feat: Add evm bytes to python converter (#357)

* feat: add evm bytes to python converter

* entry_points/tests: add tests for evm bytes parser.

* evm_bytes_to_python: Add help

* changelog

---------

Co-authored-by: spencer-tb <[email protected]>

docs/CHANGELOG.md

@@ -13,6 +13,7 @@ Test fixtures for use by clients are available for each release on the [Github r
 - ✨ Add a `--single-fixture-per-file` flag to generate one fixture JSON file per test case ([#331](https://github.com/ethereum/execution-spec-tests/pull/331)).
 - 🔀 Rename test fixtures names to match the corresponding pytest node ID as generated using `fill` ([#342](https://github.com/ethereum/execution-spec-tests/pull/342)).
 - 💥 Replace "=" with "_" in pytest node ids and test fixture names ([#342](https://github.com/ethereum/execution-spec-tests/pull/342)).
+- ✨ Add `evm_bytes_to_python` command which converts EVM bytes to Python Opcodes ([#357](https://github.com/ethereum/execution-spec-tests/pull/357))
 - 🔀 Locally calculate the transactions list's root instead of using the one returned by t8n when producing BlockchainTests ([#353](https://github.com/ethereum/execution-spec-tests/pull/353))
 - ✨ Fork objects used to write tests can now be compared using the `>`, `>=`, `<`, `<=` operators, to check for a fork being newer than, newer than or equal, older than, older than or equal, respectively when compared against other fork ([#367](https://github.com/ethereum/execution-spec-tests/pull/367))
 - 🐞 Storage type iterator is now fixed ([#369](https://github.com/ethereum/execution-spec-tests/pull/369))

setup.cfg

@@ -50,6 +50,7 @@ console_scripts =
     pyspelling_soft_fail = entry_points.pyspelling_soft_fail:main
     markdownlintcli2_soft_fail = entry_points.markdownlintcli2_soft_fail:main
     create_whitelist_for_flake8_spelling = entry_points.create_whitelist_for_flake8_spelling:main
+    evm_bytes_to_python = entry_points.evm_bytes_to_python:main
 
 [options.extras_require]
 test =

src/entry_points/evm_bytes_to_python.py

@@ -0,0 +1,57 @@
+"""
+Define an entry point wrapper for pytest.
+"""
+
+import sys
+from typing import Any, List, Optional
+
+from ethereum_test_tools import Opcodes as Op
+
+
+def process_evm_bytes(evm_bytes_hex_string: Any) -> str:  # noqa: D103
+    if evm_bytes_hex_string.startswith("0x"):
+        evm_bytes_hex_string = evm_bytes_hex_string[2:]
+
+    evm_bytes = bytearray(bytes.fromhex(evm_bytes_hex_string))
+
+    opcodes_strings: List[str] = []
+
+    while evm_bytes:
+        opcode_byte = evm_bytes.pop(0)
+
+        opcode: Optional[Op] = None
+        for op in Op:
+            if op.int() == opcode_byte:
+                opcode = op
+                break
+
+        if opcode is None:
+            raise ValueError(f"Unknown opcode: {opcode_byte}")
+
+        if opcode.data_portion_length > 0:
+            data_portion = evm_bytes[: opcode.data_portion_length]
+            evm_bytes = evm_bytes[opcode.data_portion_length :]
+            opcodes_strings.append(f'Op.{opcode._name_}("0x{data_portion.hex()}")')
+        else:
+            opcodes_strings.append(f"Op.{opcode._name_}")
+
+    return " + ".join(opcodes_strings)
+
+
+def print_help():  # noqa: D103
+    print("Usage: evm_bytes_to_python <EVM bytes hex string>")
+
+
+def main():  # noqa: D103
+    if len(sys.argv) != 2:
+        print_help()
+        sys.exit(1)
+    if sys.argv[1] in ["-h", "--help"]:
+        print_help()
+        sys.exit(0)
+    evm_bytes_hex_string = sys.argv[1]
+    print(process_evm_bytes(evm_bytes_hex_string))
+
+
+if __name__ == "__main__":
+    main()

src/entry_points/tests/__init__.py

@@ -0,0 +1,3 @@
+"""
+Basic pytest applications `entry_points` unit tests.
+"""

src/entry_points/tests/test_evm_bytes_to_python.py

@@ -0,0 +1,55 @@
+"""
+Test suite for `entry_points.evm_bytes_to_python` module.
+"""
+
+import pytest
+from evm_bytes_to_python import process_evm_bytes
+
+from ethereum_test_tools import Opcodes as Op
+
+basic_vector = [
+    "0x60008080808061AAAA612d5ff1600055",
+    'Op.PUSH1("0x00") + Op.DUP1 + Op.DUP1 + Op.DUP1 + Op.DUP1 + Op.PUSH2("0xaaaa") + Op.PUSH2("0x2d5f") + Op.CALL + Op.PUSH1("0x00") + Op.SSTORE',  # noqa: E501
+]
+complex_vector = [
+    "0x7fa0a1a2a3a4a5a6a7a8a9aaabacadaeafb0b1b2b3b4b5b6b7b8b9babbbcbdbebf5f527fc0c1c2c3c4c5c6c7c8c9cacbcccdcecfd0d1d2d3d4d5d6d7d8d9dadbdcdddedf6020527fe0e1e2e3e4e5e6e7e8e9eaebecedeeeff0f1f2f3f4f5f6f7f8f9fafbfcfdfeff60405260786040356020355f35608a565b5f515f55602051600155604051600255005b5e56",  # noqa: E501
+    'Op.PUSH32("0xa0a1a2a3a4a5a6a7a8a9aaabacadaeafb0b1b2b3b4b5b6b7b8b9babbbcbdbebf") + Op.PUSH0 + Op.MSTORE + Op.PUSH32("0xc0c1c2c3c4c5c6c7c8c9cacbcccdcecfd0d1d2d3d4d5d6d7d8d9dadbdcdddedf") + Op.PUSH1("0x20") + Op.MSTORE + Op.PUSH32("0xe0e1e2e3e4e5e6e7e8e9eaebecedeeeff0f1f2f3f4f5f6f7f8f9fafbfcfdfeff") + Op.PUSH1("0x40") + Op.MSTORE + Op.PUSH1("0x78") + Op.PUSH1("0x40") + Op.CALLDATALOAD + Op.PUSH1("0x20") + Op.CALLDATALOAD + Op.PUSH0 + Op.CALLDATALOAD + Op.PUSH1("0x8a") + Op.JUMP + Op.JUMPDEST + Op.PUSH0 + Op.MLOAD + Op.PUSH0 + Op.SSTORE + Op.PUSH1("0x20") + Op.MLOAD + Op.PUSH1("0x01") + Op.SSTORE + Op.PUSH1("0x40") + Op.MLOAD + Op.PUSH1("0x02") + Op.SSTORE + Op.STOP + Op.JUMPDEST + Op.MCOPY + Op.JUMP',  # noqa: E501
+]
+
+
+@pytest.mark.parametrize(
+    "evm_bytes, python_opcodes",
+    [
+        (basic_vector[0], basic_vector[1]),
+        (basic_vector[0][2:], basic_vector[1]),  # no "0x" prefix
+        (complex_vector[0], complex_vector[1]),
+        (complex_vector[0][2:], complex_vector[1]),  # no "0x" prefix
+    ],
+)
+def test_evm_bytes_to_python(evm_bytes, python_opcodes):
+    """Test evm_bytes_to_python using the basic and complex vectors"""
+    assert process_evm_bytes(evm_bytes) == python_opcodes
+
+
+@pytest.mark.parametrize("opcode", list(Op))
+def test_individual_opcodes(opcode):
+    """Test each opcode individually"""
+    if opcode.data_portion_length > 0:
+        expected_output = f'Op.{opcode._name_}("0x")'
+    else:
+        expected_output = f"Op.{opcode._name_}"
+
+    bytecode = opcode.int().to_bytes(1, byteorder="big").hex()
+    assert process_evm_bytes("0x" + bytecode) == expected_output
+
+
+def test_invalid_opcode():
+    """Invalid hex string"""
+    with pytest.raises(ValueError):
+        process_evm_bytes("0xZZ")
+
+
+def test_unknown_opcode():
+    """Opcode not defined in Op"""
+    with pytest.raises(ValueError):
+        process_evm_bytes("0x0F")