db-derby-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Rick Hillegas <Richard.Hille...@Sun.COM>
Subject Re: Importing delimited files with header
Date Thu, 29 May 2008 17:03:02 GMT
Hi Patrick,

Under the hood, the SYSCS_IMPORT_TABLE procedure prepares and executes 
an INSERT INTO ... SELECT * FROM statement using its own custom table 
function. So you should get the same good performance from your own 
table function as you get from the IMPORT procedure.

Hope this helps,

Patrick Meyer wrote:
> Thanks for your response. Do you know if extending FlatFileVTI will 
> work as fast as the built-in IMPORT_TABLE function? Before using CALL 
> SYSCS_UTIL.SYSCS_IMPORT_TABLE, I had written my own class that would 
> read a delimited file and use the INSERT INTO statement to populate my 
> Derby db. However, it took my class 1 hour to import a file but it 
> only took one minute when I used SYSCS_UTIL.SYSCS_IMPORT_TABLE. Will 
> extending FlatFileVTI work a quickly as using 
> Thanks
> Rick Hillegas wrote:
>> Hi Patrick,
>> You can write a table function to help you out here. Your table 
>> function will open your file, throw away the header, and then loop 
>> through the lines in the file. I would recommend extending something 
>> like FlatFileVTI (in Derby's demo directory). You can look at 
>> PropertyFileVTI (in the same directory subtree), which will show you 
>> how to extend FlatFileVTI. Once you declare your table function 
>> (let's call it MyFileReader), you can then use it to populate your 
>> table as follows:
>> insert into STAFF
>> select * from table( MyFileReader( 'myfile.del' ) ) s
>> Hope this helps,
>> -Rick
>> Patrick Meyer wrote:
>>> I have a comma delimited file that contains a header in the first 
>>> row of the file. The header contains column names. If I manually 
>>> create a new file without the header, I can successfully import the 
>>> new file with
>>> CALL 
>>> SYSCS_UTIL.SYSCS_IMPORT_TABLE(null,'STAFF','myfile.del',null,null,null,0); 
>>> Creating the new file in my Java application is too time consuming 
>>> for it to be a good solution. Is there a way of importing a 
>>> delimited file with a header in the first row? Or, a way to import 
>>> and ignore the first row of the file being imported?
>>> Thanks

View raw message