;+ ; FUNCTION cdfi = cdf_load_vars_partial(file) ; Keywords: ; VARFORMAT = string or string array: a string or string array (which may contain wildcard ; characters) that specifies the CDF variable names to load. Use ; 'VARFORMAT='*' to load all variables. ; VARNAMES = named variable ;output variable for variable names that were loaded. ; SPDF_DEPENDENCIES : Set to 1 to have SPDF defined dependent variables also loaded. ; VAR_TYPE = string or string array; Variables that have a VAR_TYPE matching these strings will ; be loaded. ; CONVERT_INT1_TO_INT2 Set this keyword to convert signed one byte to signed 2 byte integers. ; This is useful because IDL does not have the equivalent of INT1 (bytes are unsigned) ; RECORD: Specify the record index where you want to start reading. By default, this option will read one record. ; NUMBER_RECORDS: Specify the number of records that you want to read. By default, this option will begin at record zero. ; ; Note: Record & Number_Records can be used together to specify a range of records to be read. ; ; Author: Davin Larson - 2006 ; ; $LastChangedBy: peters $ ; $LastChangedDate: 2013-10-08 12:43:36 -0700 (Tue, 08 Oct 2013) $ ; $LastChangedRevision: 13281 $ ; $URL: svn+ssh://thmsvn@ambrosia.ssl.berkeley.edu/repos/spdsoft/trunk/general/missions/rbsp/efw/cdf_partial_load/cdf_load_vars_partial.pro $ ; ;- function cdf_load_vars_partial,files,varnames=vars,varformat=vars_fmt,info=info,verbose=verbose,all=all, $ record=record,convert_int1_to_int2=convert_int1_to_int2, $ spdf_dependencies=spdf_dependencies, $ var_type=var_type, $ no_attributes=no_attributes,$ number_records=number_records vb = keyword_set(verbose) ? verbose : 0 vars='' info = 0 dprint,dlevel=4,verbose=verbose,'$Id: cdf_load_vars_partial.pro 13281 2013-10-08 19:43:36Z peters $' on_ioerror, ferr for fi=0,n_elements(files)-1 do begin if file_test(files[fi]) eq 0 then begin dprint,dlevel=1,verbose=verbose,'File not found: "'+files[fi]+'"' continue endif id=cdf_open(files[fi]) if not keyword_set(info) then begin info = cdf_info(id,verbose=verbose) ;, convert_int1_to_int2=convert_int1_to_int2) endif ; if there are no variables loaded if info.nv eq 0 or ~is_struct(info.vars) then begin dprint,verbose=verbose,'No valid variables in the CDF file!' return,info endif if n_elements(spdf_dependencies) eq 0 then spdf_dependencies =1 if not keyword_set(vars) then begin if keyword_set(all) then vars_fmt = '*' if keyword_set(vars_fmt) then vars = [vars, strfilter(info.vars.name,vars_fmt,delimiter=' ')] if keyword_set(var_type) then begin vtypes = strarr(info.nv) for v=0,info.nv-1 do begin vtypes[v] = cdf_var_atts(id,info.vars[v].num,zvar=info.vars[v].is_zvar,'VAR_TYPE',default='') endfor vars= [vars, info.vars[strfilter(vtypes,var_type,delimiter=' ',/index)].name] endif vars = vars[uniq(vars,sort(vars))] if n_elements(vars) le 1 then begin dprint,verbose=verbose,'No valid variables selected to load!' return,info endif else vars=vars[1:*] vars2=vars ; if vb ge 4 then printdat,/pgmtrace,vars,width=200 if keyword_set(spdf_dependencies) then begin ; Get all the variable names that are dependents depnames = '' edepnames = strarr(n_elements(vars)) for i=0,n_elements(vars)-1 do begin vnum = where(vars[i] eq info.vars.name,nvnum) if nvnum eq 0 then message,'This should never happen, report error to D. Larson: davin@ssl.berkeley.edu' vi = info.vars[vnum] depnames = [depnames, cdf_var_atts(id,vi.num,zvar=vi.is_zvar,'DEPEND_TIME',default='')] ;bpif vars[i] eq 'tha_fgl' depnames = [depnames, cdf_var_atts(id,vi.num,zvar=vi.is_zvar,'DEPEND_0',default='')] edepnames[i] = cdf_var_atts(id,vi.num,zvar=vi.is_zvar,'DEPEND_0',default='') ndim = vi.ndimen for j=1,ndim do begin depnames = [depnames, cdf_var_atts(id,vi.num,zvar=vi.is_zvar,'DEPEND_'+strtrim(j,2),default='')] endfor endfor if keyword_set(depnames) then depnames=depnames[[where(depnames)]] depnames = depnames[uniq(depnames,sort(depnames))] vars2 = [vars2,depnames] vars2 = vars2[uniq(vars2,sort(vars2))] vars2 = vars2[where(vars2)] if keyword_set(edepnames) then edepnames=edepnames[[where(edepnames)]] edepnames = edepnames[uniq(edepnames,sort(edepnames))] ; if vb ge 4 then printdat,/pgmtrace,depnames,width=200 endif endif if keyword_set(edepnames) then cdf_partial_epoch_finder,id,edepnames,epoch_indices dprint,dlevel=2,verbose=verbose,'Loading file: "'+files[fi]+'"' for j=0,n_elements(vars2)-1 do begin w = (where( strcmp(info.vars.name, vars2[j]) , nw))[0] if nw ne 0 then begin vi = info.vars[w] dprint,verbose=verbose,dlevel=7,vi.name ; if vb ge 9 then wait,.2 ; if vi.recvary or 1 then begin ;disabling logic that does nothing, pcruce@igpp.cula.edu q=!quiet & !quiet=1 & cdf_control,id,variable=vi.name,get_var_info=vinfo & !quiet=q thisedepname = cdf_var_atts(id,vi.num,zvar=vi.is_zvar,'DEPEND_0',default='') whichepoch = where(edepnames eq thisedepname, whichepoch_cnt) if whichepoch_cnt eq 0 then begin if (vi.datatype eq 'CDF_EPOCH') or (vi.datatype eq 'CDF_EPOCH16') or (vi.datatype eq 'CDF_TIME_TT2000') then begin whichepoch = where(edepnames eq vi.name, whichepoch_cnt) endif endif ;adding logic to select the number of records that are loaded. Helps for testing with large CDFs, can be used with the record= keyword if n_elements(number_records) ne 0 then begin numrec=number_records<(vinfo.maxrec+1) endif else begin if n_elements(record) ne 0 then begin numrec=1<(vinfo.maxrec+1) endif else begin numrec = vinfo.maxrec+1 endelse endelse if whichepoch_cnt ne 0 then begin numrec = epoch_indices[whichepoch,1] - epoch_indices[whichepoch,0] + 1 record = epoch_indices[whichepoch,0] endif ; dprint,verbose=vb,dlevel=7,vi.name ; if vb ge 9 then wait,.2 ; endif else numrec = 0 if numrec gt 0 then begin q = !quiet !quiet = keyword_set(convert_int1_to_int2) if n_elements(record) ne 0 then begin value = 0 ;THIS line TO AVOID A CDF BUG IN CDF VERSION 3.1 cdf_varget,id,vi.name,value ,/string ,rec_start=record,rec_count=numrec endif else begin if vi.is_zvar then begin value = 0 ;THIS Line TO AVOID A CDF BUG IN CDF VERSION 3.1 cdf_varget,id,vi.name,value ,/string ,rec_count=numrec ;CDF_varget,id,CDF_var,x,REC_COUNT=nrecs,zvariable = zvar,rec_start=rec_start endif else begin if 1 then begin ; this cluge works but is not efficient! vinq = cdf_varinq(id,vi.num,zvar=vi.is_zvar) dimc = vinq.dimvar * info.inq.dim dimw = where(dimc eq 0,c) if c ne 0 then dimc[dimw] = 1 ;bpif vi.name eq 'ion_vel' endif value = 0 ;THIS Line TO AVOID A CDF BUG IN CDF VERSION 3.1 CDF_varget,id,vi.num,zvar=0,value,/string,COUNT=dimc,REC_COUNT=numrec ;,rec_start=rec_start value = reform(value,/overwrite) dprint,phelp=2,dlevel=5,vi,dimc,value endelse endelse !quiet = q if vi.recvary then begin if (vi.ndimen ge 1 and ((n_elements(record) eq 0) or keyword_set(whichepoch_cnt))) then begin if numrec eq 1 then begin dprint,dlevel=3,'Warning: Single record! ',vi.name,vi.ndimen,vi.d value = reform(/overwrite,value, [1,size(/dimensions,value)] ) ; Special case for variables with a single record endif else begin transshift = shift(indgen(vi.ndimen+1),1) value=transpose(value,transshift) endelse endif else value = reform(value,/overwrite) if not keyword_set(vi.dataptr) then vi.dataptr = ptr_new(value,/no_copy) $ else *vi.dataptr = [*vi.dataptr,temporary(value)] endif else begin if not keyword_set(vi.dataptr) then vi.dataptr = ptr_new(value,/no_copy) endelse endif if not keyword_set(vi.attrptr) then $ vi.attrptr = ptr_new( cdf_var_atts(id,vi.name,convert_int1_to_int2=convert_int1_to_int2) ) info.vars[w] = vi endif else dprint,dlevel=1,verbose=verbose,'variable "'+vars2[j]+'" not found!' endfor cdf_close,id endfor if keyword_set(info) and keyword_set(convert_int1_to_int2) then begin w = where(info.vars.datatype eq 'CDF_INT1',nw) for i=0,nw-1 do begin v = info.vars[w[i]] if ptr_valid(v.dataptr) then begin dprint,dlevel=5,verbose=verbose,'Warning: Converting from INT1 to INT2 (',v.name ,')' val = *v.dataptr *v.dataptr = fix(val) - (val ge 128) * 256 endif endfor endif return,info ferr: dprint,dlevel=0,"CDF FILE ERROR in: ",files[fi] msg = !error_state.msg ;copy to keep system var from being mutated when MESSAGE is called message, msg return,0 end