If merely gaining the ability to write scripts is your goal, your approach poses no problem at all. However, if building scripts that compile and appear to work (on a limited sample of data) is the extent of your ambition, you'll be missing a bet. One of the most highly valuable aspects of the Monastery is that you can call for help upon a community whose population includes some extraordinarily skilled programmers and whose topic is a wonderfully versatile language.

High level programmer skills go well beyond syntax.

If your data set is narrowly constrained, and you won't have to do too many s/// (where the value of "many" is "small" or "tolerable to you") that's a perfectly valid approach.

Answering that (for yourself; I can't without a better grasp of the data set) requires analysis:

• What are the elements in your data that must retain all UC form?
• Is retaining the UC form really required?
  "Ltd." (mixed case with period/fullstop to denote abbreviation) seems quite as comprehensible as "LTD" -- unless your transformation is a step toward some other, later process which requires the form "LTD" (say, during the use of a program you can't modify).

As for the rest of your questions... go for it! I too tend to learn in exactly that mode.

However, as a favor to yourself AND to those who seek to help, please look hard at the preview of your posts for typos and/or language errors (for ex., "uncial" and "on" where "in" appears to be intended, above) that may obscure your meaning.