Extend parser to include specialized EntryData sections

[ka-sarthak]

The definition of the EntryData class that is generated by the MatchingParser has been abstracted out into an attribute entrydata_definition. This is set in the set_entrydata_definition method.

Benefit:

• In a situation where a specialized EntryData section is defined for the XRD entry, the XRDParser can be reused by overwriting set_entrydata_definition method. For example:

class SpecializedELNXRayDiffraction(ELNXRayDiffraction):
 # additional quantities
 def normalize(self, archive, logger):
   # additional normalization
   pass

class SpecializedXRDParser(XRDParser):
 def set_entrydata_definition(self):
   self.entrydata_definition = SpecializedELNXRayDiffraction

[ka-sarthak]

@aalbino2 can you check if this works in your case?

[aalbino2]

I will implement this solution in a parser that will be included in nomad-measurements soon. I think this can be merged. Thanks for figuring out a solution that will be useful for all the parsers.

[aalbino2]

hey @ka-sarthak, apparently @JonathanNoky cannot make this work. he need to call the function inside the parse function:

   def parse(self, mainfile: str, archive: EntryArchive, logger) -> None:
        self.set_entrydata_definition()

what do you think? shouldn't it be already executed without calling it?

or do we need to really include it into the init method that is always run?

Indeed the linter says "W0201:attribute-defined-outside-init"

[JonathanNoky]

Hey @ka-sarthak , after looking into your example again I figured it out. I missed the part where the set_entrydata_definition is called at the beginning of parse().
It is working now.

[ka-sarthak]

The idea is to override the setter function that sets the entrydata definition. In a specialized MatchingParser class, you override this setter function and that's it.

@aalbino2 you are right, self.entrydata_definition is being defined outside the init method in this case. Maybe we can make it an attribute of the parent parser and update it in the child's parse method. Like:

class XRDParser(MatchingParser):
  entrydata_definition: str = ELNXRayDiffraction
  def parse(self):
    ...

class SpecializedXRDParser(XRDParser):
  def parse(self):
    self.entrydata_definition = SpecializedELNXRayDiffraction
    super().parse()

[aalbino2]

The idea is to override the setter function that sets the entrydata definition. In a specialized MatchingParser class, you override this setter function and that's it.

@aalbino2 you are right, self.entrydata_definition is being defined outside the init method in this case. Maybe we can make it an attribute of the parent parser and update it in the child's parse method. Like:

class XRDParser(MatchingParser):
  entrydata_definition: str = ELNXRayDiffraction
  def parse(self):
    ...

class SpecializedXRDParser(XRDParser):
  def parse(self):
    self.entrydata_definition = SpecializedELNXRayDiffraction
    super().parse()

okay, I think this alternative can make sense.

@JonathanNoky will you try this solution in ppms instead of the one previously proposed?

[ka-sarthak]

I haven't tested it, so please take it with a grain of salt

[JonathanNoky]

Yes, this one also works. Its shorter than the other method, as there is just one line in the parse function.

[JonathanNoky]

Actually, after some thinking about it, I think, the method of just setting the attribute in the parse function does not work for more nested parser classes like this:

  entrydata_definition: str = ELNXRayDiffraction
  def parse(self):
    ...

class SpecializedXRDParser(XRDParser):
  def parse(self):
    self.entrydata_definition = SpecializedELNXRayDiffraction
    super().parse()

class EvenMoreSpecializedXRDParser(SpecializedXRDParser):
  def parse(self):
    self.entrydata_definition = EvenMoreSpecializedELNXRayDiffraction
    super().parse()

Here, the EvenMoreSpecializedXRDParser would via the super().parse() function still get the SpecializedELNXRayDiffraction defintion, right?

So I think, I will have to go back to the first proposed solution with the separate definition function. I will try it in my CPFSPPMSParser plugin, as this has this kind of deeper nested parsers when I let it depend on the nomad-measurements plugin.

[ka-sarthak]

@JonathanNoky That's a good point. Having a setter function allows for more flexibility and deeper nesting.