Parse backslash-delimited hierarchy path into separate columns

Normalization

A fully-normalized schema would not have four path component columns in a table; it would have one table with a string column for the path component, and a foreign key to its parent (and/or child), conceptually similar to a linked list. Advantages to this approach would be support for an arbitrarily long path. Disadvantage is likely decreased performance, though the scale of your problem is not clear.

Otherwise, I find it odd that you're parsing a path, and storing the result of the parse back into the database alongside the original data. This is a fast-enough operation that surely any advantage to caching parse results is outweighed by the disadvantage of increased storage and fetch time.

The only solid justification I can think of for such a thing is when (1) you know that certain interesting strings exist in fixed locations in the path, and (2) you need very fast, indexed querying on them, and (3) such querying is too slow when doing the usual pattern-matching approach on the original path.

Iteration

Don't use range for your loop; use enumerate. Also use a raw string due to your backslash.

for i, part in enumerate(s.split(r' \ '), 1):

Bare except

Generally this is a bad idea. At the least, catch Exception instead, or more preferably the actual exception type you expect to see. If it's only an index error because you're trying to index into a split that has fewer than four components, just avoid that situation altogether:

parts = s.split(r' \ ')
for i, part in enumerate(parts, 1):
    mbo.setValue(col_prefix + str(i), part)
for i in range(len(parts)+1, 5):
    mbo.setValueNull(col_prefix + str(i))