I can't build an xsd for a flat file (txt) to handle repeating records

Question

Hi - have looked at many posts around flat file schema and they don't seem to address my question.
I have a flat file that is \n delimited
the pattern of the data is simple:
record1 - 90 characters
record2 - 20 characters
record3 - n 248 characters - each of these records is parsed into children by the positional method
record n+1 10 characters
record n+2 20 characters
so I used the flat file schema generator to generate the schema and built a map mapping the flat file schema to another xml schema. The schema looks ok - record1, record2, record n+1, record n+2 are child elements of the root. the repeating record
section is showing up as a node with the parsed children.
The transform is only mapping the children of the repeating records. When I test the map only the first repeating record gets parsed. No repeating happens (the actual flat file has 400+ repeating records). When I run the map in debug mode, the input
xml shows that record1 is read in correctly, record2 is read in correctly, record3 is read in and parsed and record4 is treated like record n+1 and record5 is treated like record n+2 and the map thinks it's all finished.
the section of the repeat part of the schema is and you can see that I set the minOccurs=1 and maxOccurs=unbounded for the node (INVOICE) and the complexType but this is not an affective syntax. I have looked at how the EDI X12 schema look and how they handle
looping and it is a lot different than what the Flat File schema wizard is doing. Is there a good set of rules published that would guide me though this? otherwise I will basically have to read in the lines from the file and parse them out with functoids -
seems so inelegant. Thanks in advance.
<xs:element minOccurs="1" maxOccurs="unbounded" name="INVOICE">
          <xs:annotation>
            <xs:appinfo>
              <b:recordInfo structure="positional" sequence_number="3" preserve_delimiter_for_empty_data="true" suppress_trailing_delimiters="false"
/>
            </xs:appinfo>
          </xs:annotation>
          <xs:complexType>
            <xs:sequence minOccurs="1" maxOccurs="unbounded">
              <xs:annotation>
                <xs:appinfo>
                  <groupInfo sequence_number="0" xmlns="http://schemas.microsoft.com/BizTalk/2003"
/>
                </xs:appinfo>
              </xs:annotation>
              <xs:element name="SegmentType" type="xs:string">
                <xs:annotation>
                  <xs:appinfo>
                    <b:fieldInfo justification="left" pos_offset="0" pos_length="2" sequence_number="1" />
                  </xs:appinfo>
                </xs:annotation>
              </xs:element>....... more children elements
Harold Rosenkrans

Answer 1

Thanks for responding
I gave up trying to parse the repeating record into fields. Instead I just loop through the repeating record section with an <xs:for-each> block in the xsl and use functoids to grab the fields.
So that works for having the two, shorter header records (structure is positional) before the section of repeating records. Now I just have to figure out how to get the schema to handle the two, shorter trailer (or footer, whatever you prefer) records after
the section of repeating records
the error I get in VS when I test the map is [BTW I changed the element names in the schema which is why you don't see INVOICE in the error]
When I declare the last element as being positional with a character length of 10 I get the error:
Error 18 Native Parsing Error: Unexpected end of stream while looking for:
'\r\n'
The current definition being parsed is SAPARData. The stream offset where the error occured is 1359. The line number where the error occured is 9. The column where the error occured is 0. 
so the first record is 77 char in length and the second is 16 char and then the repeating records (5 in the file) are 248 char and the last record is 10 char
so an offset of 1359 puts it beyond the last record by 16 characters - so the stream reader is looking for the next repeating record.
if I try to declare the last element as delimited I get the error:
Error 14 Native Parsing Error: Unexpected data found while looking for:
'\r\n'
The current definition being parsed is SAPARData. The stream offset where the error occured is 597. The line number where the error occured is 5. The column where the error occured is 0. 
so the first record is 77 char in length and the second is 16 char and then the repeating records are 248 char.
a stream offset of 597 puts me 8 characters into the third repeating record - at this point I have only declared one trailer record in the  schema, 10 characters long.
Why is stream reader stopping at such a weird spot?
The bottom line is I still haven't discovered the correct schema to handle the trailer records. even if I set the maxOccurs="4" (for the repeat record declaration) it still gets the first error. How does it find an unexpected end of stream looking
for \r\n when the maxOccurs for the repeat record declaration should have the stream pointer in the 5th repeat record.
I unfortunately don't have any options concerning the file structure.
I have read a lot of posts concerning the trailer issue. I have seen a couple that looked interesting. I guess I'll just have to give them a try. The other option is to create a custom pipeline that will only take file lines of 248 characters.
That's just disgusting !
Harold Rosenkrans