[kxml_compiler]Fix segfault when parsing an XSD containing a sequence type rootitem #49
Conversation
martonmiklos
force-pushed
the
fix_sequence_root_item_parsing
branch
2 times, most recently
from
bb0b86e
to
2e65897
Compare
XSD where the root element was a complexType which was a sequence of elements. This was becasuse the subelements was not added by the schema/ parser class during the parse.
martonmiklos
force-pushed
the
fix_sequence_root_item_parsing
branch
from
2e65897
to
070423a
Compare
| @@ -260,7 +260,8 @@ ClassDescription Creator::createClassDescription(const Schema::Element &element) | |||
|
|
|||
| QString targetClassName = Namer::getClassName(targetElement.name()); | |||
|
|
|||
| if (targetElement.text() && !targetElement.hasAttributeRelations() && !r.isList()) { | |||
| if ((targetElement.text() || targetElement.type() < Schema::Element::ComplexType) | |||
There was a problem hiding this comment.
Why < and not just !=?
There was a problem hiding this comment.
Why
<and not just!=?
Makes sense I will fix it!
There was a problem hiding this comment.
This probably means there's no need for it to be last anymore, i.e. you could revert the change to the enum, unless I'm missing something?
There was a problem hiding this comment.
Yeah. When I started to work on this I was not 100% confident that no other kind of types needs to be treated similar to complexTypes. But at the end it turned out that no need for that.
Also it might be useful to remove/refactor the Schema::Element::text in this PR because I think the name 'text' is a bit misleading and from the type (complexType or not) and from the number of attributes and relations it is possible to determine that a simple C++ type (QString/int/something similar) member or a class representing the questioned element shall be generated.
The kxml_compiler segfaulted when generated code from an XSD where the root element was a complexType which was a sequence of elements. This was becasuse the subelements was not added by the schema/parser class during the parse.