source: CMIP6dreqbuild/trunk/srcMisc/dreqSX.py @ 376

import shelve, string, collections
from utils_wb import workbook, uniCleanFunc
import cmip5so
from sets import Set

vdate = "20150827"

sh = shelve.open( 'dreq_consol_tables_shelve_v%s' % vdate, 'r' )
shnv = shelve.open( 'dreq_consol_tables_nv_shelve_v%s' % vdate, 'r' )
## from vlsc5b --- includes new names from OMIP
shold = shelve.open( 'dreq_consol_tables_old_shelve_v20150713', 'r' )
shlist = [sh,shnv,shold]

##MIP columns: recordid, ['comment', 'deflate_level', 'shuffle', 'ok_max_mean_abs', 'flag_meanings', 'type', 'ok_min_mean_abs', 'standard_name', 'deflate', 'long_name', 'valid_min', 'cell_methods', 'flag_values', 'cell_measures', 'out_name', 'modeling_realm', 'units', '#cell_methods', 'valid_max', 'positive'] , var, mip table, dims

##ee0: Short name of group      Variable short name     Table   Frequency       Description extension (optional)        Shape   Levels  Time mean, point or climatology Mask (optional), priority, mip, uuid
## vgrp, var, table, freq, comment, shape, levels, tform, mask, priority, mip, recordId
## e.g. 'PMIP-Lmon', u'rh_c14', u'new', u'mon', '', u'XYT', '', '', '', 2, 'PMIP', '21de1a34-1597-11e5-861b-ac72891c3257')
#####
###['group','var','table','frequency','shape',''....
### Want to map shape onto a "feature type" which will reference a set of dimensions.
##XYT: latitude, longitude, time
##'4 element vector' : seaice transect x time.
## KZT: RFMIP: not clear what K is.
## '2D vector field': from SIMIP, vector as in list: extra dimension is a list of thickness categories (

##eenv: 
## MIP  standard name check     Review comments Shape etc       Short name      CF standard_name        standard name status    Native grid     units   Long Name       description/comments    Priority        associated observational dataset, ..., recordID (last)
##e.g. [2, u'FAFMIP', -1.0, '', u'yr|XYZ|mean', u'ocontemppsmadvect', u'tendency_of_sea_water_conservative_temperature_expressed_as_heat_content_due_to_parameterized_submesoscale_advection', -1.0, u'ocean', u'W/m2', '', u'Level 2: priority unclear', 2.0, '', '', '', '', 'b1feb8a2-1598-11e5-92ad-ac72891c3257']

##eeold:
##e.g. ['ef36eefae77f82455c0a925167a0d4ca', 'rss', 'Net shortwave surface radiation', 'surface_net_downward_shortwave_flux', 'unset', 'W m-2', 'SPECS_day', 'mv.001.0094', 'mv.001.0394', '', '']
##ee_xref
##hash, uuid list -- list of mip tables for each parameter key (ee_old)

ee_xref = {}
ee_xref_bck = collections.defaultdict( list )
ee_miptables_old = collections.defaultdict( list )
ee_miptables_vars_old = collections.defaultdict( list )
ee_ovars_old = collections.defaultdict( list )
ee_mip = {}
ee0 = {}
eenv = {}
## dictionary to carry short names of old variables.
eeold = {}
###for ee,shv in [(ee0,sh),(eenv,shnv),(eeold,shold)]:
for ee,shv in [(ee0,sh),(eenv,shnv)]:
  for k in shv.keys():
    ee[k] = shv[k]

rmod = 'csv'
eeoldl = collections.defaultdict( list )
if rmod == 'xls':
  for ee,fn in [(ee_xref,'vlsc5_xref_v20150826.xls'),(ee_mip,'vlsc5_mip_v20150826.xls'),(eeold,'vlsc5b_v20150826.xls')]:
    wb = workbook( fn )
    sh = wb.book.sheet_by_name( 'Sheet1' )
    for i in range(sh.nrows):
      r = map( lambda x: string.strip(str(uniCleanFunc(x.value))), sh.row(i) )
      ee[r[0]] = r
else:
  for ee,fn,sep,xxx in [(ee_xref,'vlsc5_xref_v20150826.csv',',',False),(ee_mip,'vlsc5_mip_v20150826.csv', '\t',False),(eeoldl,'vlsc5b_v20150826.csv', '\t',True)]:
    for i in open(fn).readlines():
      r = map( lambda x: string.strip(str(uniCleanFunc(x))), string.split( string.strip(i), sep ) )
      while len(r) < 9:
        r.append( '' )
      if xxx:
        ee[r[0]].append( r )
      else:
        ee[r[0]] = r


for h in eeoldl.keys():
  if len( eeoldl[h] ) == 1:
    eeold[h] = eeoldl[h][0]
  else:
    ll = []
    for k in range( len( eeoldl[h] ) ):
      if eeoldl[h][k][8] == '':
        ll.append(k)
    if len(ll) == 0:
      print 'ERROR.004.0001: no unforwarded record in vlsc5b for h=%s' % h
      eeold[h] = eeoldl[h][0]
    elif len(ll) > 1:
      print 'ERROR.004.0002: multiple unforwarded records in vlsc5b for h=%s (%s)' % (h,eeoldl[h][ll[0]][1])
      eeold[h] = eeoldl[h][ll[0]]
    else:
      eeold[h] = eeoldl[h][ll[0]]

k2m = []
km = []
for k in ee_xref.keys():
  if eeold.has_key(k):
    for k2 in ee_xref[k][1:]:
      if k2 != '':
        if ee_mip.has_key(k2):
          ee_xref_bck[k2].append(k)
        else:
          k2m.append( (k2,k) )
  else:
    km.append(k)

##
## had bug here due to extension of ee_mip to include dimensions.
##
ee_byvar = collections.defaultdict(list)
for k in ee_mip.keys():
  tid = ee_mip[k][-2]
  ee_miptables_old[tid].append(  ee_mip[k][0] )
  ee_miptables_vars_old[tid].append(  ee_mip[k][-3] )
  ee_ovars_old['%s.%s' % (tid,ee_mip[k][-3]) ].append(  k )
  ee_byvar[ ee_mip[k][-3] ].append( k )

## check cmip5 vs. standard output
s1 = Set()
for k in ee_miptables_vars_old.keys():
  if k[:6] == 'CMIP5_':
    s2 = Set( ee_miptables_vars_old[k] )
    kk = k[6:]
    s1.add( kk )
    assert cmip5so.so.has_key(kk ), 'Table %s not found in standard output' % kk
    lnf = []
    for v in ee_miptables_vars_old[k]:
      if not cmip5so.so[kk].a.has_key(v):
         lnf.append(v)
    if len(lnf) > 0:
      print 'ERROR.010.0001: Variables not found in %s:: %s' % (kk,str(lnf))

    lnf = []
    for v in cmip5so.so[kk].a.keys():
      if not v in s2:
         lnf.append(v)
    if len(lnf) > 0:
      print 'ERROR.010.0002: SO Variables not found in %s:: %s' % (kk,str(lnf))

for s in cmip5so.so.keys():
  if not s in s1:
    print 'ERROR.010.0002: Table %s not in old mip tables' % s

for s in shlist:
  s.close()