Merge pull request #130 from ethereum/storagedecoder

Extract state info from AST
8 years ago · a3b953ecd4
parent 7df62fa39a 15c33e7518
commit a3b953ecd4
6 changed files with 490 additions and 0 deletions
--- a/src/index.js
+++ b/src/index.js
@ -7,6 +7,8 @@ var CodeManager = require('./code/codeManager')
 var disassembler = require('./code/disassembler')
 var SourceMappingDecoder = require('./util/sourceMappingDecoder')
 var AstWalker = require('./util/astWalker')
+var decodeInfo = require('./solidity/decodeInfo')
+var astHelper = require('./solidity/astHelper')

 if (typeof (module) !== 'undefined' && typeof (module.exports) !== 'undefined') {
  module.exports = modules()
@ -33,6 +35,10 @@ function modules () {
    util: {
      SourceMappingDecoder: SourceMappingDecoder,
      AstWalker: AstWalker
+    },
+    solidity: {
+      decodeInfo: decodeInfo,
+      astHelper: astHelper
    }
  }
 }
--- a/src/solidity/astHelper.js
+++ b/src/solidity/astHelper.js
@ -0,0 +1,64 @@
+'use strict'
+var AstWalker = require('../util/astWalker')
+
+/**
+  * return all contract definitions of the given @astList
+  *
+  * @param {Object} sourcesList - sources list (containing root AST node)
+  * @return {Object} - returns a mapping from AST node ids to AST nodes for the contracts
+  */
+function extractContractDefinitions (sourcesList) {
+  var ret = {
+    contractsById: {},
+    contractsByName: {}
+  }
+  var walker = new AstWalker()
+  walker.walkAstList(sourcesList, { 'ContractDefinition': function (node) {
+    ret.contractsById[node.id] = node
+    ret.contractsByName[node.attributes.name] = node
+    return false
+  }})
+  return ret
+}
+
+/**
+  * returns the linearized base contracts of the contract @arg id
+  *
+  * @param {Int} id - contract id to resolve
+  * @param {Map} contracts  - all contracts defined in the current context
+  * @return {Array} - array of base contracts in derived to base order as AST nodes.
+  */
+function getLinearizedBaseContracts (id, contractsById) {
+  return contractsById[id].attributes.linearizedBaseContracts.map(function (id) { return contractsById[id] })
+}
+
+/**
+  * return state var and type definition of the given contract
+  *
+  * @param {String} contractName - contract for which state var should be resolved
+  * @param {Object} sourcesList - sources list (containing root AST node)
+  * @return {Array} - return an array of AST node of all state variables (including inherited) (this will include all enum/struct declarations)
+  */
+function extractStateVariables (contractName, sourcesList) {
+  var contracts = extractContractDefinitions(sourcesList)
+  var node = contracts.contractsByName[contractName]
+  if (node) {
+    var stateVar = []
+    var baseContracts = getLinearizedBaseContracts(node.id, contracts.contractsById)
+    baseContracts.reverse()
+    for (var k in baseContracts) {
+      var ctr = baseContracts[k]
+      for (var i in ctr.children) {
+        stateVar.push(ctr.children[i])
+      }
+    }
+    return stateVar
+  }
+  return null
+}
+
+module.exports = {
+  extractStateVariables: extractStateVariables,
+  extractContractDefinitions: extractContractDefinitions,
+  getLinearizedBaseContracts: getLinearizedBaseContracts
+}
--- a/src/solidity/decodeInfo.js
+++ b/src/solidity/decodeInfo.js
@ -0,0 +1,302 @@
+'use strict'
+/**
+  * Uint decode the given @arg type
+  *
+  * @param {String} type - type given by the AST (e.g uint256, uint32)
+  * @return {Object} returns decoded info about the current type: { needsFreeStorageSlot, storageBytes, typeName}
+  */
+function Uint (type) {
+  type === 'uint' ? 'uint256' : type
+  return {
+    needsFreeStorageSlot: false,
+    storageBytes: parseInt(type.replace('uint', '')) / 8,
+    typeName: type
+  }
+}
+
+/**
+  * Int decode the given @arg type
+  *
+  * @param {String} type - type given by the AST (e.g int256, int32)
+  * @return {Object} returns decoded info about the current type: { needsFreeStorageSlot, storageBytes, typeName}
+  */
+function Int (type) {
+  type === 'int' ? 'int256' : type
+  return {
+    needsFreeStorageSlot: false,
+    storageBytes: parseInt(type.replace('int', '')) / 8,
+    typeName: type
+  }
+}
+
+/**
+  * Address decode the given @arg type
+  *
+  * @param {String} type - type given by the AST (e.g address)
+  * @return {Object} returns decoded info about the current type: { needsFreeStorageSlot, storageBytes, typeName}
+  */
+function Address (type) {
+  return {
+    needsFreeStorageSlot: false,
+    storageBytes: 20,
+    typeName: 'address'
+  }
+}
+
+/**
+  * Bool decode the given @arg type
+  *
+  * @param {String} type - type given by the AST (e.g bool)
+  * @return {Object} returns decoded info about the current type: { needsFreeStorageSlot, storageBytes, typeName}
+  */
+function Bool (type) {
+  return {
+    needsFreeStorageSlot: false,
+    storageBytes: 1,
+    typeName: 'bool'
+  }
+}
+
+/**
+  * DynamicByteArray decode the given @arg type
+  *
+  * @param {String} type - type given by the AST (e.g bytes storage ref)
+  * @return {Object} returns decoded info about the current type: { needsFreeStorageSlot, storageBytes, typeName}
+  */
+function DynamicByteArray (type) {
+  return {
+    needsFreeStorageSlot: true,
+    storageBytes: 32,
+    typeName: 'bytes'
+  }
+}
+
+/**
+  * FixedByteArray decode the given @arg type
+  *
+  * @param {String} type - type given by the AST (e.g bytes16)
+  * @return {Object} returns decoded info about the current type: { needsFreeStorageSlot, storageBytes, typeName}
+  */
+function FixedByteArray (type) {
+  return {
+    needsFreeStorageSlot: false,
+    storageBytes: parseInt(type.replace('bytes', '')),
+    typeName: type
+  }
+}
+
+/**
+  * StringType decode the given @arg type
+  *
+  * @param {String} type - type given by the AST (e.g string storage ref)
+  * @return {Object} returns decoded info about the current type: { needsFreeStorageSlot, storageBytes, typeName}
+  */
+function StringType (type) {
+  return {
+    needsFreeStorageSlot: true,
+    storageBytes: 32,
+    typeName: 'string'
+  }
+}
+
+/**
+  * ArrayType decode the given @arg type
+  *
+  * @param {String} type - type given by the AST (e.g int256[] storage ref, int256[] storage ref[] storage ref)
+  * @return {Object} returns decoded info about the current type: { needsFreeStorageSlot, storageBytes, typeName, arraySize, subArray}
+  */
+function ArrayType (type, stateDefinitions) {
+  var arraySize
+  var storageBytes
+
+  var match = type.match(/(.*)\[(.*?)\]( storage ref| storage pointer| memory| calldata)?$/)
+  if (!match || match.length < 3) {
+    console.log('unable to parse type ' + type)
+    return null
+  }
+
+  arraySize = match[2] === '' ? 'dynamic' : parseInt(match[2])
+
+  var underlyingType = decode(match[1], stateDefinitions)
+  if (underlyingType === null) {
+    console.log('unable to parse type ' + type)
+    return null
+  }
+
+  if (arraySize === 'dynamic') {
+    storageBytes = 32
+  } else {
+    storageBytes = underlyingType.storageBytes
+    if (storageBytes > 32) {
+      storageBytes = 32 * arraySize * Math.ceil(storageBytes / 32)
+    } else {
+      storageBytes = 32 * (arraySize / Math.floor(32 / storageBytes))
+    }
+  }
+
+  return {
+    needsFreeStorageSlot: true,
+    storageBytes: storageBytes,
+    typeName: type,
+    arraySize: arraySize,
+    underlyingType: underlyingType
+  }
+}
+
+/**
+  * Enum decode the given @arg type
+  *
+  * @param {String} type - type given by the AST (e.g enum enumDef)
+  * @return {Object} returns decoded info about the current type: { needsFreeStorageSlot, storageBytes, typeName, enum}
+  */
+function Enum (type, stateDefinitions) {
+  var enumDef = getEnum(type, stateDefinitions)
+  if (enumDef === null) {
+    console.log('unable to retrieve decode info of ' + type)
+    return null
+  }
+  var length = enumDef.children.length
+  var storageBytes = 0
+  while (length > 1) {
+    length = length / 256
+    storageBytes++
+  }
+  return {
+    needsFreeStorageSlot: false,
+    storageBytes: storageBytes,
+    typeName: type,
+    enum: enumDef
+  }
+}
+
+/**
+  * Struct decode the given @arg type
+  *
+  * @param {String} type - type given by the AST (e.g struct structDef storage ref)
+  * @return {Object} returns decoded info about the current type: { needsFreeStorageSlot, storageBytes, typeName, members}
+  */
+function Struct (type, stateDefinitions) {
+  var match = type.match(/struct (.*?)( storage ref| storage pointer| memory| calldata)?$/)
+  if (!match) {
+    return null
+  }
+  var memberDetails = getStructMembers(match[1], stateDefinitions) // type is used to extract the ast struct definition
+  if (!memberDetails) return null
+  return {
+    needsFreeStorageSlot: true,
+    storageBytes: memberDetails.storageBytes,
+    typeName: type,
+    members: memberDetails.members
+  }
+}
+
+/**
+  * retrieve enum declaration of the given @arg type
+  *
+  * @param {String} type - type given by the AST (e.g enum enumDef)
+  * @param {Object} stateDefinitions  - all state declarations given by the AST (including struct and enum type declaration)
+  * @return {Array} - containing all value declaration of the current enum type
+  */
+function getEnum (type, stateDefinitions) {
+  for (var k in stateDefinitions) {
+    var dec = stateDefinitions[k]
+    if (type === 'enum ' + dec.attributes.name) {
+      return dec
+    }
+  }
+  return null
+}
+
+/**
+  * retrieve memebers declared in the given @arg tye
+  *
+  * @param {String} typeName - name of the struct type (e.g struct <name>)
+  * @param {Object} stateDefinitions  - all state definition given by the AST (including struct and enum type declaration)
+  * @return {Array} containing all members of the current struct type
+  */
+function getStructMembers (typeName, stateDefinitions) {
+  var members = []
+  var storageBytes = 0
+  for (var k in stateDefinitions) {
+    var dec = stateDefinitions[k]
+    if (dec.name === 'StructDefinition' && typeName === dec.attributes.name) {
+      for (var i in dec.children) {
+        var member = dec.children[i]
+        var decoded = decode(member.attributes.type, stateDefinitions)
+        if (!decoded) {
+          console.log('unable to retrieve decode info of ' + member.attributes.type)
+          return null
+        }
+        members.push(decoded)
+        if (decoded.needsFreeStorageSlot) {
+          storageBytes = Math.ceil(storageBytes / 32) * 32
+        }
+        storageBytes += decoded.storageBytes
+      }
+      break
+    }
+  }
+  return {
+    members: members,
+    storageBytes: storageBytes
+  }
+}
+
+/**
+  * parse the full type
+  *
+  * @param {String} fullType - type given by the AST (ex: uint[2] storage ref[2])
+  * @return {String} returns the token type (used to instanciate the right decoder) (uint[2] storage ref[2] will return 'array', uint256 will return uintX)
+  */
+function typeClass (fullType) {
+  if (fullType.indexOf(']') !== -1) {
+    return 'array'
+  }
+  if (fullType.indexOf(' ') !== -1) {
+    fullType = fullType.split(' ')[0]
+  }
+  var char = fullType.indexOf('bytes') === 0 ? 'X' : ''
+  return fullType.replace(/[0-9]+/g, char)
+}
+
+/**
+  * parse the type and return an object representing the type
+  *
+  * @param {Object} type - type name given by the ast node
+  * @param {Object} stateDefinitions - all state stateDefinitions given by the AST (including struct and enum type declaration)
+  * @return {Object} - return the corresponding decoder or null on error
+  */
+function decode (type, stateDefinitions) {
+  var decodeInfos = {
+    'address': Address,
+    'array': ArrayType,
+    'bool': Bool,
+    'bytes': DynamicByteArray,
+    'bytesX': FixedByteArray,
+    'enum': Enum,
+    'string': StringType,
+    'struct': Struct,
+    'int': Int,
+    'uint': Uint
+  }
+  var currentType = typeClass(type)
+  if (currentType === null) {
+    console.log('unable to retrieve decode info of ' + type)
+    return null
+  }
+  return decodeInfos[currentType](type, stateDefinitions)
+}
+
+module.exports = {
+  decode: decode,
+  Uint: Uint,
+  Address: Address,
+  Bool: Bool,
+  DynamicByteArray: DynamicByteArray,
+  FixedByteArray: FixedByteArray,
+  Int: Int,
+  StringType: StringType,
+  ArrayType: ArrayType,
+  Enum: Enum,
+  Struct: Struct
+}
--- a/src/util/astWalker.js
+++ b/src/util/astWalker.js
@ -30,6 +30,19 @@ AstWalker.prototype.walk = function (ast, callback) {
  }
 }

+/**
+ +  * walk the given @astList
+ +  *
+ +  * @param {Object} sourcesList - sources list (containing root AST node)
+ +  * @param {Function} - callback used by AstWalker to compute response
+ +  */
+AstWalker.prototype.walkAstList = function (sourcesList, callback) {
+  var walker = new AstWalker()
+  for (var k in sourcesList) {
+    walker.walk(sourcesList[k].AST, callback)
+  }
+}
+
 function manageCallBack (node, callback) {
  if (node.name in callback) {
    return callback[node.name](node)
--- a/test/decodeInfo.js
+++ b/test/decodeInfo.js
@ -0,0 +1,104 @@
+'use strict'
+var tape = require('tape')
+var compiler = require('solc')
+var index = require('../src/index')
+
+tape('solidity', function (t) {
+  t.test('astHelper, decodeInfo', function (st) {
+    var output = compiler.compile(contracts, 0)
+
+    var stateDec = index.solidity.astHelper.extractStateVariables('contractUint', output.sources)
+    var decodeInfo = index.solidity.decodeInfo.decode(stateDec[0].attributes.type, stateDec)
+    checkDecodeInfo(st, decodeInfo, false, 1, 'uint8')
+    decodeInfo = index.solidity.decodeInfo.decode(stateDec[2].attributes.type, stateDec)
+    checkDecodeInfo(st, decodeInfo, false, 32, 'uint256')
+    decodeInfo = index.solidity.decodeInfo.decode(stateDec[3].attributes.type, stateDec)
+    checkDecodeInfo(st, decodeInfo, false, 32, 'uint256')
+    decodeInfo = index.solidity.decodeInfo.decode(stateDec[4].attributes.type, stateDec)
+    checkDecodeInfo(st, decodeInfo, false, 16, 'bytes16')
+
+    stateDec = index.solidity.astHelper.extractStateVariables('contractStructAndArray', output.sources)
+    decodeInfo = index.solidity.decodeInfo.decode(stateDec[1].attributes.type, stateDec)
+    checkDecodeInfo(st, decodeInfo, true, 64, 'struct structDef storage ref')
+    decodeInfo = index.solidity.decodeInfo.decode(stateDec[2].attributes.type, stateDec)
+    checkDecodeInfo(st, decodeInfo, true, 192, 'struct structDef storage ref[3] storage ref')
+    decodeInfo = index.solidity.decodeInfo.decode(stateDec[3].attributes.type, stateDec)
+    checkDecodeInfo(st, decodeInfo, true, 64, 'bytes12[4] storage ref')
+
+    stateDec = index.solidity.astHelper.extractStateVariables('contractArray', output.sources)
+    decodeInfo = index.solidity.decodeInfo.decode(stateDec[0].attributes.type, stateDec)
+    checkDecodeInfo(st, decodeInfo, true, 4 * 5, 'uint32[5] storage ref')
+    decodeInfo = index.solidity.decodeInfo.decode(stateDec[1].attributes.type, stateDec)
+    checkDecodeInfo(st, decodeInfo, true, 32, 'int8[] storage ref')
+    decodeInfo = index.solidity.decodeInfo.decode(stateDec[2].attributes.type, stateDec)
+    checkDecodeInfo(st, decodeInfo, true, 4 * 32, 'int16[] storage ref[3] storage ref[] storage ref[4] storage ref')
+
+    stateDec = index.solidity.astHelper.extractStateVariables('contractEnum', output.sources)
+    decodeInfo = index.solidity.decodeInfo.decode(stateDec[1].attributes.type, stateDec)
+    checkDecodeInfo(st, decodeInfo, false, 2, 'enum enumDef')
+
+    stateDec = index.solidity.astHelper.extractStateVariables('contractSmallVariable', output.sources)
+    decodeInfo = index.solidity.decodeInfo.decode(stateDec[0].attributes.type, stateDec)
+    checkDecodeInfo(st, decodeInfo, false, 1, 'int8')
+    decodeInfo = index.solidity.decodeInfo.decode(stateDec[1].attributes.type, stateDec)
+    checkDecodeInfo(st, decodeInfo, false, 1, 'uint8')
+    decodeInfo = index.solidity.decodeInfo.decode(stateDec[2].attributes.type, stateDec)
+    checkDecodeInfo(st, decodeInfo, false, 2, 'uint16')
+    decodeInfo = index.solidity.decodeInfo.decode(stateDec[3].attributes.type, stateDec)
+    checkDecodeInfo(st, decodeInfo, false, 4, 'int32')
+    decodeInfo = index.solidity.decodeInfo.decode(stateDec[4].attributes.type, stateDec)
+    checkDecodeInfo(st, decodeInfo, false, 32, 'uint256')
+    decodeInfo = index.solidity.decodeInfo.decode(stateDec[5].attributes.type, stateDec)
+    checkDecodeInfo(st, decodeInfo, false, 2, 'int16')
+
+    st.end()
+  })
+})
+
+function checkDecodeInfo (st, decodeInfo, freeSlot, storageBytes, typeName, name) {
+  st.equal(decodeInfo.needsFreeStorageSlot, freeSlot)
+  st.equal(decodeInfo.storageBytes, storageBytes)
+  st.equal(decodeInfo.typeName, typeName)
+}
+
+var contracts = `
+    contract baseContract {
+        uint8 u;
+    }
+    
+    contract contractUint is baseContract {
+        uint256 ui;
+        uint ui1;
+        bytes16 b;
+    }    
+    
+    contract contractStructAndArray {       
+        struct structDef {
+            uint8 ui;
+            string str;
+        }
+        structDef structDec;
+        structDef[3] array;
+        bytes12[4] bytesArray;
+    }
+    
+    contract contractArray {
+        uint32[5] i32st;
+        int8[] i8dyn;
+        int16[][3][][4] i16dyn;    
+    }  
+    
+    contract contractEnum {
+        enum enumDef {item0,item1,item2,item3,item4,item5,item6,item7,item8,item9,item10,item11,item12,item13,item14,item15,item16,item17,item18,item19,item20,item21,item22,item23,item24,item25,item26,item27,item28,item29,item30,item31,item32,item33,item34,item35,item36,item37,item38,item39,item40,item41,item42,item43,item44,item45,item46,item47,item48,item49,item50,item51,item52,item53,item54,item55,item56,item57,item58,item59,item60,item61,item62,item63,item64,item65,item66,item67,item68,item69,item70,item71,item72,item73,item74,item75,item76,item77,item78,item79,item80,item81,item82,item83,item84,item85,item86,item87,item88,item89,item90,item91,item92,item93,item94,item95,item96,item97,item98,item99,item100,item101,item102,item103,item104,item105,item106,item107,item108,item109,item110,item111,item112,item113,item114,item115,item116,item117,item118,item119,item120,item121,item122,item123,item124,item125,item126,item127,item128,item129,item130,item131,item132,item133,item134,item135,item136,item137,item138,item139,item140,item141,item142,item143,item144,item145,item146,item147,item148,item149,item150,item151,item152,item153,item154,item155,item156,item157,item158,item159,item160,item161,item162,item163,item164,item165,item166,item167,item168,item169,item170,item171,item172,item173,item174,item175,item176,item177,item178,item179,item180,item181,item182,item183,item184,item185,item186,item187,item188,item189,item190,item191,item192,item193,item194,item195,item196,item197,item198,item199,item200,item201,item202,item203,item204,item205,item206,item207,item208,item209,item210,item211,item212,item213,item214,item215,item216,item217,item218,item219,item220,item221,item222,item223,item224,item225,item226,item227,item228,item229,item230,item231,item232,item233,item234,item235,item236,item237,item238,item239,item240,item241,item242,item243,item244,item245,item246,item247,item248,item249,item250,item251,item252,item253,item254,item255,item256,item257,item258,item259,item260,item261,item262,item263,item264,item265,item266,item267,item268,item269,item270,item271,item272,item273,item274,item275,item276,item277,item278,item279,item280,item281,item282,item283,item284,item285,item286,item287,item288,item289,item290,item291,item292,item293,item294,item295,item296,item297,item298,item299,item100000000}
+        enumDef enum1;
+    }
+    
+    contract contractSmallVariable {
+        int8 i8;
+        uint8 iu8;
+        uint16 iu18;
+        int32 i32;
+        uint ui32;
+        int16 i16;
+    }  
+`
--- a/test/tests.js
+++ b/test/tests.js
@ -7,3 +7,4 @@ require('./astwalker.js')
 require('./disassembler.js')
 require('./eventManager.js')
 require('./sourceMappingDecoder.js')
+require('./decodeInfo.js')