |
|---|
| 602 | ## |
|---|
| 603 | ## need to fill in details on dynamic groups after completion of scan of new groups. |
|---|
| 604 | ## |
|---|
| 605 | ## |
|---|
| 606 | sh.close() |
|---|
| 607 | sh0.close() |
|---|
| 608 | if len(ff) > 0: |
|---|
| 609 | oo = open( 'varGroupSuggestions.csv', 'w' ) |
|---|
| 610 | for t in sorted( list( ff ) ): |
|---|
| 611 | oo.write( '%s\t%s' % t ) |
|---|
| 612 | oo.write( '\t%s\n' % str( uuid.uuid1() ) ) |
|---|
| 613 | oo.close() |
|---|
| 614 | |
|---|
| 615 | def saveDynGrp(self,glab,tab,rgid): |
|---|
| 616 | this = self.vgss.s2[glab] |
|---|
| 617 | print 'INFO.dyngrp.00001: ',rgid,this, glab |
|---|
| 618 | for k in this: |
|---|
| 619 | if k[0] != '_': |
|---|
| 620 | if this[k] == None: |
|---|
| 621 | print 'ERROR.dyngrp.00002: variable not found: ',rgid,k, glab |
|---|
| 622 | else: |
|---|
| 623 | vid, priority = this[k] |
|---|
| 624 | rec = [vid, '%s: %s' % (glab,k), k, priority, rgid, self.mip] |
|---|
| 625 | uid = str(uuid.uuid1() ) |
|---|
| 626 | self.shdvg[uid] = rec |
|---|
| 627 | |
|---|
