file access node - Ascii vs binary

Taking in a file that is in Ascii (CR/LF at the end of every record). The file is pulled into the flow using File channel/receiver and then is written from the folder/file on the Landmark server to our LSF server which is Linux. That file is passed to a SysCmd Node that does an importdb. I cannot get the importdb to work. If I run the same script from the command prompt in LID - it works fine. I've tried binary, US-ASCII and blank for the encoding types and nothing works - I get the same result every time.

I get this error on EVERY line and no records upload - Field 21 is the Line Feed on every line - it is as if it is treating the second part of the CR/LF as a separate field

Warning: line 1, Field 21:

- Was converted to Uppercase.

Any thoughts on how to handle this?

0712131256598360.doc

Find more posts tagged with

Process Automation (IPA) - General Discussions

Accepted answers

All comments

russel-cornelson

Since you are saying there are CR/LF characters, I assume the file was built on windows which is
, and you are saying LSF on on Linux the correct line feed is

I would use "sed" or "tr" to remove the
so the correct line feed characters are in the file before processing the importdb.

The alternative is to modify whatever is creating the file so it uses the
for line feed. Aka, using an assign node so you can specify the carriage return.

laura-patenaude

Thanks for your response. Yes, you are correct. And the problem is the additional character. I don't understand though why I can run the Unix script at the command prompt in LID and the file loads fine - but if I run the script in IPA - the errors occur. I would think running the script in IPA would give the same results as running it from the command prompt. The sending system is unable to be modified unfortunately. What is the purpose of the "encoding type" in the File Access node? It seems like that would have something to do with this. I can't find any documentation on it - I've been searching.

serafino

You can use 'dos2unix' and 'unix2dos' to convert text files between the two. On the UNIX side, you can use 'col -b' to remove the extra ^M characters from new lines. All of these can be used as:

cat WINDOWSFILE.TXT | dos2unix > unixfile

col -b < WINDOWSFILE.TXT > unixfile

russel-cornelson

When using a flow, does the flow put the file on the LSF server?

When using LID, are you manually moving the file?

I would expect the same result if everything is truly the same.

0911060553270289.pdf

laura-patenaude

Yes - flow moves the file and in LID - I am manually moving the file first. But in both cases - it is the same file. All my other flows are set to "binary" because I get the files are binary from the outside sources (so just a Line Feed at the end) and then they execute a Unix script for importdb - those all work great. It's only this one that is having a problem. I wish I could find some clear documentation on what the purposes of the Encoding type is and how to use it. I've opened a ticket with Infor but I'm not optimistic about getting a lot of help since this is more of a development kind of question.

unknown

Laura -

The FileAccess node is going to create a file with the exact contents that are in whatever variable you specify in the input data. I assume that when you move the file to the LSF server from your local machine and test it in LID you are using FTP which is not the same as the FileAccess node. You are likely specifying Ascii or auto and the ftp program is automatically converting your EOL characters. I see 3 options. 1. Write the file to your LMK server and then use an FTP node to move it to LSF and specify Ascii to change the EOL characters as a part of the process. 2. modify the data in the flow before you write the file to LSF via your file access node or 3. As Sal suggests once the file is on the LSF server run a system command like dos2unix or a perl one-liner or whatever you like to translate the
to
and then run your importdb.

On a side note, how are you? It's been a LONG time since YCC.

russel-cornelson

Additional NOTE: I am going to open an enhancement request so if you create a file using a message builder; you would be able to specify the carriage return character that is automatically added at every line with a default set to
. However, this way if you are building a file with a message builder you could set the message builder to linuxunix or windows.

The workaround is still to use an assign node; and add your own carriage return character when building the output.

var1 += var + '
'

laura-patenaude

I dropped you an email - hopefully you'll get it. Thanks for your input on this - looks like I'm going to have to do more work than I wanted --- lol

0712131256598360.doc

Copyright © 2025 Infor. All rights reserved.