Invalid Data Error - From Informat

[GBL__]

Hello all,

I am importing a csv file that has a variable named bnkrptdate that has a length of 7 and I am using an informat of MMDDYY7. to read it in to SAS.

The problem I am running in to is that if the observation does NOT have a bnkrptdate, the value is 0000000, if it does have a bnkrptdate then the informat works

data have ;
	infile datalines delimiter=',' ;
	informat contrprofile 16. bnkrpt_ch $2. bnkrptdate mmddyy7. loanclass $2. int_type $1. purchasedate mmddyy7. ;
	input contrprofile bnkrpt_ch $ bnkrptdate loanclass $ int_type $ purchasedate ;
datalines ;
0000000000000,00,0000000,59,P,0081518 
0000000000000,00,0000000,59,P,0081518 
0000000000000,00,0000000,59,P,0081518 
0000000000000,00,0000000,59,P,0081518 
0000000000000,00,0000000,59,P,0081518 
0000000000000,13,0040218,59,P,0081518 
0000000000000,00,0000000,59,P,0081518 
0000000000000,00,0000000,59,P,0081518 
0000000000000,00,0000000,58,I,0081518 
0000000000000,00,0000000,59,P,0081518 
0000000000000,00,0000000,59,P,0081718 
;
	format bnkrptdate purchasedate mmddyy10. ;
run;

This is the note I get (from my full dataset):

NOTE: Invalid data for bnkrptdate in line 21 494-500.
WARNING: Limit set by ERRORS= option reached.  Further errors of this type will not be printed.

I am just curious how I can work around this issue?  Open to any suggestions, thnaks for reading!

[Reeza · Accepted solution]

1. Read it as character and fix in the next step or within the proc

2. Create a custom informat that sets 000000 to missing (untested)

proc format;
invalue myFormat
'0000000' = .
other = [yymmdd10.];
run;

[GBL__]

proc format was a great way to go, thank you!

[ballardw]

On an input statement you can add the INFORMAT modifier  ?? to suppress error messages .

BTW you can't have any code such as your Format statement after a datalines block.

data have ;
	infile datalines delimiter=',' ;
	informat contrprofile 16. bnkrpt_ch $2. bnkrptdate  mmddyy7. loanclass $2. int_type $1. purchasedate mmddyy7. ;
	input contrprofile bnkrpt_ch $ bnkrptdate ?? loanclass $ int_type $ purchasedate ;
	format bnkrptdate purchasedate mmddyy10. ;
datalines ;
0000000000000,00,0000000,59,P,0081518 
0000000000000,00,0000000,59,P,0081518 
0000000000000,00,0000000,59,P,0081518 
0000000000000,00,0000000,59,P,0081518 
0000000000000,00,0000000,59,P,0081518 
0000000000000,13,0040218,59,P,0081518 
0000000000000,00,0000000,59,P,0081518 
0000000000000,00,0000000,59,P,0081518 
0000000000000,00,0000000,58,I,0081518 
0000000000000,00,0000000,59,P,0081518 
0000000000000,00,0000000,59,P,0081718 
;
run;

[GBL__]

Thanks, ballardw, for the tip about the ??.

As far as the format statement I just was trying to recreate quickly a small sample of my data, but good to know for future reference.