Multiple Record Types Same File

A flat file I must parse has multiple record types in the same file that must be processed together. For instance, in the following example:

01TestStuff888
02TestStuff2
03TestStuff3
01TestStuff4
02TestStuff5

Each 01 record type has the records after it associated to it until the next 01 appears, so TestStuff would have TestStuff 2,3 related to it while TestStuff 4,5 belong together. In the example the 888 in the 01 record is the key to the group, but it does not appear in the following lines.

The problem is that each record type has different line formats, columns, etc, so they must be parsed differently. I have created a conditional branch on the first two characters, and written each record type out to a seperate flat file for that type, so that they can be imported again and parsed with the Flat File Source, but I am unsure how to relate them again. I tried appending the 888 to the other lines before they were written out, but I can't find a way to share the variable across the conditional split branches using a script component.

Does anyone have an idea how I could parse these files and keep the relationship intact?

Is there a way to tell the flat file wizard to use a different map based on certain characters?

Is there a way to share a variable across the different braches of a conditional split.

Strange question I know, but thanks for any help.

Travis