feat(decompile): resolve external calls, simplify internal logic around external calls

.github/workflows/tests.yml

@@ -68,7 +68,7 @@ jobs:
       - uses: dtolnay/rust-toolchain@nightly
         with:
           components: rustfmt
-      - run: cargo fmt --check --all
+      - run: cargo +nightly fmt --check --all
 
   check:
     runs-on: ubuntu-latest

crates/cache/src/lib.rs

@@ -215,8 +215,7 @@ pub fn delete_cache(key: &str) -> Result<(), Error> {
 #[allow(deprecated)]
 pub fn read_cache<T>(key: &str) -> Result<Option<T>, Error>
 where
-    T: 'static + DeserializeOwned,
-{
+    T: 'static + DeserializeOwned, {
     let home = home_dir().ok_or(Error::Generic(
         "failed to get home directory. does your os support `std::env::home_dir()`?".to_string(),
     ))?;
@@ -239,8 +238,8 @@ where
         .map_err(|e| Error::Generic(format!("failed to deserialize cache object: {:?}", e)))?;
 
     // check if the cache has expired, if so, delete it and return None
-    if cache.expiry
-        < std::time::SystemTime::now()
+    if cache.expiry <
+        std::time::SystemTime::now()
             .duration_since(std::time::UNIX_EPOCH)
             .map_err(|e| Error::Generic(format!("failed to get current time: {:?}", e)))?
             .as_secs()
@@ -267,8 +266,7 @@ where
 #[allow(deprecated)]
 pub fn store_cache<T>(key: &str, value: T, expiry: Option<u64>) -> Result<(), Error>
 where
-    T: Serialize,
-{
+    T: Serialize, {
     let home = home_dir().ok_or(Error::Generic(
         "failed to get home directory. does your os support `std::env::home_dir()`?".to_string(),
     ))?;
@@ -280,8 +278,8 @@ where
         std::time::SystemTime::now()
             .duration_since(std::time::UNIX_EPOCH)
             .map_err(|e| Error::Generic(format!("failed to get current time: {:?}", e)))?
-            .as_secs()
-            + 60 * 60 * 24 * 90,
+            .as_secs() +
+            60 * 60 * 24 * 90,
     );
 
     let cache = Cache { value, expiry };
@@ -306,8 +304,7 @@ pub async fn with_cache<T, F, Fut>(key: &str, func: F) -> eyre::Result<T>
 where
     T: 'static + Serialize + DeserializeOwned + Send + Sync,
     F: FnOnce() -> Fut + Send,
-    Fut: std::future::Future<Output = Result<T, eyre::Report>> + Send,
-{
+    Fut: std::future::Future<Output = Result<T, eyre::Report>> + Send, {
     // Try to read from cache
     match read_cache::<T>(key) {
         Ok(Some(cached_value)) => {

crates/cfg/src/core/graph.rs

@@ -61,8 +61,8 @@ pub fn build_cfg(
                 .first()
                 .ok_or_eyre("failed to get first operation")?
                 .last_instruction
-                .opcode
-                == JUMPDEST,
+                .opcode ==
+                JUMPDEST,
         )?;
     }
 

crates/cfg/src/core/mod.rs

@@ -8,17 +8,17 @@ use heimdall_vm::core::vm::VM;
 use petgraph::{dot::Dot, Graph};
 use std::time::{Duration, Instant};
 
-use super::CFGArgs;
+use super::CfgArgs;
 
 use crate::{core::graph::build_cfg, error::Error};
 use tracing::{debug, info};
 
 #[derive(Debug, Clone)]
-pub struct CFGResult {
+pub struct CfgResult {
     pub graph: Graph<String, String>,
 }
 
-impl CFGResult {
+impl CfgResult {
     pub fn as_dot(&self, color_edges: bool) -> String {
         let output = format!("{}", Dot::with_config(&self.graph, &[]));
 
@@ -44,7 +44,7 @@ impl CFGResult {
     }
 }
 
-pub async fn cfg(args: CFGArgs) -> Result<CFGResult, Error> {
+pub async fn cfg(args: CfgArgs) -> Result<CfgResult, Error> {
     // init
     let start_time = Instant::now();
 
@@ -99,5 +99,5 @@ pub async fn cfg(args: CFGArgs) -> Result<CFGResult, Error> {
     debug!("cfg generated in {:?}", start_time.elapsed());
     info!("generated cfg successfully");
 
-    Ok(CFGResult { graph: contract_cfg })
+    Ok(CfgResult { graph: contract_cfg })
 }

crates/cfg/src/interfaces/args.rs

@@ -10,8 +10,8 @@ use heimdall_config::parse_url_arg;
     after_help = "For more information, read the wiki: https://jbecker.dev/r/heimdall-rs/wiki",
     override_usage = "heimdall cfg <TARGET> [OPTIONS]"
 )]
-pub struct CFGArgs {
-    /// The target to generate a CFG for, either a file, bytecode, contract address, or ENS name.
+pub struct CfgArgs {
+    /// The target to generate a Cfg for, either a file, bytecode, contract address, or ENS name.
     #[clap(required = true)]
     pub target: String,
 
@@ -42,13 +42,13 @@ pub struct CFGArgs {
     pub timeout: u64,
 }
 
-impl CFGArgs {
+impl CfgArgs {
     pub async fn get_bytecode(&self) -> Result<Vec<u8>> {
         get_bytecode_from_target(&self.target, &self.rpc_url).await
     }
 }
 
-impl CFGArgsBuilder {
+impl CfgArgsBuilder {
     pub fn new() -> Self {
         Self {
             target: Some(String::new()),

crates/cfg/src/interfaces/mod.rs

@@ -1,4 +1,4 @@
 mod args;
 
 // re-export the public interface
-pub use args::{CFGArgs, CFGArgsBuilder};
+pub use args::{CfgArgs, CfgArgsBuilder};

crates/cfg/src/lib.rs

@@ -4,6 +4,6 @@ mod core;
 mod interfaces;
 
 // re-export the public interface
-pub use core::{cfg, CFGResult};
+pub use core::{cfg, CfgResult};
 pub use error::Error;
-pub use interfaces::{CFGArgs, CFGArgsBuilder};
+pub use interfaces::{CfgArgs, CfgArgsBuilder};

crates/cli/src/args.rs

@@ -4,7 +4,7 @@ use clap::{ArgAction, Args, ValueEnum};
 use heimdall_cache::CacheArgs;
 use heimdall_config::ConfigArgs;
 use heimdall_core::{
-    heimdall_cfg::CFGArgs, heimdall_decoder::DecodeArgs, heimdall_decompiler::DecompilerArgs,
+    heimdall_cfg::CfgArgs, heimdall_decoder::DecodeArgs, heimdall_decompiler::DecompilerArgs,
     heimdall_disassembler::DisassemblerArgs, heimdall_dump::DumpArgs,
     heimdall_inspect::InspectArgs,
 };
@@ -42,7 +42,7 @@ pub enum Subcommands {
     Decompile(DecompilerArgs),
 
     #[clap(name = "cfg", about = "Generate a visual control flow graph for EVM bytecode")]
-    CFG(CFGArgs),
+    Cfg(CfgArgs),
 
     #[clap(name = "decode", about = "Decode calldata into readable types")]
     Decode(DecodeArgs),

crates/cli/src/main.rs

@@ -160,7 +160,7 @@ async fn main() -> Result<()> {
             result.display()
         }
 
-        Subcommands::CFG(mut cmd) => {
+        Subcommands::Cfg(mut cmd) => {
             // if the user has not specified a rpc url, use the default
             if cmd.rpc_url.as_str() == "" {
                 cmd.rpc_url = configuration.rpc_url;

crates/common/src/ether/calldata.rs

@@ -4,13 +4,17 @@ use alloy::primitives::TxHash;
 use eyre::{bail, eyre, Result};
 
 /// Given a target, return calldata of the target.
-pub async fn get_calldata_from_target(target: &str, rpc_url: &str) -> Result<Vec<u8>> {
+pub async fn get_calldata_from_target(target: &str, raw: bool, rpc_url: &str) -> Result<Vec<u8>> {
     // If the target is a transaction hash, fetch the calldata from the RPC provider.
     if let Ok(address) = target.parse::<TxHash>() {
-        return get_transaction(address, rpc_url)
-            .await
-            .map(|tx| tx.input.to_vec())
-            .map_err(|_| eyre!("failed to fetch transaction from RPC provider"));
+        // if raw is true, the user specified that the target is raw calldata. skip fetching the
+        // transaction.
+        if !raw {
+            return get_transaction(address, rpc_url)
+                .await
+                .map(|tx| tx.input.to_vec())
+                .map_err(|_| eyre!("failed to fetch transaction from RPC provider"));
+        }
     }
 
     // If the target is not a transaction hash, it could be calldata.
@@ -34,6 +38,7 @@ mod tests {
 
         let calldata = get_calldata_from_target(
             "0x317907eeece00619fd4418c18a4ec4ebe5c87cdbff808f4b01cc2c6384799837",
+            false,
             &rpc_url,
         )
         .await
@@ -51,6 +56,7 @@ mod tests {
 
         let calldata = get_calldata_from_target(
             "0xf14fcbc8bf9eac48d61719f80efb268ef1099a248fa332ed639041337954647ec6583f2e",
+            false,
             &rpc_url,
         )
         .await
@@ -66,10 +72,31 @@ mod tests {
             std::process::exit(0);
         });
 
-        let calldata =
-            get_calldata_from_target("asfnsdalkfasdlfnlasdkfnalkdsfndaskljfnasldkjfnasf", &rpc_url)
-                .await;
+        let calldata = get_calldata_from_target(
+            "asfnsdalkfasdlfnlasdkfnalkdsfndaskljfnasldkjfnasf",
+            false,
+            &rpc_url,
+        )
+        .await;
 
         assert!(calldata.is_err());
     }
+
+    #[tokio::test]
+    async fn test_get_calldata_when_target_is_calldata_that_is_exactly_32_bytes() {
+        let rpc_url = std::env::var("RPC_URL").unwrap_or_else(|_| {
+            println!("RPC_URL not set, skipping test");
+            std::process::exit(0);
+        });
+
+        let calldata = get_calldata_from_target(
+            "0x317907eeece00619fd4418c18a4ec4ebe5c87cdbff808f4b01cc2c6384799837",
+            true,
+            &rpc_url,
+        )
+        .await
+        .expect("failed to get calldata from target");
+
+        assert!(calldata.len() == 32);
+    }
 }

crates/common/src/ether/signatures.rs

@@ -339,18 +339,18 @@ pub fn score_signature(signature: &str, num_words: Option<usize>) -> u32 {
 
     // prioritize signatures with less numbers
     score -= (signature.split('(').next().unwrap_or("").matches(|c: char| c.is_numeric()).count()
-        as u32)
-        * 3;
+        as u32) *
+        3;
 
     // prioritize signatures with parameters
     let num_params = signature.matches(',').count() + 1;
     score += num_params as u32 * 10;
 
     // count the number of parameters in the signature, if enabled
     if let Some(num_words) = num_words {
-        let num_dyn_params = signature.matches("bytes").count()
-            + signature.matches("string").count()
-            + signature.matches('[').count();
+        let num_dyn_params = signature.matches("bytes").count() +
+            signature.matches("string").count() +
+            signature.matches('[').count();
         let num_static_params = num_params - num_dyn_params;
 
         // reduce the score if the signature has less static parameters than there are words in the

crates/common/src/ether/tokenize.rs

@@ -130,17 +130,17 @@ pub fn tokenize(s: &str) -> Token {
                 let mut op = ch.to_string();
                 iter.next();
                 if let Some(&next_ch) = iter.peek() {
-                    if (ch == '=' && (next_ch == '=' || next_ch == '>'))
-                        || (ch == '&' && next_ch == '&')
-                        || (ch == '|' && next_ch == '|')
-                        || (ch == '<' && next_ch == '=')
-                        || (ch == '>' && next_ch == '=')
-                        || (ch == '!' && next_ch == '=')
-                        || (ch == '+' && next_ch == '+')
-                        || (ch == '-' && next_ch == '-')
-                        || (ch == '*' && next_ch == '*')
-                        || (ch == '>' && next_ch == '>')
-                        || (ch == '<' && next_ch == '<')
+                    if (ch == '=' && (next_ch == '=' || next_ch == '>')) ||
+                        (ch == '&' && next_ch == '&') ||
+                        (ch == '|' && next_ch == '|') ||
+                        (ch == '<' && next_ch == '=') ||
+                        (ch == '>' && next_ch == '=') ||
+                        (ch == '!' && next_ch == '=') ||
+                        (ch == '+' && next_ch == '+') ||
+                        (ch == '-' && next_ch == '-') ||
+                        (ch == '*' && next_ch == '*') ||
+                        (ch == '>' && next_ch == '>') ||
+                        (ch == '<' && next_ch == '<')
                     {
                         op.push(next_ch);
                         iter.next();
@@ -188,9 +188,9 @@ fn parse_literal(iter: &mut std::iter::Peekable<std::str::Chars>) -> String {
     }
 
     // literal validation
-    if literal.starts_with("0x")
-        && literal.len() > 2
-        && literal[2..].chars().all(|c| c.is_ascii_hexdigit())
+    if literal.starts_with("0x") &&
+        literal.len() > 2 &&
+        literal[2..].chars().all(|c| c.is_ascii_hexdigit())
     {
         return literal;
     }

crates/common/src/utils/hex.rs

@@ -1,5 +1,5 @@
 use super::strings::encode_hex;
-use alloy::primitives::{Address, Bytes, FixedBytes, I256, U256};
+use alloy::primitives::{Address, Bytes, FixedBytes, U256};
 
 /// A convenience function which encodes a given EVM type into a sized, lowercase hex string.
 pub trait ToLowerHex {
@@ -20,13 +20,7 @@ impl ToLowerHex for bytes::Bytes {
 
 impl ToLowerHex for U256 {
     fn to_lower_hex(&self) -> String {
-        format!("{:#032x}", self)
-    }
-}
-
-impl ToLowerHex for I256 {
-    fn to_lower_hex(&self) -> String {
-        format!("{:#032x}", self)
+        encode_hex(&self.to_be_bytes_vec())
     }
 }
 

crates/common/src/utils/strings.rs

@@ -330,8 +330,8 @@ pub fn tokenize(s: &str) -> Vec<String> {
             // Check if current character and last character form a compound operator (like "==",
             // ">=", "&&", "||")
             if let Some(last) = last_char {
-                if compound_operator_first_chars.contains(&last)
-                    && (c == '=' || c == '&' || c == '|')
+                if compound_operator_first_chars.contains(&last) &&
+                    (c == '=' || c == '&' || c == '|')
                 {
                     // Remove the last character as a single token
                     tokens.pop();

crates/common/src/utils/sync.rs

@@ -1,7 +1,10 @@
+use std::{future::Future, pin::Pin};
+
 /// Take in a non-async function and await it. This functions should be blocking.
 pub fn blocking_await<F, T>(f: F) -> T
 where
-    F: FnOnce() -> T,
-{
+    F: FnOnce() -> T, {
     tokio::task::block_in_place(f)
 }
+
+pub type BoxFuture<'a, T> = Pin<Box<dyn Future<Output = T> + 'a>>;