Merge pull request #29 from Jon-Becker/feat/decompile

✨ feat: decompile

.github/workflows/rust_build.yml

@@ -12,12 +12,13 @@ jobs:
 
     steps:
     - uses: actions/checkout@v3
-    - name: Build
+    - name: Compile
       working-directory: ./heimdall
       run: cargo build --verbose
-    - name: Run tests
+
+    - name: Run Tests
       working-directory: ./heimdall
       run: |
-        cargo test --package heimdall
-        cargo test --package heimdall-config
-        cargo test --package heimdall-common
+        cargo test --package heimdall -- test_ --nocapture
+        cargo test --package heimdall-config -- test_ --nocapture
+        cargo test --package heimdall-common -- test_ --nocapture

.gitignore

@@ -13,4 +13,7 @@ output
 output/*
 */output
 *.evm
-*.asm
+*.asm
+
+heimdall/scripts/update
+heimdall/scripts/git

common/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "heimdall-common"
-version = "0.1.7"
+version = "0.2.0"
 edition = "2021"
 license = "MIT"
 readme = "README.md"
@@ -9,12 +9,12 @@ keywords = ["ethereum", "web3", "decompiler", "evm", "crypto"]
 
 [dependencies]
 colored = "2"
-regex = "1.1.5"
+fancy-regex = "0.10.0"
 lazy_static = "1.4.0"
 clap-verbosity-flag = "1.0.0"
 indicatif = "0.17.0"
 tokio = { version = "1", features = ["full"] }
 clap = { version = "3.1.18", features = ["derive"] }
 ethers = "1.0.0"
 reqwest = { version = "0.11.11", features = ["blocking"] }
-serde_json = "1.0"
+serde_json = "1.0"

common/src/consts.rs → common/src/constants.rs

@@ -1,4 +1,4 @@
-use regex::Regex;
+use fancy_regex::Regex;
 use lazy_static::lazy_static;
 
 lazy_static! {
@@ -16,4 +16,13 @@ lazy_static! {
 
     // The following regex is used to reduce null byte prefixes
     pub static ref REDUCE_HEX_REGEX: Regex = Regex::new(r"^0x(00)*").unwrap();
+
+    // The following regex is used as a search pattern for words
+    pub static ref WORD_REGEX: Regex = Regex::new(r"0x[0-9a-fA-F]{0,64}").unwrap();
+
+    // The following regex is used to find type castings
+    pub static ref TYPE_CAST_REGEX: Regex = Regex::new(r"(address\(|string\(|bool\(|bytes(\d*)\(|uint(\d*)\(|int(\d*)\()").unwrap();
+
+    // The following regex is used to find memory accesses
+    pub static ref MEMLEN_REGEX: Regex = Regex::new(r"memory\[memory\[[0-9x]*\]\]").unwrap();
 }

common/src/ether/evm/disassemble.rs

@@ -7,7 +7,7 @@ use ethers::{
     providers::{Middleware, Provider, Http},
 };
 use crate::{
-    consts::{ ADDRESS_REGEX, BYTECODE_REGEX },
+    constants::{ ADDRESS_REGEX, BYTECODE_REGEX },
     io::{ logging::*, file::* },
     ether::evm::{ opcodes::opcode }
 };
@@ -66,7 +66,7 @@ pub fn disassemble(args: DisassemblerArgs) -> String {
     }
 
     let contract_bytecode: String;
-    if ADDRESS_REGEX.is_match(&args.target) {
+    if ADDRESS_REGEX.is_match(&args.target).unwrap() {
 
         // push the address to the output directory
         if &output_dir != &args.output {
@@ -118,7 +118,7 @@ pub fn disassemble(args: DisassemblerArgs) -> String {
         });
 
     }
-    else if BYTECODE_REGEX.is_match(&args.target) {
+    else if BYTECODE_REGEX.is_match(&args.target).unwrap() {
         contract_bytecode = args.target.clone();
     }
     else {
@@ -131,7 +131,7 @@ pub fn disassemble(args: DisassemblerArgs) -> String {
         // We are disassembling a file, so we need to read the bytecode from the file.
         contract_bytecode = match fs::read_to_string(&args.target) {
             Ok(contents) => {                
-                if BYTECODE_REGEX.is_match(&contents) && contents.len() % 2 == 0 {
+                if BYTECODE_REGEX.is_match(&contents).unwrap() && contents.len() % 2 == 0 {
                     contents.replacen("0x", "", 1)
                 }
                 else {
@@ -178,11 +178,11 @@ pub fn disassemble(args: DisassemblerArgs) -> String {
         program_counter += 1;
     }
 
-    logger.success(&format!("disassembled {} bytes successfully.", program_counter).to_string());
+    logger.info(&format!("disassembled {} bytes successfully.", program_counter).to_string());
 
     write_file(&String::from(format!("{}/bytecode.evm", &output_dir)), &contract_bytecode);
     let file_path = write_file(&String::from(format!("{}/disassembled.asm", &output_dir)), &output);
-    logger.info(&format!("wrote disassembled bytecode to '{}' .", file_path).to_string());
+    logger.success(&format!("wrote disassembled bytecode to '{}' .", file_path).to_string());
 
     logger.debug(&format!("disassembly completed in {} ms.", now.elapsed().as_millis()).to_string());
 

common/src/ether/evm/memory.rs

@@ -18,6 +18,7 @@ impl Memory {
     }
 
     pub fn extend(&mut self, offset: u128, size: u128) {
+
         // calculate the new size of the memory
         let r = (offset + size) % 32;
         let new_mem_size: u128;
@@ -65,11 +66,7 @@ impl Memory {
     }
 
     // read a value from the memory at the given offset, with a fixed size
-    pub fn read(&self, mut offset: usize, size: usize) -> String {
-        // cap offset to 2**16 for optimization
-        if offset > 65536 {
-            offset = 65536;
-        }
+    pub fn read(&self, offset: usize, size: usize) -> String {
 
         // if the offset + size will be out of bounds, append null bytes until the size is met
         if offset + size > self.size() as usize {

common/src/ether/evm/stack.rs

@@ -90,6 +90,10 @@ impl Stack {
         values
     }
 
+    pub fn size(&self) -> usize {
+        self.stack.len()
+    }
+
     // Check if the stack is empty.
     pub fn is_empty(&self) -> bool {
         self.stack.is_empty()

common/src/ether/evm/types.rs

@@ -1,7 +1,7 @@
 use colored::Colorize;
 use ethers::abi::{ParamType, Token, AbiEncode};
 
-use crate::utils::strings::replace_last;
+use crate::{utils::strings::{replace_last, find_balanced_encapsulator}, constants::TYPE_CAST_REGEX};
 
 use super::vm::Instruction;
 
@@ -232,4 +232,21 @@ pub fn byte_size_to_type(byte_size: usize) -> (usize, Vec<String>) {
 
     // return list of potential type castings, sorted by likelihood descending
     (byte_size, potential_types)
+}
+
+pub fn find_cast(line: String) -> (usize, usize, Option<String>) {
+
+    // find the start of the cast
+    match TYPE_CAST_REGEX.find(&line).unwrap() {
+        Some(m) => {
+            let start = m.start();
+            let end = m.end() - 1;
+            let cast_type = line[start..].split("(").collect::<Vec<&str>>()[0].to_string();
+
+            // find where the cast ends
+            let (a, b, _) = find_balanced_encapsulator(line[end..].to_string(), ('(', ')'));
+            return (end+a, end+b, Some(cast_type))
+        },
+        None => return (0, 0, None),
+    }
 }

common/src/ether/evm/vm.rs

@@ -67,6 +67,7 @@ pub struct Instruction {
 }
 
 impl VM {
+
     // Creates a new VM instance
     pub fn new(
         bytecode: String,
@@ -122,6 +123,7 @@ impl VM {
 
     // Steps to the next PC and executes the instruction
     fn _step(&mut self) -> Instruction {
+
         // sanity check
         if self.bytecode.len() < (self.instruction * 2 + 2) as usize {
             self.exit(2, "0x");
@@ -1607,6 +1609,7 @@ impl VM {
                 };
             }
             _ => {
+
                 // we reached an INVALID opcode, consume all remaining gas
                 self.exit(4, "0x");
                 return Instruction {

common/src/ether/signatures.rs

@@ -28,7 +28,7 @@ pub struct ResolvedLog {
 pub fn resolve_function_signature(signature: &String) -> Option<Vec<ResolvedFunction>> {
 
     // get function possibilities from 4byte
-    let signatures = match get_json_from_url(format!("https://sig.eth.samczsun.com/api/v1/signatures?all=true&function=0x{}", &signature)) {
+    let signatures = match get_json_from_url(format!("https://sig.eth.samczsun.com/api/v1/signatures?all=true&function=0x{}", &signature), 3) {
         Some(signatures) => signatures,
         None => return None
     };
@@ -84,7 +84,7 @@ pub fn resolve_function_signature(signature: &String) -> Option<Vec<ResolvedFunc
 pub fn resolve_error_signature(signature: &String) -> Option<Vec<ResolvedError>> {
 
     // get function possibilities from 4byte
-    let signatures = match get_json_from_url(format!("https://sig.eth.samczsun.com/api/v1/signatures?all=true&function=0x{}", &signature)) {
+    let signatures = match get_json_from_url(format!("https://sig.eth.samczsun.com/api/v1/signatures?all=true&function=0x{}", &signature), 3) {
         Some(signatures) => signatures,
         None => return None
     };
@@ -140,7 +140,7 @@ pub fn resolve_error_signature(signature: &String) -> Option<Vec<ResolvedError>>
 pub fn resolve_event_signature(signature: &String) -> Option<Vec<ResolvedLog>> {
 
     // get function possibilities from 4byte
-    let signatures = match get_json_from_url(format!("https://sig.eth.samczsun.com/api/v1/signatures?all=true&event=0x{}", &signature)) {
+    let signatures = match get_json_from_url(format!("https://sig.eth.samczsun.com/api/v1/signatures?all=true&function=0x{}", &signature), 3) {
         Some(signatures) => signatures,
         None => return None
     };