Hmm, I know it is hard to suggest anything without knowing the AUT.
Stripping would take ages because even the custom tagnames are treated as sensitive ...
I try to explain further:
What this is all about, is comparing elements in a table with known values.
But - of course that would be too easy - it is not a table webelement, it has no table capabilities, there are no IDs or other unique properties for any cell.
It is just a bunch of divs basically.
Headers are not linked in any way to data cells, all needs to be built up by hand.
In addition, the tests have to run in 2 different environments (A, B) , which are basically the same except the fact, that there is an additional icon column added as first column in environment B, which i need to skip.
So, in environment A this table looks like:
Code: Select all
Column 1 | Column 2 | Column 3 | ...
value1 | value2 | value3 | ...
environment B looks the same, but has an icon colum added as first column:
Code: Select all
Icon | Column 1 | Column 2 | Column 3 | ...
| value1 | value2 | value3 | ...
In environment A the data row looks like:
Code: Select all
<dataRow>
<tag> --> data cell
<div>
<div> --> value1
<tag> --> data cell
<div>
<div> --> value2
...
and in environment B:
Code: Select all
<dataRow>
<tag> --> icon cell
<div>
<tag>
<tag> --> data cell
<div>
<div> --> value1
<tag> --> data cell
<div>
<div> --> value2
...
So for environment A the cell values could be easily linked statically as repository items like:
Code: Select all
tag[1]/div/div --> value1
tag[2]/div/div --> value2
...
which does not work anymore in environment B, because tag[1]/div/div is not found there.
Without making situation more complicated and using different recording and other code-based stuff I would like to set the RXpath accordingly, so object recognition works in each of these environments.
Is this more understandable now?