[C#]:Type recovery and data flow completion in CPG about joern HOT 6 OPEN

Hi @chengkenyong, congrats! So the simplest would be to have a look at how javasrc2cpg tests a similar structure, and to check using something like dotAst/dotCfg/dotDdg to determine what the differences are. My guess is, usually the equivalent data-flow is supported in Java, but that the resulting AST in C# might be a bit off and need adjusting in order to produce the same CFG/DDG that will eventually give you your flow. If your finding isn't obvious, feel free to post the comparison in this thread and I can continue to guide you

from joern.

Hi @chengkenyong, development of C# frontend has been largely funded by Privado, but they have since scaled it back a bit due to priorities on other languages. The C# frontend relies on two components (1) the Roslyn parser wrapped in a neat binary which can be found here https://github.com/joernio/DotNetAstGen, and (2) the AstCreation in Joern.

For the missing AST features you've described, it could be that they need to be filtered in on the parser. So that change would be in DotNetAstGen. One way to check is to create a JSON AST dump of the file you're interested in, and see if it is present.

If the entity is there, then it probably needs a handler in one of the classes within the astcreation package in csharpsrc2cpg.

Regarding type information, it is not uncommon for fieldAccess nodes to not have type info, especially if the receiver is an object, since this might have multiple types in the case of polymorphism. This may require a bit more of a "dynamic" query that looks at the fieldAccess.argument(1).evalType.baseTypeDeclTransitive step to find all possible types, and then determine which of those types have members matching the canonicalName of the fieldIdentifier.

from joern.

Hi @DavidBakerEffendi , Your explanation has been very helpful to me, and I will follow your advice to learn more about the relevant topics. Thank you for your assistance.

from joern.

Hi @DavidBakerEffendi ，Based on your suggestion, I have restored the missing type of the fieldAccess node to the desired state in the first case. However, I am still unclear about how to address the third case of missing data flow. Specifically, I want to restore the data flow in the third case where the variable childSqlMapSource cannot reach sqlMapPath through the foreach loop. Could you please provide me with some more advice on how to achieve this? Thank you!

from joern.