leethomason
GetText() on element containing only whitespace characters returns null
Not sure if it's a bug or feature. Yet it can then be a little bit tricky when you really need to work with the exact text within an element.
Code:
std::string str = "<textData>example</textData>";
tinyxml2::XMLDocument doc;
doc.Parse(str.c_str());
doc.FirstChildElement()->GetText();
Works as expected with:
str = "<textData>example</textData>";
// GetText() gets: "example"
str = "<textData> example </textData>";
// GetText() gets: " example "
But returns NULL when there are only whitespace characters inside the element:
str = "<textData> </textData>";
// GetText() returns NULL
Feature, in the broad sense, in that TinyXML-2 skips all whitespace elements. In this case, it's obvious that the whitespace is intended (to a human reading the file). On the other hand:
<textData>
</textData>
Isn't clear - should that newline be preserved? Extra trailing spaces?
I think eliminating all whitespace areas is usually the correct choice, but I'd be open to an algorithm that tried to be a little smarter, if it doesn't make the rules too weird.
Closing for now. May reopen if a good strategy to handle whitespace nodes is proposed.
An option over whether leaf nodes with all whitespace are preserved or collapsed seems like a reasonable compromise approach. That's the essential difference between the different kinds of whitespace in fragments like the following.
<body>
<text> </text>
</body>`
It also lets the user choose how to treat
<textData>
</textData>
Re-opening. I don't know if we'll ever make everyone happy on this issue, but this is a nice summary.
@leethomason is there any plan to fix this issue? I think in both cases, we should preserve the space if PRESERVE_WHITESPACE flag is set.
