add abi.decode strictness flag and catch InsufficientDataBytes error …

[pacrob]

…when parsing logs

This should use the existing global strictness flag.

What was wrong?

Related to Issue #1441 , if a user is decoding logs with length not a multiple of 32 bytes, it throws an eth-abi InsufficientBytesLength error which is not properly caught.

How was it fixed?

Add catching the error type InsuffiecientDataBytes and allow the passing of strict flag through to eth_abi.decode in process_receipt.

Since the flag is then passed to get_event_data, I considered making the flag available everywhere get_event_data is called, but that would greatly expand the scope and doesn't seem necessary, given the original issue.

Todo:

• Add entry to the release notes

Cute Animal Picture

[fselmo]

Instead of creating a new flag and passing that flag around, we should let this be determined by the Web3.strict_bytes_type_checking flag. We should use this flag in the decoder for web3 to mark the strict flag value for the abi methods. Thoughts there?

[fselmo]

Instead of creating a new flag and passing that flag around, we should let this be determined by the Web3.strict_bytes_type_checking flag. We should use this flag in the decoder for web3 to mark the strict flag value for the abi methods. Thoughts there?

Building on this, it looks like the previous version of the global strict bytes checking only re-builds encoders with is_strict set to False. We could, and maybe should, build decoders with these values as well and override certain areas where the strictness is checked. I'm not sure how much more effort this would take nor that it is the right avenue since it might just re-write a lot of the eth-abi code.

If we can find a good balance to set the strictness at the decoder level, then we can maybe even solve issues like #2489 when we pass in the w3.codec if w3.strict_bytes_type_checking is set to False. This would ensure the decoder itself is already essentially using strict=False when it decodes.

I think we should take a step back and re-evaluate how we want to add this globally since it doesn't seem like the global flag is handling decoding strictness very gracefully.

---

edit: Looking at the eth-abi code again, I think we just need to override the strict property on each decoder when we build the non-strict registry which should be pretty simple. But don't want to speak too soon.

Curious on your thoughts there.

[pacrob]

As far as I can tell, I don't think this will work without updates to eth-abi. The strict flag only has any effect on decoding bytes and string types (not bytes[], not bytes2, etc.) Since logs can be anything, it wouldn't work for most issues, the example in #2489 being one.

From eth_abi/registry.py:

    def _get_decoder_uncached(self, type_str, strict=True):
        decoder = self._get_registration(self._decoders, type_str)

        if hasattr(decoder, "is_dynamic") and decoder.is_dynamic:
            # Set a transient flag each time a call is made to ``get_decoder()``.
            # Only dynamic decoders should be allowed these looser constraints. All
            # other decoders should keep the default value of ``True``.
            decoder.strict = strict

        return decoder

[pacrob]

Let me know if I'm missing something. Otherwise, I think scaling this back to just catching the InsufficientDataBytes error is the way to go for now.