unable to read invalid characters from xml

ArndW · Post by **ArndW** » Thu Jan 08, 2009 9:04 am

What error message is being generated, and from which stage? Do you have NLS installed and if so, which character set are you using?

krishna81 · Post by **krishna81** » Thu Jan 08, 2009 9:53 am

It's not issuing any errors/warnings.The job has been successfully compleated but it processed one record instead of two.And NLS is disabled in project level.

chucksmith · Post by **chucksmith** » Thu Jan 08, 2009 9:56 am

On the server job side, I have had to put a transformer stage between the sequential, in my case a folder stage, and the xml stage. In the transformer, I use the convert() function to change or remove the problem characters. The key is to identify those characters.

miwinter · Post by **miwinter** » Thu Jan 08, 2009 9:58 am

Are you using an XML input stage? Try highlighting "Log reject errors" on the General tab

Maybe also change the transformation error mappings to "Warning" or "Fatal"

krishna81 · Post by **krishna81** » Thu Jan 08, 2009 10:24 am

yes iam using xml input stage.I hve tried with highlighting "Log reject errors" on the General tab but i didnot see any change.
even i tried with convert function using ASCII value 179 but no luck.

miwinter · Post by **miwinter** » Thu Jan 08, 2009 10:27 am

So there is no mention in the log of your rejected row then, once the "Log reject errors" has been turned on? Is it definitely recognising the two rows as distinct records?

eostic · Post by **eostic** » Thu Jan 08, 2009 12:52 pm

How certain are you that XMLInput is losing the row....could it be that the Trx is dropping it...perhaps due to nulls somewhere or other datatype issue?

Ernie

krishna81 · Post by **krishna81** » Thu Jan 08, 2009 1:46 pm

At the outset it's not able to read the record.

From the XML itself Dropping the Rows.

Seq------>XML---------->Tx----------->Seq.
0Rows 0 Rows

I believe this makes sense.

eostic · Post by **eostic** » Thu Jan 08, 2009 7:59 pm

...try two things as debugging aids:

a) Use the External Source stage to read the input name of the xml document... see www.dsrealtime.com for instructions (go to the table of contents and find XML ... reading xml content for details)

b) try it in a Server job for test purposes, reading the xml document using a folder stage. If XMLInput was dying, you'd get "1" row from Seq and 0 rows from XML.

...can't say what it is exactly, but I don't trust how the Sequential Stage in EE might be reading the document.

Ernie

krishna81 · Post by **krishna81** » Fri Jan 09, 2009 8:06 am

Here the sequential file doing the same what the External source stage.i mean we are using Seq file stage to read the input name of the xml document.

eostic · Post by **eostic** » Sat Jan 10, 2009 8:23 am

and does the same logic fail when written in Server?

eostic · Post by **eostic** » Sat Jan 10, 2009 1:08 pm

Looking at this again, what "exactly" are the characters? Can you share the string in hex here, just in case we're not seeing exactly what is represented?

Ernie

l_homme_de_rio · Post by **l_homme_de_rio** » Thu Jan 15, 2009 2:55 am

Hi
I guess i m facing the same problem ...
Open your xml with Ultraedit You gonna see some garbage char before "<xml version="1.0">"
if you remove them it should work
Michel