0.5.2 Feature Request: Reader parse-only mode.
qg020 opened this issue · 7 comments
I am reading ARXML we have generated and it is not to the Reader's liking. I have iterated through fixing each raised exception. A nice feature would be a 'parse-only' flag that would return a list of all lines in an ARXML file needing attention and why it is a problem.
Does the parsing stop because an exception is raised by the parser? In that case I can add an option to continue parsing and not stop by raising an error.
Also, have you verified your ARXML file against the schema file before trying to parse?
I always use xmllint tool to verify any XML file first. I always check against schema versions 48, 49, 50 and 51. If your ARXML uses any other version I cannot give any support.
Finally, the new code base is still in early development. I haven't tried it myself yet to parse any XML generated by other tool or script. I will start testing it later, somewhere after v0.5.4.
Hi cogu, thanks for the response. ARXML input files mention r4.0 AUTOSAR_4-2-2. The problem is definitely in our ARMXL which have duplicate entries. All errors reported by the Reader indicate that. Which is great. Just a convenience thing if you could ask to see all the parsing problems at once.
Low hanging fruit for when you have time (haha). Thanks!
When I have time I plan to add a boolean flag to the reader class named something like "stop_on_error" which defaults to True.
Manually setting it to False should get the effect you want.
But before I can add that I have to refactor all the places in the Reader code that raises an exception to only print a warning based on this flag. It's gonna take some time to do.
@rcd09, Actually when I think about it I would prefer to keep most of the exceptions and instead catch them in a try-except block in the method _read_package_elements.
- If the element is too broken to parse then we just skip it and move on to the next element
- If the element is a duplicate, as seems to be the case for @qg020 then the line
package.append(element)inside_read_package_elementswill raise an exception instead, which we can catch in the same place.
I can make the change to the _read_package_elements. It should be a quick fix.
@qg020, @rcd09, I've made a first attempt of the change on latest master. There is still a lot of improvements that can be made but it should be good enough for now.
I decided to let stop_on_error be False as default. Now you have to explicitly set it to True to stop reading the XML.
See print_errors.py example with an invalid ARXML containing duplicates.