Support read unstructured excel file

[khm0651]

It has been confirmed that the current Dataframe.readExcel function only supports structured Excel formats.

Of course, it would be ideal if everyone created Excel files in a structured format, but as shown in the image, when dealing with unstructured Excel formats, the current DataFrame approach of always designating the first row as the header causes difficulties in usage.

The Python pandas library supports this, making it very efficient to use.

As a result, for unstructured Excel formats, i've implemented support by using a value called withDefaultHeader.

When set to true, it automatically generates headers using NameRepairStrategy,
thus enabling support for unstructured Excel formats.

when withDefaultHeader is set to true, it operates as [NameRepairStrategy.MAKE_UNIQUE]

Is there a better approach?

[Jolanrensen]

Thanks for your contribution!

Some thoughts:

So if I'm correct, withDefaultHeader = true will make the columns be named according to excel columns, like "A", "B", "C" etc.? It's not really clear to me this will happen from your description of the argument.

I might change it to firstRowIsHeader = true as the argument, then describing that when that's set to false explicitly, DF will fall-back to excel letter column names, else it will take the first row (after skipRows) as the header.

Also, am I correct that in the current implementation, column J will not be included in the result?

[khm0651]

Oh, it's a bit different. When withDefaultHeader is set to true, it doesn't use the row specified by skipRow as the header. If skipRow is set, it retrieves data starting from the row specified by skipRow, using automatically generated headers.
Excel columns are automatically generated as "A", "B", "C", etc.. and so on.

However, as @Jolanrensen pointed out, I hadn't considered that particular case.
It seems we need to address this case as well
to accommodate this case, should calculate and incorporate a maximum value.

and i agree change it to firstRowIsHeader = true as the argument too

[Jolanrensen]

I might name this largestRow or something like that. Good solution though :) didn't know about lastCellNum

[Jolanrensen]

I recently learned about bytecode compatibility which will break when adding this extra argument (https://kotlinlang.org/docs/api-guidelines-backward-compatibility.html#avoid-adding-arguments-to-existing-api-functions). This means that when you use a library that depends on DataFrame 0.14 while your own project depends on DataFrame 0.15 and that library calls readExcel(), you'll get a NoSuchMethodError. We're adding a detector plugin for these cases.
We can solve this, however, by adding some extra DeprecationLevel.HIDDEN overloads of the old version of these functions. I could do this for you if you like :) The rest looks good, so afterwards we could merge it.

[khm0651]

Thank you so much for the review and for accepting my answer!
I truly appreciate it.