feat(sighash): implement sighash bitmask [part 1/7]

[glevco]

Depends on #851

Motivation

Implement the new SIGHASH structure. This PR implements everything necessary for creating P2PKH inputs/outputs using OP_SIGHASH_BITMASK. That is, creating atomic swap txs is already fully functional, but gated to prevent a fork (this will later be toggled by Feature Activation).

Adding support for OP_SIGHASH_RANGE and MultiSig is straightforward and will be done in separate PRs. Tests for new functionality will also be implemented in a separate PR.

Acceptance Criteria

• New models SighashAll, SighashBitmask, SighashRange, and SighashLimit.
• Implement evaluate_final_context() and change execute_eval() to call it.
• Change execute_eval() and script_eval() to return the selected outputs from the context.
• Create new opcodes and implement op_sighash_bitmask() and op_max_inputs_outputs().
• Update op_checksig() to use the ScriptContext for signature verification.
• Implement is_opcode_valid() and use it to gate new opcodes in execute_op_code().
• Update P2PKH.create_input_data() to support SighashBitmask and SighashLimit.
• Update ScriptContext with methods for interacting with sighash.
• Implement Transaction.get_custom_sighash_data() and change get_sighash_all() to unify implementations.
• Update TransactionVerifier.verify_inputs() to check for unsigned outputs.
• Update existing tests.

[codecov]

Codecov Report

Attention: Patch coverage is 63.33333% with 66 lines in your changes missing coverage. Please review.

Project coverage is 84.12%. Comparing base (dbcff55) to head (922055e).

Files with missing lines	Patch %	Lines
hathor/transaction/scripts/opcode.py	29.78%	32 Missing and 1 partial ⚠️
hathor/transaction/scripts/script_context.py	54.83%	12 Missing and 2 partials ⚠️
hathor/transaction/scripts/p2pkh.py	20.00%	6 Missing and 2 partials ⚠️
hathor/transaction/transaction.py	66.66%	6 Missing ⚠️
hathor/transaction/scripts/sighash.py	90.00%	3 Missing ⚠️
hathor/verification/transaction_verifier.py	84.61%	1 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #911      +/-   ##
==========================================
- Coverage   84.26%   84.12%   -0.15%     
==========================================
  Files         320      321       +1     
  Lines       24455    24601     +146     
  Branches     3744     3773      +29     
==========================================
+ Hits        20608    20696      +88     
- Misses       3113     3168      +55     
- Partials      734      737       +3     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[msbrogli]

How can we differentiate a code bug from a bad input? I feel that we should capture only the exceptions raised by the bytes_to_int() and let all others blow up.

[glevco]

Changed in b611297 to catch only pydantic.ValidationError. The bytes_to_int() function does not raise any exceptions explicitly, so the only expected exception to be raised here is the model validation from pydantic.

However, I believe this is a shortcoming caused by the try-except model itself. There's no way to formally differentiate between exceptions caused by bugs or bad inputs, because there's no formally typed list of expected exceptions. In my opinion we should improve this by using result monads throughout the code.

[msbrogli]

Same here.

[glevco]

Done in b611297, see #911 (comment)

[github-actions]

Bencher Report

Branch	feat/sighash/bitmask
Testbed	ubuntu-22.04

Click to view all benchmark results

Benchmark	Latency	Benchmark Result minutes (m) (Result Δ%)	Lower Boundary minutes (m) (Limit %)	Upper Boundary minutes (m) (Limit %)
sync-v2 (up to 20000 blocks)	📈 view plot 🚷 view threshold	1.50 m (-7.99%) Baseline: 1.63 m	1.47 m (97.82%)	1.79 m (83.65%)

🐰 View full continuous benchmarking report in Bencher

[msbrogli]

As you're already changing the method signature, you can put it all as kwargs only, right? The docstring is also outdated with the new signature.

[glevco]

Done in 2f8579f

[msbrogli]

You can use all_selected_outputs.update(selected_outputs) to prevent creating a new set.

[glevco]

Done