Resolving names vs. Parsing quickly

[jalanmiddleton]

When parsing formulas, the POI FormulaParser asks for the spreadsheet context (including file, sheet number) of a formula before breaking it down, so it can resolve user-defined names.

• If you pass in the actual workbook, it can resolve names and use the formula as more data at the cost of more time in parsing.
• If you pass in a blank workbook, every formula will parse in a jiffy but functions with names will be lost.

First step: Find the portion of the total formula corpus that uses names or other elements that rely on context..

[jalanmiddleton]

Of the 17585500 formulas in the database, 1375301 (~7.8%) fail to parse with a blank worksheet. I think it's worth looking into workarounds, since this is a big loss, yet the delay that reading in a worksheet incurs can be a burden.

Maybe I can get POI to replace names with their stored value and replace those records in the DB? It would be a single sweep across to find the troublemakers and replace them, then blank-workbook parsing would be quick afterward.

We'd lose the ability to analyze anything about name use with that approach, though, if that's important.