Have FME Desktop 2020. It turns out that writing UPRN's into a CSV doesn't work for values above 99.999.999. Have tried all data types in the writer (real64 is the one that FME sets automatically, but have tried String, int64, etc..). Nothing works.
Can somebody help me please?
First capture is the output before writing csv (upper workflow):
Second capture is the output after reading the CSV (lower workflow). Notice that there are 14400 records with null UPRN.
CSV will accept just about anything. Can you post a screenshot of some values from the inspector just before the writer?
CSV will accept just about anything. Can you post a screenshot of some values from the inspector just before the writer?
Hi, @jlbaker2779, have added screen captures, as requested.
Actually, all values missing are like 10.000.000.000 and 200.000.000.000. So quite larger than the 99.999.999 previously stated
Hi, @jlbaker2779, have added screen captures, as requested.
If it does have a value, use a tester to test out a single record from the bottom and top like; 4 Sands Terrace Leeds Road. Check if the attribute is only digits, no spaces etc. Then use a expression evaluator to multiply the number by 1 and write to the CSV.
Kind of a work around that should work, but if it's not null and the attribute is all numerical values it should be writing without issue to the column and may need to be looked at by Safe.
Hi @miquelrs, have you perhaps set data type of UPRN field manually in the CSV reader?
If you set a numeric type (e.g. int32) to the field manually and the value was out of range for the data type, the CSV reader would read it as <null>.
If you are trying to write exactly this value 10.000.000.000 to a field that is set to be a number it will not be written. Are the points actually present in your data?
If you are trying to write exactly this value 10.000.000.000 to a field that is set to be a number it will not be written. Are the points actually present in your data?
I added the .'s manually in this post. UPRNs that are missing look like 200003790604 and 10094968720, while the ones that go through are like 83000014
Hi @miquelrs, have you perhaps set data type of UPRN field manually in the CSV reader?
If you set a numeric type (e.g. int32) to the field manually and the value was out of range for the data type, the CSV reader would read it as <null>.
Hi @takashi, the reader is a SQL database, and the datatype is set to "float" by default in the reader.
If it does have a value, use a tester to test out a single record from the bottom and top like; 4 Sands Terrace Leeds Road. Check if the attribute is only digits, no spaces etc. Then use a expression evaluator to multiply the number by 1 and write to the CSV.
Kind of a work around that should work, but if it's not null and the attribute is all numerical values it should be writing without issue to the column and may need to be looked at by Safe.
All records have originally a value in the step previous to the writer (all UPRNs listed in the first screenshot get to "null" once written as csv.
UPRNs that are missing look like 200003790604 and 10094968720, while the ones that go through are like 83000014
Hi @miquelrs, have you perhaps set data type of UPRN field manually in the CSV reader?
If you set a numeric type (e.g. int32) to the field manually and the value was out of range for the data type, the CSV reader would read it as <null>.
I meant this one. Have you checked data type setting in this CSV reader feature type?
I meant this one. Have you checked data type setting in this CSV reader feature type?
Great, so that was the issue, with the reader, not with the writer. It was set automatically to int32. I have changed it to int64 and string and those missing UPRNs come through. Thank you very much for your help!
Enter your E-mail address. We'll send you an e-mail with instructions to reset your password.