On Sat, 28 Jan 2017, Greg Woolsey wrote:
As noted in one of the method JavaDocs, we also need to expose and make use
of the ConditionalFormattingRule "priority" attribute. That's key to
matching the right rule when more than one rule applies to a cell. Only
the first match in priority order is applied.
Your slight challenge is that not all Conditional Formatting rules have a
priority... XLSX ones do, and newer XLS ones based on CFRule12Record (sid
= 0x087A) do, but the older XLS ones (CFRuleRecord / 0x01B1) don't. I'm
not sure what Excel does for those, but my hunch (based on our API) is
that it uses their order as a priority.
I've created a fork in GitHub for this, and committed a stab at
high-level API methods that could be added to the FormulaEvaluator
interface:
https://github.com/WoozyG/poi/commit/d44fee7bd03ed450af589467ec90e2581b9f2b16$
FormulaEvaluator is an interface, which we have 4 implementations of in
our codebabse, and I'd guess that other complex users of POI will have
dozens more. I'm not sure, therefore, that we want to be putting all of
the CF and DV logic methods on there, especially as it'll be common to all
implementations
The HSSF classes for CF all use org.apache.poi.ss.formula.Formula which is
PTG based. The HSSF classes for DV seem to store the raw PTGs.
If we added two new SS usermodel classes, eg
ConditionalFormattingEvaluator and DataValidationEvaluator, these could be
classes (not interfaces) with your proposed new methods on. They could
hold the logic (once) for all formats (as it's basically the same on all)
for priority, checking etc
Doing that would also mean that "our" new classes could call out to our
existing low-level ones to evaluate formulas. That would mean we wouldn't
have to make a breaking change to the FormulaEvaluator interface too
Might that work for you?
No implementations have been done yet, and the Vaadin comments indicate
HSSF doesn't parse conditional formatting properly or something, and can't
be evaluated correctly currently. I don't know exactly what they found
wrong, and it's rather annoying they didn't file any bugs.
I think that comment is out of date, from before the CF work in 3.13
Nick
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
