09-09-2016 11:52 AM
We recieve data shipped to us as large complex XML files with an XML schema included. Using the SAS XML Mapper 9.4 and the schema, I was able to use the "Auto Generate" process to create an XML Map (with auto generated keys) from the XML schema. This resulted in the XML being divied up into over 300 tables. I ran PROC COPY to pull all the tables into SAS... and it took about 2 days to finish. I tried the same XML file and XML map and ran a DATA step for one table. That step took a little more than 5 minutes. It appears based on the log that for each table, SAS reads through the entire XML file again. Is there a way to output to multiple SAS datasets with only one read through of the XML?
Code run on SAS 9.4 UNIX server:
filename DataWare "/sas/data/LARGE.xml";
filename SXLEMAP "/sas/data/XML Maps/LARGE_AUTO.map";
libname DataWare xmlv2 xmlmap=SXLEMAP access=READONLY;
libname tempo "%sysfunc(getoption(WORK))";
proc copy in=DataWare out=tempo;
09-09-2016 10:30 PM
If you don't really need the 300 different tables, manually construct the xml map by drag and drop of items to create a single larger denormalised table.