B1i Truncating space characters in elements from RegEx flat file SLD

Question

Task:
I'm importing from a fixed-width file that's specifically 94 characters wide consisting of multiple columns.  My only concern at this point is the number of lines and their width.  I have to pad the block out to a specific number of lines, and all existing lines in my source file already fill all 94 characters.  Multiple headers and footers exist in the same flat file, so defining the format explicitly based on column width in RegEx will add significant time to the project (though it's not necessarily impossible).
This file ends lines with a single LF character (no CR), which apparently is too much for TXT and CSV formats to handle (the entire file comes in as one line).  So I turned to Regex, where I can split the document on arbitrary characters like LFs.
I've recorded a message and the lines look something like this:
"ABB CCCCCCCCC DDDDDDDDDEEEEEEFFFFGHHHIIJKKKKKK KKKK K KKKKK    KKKKKK KKKK K KKKKK    LLLLLLLL"
and column L is occasionally blank (just spaces).  This is one of the row formats present in the file.  Most of the columns use spaces as fill characters and are left justified.
Tag definition looks like this:
<tagDefinition xmlns="urn:com.sap.b1i.bizprocessor:pltdefinition" regex="\n" schemaName="" tagName="line" matchSplit="S" stackSize="10000" DOTALL="true" MULTILINE="false">
</tagDefinition>
Problems:
With a recorded test messsage, the lines build properly:
<Payload Role="S" intype="rgx_ruledoc">
<bfa:io xmlns:bfa="urn:com.sap.b1i.bizprocessor:bizatoms" pltype="rgx" schemaName="">
  <line xmlns="">ABB CCCCCCCCC DDDDDDDDDEEEEEEFFFFGHHHIIJKKKKKK KKKK K KKKKK    KKKKKK KKKK K KKKKK            </line>
</bfa:io>
</Payload>
But by the time it reaches my first XSL Transformation atom, I receive the data in this form:
<Payload Role="S" intype="rgx_ruledoc">
<bfa:io xmlns:bfa="urn:com.sap.b1i.bizprocessor:bizatoms" pltype="rgx" schemaName="">
  <line xmlns="">ABB CCCCCCCCC DDDDDDDDDEEEEEEFFFFGHHHIIJKKKKKK KKKK K KKKKK KKKKKK KKKK K KKKKK </line>
</bfa:io>
</Payload>
In other words, I'm missing 3 spaces from within the middle of K and 11 from the end of the example line.  Multiple lines are affected by this.  I've built the Regex import portion properly, judging by the test message.  But the data I build is not the data I'm given.  And the truncation seems to be happening somewhere outside my control.
Questions:
1. Is there a way to specify that all spaces are non-breaking in the input/output format descriptors?
2. Is there a way I can specify that B1i doesn't modify the data before handing it to me?
3. Is there an easier/more correct way to do this?
Also possibly 4. Can you specify line endings in TXT/CSV imports?

Answer 1

As far as I've found, there's no way to specify that B1i doesn't normalize spaces for you in an input Msg.  I ended up splitting the incoming message line to its individual columns using the Regex format descriptor.  That way, I was able to track the integrity of each component of the message.
However, when writing out to the same file format using the TXT file type, B1i is inserting commas between the columns, despite:
1. Being TXT format, which shouldn't use delimiters.
2. Having the delimiter field empty.  It correctly ignores this field, even if it does insert a delimiter.
3. Specifying no delimiter explicitly using <FileOut type="file_full">/<Control><deli>.
It seems odd that the included means for writing raw text would mangle the output message so.
Has anyone successfully written data to a TXT file with no delimiter?