blob: a8e654d8e31f9c151885887a597b530dd1241be8 [file] [log] [blame] [raw]
// Copyright (c) 2022, Compiler Explorer Authors
// All rights reserved.
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are met:
//
// * Redistributions of source code must retain the above copyright notice,
// this list of conditions and the following disclaimer.
// * Redistributions in binary form must reproduce the above copyright
// notice, this list of conditions and the following disclaimer in the
// documentation and/or other materials provided with the distribution.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
// AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
// ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
// LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
// CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
// SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
// INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
// CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
// ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
// POSSIBILITY OF SUCH DAMAGE.
import * as fs from 'fs';
import path from 'path';
import Semver from 'semver';
import {BaseCompiler} from '../base-compiler';
import {asSafeVer} from '../utils';
import {ClangParser} from './argument-parsers';
export class SolidityCompiler extends BaseCompiler {
static get key() {
return 'solidity';
}
getSharedLibraryPathsAsArguments() {
return [];
}
getArgumentParser() {
return ClangParser;
}
optionsForFilter(filters, outputFilename, userOptions) {
return [
// We use --combined-json instead of `--asm-json` to have compacted json
'--combined-json',
Semver.lt(asSafeVer(this.compiler.semver), '0.8.0', true)
? 'asm,ast'
: 'asm,ast,generated-sources,generated-sources-runtime',
'-o',
'contracts',
];
}
isCfgCompiler(/*compilerVersion*/) {
return false;
}
getOutputFilename(dirPath) {
return path.join(dirPath, 'contracts/combined.json');
}
processAsm(result) {
// Handle "error" documents.
if (!result.asm.includes('\n') && result.asm[0] === '<') {
return {asm: [{text: result.asm}]};
}
// solc gives us a character range for each asm instruction,
// so open the input file and figure out what line each
// character is on
const inputFile = fs.readFileSync(result.inputFilename);
let currentLine = 1;
const charToLine = inputFile.map(c => {
const line = currentLine;
if (c === '\n'.codePointAt(0)) {
++currentLine;
}
return line;
});
const hasOldJSONLayout = Semver.lt(asSafeVer(this.compiler.semver), '0.8.0', true);
const hasGeneratedSources = Semver.gte(asSafeVer(this.compiler.semver), '0.8.0', true);
const asm = JSON.parse(result.asm);
return {
asm: Object.entries(asm.contracts)
.sort(([_name1, data1], [_name2, data2]) => data1.asm['.code'][0].begin - data2.asm['.code'][0].begin)
.map(([name, data]) => {
// name is in the format of file:contract
// e.g. MyFile.sol:MyContract
const [sourceName, contractName] = name.split(':');
// to make the asm more readable, we rename the
// tags (jumpdests) to show what function they're
// part of. here we parse the AST so we know what
// range of characters belongs to each function.
let contractFunctions;
// the layout of this JSON has changed between versions...
if (hasOldJSONLayout) {
contractFunctions = asm.sources[sourceName].AST.children
.find(node => {
return node.name === 'ContractDefinition' && node.attributes.name === contractName;
})
.children.filter(node => {
return node.name === 'FunctionDefinition';
})
.map(node => {
const [begin, length] = node.src.split(':').map(x => parseInt(x));
let name = node.attributes.isConstructor ? 'constructor' : node.attributes.name;
// encode the args into the name so we can
// differentiate between overloads
if (node.children[0].children.length > 0) {
name +=
'_' +
node.children[0].children
.map(paramNode => {
return paramNode.attributes.type;
})
.join('_');
}
return {
name: name,
begin: begin,
end: begin + length,
tagCount: 0,
};
});
} else {
contractFunctions = asm.sources[sourceName].AST.nodes
.find(node => {
return node.nodeType === 'ContractDefinition' && node.name === contractName;
})
.nodes.filter(node => {
return node.nodeType === 'FunctionDefinition';
})
.map(node => {
const [begin, length] = node.src.split(':').map(x => parseInt(x));
let name = node.kind === 'constructor' ? 'constructor' : node.name;
// encode the args into the name so we can
// differentiate between overloads
if (node.parameters.parameters.length > 0) {
name +=
'_' +
node.parameters.parameters
.map(paramNode => {
return paramNode.typeName.name;
})
.join('_');
}
return {
name: name,
begin: begin,
end: begin + length,
tagCount: 0,
};
});
}
// solc generates some code, for things like detecting
// and reverting if a multiplication results in
// integer overflow, etc.
const processGeneratedSources = generatedSourcesData => {
let generatedSources = {};
for (const generatedSource of generatedSourcesData) {
generatedSources[generatedSource.id] = generatedSource.ast.statements.map(statement => {
const [begin, length] = statement.src.split(':').map(x => parseInt(x));
return {
name: statement.name,
begin: begin,
end: begin + length,
tagCount: 0,
};
});
}
return generatedSources;
};
// (0.8.x onwards only!)
// there are two sets of generated sources, one for the code which deploys
// the contract (i.e. the constructor) 'generated-sources', and the code
// which is deployed and stored on-chain 'generated-sources-runtime'
const generatedSources = hasGeneratedSources
? processGeneratedSources(data['generated-sources'])
: {};
const generatedSourcesRuntime = hasGeneratedSources
? processGeneratedSources(data['generated-sources-runtime'])
: {};
const processOpcodes = (opcodes, indent, generatedSources) => {
// first iterate the opcodes to find all the tags,
// and assign human-readable names to as many of
// them as we can
let tagNames = {};
const processPossibleTagOpcode = (opcode, funcList) => {
if (opcode.name === 'tag') {
const func = funcList.find(func => {
return opcode.begin >= func.begin && opcode.end <= func.end;
});
if (func !== undefined) {
// a function can have multiple tags, so append
// a number to each
const tagName = `${func.name}_${func.tagCount}`;
++func.tagCount;
tagNames[opcode.value] = tagName;
opcode.value = tagName;
}
}
};
for (const opcode of opcodes) {
// source 0 is the .sol file the user is
// editing, everything else is generated
// sources (from version 0.8.x onwards).
// if source is undefined, then this is
// a compiler version which doesn't
// provide one (< 0.6.x), but we can
// infer 0 in this case.
if (opcode.source === 0 || opcode.source === undefined) {
opcode.line = charToLine[opcode.begin];
processPossibleTagOpcode(opcode, contractFunctions);
} else {
processPossibleTagOpcode(opcode, generatedSources[opcode.source]);
}
}
return opcodes.map(opcode => {
const name = `${opcode.name.startsWith('tag') ? indent : `${indent}\t`}${opcode.name}`;
let value = opcode.value || '';
if (opcode.name === 'PUSH [tag]') {
if (tagNames[value] !== undefined) {
value = tagNames[value];
}
}
return {
text: `${name} ${value}`,
source: {line: opcode.line, file: null},
};
});
};
return [
{text: `// ${contractName}`},
// .code section is the code only run when deploying - the constructor
{text: '.code'},
processOpcodes(data.asm['.code'], '', generatedSources),
{text: ''},
// .data section is deployed bytecode - everything else
{text: '.data'},
Object.entries(data.asm['.data']).map(([id, {'.code': code}]) => [
{text: `\t${id}:`},
processOpcodes(code, '\t', generatedSourcesRuntime),
]),
];
})
.flat(Infinity),
};
}
}