prepare_report.py: Make parsing more lax to handle output from older compiler versions

scripts/bytecodecompare/prepare_report.js

@@ -16,6 +16,13 @@ function loadSource(sourceFileName, stripSMTPragmas)
     return source
 }
 
+function cleanString(string)
+{
+    if (string !== undefined)
+        string = string.trim()
+    return (string !== '' ? string : undefined)
+}
+
 
 let stripSMTPragmas = false
 let firstFileArgumentIndex = 2
@@ -78,10 +85,15 @@ for (const optimize of [false, true])
                         let bytecode = '<NO BYTECODE>'
                         let metadata = '<NO METADATA>'
 
-                        if ('evm' in contractResults && 'bytecode' in contractResults['evm'] && 'object' in contractResults['evm']['bytecode'])
+                        if (
-                            bytecode = contractResults.evm.bytecode.object
+                            'evm' in contractResults &&
+                            'bytecode' in contractResults['evm'] &&
+                            'object' in contractResults['evm']['bytecode'] &&
+                            cleanString(contractResults.evm.bytecode.object) !== undefined
+                        )
+                            bytecode = cleanString(contractResults.evm.bytecode.object)
 
-                        if ('metadata' in contractResults)
+                        if ('metadata' in contractResults && cleanString(contractResults.metadata) !== undefined)
                             metadata = contractResults.metadata
 
                         console.log(filename + ':' + contractName + ' ' + bytecode)

scripts/bytecodecompare/prepare_report.py

@@ -13,9 +13,9 @@ from tempfile import TemporaryDirectory
 from typing import List, Optional, Tuple, Union
 
 
-CONTRACT_SEPARATOR_PATTERN = re.compile(r'^======= (?P<file_name>.+):(?P<contract_name>[^:]+) =======$', re.MULTILINE)
+CONTRACT_SEPARATOR_PATTERN = re.compile(r'^ *======= +(?:(?P<file_name>.+) *:)? *(?P<contract_name>[^:]+) +======= *$', re.MULTILINE)
-BYTECODE_REGEX = re.compile(r'^Binary:\n(?P<bytecode>.*)$', re.MULTILINE)
+BYTECODE_REGEX = re.compile(r'^ *Binary: *\n(?P<bytecode>.*[0-9a-f$_]+.*)$', re.MULTILINE)
-METADATA_REGEX = re.compile(r'^Metadata:\n(?P<metadata>\{.*\})$', re.MULTILINE)
+METADATA_REGEX = re.compile(r'^ *Metadata: *\n *(?P<metadata>\{.*\}) *$', re.MULTILINE)
 
 
 class CompilerInterface(Enum):
@@ -32,7 +32,7 @@ class SMTUse(Enum):
 @dataclass(frozen=True)
 class ContractReport:
     contract_name: str
-    file_name: Path
+    file_name: Optional[Path]
     bytecode: Optional[str]
     metadata: Optional[str]
 
@@ -72,6 +72,11 @@ def load_source(path: Union[Path, str], smt_use: SMTUse) -> str:
     return file_content
 
 
+def clean_string(value: Optional[str]) -> Optional[str]:
+    value = value.strip() if value is not None else None
+    return value if value != '' else None
+
+
 def parse_standard_json_output(source_file_name: Path, standard_json_output: str) -> FileReport:
     decoded_json_output = json.loads(standard_json_output.strip())
 
@@ -98,8 +103,8 @@ def parse_standard_json_output(source_file_name: Path, standard_json_output: str
             file_report.contract_reports.append(ContractReport(
                 contract_name=contract_name,
                 file_name=Path(file_name),
-                bytecode=contract_results.get('evm', {}).get('bytecode', {}).get('object'),
+                bytecode=clean_string(contract_results.get('evm', {}).get('bytecode', {}).get('object')),
-                metadata=contract_results.get('metadata'),
+                metadata=clean_string(contract_results.get('metadata')),
             ))
 
     return file_report
@@ -122,10 +127,10 @@ def parse_cli_output(source_file_name: Path, cli_output: str) -> FileReport:
 
         assert file_report.contract_reports is not None
         file_report.contract_reports.append(ContractReport(
-            contract_name=contract_name,
+            contract_name=contract_name.strip(),
-            file_name=Path(file_name),
+            file_name=Path(file_name.strip()) if file_name is not None else None,
-            bytecode=bytecode_match['bytecode'] if bytecode_match is not None else None,
+            bytecode=clean_string(bytecode_match['bytecode'] if bytecode_match is not None else None),
-            metadata=metadata_match['metadata'] if metadata_match is not None else None,
+            metadata=clean_string(metadata_match['metadata'] if metadata_match is not None else None),
         ))
 
     return file_report

test/scripts/fixtures/solc_0.4.0_cli_output.txt

@@ -0,0 +1,8 @@
+contract.sol:1:1: Warning: Source file does not specify required compiler version! Consider adding "pragma solidity ^0.4.0;".
+contract C {}
+^
+Spanning multiple lines.
+
+======= C =======
+Binary:
+6060604052600c8060106000396000f360606040526008565b600256

test/scripts/fixtures/solc_0.4.8_cli_output.txt

@@ -0,0 +1,10 @@
+contract.sol:1:1: Warning: Source file does not specify required compiler version!Consider adding "pragma solidity ^0.4.8
+contract C {}
+^
+Spanning multiple lines.
+
+======= C =======
+Binary:
+6060604052346000575b60358060166000396000f30060606040525b60005600a165627a7a72305820ccf9337430b4c4f7d6ad41efb10a94411a2af6a9f173ef52daeadd31f4bf11890029
+Metadata:
+{"compiler":{"version":"0.4.8+commit.60cc1668.mod.Darwin.appleclang"},"language":"Solidity","output":{"abi":[],"devdoc":{"methods":{}},"userdoc":{"methods":{}}},"settings":{"compilationTarget":{"contract.sol":"C"},"libraries":{},"optimizer":{"enabled":false,"runs":200},"remappings":[]},"sources":{"contract.sol":{"keccak256":"0xbe86d3681a198587296ad6d4a834606197e1a8f8944922c501631b04e21eeba2","urls":["bzzr://af16957d3d86013309d64d3cc572d007b1d8b08a821f2ff366840deb54a78524"]}},"version":1}

test/scripts/test_bytecodecompare_prepare_report.py

@@ -41,6 +41,9 @@ STACK_TOO_DEEP_CLI_OUTPUT = load_fixture('stack_too_deep_cli_output.txt')
 CODE_GENERATION_ERROR_JSON_OUTPUT = load_fixture('code_generation_error_json_output.json')
 CODE_GENERATION_ERROR_CLI_OUTPUT = load_fixture('code_generation_error_cli_output.txt')
 
+SOLC_0_4_0_CLI_OUTPUT = load_fixture('solc_0.4.0_cli_output.txt')
+SOLC_0_4_8_CLI_OUTPUT = load_fixture('solc_0.4.8_cli_output.txt')
+
 
 class PrepareReportTestBase(unittest.TestCase):
     def setUp(self):
@@ -334,6 +337,34 @@ class TestParseStandardJSONOutput(PrepareReportTestBase):
 
 
 class TestParseCLIOutput(PrepareReportTestBase):
+    def test_parse_standard_json_output_should_report_missing_if_value_is_just_whitespace(self):
+        compiler_output = dedent("""\
+            {
+                "contracts": {
+                    "contract.sol": {
+                        "A": {
+                            "evm": {"bytecode": {"object": ""}},
+                            "metadata": ""
+                        },
+                        "B": {
+                            "evm": {"bytecode": {"object": "  "}},
+                            "metadata": "  "
+                        }
+                    }
+                }
+            }
+        """)
+
+        expected_report = FileReport(
+            file_name=Path('contract.sol'),
+            contract_reports=[
+                ContractReport(contract_name='A', file_name=Path('contract.sol'), bytecode=None, metadata=None),
+                ContractReport(contract_name='B', file_name=Path('contract.sol'), bytecode=None, metadata=None),
+            ]
+        )
+
+        self.assertEqual(parse_standard_json_output(Path('contract.sol'), compiler_output), expected_report)
+
     def test_parse_cli_output(self):
         expected_report = FileReport(
             file_name=Path('syntaxTests/scoping/library_inherited2.sol'),
@@ -431,3 +462,107 @@ class TestParseCLIOutput(PrepareReportTestBase):
         expected_report = FileReport(file_name=Path('file.sol'), contract_reports=None)
 
         self.assertEqual(parse_cli_output(Path('file.sol'), CODE_GENERATION_ERROR_CLI_OUTPUT), expected_report)
+
+    def test_parse_cli_output_should_handle_output_from_solc_0_4_0(self):
+        expected_report = FileReport(
+            file_name=Path('contract.sol'),
+            contract_reports=[
+                ContractReport(
+                    contract_name='C',
+                    file_name=None,
+                    bytecode='6060604052600c8060106000396000f360606040526008565b600256',
+                    metadata=None,
+                )
+            ]
+        )
+
+        self.assertEqual(parse_cli_output(Path('contract.sol'), SOLC_0_4_0_CLI_OUTPUT), expected_report)
+
+    def test_parse_cli_output_should_handle_output_from_solc_0_4_8(self):
+        expected_report = FileReport(
+            file_name=Path('contract.sol'),
+            contract_reports=[
+                # pragma pylint: disable=line-too-long
+                ContractReport(
+                    contract_name='C',
+                    file_name=None,
+                    bytecode='6060604052346000575b60358060166000396000f30060606040525b60005600a165627a7a72305820ccf9337430b4c4f7d6ad41efb10a94411a2af6a9f173ef52daeadd31f4bf11890029',
+                    metadata='{"compiler":{"version":"0.4.8+commit.60cc1668.mod.Darwin.appleclang"},"language":"Solidity","output":{"abi":[],"devdoc":{"methods":{}},"userdoc":{"methods":{}}},"settings":{"compilationTarget":{"contract.sol":"C"},"libraries":{},"optimizer":{"enabled":false,"runs":200},"remappings":[]},"sources":{"contract.sol":{"keccak256":"0xbe86d3681a198587296ad6d4a834606197e1a8f8944922c501631b04e21eeba2","urls":["bzzr://af16957d3d86013309d64d3cc572d007b1d8b08a821f2ff366840deb54a78524"]}},"version":1}',
+                )
+                # pragma pylint: enable=line-too-long
+            ]
+        )
+
+        self.assertEqual(parse_cli_output(Path('contract.sol'), SOLC_0_4_8_CLI_OUTPUT), expected_report)
+
+    def test_parse_cli_output_should_handle_leading_and_trailing_spaces(self):
+        compiler_output = (
+            ' =======  contract.sol : C  ======= \n'
+            ' Binary: \n'
+            ' 60806040523480156 \n'
+            ' Metadata: \n'
+            ' {  } \n'
+        )
+
+        expected_report = FileReport(
+            file_name=Path('contract.sol'),
+            contract_reports=[ContractReport(contract_name='C', file_name=Path('contract.sol'), bytecode='60806040523480156', metadata='{  }')]
+        )
+
+        self.assertEqual(parse_cli_output(Path('contract.sol'), compiler_output), expected_report)
+
+    def test_parse_cli_output_should_handle_empty_bytecode_and_metadata_lines(self):
+        compiler_output = dedent("""\
+            ======= contract.sol:C =======
+            Binary:
+            60806040523480156
+            Metadata:
+
+
+            ======= contract.sol:D =======
+            Binary:
+
+            Metadata:
+            {}
+
+
+            ======= contract.sol:E =======
+            Binary:
+
+            Metadata:
+
+
+        """)
+
+        expected_report = FileReport(
+            file_name=Path('contract.sol'),
+            contract_reports=[
+                ContractReport(contract_name='C', file_name=Path('contract.sol'), bytecode='60806040523480156', metadata=None),
+                ContractReport(contract_name='D', file_name=Path('contract.sol'), bytecode=None, metadata='{}'),
+                ContractReport(contract_name='E', file_name=Path('contract.sol'), bytecode=None, metadata=None),
+            ]
+        )
+
+        self.assertEqual(parse_cli_output(Path('contract.sol'), compiler_output), expected_report)
+
+    def test_parse_cli_output_should_handle_link_references_in_bytecode(self):
+        compiler_output = dedent("""\
+            ======= contract.sol:C =======
+            Binary:
+            73123456789012345678901234567890123456789073__$fb58009a6b1ecea3b9d99bedd645df4ec3$__5050
+            ======= contract.sol:D =======
+            Binary:
+            __$fb58009a6b1ecea3b9d99bedd645df4ec3$__
+        """)
+
+        # pragma pylint: disable=line-too-long
+        expected_report = FileReport(
+            file_name=Path('contract.sol'),
+            contract_reports=[
+                ContractReport(contract_name='C', file_name=Path('contract.sol'), bytecode='73123456789012345678901234567890123456789073__$fb58009a6b1ecea3b9d99bedd645df4ec3$__5050', metadata=None),
+                ContractReport(contract_name='D', file_name=Path('contract.sol'), bytecode='__$fb58009a6b1ecea3b9d99bedd645df4ec3$__', metadata=None),
+            ]
+        )
+        # pragma pylint: enable=line-too-long
+
+        self.assertEqual(parse_cli_output(Path('contract.sol'), compiler_output), expected_report)